CN110850298B - Lithium battery SOH estimation method and system based on data driving - Google Patents

Lithium battery SOH estimation method and system based on data driving Download PDF

Info

Publication number
CN110850298B
CN110850298B CN201911038721.0A CN201911038721A CN110850298B CN 110850298 B CN110850298 B CN 110850298B CN 201911038721 A CN201911038721 A CN 201911038721A CN 110850298 B CN110850298 B CN 110850298B
Authority
CN
China
Prior art keywords
network
battery
data
input
fuzzy
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
Application number
CN201911038721.0A
Other languages
Chinese (zh)
Other versions
CN110850298A (en
Inventor
张希
刘良俊
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.)
Shanghai Jiaotong University
Original Assignee
Shanghai Jiaotong University
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 Shanghai Jiaotong University filed Critical Shanghai Jiaotong University
Priority to CN201911038721.0A priority Critical patent/CN110850298B/en
Publication of CN110850298A publication Critical patent/CN110850298A/en
Application granted granted Critical
Publication of CN110850298B publication Critical patent/CN110850298B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • 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]
    • G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
    • 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]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Secondary Cells (AREA)

Abstract

The invention provides a lithium battery SOH estimation method and system based on data driving, which comprises the following steps: step 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit; step 2: cleaning the acquired data, and normalizing the data; and step 3: taking the load current, temperature and SOC data of the lithium ion battery as the input of a fuzzy neural network; and 4, step 4: establishing a fuzzy neural network structure, and determining the number of nodes in each layer; and 5: forward transmission is carried out to obtain the output error of the network; adjusting network parameters to minimize network output errors; step 6: calculating a network prediction error, and evaluating the accuracy of network characterization battery dynamic characteristics; and 7: and obtaining a virtual voltage response curve of the battery, and further calculating the capacity of the battery. The invention does not need to be based on a specific physical model, but is based on data modeling, and can shorten the modeling time.

Description

Lithium battery SOH estimation method and system based on data driving
Technical Field
The invention relates to the field of battery management systems, in particular to a lithium battery SOH estimation method and system based on data driving.
Background
Under the background of the increasingly serious energy problems and environmental problems in the current society, new energy vehicles such as hybrid vehicles and pure electric vehicles are gradually becoming mainstream in the industry of the automobile industry. Lithium ion batteries are important core components of electric vehicles. The Battery Management System (BMS) functions to ensure safe and stable operation of the battery. The state quantity of the battery, such as the capacity, the internal resistance and the like of the battery, cannot be directly measured by the vehicle-mounted sensor. Therefore, the BMS can only indirectly estimate the state of the battery through signals that can be directly measured by the vehicle-mounted sensors, such as the terminal voltage of the battery, the load current of the battery, and the temperature of the surface of the battery, in order to manage and monitor the battery. In addition, China is about to meet the retirement peak of the vehicle-mounted lithium ion power battery, and a large number of lithium ion batteries are utilized in a gradient manner at the time. Lithium ion batteries vary in capacity loss and in cases where they are reused. Therefore, before the gradient utilization of the retired lithium battery, the capacity test and the internal resistance test of the battery are required. The method for estimating the SOH of the battery at home and abroad mainly comprises an experimental analysis method and a method based on a battery model. Experimental analysis SOH was estimated by performing standard test experiments on the cells. The model-based method establishes a dynamic model of the lithium ion battery, and performs parameter identification and state identification based on the model. The lithium battery dynamic model comprises an equivalent circuit model and an electrochemical model. Most of the equivalent circuit parameters of the lithium battery are identified by using a least square method. The parameter identification of the electrochemical model mostly uses particle swarm algorithm, genetic algorithm and other optimization algorithms suitable for optimizing highly nonlinear objective functions.
The SOH is estimated by using an experimental analysis method, which is relatively direct and has higher precision, but professional experimental equipment such as a chemical workstation is expensive, and the time required by the test is longer. Under the background that a large number of lithium batteries are about to be retired and need to be tested, the efficiency of the lithium ion batteries in echelon utilization can be greatly influenced. The equivalent circuit model is simpler and has high calculation efficiency, but the equivalent circuit model is an approximate depiction of the dynamic characteristics of the battery, and the model precision is relatively lower. The electrochemical model can accurately describe the dynamic response characteristics of the battery, but the calculation of the electrochemical model involves a plurality of partial differential equations, the calculation amount is large, and the calculation amount is difficult to realize on-board BMS systems.
Patent document 110095732a discloses a lithium battery SOH estimation method considering the influence of ambient humidity on internal resistance, (1) by using a method of controlling variables, testing to obtain a change curve of internal resistance of a battery when the battery is discharged at the same rate, charge state and temperature under different ambient humidity, and fitting a change formula of the ambient humidity and the internal resistance of the battery; (2) adding self-discharge internal resistance influence factors into the current measurement of the internal resistance of the battery, and when the density of water molecules in the environment changes, correspondingly increasing or decreasing the self-discharge internal resistance of the battery according to the fitting relation between the self-discharge internal resistance of the battery and the environmental humidity, so that the internal resistance of the battery is updated according to the change of the environmental humidity, and the estimation formula of the SOH of the lithium battery is obtained. The method still has room to be perfected for the SOH estimation of the lithium battery based on data driving.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a lithium battery SOH estimation method and system based on data driving.
The invention provides a lithium battery SOH estimation method based on data driving, which comprises the following steps:
step 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
step 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and step 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and 4, step 4: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
step 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data.
Preferably, the step 2 includes:
step 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
step 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
step 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure BDA0002252266390000031
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of this state component for all samples.
Preferably, the step 4 comprises:
step 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33SOC (k) is an input component of SOC; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure BDA0002252266390000032
The calculation formula of (2) is as follows:
Figure BDA0002252266390000033
wherein, i is 1,2,3,
Figure BDA0002252266390000034
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure BDA0002252266390000035
Indicating network layer kiA node output of each node, the output value being equal to
Figure BDA0002252266390000036
Figure BDA0002252266390000037
Representing the ith input component xiMembership to fuzzy sets
Figure BDA0002252266390000038
Degree of membership of; in the present invention, the membership function is characterized by a general bell-shaped membership function, wherein
Figure BDA0002252266390000039
To belong to fuzzy sets
Figure BDA00022522663900000310
The center of membership of the generally bell-shaped membership function,
Figure BDA00022522663900000311
to belong to fuzzy sets
Figure BDA00022522663900000312
Is generally a standard deviation of a bell-shaped membership function,
Figure BDA00022522663900000313
the parameters are used for controlling the width of a general bell-shaped membership function curve;
step 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure BDA0002252266390000041
output of ith node of second layer of network
Figure BDA0002252266390000042
The calculation formula of (2) is as follows:
Figure BDA0002252266390000043
in the formula (II)
Figure BDA0002252266390000044
Is marked as
Figure BDA0002252266390000045
Figure BDA0002252266390000046
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure BDA0002252266390000047
Figure BDA0002252266390000048
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure BDA0002252266390000049
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure BDA00022522663900000410
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure BDA00022522663900000411
representing the input x to the current2All fuzzy subsets of i (k) partition, Bk2K-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure BDA00022522663900000412
representing an input x to SOC3All fuzzy subsets of the soc (k) partition,
Figure BDA00022522663900000413
k-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure BDA00022522663900000414
is the k-th layer of the networkiThe output of each node;
step 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure BDA00022522663900000415
The calculation formula of (2) is as follows:
Figure BDA00022522663900000416
step 4.4: constructing a fourth layer of the fuzzy neural network, and outputting the ith node of the fourth layer of the network
Figure BDA00022522663900000417
The calculation formula of (2) is as follows:
Figure BDA00022522663900000418
step 4.5: constructing fuzzy neural network fifth layer, network fifth layer ith node output
Figure BDA0002252266390000051
The calculation formula of (2) is as follows:
Figure BDA0002252266390000052
the input and output relations of the fuzzy neural network are as follows:
Figure BDA0002252266390000053
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
Preferably, the step 5 comprises:
step 5.1: for a given number n of samples (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure BDA0002252266390000054
Step 5.2: conditional parameters of fixed networks
Figure BDA0002252266390000055
Wherein the content of the first and second substances,
Figure BDA0002252266390000056
the network conclusion parameters are:
Figure BDA0002252266390000057
wherein the content of the first and second substances,
Figure BDA0002252266390000058
the output of the network can be rewritten as:
Figure BDA0002252266390000061
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure BDA0002252266390000062
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure BDA0002252266390000063
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure BDA0002252266390000064
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure BDA0002252266390000065
then this time
Figure BDA0002252266390000066
And in the foregoing
Figure BDA0002252266390000067
Representing the same quantity; and n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure BDA0002252266390000068
Namely:
Figure BDA0002252266390000069
obtaining conclusion parameter vector calculation result information;
step 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure BDA00022522663900000610
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure BDA00022522663900000611
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure BDA00022522663900000612
step 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Step 5.5: and (5.1) repeating the steps from 5.1 to 5.4 until the error of the network is lower than a preset value or a limited training round number is reached.
Preferably, the method comprises the following steps: the step 7 comprises the following steps:
step 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure BDA0002252266390000071
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure BDA0002252266390000072
therein, SOCinitThe initial value of the SOC of the battery is set to 100%, η represents the charging and discharging coulombic efficiency of the battery, and the value is generally determined through experiments, and the value is considered to be the value when the battery leaves the factory and is not changed, and can be a value provided by a battery manufacturer or a value of the charging and discharging coulombic efficiency of the battery set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
The invention provides a lithium battery SOH estimation system based on data driving, which comprises:
module 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
and (3) module 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and a module 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and (4) module: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and a module 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
and a module 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and a module 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data.
Preferably, the module 2 comprises:
module 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
module 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
module 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure BDA0002252266390000081
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of this state component for all samples.
Preferably, said module 4 comprises:
module 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33SOC (k) is SOCThis input component; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure BDA0002252266390000082
The calculation formula of (2) is as follows:
Figure BDA0002252266390000083
wherein, i is 1,2,3,
Figure BDA0002252266390000084
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure BDA0002252266390000085
Indicating network layer kiA node output of each node, the output value being equal to
Figure BDA0002252266390000086
Figure BDA0002252266390000087
Representing the ith input component xiMembership to fuzzy sets
Figure BDA0002252266390000088
Degree of membership of; in the present invention, the membership function is characterized by a general bell-shaped membership function, wherein
Figure BDA0002252266390000089
To belong to fuzzy sets
Figure BDA00022522663900000810
The center of membership of the generally bell-shaped membership function,
Figure BDA00022522663900000811
to belong to fuzzy sets
Figure BDA00022522663900000812
Is generally a standard deviation of a bell-shaped membership function,
Figure BDA0002252266390000091
the parameters are used for controlling the width of a general bell-shaped membership function curve;
module 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure BDA0002252266390000092
output of ith node of second layer of network
Figure BDA0002252266390000093
The calculation formula of (2) is as follows:
Figure BDA0002252266390000094
in the formula (II)
Figure BDA0002252266390000095
Is marked as
Figure BDA0002252266390000096
Figure BDA00022522663900000918
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure BDA0002252266390000097
Figure BDA0002252266390000098
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure BDA0002252266390000099
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure BDA00022522663900000910
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure BDA00022522663900000911
representing the input x to the current2All fuzzy subsets of the partition i (k),
Figure BDA00022522663900000912
k-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure BDA00022522663900000913
representing an input x to SOC3All fuzzy subsets of soc (k) partition, Ck3K-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure BDA00022522663900000914
is the k-th layer of the networkiThe output of each node;
module 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure BDA00022522663900000915
The calculation formula of (2) is as follows:
Figure BDA00022522663900000916
module 4.4: fourth layer of fuzzy neural networkOutput of ith node of fourth layer of network
Figure BDA00022522663900000917
The calculation formula of (2) is as follows:
Figure BDA0002252266390000101
module 4.5: constructing fuzzy neural network fifth layer, network fifth layer ith node output
Figure BDA0002252266390000102
The calculation formula of (2) is as follows:
Figure BDA0002252266390000103
the input and output relations of the fuzzy neural network are as follows:
Figure BDA0002252266390000104
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
Preferably, said module 5 comprises:
module 5.1: for a given number n of samples (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure BDA0002252266390000105
Module 5.2: conditional parameters of fixed networks
Figure BDA0002252266390000106
Wherein the content of the first and second substances,
Figure BDA0002252266390000107
the network conclusion parameters are:
Figure BDA0002252266390000108
wherein the content of the first and second substances,
Figure BDA0002252266390000109
the output of the network can be rewritten as:
Figure BDA0002252266390000111
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure BDA0002252266390000112
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure BDA0002252266390000113
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure BDA0002252266390000114
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure BDA0002252266390000115
then this time
Figure BDA0002252266390000116
And in the foregoing
Figure BDA0002252266390000117
Representing the same quantity; and n is sample (x)i,yi),i＝1,2,…N, m is the total number of fuzzy rules, p + q + r;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure BDA0002252266390000118
Namely:
Figure BDA0002252266390000119
obtaining conclusion parameter vector calculation result information;
module 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure BDA00022522663900001110
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure BDA00022522663900001111
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure BDA00022522663900001112
module 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Module 5.5: the blocks 5.1 to 5.4 are repeated until the error of the network is lower than a preset value or a defined number of training rounds is reached.
Preferably, the method comprises the following steps: the module 7 comprises:
module 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure BDA0002252266390000121
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure BDA0002252266390000122
therein, SOCinitThe initial value of the SOC of the battery is set to 100%, η represents the charging and discharging coulombic efficiency of the battery, and the value is generally determined through experiments, and the value is considered to be the value when the battery leaves the factory and is not changed, and can be a value provided by a battery manufacturer or a value of the charging and discharging coulombic efficiency of the battery set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention does not need professional experimental equipment such as a chemical workstation and the like, the required time is relatively short, and the cost is relatively low;
2. the method is based on data modeling instead of a specific physical model, so that the modeling time can be shortened;
3. the method has the advantages of moderate calculation time and relatively high calculation precision, and can finish the estimation of the SOH of the battery at relatively high speed on the premise of not sacrificing the precision
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
fig. 1 is a schematic diagram of a data acquisition process in an embodiment of the present invention.
Fig. 2 is a schematic diagram of a flow of an algorithm for obtaining battery capacity and internal resistance in an embodiment of the present invention.
Fig. 3 is a schematic diagram of a fuzzy neural network node according to an embodiment of the present invention.
Fig. 4 is a schematic structural diagram of a fuzzy neural network in the embodiment of the present invention.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that it would be obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit of the invention. All falling within the scope of the present invention.
The invention provides a lithium battery SOH estimation method based on data driving, which comprises the following steps:
step 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
step 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and step 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and 4, step 4: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
step 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data.
Preferably, the step 2 includes:
step 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
step 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
step 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure BDA0002252266390000131
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of this state component for all samples.
Preferably, the step 4 comprises:
step 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33＝SOC(k) An input component of SOC; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure BDA0002252266390000141
The calculation formula of (2) is as follows:
Figure BDA0002252266390000142
wherein, i is 1,2,3,
Figure BDA0002252266390000143
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure BDA0002252266390000144
Indicating network layer kiA node output of each node, the output value being equal to
Figure BDA0002252266390000145
Figure BDA0002252266390000146
Representing the ith input component xiMembership to fuzzy sets
Figure BDA0002252266390000147
Degree of membership of; in the present invention, the membership function is characterized by a general bell-shaped membership function, wherein
Figure BDA0002252266390000148
To belong to fuzzy sets
Figure BDA0002252266390000149
The center of membership of the generally bell-shaped membership function,
Figure BDA00022522663900001410
to belong to fuzzy sets
Figure BDA00022522663900001411
Is generally a standard deviation of a bell-shaped membership function,
Figure BDA00022522663900001412
the parameters are used for controlling the width of a general bell-shaped membership function curve;
step 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure BDA00022522663900001413
output of ith node of second layer of network
Figure BDA00022522663900001414
The calculation formula of (2) is as follows:
Figure BDA0002252266390000151
in the formula (II)
Figure BDA0002252266390000152
Is marked as
Figure BDA0002252266390000153
Figure BDA0002252266390000154
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure BDA0002252266390000155
Figure BDA0002252266390000156
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure BDA0002252266390000157
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure BDA0002252266390000158
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure BDA0002252266390000159
representing the input x to the current2All fuzzy subsets of the partition i (k),
Figure BDA00022522663900001510
k-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure BDA00022522663900001511
representing an input x to SOC3All fuzzy subsets of the soc (k) partition,
Figure BDA00022522663900001512
k-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure BDA00022522663900001513
is the k-th layer of the networkiThe output of each node;
step 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure BDA00022522663900001514
The calculation formula of (2) is as follows:
Figure BDA00022522663900001515
step 4.4: constructing a fourth layer of the fuzzy neural network, and outputting the ith node of the fourth layer of the network
Figure BDA00022522663900001516
The calculation formula of (2) is as follows:
Figure BDA00022522663900001517
step 4.5: constructing fuzzy neural network fifth layer, network fifth layer ith node output
Figure BDA00022522663900001518
The calculation formula of (2) is as follows:
Figure BDA00022522663900001519
the input and output relations of the fuzzy neural network are as follows:
Figure BDA0002252266390000161
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
Preferably, the step 5 comprises:
step 5.1: for a given number n of samples (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure BDA0002252266390000162
Step 5.2: conditional parameters of fixed networks
Figure BDA0002252266390000163
Wherein the content of the first and second substances,
Figure BDA0002252266390000164
the network conclusion parameters are:
Figure BDA0002252266390000165
wherein the content of the first and second substances,
Figure BDA0002252266390000166
the output of the network can be rewritten as:
Figure BDA0002252266390000167
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure BDA0002252266390000168
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure BDA0002252266390000169
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure BDA00022522663900001610
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure BDA0002252266390000171
then this time
Figure BDA0002252266390000172
And in the foregoing
Figure BDA0002252266390000173
Representing the same quantity; and n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure BDA0002252266390000174
Namely:
Figure BDA0002252266390000175
obtaining conclusion parameter vector calculation result information;
step 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure BDA0002252266390000176
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure BDA0002252266390000177
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure BDA0002252266390000178
step 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Step 5.5: and (5.1) repeating the steps from 5.1 to 5.4 until the error of the network is lower than a preset value or a limited training round number is reached.
Preferably, the method comprises the following steps: the step 7 comprises the following steps:
step 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure BDA0002252266390000179
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure BDA00022522663900001710
therein, SOCinitThe initial value of the SOC of the battery is set to 100%, η represents the charging and discharging coulombic efficiency of the battery, and the value is generally determined through experiments, and the value is considered to be the value when the battery leaves the factory and is not changed, and can be a value provided by a battery manufacturer or a value of the charging and discharging coulombic efficiency of the battery set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
The invention provides a lithium battery SOH estimation system based on data driving, which comprises:
module 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
and (3) module 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and a module 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and (4) module: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and a module 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
and a module 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and a module 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data.
Preferably, the module 2 comprises:
module 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
module 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
module 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure BDA0002252266390000191
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of this state component for all samples.
Preferably, said module 4 comprises:
module 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33SOC (k) is an input component of SOC; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure BDA0002252266390000192
The calculation formula of (2) is as follows:
Figure BDA0002252266390000193
wherein, i is 1,2,3,
Figure BDA0002252266390000194
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure BDA0002252266390000195
Indicating network layer kiA node output of each node, the output value being equal to
Figure BDA0002252266390000196
Figure BDA0002252266390000197
Representing the ith input component xiMembership to fuzzy sets
Figure BDA0002252266390000198
Degree of membership of; in the present invention, the membership function is characterized by a general bell-shaped membership function, wherein
Figure BDA0002252266390000199
To belong to fuzzy sets
Figure BDA00022522663900001910
The center of membership of the generally bell-shaped membership function,
Figure BDA00022522663900001911
to belong to fuzzy sets
Figure BDA00022522663900001912
Is generally a standard deviation of a bell-shaped membership function,
Figure BDA00022522663900001913
the parameters are used for controlling the width of a general bell-shaped membership function curve;
module 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure BDA0002252266390000201
output of ith node of second layer of network
Figure BDA0002252266390000202
The calculation formula of (2) is as follows:
Figure BDA0002252266390000203
in the formula (II)
Figure BDA0002252266390000204
Is marked as
Figure BDA0002252266390000205
Figure BDA0002252266390000206
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure BDA0002252266390000207
Figure BDA0002252266390000208
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure BDA0002252266390000209
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure BDA00022522663900002010
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure BDA00022522663900002011
representing the input x to the current2All fuzzy subsets of the partition i (k),
Figure BDA00022522663900002012
k-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure BDA00022522663900002013
representing an input x to SOC3All fuzzy subsets of the soc (k) partition,
Figure BDA00022522663900002014
k-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure BDA00022522663900002015
is the k-th layer of the networkiThe output of each node;
module 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure BDA00022522663900002016
The calculation formula of (2) is as follows:
Figure BDA00022522663900002017
module 4.4: constructing a fourth layer of the fuzzy neural network, and outputting the ith node of the fourth layer of the network
Figure BDA00022522663900002018
The calculation formula of (2) is as follows:
Figure BDA00022522663900002019
module 4.5: constructing fuzzy neural network fifth layer, network fifth layer ith node output
Figure BDA0002252266390000211
The calculation formula of (2) is as follows:
Figure BDA0002252266390000212
the input and output relations of the fuzzy neural network are as follows:
Figure BDA0002252266390000213
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
Preferably, said module 5 comprises:
module 5.1: for a given number n of samplesBook (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure BDA0002252266390000214
Module 5.2: conditional parameters of fixed networks
Figure BDA0002252266390000215
Wherein the content of the first and second substances,
Figure BDA0002252266390000218
the network conclusion parameters are:
Figure BDA0002252266390000216
wherein the content of the first and second substances,
Figure BDA0002252266390000217
the output of the network can be rewritten as:
Figure BDA0002252266390000221
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure BDA0002252266390000222
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure BDA0002252266390000223
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure BDA0002252266390000224
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure BDA0002252266390000225
then this time
Figure BDA0002252266390000226
And in the foregoing
Figure BDA0002252266390000227
Representing the same quantity; and n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure BDA0002252266390000228
Namely:
Figure BDA0002252266390000229
obtaining conclusion parameter vector calculation result information;
module 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure BDA00022522663900002210
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure BDA00022522663900002211
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure BDA00022522663900002212
module 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Module 5.5: the blocks 5.1 to 5.4 are repeated until the error of the network is lower than a preset value or a defined number of training rounds is reached.
Preferably, the method comprises the following steps: the module 7 comprises:
module 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure BDA0002252266390000231
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure BDA0002252266390000232
therein, SOCinitThe initial value of the SOC of the battery is set to 100%, η represents the charging and discharging coulombic efficiency of the battery, and the value is generally determined through experiments, and the value is considered to be the value when the battery leaves the factory and is not changed, and can be a value provided by a battery manufacturer or a value of the charging and discharging coulombic efficiency of the battery set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
Specifically, in one embodiment, the required battery data is first extracted from the on-board BMS. And then cleaning, sorting and preprocessing the extracted data. And then, constructing a fuzzy neural network model, and designing the structure of the fuzzy neural network. And then training parameters in the fuzzy neural network by using the sorted data. And finally, manually set input is transmitted to the trained fuzzy neural network to carry out a virtual capacity experiment, and finally an estimated value of the battery capacity is obtained to represent the SOH of the battery. The method comprises the following specific steps:
the battery data acquisition flow is shown in fig. 1. Current, voltage and temperature sensors are arranged at corresponding positions on the battery pack of the electric automobile, sensor data are transmitted to a vehicle-mounted Battery Management System (BMS) through a CAN (controller area network) wired network, and the BMS records response voltage, current and temperature data and estimates the SOC of the battery. The battery management system uploads battery operation data to the T-BOX, and the T-BOX transmits the data to the cloud big data center through the 4G wireless network for further analysis and processing;
and performing data cleaning, and deleting data points which are seriously deviated in the data sequence. And when the difference value between the numerical value at the point and the previous moment value is more than 1.5 times of the previous moment value, the data point at the current time is determined to be an abnormal data point, and the data value at the point is obtained by linearly difference the data values at the previous moment and the next moment. For missing data, the linear difference result of the data values at the previous and next time instants is also used for padding. And then, carrying out normalization processing on the data set by adopting a maximum and minimum normalization method. And finally, dividing the data set, and dividing the data set into a training set, a verification set and a test set according to 70%, 0% and 20% of the number of the data set samples. The maximum and minimum normalization method has the following calculation formula, wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of this state component for all samples:
Figure BDA0002252266390000241
and (5) carrying out structural design of the fuzzy neural network. The structure of the fuzzy neural network used is shown in fig. 4. The input of the network is selected as the temperature, load current and SOC value of the battery at the k-th moment, and the output of the network is selected as the predicted value of the terminal voltage of the battery at the k + 1-th moment. The behavior of the network satisfies the following function:
Figure BDA0002252266390000242
the first layer of the network is a fuzzification layer, which fuzzifies input variables and outputs of the ith node of the first layer of the network
Figure BDA0002252266390000243
The calculation formula of (2) is as follows:
Figure BDA0002252266390000244
wherein i is 1,2,3,
Figure BDA0002252266390000245
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided. Where p-q-r-5, three network input components;
the second layer of the network realizes the calculation of the conditional part in the fuzzy logic 'if … the …', and adopts a first-order Sugeno fuzzy rule, and p.q.r is 125 fuzzy rules. The fuzzy rule is as follows:
Figure BDA0002252266390000246
output of i nodes of second layer of network
Figure BDA0002252266390000247
The calculation formula of (2) is as follows:
Figure BDA0002252266390000251
the third layer is the normalization process of the output of the second layer of the network, and the output of the ith node of the third layer of the network
Figure BDA0002252266390000252
The calculation formula of (2) is as follows:
Figure BDA0002252266390000253
the fourth layer is a fuzzy inference layer, and the output of the ith node of the fourth layer of the network
Figure BDA0002252266390000254
The calculation formula of (2) is as follows:
Figure BDA0002252266390000255
the fifth layer is an output layer or a deblurring layer, and the output of the ith node of the network
Figure BDA0002252266390000256
The calculation formula of (2) is as follows:
Figure BDA0002252266390000257
the input and output relations of the fuzzy neural network are as follows:
Figure BDA0002252266390000258
parameters of the fuzzy neural network are trained using a hybrid training algorithm. Dividing parameters of fuzzy neural network into conditional parameters thetapAnd a conclusion parameter thetacTwo sets of parameters were trained using different methods, respectively. Mixing ofThe co-training algorithm comprises the following steps:
first, for a given number n of samples (x)i,yi) Will input xiThe vector is input into the network. When the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure BDA0002252266390000259
Then the condition parameters of the fixed network:
Figure BDA00022522663900002510
wherein
Figure BDA0002252266390000261
The fitting of the theoretical parameters is performed using a least squares algorithm. The network conclusion parameters are noted as:
Figure BDA0002252266390000262
wherein
Figure BDA0002252266390000263
The output of the network can be rewritten as:
Figure BDA0002252266390000264
wherein the matrix A, thetacAnd y is in the shape of: n × m, m × 1, n × 1, where m ═ p + q + r. In addition, the
Figure BDA0002252266390000265
Is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure BDA0002252266390000266
and the applicability corresponding to the normalized jth fuzzy rule is shown when the ith sample is used as system input.
Figure BDA0002252266390000267
And (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure BDA0002252266390000268
then this time
Figure BDA0002252266390000269
And in the foregoing
Figure BDA00022522663900002610
The same quantity is indicated. And n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
using the least squares method, the minimum mean square error, min | | A θ, can be obtainedc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure BDA00022522663900002611
Namely:
Figure BDA00022522663900002612
the calculated conclusion parameter vector
Figure BDA00022522663900002613
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure BDA00022522663900002614
Computing the error of the network output according to a mean square error criterion (MSE)：
Figure BDA00022522663900002615
Updating the conditional parameter theta of the network using an error back propagation algorithm based on the error of the network output calculated in the above equationp。
Repeating the previous four steps until the error of the network is lower than a preset value or the number of times of a limited training round is reached;
after the network training is completed, the dynamic characteristics of the network should be approximately equal to the dynamic characteristics of the actual battery, and the battery model based on the fuzzy neural network and the actual battery model can be considered to be equivalent to each other. After that, the artificial defined input quantity is applied to the battery model based on the fuzzy neural network to perform the virtual capacity test, which is equivalent to performing the capacity test on the actual battery, so that the estimated value of the capacity of the battery can be obtained. The virtual capacity test adopts a constant current discharge mode, the discharge current is 0.3C, the battery temperature is constant at 25 ℃, and the sequence sampling period is 1 s. The input sequence of the network is:
Figure BDA0002252266390000271
wherein the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure BDA0002252266390000272
wherein the SOCinitThe initial value of the SOC of the battery is set as 100%, eta represents the charging and discharging coulombic efficiency of the battery, the value is generally determined through experiments, the value is considered to be the constant value when the battery leaves a factory, and the value can be the value provided by a battery manufacturer or the value of the charging and discharging coulombic efficiency of the battery set in a BMS system; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
the input sequence is continuously input to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage (taking 2.5V). At this time, the current is integrated by an ampere-hour integration method to obtain the total number of coulombs discharged by the battery, and the value is used as an estimation value of the battery capacity, and the value is used for defining the SOH of the battery.
The invention does not need professional experimental equipment such as a chemical workstation and the like, the required time is relatively short, and the cost is relatively low; the method is based on data modeling instead of a specific physical model, so that the modeling time can be shortened; the method has the advantages of moderate calculation time and relatively high calculation precision, and can finish the estimation of the SOH of the battery at a relatively high speed on the premise of not sacrificing the precision.
Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, units provided by the present invention as pure computer readable program code, the system and its various devices, modules, units provided by the present invention can be fully implemented by logically programming method steps in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Therefore, the system and various devices, modules and units thereof provided by the invention can be regarded as a hardware component, and the devices, modules and units included in the system for realizing various functions can also be regarded as structures in the hardware component; means, modules, units for performing the various functions may also be regarded as structures within both software modules and hardware components for performing the method.
The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention. The embodiments and features of the embodiments of the present application may be combined with each other arbitrarily without conflict.

Claims (6)

1. A lithium battery SOH estimation method based on data driving is characterized by comprising the following steps:
step 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
step 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and step 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and 4, step 4: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
step 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data;
the step 2 comprises the following steps:
step 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
step 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
step 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure FDA0003044363960000011
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of the state component for all samples;
the step 4 comprises the following steps:
step 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33SOC (k) is an input component of SOC; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure FDA0003044363960000021
The calculation formula of (2) is as follows:
Figure FDA0003044363960000022
wherein, i is 1,2,3,
Figure FDA0003044363960000023
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure FDA0003044363960000024
Indicating network layer kiA node output of each node, the output value being equal to
Figure FDA0003044363960000025
Figure FDA0003044363960000026
Representing the ith input component xiMembership to fuzzy sets
Figure FDA0003044363960000027
Degree of membership of; the membership function is characterized by a generally bell-shaped membership function, wherein
Figure FDA0003044363960000028
To belong to fuzzy sets
Figure FDA0003044363960000029
The center of membership of the generally bell-shaped membership function,
Figure FDA00030443639600000210
to belong to fuzzy sets
Figure FDA00030443639600000211
Is generally a standard deviation of a bell-shaped membership function,
Figure FDA00030443639600000212
the parameters are used for controlling the width of a general bell-shaped membership function curve;
step 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure FDA00030443639600000213
output of ith node of second layer of network
Figure FDA00030443639600000214
The calculation formula of (2) is as follows:
Figure FDA00030443639600000215
in the formula (II)
Figure FDA0003044363960000031
Is marked as
Figure FDA0003044363960000032
Figure FDA0003044363960000033
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure FDA0003044363960000034
Figure FDA0003044363960000035
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure FDA0003044363960000036
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure FDA0003044363960000037
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure FDA0003044363960000038
indicates this to the currentInput x2All fuzzy subsets of the partition i (k),
Figure FDA0003044363960000039
k-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure FDA00030443639600000310
representing an input x to SOC3All fuzzy subsets of the soc (k) partition,
Figure FDA00030443639600000311
k-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure FDA00030443639600000312
is the k-th layer of the networkiThe output of each node;
step 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure FDA00030443639600000313
The calculation formula of (2) is as follows:
Figure FDA00030443639600000314
step 4.4: constructing a fourth layer of the fuzzy neural network, and outputting the ith node of the fourth layer of the network
Figure FDA00030443639600000315
The calculation formula of (2) is as follows:
Figure FDA00030443639600000316
step 4.5: constructing a fifth layer of fuzzy neural network, the ith networkOutput of the node
Figure FDA00030443639600000317
The calculation formula of (2) is as follows:
Figure FDA00030443639600000318
the input and output relations of the fuzzy neural network are as follows:
Figure FDA0003044363960000041
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
2. The method for estimating the SOH of the lithium battery based on the data driving according to claim 1, wherein the step 5 comprises:
step 5.1: for a given number n of samples (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure FDA0003044363960000042
Step 5.2: conditional parameters of fixed networks
Figure FDA0003044363960000043
Wherein the content of the first and second substances,
Figure FDA0003044363960000044
the network conclusion parameters are:
Figure FDA0003044363960000045
wherein the content of the first and second substances,
Figure FDA0003044363960000046
the output of the network can be rewritten as:
Figure FDA0003044363960000047
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure FDA0003044363960000048
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure FDA0003044363960000049
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure FDA0003044363960000051
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure FDA0003044363960000052
then this time
Figure FDA0003044363960000053
And in the foregoing
Figure FDA0003044363960000054
Representing the same quantity; and n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure FDA0003044363960000055
Namely:
Figure FDA0003044363960000056
obtaining conclusion parameter vector calculation result information;
step 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure FDA0003044363960000057
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure FDA0003044363960000058
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure FDA0003044363960000059
step 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Step 5.5: and (5.1) repeating the steps from 5.1 to 5.4 until the error of the network is lower than a preset value or a limited training round number is reached.
3. The SOH estimation method for the lithium battery based on the data driving according to claim 1, comprising: the step 7 comprises the following steps:
step 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure FDA00030443639600000510
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure FDA00030443639600000511
therein, SOCinitThe initial value of the SOC of the battery is set to be 100%, eta represents the charging and discharging coulombic efficiency of the battery, the value is determined through experiments, the value is considered to be the unchanged value of the battery when the battery leaves a factory, and the value is taken as the charging and discharging coulombic efficiency value of the battery set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
4. A lithium battery SOH estimation system based on data driving is characterized by comprising:
module 1: acquiring a recent data packet sequence of the lithium ion battery recorded by the BMS unit to acquire recent data packet information;
and (3) module 2: according to recent data packet information, data cleaning is carried out on the collected data, abnormal data points are removed, data missing points are filled, normalization processing is carried out on the data, and normalization processing result information is obtained;
and a module 3: according to the normalization processing result information, taking the load current, the temperature and the SOC data of the lithium ion battery as the input of a fuzzy neural network, taking the terminal voltage response value of the lithium ion battery as the output of the network, dividing a data set into a training set and a testing set, and acquiring input and output characteristic information;
and (4) module: establishing a fuzzy neural network structure according to the input and output characteristic information and determining the number of nodes of each layer;
and a module 5: inputting the preprocessed battery load current, temperature and SOC data into a fuzzy neural network structure after random initialization parameters, and transmitting the data in a forward direction to obtain the output error of the network; adjusting network parameters to minimize network output errors;
and a module 6: inputting the test set sample into a trained fuzzy neural network, calculating a network prediction error, and evaluating the accuracy of the network describing the dynamic characteristics of the battery;
and a module 7: inputting a standard test load into the trained fuzzy neural network, performing virtual battery capacity test to obtain a virtual voltage response curve of the battery, further calculating the capacity of the battery, and obtaining the information of the characterization result of the SOH value of the lithium battery;
the recent data packet comprises: load current data, terminal voltage data, temperature data, and SOC data;
the module 2 comprises:
module 2.1: carrying out data cleaning, and deleting data points which are seriously deviated in the data sequence, wherein the data value at the point is obtained by carrying out linear difference on the data values at the front moment and the rear moment;
module 2.2: deleting repeated redundant data, and for redundant data and missing data, using linear difference results of data values at the previous moment and the next moment to perform completion;
module 2.3: carrying out normalization processing on the data set, and calculating by adopting maximum and minimum normalization:
Figure FDA0003044363960000071
wherein X refers to the value of a certain state component of the current sample, and X isminIs the minimum of the state components of all samples, XmaxIs the maximum of the state component of all samples；
The module 4 comprises:
module 4.1: constructing the first layer of the fuzzy neural network, fuzzifying the input variable, and recording xiI is 1,2,3, i is the ith input component of the network; when i is 1, x1T (k) is the input component of temperature, and x is when i is 22I (k) is the input component of current, and x is when i is 33SOC (k) is an input component of SOC; the ith input component xiNetwork first layer kth with i ═ 1,2,3iOutput of each node
Figure FDA0003044363960000072
The calculation formula of (2) is as follows:
Figure FDA0003044363960000073
wherein, i is 1,2,3,
Figure FDA0003044363960000074
where p is the input x to temperature1T (k) number of fuzzy subsets divided, q is the input x to the current2I (k) number of fuzzy subsets divided, r is input x to SOC3Soc (k) number of fuzzy subsets divided; in the formula
Figure FDA0003044363960000075
Indicating network layer kiA node output of each node, the output value being equal to
Figure FDA0003044363960000076
Figure FDA0003044363960000077
Representing the ith input component xiMembership to fuzzy sets
Figure FDA0003044363960000078
Is subject toDegree; the membership function is characterized by a generally bell-shaped membership function, wherein
Figure FDA0003044363960000079
To belong to fuzzy sets
Figure FDA00030443639600000710
The center of membership of the generally bell-shaped membership function,
Figure FDA00030443639600000711
to belong to fuzzy sets
Figure FDA00030443639600000712
Is generally a standard deviation of a bell-shaped membership function,
Figure FDA00030443639600000713
the parameters are used for controlling the width of a general bell-shaped membership function curve;
module 4.2: constructing a second layer of the fuzzy neural network, and adopting a first-order Sugeno fuzzy rule to share p.q.r fuzzy rules; the fuzzy rule is as follows:
Figure FDA0003044363960000081
output of ith node of second layer of network
Figure FDA0003044363960000082
The calculation formula of (2) is as follows:
Figure FDA0003044363960000083
in the formula (II)
Figure FDA0003044363960000084
Is marked as
Figure FDA0003044363960000085
Figure FDA0003044363960000086
Representing the output of the ith node of the second layer of the network, with a value equal to
Figure FDA0003044363960000087
Figure FDA0003044363960000088
Indicates that the current network input x ═ x1,x2,x3) For rules
Figure FDA0003044363960000089
The degree of engagement of; a. the1,A2,…,Ai,…,ApRepresenting the input x to temperature1T (k) all fuzzy subsets of the partition,
Figure FDA00030443639600000810
k-th indicating input division into temperature1A (k)11,2, …, p) fuzzy subset;
Figure FDA00030443639600000811
representing the input x to the current2All fuzzy subsets of the partition i (k),
Figure FDA00030443639600000812
k-th representing the division of this input into currents2A (k)21,2, …, q) fuzzy subset;
Figure FDA00030443639600000813
representing an input x to SOC3All fuzzy subsets of the soc (k) partition,
Figure FDA00030443639600000814
k-th representing input division to SOC3A (k)31,2, …, r) fuzzy subset;
Figure FDA00030443639600000815
is the k-th layer of the networkiThe output of each node;
module 4.3: constructing a third layer of the fuzzy neural network, and outputting the ith node of the third layer of the network
Figure FDA00030443639600000816
The calculation formula of (2) is as follows:
Figure FDA00030443639600000817
module 4.4: constructing a fourth layer of the fuzzy neural network, and outputting the ith node of the fourth layer of the network
Figure FDA00030443639600000818
The calculation formula of (2) is as follows:
Figure FDA00030443639600000819
module 4.5: constructing fuzzy neural network fifth layer, network fifth layer ith node output
Figure FDA0003044363960000091
The calculation formula of (2) is as follows:
Figure FDA0003044363960000092
the input and output relations of the fuzzy neural network are as follows:
Figure FDA0003044363960000093
wherein g (θ, x) represents a functional relationship between the input x and the output y of the network, and θ represents a parameter to be trained in the network.
5. The system for estimating SOH of a lithium battery based on data driving of claim 4, wherein the module 5 comprises:
module 5.1: for a given number n of samples (x)i,yi) Will input xiInputting the vector into the network; when the input forward direction is transmitted to the fourth layer of the network, the calculation is suspended, and the applicability corresponding to each fuzzy rule after the normalization of the third layer of the network is obtained
Figure FDA0003044363960000094
Module 5.2: conditional parameters of fixed networks
Figure FDA0003044363960000095
Wherein the content of the first and second substances,
Figure FDA0003044363960000096
the network conclusion parameters are:
Figure FDA0003044363960000097
wherein the content of the first and second substances,
Figure FDA0003044363960000098
the output of the network can be rewritten as:
Figure FDA0003044363960000101
matrix A, thetacAnd y is in the shape of: nxm, mx 1, nx1, wherein m ═ p + q + r;
in addition, the first and second substrates are,
Figure FDA0003044363960000102
is a line vector, i denotes the ith sample, j denotes the corresponding jth fuzzy rule,
Figure FDA0003044363960000103
representing the applicability corresponding to the normalized jth fuzzy rule when the ith sample is input as a unit;
Figure FDA0003044363960000104
and (3) representing a conclusion parameter vector corresponding to the jth fuzzy rule, and setting the condition (antecedent) of the jth fuzzy rule as follows:
Figure FDA0003044363960000105
then this time
Figure FDA0003044363960000106
And in the foregoing
Figure FDA0003044363960000107
Representing the same quantity; and n is sample (x)i,yi) I is 1,2, …, the total number of n, m is p + q + r is the total number of fuzzy rules;
to obtain the minimum mean square error, i.e. min | | A θc-y||,y＝(y1,y2,…,yn)TConclusion parameter vector θ in sensecBest estimate of
Figure FDA0003044363960000108
Namely:
Figure FDA0003044363960000109
obtaining conclusion parameter vector calculation result information;
module 5.3: calculating result information according to the conclusion parameter vector, and calculating the conclusion parameter vector
Figure FDA00030443639600001010
Fixing, the input continues to forward pass from the fourth layer of the network until passing to the output layer (i.e. the fifth layer of the network), and the output of the network is obtained
Figure FDA00030443639600001011
Calculating the error of the network output according to a mean square error criterion (MSE):
Figure FDA00030443639600001012
module 5.4: updating the conditional parameter θ of the network using an error back propagation algorithm based on the error of the network output calculated in S53p；
Module 5.5: the blocks 5.1 to 5.4 are repeated until the error of the network is lower than a preset value or a defined number of training rounds is reached.
6. The SOH estimation system for lithium batteries based on data driving according to claim 4, characterized in that it comprises: the module 7 comprises:
module 7.1: by adopting a constant-current discharging mode, the discharging current is 0.3C, the battery temperature is constant at 25 ℃, the sequence sampling period is 1s, and the input sequence of the network is as follows:
Figure FDA0003044363960000111
wherein, the SOC sequence is obtained by an ampere-hour integration method, namely:
Figure FDA0003044363960000112
therein, SOCinitRepresents an initial value of SOC of the battery set to 100%, and η represents a coulombic efficiency of charge and discharge of the battery, and the value is determined by experimentsConsidering that the value is unchanged when the battery leaves the factory, and taking the value as the battery charging and discharging coulombic efficiency value set in the BMS unit; cmaxThe maximum available capacity of the battery is set as the battery capacity obtained after SOH estimation is carried out on the battery for the last time;
continuously inputting the input sequence to the fuzzy neural network until the output of the network, namely the estimated battery terminal voltage response value reaches the set cut-off voltage; and integrating the current to obtain the total number of discharged coulombs of the battery, taking the value as the estimation of the battery capacity, defining the SOH of the battery by the value, and obtaining the information of the characterization result of the SOH value of the lithium battery.
CN201911038721.0A 2019-10-29 2019-10-29 Lithium battery SOH estimation method and system based on data driving Active CN110850298B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911038721.0A CN110850298B (en) 2019-10-29 2019-10-29 Lithium battery SOH estimation method and system based on data driving

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911038721.0A CN110850298B (en) 2019-10-29 2019-10-29 Lithium battery SOH estimation method and system based on data driving

Publications (2)

Publication Number Publication Date
CN110850298A CN110850298A (en) 2020-02-28
CN110850298B true CN110850298B (en) 2021-08-17

Family

ID=69598963

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911038721.0A Active CN110850298B (en) 2019-10-29 2019-10-29 Lithium battery SOH estimation method and system based on data driving

Country Status (1)

Country Link
CN (1) CN110850298B (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111832220A (en) * 2020-06-16 2020-10-27 天津大学 Lithium ion battery health state estimation method based on codec model
CN113406503A (en) * 2021-06-01 2021-09-17 重庆大学 Lithium battery SOH online estimation method based on deep neural network
CN113740739B (en) * 2021-07-22 2023-05-02 西安交通大学 Vehicle-mounted starting maintenance-free lead-acid storage battery residual life prediction method
CN113589183A (en) * 2021-08-09 2021-11-02 湖北亿纬动力有限公司 Self-discharge detection method, device and system based on big data
CN113484774B (en) * 2021-08-18 2024-01-26 昆明理工大学 Lithium battery pack capacity estimation method based on temperature calibration and neural network
CN113820604B (en) * 2021-08-30 2024-04-26 昆明理工大学 Lithium battery SOH estimation method based on temperature prediction
CN116068449B (en) * 2023-03-23 2023-07-14 中国科学技术大学 Lithium battery evaluation system and method for physical model constraint deep neural network
CN117825975A (en) * 2024-03-05 2024-04-05 烟台海博电气设备有限公司 Data-driven lithium ion battery SOH evaluation method and system
CN117949832B (en) * 2024-03-27 2024-06-18 湖南大学 Battery SOH analysis method based on optimized neural network

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7072871B1 (en) * 2001-08-22 2006-07-04 Cadex Electronics Inc. Fuzzy logic method and apparatus for battery state of health determination
CN104348205A (en) * 2013-07-31 2015-02-11 杭州锂软科技有限公司 SOC-SOH (state of charge-state of health)-based distributed BMS (Battery Management System)
CN107436411A (en) * 2017-07-28 2017-12-05 南京航空航天大学 Battery SOH On-line Estimation methods based on fractional order neural network and dual-volume storage Kalman
CN109061505A (en) * 2018-08-28 2018-12-21 淮阴工学院 A kind of detection method of lithium battery SOH
CN110095732A (en) * 2019-05-23 2019-08-06 合肥工业大学智能制造技术研究院 A kind of lithium battery SOH estimation method for considering ambient humidity and internal resistance being influenced

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7072871B1 (en) * 2001-08-22 2006-07-04 Cadex Electronics Inc. Fuzzy logic method and apparatus for battery state of health determination
CN104348205A (en) * 2013-07-31 2015-02-11 杭州锂软科技有限公司 SOC-SOH (state of charge-state of health)-based distributed BMS (Battery Management System)
CN107436411A (en) * 2017-07-28 2017-12-05 南京航空航天大学 Battery SOH On-line Estimation methods based on fractional order neural network and dual-volume storage Kalman
CN109061505A (en) * 2018-08-28 2018-12-21 淮阴工学院 A kind of detection method of lithium battery SOH
CN110095732A (en) * 2019-05-23 2019-08-06 合肥工业大学智能制造技术研究院 A kind of lithium battery SOH estimation method for considering ambient humidity and internal resistance being influenced

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Estimation of State-of-Charge and State-of-Health for Lithium-Ion Degraded Battery Considering Side Reactions;Yizhao Gao等;《Journal of The Electrochemical Society》;20181231;第165卷(第16期);4018-4025 *
基于模糊辨识的蓄电池SOH测量方法;唐骏宇等;《电源技术》;20180228;第42卷(第8期);244-246 *
电动汽车动力电池SOH在线实时估计算法研究;汤露曦;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20151015(第10期);I、34-40、50 *

Also Published As

Publication number Publication date
CN110850298A (en) 2020-02-28

Similar Documents

Publication Publication Date Title
CN110850298B (en) Lithium battery SOH estimation method and system based on data driving
Lipu et al. A review of state of health and remaining useful life estimation methods for lithium-ion battery in electric vehicles: Challenges and recommendations
Zhang et al. An on-line estimation of battery pack parameters and state-of-charge using dual filters based on pack model
Yang et al. A novel method on estimating the degradation and state of charge of lithium-ion batteries used for electrical vehicles
CN110568361B (en) Method for predicting health state of power battery
Zhang et al. A novel approach of battery pack state of health estimation using artificial intelligence optimization algorithm
Farmann et al. A comprehensive review of on-board State-of-Available-Power prediction techniques for lithium-ion batteries in electric vehicles
CN113419187B (en) Lithium ion battery health estimation method
CN111007399B (en) Lithium battery state of charge prediction method based on improved generation countermeasure network
Tang et al. Joint estimation of state-of-charge and state-of-health for all cells in the battery pack using “leader-follower” strategy
Zhang et al. Remaining useful life estimation of lithium-ion batteries based on thermal dynamics
Mamo et al. Long short-term memory with attention mechanism for state of charge estimation of lithium-ion batteries
CN108872869B (en) Lithium ion battery degradation classification method based on BP neural network
CN104577242B (en) A kind of batteries management system and method
CN111948560A (en) Lithium battery health state estimation method based on multi-factor evaluation model
CN109001640B (en) Data processing method and device for power battery
CN110658459B (en) Lithium ion battery state of charge estimation method based on bidirectional cyclic neural network
CN110888056B (en) Online SOC observer building method and system suitable for vehicle-mounted power lithium ion battery
CN109839599B (en) Lithium ion battery SOC estimation method based on second-order EKF algorithm
CN113109717B (en) Lithium battery state of charge estimation method based on characteristic curve optimization
CN106777786A (en) A kind of lithium ion battery SOC estimation method
CN111983459A (en) Health state testing and evaluating method based on vehicle lithium ion battery
Chen et al. Parameter identification of equivalent circuit models for Li-ion batteries based on tree seeds algorithm
Chen et al. State of health estimation for lithium-ion battery based on long short term memory networks
Sun et al. Battery state of health estimation method based on sparse auto‐encoder and backward propagation fading diversity among battery cells

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