CN119442857B - Rocket trajectory optimization and real-time adjustment system based on artificial intelligence - Google Patents
Rocket trajectory optimization and real-time adjustment system based on artificial intelligenceInfo
- Publication number
- CN119442857B CN119442857B CN202411457458.XA CN202411457458A CN119442857B CN 119442857 B CN119442857 B CN 119442857B CN 202411457458 A CN202411457458 A CN 202411457458A CN 119442857 B CN119442857 B CN 119442857B
- Authority
- CN
- China
- Prior art keywords
- track
- time
- rocket
- real
- state
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Automation & Control Theory (AREA)
- Aviation & Aerospace Engineering (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Mathematics (AREA)
- Evolutionary Biology (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of artificial intelligence, in particular to a rocket track optimization and real-time adjustment system based on artificial intelligence, which comprises a track prediction module, a real-time adjustment control module and an intelligent monitoring module, wherein a jump mechanism is introduced into the track prediction module, so that the interference of an intermediate state is effectively reduced, a key time step is directly focused, and a forward diffusion and inverse process denoising mechanism is introduced, so that the sensitivity of a model to noise is effectively reduced; based on model predictive control in a real-time adjustment control module, rocket attitude and thrust are automatically adjusted, state deviation is minimized by optimizing an objective function, and a constrained quadratic programming problem is adopted, so that the solving process is more efficient, the computational complexity is reduced, the decision speed is increased, the trajectory optimization and real-time adjustment capability of the rocket are remarkably improved, and powerful support is provided for realizing complex space missions.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a rocket track optimization and real-time adjustment system based on artificial intelligence.
Background
Along with the development of artificial intelligence, the application in the rocket field is gradually deepening, but the general rocket track optimization system has the problems of low prediction precision and poor track generation quality, lacks a denoising mechanism, is more sensitive to noise and abnormal values in input data, so that inaccurate prediction results can be possibly caused, meanwhile, an effective initialization and denoising means are lacked, the generated multi-mode future track is possibly low in quality and cannot accurately reflect real conditions, subsequent planning and execution are affected, the problem of large state deviation and high calculation complexity exists in an adjusting part in the general rocket track optimization system, the actual state of a rocket is possibly deviated from an expected state, the track is inaccurate, the solving process is complex, the calculation time and resource consumption are increased, and the decision making process is delayed.
Disclosure of Invention
Aiming at the problems of low prediction precision and poor track generation quality, the invention introduces a jump mechanism in a track prediction module, effectively reduces the interference of an intermediate state, directly focuses on key time steps, reduces noise influence, improves the track prediction precision, a jump initializer generates a more accurate multi-mode future track by taking a denoised deterministic track as a basis, improves the track planning quality, introduces forward diffusion and inverse process denoising mechanisms, effectively reduces the sensitivity of a model to noise, improves the stability of a prediction result, aims at the problems of large state deviation and high calculation complexity of an adjustment part, automatically adjusts the rocket attitude and thrust based on model prediction control in a real-time adjustment control module, minimizes the state deviation by optimizing an objective function, effectively controls the use of the thrust, adopts a constrained quadratic programming problem, ensures that the solving process is more efficient, reduces the calculation complexity, quickens the speed, obviously improves the track optimization and real-time adjustment decision making capability of the rocket, and provides a strong supporting task for realizing complex spaceflight.
The invention provides an artificial intelligence-based rocket track optimization and real-time adjustment system, which comprises a track prediction module, a real-time adjustment control module and an intelligent monitoring module, and specifically comprises the following contents:
the track prediction module predicts the optimal flight track of the rocket in real time by constructing a track prediction model;
The real-time adjustment control module automatically adjusts the attitude and thrust of the rocket based on model predictive control, so as to ensure real-time optimization of the flight trajectory;
the intelligent monitoring module provides real-time visual display of the flight process and helps ground control personnel monitor the flight track and state at any time.
Further, a track prediction model is constructed in the track prediction planning module, and the method specifically comprises the following steps:
step S1, collecting historical data, namely collecting historical track data and historical environment data of a rocket, wherein the historical track data comprise positions, speeds, accelerations and attitude angles, and the historical environment data comprise wind speeds, wind directions, air pressures and temperatures;
Step S2, preprocessing data and extracting features, namely preprocessing historical track data and historical environment data of a rocket, including outlier processing, filling missing values and denoising data, extracting key moment features in the rocket track by using a dynamic attention mechanism and generating high-dimensional space-time feature vectors;
step S3, introducing a mode initializer, and generating a starting state of a future track through a high-dimensional space-time feature vector;
s4, a multi-mode track prediction unit generates a multi-mode future track by constructing a track prediction model according to the initial state of the future track;
step S5, introducing an adaptive reasoning accelerator, and constructing the adaptive reasoning accelerator by using a cyclic neural network as a strategy network to optimize the diffusion model in the step S4;
S6, collecting real-time data and collecting real-time flight data of the rocket;
Step S7, dynamically adjusting, namely introducing a reinforcement learning Adam optimizer to analyze real-time flight data of the rocket, and adjusting a track prediction model through a real-time feedback mechanism;
further, step S3 specifically includes the following:
The construction field self-adaptive module is based on the proprietary knowledge of rocket field, including weather change, flight mode and thrust adjustment, and is used for inputting the preprocessed historical environment data into the field self-adaptive module, and adjusting the weight of the initial state of generating future tracks by learning the performances of the rocket under different environments;
Constructing a mode initializer, namely constructing the mode initializer by using a contrast learning method and combining a field self-adaptive module to generate the initial state of a future track;
further, step S4 specifically includes the following steps:
s41, inputting initial conditions, collecting initial positions, speeds, angles, masses, thrust and environmental factors of a rocket, and generating a track initial state;
Step S42, forward diffusion, using Representing the initial state of the track, introducing a diffusion model, adding noise on the initial state of the track by using Gaussian noise, and adding noise according to the following formula:
;
Wherein, the Is the coefficient of attenuation which is the coefficient of attenuation,For the time step size of the time step,Is a standard normally distributed noise;
Step S43 of generating a track profile by stepwise adding time steps Obtaining future track distribution of multiple modes;
Step S44, performing noise removal on future track distribution of a plurality of modes in an inverse process to obtain a deterministic future track;
Step S44 specifically includes the following:
future trajectory distribution for multiple modalities, from The inverse denoising is started, and the trajectory is updated by using an inverse diffusion equation, wherein the inverse diffusion equation is as follows:
;
Wherein, the In order to adjust the parameters of the denoising strength,Is a denoising function;
step-wise reduction of time step Up to the value of (2)Obtaining future track distribution of the most modes after denoising;
step S45, introducing a jump mechanism, skipping the middle time, setting a jump step length, and skipping in the forward diffusion and reverse processes The skipped formula for each time step is as follows:
;
Wherein, the Is the time derivative of the trace state;
step S46, jump initializer utilizing the deterministic future track generated in step S44 As a basis, initializing a multi-modal future track modality;
And S47, denoising the initialized multi-mode future track mode by a denoising unit, and strengthening the denoising effect through residual connection to obtain a non-noisy multi-mode future track.
Further, step S44 specifically includes the following:
future trajectory distribution for multiple modalities, from The inverse denoising is started, and the trajectory is updated by using an inverse diffusion equation, wherein the inverse diffusion equation is as follows:
;
Wherein, the In order to adjust the parameters of the denoising strength,Is a denoising function;
step-wise reduction of time step Up to the value of (2)And obtaining future track distribution of the most modes after denoising.
Further, in the real-time adjustment control module, the rocket attitude and the thrust are automatically adjusted based on model prediction control, and the method specifically comprises the following steps:
Q1, building a model, namely building a state space model by using the physical characteristics of the rocket and Newton's law of motion, wherein state variables comprise the position, the speed, the acceleration and the attitude angle of the rocket, input variables are thrust and attitude adjustment amounts, and a state equation is as follows:
;
the output equation is as follows:
;
Wherein, the Is a state vector, is a vector containing all state variables,Is the control input vector, is the vector containing all the input variables,Is an output of the device and is,AndIs process noise and measurement noise;
Step Q2, setting an objective function and constraint setting, wherein the objective function is set to optimally minimize the state deviation and the use of control input, and define the state and the input constraint, and the objective function is as follows:
;
;
;
;
;
Wherein, the For a time ofThe state vector of the time-dependent state vector,In order to achieve the desired state vector,Is shown inThe control input vector at the time of the time,、For the minimum and maximum allowed values of each state variable in the state vector,、Minimum and maximum allowable values for each input variable in the control input vector;
step Q3, optimizing problem construction, and determining a prediction time range Combining the objective function and the constraint, solving the optimal control input to form a quadratic programming problem with the constraint, wherein the method comprises the following steps:
;
;
;
step Q4, solving an optimization problem, and obtaining optimal thrust and attitude adjustment quantity by adopting an initialization strategy;
And Q5, applying control input, and adjusting the thrust and the posture of the rocket according to the optimal thrust and posture adjustment quantity.
By adopting the scheme, the beneficial effects obtained by the invention are as follows:
(1) Aiming at the problems of low prediction precision and poor track generation quality, a jump mechanism is introduced into a track prediction module, so that the interference of an intermediate state is effectively reduced, key time steps are directly concerned, the noise influence is reduced, the precision of track prediction is improved, a jump initializer uses a denoised deterministic track as a basis to generate a more accurate multi-mode future track, the quality of track planning is improved, and forward diffusion and inverse process denoising mechanisms are introduced, so that the sensitivity of a model to noise is effectively reduced, and the stability of a prediction result is improved;
(2) Aiming at the problems of large state deviation and high calculation complexity of an adjusting part, the invention automatically adjusts the attitude and the thrust of the rocket based on model prediction control in a real-time adjusting control module, minimizes the state deviation by optimizing an objective function, can effectively control the use of the thrust, adopts the constrained quadratic programming problem, ensures that the solving process is more efficient, reduces the calculation complexity, accelerates the decision-making speed, obviously improves the track optimization and the real-time adjusting capability of the rocket, and provides powerful support for realizing complex space missions.
FIG. 1 is a schematic diagram of an artificial intelligence based rocket trajectory optimization and real-time adjustment system provided by the invention.
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments, and all other embodiments obtained by those skilled in the art without making any inventive effort based on the embodiments of the present invention are within the scope of protection of the present invention.
In the description of the present invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate orientation or positional relationships based on those shown in the drawings, merely to facilitate description of the invention and simplify the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention.
The invention provides an artificial intelligence-based rocket trajectory optimization and real-time adjustment system, which comprises a trajectory prediction module, a real-time adjustment control module and an intelligent monitoring module, and specifically comprises the following contents:
the track prediction module predicts the optimal flight track of the rocket in real time by constructing a track prediction model;
The real-time adjustment control module automatically adjusts the attitude and thrust of the rocket based on model predictive control, so as to ensure real-time optimization of the flight trajectory;
the intelligent monitoring module provides real-time visual display of the flight process and helps ground control personnel monitor the flight track and state at any time.
In a second embodiment, the track prediction model is built in the track prediction planning module based on the above embodiment, and specifically includes the following steps:
step S1, collecting historical data, namely collecting historical track data and historical environment data of a rocket, wherein the historical track data comprise positions, speeds, accelerations and attitude angles, and the historical environment data comprise wind speeds, wind directions, air pressures and temperatures;
Step S2, preprocessing data and extracting features, namely preprocessing historical track data and historical environment data of a rocket, including outlier processing, filling missing values and denoising data, extracting key moment features in the rocket track by using a dynamic attention mechanism and generating high-dimensional space-time feature vectors;
step S3, introducing a mode initializer, and generating a starting state of a future track through a high-dimensional space-time feature vector;
s4, a multi-mode track prediction unit generates a multi-mode future track by constructing a track prediction model according to the initial state of the future track;
step S5, introducing an adaptive reasoning accelerator, and constructing the adaptive reasoning accelerator by using a cyclic neural network as a strategy network to optimize the diffusion model in the step S4;
S6, collecting real-time data and collecting real-time flight data of the rocket;
And S7, dynamically adjusting, namely introducing a reinforcement learning Adam optimizer to analyze real-time flight data of the rocket, and adjusting the track prediction model through a real-time feedback mechanism.
In this embodiment, in step S2, a statistical method is used to identify and remove obvious outliers in the dataset, an interpolation method is applied to fill in missing trajectory data, a low-pass filter is used to denoise flight data, smoothness of the data is ensured, a dynamic attention model is constructed by using a dynamic attention mechanism, key moment features are extracted in an important way, high-dimensional space-time feature vectors are generated, and important changes in the flight process are captured.
In the embodiment, a cyclic neural network is adopted as a strategy network, a track prediction model is optimized, and the model is enabled to rapidly adjust the prediction strategy under different environmental conditions by training a self-adaptive reasoning accelerator, so that the instantaneity is improved.
Embodiment three, this embodiment is based on the above embodiment, and step S3 specifically includes the following:
The construction field self-adaptive module is based on the proprietary knowledge of rocket field, including weather change, flight mode and thrust adjustment, and is used for inputting the preprocessed historical environment data into the field self-adaptive module, and adjusting the weight of the initial state of generating future tracks by learning the performances of the rocket under different environments;
and constructing a mode initializer, namely constructing the mode initializer by using a contrast learning method and combining a field self-adaptive module to generate the initial state of the future track.
Embodiment four, this embodiment is based on the above embodiment, and step S4 specifically includes the following steps:
s41, inputting initial conditions, collecting initial positions, speeds, angles, masses, thrust and environmental factors of a rocket, and generating a track initial state;
Step S42, forward diffusion, using Representing the initial state of the track, introducing a diffusion model, adding noise on the initial state of the track by using Gaussian noise, and adding noise according to the following formula:
;
Wherein, the Is the coefficient of attenuation which is the coefficient of attenuation,For the time step size of the time step,Is a standard normally distributed noise;
Step S43 of generating a track profile by stepwise adding time steps Obtaining future track distribution of multiple modes;
Step S44, performing noise removal on future track distribution of a plurality of modes in an inverse process to obtain a deterministic future track;
Step S44 specifically includes the following:
future trajectory distribution for multiple modalities, from The inverse denoising is started, and the trajectory is updated by using an inverse diffusion equation, wherein the inverse diffusion equation is as follows:
;
Wherein, the In order to adjust the parameters of the denoising strength,Is a denoising function;
step-wise reduction of time step Up to the value of (2)Obtaining future track distribution of the most modes after denoising;
step S45, introducing a jump mechanism, skipping the middle time, setting a jump step length, and skipping in the forward diffusion and reverse processes The skipped formula for each time step is as follows:
;
Wherein, the Is the time derivative of the trace state;
step S46, jump initializer utilizing the deterministic future track generated in step S44 As a basis, initializing a multi-modal future track modality;
And S47, denoising the initialized multi-mode future track mode by a denoising unit, and strengthening the denoising effect through residual connection to obtain a non-noisy multi-mode future track.
In the present embodiment, the initial conditions are set as follows:
An initial position;
An initial speed;
the attitude angles are 0,0 and 0;
the mass was 15000 kg;
thrust is 300000N;
Environmental factors, wind speed 5 m/s, wind direction = 30 degrees, air pressure 101325 pa, temperature 288 kelvin;
The initial state of the generated track is [1000,2000,3000,50,60,70,0,0,0,15000,300000,5,30,101325,288];
setting the attenuation coefficient to 0.9 in forward diffusion, generating gaussian noise of [0.5, -0.2,0.1,0,0.3, -0.4,0,0,0,0,0,0,0,0,0,0];
The calculation of the added noise is performed, Performing element-by-element calculation to obtain[948.85,1899.92,2849.99,47.44,59.24,67.56,0,0,0,15000,300000,5,30,101325,288] Is shown;
Assuming a time step For 1 second, generate 3 modes;
The generated track modes are as follows:
=[1005,2000,3001,50.1,60.5,69.8,0,0,0,15002,300001,5.1,30,101320,288];
=[1002,1998,3003,49.5,59.9,70.1,0,0,0,15001,300002,5.2,30,101325,288];
=[1001,2001,3000,50.2,60.1,69.5,0,0,0,15003,300000,5,30,101300,288];
Set to 0.1, denoising function The trace after denoising is as follows:
;
Suppose a jump step size For a period of 2 seconds,;
=[1004,2001,3000,50,60,70,0,0,0,15000,300000,5,30,101325,288];
The final generated modes are:
=[1004,2001,3000,50,60,70,0,0,0,15000,300000,5,30,101325,288];
=[1003,2000,3001,50,60,70,0,0,0,15001,300001,5,30,101320,288];
=[1005,2002,3002,50,60,70,0,0,0,15002,300002,5,30,101330,288]
the steps above demonstrate how to generate future trajectories of rockets through specific parameter definitions and calculations.
An embodiment five, which is based on the above embodiment, in the real-time adjustment control module, automatically adjusts the rocket attitude and the thrust based on model prediction control, and specifically includes the following steps:
Q1, building a model, namely building a state space model by using the physical characteristics of the rocket and Newton's law of motion, wherein state variables comprise the position, the speed, the acceleration and the attitude angle of the rocket, input variables are thrust and attitude adjustment amounts, and a state equation is as follows:
;
the output equation is as follows:
;
Wherein, the Is a state vector, is a vector containing all state variables,Is the control input vector, is the vector containing all the input variables,Is an output of the device and is,AndIs process noise and measurement noise;
Step Q2, setting an objective function and constraint setting, wherein the objective function is set to optimally minimize the state deviation and the use of control input, and define the state and the input constraint, and the objective function is as follows:
;
;
;
;
;
Wherein, the For a time ofThe state vector of the time-dependent state vector,In order to achieve the desired state vector,Is shown inThe control input vector at the time of the time,、For the minimum and maximum allowed values of each state variable in the state vector,、Minimum and maximum allowable values for each input variable in the control input vector;
step Q3, optimizing problem construction, and determining a prediction time range Combining the objective function and the constraint, solving the optimal control input to form a quadratic programming problem with the constraint, wherein the method comprises the following steps:
;
;
;
step Q4, solving an optimization problem, and obtaining optimal thrust and attitude adjustment quantity by adopting an initialization strategy;
And Q5, applying control input, and adjusting the thrust and the posture of the rocket according to the optimal thrust and posture adjustment quantity.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
The invention and its embodiments have been described above with no limitation, and the actual construction is not limited to the embodiments of the invention as shown in the drawings. In summary, if one of ordinary skill in the art is informed by this disclosure, a structural manner and an embodiment similar to the technical solution should not be creatively devised without departing from the gist of the present invention.
Claims (4)
1. The rocket track optimizing and real-time adjusting system based on artificial intelligence is characterized by comprising a track predicting module, a real-time adjusting control module and an intelligent monitoring module, and specifically comprises the following contents:
the track prediction module predicts the optimal flight track of the rocket in real time by constructing a track prediction model;
The real-time adjustment control module automatically adjusts the attitude and thrust of the rocket based on model predictive control, so as to ensure real-time optimization of the flight trajectory;
the intelligent monitoring module provides real-time visual display of the flight process and helps ground control personnel monitor the flight track and state at any time;
The track prediction model is constructed in the track prediction planning module, and specifically comprises the following steps:
step S1, collecting historical data, namely collecting historical track data and historical environment data of a rocket, wherein the historical track data comprise positions, speeds, accelerations and attitude angles, and the historical environment data comprise wind speeds, wind directions, air pressures and temperatures;
Step S2, preprocessing data and extracting features, namely preprocessing historical track data and historical environment data of a rocket, including outlier processing, filling missing values and denoising data, extracting key moment features in the rocket track by using a dynamic attention mechanism and generating high-dimensional space-time feature vectors;
step S3, introducing a mode initializer, and generating a starting state of a future track through a high-dimensional space-time feature vector;
s4, a multi-mode track prediction unit generates a multi-mode future track by constructing a track prediction model according to the initial state of the future track;
step S5, introducing an adaptive reasoning accelerator, and constructing the adaptive reasoning accelerator by using a cyclic neural network as a strategy network to optimize the diffusion model in the step S4;
S6, collecting real-time data and collecting real-time flight data of the rocket;
And S7, dynamically adjusting, namely introducing a reinforcement learning Adam optimizer to analyze real-time flight data of the rocket, and adjusting the track prediction model through a real-time feedback mechanism.
2. The artificial intelligence-based rocket trajectory optimization and real-time adjustment system according to claim 1, wherein the step S3 comprises the following specific steps:
The construction field self-adaptive module is based on the proprietary knowledge of rocket field, including weather change, flight mode and thrust adjustment, and is used for inputting the preprocessed historical environment data into the field self-adaptive module, and adjusting the weight of the initial state of generating future tracks by learning the performances of the rocket under different environments;
and constructing a mode initializer, namely constructing the mode initializer by using a contrast learning method and combining a field self-adaptive module to generate the initial state of the future track.
3. The artificial intelligence-based rocket trajectory optimization and real-time adjustment system according to claim 1, wherein the step S4 comprises the following steps:
s41, inputting initial conditions, collecting initial positions, speeds, angles, masses, thrust and environmental factors of a rocket, and generating a track initial state;
Step S42, forward diffusion, using Representing the initial state of the track, introducing a diffusion model, adding noise on the initial state of the track by using Gaussian noise, and adding noise according to the following formula:
;
Wherein, the Is the coefficient of attenuation which is the coefficient of attenuation,For the time step size of the time step,Is a standard normally distributed noise;
Step S43 of generating a track profile by stepwise adding time steps Obtaining future track distribution of multiple modes;
Step S44, performing noise removal on future track distribution of a plurality of modes in an inverse process to obtain a deterministic future track;
Step S44 specifically includes the following:
future trajectory distribution for multiple modalities, from The inverse denoising is started, and the trajectory is updated by using an inverse diffusion equation, wherein the inverse diffusion equation is as follows:
;
Wherein, the In order to adjust the parameters of the denoising strength,Is a denoising function;
step-wise reduction of time step Up to the value of (2)Obtaining future track distribution of the most modes after denoising;
step S45, introducing a jump mechanism, skipping the middle time, setting a jump step length, and skipping in the forward diffusion and reverse processes The skipped formula for each time step is as follows:
;
Wherein, the Is the time derivative of the trace state;
step S46, jump initializer utilizing the deterministic future track generated in step S44 As a basis, initializing a multi-modal future track modality;
And S47, denoising the initialized multi-mode future track mode by a denoising unit, and strengthening the denoising effect through residual connection to obtain a non-noisy multi-mode future track.
4. The artificial intelligence-based rocket trajectory optimization and real-time adjustment system according to claim 1, wherein in the real-time adjustment control module, the rocket attitude and thrust are automatically adjusted based on model prediction control, and the method specifically comprises the following steps:
Q1, building a model, namely building a state space model by using the physical characteristics of the rocket and Newton's law of motion, wherein state variables comprise the position, the speed, the acceleration and the attitude angle of the rocket, input variables are thrust and attitude adjustment amounts, and a state equation is as follows:
;
the output equation is as follows:
;
Wherein, the Is a state vector, is a vector containing all state variables,Is the control input vector, is the vector containing all the input variables,Is an output of the device and is,AndIs process noise and measurement noise;
Step Q2, setting an objective function and constraint setting, wherein the objective function is set to optimally minimize the state deviation and the use of control input, and define the state and the input constraint, and the objective function is as follows:
;
;
;
;
;
Wherein, the For a time ofThe state vector of the time-dependent state vector,In order to achieve the desired state vector,Is shown inThe control input vector at the time of the time,、For the minimum and maximum allowed values of each state variable in the state vector,、Minimum and maximum allowable values for each input variable in the control input vector;
step Q3, optimizing problem construction, and determining a prediction time range Combining the objective function and the constraint, solving the optimal control input to form a quadratic programming problem with the constraint, wherein the method comprises the following steps:
;
;
;
step Q4, solving an optimization problem, and obtaining optimal thrust and attitude adjustment quantity by adopting an initialization strategy;
And Q5, applying control input, and adjusting the thrust and the posture of the rocket according to the optimal thrust and posture adjustment quantity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411457458.XA CN119442857B (en) | 2024-10-18 | 2024-10-18 | Rocket trajectory optimization and real-time adjustment system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411457458.XA CN119442857B (en) | 2024-10-18 | 2024-10-18 | Rocket trajectory optimization and real-time adjustment system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119442857A CN119442857A (en) | 2025-02-14 |
| CN119442857B true CN119442857B (en) | 2025-08-12 |
Family
ID=94522992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411457458.XA Active CN119442857B (en) | 2024-10-18 | 2024-10-18 | Rocket trajectory optimization and real-time adjustment system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119442857B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120447384B (en) * | 2025-05-08 | 2025-11-25 | 烟台海星天箭航天科技合伙企业(有限合伙) | A self-supervised learning-based adaptive sea state lift control method for rocket platforms |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112550770A (en) * | 2020-12-15 | 2021-03-26 | 北京航天自动控制研究所 | Rocket soft landing trajectory planning method based on convex optimization |
| CN115289917A (en) * | 2022-08-12 | 2022-11-04 | 中山大学 | Real-time optimal guidance method and system for rocket sub-stage landing based on deep learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210178600A1 (en) * | 2019-12-12 | 2021-06-17 | Mitsubishi Electric Research Laboratories, Inc. | System and Method for Robust Optimization for Trajectory-Centric ModelBased Reinforcement Learning |
-
2024
- 2024-10-18 CN CN202411457458.XA patent/CN119442857B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112550770A (en) * | 2020-12-15 | 2021-03-26 | 北京航天自动控制研究所 | Rocket soft landing trajectory planning method based on convex optimization |
| CN115289917A (en) * | 2022-08-12 | 2022-11-04 | 中山大学 | Real-time optimal guidance method and system for rocket sub-stage landing based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119442857A (en) | 2025-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US12067491B2 (en) | Multi-agent reinforcement learning with matchmaking policies | |
| EP3791324B1 (en) | Sample-efficient reinforcement learning | |
| JP4803212B2 (en) | Data processing apparatus, data processing method, and program | |
| EP4014162B1 (en) | Controlling agents using causally correct environment models | |
| KR20210011422A (en) | Stacked convolutional long-term memory for modelless reinforcement learning | |
| KR20230119023A (en) | Attention neural networks with short-term memory | |
| CN119442857B (en) | Rocket trajectory optimization and real-time adjustment system based on artificial intelligence | |
| US20230083486A1 (en) | Learning environment representations for agent control using predictions of bootstrapped latents | |
| JP7646870B2 (en) | Reinforcement learning using an ensemble of discriminator models | |
| CN119644704B (en) | Biped robot complex terrain self-adaptive gait planning method and biped robot | |
| Hafez et al. | Efficient intrinsically motivated robotic grasping with learning-adaptive imagination in latent space | |
| CN119758719B (en) | Inverted Pendulum Stabilization Method for Quadruped Robot Based on Hybrid State Estimation and Reinforcement Learning | |
| CN119159582B (en) | Multi-axis mechanical arm prediction control method based on information physical neural network | |
| Detroja | Parameterized Adaptive Controller Design using Reinforcement Learning and Deep Neural Networks | |
| JP2009116770A (en) | Data processing apparatus and method, program, and recording medium | |
| Chen et al. | Brief communication: Uniform ultimate boundedness of a fuzzy logic controlled industrial robot | |
| CN120533718B (en) | Exoskeleton robot self-adaptive control method and related equipment | |
| CN120491449A (en) | Robust fault-tolerant control method based on deep reinforcement learning | |
| Tiong et al. | Process Proportional-Integral PI Control with Deep Reinforcement Learning | |
| Zhu et al. | Learning of Quadruped Robot Motor Skills Based on Policy Constrained TD3 | |
| Niu et al. | Enhancing Control Performance through ESN-Based Model Compensation in MPC for Dynamic Systems | |
| Harib | Deep Reinforcement Learning for Robust Control of 6-DOF Robotic Manipulators | |
| CN118195077A (en) | Track prediction and model training method, device, storage medium and equipment thereof | |
| Li et al. | Single degree of freedom control based on deep reinforcement learning for underwater unmanned vehicle | |
| Vaiuso et al. | Methods for Multi-objective Optimization PID Controller for quadrotor UAVs |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |