[go: up one dir, main page]

WO2019006995A1 - Système de prédiction intelligent du soc d'une batterie d'alimentation de véhicule électrique - Google Patents

Système de prédiction intelligent du soc d'une batterie d'alimentation de véhicule électrique Download PDF

Info

Publication number
WO2019006995A1
WO2019006995A1 PCT/CN2017/116819 CN2017116819W WO2019006995A1 WO 2019006995 A1 WO2019006995 A1 WO 2019006995A1 CN 2017116819 W CN2017116819 W CN 2017116819W WO 2019006995 A1 WO2019006995 A1 WO 2019006995A1
Authority
WO
WIPO (PCT)
Prior art keywords
battery
prediction
model
neural network
soc
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.)
Ceased
Application number
PCT/CN2017/116819
Other languages
English (en)
Chinese (zh)
Inventor
马从国
王业琴
王建国
陈亚娟
杨玉东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huaiyin Institute of Technology
Original Assignee
Huaiyin Institute of Technology
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 Huaiyin Institute of Technology filed Critical Huaiyin Institute of Technology
Publication of WO2019006995A1 publication Critical patent/WO2019006995A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
    • B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
    • B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/3644—Constructional arrangements
    • G01R31/3648—Constructional arrangements comprising digital calculation means, e.g. for performing an algorithm
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/385—Arrangements for measuring battery or accumulator variables
    • G01R31/387—Determining ampere-hour charge capacity or SoC
    • G01R31/388—Determining ampere-hour charge capacity or SoC involving voltage measurements
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/70—Energy storage systems for electromobility, e.g. batteries

Definitions

  • the invention relates to the technical field of battery testing equipment, in particular to an intelligent prediction system for an electric vehicle power battery SOC.
  • SOC state of charge
  • the commonly used estimation methods of SOC mainly include: ampere integration method, open circuit voltage method, Kalman filter method, neural network method and the like.
  • the An-time integration method obtains the power consumption of the battery pack by calculating the current-to-time integral, and then obtains the remaining power, but it is essentially an open-loop prediction. The existence of the pure integral link makes the error increase with time.
  • the open circuit voltage method obtains the remaining power by detecting the open circuit voltage of the battery, and requires the battery to stand for a long time without external power supply, which is not suitable for online real-time measurement.
  • the Kalman filter method needs to establish the internal model of the battery to obtain the state equation.
  • the accuracy of the battery model is high, and it has certain limitations in practical applications.
  • the neural network method can obtain better precision by using a large number of sample data for training learning according to the established network model, but the network is more sensitive to the initial weight selection, generally converges to the local minimum near the initial value, and the initial value changes. Affect the convergence speed and accuracy of the network. Domestic time, etc.
  • the invention patent design is an intelligent predictive device for electric vehicle power battery SOC, which realizes the collection of parameters such as voltage, current and temperature of electric vehicle battery parameters and accurate prediction of electric vehicle battery SOC.
  • the invention provides a SOC intelligent prediction system for an electric vehicle power battery, and the invention effectively solves the problem that the battery SOC is a nonlinear, delayed, multivariable coupling and complex real-time system, real-time Sexual requirements are very high, and conventional control methods are difficult to achieve the desired results.
  • An intelligent vehicle SOC intelligent prediction system is characterized in that: the intelligent prediction system comprises a battery parameter acquisition platform and a battery SOC prediction system, and the battery parameter collection platform is configured to collect the voltage, current, temperature and The real-time parameter of the ambient temperature, the battery SOC prediction system predicts the battery SOC value through the collected real-time parameters;
  • the battery parameter collection platform is composed of a current sensor, a voltage detection circuit, a battery pack temperature sensor, an ambient temperature sensor, a load and a measurement and control unit, wherein the measurement and control unit comprises a single battery data acquisition module, a CPU processor, a touch screen, an RS232 interface, and a CAN.
  • the interface, the A/D conversion unit and the equalizer, the battery parameter acquisition platform collects the battery pack voltage and current, the battery temperature and the ambient temperature, and performs information interaction with the electric vehicle control system through the CAN bus interface;
  • the battery SOC prediction system includes a GM (1, 1) voltage prediction model, a GM (1, 1) current prediction model, a GM (1, 1) temperature prediction model, a SOM neural network classifier, and multiple RBF fuzzy neural network estimations.
  • the SOM neural network classifier outputs a value of a GM (1, 1) voltage prediction model output value of an electric vehicle battery, a GM (1, 1) current prediction model output value, a predicted voltage of a GM (1, 1) temperature prediction model output value, Predicting current and predicted temperature characteristic parameters for reasonable sample subset partitioning, different subset characteristics
  • the parameter input corresponds to the RBF fuzzy neural network estimation model to achieve accurate prediction of the SOC value of the electric vehicle battery.
  • the output value of the ANFIS estimation compensation model is based on an electric vehicle battery GM (1, 1) ambient temperature change prediction model output value, a GM (1, 1) battery internal resistance change prediction model output value, and an RBF fuzzy neural network estimation model output.
  • the value of the value compensates the output values of the multiple RBF fuzzy neural network estimation models, and improves the accuracy of the electric vehicle SOC intelligent prediction device for predicting the SOC value of the electric vehicle battery.
  • the RBF fuzzy neural network estimation model output subtracts the k difference of the output value of the ANFIS compensation estimation model as the input of the ARIMA dynamic prediction model, and the ARIMA dynamic prediction model outputs the predicted value of the battery SOC to improve the prediction of the SOC value of the battery. Accuracy.
  • the present invention has the following significant advantages:
  • the SOM neural network classifier used in the present invention is a data classification method.
  • the purpose is to divide a set of data in a prediction data space such as voltage, current and temperature of an electric vehicle battery into a plurality of subsets according to a similarity criterion, so that each subset of the vehicle battery characteristic parameters represents one of the entire predicted data sample sets.
  • Characteristics, the establishment of SOM neural network classifier to classify the parameters of electric vehicle battery characteristic prediction is to find a reasonable sample subset division, the characteristics of different subsets of root root prediction parameters input RBF fuzzy neural network estimation model to predict battery SOC value, improve detection SOC The accuracy of the prediction of the value.
  • the SOM neural network classifier is used to divide the subset of electric vehicle battery feature prediction parameters before using the RBF fuzzy neural network estimation model, and each subset adopts a corresponding RBF fuzzy neural network.
  • this method can use the corresponding estimation sub-model according to the characteristics of each sub-parameter subset to improve the prediction accuracy and operation speed of the RBF fuzzy neural network estimation model.
  • the prediction method has better fitting precision and generalization ability. .
  • the present invention utilizes the ANFIS compensation estimation model to accurately predict the gray temperature prediction of environmental temperature changes, the gray prediction amount of the internal resistance of the battery, and the output of the RBF fuzzy neural network estimation model.
  • the input and output characteristics of SOC value have good nonlinear approximation ability.
  • ANFIS not only has the inference function of fuzzy inference system, but also has the training and learning function of neural network. Combining the advantages of the two, it overcomes the characteristics of the simple neural network black box and has a certain transparency. Through a large number of experiments, it is verified that the ANFIS compensation estimation model is faster than the general BP neural network training, and the training times are greatly reduced, overcoming the local optimal problem. Therefore, the AN FIS compensation estimation model is used to establish an accurate input and output relationship that affects the SOC value of the battery.
  • the ANFIS compensation estimation model adopted by the present invention is a fuzzy inference system based on the Takagi-Sugeno model, which is a novel fuzzy inference system structure that combines fuzzy logic and a neural network, using a back propagation algorithm and a least squares method.
  • the hybrid algorithm adjusts the premise parameters and conclusion parameters and automatically generates the If-Then rules.
  • the ANFIS compensation estimation model also has the function of approximating the arbitrary linear and nonlinear functions of the battery SOC with arbitrary precision, and the convergence speed is fast and the sample requirement is small.
  • the ANFIS compensation estimation model has a fast calculation speed, reliable results, and good results.
  • the present invention combines the artificial neural network and the fuzzy theory organically by the ANFIS compensation estimation model, constructs the fuzzy system by using the neural network, and automatically designs and adjusts according to the input and output samples affecting the SOC value of the battery by using the neural network learning method.
  • the parameters of the fuzzy system realize the self-learning and self-adaptive functions of the fuzzy system, and can fit the linear and nonlinear mapping relationship between the input and output that affect the SOC value of the battery. It is especially suitable for complex nonlinear battery SOC systems and compensation batteries. The predicted value is higher.
  • the RBF fuzzy neural network estimation model adopted by the present invention uses a radial basis (RBF) neural network to have a fast learning speed, has a good generalization ability, and can approximate a nonlinear function with arbitrary precision. And with global approximation ability, it fundamentally solves the local optimal problem of BP network, and the topology is compact, the structural parameters can be separated and learned, and the convergence speed is fast. Fuzzy logic systems have strong inference adaptive performance for arbitrary complexity systems.
  • the RBF fuzzy neural network combines the advantages of both to achieve functional and structural complementarity.
  • the RBF fuzzy neural network estimation model has high adaptability and high learning accuracy for predicting battery SOC value.
  • the prediction accuracy of the invention is high, and the battery characteristic parameters voltage, current and temperature are three GM modes.
  • the RBF fuzzy neural network estimation model and the SOM neural network classifier are combined to establish a battery SOC value estimation prediction model.
  • the historical data of temperature, current and voltage characteristic parameters affecting the SOC value of the battery are differently selected as the initial data input.
  • the GM model of the parameters, the outputs of the three GM models are classified as inputs of the SOM neural network classifier, and each type of predicted value is input into the RBF fuzzy neural network estimation model.
  • the battery SOC value estimation method combines the advantages of the gray prediction GM model with less original data and simple method, and the strong nonlinear fitting ability of the RBF fuzzy neural network.
  • the original data is accumulated by the gray prediction theory, and the trend is highlighted.
  • the influence of the nonlinear excitation function of the RBF fuzzy neural network estimation model is easier to approximate, and the influence of uncertain components on the grey theory prediction value is reduced.
  • the accuracy of the gray GM prediction model is low and the training data required by the RBF fuzzy neural network are more.
  • the shortcomings effectively avoid the lack of information loss in a single model, thus improving the accuracy of the prediction results.
  • the SOM neural network classifier is used to classify each type of battery characteristic prediction parameters, and each type of parameter is input into a class of RBF fuzzy neural network estimation. Model, the residual is small, and the generalization ability of the network is better.
  • the learning time and convergence speed of the RBF fuzzy neural network estimation model are faster, more stable, and the prediction accuracy is higher.
  • the GM prediction model of the battery ambient temperature change and the GM prediction model of the battery internal resistance change and the output value of the RBF fuzzy neural network estimation model are used as inputs to the ANFIS compensation estimation model, and the ANFIS compensation estimation model is compensated for the RBF fuzzy neural network estimation model output value.
  • the accuracy which greatly improves the accuracy and accuracy of battery SOC prediction.
  • the robustness of the present invention is strong, and the SOC prediction model of the electric vehicle battery with the gray fuzzy neural optimization combination is established, which embodies the gray system behavior of the SOC value of the battery, and can dynamically predict, with high precision and stability, and
  • the combination of grey theory, neural network and fuzzy logic can make good use of the advantages of each single-item algorithm, and give full play to the advantages of grey prediction, neural network and fuzzy logic, and improve prediction accuracy, stability and rapidity in essence;
  • gray system The new data is obtained by accumulating or subtracting the sample data, which weakens the randomness of the original sample to a certain extent, and has less demand for sample capacity; the patent portfolio prediction can autonomy the inherent law in the sample data.
  • the present invention predicts the time span of the SOC value of the battery.
  • the GM model can predict the temperature and voltage of the battery at a future time according to the temperature, voltage, current, ambient temperature change and the amount of change in the internal resistance of the battery that affect the SOC value of the battery at the previous time.
  • current, ambient temperature change and battery internal resistance change input battery SOC prediction system can predict the battery SOC value in the future, after using the above method to predict the battery SOC value, the battery temperature, voltage, current, ambient temperature changes
  • the amount and the internal resistance change parameter value of the battery are added to the original series, and a data model at the beginning of the series is removed correspondingly to predict the battery SOC value.
  • This method is called an equal-dimensional gray number replenishment model, which can achieve long-term prediction.
  • the user can grasp the changing trend of the SOC value of the battery more accurately, and fully prepare for the safe and reliable operation or maintenance of the electric vehicle.
  • the present invention uses the ARIMA dynamic prediction model to predict the SOC value of the battery, integrates the original time series variables of the trend factors, periodic factors and random errors of the SOC value of the battery, and converts the non-stationary sequence to zero mean by differential data conversion and the like.
  • the stationary random sequence is compared and compared with the model diagnosis and the ideal model is selected for battery SOC value data fitting and prediction.
  • This method combines the advantages of autoregressive and moving average methods, has the characteristics of being unconstrained by data types and strong applicability, and is a model for predicting the short-term prediction of battery SOC value, improving the prediction accuracy of battery SOC value. , time span and robustness.
  • 1 is a battery parameter collection platform of the present invention
  • FIG. 3 is a schematic diagram showing the function of the software of the measurement and control unit of the present invention.
  • FIG. 4 is a plan view of a battery management system of the present invention.
  • the battery SOC intelligent prediction system should have the following functions: 1) Parameter detection. The battery is charged and discharged in real time, and the data of the collected battery includes voltage, battery current, battery temperature, and battery voltage of the single module; 2) Residual power (SOC) prediction. The system should immediately collect parameters such as charge and discharge current and voltage, and estimate the SOC by the corresponding algorithm. The remaining energy of the battery is equivalent to the oil of the traditional car; 3) heat management. Collect the temperature of the battery in real time, prevent the battery temperature from being too high by controlling the heat sink; 4) Balance control.
  • the system should be able to judge and automatically perform equalization processing; 5) information monitoring.
  • the main information of the battery is displayed in real time through the RS232 interface on the touch screen display terminal; 6) CAN interface.
  • the battery measurement and control unit shares information with other systems of the vehicle through the CAN interface.
  • the present invention patents a battery parameter acquisition platform in an electric vehicle power battery SOC intelligent prediction device.
  • the battery parameter acquisition platform is composed of a current sensor, a voltage detection circuit, a battery temperature sensor, an ambient temperature sensor, a load, and a measurement and control unit.
  • the measurement and control unit includes a single battery data acquisition module, a CPU processor, a touch screen, an RSS32 interface, a CAN interface, The A/D conversion unit and the equalizer, the battery parameter collection platform collects the battery pack voltage, current, battery temperature and ambient temperature, and performs information interaction with the electric vehicle control system through the CAN bus interface; the electric vehicle power battery SOC intelligent prediction device Figure 1 shows.
  • the battery management system CPU processor is the core of the whole system.
  • the CPU processor selects the DSP56F807 chip integrated with the CAN controller module to realize the CAN interface, the CAN interface transceiver selects the PCA82C250 as the transceiver, and the battery equalizer adopts the distributed dynamic equalization control.
  • DC/DC chopper circuit isolated drive, PWM controller and matrix switch type channel selection circuit; using AV100-150 Hall voltage sensor and CHB-200SF Hall current sensor for total voltage and current detection of the battery pack.
  • the single-cell data acquisition module monitors the voltage and temperature data of each single cell in real time, and the equalizer sends a strobe signal to the channel selection circuit to realize dynamic equalization charging and discharging of the single cells in each battery module; Realize communication with the touch screen and calibration of the system.
  • the battery measurement and control module microcontroller uses a 2-channel 12-bit precision A/D conversion unit.
  • the battery temperature sensor and the ambient temperature sensor use the digital temperature sensor DS18B20 to collect the battery test point temperature and the battery pack operating environment temperature.
  • the measurement and control unit software adopts modular programming, and the CPU processor program is written in C language. According to the functions of the system, it is divided into several sub-programs, including: program parameters and control parameter initialization module, parameter and control module and display module to achieve battery voltage, current, temperature and ambient temperature acquisition, battery equalization control, SOC estimation. , curve display and data display and other functions.
  • the software function is shown in Figure 3.
  • the battery SOC prediction system is designed to predict the battery SOC value in the CPU processor of the measurement and control unit.
  • the battery SOC prediction system includes the gray prediction GM (1, 1) model, the SOM neural network classifier, multiple RBF fuzzy neural network estimation models, and ANFIS estimation.
  • the compensation model and the ARIMA dynamic prediction model are composed.
  • the battery SOC prediction system is shown in Figure 2, and is designed as follows:
  • the SOM neural network classifier is called a self-organizing feature mapping network.
  • the network is a non-teacher self-organizing and self-learning network composed of fully connected neuron arrays.
  • a neural network accepts the external input mode, it will be divided into different The reaction area, each area has different response characteristics to the input mode.
  • the invention patent uses SOM neural network classifier to classify samples of voltage, current and temperature of predicted characteristic parameters affecting battery power, and various sample parameters input corresponding RBF fuzzy neural network estimation model to predict battery SOC, SOM neural network learning algorithm as follows:
  • Fuzzy neural network is an intelligent technology that combines the powerful structural knowledge representation of fuzzy logic reasoning with the powerful self-learning ability of neural network.
  • This patent adopts an RBF fuzzy neural network with simple structure, good approximation ability and functional equivalence.
  • the RBF fuzzy neural network is a 4-layer structure, which is an input layer, a fuzzy layer, a fuzzy rule layer and a deblurring layer.
  • the first layer is the input layer.
  • the second layer is the blur layer.
  • the input parameters are blurred, where the three inputs are each divided into three fuzzy subsets ⁇ positive, positive, and zero ⁇ , so the layer has a total of nine nodes.
  • the membership degree of the jth fuzzy subset of the corresponding i-th input variable for each node pair For calculation, the membership function uses a Gaussian function.
  • the third layer is a fuzzy rule layer, which is used to match the fuzzy rule fronts and calculate the applicability of each rule.
  • the rule fitness of each node is obtained by a minimum operation.
  • the fourth layer is the deblurring layer, and the output of the fuzzy neural network is calculated by the weighted average method.
  • RBF neural network RBF-FNN
  • the membership function function of the RBF fuzzy neural network parameters, the membership function width, and the connection weights c ij , ⁇ ij between the rule layer and the de-blur layer The intensive learning adjustment of w mn is mainly divided into the following two stages. 1 In the practical application, the initial training adjustment of the parameters of the fuzzy neural network is carried out.
  • the fuzzy neural network under the current parameters is used to predict the battery SOC;
  • the initially trained fuzzy neural network adjusts the parameters of the fuzzy neural network online to dynamically adapt to the changes of the characteristic parameters of the network battery to achieve better battery load prediction.
  • ANFIS neural network-based adaptive fuzzy inference system
  • Adaptive Neuro-Fuzzy Inference System combines the two to combine the advantages of both and to compensate for their respective insufficient.
  • the fuzzy membership function and fuzzy rules in adaptive neural network fuzzy systems are obtained by learning a large amount of known data.
  • the biggest feature of ANFIS is the data-based modeling method, rather than based on experience or intuition. . This is especially important for systems where the features are not fully understood or the features are very complex.
  • the input of the ANFIS compensation estimation model is the output of the RBF fuzzy neural network estimation model, the predicted value of the internal resistance change of the battery and the predicted value of the environmental temperature change.
  • the output is the predicted amount of the battery SOC compensation.
  • Layer 1 Blurring the input data, the corresponding output of each node can be expressed as:
  • the invention patent has three nodes, which are the output of the RBF fuzzy neural network estimation model, the predicted value of the battery internal resistance change and the predicted value of the environmental temperature change.
  • Equation n is the number of each input membership function, and the membership function uses a Gaussian membership function.
  • Layer 2 Implementing rule operations, applicability of output rules, and multiplication of rule operations for ANFIS compensation estimation models.
  • Layer 4 The transfer function of each node is a linear function, representing a local linear model, each adaptive The output of node i should be:
  • Layer 5 The single node of this layer is a fixed node.
  • the total output of the compensated prediction value of the ANFIS compensation estimation model is calculated as:
  • conditional parameters determining the shape of the membership function and the conclusion parameters of the inference rule in the ANFIS compensation estimation model can be trained through the learning process.
  • the parameters are adjusted by a linear least squares estimation algorithm combined with gradient descent algorithm.
  • the input signal is forwarded along the network forward until the fourth layer.
  • the condition parameters are fixed, and the least squares estimation algorithm is used to adjust the conclusion parameters; the signal continues to be transmitted along the network forward until the output layer (ie, 5th floor).
  • the ANFIS compensation estimation model propagates the error signal back along the network and updates the condition parameters with the gradient method.
  • the global maximum advantage of the conclusion parameters can be obtained, which can not only reduce the dimension of the search space in the gradient method, but also improve the convergence speed of the ANFIS compensation estimation model parameters. .
  • the modeling process of the gray prediction GM(1,1) model is to accumulate the original data of the variables to be predicted, such as the irregular voltage, current, temperature, temperature change, and internal resistance change, to obtain the sequence with strong regularity. After modeling, the data obtained by the generated model is further subtracted to obtain the predicted value of the original data, and then predicted.
  • the original number of parameters as:
  • a becomes the development gray number, which reflects the development trend of x (1) and x (0) ;
  • u is the endogenous control gray number, reflecting the change relationship between the data.
  • the gray prediction model of the original sequence x (0) is obtained by the reduction of the following formula:
  • the gray prediction GM(1,1) model By constructing the gray prediction GM(1,1) model, the prediction of voltage, current, temperature and internal resistance change and ambient temperature variation of the patent power supply can be realized, and the gray prediction GM (1,1) corresponding to the battery characteristic parameters can be constructed. model.
  • the ARIMA model is a method for predicting modeled objects based on time series proposed by Box et al., which can be extended to analyze the time series of predicted objects.
  • This patent studies the time series characteristics of the ARIMA dynamic prediction model.
  • Three parameters are used to analyze the time series of changes in the SOC value of the battery, namely the autoregressive order (p), the difference order (d) and the moving average order (q). ).
  • the ARIMA dynamic prediction model is written as: ARIMA(p,d,q).
  • the ARIMA dynamic prediction battery SOC model equation with p, d, q as parameters can be expressed as follows:
  • ⁇ d y t represents the sequence of y t after d differential conversion
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • ⁇ t is the random error of time
  • a mutually independent white noise sequence a mutually independent white noise sequence
  • the battery SOC data sequence is non-stationary, if there is a certain increase or decrease trend, etc., the data needs to be differentially processed.
  • Commonly used tools are autocorrelation function graphs and partial autocorrelation function graphs. If the autocorrelation function quickly approaches zero, the battery SOC time series is a stationary time series. If there is a certain trend in the time series, the battery SOC data needs to be differentially processed. If there is a seasonal law, seasonal difference is needed. If the time series is heteroscedastic, the battery SOC data needs to be logarithmically converted. (2), model identification. The orders p, d and q of the ARIMA dynamic prediction battery SOC model are determined mainly by the autocorrelation coefficient and the partial autocorrelation coefficient.
  • the electric vehicle battery management system is arranged according to the components of the battery management system, and the system is arranged with a plane arrangement installation diagram of the current sensor, the voltage detection circuit, the load, the ambient temperature sensor, the battery temperature sensor, the battery pack and the measurement and control unit, wherein the ambient temperature sensor is arranged in the In the working environment of the detection battery pack, the battery temperature sensor is arranged in the outer casing of the battery pack, and the whole system is arranged in a plane as shown in FIG. 4, and the battery management parameter of the electric vehicle is collected and predicted by the system.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Sustainable Development (AREA)
  • Sustainable Energy (AREA)
  • Power Engineering (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Secondary Cells (AREA)

Abstract

La présente invention concerne un système de prédiction intelligent du SOC d'une batterie d'alimentation d'un véhicule électrique, le système de prédiction intelligent comprenant une plateforme de collecte de paramètres de batterie et un système de prédiction de SOC de batterie, la plateforme de collecte de paramètres de batterie étant utilisée pour collecter des paramètres en temps réel, tels que la tension, le courant et la température, d'un bloc-batterie d'alimentation de véhicule et une température ambiante; et le système de prédiction de SOC de batterie prédit, grâce aux paramètres collectés en temps réel, une valeur de SOC de batterie. Le SOC d'une batterie est un système en temps réel qui est non linéaire, retardé dans le temps, couplé à plusieurs variables et complexe, avec des demandes élevées en performances en temps réel. Le système de prédiction intelligent résout efficacement le problème selon lequel il est difficile pour un dispositif de prédiction classique d'obtenir un effet idéal de précision de prédiction de SOC d'une batterie.
PCT/CN2017/116819 2017-07-07 2017-12-18 Système de prédiction intelligent du soc d'une batterie d'alimentation de véhicule électrique Ceased WO2019006995A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710548650.3 2017-07-07
CN201710548650.3A CN107436409B (zh) 2017-07-07 2017-07-07 一种电动汽车动力电池soc智能预测装置

Publications (1)

Publication Number Publication Date
WO2019006995A1 true WO2019006995A1 (fr) 2019-01-10

Family

ID=60459809

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/116819 Ceased WO2019006995A1 (fr) 2017-07-07 2017-12-18 Système de prédiction intelligent du soc d'une batterie d'alimentation de véhicule électrique

Country Status (2)

Country Link
CN (1) CN107436409B (fr)
WO (1) WO2019006995A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11498450B2 (en) * 2019-05-21 2022-11-15 Rolls-Royce Plc Forecast of electric vehicle state of charge and energy storage capacity

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107436409B (zh) * 2017-07-07 2019-12-31 淮阴工学院 一种电动汽车动力电池soc智能预测装置
CN108445410B (zh) * 2018-04-02 2021-02-26 国家计算机网络与信息安全管理中心 一种监测蓄电池组运行状态的方法及装置
CN108226809A (zh) * 2018-04-13 2018-06-29 淮阴工学院 一种多模型并用的电池soc估算方法
CN111301222B (zh) * 2020-02-17 2021-07-16 北京嘀嘀无限科技发展有限公司 车辆电芯电压下降预警方法、电子设备及存储介质
CN111695301A (zh) * 2020-06-16 2020-09-22 中国科学院深圳先进技术研究院 电池电荷状态的预测方法及预测装置、存储介质、设备
CN112083346B (zh) * 2020-08-03 2021-11-09 山东大学 基于lstm的并联电池组内部电流分布估计方法及系统
CN114692494A (zh) * 2022-03-17 2022-07-01 广东工业大学 一种锂电池温度场在线建模方法及系统
CN116278957B (zh) * 2023-05-09 2025-09-02 赛力斯汽车有限公司 一种电池温度的预测方法、装置、车辆及存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101198922A (zh) * 2005-06-13 2008-06-11 Lg化学株式会社 用于测试电池充电状态的装置和方法
CN102749585A (zh) * 2011-04-21 2012-10-24 李昌 一种基于多层支持向量机的蓄电池在线监测方法
CN103744030A (zh) * 2014-01-12 2014-04-23 中国科学院电工研究所 电池组健康状态和荷电状态在线估算装置及估算方法
US20150066406A1 (en) * 2013-08-27 2015-03-05 The Regents Of The University Of Michigan On-board state of health monitoring of batteries using incremental capacity analysis
CN105093114A (zh) * 2015-03-02 2015-11-25 北京交通大学 一种电池在线建模与荷电状态的联合估计方法及系统
CN106777786A (zh) * 2017-01-12 2017-05-31 常州工学院 一种锂离子电池soc估算方法
CN107436409A (zh) * 2017-07-07 2017-12-05 淮阴工学院 一种电动汽车动力电池soc智能预测装置

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105929340B (zh) * 2016-06-30 2019-08-20 四川普力科技有限公司 一种基于arima估算电池soc的方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101198922A (zh) * 2005-06-13 2008-06-11 Lg化学株式会社 用于测试电池充电状态的装置和方法
CN102749585A (zh) * 2011-04-21 2012-10-24 李昌 一种基于多层支持向量机的蓄电池在线监测方法
US20150066406A1 (en) * 2013-08-27 2015-03-05 The Regents Of The University Of Michigan On-board state of health monitoring of batteries using incremental capacity analysis
CN103744030A (zh) * 2014-01-12 2014-04-23 中国科学院电工研究所 电池组健康状态和荷电状态在线估算装置及估算方法
CN105093114A (zh) * 2015-03-02 2015-11-25 北京交通大学 一种电池在线建模与荷电状态的联合估计方法及系统
CN106777786A (zh) * 2017-01-12 2017-05-31 常州工学院 一种锂离子电池soc估算方法
CN107436409A (zh) * 2017-07-07 2017-12-05 淮阴工学院 一种电动汽车动力电池soc智能预测装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI, SHAOMEI: "The Research of Battery SOC Estimation for Plug-In Hybrid Electric Vehicles", CHINA MASTER'S THESES FULL-TEXT DATABASE, 15 September 2010 (2010-09-15), pages 23 - 28, 43-52 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11498450B2 (en) * 2019-05-21 2022-11-15 Rolls-Royce Plc Forecast of electric vehicle state of charge and energy storage capacity

Also Published As

Publication number Publication date
CN107436409B (zh) 2019-12-31
CN107436409A (zh) 2017-12-05

Similar Documents

Publication Publication Date Title
CN107422272B (zh) 一种电动汽车动力电池soc智能化检测装置
WO2019006995A1 (fr) Système de prédiction intelligent du soc d'une batterie d'alimentation de véhicule électrique
Yang et al. A deep learning approach to state of charge estimation of lithium-ion batteries based on dual-stage attention mechanism
CN107367693B (zh) 一种电动汽车动力电池soc检测系统
Zafar et al. Hybrid deep learning model for efficient state of charge estimation of Li-ion batteries in electric vehicles
Wang et al. Energy management strategy of hybrid energy storage system for electric vehicles based on genetic algorithm optimization and temperature effect
CN108872866B (zh) 一种锂离子电池荷电状态动态评估与长效预测融合方法
Zhang et al. Data-driven state of health estimation in retired battery based on low and medium-frequency electrochemical impedance spectroscopy
CN112036084B (zh) 一种相似产品寿命迁移筛选方法和系统
CN113253116A (zh) 锂离子电池荷电状态估计方法、存储介质
CN113064093A (zh) 储能电池荷电状态与健康状态联合估算方法及系统
CN108519556A (zh) 一种基于循环神经网络的锂离子电池soc预测方法
CN111398832A (zh) 一种基于anfis模型的公交车电池soc预测方法
CN113093014B (zh) 一种基于阻抗参数的soh与soc的在线协同估计方法及系统
CN111856287A (zh) 基于堆叠残差因果卷积神经网络的锂电池健康状态检测方法
CN112630659A (zh) 一种基于改进bp-ekf算法的锂电池soc估算方法
CN112083333A (zh) 一种基于机器学习模型的动力电池组荷电状态估计方法
CN112883637A (zh) 锂离子电池rul的混沌时序非线性组合预测方法
CN114295987B (zh) 一种基于非线性卡尔曼滤波的电池soc状态估计方法
CN115841176A (zh) 充电桩变频负荷的安全态势感知方法及系统
AU2021101964A4 (en) Artificial intelligence based smart electric vehicle battery management system
CN120254651B (zh) 基于动态不确定性建模的锂电池剩余寿命实时预测方法
Niu et al. State of charge prediction study of vanadium redox-flow battery with BP neural network
Geng et al. SOC Prediction of power lithium battery using BP neural network theory based on keras
Li et al. Real-world data-driven charging strategies for incorporating health awareness in electric buses

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17917061

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17917061

Country of ref document: EP

Kind code of ref document: A1