CN114966407A - Estimation method for lithium battery multi-sensor information fusion state of charge - Google Patents

Estimation method for lithium battery multi-sensor information fusion state of charge Download PDF

Info

Publication number
CN114966407A
CN114966407A CN202210456586.7A CN202210456586A CN114966407A CN 114966407 A CN114966407 A CN 114966407A CN 202210456586 A CN202210456586 A CN 202210456586A CN 114966407 A CN114966407 A CN 114966407A
Authority
CN
China
Prior art keywords
filter
sensors
voltage
lithium battery
sensor
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.)
Pending
Application number
CN202210456586.7A
Other languages
Chinese (zh)
Inventor
吴忠强
徐伟杰
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.)
Yanshan University
Original Assignee
Yanshan 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 Yanshan University filed Critical Yanshan University
Priority to CN202210456586.7A priority Critical patent/CN114966407A/en
Publication of CN114966407A publication Critical patent/CN114966407A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
    • 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]
    • G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
    • 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]
    • G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/70—Energy storage systems for electromobility, e.g. batteries

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Tests Of Electric Status Of Batteries (AREA)

Abstract

The invention discloses a method for estimating the state of charge of multi-sensor information fusion of a lithium battery, which aims at the problem that the traditional method for estimating the SOC of the lithium battery only uses a single sensor and can seriously influence the estimation effect if the sensor fails. And then, an improved D-S evidence theory is introduced to update the fusion information weight of the multiple sensors in real time, so that the defect that the traditional combined extended Kalman filtering algorithm is given by experience and is unchanged when the weight is distributed to the fusion information is overcome. The two improvements are combined, so that the robustness of the estimation algorithm can be greatly improved.

Description

Estimation method for lithium battery multi-sensor information fusion state of charge
Technical Field
The invention relates to a method for estimating the State of Charge (SOC) of a lithium battery, belongs to the technical field of new energy, and particularly relates to a method for estimating the SOC of a lithium battery through multi-sensor information fusion.
Background
The automobile industry is developing vigorously nowadays, and plays an important role in national economy and social development as an important solid economy industry. As a strategic emerging industry, the new energy automobile industry will undoubtedly become the development direction of future automobiles. The lithium battery is one of core components of the electric automobile, is a complex chemical system, needs a battery management system to accurately manage the lithium battery, can provide prediction for the remaining endurance mileage by SOC estimation, and is one of key links in the battery management system.
Currently, many methods have been proposed to estimate the SOC of lithium batteries: such as open circuit voltage method, ampere-hour method and kalman series algorithm. The lithium battery state of charge estimated by the open circuit voltage method is high in accuracy, but the method needs the lithium battery to stand for a long enough time and is not suitable for on-line estimation. The ampere-hour method calculates the charging and discharging current in a certain time and the integral of the corresponding time to obtain the residual electric quantity. The method is simple in calculation and easy to implement, but due to the open-loop property of the method, the integral error cannot be corrected in time, and the algorithm is easy to disperse after long-time operation. Kalman Filter (KF) series algorithms are efficient autoregressive filtering algorithms, and the state of charge estimation of lithium batteries using these algorithms has become one of the mainstream directions. When an Extended Kalman Filter (EKF) algorithm is used for estimation, a nonlinear part in a battery model needs to be linearized, but linearization errors are introduced, so that the estimation accuracy of the state of charge of the lithium battery is influenced. The Unscented Kalman Filter (Unscented Kalman Filter, UKF) algorithm uses Unscented transformation to approximate statistics of the mean and variance of the obtained state, avoids the error caused by linearization, overcomes the limitation of the extended Kalman Filter algorithm, and has high estimation precision. However, the unscented kalman filter algorithms diverge in use if the covariance matrix is not guaranteed to be positive.
Disclosure of Invention
The invention provides a method for estimating the multi-sensor information fusion state of charge of a lithium battery, which is characterized in that parameters of the lithium battery are identified on line by using an FFRLS algorithm, and two links of parameter identification and SOC estimation work on different time scales (firstly, battery parameter estimation is carried out and then, battery SOC estimation is carried out), so that the parameters are continuously corrected by the algorithm in the SOC estimation process, the estimation error caused by the change of the battery parameters is reduced, and the accuracy of the algorithm is improved.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows: a method for estimating the state of charge of multi-sensor information fusion of a lithium battery comprises the following steps:
s1, establishing a second-order RC model of the lithium battery based on the second-order RC model, determining the number of used voltage and current sensors, collecting data of all groups of voltage and current sensors, and performing online identification on parameters of the lithium battery by using a least square method based on forgetting factors;
step S2, sending all the groups of voltage and current sensor data into sub-filters, performing SOC estimation on the sub-filters by adopting a DS-AFEKF algorithm, performing parallel work on each sub-filter to respectively obtain different local optimal values, sending the local optimal values into a main filter, performing information fusion on the local optimal values by the main filter to obtain a global optimal estimation value, and outputting the global optimal estimation value;
step S3, distributing respective weights for each voltage sensor and each current sensor according to the credibility of the voltage sensors and the current sensors, dividing all the voltage sensors and the current sensors into target sensors and evidence sensors during each algorithm iteration, defining basic trust distribution of new evidence according to the relative size relation between the evidence and the target sensors, and updating the fusion weights of all the voltage sensors and the current sensors in the AFEKF algorithm in real time by adopting an improved D-S evidence theory;
s4, outputting the basic trust distribution weight of the new evidence to each sub-filter, and performing the next algorithm iteration;
and S5, when estimating the SOC of the lithium battery, repeating the same optimization and iteration operations as the steps S2-S4 until the iteration is finished, outputting the estimated SOC value, and drawing a change curve of the SOC value.
The technical scheme of the invention is further improved as follows: the expression of the second-order RC model of the lithium battery in step S1 is:
Figure BDA0003618961910000031
Figure BDA0003618961910000032
wherein u is oc Is the open circuit voltage of the battery, u is the terminal voltage of the battery, i is the current in the battery, R S Is the ohmic internal resistance, R, of the cell τ And R L Respectively showing concentration polarization internal resistance and electrochemical polarization internal resistance, C τ And C L Respectively representing the concentration polarization capacitance and the electrochemical polarization capacitance of the cell, u τ And u L Respectively representing concentration polarization voltage and electrochemical polarization voltage of the cell, eta is coulombic efficiency, Q C Rated capacity of battery, t 0 Is the initial time, and t is the current time.
The technical scheme of the invention is further improved as follows: in the step S1, the second-order RC model of the lithium battery is discretized and linearized, voltages of two RC loops and a state of charge of the battery are selected as state variables, and random disturbance is considered, so that a discrete state equation and an observation equation are obtained as follows:
Figure BDA0003618961910000033
wherein X (k) is [ u ] τ (k) u L (k) SOC(k)] T ,U(k)＝i(k)，Z(k)＝u(k)，
Figure BDA0003618961910000034
Figure BDA0003618961910000035
Wherein the partial differential term is formed by H [ SOC (k)]Performing a first order Taylor expansion around SOC (k) to obtain D ═ R S ，T S For the sampling period, W (k) is process noise, V (k) is observation noiseSound, k is time; the noise statistics of the system are described as:
Figure BDA0003618961910000041
where E [. cndot. ] represents the mathematical expectation of the correlated noise, Q (k) is the process noise variance, R (k) is the observed noise variance, and η (k, j) is the Crohn's function, defined as:
Figure BDA0003618961910000042
the technical scheme of the invention is further improved as follows: in step S2, the ith sub-filter is described as:
Figure BDA0003618961910000043
the temporal update and the prediction of the observed quantity of the ith sub-filter and the main filter are as follows:
Figure BDA0003618961910000044
wherein, when i is 1,2, …, M represents a sub-filter, and when i is g represents a main filter;
the measurements of the ith sub-filter and the main filter are updated as:
K i ＝P i (k+1|k)C T (CP i (k+1|k)C T +R i ) -1
Figure BDA0003618961910000051
P i (k+1|k+1)＝P i (k+1|k)-P i (k+1|k)K i C T
the covariance and state assignment formula of the ith sub-filter and the main filter is as follows:
Figure BDA0003618961910000052
Figure BDA0003618961910000053
Figure BDA0003618961910000054
in the formula, Q i Is the observed noise variance, Q, of the sub-filter g Is the observed noise variance, P, of the main filter i Is a covariance matrix of sub-filter systems, P g Is a systematic covariance matrix, beta, of the main filter i The feedback weight after each filtering needs to satisfy the information conservation principle:
Figure BDA0003618961910000055
and beta is not less than 0 i ≤1；
Summarizing the local optimal value of each filter to a main filter for information fusion, wherein the fusion method comprises the following steps:
Figure BDA0003618961910000056
the global optimum state update is given by:
Figure BDA0003618961910000057
the global state covariance matrix can be obtained from the following equation:
Figure BDA0003618961910000058
the technical scheme of the invention is further improved as follows: in the step S1 and the step S2, the FFRLS online identification system parameters and the SOC estimation are alternately performed to obtain the adaptive joint extended kalman filter algorithm, where the specific formula of the adaptive joint extended kalman filter algorithm is as follows:
Figure BDA0003618961910000061
in the formula (I), the compound is shown in the specification,
Figure BDA0003618961910000062
for the parameter variables to be identified, P (k) is the estimated covariance matrix of the system, h (k) is the gain matrix of the system,
Figure BDA0003618961910000063
is a system data variable, n 1 ，n 2 Is positive number, lambda is introduced forgetting factor, and the value interval is 0.95-1.
The technical scheme of the invention is further improved as follows: in the step S3, the assignment process in which the credibility of each voltage and current sensor assigns their respective weights is a basic confidence assignment, and the formula is as follows:
Figure BDA0003618961910000064
where m (-) is the basic confidence distribution function on the recognition framework Θ, and α, which makes m (α) >0, is called the focal element, φ is the empty set.
The technical scheme of the invention is further improved as follows: the process of improving the D-S evidence theory and updating the fusion weights of all the voltage and current sensors in the AFEKF algorithm in real time in step S3 includes:
at a certain moment, the set of voltage sensor measurements of each sub-filter is t:
Τ＝[Z 1 (k),…,Z i (k),…,Z M (k)]，
1) randomly selecting one of the sensors Z i (k) For a target sensor set T o Then all remaining sensors constitute evidence sensorsCollection T m ：
Figure BDA0003618961910000065
2) Defining the magnitude between the target sensor and the evidence sensor measurements as d, then the magnitude between the target sensor and each evidence sensor measurement is:
d j ＝|Z i (k)-Z j (k)|，j＝1,2,…,M,j≠i
3) obtaining a basic confidence distribution function of each voltage sensor corresponding to the sub-filter:
Figure BDA0003618961910000071
in the formula, alpha i I is 1,2, …, M, i is the ith focal element;
4) fusing all evidences to obtain the distribution weight of each voltage sensor corresponding to the sub-filter:
Figure BDA0003618961910000072
in the formula (I), the compound is shown in the specification,
Figure BDA0003618961910000073
for an average trust distribution of all evidence, β i Is the weight of probability distribution, gamma represents the collision coefficient between evidences, and its value represents the correlation degree between evidences.
Due to the adoption of the technical scheme, the invention has the technical progress that:
1. the battery parameter is identified on line by using the FFRLS algorithm, and two links of parameter identification and SOC estimation work on different time scales (firstly, battery parameter estimation is carried out, and then, the SOC estimation of the battery is carried out), so that the parameters are continuously corrected by the algorithm in the SOC estimation process, the estimation error caused by the parameter change of the battery is reduced, and the accuracy of the algorithm is improved;
2. the invention creates a new basic trust distribution process according to the technical field to be used when improving the D-S evidence theory, and specifically comprises the following steps: (1) the voltage sensors are randomly assigned as evidence sensors and target sensors. The randomness is ensured, and the condition that the estimation precision is influenced if a single target sensor is set and the target sensor is damaged is avoided; (2) the characteristics of the sensor in the battery field are fully combined: because the data of the sensors are consistent, the measurement results of the sensors can be mutually verified, and therefore, the sensors can be assigned with respective weights according to the credibility of the sensors. If an abnormal value occurs successively in a certain sensor, the sensor is not considered to have high credibility. Considering that the measured value of a voltage sensor may fluctuate above and below the true value if the voltage sensor fails, dividing all sensors into a target sensor and an evidence sensor during each algorithm iteration, and defining the basic trust distribution of new evidence according to the relative size relationship between the evidence and the target sensor;
3. the invention introduces an improved D-S evidence theory to dynamically distribute the weight value of each sub-filter, thereby reducing the influence on the estimation result if the sensor fails. Under the condition that the voltage sensor cannot be replaced in time, the SOC estimation method provided by the invention still has better estimation accuracy and stronger robustness compared with other SOC estimation algorithms.
Drawings
FIG. 1 is a diagram of the algorithm architecture of the present invention;
FIG. 2 is a second order RC model of a lithium battery of the present invention;
FIG. 3 is a diagram of the parameter identification result of the present invention;
FIG. 4 is a diagram of the estimation results of the estimation method of the present invention and other methods;
FIG. 5 is a graph of the present invention's estimation results with other methods, assuming sensor damage.
Detailed Description
The invention aims to provide a multi-sensor information fusion SOC estimation method, which aims to solve the problem that the estimation effect is seriously influenced if a sensor fails due to the fact that only a single sensor is used in the traditional lithium battery SOC estimation method. Firstly, a FeKF (Federal Extended Kalman Filter) algorithm is improved, a self-adaptive function is introduced, and a least square method with a forgetting factor is used for identifying lithium battery parameters, so that the parameters are continuously corrected in the process of estimating the SOC of the lithium battery by the algorithm, the estimation error caused by the change of the battery parameters is reduced, and the accuracy is improved. And then, an improved D-S evidence theory is introduced to update the fusion information weight of the multiple sensors in real time, so that the defect that the traditional combined extended Kalman filtering algorithm is given by experience and is unchanged when the weight is distributed to the fusion information is overcome. The two improvements are combined, so that the robustness of the estimation algorithm can be greatly improved.
In order to achieve the purpose, the technical scheme of the invention is as follows:
as shown in fig. 1, a method for estimating a state of charge of a lithium battery by multi-sensor information fusion includes the following steps:
s1, establishing a second-order RC model of the lithium battery based on the second-order RC model, determining the number of used voltage and current sensors, collecting data of all groups of voltage and current sensors, and performing online identification on parameters of the lithium battery by using a least square method based on forgetting factors;
the second-order RC model of the lithium battery is shown in FIG. 2, and the expression is as follows:
Figure BDA0003618961910000091
Figure BDA0003618961910000092
wherein u is oc Is the open circuit voltage of the battery, u is the terminal voltage of the battery, i is the current in the battery, R S Is the ohmic internal resistance, R, of the cell τ And R L Respectively showing concentration polarization internal resistance and electrochemical polarization internal resistance, C τ And C L Respectively representing the concentration polarization capacitance and the electrochemical polarization capacitance of the cell, u τ And u L Respectively representing concentration polarization voltage and electrochemical polarization voltage of the cell, eta is coulombic efficiency, Q C Rated capacity of battery, t 0 Is the initial time, and t is the current time.
The second-order RC model of the lithium battery is subjected to discretization and linearization, the voltages of two RC loops and the state of charge of the battery are selected as state variables, random disturbance is considered, and a discretization state equation and an observation equation are obtained as follows:
Figure BDA0003618961910000093
wherein X (k) is [ u ] τ (k) u L (k) SOC(k)] T ,U(k)＝i(k)，Z(k)＝u(k)，
Figure BDA0003618961910000101
Figure BDA0003618961910000102
Wherein partial derivative term is represented by H [ SOC (k)]Performing a first order Taylor expansion around SOC (k) to obtain D ═ R S ，T S For the sampling period, w (k) is process noise, v (k) is observation noise, and k is time; the noise statistics of the system are described as:
Figure BDA0003618961910000103
where E [. cndot. ] represents the mathematical expectation of the correlated noise, Q (k) is the process noise variance, R (k) is the observed noise variance, and η (k, j) is the Crohn's function, defined as:
Figure BDA0003618961910000104
and S2, sending all the groups of voltage and current sensor data into sub-filters, performing SOC estimation on the sub-filters by adopting a DS-AFEKF algorithm, enabling each sub-filter to work in parallel to respectively obtain different local optimal values, sending the local optimal values into a main filter, and performing information fusion on the local optimal values by the main filter to obtain a global optimal estimation value and outputting the global optimal estimation value. And each sub-filter distributes the weight according to different shared information, distributes the updated variance of the main filter and is used for the next operation. The method of firstly blocking and then carrying out global fusion utilizes the available data to the maximum extent so as to improve the accuracy of final estimation.
For a lithium battery system mathematically modeled according to the present invention, the ith sub-filter can be described as:
Figure BDA0003618961910000111
the temporal update and the prediction of the observed quantity of the ith sub-filter and the main filter are as follows:
Figure BDA0003618961910000112
wherein, when i is 1,2, …, M represents a sub-filter, and when i is g represents a main filter;
the measurements of the ith sub-filter and the main filter are updated as:
K i ＝P i (k+1|k)C T (CP i (k+1|k)C T +R i ) -1
Figure BDA0003618961910000113
P i (k+1|k+1)＝P i (k+1|k)-P i (k+1|k)K i C T
the covariance and state assignment formula of the ith sub-filter and the main filter is:
Figure BDA0003618961910000114
Figure BDA0003618961910000115
Figure BDA0003618961910000116
in the formula, Q i Is the observed noise variance, Q, of the sub-filter g Is the observed noise variance, P, of the main filter i Is a covariance matrix of sub-filter systems, P g Is a systematic covariance matrix, beta, of the main filter i The feedback weight after each filtering needs to satisfy the information conservation principle:
Figure BDA0003618961910000121
and 0 is not less than beta i ≤1。
Summarizing the local optimal value of each filter to a main filter for information fusion, wherein the fusion method comprises the following steps:
Figure BDA0003618961910000122
the global optimum state update is given by:
Figure BDA0003618961910000123
the global state covariance matrix can be obtained from the following equation:
Figure BDA0003618961910000124
as shown in fig. 3, a result of online identification of parameters of a lithium battery by using the least square method with forgetting factor of the present invention is shown. Considering that the battery parameters can change along with the aging of the battery, the invention alternately carries out FFRLS on-line identification system parameters and SOC estimation to obtain the self-adaptive combined extended Kalman filtering algorithm, and the specific formula of the self-adaptive combined extended Kalman filtering algorithm is as follows:
Figure BDA0003618961910000125
in the formula (I), the compound is shown in the specification,
Figure BDA0003618961910000126
for the parameter variables to be identified, P (k) is the estimated covariance matrix of the system, h (k) is the gain matrix of the system,
Figure BDA0003618961910000127
is a system data variable, n 1 ，n 2 Is positive number, lambda is introduced forgetting factor, and the value interval is 0.95-1.
Step S3, distributing respective weight values to each voltage sensor and each current sensor according to the credibility of the voltage sensors and the current sensors, considering that the measured values of the voltage sensors may fluctuate above and below the true values if the voltage sensors fail, dividing all the voltage sensors and the current sensors into target sensors and evidence sensors during each algorithm iteration, defining the basic trust distribution of new evidence according to the relative size relationship between the evidence and the target sensors, and updating the fusion weight values of all the voltage sensors and the current sensors in the AFE KF algorithm in real time by adopting an improved D-S evidence theory;
weight beta distributed by FEKF algorithm i Is generally given empirically and is fixed, and each beta i The value has a large influence on the estimation result. The improved D-S evidence theory of the invention updates the weight value of information distribution in real time. The D-S evidence theory is an important theory capable of processing multi-source information, and the biggest characteristic is that the description of uncertainty information adopts 'interval estimation' rather than 'point estimation'. Great flexibility is shown in distinguishing between unknown and uncertain aspects and in accurately reflecting evidence collection. For objective evidence and subjective evaluation, results from either party should not be biased in the collection of information, but their importance should be noted. Can charge by D-S theoryThe uncertainty of both results is retained, for all possible sets of uncertainty, called the recognition framework in theory, denoted by Θ. In which the internal elements are mutually exclusive, and the combination of all possible problems is 2 Θ The D-S theory assigns probabilities to the information, namely, the capability of reserving and processing multi-source information is possessed. The Assignment process is called Basic Belief Assignment (BBA), generally called mass function, and in the power set of the recognition framework is:
Figure BDA0003618961910000131
where m (-) is the basic confidence distribution function on the recognition framework Θ, and α, which makes m (α) >0, is called the focal element, φ is the empty set.
The process of improving the D-S evidence theory to update the fusion weights of all the voltage and current sensors in the AFEKF algorithm in real time comprises the following steps:
because the data of the sensors are consistent, the measurement results of the sensors can be mutually verified, and therefore, the sensors can be assigned with respective weights according to the credibility of the sensors. If an abnormal value occurs successively in a certain sensor, the sensor is not considered to have high credibility. Considering that the measured value of the voltage sensor may fluctuate above and below the true value if the voltage sensor fails, all sensors are divided into a target sensor and an evidence sensor during each algorithm iteration, and the basic trust distribution of new evidence can be defined according to the relative size relationship between the evidence and the target sensor. The weight adjustment based on the D-S theory is as follows:
at a certain moment, the set of voltage sensor measurements of each sub-filter is t:
Τ＝[Z 1 (k),…,Z i (k),…,Z M (k)]，
1) randomly selecting one of the sensors Z i (k) For a target sensor set T o Then all remaining sensors make up the evidence sensor set T m ：
Figure BDA0003618961910000141
2) Defining the magnitude between the target sensor and the evidence sensor measurements as d, then the magnitude between the target sensor and each evidence sensor measurement is:
d j ＝|Z i (k)-Z j (k)|，j＝1,2,…,M,j≠i
3) obtaining a basic confidence distribution function of each voltage sensor corresponding to the sub-filter:
Figure BDA0003618961910000142
in the formula, alpha i I is 1,2, …, M, i is the ith focal element;
4) fusing all evidences to obtain the distribution weight of each voltage sensor corresponding to the sub-filter:
Figure BDA0003618961910000143
in the formula (I), the compound is shown in the specification,
Figure BDA0003618961910000144
an average trust distribution for all evidences. Beta is a i Are the weights of the probability assignments. Gamma represents a conflict coefficient between evidences, the value of the conflict coefficient represents the correlation degree between the evidences, and the greater the conflict value is, the greater the dissimilarity between the measured evidences is; conversely, it indicates that the greater the correlation between the measured evidences. Therefore, the data fusion of the measuring evidence can be modified according to the actual situation, so as to obtain the fused weight beta i 。
S4, outputting the basic trust distribution weight of the new evidence to each sub-filter, and performing the next algorithm iteration;
and S5, when estimating the SOC of the lithium battery, repeating the same optimization and iteration operations as the steps S2-S4 until the iteration is finished, outputting the estimated SOC value, and drawing a change curve of the SOC value.
As shown in fig. 4, in order to show the advantages of the algorithm, the AUKF algorithm (only 1 set of sensors, i.e., one voltage and current sensor) and the FEKF algorithm (3 sets of sensors are used) were chosen for comparison. The initial value of the SOC is set to 0.9 collectively. Wherein, the estimation error of the DS-AFEKF algorithm is minimum and is within 1.12 percent. The estimation error of the FEKF algorithm is second only to DS-AFKF and is within 1.94%. The estimation error of AUKF is maximum and is within 2.11%. Compared with AUKF and FEKF, the DS-AFEKF algorithm has better estimation accuracy due to the adoption of the optimal fusion criterion of improved D-S weighting.
As shown in fig. 5, assume that 1 of the 3 voltage sensors failed, i.e., a perturbation is added to the voltage data it collects (constant value reduced by 0.12V). When 1 voltage sensor fails, the estimation error of the DS-AFEKF algorithm is minimum and within 5.87%, the estimation error of the FEKF algorithm is second and within 9.13%, and the estimation error of the AUKF algorithm is maximum and within 16.59%. It can be seen that, in actual operation, if a sensor fails, the deviation between the measured data and the actual value is large, and only one sensor has a large influence on the result (for example, the error of the estimated result of the AUKF algorithm is large). And the DS-AFKF and FEKF algorithms have 3 sensors, and have better anti-interference capability compared with the AUKF algorithm only using one filter. Meanwhile, the DS-AFEKF algorithm is based on an improved D-S theory, more reasonable weight can be dynamically distributed, and compared with the FEKF algorithm distributed by fixed weight, the method has higher estimation precision.

Claims (7)

1. A method for estimating the state of charge of multi-sensor information fusion of a lithium battery is characterized by comprising the following steps: the method comprises the following steps:
s1, establishing a second-order RC model of the lithium battery based on the second-order RC model, determining the number of used voltage and current sensors, collecting data of all groups of voltage and current sensors, and performing online identification on parameters of the lithium battery by using a least square method based on forgetting factors;
step S2, sending all groups of voltage and current sensor data into sub-filters, performing SOC estimation on the sub-filters by adopting a DS-AFEKF algorithm, enabling each sub-filter to work in parallel to respectively obtain different local optimal values, sending the local optimal values into a main filter, and performing information fusion on the local optimal values by the main filter to obtain a global optimal estimation value and outputting the global optimal estimation value;
step S3, distributing respective weights for each voltage sensor and each current sensor according to the credibility of the voltage sensors and the current sensors, dividing all the voltage sensors and the current sensors into target sensors and evidence sensors during each algorithm iteration, defining basic trust distribution of new evidence according to the relative size relation between the evidence and the target sensors, and updating the fusion weights of all the voltage sensors and the current sensors in the AFEKF algorithm in real time by adopting an improved D-S evidence theory;
s4, outputting the basic trust distribution weight of the new evidence to each sub-filter, and performing the next algorithm iteration;
and step S5, when estimating the SOC of the lithium battery, repeatedly carrying out the same optimization and iteration operations as the steps S2-S4 until the iteration is finished, outputting the estimated SOC value, and drawing a change curve of the SOC value.
2. The method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 1, wherein the method comprises the following steps: the expression of the second-order RC model of the lithium battery in step S1 is:
Figure FDA0003618961900000011
Figure FDA0003618961900000021
wherein u is oc Is the open circuit voltage of the battery, u is the terminal voltage of the battery, i is the current in the battery, R S Is the ohmic internal resistance, R, of the battery τ And R L Respectively showing concentration polarization internal resistance and electrochemical polarization internal resistance, C τ And C L Respectively representing the concentration polarization capacitance and the electrochemical polarization capacitance of the cell, u τ And u L Respectively representing concentration polarization voltage and electrochemical property of the cellChemical polarization voltage, η coulombic efficiency, Q C Rated capacity of battery, t 0 Is the initial time, and t is the current time.
3. The method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 2, wherein the method comprises the following steps: in the step S1, the second-order RC model of the lithium battery is discretized and linearized, voltages of two RC loops and a state of charge of the battery are selected as state variables, and random disturbance is considered, so as to obtain a discretized state equation and an observation equation as follows:
Figure FDA0003618961900000022
wherein X (k) is [ u ] τ (k) u L (k) SOC(k)] T ,U(k)＝i(k)，Z(k)＝u(k)，
Figure FDA0003618961900000023
Figure FDA0003618961900000024
Wherein the partial differential term is formed by H [ SOC (k)]Performing a first order Taylor expansion around SOC (k) to obtain D ═ R S ，T S For the sampling period, w (k) is process noise, v (k) is observation noise, and k is time; the noise statistics of the system are described as:
Figure FDA0003618961900000031
where E [. cndot. ] represents the mathematical expectation of the correlated noise, Q (k) is the process noise variance, R (k) is the observed noise variance, and η (k, j) is the Crohn's function, defined as:
Figure FDA0003618961900000032
4. the method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 3, wherein the method comprises the following steps: in step S2, the ith sub-filter is described as:
Figure FDA0003618961900000033
the temporal update and the prediction of the observed quantity of the ith sub-filter and the main filter are as follows:
Figure FDA0003618961900000034
wherein, when i is 1,2, …, M represents a sub-filter, and when i is g represents a main filter;
the measurements of the ith sub-filter and the main filter are updated as:
K i ＝P i (k+1|k)C T (CP i (k+1|k)C T +R i ) -1
Figure FDA0003618961900000035
P i (k+1|k+1)＝P i (k+1|k)-P i (k+1|k)K i C T
the covariance and state assignment formula of the ith sub-filter and the main filter is as follows:
Figure FDA0003618961900000041
Figure FDA0003618961900000042
Figure FDA0003618961900000043
in the formula, Q i Is the observed noise variance, Q, of the sub-filter g Is the observed noise variance, P, of the main filter i Is a covariance matrix of sub-filter systems, P g Is a systematic covariance matrix, beta, of the main filter i The feedback weight after each filtering needs to satisfy the information conservation principle:
Figure FDA0003618961900000044
and 0 is not less than beta i ≤1；
Summarizing the local optimal value of each filter to a main filter for information fusion, wherein the fusion method comprises the following steps:
Figure FDA0003618961900000045
the global optimum state update is given by:
Figure FDA0003618961900000046
the global state covariance matrix can be obtained from the following equation:
Figure FDA0003618961900000047
5. the method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 4, wherein the method comprises the following steps: in the step S1 and the step S2, the FFRLS online identification system parameters and the SOC estimation are alternately performed to obtain the adaptive joint extended kalman filter algorithm, where the specific formula of the adaptive joint extended kalman filter algorithm is as follows:
Figure FDA0003618961900000051
in the formula (I), the compound is shown in the specification,
Figure FDA0003618961900000052
for the parameter variables to be identified, P (k) is the estimated covariance matrix of the system, h (k) is the gain matrix of the system,
Figure FDA0003618961900000053
is a system data variable, n 1 ，n 2 Is positive number, lambda is introduced forgetting factor, and the value interval is 0.95-1.
6. The method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 5, wherein the method comprises the following steps: in the step S3, the assignment process in which the credibility of each voltage and current sensor assigns their respective weights is a basic confidence assignment, and the formula is as follows:
Figure FDA0003618961900000054
where m (-) is a basic confidence assignment function on the recognition framework Θ, and α such that m (α) >0 is called a focal element and φ is an empty set.
7. The method for estimating the information fusion state of charge of the multiple sensors of the lithium battery according to claim 6, wherein the method comprises the following steps: the process of improving the D-S evidence theory and updating the fusion weights of all the voltage and current sensors in the AFEKF algorithm in real time in step S3 includes:
at a certain moment, the set of voltage sensor measurements of each sub-filter is t:
Τ＝[Z 1 (k),...,Z i (k),...,Z M (k)]，
1) randomly selecting one of the sensors Z i (k) For a target sensor set T o Then all remaining sensors make up the evidence sensor set T m ：
Figure FDA0003618961900000061
2) Defining the magnitude between the target sensor and the evidence sensor measurements as d, then the magnitude between the target sensor and each evidence sensor measurement is:
d j ＝|Z i (k)-Z j (k)|，j＝1,2,…,M,j≠i；
3) obtaining a basic confidence distribution function of each voltage sensor corresponding to the sub-filter:
Figure FDA0003618961900000062
in the formula, alpha i I is 1,2, …, M, i is the ith focal element;
4) fusing all evidences to obtain the distribution weight of each voltage sensor corresponding to the sub-filter:
Figure FDA0003618961900000063
in the formula (I), the compound is shown in the specification,
Figure FDA0003618961900000064
for an average trust distribution of all evidence, β i Is the weight of probability distribution, gamma represents the collision coefficient between evidences, and its value represents the correlation degree between evidences.
CN202210456586.7A 2022-04-27 2022-04-27 Estimation method for lithium battery multi-sensor information fusion state of charge Pending CN114966407A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210456586.7A CN114966407A (en) 2022-04-27 2022-04-27 Estimation method for lithium battery multi-sensor information fusion state of charge

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210456586.7A CN114966407A (en) 2022-04-27 2022-04-27 Estimation method for lithium battery multi-sensor information fusion state of charge

Publications (1)

Publication Number Publication Date
CN114966407A true CN114966407A (en) 2022-08-30

Family

ID=82978346

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210456586.7A Pending CN114966407A (en) 2022-04-27 2022-04-27 Estimation method for lithium battery multi-sensor information fusion state of charge

Country Status (1)

Country Link
CN (1) CN114966407A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115392139A (en) * 2022-10-27 2022-11-25 西北工业大学 Fuel cell parameter identification method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115392139A (en) * 2022-10-27 2022-11-25 西北工业大学 Fuel cell parameter identification method

Similar Documents

Publication Publication Date Title
CN107957562B (en) Online prediction method for residual life of lithium ion battery
Pan et al. State of charge estimation of lithium-ion batteries using a grey extended Kalman filter and a novel open-circuit voltage model
Ouyang et al. An online prediction of capacity and remaining useful life of lithium-ion batteries based on simultaneous input and state estimation algorithm
CN111505506A (en) Battery SOC estimation method integrating multi-scale Kalman filtering and unscented Kalman filtering
Li et al. Review of lithium-ion battery state of charge estimation
CN104714188A (en) Method and system for estimating measured noise variance matrix matched battery state of charge (SOC)
CN111426957A (en) SOC estimation optimization method for power battery under simulated vehicle working condition
Wang et al. Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile
CN113191547B (en) New energy vehicle power battery charging optimization method and system
CN112269133B (en) SOC estimation method based on pre-charging circuit model parameter identification
CN113554200B (en) Method, system and equipment for predicting voltage inconsistency of power battery
Chen et al. State of health estimation for lithium-ion battery based on long short term memory networks
He et al. A novel adaptive particle swarm optimization algorithm based high precision parameter identification and state estimation of lithium-ion battery
CN114966407A (en) Estimation method for lithium battery multi-sensor information fusion state of charge
CN115754741A (en) SOC estimation method and device of energy storage system, equipment and storage medium
CN116125278A (en) Lithium battery SOC estimation method and system based on LSTM-EKF algorithm
Zhao et al. Digital twin-driven estimation of state of charge for Li-ion battery
Zhang et al. Interpretable battery cycle life range prediction using early cell degradation data
Zhou et al. Data-driven battery state-of-health estimation and prediction using IC based features and coupled model
CN117665620A (en) New energy automobile data-based battery health evaluation method
CN115308611B (en) Lithium ion battery residual life prediction method considering temperature compensation
Meng et al. Multi-segment state of health estimation of lithium-ion batteries considering short partial charging
CN114460471A (en) Power battery multi-algorithm fusion SOC estimation method based on Bayesian probability
CN115389954A (en) Battery capacity estimation method, electronic equipment and readable storage medium
CN113935222A (en) Power battery multi-model fusion estimation method based on ordered weighted average operator

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