[go: up one dir, main page]

NL2038172A - Dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors - Google Patents

Dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors Download PDF

Info

Publication number
NL2038172A
NL2038172A NL2038172A NL2038172A NL2038172A NL 2038172 A NL2038172 A NL 2038172A NL 2038172 A NL2038172 A NL 2038172A NL 2038172 A NL2038172 A NL 2038172A NL 2038172 A NL2038172 A NL 2038172A
Authority
NL
Netherlands
Prior art keywords
state
observation
follows
observation device
meteorological
Prior art date
Application number
NL2038172A
Other languages
Dutch (nl)
Other versions
NL2038172B1 (en
Inventor
Wang Lingtao
Zhang Sheng
Zhai Chongzhi
Peng Chao
Chen Yang
Wang Xiaochen
Du Min
Li Zhenliang
Original Assignee
Chongqing Res Academy Of Eco Environmental Sciences
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing Res Academy Of Eco Environmental Sciences filed Critical Chongqing Res Academy Of Eco Environmental Sciences
Publication of NL2038172A publication Critical patent/NL2038172A/en
Application granted granted Critical
Publication of NL2038172B1 publication Critical patent/NL2038172B1/en

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01W—METEOROLOGY
    • G01W1/00—Meteorology
    • G01W1/10—Devices for predicting weather conditions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01W—METEOROLOGY
    • G01W1/00—Meteorology
    • G01W1/02—Instruments for indicating weather conditions by measuring two or more variables, e.g. humidity, pressure, temperature, cloud cover or wind speed
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01W—METEOROLOGY
    • G01W1/00—Meteorology
    • G01W2001/006—Main server receiving weather information from several sub-stations

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Environmental & Geological Engineering (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Computation (AREA)
  • Ecology (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Environmental Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Atmospheric Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

Disclosed is a dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors. The device includes an agent- based modeling (ABM) framework module, which is configured to define observation devices, where each observation device represents an observation station and has a state and a behavior; a particle swarm optimization module, which is configured to design an objective function, and correlate states of the observation devices with an optimization target, a nesting module of a weather research and forecasting (WRF) model, which is configured to nest the WRF model in an ABM framework as a prediction model of meteorological factors, and a decision making and optimization module, which is configured to execute, on the basis of an update state of each observation device, a decision of adjusting sampling frequency.

Description

DYNAMIC OPTIMIZATION TYPE AUTOMATIC MONITORING DEVICE
COUPLING METEOROLOGICAL AND ATMOSPHERIC
ENVIRONMENTAL FACTORS
TECHNICAL FIELD
[01] The present invention relates to the technical field of environmental monitoring, and particularly relates to a dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors.
BACKGROUND ART
[02] As a field of science of studying various physical phenomena and climate changes in the atmosphere, meteorology involve various phenomena in the atmosphere, such as temperatures, relative humidity, air pressure, wind power and precipitation.
Atmospheric environmental factors refer to various elements that affect the atmospheric environment, such as fine particulate matter, ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide and carbon dioxide, which can affect meteorological changes and climate formation. However, meteorological and atmospheric environmental factors cannot be fully dynamically optimized, so real environmental changes cannot be better dealt with.
SUMMARY
[03] In order to solve the above problems, the present invention provides a dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors.
[04] In order to achieve the above objective, a technical solution employed by the present invention is as follows:
[05] A dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors includes:
[06] an agent-based modeling (ABM) framework module, which is configured to execute steps as follows:
[07] defining observation devices, where each observation device represents an observation station and has a state and a behavior;
[08] in a case of a state representation, observation indexes of each observation device include current meteorological factors and atmospheric environmental factors; and in a case of a behavior representation, each observation device executes a decision behavior according to a current state and environmental change;
[09] a particle swarm optimization module, which is configured to execute steps as follows:
[10] defining an objective function, specifically, designing the objective function, and correlating states of the observation devices with an optimization target;
[11] initializing a particle swarm, specifically, generating a group of particles randomly, where each particle represents decision parameters of a group of observation devices, and
[12] updating positions of the particles, specifically, updating the positions and velocities of the particles through a particle swarm optimization algorithm according to a current objective function value and a historical optimal value, and searching for an optimal solution;
[13] a nesting module of a weather research and forecasting (WRF) model, which is configured to execute steps as follows:
[14] using a WRF model, specifically, nesting the WRF model in an ABM framework as a prediction model of meteorological factors; and
[15] acquiring model output, specifically, acquiring predicted meteorological factor values from the WRF model;
[16] a data assimilation and state update module, which is configured to execute steps as follows:
[17] assimilating data, specifically, assimilating the output of the WRF model to actually-observed data, and obtaining more accurate state data; and
[18] updating the states of the observation devices, specifically, updating the state of each observation device according to the assimilated data and a behavior model of the observation devices; and
[19] a decision making and optimization module, which is configured to execute steps as follows:
[20] making a decision, specifically, executing, on the basis of an update state of each observation device, a decision of adjusting sampling frequency; and
[21] optimizing the particle swarm, specifically, running the particle swarm optimization algorithm again, updating a position of the particle swarm, responding to a new state and decision situation, and searching for a better parameter configuration.
[22] Further, in the ABM framework module,
[23] each observation device is represented by a group of variables, and the variables represent the state and the behavior of the observation device. It is assumed that a state variable of the observation device is S and a behavior variable is A.
[24] A state representation of each observation device includes the current meteorological factors and atmospheric environmental factors, which include a temperature (T), relative humidity (RH), a wind velocity (V), precipitation (p), and mass concentrations of fine particulate matter (PM2.5), ozone (O3), nitrogen dioxide (NO>), sulfur dioxide (SO), carbon monoxide (CO) and carbon dioxide (CO2).
[25] The state representation 1s as follows: S=(T, RH, V, p, PM2.5, O3, NO2, SOs,
CO, CO)
[26] The observation device executes the decision behavior according to the current state and environmental change, and a strategy is represented by a function.
[27] A behavior model is as follows: A=f(S).
[28] Further, in the particle swarm optimization module,
[29] the optimization target includes nonlinear optimization fitting of a minimum of the objective function. The objective function is as follows: , where
[30] f(x) = wil + (PM2.5 — PM2.54gesireg)? + W2 * (03 — 03 gesireq)? + W3 * (NO2 — NO2gesireq)? + W4 * (S02 — SO2gesireg)? + W5 * (CO — COgesireg)? +
W6 * (CO2 — CO24esirea)® + W7 * (T — Taesired)? + W8 * (RH — RHgesireg): + W9 * (V — Viesirea)® + W10 * (p — Paesirea)?
[31] xis the decision parameter of the observation device, PM2.5, O3, NO», SO, CO and CO2 are the mass concentrations of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, T, RH,
V and p are current temperature, relative humidity, wind velocity and precipitation,
PM2 desired, O3desired, NO2desired, SO2desired, COdesirea and COadesired, and Taesired, RHdesired,
Viesired and Pdesired are desired mass concentration values of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, and desired temperature, relative humidity, wind velocity and precipitation values, which are acquired on the basis of historical simulation data, and wl, w2, w3, wd, w5, wo, w7, w8, w9 and wl0 are weights for balancing significance of different factors.
[32] A group of particles are randomly generated, and each particle represents the decision parameters of the group of observation devices. It is assumed that the decision parameter of the particle is represented as vector x=[X1, X2, X3,..., Xn], where n is a number of the decision parameter.
[33] The positions and velocities of the particles are updated according to the current objective function value and the historical optimal value, and the optimal solution is searched for. Particle velocity and position update formulas are as follows:
[34] vigry=wHvigtcl rand *(Poesi-Xi)+c2 rand (Soes-Xi)), and
[35] Xie =Xio Vin). where
[36] vo is a velocity of particle i at time t, xiqt Is a position of particle i at time t, Poesti is a historical optimal position of particle i, Sbest is a historical optimal position of the entire particle swarm, w is an inertia weight, cl and c2 are learning factors, and rand () is a random number.
[37] Further, in the nesting module of the WRF model,
[38] a change in air quality is described by a quality equation, and a mathematical expression of the quality equation is as follows: , where 39] LV (pu) =0
[40] pis density of the air, t is time, u is a velocity vector of the air, and V represents a divergence operator for representing divergence of a vector field.
[41] A continuity equation is discretized, a numerical format is obtained, and solving is carried out on a computer.
[42] Within each time step, according to density distribution and a velocity field at current time, the quality equation is solved by using the numerical format, and density distribution at next time is updated.
[43] Required meteorological factors are extracted from a simulation result after each time step of simulation.
[44] It is assumed that state variables of the concerned observation device is temperature T, relative humidity RH, wind velocity V and precipitation p. The state representation of the observation device is S=(T, RH, V, p).
[45] Within each time step, the WRF model simulates temperature T, relative humidity RH, wind velocity V and precipitation p at the current time, and Tsim, RHsim,
Vsim and psim are obtained.
[46] Simulated temperature Tsim, relative humidity RHsim, wind velocity Vsim and precipitation Psim as state updates of the observation device. In a state update model of the observation device, weights of simulated values and observed values are balanced by using weighting factors. 5 [47] A temperature state is updated as follows: ,
[48] Thew = 0 X Tsim + (1 = 0) X Topservea
[49] a relative humidity state is updated as follows: ,
[50] RHpew = BX RHsm + (1 — B) X RHopserved
[51] a wind velocity state is updated as follows: , and
[52] Vaew = YX Vsim + (1 = ¥) X Vopservea
[53] a precipitation state is updated as follows: , where
[54] Phew = 8 X Psim + (1 = 8) X Popserved
[55] a, B, y and ô are weighting factors, Topserved, RHopservea, Vobserved and Pobserved are actually-observed temperature, relative humidity, wind velocity and precipitation values.
[56] State S=(Tnew, RHuew, View, Pnew) of the observation device is updated.
[57] Compared with the prior art, the present invention has technical progress as follows:
[58] ABM in the present invention allows modeling to be carried out on individual observation devices in the system. These observation devices may change according to influences of environments and other observation devices. The dynamic optimization problem of coupling meteorological and atmospheric environmental factors is solved.
Each observation device may represent an observation station or region. The ABM provides a framework for simulating interactions between observation devices and adapting to environmental changes such that complex system behaviors can be better reflected.
[59] Particle swarm optimization (PSO) is an optimization algorithm based on swarm intelligence, and is suitable for searching for a global optimal solution of a complex problem. In the method, each particle represents a group of solutions, and the entire particle swarm searches for the optimal solution through constant update. In order to solve the dynamic optimization problem, the PSO can adjust positions of the particles to search for changes in optimal solutions by adapting to new states and data. Through the
PSO algorithm, the decision parameters of the observation devices can dynamically adapt to environmental changes such that system performance can be better optimized.
[60] The WRF model is used to predict changes in meteorological factors in a period of time in the future such that more accurate meteorological data can be provided to guide decisions of the observation devices. By nesting the WRF model, more accurate meteorological data prediction results can be obtained such that the observation devices can better make decisions in dynamic environments. A simulation result of the WRF model is assimilated to the actually-observed data, such that accuracy of the state data is further improved.
[61] Cooperation between the agent-based modeling (ABM) and the PSO is as follows: the ABM provides a mechanism for interactions between the observation devices such that the system can be dynamically adjusted according to individual adaptability. The PSO optimizes the decision parameter of each observation device through such interactions, so as to achieve a global optimization target. The ABM provides real-time data and feedback, and the PSO guides, by using such information, the observation devices to search for optimal solutions in complex environments.
[62] Cooperation between the ABM and the WRF model is as follows: the ABM guides the decisions of the observation devices by using meteorological factors predicted by the WRF model. By nesting the WRF model, the observation devices can obtain more accurate meteorological prediction information, so as to better adapt to environmental changes. Such cooperation enables the observation devices to more accurately consider meteorological factors during decision making.
[63] Cooperation between the PSO and the WRF model is as follows: the PSO algorithm can take a simulation result of the WRF model as external information to guide a search of the particle swarm. The decision parameters of the observation devices can optimize, through the PSO algorithm, the prediction data provided by the WRF model.
Such cooperation makes decisions of the observation devices more targeted and better able to adapt to changes in dynamic environments.
[64] To sum up, the cooperations between the ABM, the PSO and the WRF model enable the observation devices to better deal with environmental changes in dynamic optimization problems, and better decisions and system performance can be achieved.
BRIEF DESCRIPTION OF THE DRAWINGS
[65] The accompanying drawings serve as a constituent part of the description to provide a further understanding of the present invention. The accompanying drawings and examples of the present invention are used for explaining the present invention and do not constitute a limitation on the present invention.
[66] In the figure,
[67] FIG. 11s a system structure diagram of the present invention.
DETAILED DESCRIPTION OF THE EMBODIMENTS
[68] The following several particular examples can be combined with each other, and the same or similar concepts or processes may not be repeated in some examples.
Examples of the present invention will be described below in combination with accompanying drawings.
[69] Asshownin FIG. 1, the present invention provides a dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors. The device includes:
[70] an agent-based modeling (ABM) framework module, which is configured to execute steps as follows:
[71] Define observation devices, where each observation device represents a specific region or observation station and has a state and a behavior.
[72] A state representation is as follows: observation indexes of each observation device include the current meteorological factors and atmospheric environmental factors, such as a temperature, relative humidity, a wind velocity, precipitation, and mass concentrations of fine particulate matter, ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide and carbon dioxide.
[73] A behavior model is as follows: the observation device executes the decision behavior according to the current state and environmental change.
[74] The device further includes a particle swarm optimization module, which is configured to execute steps as follows:
[75] Define an objective function, specifically, design the objective function, and correlate states of the observation devices with an optimization target (nonlinear optimization fitting of a minimum of the objective function).
[76] [Initialize a particle swarm, specifically, generate a group of particles randomly, where each particle represents decision parameters of a group of observation devices. [771 Update positions of the particles, specifically, update the positions and velocities of the particles through a particle swarm optimization (PSO) algorithm according to a current objective function value and a historical optimal value, and search for an optimal solution.
[78] The device further includes a nesting module of a weather research and forecasting (WRF) model, which is configured to execute steps as follows:
[79] Use a WRF model, specifically, nest the WRF model in an ABM framework as a prediction model of meteorological factors. The WRF will simulate weather changes in a period of time in the future on the basis of an initial condition and a boundary condition.
[80] Acquire model output, specifically, acquire predicted meteorological factor values, such as a temperature, relative humidity, a wind velocity and precipitation, from the WRF model.
[81] The device further includes a data assimilation and state update module, which is configured to execute steps as follows:
[82] Assimilate data, specifically, assimilate the output of the WRF model to actually-observed data, and obtain more accurate state data.
[83] Update the states of the observation devices, specifically, update the state, which includes meteorological factors and atmospheric environmental factors, of each observation device according to the assimilated data and a behavior model of the observation devices.
[84] The device further includes a decision making and optimization module, which is configured to execute steps as follows:
[85] Make a decision, specifically, execute, on the basis of an update state of each observation device, a corresponding decision such as adjustment of sampling frequency, and achieve the optimization target.
[86] Carry out PSO optimization, specifically, run the PSO algorithm again, update a position of the particle swarm, respond to a new state and decision situation, and search for a better parameter configuration.
[87] Specifically, in the ABM framework module,
[88] 1.1 The observation devices are defined as follows:
[89] Each observation device is represented by a group of variables, and the variables represent the state and the behavior of the observation device. It is assumed that a state variable of the observation device is S and a behavior variable is A.
[90] 1.2 A state representation is as follows:
[91] observation indexes of a state representation of each observation device include the current meteorological factors and atmospheric environmental factors, such as a temperature (T), relative humidity (RH), a wind velocity (V), precipitation (p), and mass concentrations of fine particulate matter (PM2.5), ozone (Os), nitrogen dioxide (NO2), sulfur dioxide (SO2), carbon monoxide (CO) and carbon dioxide (CO).
[92] The state representation is as follows: S=(T, RH, V, p, PM2.5, O3, NO2, SOs,
CO, CO»).
[93] 1.3 A behavior model is as follows:
[94] the observation device executes the decision behavior according to the current state and environmental change. It is assumed that the behavior of the observation device is carried out on the basis of a current state of the observation device and a strategy. The strategy may be represented by a function.
[95] A behavior model is as follows: A=f(S).
[96] Behavior A of the observation device may be a result of a series of decisions, such as adjustment of frequency of the observation device, and choice of a method for transmitting data. Specific function f may be defined according to a specific situation of the problem, which may involve some optimization strategies.
[97] For instance, a decision of a behavior may be represented by a linear function as follows:
[98] A=wl*T+w2*RH+w3*V+w4*p+wS*PM2.5+wo6*O1+w7*NO,+w8*SOr+w9*
CO+w10*CO2, where
[99] wl, w2, w3, w4, w5, wo, w7, w8, w9 and wl0 are weights that may be determined through the PSO algorithm.
[100] Specifically, in the particle swarm optimization module,
[101] 2.1 The objective function is defined as follows: the optimization target is nonlinear optimization fitting of a minimum of the objective function. The objective function may be defined as follows: where
[102] f(x) = wl + (PM2.5 — PM2.54gesireg)? + W2 * (03 — 03 gesireq)? + W3 * (NO2 — NO2 gesired)? + W4 * (SO2 — SO24esired)? + W5 * (CO — COgesirea)? + wb * (CO2 — CO24esirea)® + W7 + (T — Taesireg)? + W8 * (RH — RHgesirea): + W9 * (V — Viesirea)® + W10 * (P — Paesired)’
[103] x isthe decision parameter of the observation device, PM2.5, O3, NO», SO, CO and CO: are mass concentrations of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, T, RH, V and p are current temperature, relative humidity, wind velocity and precipitation,
PM2 desired, O3desired, NO2desired, SO2desired, COdesirea and COadesired, and Taesired, RHdesired,
Vdesired and Pdesirea are desired mass concentration values of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, and desired temperature, relative humidity, wind velocity and precipitation values, which are acquired on the basis of historical simulation data, and wl, w2, w3, wA4, w5, wo, w7, w8, w9 and wl0 are weights for balancing significance of different factors;
[104] 2.2 A particle swarm is initialized as follows:
[105] a group of particles are randomly generated, and each particle represents the decision parameters of the group of observation devices. It is assumed that the decision parameter of the particle is represented as vector x=[x1, x2, x3,..., Xu], where n is a number of the decision parameter.
[106] 2.3 The positions of the particles are updated as follows:
[107] the PSO algorithm updates the positions and velocities of the particles according to the current objective function value and the historical optimal value, and searches for the optimal solution. Particle velocity and position update formulas are as follows:
[108] vi =w*viptcl *rand()* (poesti-Xiay)+c2 *rand() *( goest-Xiay), and
[109] Xie =Xi0 Vi), where
[110] vig 1s a velocity of particle 1 at time t, Xj) 15 a position of particle 1 at time t, poesii is a historical optimal position of particle i, goes is a historical optimal position of the entire particle swarm, w is an inertia weight, cl and c2 are learning factors, and rand () is a random number.
[111] By iteratively updating the velocities and positions of the particles, the particle swarm will converge to a vicinity of the optimal solution.
[112] Specifically, in the nesting module of the WRF model,
[113] a quality equation (continuity equation) in the WRF model is used as an instance to illustrate how the WRF model is nested in the ABM framework such that weather changes in a period of time in the future can be simulated. A change in air quality is described by the quality equation, and a mathematical expression of the quality equation is as follows: , where
[114] LV (pew) 0
[115] pis density of the air, tis time, u is a velocity vector of the air, and V represents a divergence operator for representing divergence of a vector field.
[116] The WRF model may be nested specifically as follows:
[117] the quality equation is converted into a numerical format as follows: the continuity equation is discretized, a numerical format is obtained, and solving is carried out on a computer. Spaces and time are generally divided into grids, and a difference method is used to approximate derivatives.
[118] An initial condition and a boundary condition are set as follows: before simulation, the initial condition (density distribution at initial time) and the boundary condition (a condition at a boundary of a simulation region) are required to be set for the model. These conditions are generally based on real observed data.
[119] Time stepping is carried out as follows: within each time step, according to density distribution and a velocity field at current time, the quality equation is solved by using the numerical format, and density distribution at next time is updated.
[120] A simulation result is acquired as follows: required meteorological factors, such as a temperature, relative humidity, a wind velocity and precipitation, are extracted from the simulation result after each time step of simulation.
[121] In a process of nesting the WRF model, a simulation result is applied to state updates of the observation devices such that the observation devices can operate on the basis of more accurate meteorological forecast information during decision making and optimization.
[122] When the output of the WRF model is acquired as one of the states of the observation device, it can be specifically described how a simulation result of the continuity equation is applied to the state update of the observation device. It is assumed that state variables of the concerned observation device are temperature T, relative humidity RH, wind velocity V and precipitation p, and the state representation of the observation device 1s S=(T, RH, V, p). Specific steps are as follows:
[123] Within each time step, the WRF model simulates temperature T, relative humidity RH, wind velocity V and precipitation p at the current time, and Tsim, RHsim,
Vsim and Psim are obtained.
[124] Simulated temperature Tsm, relative humidity RHsim, wind velocity Vsim and precipitation psim serve as state updates of the observation device, and in a state update model of the observation devices, weights of simulated values and observed values are balanced by using weighting factors.
[125] A temperature state is updated as follows: ,
[126] Thew = aX Tsim + (1 — 0) X Topserved
[127] a relative humidity state is updated as follows: ,
[128] RHpew = BX RHsim + (1 — B) X RHopserveg
[129] a wind velocity state is updated as follows: ,
[130] View = YX Vsim + (1 — ¥) X Vopserved
[131] a precipitation state is updated as follows: , where
[132] Phew = 8 X Psim + (1 — 8) X Pobserved
[133] a, B, y and are weighting factors, Topservea, RHobserved, Vobserved and Pobserved are actually-observed temperature, relative humidity, wind velocity and precipitation values.
[134] State S=(Tnew, RHuew, View, Prew) is updated of the observation device.
[135] Through the above steps, the simulation output of the WRF model is combined with the state update of the observation device such that a more accurate state of the observation device can be obtained. The observation devices can carry out decision making and optimization on the basis of actually-simulated meteorological factor information, so as to better respond to dynamically-changing environments.
[136] Specifically, in a data assimilation and state update module,
[137] 4.1 Data assimilation is as follows:
[138] data assimilation is a process for combining a model simulation result with actually-observed data such that state data can be improved. An update is carried out through a Kalman filter method. A state update and an observation update during
Kalman filtering may be represented by formulas as follows:
[139] state prediction (a time update) is as follows: ,
[140] ti = FR ije-1 + Bugg
[141] error covariance prediction (a time update) is as follows: ,
[142] Pi 1= FP jp. F" +Q
[143] a Kalman gain is as follows: ,
[144] K,=P_,HT(P_,HT +R)!
[145] state correction (an observation update) is as follows: , and
[146] & =p + Ke(Ze 7 H8t-1)
[147] error covariance correction (an observation update) is as follows: , where
[148] Pit = (1 —- KH) Pt;
[149] £t-1 is the state data after the time update, P: ‚1-1 is an error covariance of the state data after the time update, F is a state transition matrix describing how a state changes with time, B is a control input matrix for taking a control signal into account, ue-1 is the control signal, Q is a process noise covariance matrix, K; is the Kalman gain,
RH is an observation matrix describing how observation is correlated to the state, R 1s an observation noise covariance matrix, and z; is actually-observed data.
[150] 4.2 State updates of the observation devices are as follows:
[151] according to the assimilated state data, the state including meteorological factors and atmospheric environmental factors of each observation device is updated.
[152] For instance, itis assumed that a state of a concerned observation device is S=(T,
RH, V, p, PM2.5, Os, NO;, SO», CO, CO), where T is a temperature, RH is relative humidity, V is a wind velocity, p is precipitation, and PM2.5, O3, NO2, SO;, CO and
CO: are mass concentrations of fine particulate matter, ozone, nitrogen dioxide, sulfur dioxide, carbon monoxide and carbon dioxide. The assimilated state data may be applied to the state updates of the observation devices by using a state correction formula of
Kalman filtering.
[153] Thew = Tet
[154] RHaew = RB
[55] Vrew = Vet
[156] Pnew = Vet
[157] PM2.5,., = PM2.5,,
[158] OZnew = 03
[159] NOZnew = NOZy,
[160] SOZnew = SOZ
[161] COnew = Oy
[162] COZnew = CÔ2t
[163] Through the above steps, the simulation result of the WRF model can be assimilated to the actually-observed data, and then the assimilated state data can be applied to the state updates of the observation devices such that more accurate state data and decision making can be achieved.
[164] Specifically, in a decision making and optimization module,
[165] 5.1 Decision making is as follows: sampling frequency is adjusted.
[166] Each observation device is responsible for observing meteorological factors and atmospheric environmental factors in an environment. The optimization target can be achieved by adjusting the sampling frequency of the observation devices. The sampling frequency of the observation devices may be defined as variable fsampte. The optimization problem is the nonlinear optimization fitting of the minimum of the objective function:
[167] An optimized objective function is as follows: where
US f(x) = wl * (PM2.5 — PM2.5gesired)? + W2 * (03 — O3gesirea)? + W3 * (NO2 — NO2gesired)® + W4 * (S02 — SO24esired)? + W5 * (CO — COgesirea)? +
W6 * (CO2 — CO2gesired)® + W7 * (T — Tyesired)® + W8 * (RH — RHaesireg)? + W9 * (V — Viesirea)® + W10 * (p — Pdesirea)*
[169] xis the decision parameter of the observation device, PM2.5, O3, NO2, SO,, CO and CO: are mass concentrations of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, T, RH, V and p are current temperature, relative humidity, wind velocity and precipitation,
PM2 Sdesired, Osdesired, NO2desired, SO2desired, COdesired and COzdesired, and Tdesired, RHdesired,
Vesa and Pdesirea are desired mass concentration values of the fine particulate matter, the ozone, the nitrogen dioxide, the sulfur dioxide, the carbon monoxide and the carbon dioxide, and desired temperature, relative humidity, wind velocity and precipitation values, which are acquired on the basis of historical simulation data, and wl, w2, w3, wA4, w5, wo, w7, w8, w9 and wl0 are weights for balancing significance of different factors;
[170] 5.2 PSO Optimization is as follows: sampling frequency is adjusted.
[171] Optimal sampling frequency of the observation device may be searched for through the PSO algorithm such that the optimization target can reach the minimum.
The PSO algorithm will update the position of particle swarm to optimize the objective function. The sampling frequency of the observation devices may serve as the decision parameters of particles, and is updated according to the optimization target.
[172] During each PSO iteration, the positions and velocities of the particles are updated such that a global optimal solution can be approached. Therefore, the sampling frequency of the observation device can be adjusted, and the optimization target can reach the minimum.
[173] To sum up, through decision making and PSO optimization, the sampling frequency of the observation devices can be adjusted to be based on the optimization target, so as to better respond to the dynamic changes in meteorological and atmospheric environments.
[174] Finally, it should be noted that the above examples are merely the preferred examples of the present invention and are not intended to limit the present invention.
Although the present invention is described in detail with reference to the foregoing examples, a person skilled in the art can still make modifications to the technical solutions described in various foregoing examples, or make equivalent substitution to some technical features in the technical solutions. Any modification, equivalent substitution, improvement, etc. within the spirit and principles of the present invention shall fall within the scope of protection of the present invention.

Claims (4)

Conclusies l. Automatische bewakingsinrichting van het dynamischeoptimalisatietype die meteorologische en atmosferische omgevingsfactoren koppelt, die het volgende omvat: een agent gebaseerde modelering (“agent-based modeling”, ABM) raamwerkmodule, die is geconfigureerd om de stappen als volgt uit te voeren: het definiéren van observatie-inrichtingen, waarbij elke observatie-inrichting een observatiestation vertegenwoordigt en een toestand en een gedrag heeft; in het geval van een toestandsrepresentatie omvatten observatie-indexen van elke observatie-inrichting huidige meteorologische factoren en atmosferische omgevingsfactoren; en in het geval van een gedragsrepresentatie voert elke observatie- inrichting een beslissingsgedrag uit in overeenstemming met een huidige toestand en omgevingsverandering; een deeltjeszwermoptimalisatiemodule, die is geconfigureerd om als volgt stappen uit te voeren: het definiëren van een objectieve functie, in het bijzonder het ontwerpen van de objectieve functie, en het correleren van de toestanden van de observatie-inrichtingen met een optimalisatiedoel; het initialiseren van een deeltjeszwerm, in het bijzonder, het willekeurig genereren van een groep deeltjes, waarbij elk deeltje beslissingsparameters vertegenwoordigt van een groep observatie-inrichtingen; en het bijwerken van de posities van de deeltjes, in het bijzonder het bijwerken van de posities en snelheden van de deeltjes door middel van een deeltjeszwermoptimalisatiealgoritme volgens een huidige objectieve functiewaarde en een historische optimale waarde, en het zoeken naar een optimale oplossing; een inbeddingsmodule van een weeronderzoek- en voorspellingsmodel (“weather research and forecasting”, WRF), die is geconfigureerd om als volgt stappen uit te voeren: het gebruiken van een WRF-model, in het bijzonder, het inbedden van het WRF- model in een ABM-framework als een voorspellingsmodel van meteorologische factoren; en het verwerven van modeluitvoer, in het bijzonder, het verwerven van voorspelde meteorologische factorwaarden uit het WRF-model,Conclusions l. Automatic monitoring device of the dynamic optimization type coupling meteorological and atmospheric environmental factors, comprising: an agent-based modeling (“ABM”) framework module, configured to perform the steps as follows: defining observation devices, each observation device representing an observation station and having a state and a behavior; in the case of a state representation, observation indexes of each observation device include current meteorological factors and atmospheric environmental factors; and in the case of a behavior representation, each observation device performs a decision behavior in accordance with a current state and environmental change; a particle swarm optimization module, configured to perform the steps as follows: defining an objective function, in particular designing the objective function, and correlating the states of the observation devices with an optimization goal; initializing a particle swarm, in particular, randomly generating a group of particles, each particle representing decision parameters of a group of observation devices; and updating the positions of the particles, in particular updating the positions and velocities of the particles by means of a particle swarm optimization algorithm according to a current objective function value and a historical optimal value, and searching for an optimal solution; a weather research and forecasting model embedding module (“WRF”) configured to perform the following steps: using a WRF model, in particular, embedding the WRF model into an ABM framework as a meteorological factor prediction model; and acquiring model output, in particular, acquiring forecast meteorological factor values from the WRF model, een data-assimilatie- en toestandbijwerkingsmodule voor, die is geconfigureerd om de stappen als volgt uit te voeren: het assimileren van data, in het bijzonder, het assimileren van de uitvoer van het WRF-model met feitelijk waargenomen data, en het verkrijgen van nauwkeurigere toestanddata; en het bijwerken van de toestanden van de observatie-inrichtingen, in het bijzonder, het bijwerken van de toestand van elke observatie-inrichting volgens de geassimileerde data en een gedragsmodel van de observatie-inrichtingen; en een besluitvormings- en optimalisatiemodule, die 1s geconfigureerd om als volgt stappen uit te voeren: het nemen van een beslissing, in het bijzonder het uitvoeren, op basis van een bijwerkingstoestand van elke observatie-inrichting, van een beslissing over het aanpassen van de bemonsteringsfrequentie; en het optimaliseren van de deeltjeszwerm, in het bijzonder, het opnieuw uitvoeren van het deeltjeszwermoptimalisatie-algoritme, het bijwerken van een positie van de deeltjeszwerm, het reageren op een nieuwe toestand en beslissingssituatie, en het zoeken naar een betere parameterconfiguratie.a data assimilation and state updating module, which is configured to perform the steps as follows: assimilating data, in particular, assimilating the output of the WRF model with actual observed data, and obtaining more accurate state data; and updating the states of the observation devices, in particular, updating the state of each observation device according to the assimilated data and a behavioral model of the observation devices; and a decision making and optimization module, which is configured to perform the steps as follows: making a decision, in particular, executing, based on an update state of each observation device, a decision on adjusting the sampling frequency; and optimizing the particle swarm, in particular, re-executing the particle swarm optimization algorithm, updating a position of the particle swarm, responding to a new state and decision situation, and searching for a better parameter configuration. 2. Automatische bewakingsinrichting van het dynamischeoptimalisatietype die meteorologische en atmosferische omgevingsfactoren koppelt volgens conclusie 1, waarbij in de ABM-frameworkmodule, elke observatie-inrichting wordt vertegenwoordigd door een groep variabelen, waarbij de variabelen de toestand en het gedrag van de observatie-inrichting vertegenwoordigen, en er wordt aangenomen dat een toestandvariabele van de observatie-inrichting S is en een gedragsvariabele A is; een toestandsweergave van elke observatie-inrichting omvat de huidige meteorologische factoren en atmosferische omgevingsfactoren, waaronder een temperatuur (T), relatieve vochtigheid (“relative humidity”, RH), een windsnelheid “velocity”, V), neerslag (“precipitation”, p) en massaconcentraties van fijn stof (PM2,5), ozon (O3), stikstofdioxide (NO:), zwaveldioxide (SO2), koolmonoxide (CO) en kooldioxide (CO2); de toestandsweergave is als volgt: S=(T, RH, V, p, PM2,5, Os, NO:;, SO, CO, CO»)2. An automatic monitoring device of the dynamic optimization type linking meteorological and atmospheric environmental factors according to claim 1, wherein in the ABM framework module, each observation device is represented by a group of variables, the variables representing the state and behavior of the observation device, and it is assumed that a state variable of the observation device is S and a behavior variable is A; a state representation of each observation device includes the current meteorological factors and atmospheric environmental factors, including a temperature (T), relative humidity (RH), a wind velocity (V), precipitation (p) and mass concentrations of particulate matter (PM2.5), ozone (O3), nitrogen dioxide (NO:), sulfur dioxide (SO2), carbon monoxide (CO) and carbon dioxide (CO2); the state representation is as follows: S=(T, RH, V, p, PM2.5, Os, NO:;, SO, CO, CO») de observatie-inrichting voert het beslissingsgedrag uit volgens de huidige toestand en veranderingen in de omgeving, en een strategie wordt weergegeven door een functie; en een gedragsmodel is als volgt: A=f(S).the observation device carries out the decision behavior according to the current state and changes in the environment, and a strategy is represented by a function; and a behavior model is as follows: A=f(S). 3. Automatische bewakingsinrichting van het dynamischeoptimalisatietype die meteorologische en atmosferische omgevingsfactoren koppelt volgens conclusie 2, waarbij in de deeltjeszwermoptimalisatiemodule, het optimalisatiedoel omvat niet-lineaire optimalisatieaanpassing van een minimum van de objectieve functie, en de objectieve functie 1s als volgt: waarbij f(x) = wl * (PM2,5 — PM2,5gewenst)? + W2 * (03 — O3gewenst)? + W3 * (NO2 — NOZgewenst)* + W4 * (SO2 — SOZgewenst)? + W5 * (CO — COgewenst)? + W6 * (CO2 — CO2gewenst)? + W7 * (T — Tgewenst)? + WB * (RH — RHgewenst)? + w9 x (V - Veewenst)? + w10 * (p - Pgewenst)? x de beslissingsparameter is van de observatie-inrichting, PM2,5, O3, NO, SO, CO en CO: de massaconcentraties zijn van fijn stof, de ozon, de stikstofdioxide, de zwaveldioxide, de koolmonoxide en de koolstof dioxide, T, RH, V en p zijn huidige temperatuur, relatieve vochtigheid, windsnelheid en neerslag, PM2,5gewenst, O3gewenst, NO2gewenst, SO2gewenst, COgewenst en CO2pewenst, en Tgewenst, RHgewenst, V gewenst EN Pgewenst Zijn de gewenste massaconcentratiewaarden van de fijn stof, de ozon, de stikstofdioxide, de zwaveldioxide, de koolmonoxide en de kooldioxide, en gewenste temperatuur-, relatieve vochtigheid-, windsnelheid- en neerslagwaarden, die zijn verkregen op basis van historische simulatiedata, en wl, w2, w3, w4, w5, wo, w7, w8, w9 en w10 zijn gewichten voor het balanceren van de significantie van verschillende factoren; een groep deeltjes willekeurig gegenereerd wordt, en elk deeltje vertegenwoordigt de beslissingsparameters van de groep observatie-inrichtingen; er aangenomen wordt dat de beslissingsparameter van het deeltje wordt weergegeven als vector x=[xX1, X2, X3,..., Xu], waarbij n een getal van de beslissingsparameter is; de posities en snelheden van de deeltjes worden bijgewerkt volgens de huidige objectieve functiewaarde en de historische optimale waarde, en er naar de optimale oplossing gezocht wordt, en formules voor het bijwerken van de deeltjessnelheid en positie als volgt zijn: Vier =w*vigtel “rand (Poesi-xiw)+e2 *rand () "“(Sbest-Xi), en Kite Ki Vie D, waarbij vig 1s een snelheid van deeltje i op tijdstip t, xi) is een positie van deeltje 1 op tijdstip t, Poes Is een historisch optimale positie van deeltje i, Soest Is een historisch optimale positie van de gehele deeltjeszwerm, w is een traagheidsgewicht, cl en C2 zijn leerfactoren, en rand () is een willekeurig getal.3. An automatic monitoring device of the dynamic optimization type coupling meteorological and atmospheric environmental factors according to claim 2, wherein in the particle swarm optimization module, the optimization objective comprises nonlinear optimization adjustment of a minimum of the objective function, and the objective function 1s as follows: where f(x) = w1 * (PM2.5 — PM2.5desired)? + w2 * (O3 — O3desired)? + w3 * (NO2 — NOZdesired)* + w4 * (SO2 — SOZdesired)? + w5 * (CO — COdesired)? + w6 * (CO2 — CO2desired)? + w7 * (T — Tdesired)? + wB * (RH — RHdesired)? + w9 x (V — Vedesired)? + w10 * (p — Pdesired)? x is the decision parameter of the observation device, PM2.5, O3, NO, SO, CO and CO are the mass concentrations of particulate matter, ozone, nitrogen dioxide, sulphur dioxide, carbon monoxide and carbon dioxide, T, RH, V and p are current temperature, relative humidity, wind speed and precipitation, PM2.5desired, O3desired, NO2desired, SO2desired, COdesired and CO2pedesired, and Tdesired, RHdesired, Vdesired AND Pdesired are the desired mass concentration values of particulate matter, ozone, nitrogen dioxide, sulphur dioxide, carbon monoxide and carbon dioxide, and desired temperature, relative humidity, wind speed and precipitation values, which are obtained from historical simulation data, and wl, w2, w3, w4, w5, wo, w7, w8, w9 and w10 are weights for balancing the significance of different factors; a group of particles is randomly generated, and each particle represents the decision parameters of the group of observation devices; it is assumed that the decision parameter of the particle is represented as vector x=[xX1, X2, X3,..., Xu], where n is a number of the decision parameter; the positions and velocities of the particles are updated according to the current objective function value and the historical optimal value, and the optimal solution is sought, and formulas for updating the particle velocity and position are as follows: Vier =w*vigtel “rand (Poesi-xiw)+e2 *rand () "“(Sbest-Xi), and Kite Ki Vie D, where vig 1s is a velocity of particle i at time t, xi) is a position of particle 1 at time t, Poes Is a historical optimal position of particle i, Soest Is a historical optimal position of the entire particle swarm, w is an inertial weight, cl and C2 are learning factors, and rand () is a random number. 4. Automatische bewakingsinrichting van het dynamischeoptimalisatietype die meteorologische en atmosferische omgevingsfactoren koppelt volgens conclusie 3, waarbij: de inbeddingsmodule van het WRF-model omvat: een verandering in de luchtkwaliteit wordt beschreven door een kwaliteitsvergelijking, en een wiskundige uitdrukking van de kwaliteitsvergelijking is als volgt: waarbij dp4. An automatic monitoring device of the dynamic optimization type linking meteorological and atmospheric environmental factors according to claim 3, wherein: the embedding module of the WRF model comprises: a change in air quality is described by a quality equation, and a mathematical expression of the quality equation is as follows: where dp 2. +Vx(p*xu)=0 p de dichtheid van de lucht is, t tijd is, u een snelheidsvector van de lucht is, en V een divergentie-operator vertegenwoordigt voor het weergeven van de divergentie van een vectorveld; een ononderbrokensvergelijking wordt gediscretiseerd, een numeriek formaat wordt verkregen en het oplossen wordt uitgevoerd op een computer; binnen elke tijdstap wordt, afhankelijk van de dichtheidsverdeling en een snelheidsveld op het huidige tijdstip, de kwaliteitsvergelijking opgelost door gebruik te maken van het numerieke formaat, en wordt de dichtheidsverdeling de volgende keer bijgewerkt; de vereiste meteorologische factoren worden na elke tijdstap van de simulatie uit een simulatieresultaat gehaald; er wordt aangenomen dat de toestandsvariabelen van de betreffende observatie- inrichting temperatuur T, relatieve vochtigheid RH, windsnelheid V en neerslag p zijn, en dat de toestandsweergave van de observatie-inrichting S=(T, RH, V, p) is; binnen elke tijdstap simuleert het WRF-model temperatuur T, relatieve vochtigheid RH, windsnelheid V en neerslag p op het huidige tijdstip, en Tsim, RHsim, Vsim en Psim worden verkregen; gesimuleerde temperatuur Tsim, relatieve vochtigheid RHsm, windsnelheid Vsim en neerslag Psm dienen als toestandbijwerking van de observatie-inrichting, en in een toestandbijwerkingsmodel van de observatie-inrichting worden de gewichten van gesimuleerde waarden en waargenomen waarden in evenwicht gebracht door gebruik te maken van weegfactoren; een temperatuurtoestand wordt als volgt bijgewerkt: , Thieuw =aX Tsim + (1 - a) X Twaargenomen een relatieve vochtigheidstoestand wordt als volgt bijgewerkt: , RHpjeuw = B x RHsim + (1 = B) x RHywaargenomen een windsnelheidstoestand wordt als volgt bijgewerkt: , Vhieaw = VX Vsim + (1 — Y) x Vwaargenomen een neerslagtoestand wordt als volgt bijgewerkt: , waarin Pnieaw = ôx Psim + (1 =_ 6) X Pwaargenomen a, B, y en d zijn weegfactoren, Twaargenomen, RHwaargenomen, Vwaargenomen €N Pwaargenomen Zijn feitelijk waargenomen waarden voor temperatuur, relatieve vochtigheid, windsnelheid en neerslag; en toestand S=(Tnieuw, RHnieuw, Vnieuw, Pnieuw) van de observatie-inrichting wordt bijgewerkt.2. +Vx(p*xu)=0 where p is the density of air, t is time, u is a velocity vector of air, and V represents a divergence operator for representing the divergence of a vector field; a continuous equation is discretized, a numerical format is obtained, and the solution is performed on a computer; within each time step, depending on the density distribution and a velocity field at the current time, the quality equation is solved by using the numerical format, and the density distribution is updated the next time; the required meteorological factors are extracted from a simulation result after each time step of the simulation; it is assumed that the state variables of the respective observation device are temperature T, relative humidity RH, wind speed V, and precipitation p, and the state representation of the observation device is S=(T, RH, V, p); within each time step, the WRF model simulates temperature T, relative humidity RH, wind speed V and precipitation p at the current time, and Tsim, RHsim, Vsim and Psim are obtained; simulated temperature Tsim, relative humidity RHsm, wind speed Vsim and precipitation Psm serve as state update of the observation device, and in a state update model of the observation device, the weights of simulated values and observed values are balanced by using weighting factors; a temperature state is updated as follows: , Thieuw =aX Tsim + (1 - a) X Tobserved a relative humidity state is updated as follows: , RHpjeuw = B x RHsim + (1 = B) x RHyobserved a wind speed state is updated as follows: , Vhieaw = VX Vsim + (1 — Y) x Vobserved a precipitation state is updated as follows: , where Pnieaw = ôx Psim + (1 =_ 6) X Pobserved a, B, y and d are weighting factors, Tobserved, RHobserved, Vobserved €N Pobserved are actually observed values for temperature, relative humidity, wind speed and precipitation; and state S=(Tnew, RHnew, Vnew, Pnew) of the observation facility is updated.
NL2038172A 2023-09-22 2024-07-09 Dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors NL2038172B1 (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311237482.8A CN117310840A (en) 2023-09-22 2023-09-22 Dynamic optimization type automatic monitoring equipment for coupling weather and atmospheric environmental factors

Publications (2)

Publication Number Publication Date
NL2038172A true NL2038172A (en) 2024-08-27
NL2038172B1 NL2038172B1 (en) 2025-05-19

Family

ID=89296500

Family Applications (1)

Application Number Title Priority Date Filing Date
NL2038172A NL2038172B1 (en) 2023-09-22 2024-07-09 Dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors

Country Status (2)

Country Link
CN (1) CN117310840A (en)
NL (1) NL2038172B1 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118311207B (en) * 2024-04-16 2025-06-20 北京城建勘测设计研究院有限责任公司 Environmental monitoring method, system and storage medium

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103227833B (en) * 2013-04-28 2016-09-28 北京农业信息技术研究中心 Soil humidity sensor network system and information getting method thereof
CN115203189A (en) * 2022-07-15 2022-10-18 中国环境监测总站 Method for improving atmospheric transmission quantification capability by fusing multi-source data and visualization system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103227833B (en) * 2013-04-28 2016-09-28 北京农业信息技术研究中心 Soil humidity sensor network system and information getting method thereof
CN115203189A (en) * 2022-07-15 2022-10-18 中国环境监测总站 Method for improving atmospheric transmission quantification capability by fusing multi-source data and visualization system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
MURILLO-ESCOBAR J. ET AL: "Forecasting concentrations of air pollutants using support vector regression improved with particle swarm optimization: Case study in Aburr� Valley, Colombia", URBAN CLIMATE, vol. 29, 18 May 2019 (2019-05-18), pages 100473, XP055795150, ISSN: 2212-0955, Retrieved from the Internet <URL:http://dx.doi.org/10.1016/j.uclim.2019.100473> DOI: 10.1016/j.uclim.2019.100473 *

Also Published As

Publication number Publication date
NL2038172B1 (en) 2025-05-19
CN117310840A (en) 2023-12-29

Similar Documents

Publication Publication Date Title
Hou et al. From model-based control to data-driven control: Survey, classification and perspective
CN113325721B (en) Model-free adaptive control method and system for industrial system
CN105298734B (en) A Parameter Identification Method for Water Turbine Regulating System
Darus et al. Parametric modelling of a twin rotor system using genetic algorithms
NL2038172B1 (en) Dynamic optimization type automatic monitoring device coupling meteorological and atmospheric environmental factors
CN109885077A (en) Attitude control method and controller of a quadrotor aircraft
CN118523200B (en) Suspension cable type crossing frame sag self-adaption method and system
Fuchs et al. Towards more specific estimation of membership functions for data-driven fuzzy inference systems
Xu et al. DBN based SD-ARX model for nonlinear time series prediction and analysis
Aslam et al. Dynamic Optimization of Recurrent Networks for Wind Prediction on Edge Devices
Arshad et al. Deep Deterministic Policy Gradient to Regulate Feedback Control Systems Using Reinforcement Learning.
Derner et al. Data-driven construction of symbolic process models for reinforcement learning
Fidêncio et al. Application of Reinforcement Learning to a Mining System
CN114330819B (en) Wind turbine generator free inflow wind speed prediction method
Banakar et al. Parameter identification of TSK neuro-fuzzy models
CN113705932B (en) Short-term load prediction method and device
CN119105342A (en) H∞ control method for multi-drive servo system based on event triggering
Degachi et al. Nonlinear model predictive control for pH neutralization process based on SOMA algorithm
Grubinger et al. Online transfer learning for climate control in residential buildings
Dang et al. Stochastic configuration networks for adaptive inverse dynamics modeling
Kumar et al. Nonlinear dynamic engineering processes modeling using a lyapunov-stability based novel locally connected recurrent pi-sigma neural network: design, simulation, and a comparative study
Khan et al. A highly optimized multi-stage teacher-learner inspired particle swarm optimizer system
Wen et al. Implicit generalized predictive control-based fractional-order PID strategy for BTTGU regulation system
Al-Hiary et al. Identification of a chemical process reactor using soft computing techniques
Mohammadzadeh et al. Training interval type-2 fuzzy systems based on error backpropagation