CN112816884A - Method, device and equipment for monitoring health state of satellite lithium ion battery - Google Patents
Method, device and equipment for monitoring health state of satellite lithium ion battery Download PDFInfo
- Publication number
- CN112816884A CN112816884A CN202110225533.XA CN202110225533A CN112816884A CN 112816884 A CN112816884 A CN 112816884A CN 202110225533 A CN202110225533 A CN 202110225533A CN 112816884 A CN112816884 A CN 112816884A
- Authority
- CN
- China
- Prior art keywords
- ion battery
- lithium ion
- neural network
- self
- input
- 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.)
- Granted
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]
- G01R31/3644—Constructional arrangements
- G01R31/3648—Constructional arrangements comprising digital calculation means, e.g. for performing an algorithm
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
-
- 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
- G06N3/08—Learning methods
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (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)
- Charge And Discharge Circuits For Batteries Or The Like (AREA)
- Secondary Cells (AREA)
Abstract
The application relates to a method, a device and equipment for monitoring the health state of a satellite lithium ion battery, wherein the method comprises the following steps: acquiring real-time monitoring data of the satellite lithium ion battery; inputting real-time monitoring data into a self-organizing mapping neural network obtained by training, and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery; calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery. Through the technical scheme, the purpose of extracting the CV virtual index from the parameters of the satellite lithium ion battery by adopting the self-organizing mapping neural network is achieved, and the technical effect of carrying out high-accuracy monitoring on the health state of the satellite lithium ion battery is achieved.
Description
Technical Field
The application relates to the technical field of satellite measurement and control, in particular to a method, a device and equipment for monitoring the health state of a satellite lithium ion battery.
Background
In the whole satellite operation process, the battery plays a very crucial role in the whole satellite system and is a core single-machine component for the whole system operation. The battery health management of the satellite is the basis for guaranteeing the normal work of the satellite. The selection of an appropriate Health Indicator (HI) is a prerequisite for an accurate assessment of the state of health of the battery. Currently, a lithium ion battery is generally used in a third-generation satellite energy storage battery, and in the field of Health assessment of lithium ion batteries, the battery capacity, the State of Charge (SOC) value and the State of Health (SOH) value of the battery are generally used in the industry to comprehensively measure the Health condition of the battery.
However, in the process of implementing the present invention, the inventor finds that for the three common lithium ion battery health indexes, when the satellite lithium ion battery works, firstly, the working conditions of the lithium ion battery in a laboratory cannot be achieved, for example, the battery is not completely charged and discharged in the whole charging and discharging period, and the constant current and constant voltage conditions cannot be guaranteed; secondly, on-line monitoring cannot acquire a plurality of important health characteristic parameters of the battery, which also leads to the fact that health evaluation indexes such as SOC and SOH which are commonly used for the lithium ion battery are not under the control of a satellite lithium ion battery. Therefore, most of the existing documents select a single parameter as a health index, or select data features extracted from several parameters as the health index, and have the technical problem of poor health state monitoring accuracy of the satellite lithium ion battery.
Disclosure of Invention
In view of the above, it is necessary to provide a method for monitoring a health state of a satellite lithium ion battery, a device for monitoring a health state of a satellite lithium ion battery, a computer device, and a computer-readable storage medium, which can implement high-accuracy monitoring of a health state of a satellite lithium ion battery.
In order to achieve the above purpose, the embodiment of the invention adopts the following technical scheme:
on one hand, the embodiment of the invention provides a method for monitoring the health state of a satellite lithium ion battery, which comprises the following steps:
acquiring real-time monitoring data of the satellite lithium ion battery;
inputting real-time monitoring data into a self-organizing mapping neural network obtained by training, and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery;
calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
On the other hand, still provide a satellite lithium ion battery's health status monitoring devices, include:
the data acquisition module is used for acquiring real-time monitoring data of the satellite lithium ion battery;
the winning determining module is used for inputting the real-time monitoring data into the self-organizing mapping neural network obtained by training and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery;
the index calculation module is used for calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
In another aspect, a computer device is further provided, which includes a memory and a processor, where the memory stores a computer program, and the processor implements the steps of the above method for monitoring the health status of a satellite lithium ion battery when executing the computer program.
In still another aspect, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program, when executed by a processor, implements the steps of the above-mentioned method for monitoring the state of health of a satellite lithium ion battery.
One of the above technical solutions has the following advantages and beneficial effects:
according to the method, the device and the equipment for monitoring the health state of the satellite lithium ion battery, the self-organization mapping neural network obtained by training based on the historical monitoring data of the satellite lithium ion battery is adopted, the newly obtained real-time monitoring data of the satellite lithium ion battery is input into the neural network, so that the winning neuron corresponding to the real-time monitoring data is determined, and then the CV virtual index indicating the health state of the satellite lithium ion battery is obtained through calculation by utilizing the set parameters and the dot product between each input neuron and the winning neuron of the self-organization mapping neural network. The purpose of extracting the CV virtual index from the parameters of the satellite lithium ion battery by adopting the self-organizing mapping neural network is achieved, and the technical effect of monitoring the health state of the satellite lithium ion battery with high accuracy is achieved.
Drawings
FIG. 1 is a schematic flow chart illustrating a method for monitoring the state of health of a satellite lithium ion battery according to an embodiment;
FIG. 2 is a schematic flow chart illustrating a configuration of an embodiment of a self-organizing map neural network before training;
FIG. 3 is a schematic diagram illustrating a training process of the SOMO neural network in one embodiment;
FIG. 4 is a graphical illustration of a lithium ion battery degradation trend in one embodiment;
FIG. 5 is a schematic diagram showing CV virtual indexes of lithium ion battery No. 6 in one embodiment;
FIG. 6 is a diagram illustrating a comparison between the virtual index of CV and the index SOH of the lithium ion battery No. 5 in one embodiment;
FIG. 7 is a diagram illustrating a comparison between the virtual index of CV and the index SOH of the lithium ion battery No. 6 in one embodiment;
FIG. 8 is a diagram illustrating a comparison between the virtual index of CV and the index SOH of the lithium ion battery No. 7 in one embodiment;
fig. 9 is a schematic block diagram of a health status monitoring device of a satellite lithium ion battery according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
Currently, a lithium ion battery is generally used in a third-generation satellite energy storage battery, and in the field of Health assessment of lithium ion batteries, the battery capacity, the State of Charge (SOC) value and the State of Health (SOH) value of the battery are generally used in the industry to comprehensively measure the Health condition of the battery. Reference may be made to the existing literature: "Zhangyang, lithium ion battery online residual life prediction method based on correlation vector machine research [ D ]. national defense science and technology university, 2016"; "Gunn, king friend. storage battery SOH estimation method research reviews [ J ]. machinery manufacturing and automation, 2019,48(01): 204-.
Battery capacity: in state of health monitoring, the monitoring system is concerned with the actual capacity of the battery, and it is generally considered that the failure threshold is 70% to 80% of the battery capacity. The charge capacity of the battery can be divided into two parts, CC charge Capacity (CQ)cc) And CV charging Capacity (CQ)cv) They can be represented as:
wherein t is0And tccRespectively representing the starting time and the ending time of constant-current charging; i represents a current variable; i isccA current value representing constant current charging; t is tcvIndicating the termination time of the constant voltage charging. The above formula can be used in laboratory degradation experiments to estimate battery capacity.
State of health (SOH): the state of health refers to the deviation of the current main performance of the storage battery from the rated design performance and indexes, such as the capacity of a single battery in a fully charged state, the internal resistance, the terminal voltage of heavy current discharge, the capacity of the battery during charging, different temperature characteristics and the like, and represents the performance state of the battery, which is defined by the following formula:
wherein, CFruit of Chinese wolfberryRepresenting the actual capacity during use, CForehead (forehead)Indicating rated capacity, CEOLThe representation is an artificially defined failure threshold during use.
State of charge (SOC): the relative value of the remaining capacity of the battery, i.e., the state of charge, is generally expressed as SOC, and refers to the ratio of the amount of electricity that the battery can still discharge to the rated capacity.
CremainIndicates the remaining capacity of the battery, CratedIndicating the rated capacity of the battery.
In implementing the present invention, the inventor finds that a CV (confidence value) virtual index is a health index based on data, and the basic principle of the CV virtual index is to measure the health state of a device by calculating the distance between the current state of the device and a specified standard state through a certain method. The CV virtual index achieves a good effect on bearing parts [ reference: hong S Zhou Z Zio E et al
35: 117-. The inventor takes the above as a inspiration, considers that the CV virtual index can be based on the original monitoring data, and adopts a neural network method to convert each parameter reflecting the health condition of the lithium ion battery into a Confidence Value (CV), so that the index is used as the health index of the satellite lithium ion battery, a brand new technical scheme capable of accurately acquiring the health condition of the satellite lithium ion battery according to real-time monitoring data is obtained, and the method has great significance for solving the practical engineering problem.
In order to solve the technical problem of poor health state monitoring accuracy of a satellite lithium ion battery in the conventional monitoring technology, the embodiment of the invention provides the following technical scheme:
referring to fig. 1, in an embodiment, the present invention provides a method for monitoring a state of health of a satellite lithium ion battery, including the following steps S12 to S16:
and S12, acquiring real-time monitoring data of the satellite lithium ion battery.
It is understood that the real-time monitoring data of the satellite lithium ion battery can be, but is not limited to, obtained directly by receiving telemetry data transmitted back by the satellite and extracting the telemetry data, or obtained by direct input of staff, or obtained by request from a networked database server. The real-time monitoring data may be newly acquired monitoring data (as compared to historical monitoring data for the satellite) as of the current time.
S14, inputting the real-time monitoring data into the self-organizing mapping neural network obtained by training, and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery.
It can be understood that, in the embodiment, a Self-organizing mapping neural network (i.e., (Self-organizing mapping Maps, SOM)) neural network is used to extract a CV virtual index from battery parameters of a satellite lithium ion battery. The historical monitoring data of the satellite lithium ion battery refers to monitoring data which is earlier than the real-time monitoring data in time, and can refer to monitoring data when the health state of the satellite lithium ion battery is normal. The self-organizing mapping neural network obtained by training may be obtained by performing pre-training or in-situ training by using the historical monitoring data of the satellite lithium ion battery, and may also be obtained by receiving from a database server or a platform that has been trained to obtain the SOM network, which is not specifically limited in this embodiment.
Specifically, in the trained SOM network, the obtained real-time monitoring data is input into the SOM network, after the search process of the input layer neurons and the output layer neurons of the SOM network, dot product data of the weight vectors of the output neurons and the input neurons corresponding to the input data is obtained, and finally required winning neurons can be determined according to the maximum value of the dot product values from the obtained dot product data.
S16, calculating to obtain CV virtual indexes according to dot products and set parameters between input neurons and winning neurons of the self-organizing mapping neural network; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
It can be understood that after the corresponding winning neuron is determined, the dot product between each input neuron of the SOM network and the winning neuron can be obtained, and the CV virtual index can be calculated by using the setting parameters according to the dot product:
wherein c represents the aforementioned setting parameter, gYBRepresenting the dot product, CV, between each input neuron and the winning neuronYBAnd a value representing a virtual index of CV corresponding to the real-time monitoring data. The set parameter is an empirical value parameter, because the value of the CV virtual index is a value between 0 and 1, the more approaching 1 indicates that the state is more normal, the more approaching 0 indicates that the state is worse, the value of the set parameter can conveniently adjust the value range of the value of the CV virtual index, the size of the set parameter is associated with the confidence coefficient CV and the division of the state interval, and particularly, the value of the CV virtual index can be determined according to gYBThe value is set to a specific value. The CV virtual index reflects the difference between the real-time monitoring data of the satellite lithium ion battery and the standard mode, and the larger the difference is, the larger the table isShowing that the health status of the satellite lithium ion battery is worse. The standard patterns are a plurality of standard patterns formed by clustering historical monitoring data of the satellite lithium ion battery.
According to the method for monitoring the health state of the satellite lithium ion battery, the newly acquired real-time monitoring data of the satellite lithium ion battery is input into the neural network through the self-organizing mapping neural network obtained by training based on the historical monitoring data of the satellite lithium ion battery, so that the winning neuron corresponding to the real-time monitoring data is determined, and then the CV virtual index indicating the health state of the satellite lithium ion battery is calculated by using the set parameters and the dot product between each input neuron and the winning neuron of the self-organizing mapping neural network. The purpose of extracting the CV virtual index from the parameters of the satellite lithium ion battery by adopting the self-organizing mapping neural network is achieved, and the technical effect of monitoring the health state of the satellite lithium ion battery with high accuracy is achieved.
Referring to fig. 2, in an embodiment, the process of training the self-organizing map neural network in the step S14 may include the following processing steps S20 to S25:
s20, acquiring historical monitoring data of the satellite lithium ion battery;
s21, determining the total number m of the battery parameters according to the historical monitoring data, and clustering the historical monitoring data to obtain a clustering number r;
s23, determining that the number of input layer neurons and the number of output layer neurons of the self-organizing mapping neural network structure are m and r according to the total number m of the battery parameters and the number r of clusters;
s25, setting a weight network between neurons of an output layer and an input layer of the self-organizing map neural network; the weight network is:
wherein, wijRepresenting the weight value between the ith input neuron and the jth output neuron.
It is understood that, in this embodiment, the historical monitoring data of the satellite lithium ion battery may also be, but is not limited to, obtained by directly receiving telemetry data transmitted back by the satellite and extracting from the telemetry data, or may be obtained by direct input by a worker, or may be obtained by request from a networked database server.
Specifically, if the input data of the satellite lithium ion battery (i.e. the historical monitoring data or the data set protecting the historical monitoring data) is:
where m represents the total number of battery parameters and n represents the total number of telemetry data for the aforementioned satellite, where xijRepresenting the value of the jth parameter in the ith telemetry datum. The input layer of the structure of the SOM neural network can be defined as m neural units and the output layer as r neural units. The value-taking principle of r is that the monitoring data with normal health state of the satellite lithium ion battery in the monitoring data are clustered to form a plurality of standard modes, and r is the cluster number. Thereafter, based on the set input layer neurons and output layer neurons, a weighting network between the output layer neurons and the input layer neurons is set as described above.
Through the processing steps, the early preparation process of the SOM neural network training can be quickly completed.
Referring to fig. 3, in an embodiment, the process of training the neural network to obtain the required self-organizing map may specifically include the following processing steps S27 to S30:
and S27, performing weight initialization on the weight network between the output layer neuron and the input layer neuron of the self-organizing map neural network.
It can be understood that, in this step, it is necessary to assign an initial value to each weight value in the weight network and normalize the weight values. Those skilled in the art will understand that the assignment method known in the art can be used to implement the assignment of initial values of weight values required in the present embodiment.
In some real-time manners, optionally, a random value method may be adopted to assign an initial value to each weight value in the weight network and normalize the weight values. Preferably, if the selectable value range of the random value is a certain value range, the random value can be set to be a smaller value, so that the calculation efficiency of the subsequent processing steps can be effectively improved, and the time consumed by training is saved.
S28, extracting input sample data from the historical monitoring data and inputting the sample data into the self-organizing map neural network, and determining a winning neuron corresponding to the input sample data;
it is to be understood that after the initialization weight is completed, the sample may be input for calculation. Specifically, an input sample set is extracted from the historical monitoring data, for example, the ith row of data is randomly extracted from the input data X as input sample data xw, that is:
xw=[xi1,xi2,…,xim]
and outputting the extracted input sample data to the SOM neural network to be trained to obtain the dot product of the weight vector of each output neuron and the input neuron. And from the obtained each dot product, the output neuron corresponding to the maximum dot product value is the winning neuron.
Optionally, in some embodiments, the step S28 may specifically include the following processing steps:
extracting input sample data from historical monitoring data, and carrying out normalized processing on the input sample data by using an Euclidean norm method to obtain an Euclidean norm corresponding to the input sample data;
obtaining the results of performing dot product calculation on the obtained Euclidean norm and each output layer neuron of the self-organizing mapping neural network respectively to obtain dot product vectors;
and determining the dot product maximum value in the dot product vector, and determining a winning neuron corresponding to the input sample data according to the dot product maximum value.
Specifically, the ith row of data is randomly extracted from the input data X as input sample data xw, that is:
xw=[xi1,xi2,…,xim]
and input sample data xw is subjected to normalized processing by a Euclidean norm method to obtain a corresponding Euclidean norm xw':
xw′=xw/||xw||
where xw' represents the Euclidean norm of xw.
The processed xw' will perform dot product with each output layer neuronCalculating outAnd obtaining a dot product vector:
g=[g1,g2,…,gr]
wherein, gjThe dot product of the weight vector representing the jth output layer neuron and the input neuron.
gj=w1j*xi1+w2j*xi2+…+wnj*xin
Finding out the maximum value of g from the dot product vector g and finding out the corresponding output neuron, which is the winning neuron gwin。
S29, updating the connection weight between the winning neuron and each input neuron according to the learning rate, the input sample data and the connection weight between the winning neuron and each input layer neuron corresponding to the input sample data;
it can be understood that, after the input calculation of the input sample data is completed to obtain the corresponding winning neuron, the connection weights between the winning neuron and each input neuron need to be updated so as to determine whether the training can be finished by the current training progress.
Alternatively, in some embodiments, regarding the process of updating the connection weights between the winning neuron and each input neuron in the foregoing step S29, the foregoing connection weights may be updated by the following update formula:
Wwin *=Wwin+η(t)(xw′-Wwin)
wherein, Wwin *Representing updated winning neurons and input nervesConnection weight between elements, WwinRepresenting the weight of the connection between the winning neuron and each input neuron, η (t) representing the learning rate, as a function of the training time t and the topological distance n of the winning neuron: η (t, n) ═ η (t, n) e-nAnd xw' represents the euclidean norm of the input sample data.
And S30, if the learning rate meets the preset training termination condition, obtaining the self-organizing map neural network after the training is finished.
It is understood that the training termination condition may be that the learning rate is less than a set value (η (t) < ηmin) Or a predetermined number of iterations, ηminIndicating the set minimum learning rate. After the above steps S28 and S29 are performed each time, it may be determined whether the current training schedule can satisfy a training termination condition, for example, a set training number N corresponding to the training termination condition, and it is determined whether the training number up to the current time is equal to the set training number N, so that the learning rate is less than the minimum learning rate, and if so, the training is ended to obtain the trained SOM neural network.
As shown in fig. 3, if the learning rate does not satisfy the predetermined training termination condition, the process returns to step S28 until the learning rate satisfies the predetermined training termination condition.
Through the processing steps, the needed SOM neural network can be obtained through efficient training based on normal monitoring data of the satellite lithium ion battery.
In order to more intuitively explain the above-described embodiments of the method of the present invention, the following specific implementation examples are given. It should be noted that the following examples are not intended to be the only limitations of the above-described embodiments of the method of the present invention, but rather are exemplary embodiments of the present invention:
in the past research, the CV virtual index is generally used for mechanical stand-alone devices such as bearings, and no experiment or application has been performed on a satellite lithium ion battery, and in order to check the problems of how to effect and whether the CV virtual index is suitable for the stand-alone devices, the health index is checked in the specification based on NASA battery data, and the NASA battery data is derived from ground laboratory data, so that the health index SOH recognized by the lithium ion battery can be calculated. And verifying the effectiveness of the value of the CV virtual index on the health state evaluation of the lithium ion battery by comparing the value of the CV virtual index with the SOH trend curve.
The specific test process is as follows:
firstly, selecting lithium battery performance degradation data in an open source database of NASA (national institutes of health), and checking the effectiveness of a model, wherein the data come from an aging experiment, a 18650 lithium ion battery is selected to perform a cyclic charge and discharge process, and each complete charge and discharge cycle comprises three processes: charging, discharging and impedance measurement, and cyclic charging and discharging lead to the degradation of the performance of the lithium ion battery and the reduction of the actual capacity. Based on different experimental conditions such as discharge depth, external temperature, failure threshold and the like, the NASA performs 9-batch lithium ion battery aging experiments, selects 3 batteries of the first batch, and adopts the battery numbers of 5, 6 and 7 respectively, the experimental temperature of 24 ℃, the discharge current of 2A, the cut-off voltage of 2.7V, 2.5V and 2.2V respectively, and the experimental termination condition of 30% degradation.
In the process of an experiment, a NASA experiment platform collects a plurality of data including battery section voltage, battery output current and the like, and based on the data, researchers extract 5 indirect indexes, namely an isobaric charging time interval K1, an isobaric charging time interval K2, an isobaric discharging time interval K3, a charging-period battery average temperature K4 and a discharging-period battery average temperature K5. The data used contained 168 days of data for three cells, with 5 indirect indicators per day and cell capacity. The accepted battery health indicator SOH can be calculated based on the battery capacity to characterize the degradation trend of the battery, the degradation trends of the three batteries being shown in fig. 4.
On the basis of the above-mentioned studies by the predecessors, the following experiments were conducted.
Because the data condition provided in the NASA database is relatively good, the data requirement of the proposed model is met, so that data preprocessing is not needed, and then the K4 index and the K5 index which have small correlation with the battery health are screened out through parameter screening, thereby avoiding influencing the accuracy of the value of the CV virtual index. And finally, performing index fusion on the K1 index, the K2 index and the K3 index to obtain a new comprehensive index CV.
For the test, the SOM neural network method is used for training the model of the data in the first 80 th period, the trained SOM neural network model is used for calculating the data from the 81 th period to the 168 th period, and based on the three indexes, the three batteries of 5, 6 and 7 are respectively tested to obtain the comprehensive index CV. The experimental CV value establishment procedure is shown below for battery No. 6.
Since 3 parameters are selected: the CV virtual index is constructed by the K1 index, the K2 index and the K3 index, the number of neurons in an input layer of the SOM neural network is 3, battery data are simple and do not need clustering, the standard state of the data can be defined as failure and non-failure, and the number of the neurons in an output layer of the SOM neural network is 2. And (3) building an SOM neural network with 3 input layer neurons and 2 output layer neurons.
The first step is as follows: initialization weights
By means of a random value method, an initial value is given to each weight value in the WEI, and an initial weight matrix is obtained as follows:
the normalization of the initial weight matrix is:
the second step to the fourth step: training SOM neural networks
Inputting data of No. 6 battery in the first 80 periods for training iteration, setting training termination conditions that the training times reach 2000 times, and updating the weight matrix after training is completed as follows:
the fifth step: calculating a value of a CV virtual index
Data of battery No. 6 from 81 to 168 periods are input into the trained SOM neural network, winning neurons and dot product values corresponding to the data of each period are obtained, the value of the setting parameter c is set to 1.5, and the calculated value of the CV virtual index is shown in fig. 5.
In order to prove the effectiveness of the CV virtual index, the CV virtual index from 81 th to 168 th periods is compared with the accepted index SOH of the lithium ion battery, and comparative graphs are obtained as shown in fig. 6 to 8 respectively.
As can be seen from fig. 6 to 8, the proposed CV virtual index achieves better effects on three batteries (No. 5 battery to No. 7 battery), which are mainly shown in the following: the change trend of the CV virtual index is similar to the change trend of the index SOH, the change of a plurality of peaks is well matched with the change of the index SOH, and the battery degradation trend represented by the CV virtual index is consistent with the actual condition. Through experiments on NASA battery data, the CV virtual index can achieve the effect of representing the health condition of the lithium ion battery similar to the index SOH. Health indexes such as indexes SOH and the like are difficult to apply to the field of satellite lithium ion batteries due to the limitation of calculation conditions, and the CV virtual index is applied to well solve the problem and has great practical significance for health evaluation of the satellite lithium ion batteries.
It should be understood that, although the steps in the flowcharts of fig. 1 to 3 are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 1-3 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performing the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.
Referring to fig. 9, on the other hand, a health status monitoring apparatus 100 for a satellite lithium ion battery is further provided, which includes a data obtaining module 13, a winning determining module 15, and an index calculating module 17. The data acquisition module 13 is configured to acquire real-time monitoring data of the satellite lithium ion battery. The winning determining module 15 is used for inputting the real-time monitoring data into the self-organizing mapping neural network obtained by training and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery. The index calculation module 17 is configured to calculate a CV virtual index according to a dot product between each input neuron and a winning neuron of the self-organizing map neural network and a set parameter; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
The health state monitoring device 100 for the satellite lithium ion battery adopts the self-organizing map neural network obtained by training based on the historical monitoring data of the satellite lithium ion battery through the cooperation of all modules, inputs the newly acquired real-time monitoring data of the satellite lithium ion battery into the neural network so as to determine the winning neuron corresponding to the real-time monitoring data, and then calculates the CV virtual index indicating the health state of the satellite lithium ion battery by using the set parameters and the dot product between each input neuron and the winning neuron of the self-organizing map neural network. The purpose of extracting the CV virtual index from the parameters of the satellite lithium ion battery by adopting the self-organizing mapping neural network is achieved, and the technical effect of monitoring the health state of the satellite lithium ion battery with high accuracy is achieved.
In one embodiment, the health status monitoring apparatus 100 for a satellite lithium ion battery may further include a network training module. The network training module comprises a historical data submodule, a parameter clustering submodule, a network structure determining submodule and a weight setting submodule, wherein the historical data submodule is used for acquiring historical monitoring data of the satellite lithium ion battery. And the parameter clustering submodule is used for determining the total number m of the battery parameters according to the historical monitoring data, clustering the historical monitoring data and obtaining a clustering number r. And the network structure determining submodule is used for determining that the number of input layer neurons and the number of output layer neurons of the self-organizing mapping neural network are m according to the total number m of the battery parameters and the clustering number r. The weight setting submodule is used for setting a weight network between neurons of an output layer and neurons of an input layer of the self-organizing mapping neural network; the weight network is:
wherein, wijRepresenting the weight value between the ith input neuron and the jth output neuron.
In one embodiment, the network training module further comprises an initialization sub-module, an input determination sub-module, a weight update sub-module, and a training judgment sub-module. The initialization submodule is used for carrying out weight initialization on a weight network between an output layer neuron and an input layer neuron of the self-organizing mapping neural network. The input determining submodule is used for extracting input sample data from the historical monitoring data, inputting the input sample data into the self-organizing mapping neural network, and determining a winning neuron corresponding to the input sample data. And the weight updating submodule is used for updating the connection weight between the winning neuron and each input neuron according to the learning rate, the input sample data and the connection weight between the winning neuron and each input layer neuron corresponding to the input sample data. And the training judgment submodule is used for obtaining the self-organizing mapping neural network after the training is finished when the learning rate meets the preset training termination condition. The training judgment sub-module is further used for triggering the input determination sub-module to start the next round of processing when the learning rate does not meet the predetermined training termination condition until the learning rate meets the predetermined training termination condition.
In an embodiment, the initialization sub-module may be specifically configured to assign an initial value to each weight value in the weight network and normalize the weight values by using a random value method.
In one embodiment, the input determining submodule is configured to extract input sample data from the historical monitoring data, and perform normalized processing on the input sample data by using a euclidean norm method to obtain a euclidean norm corresponding to the input sample data; obtaining the results of performing dot product calculation on the obtained Euclidean norm and each output layer neuron of the self-organizing mapping neural network respectively to obtain dot product vectors; and determining the dot product maximum value in the dot product vector, and determining a winning neuron corresponding to the input sample data according to the dot product maximum value.
In an embodiment, the weight updating sub-module may be specifically configured to update the weight by using the following update algorithm:
Wwin *=Wwin+η(t)(xw′-Wwin)
wherein, Wwin *Representing the updated connecting weight between the winning neuron and each input neuron, WwinRepresents the connection weight between the winning neuron and each input neuron, η (t) represents the learning rate, and xw' represents the euclidean norm of the input sample data.
For specific limitations of the satellite lithium ion battery health status monitoring apparatus 100, reference may be made to the above corresponding limitations of the satellite lithium ion battery health status monitoring method, and details are not repeated here. All or part of the modules in the health status monitoring device 100 for satellite lithium ion batteries can be implemented by software, hardware and a combination thereof. The modules may be embedded in a hardware form or a device independent of a specific data processing function, or may be stored in a memory of the device in a software form, so that a processor may invoke and execute operations corresponding to the modules, where the device may be, but is not limited to, a measurement and control terminal of a satellite control system, a control device on a satellite, or a personal computer.
In still another aspect, a computer device is provided, which includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the following steps: acquiring real-time monitoring data of the satellite lithium ion battery; inputting real-time monitoring data into a self-organizing mapping neural network obtained by training, and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery; calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
In one embodiment, the processor, when executing the computer program, may further implement the additional steps or substeps in the embodiments of the method for monitoring the state of health of a satellite lithium ion battery.
In yet another aspect, there is also provided a computer readable storage medium having a computer program stored thereon, the computer program when executed by a processor implementing the steps of: acquiring real-time monitoring data of the satellite lithium ion battery; inputting real-time monitoring data into a self-organizing mapping neural network obtained by training, and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training historical monitoring data based on a satellite lithium ion battery; calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
In one embodiment, when being executed by a processor, the computer program may further implement the additional steps or substeps in each embodiment of the health status monitoring method for a satellite lithium ion battery.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), synchronous link DRAM (Synchlink) DRAM (SLDRAM), Rambus DRAM (RDRAM), and interface DRAM (DRDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above examples only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for those skilled in the art, various changes and modifications can be made without departing from the spirit of the present application, and all of them fall within the scope of the present application. Therefore, the protection scope of the present patent should be subject to the appended claims.
Claims (10)
1. A health state monitoring method of a satellite lithium ion battery is characterized by comprising the following steps:
acquiring real-time monitoring data of the satellite lithium ion battery;
inputting the real-time monitoring data into a self-organizing mapping neural network obtained by training, and determining a winning neuron corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training based on historical monitoring data of the satellite lithium ion battery;
calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing map neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
2. The method for monitoring the state of health of a satellite lithium ion battery according to claim 1, wherein training the process of obtaining the self-organizing map neural network comprises:
performing weight initialization on a weight network between output layer neurons and input layer neurons of the self-organizing map neural network;
extracting input sample data from the historical monitoring data, inputting the input sample data into the self-organizing mapping neural network, and determining a winning neuron corresponding to the input sample data;
updating the connection weight between the winning neuron and each input neuron according to the learning rate, the input sample data and the connection weight between the winning neuron and each input layer neuron corresponding to the input sample data;
and if the learning rate meets a preset training termination condition, obtaining the self-organizing mapping neural network after the training is finished.
3. The method for monitoring the state of health of a satellite lithium ion battery according to claim 2, wherein training the process of obtaining the self-organizing map neural network further comprises:
and if the learning rate does not meet the preset training termination condition, returning to the step of extracting input sample data from the monitoring data, inputting the input sample data into the self-organizing map neural network, and determining the winning neuron corresponding to the input sample data until the learning rate meets the preset training termination condition.
4. The method for monitoring the state of health of a satellite lithium ion battery according to claim 2, wherein the step of initializing weights of the weight network between the output layer neurons and the input layer neurons of the self-organizing map neural network comprises:
and assigning an initial value to each weight value in the weight network by adopting a random value method and carrying out normalization processing on the weight values.
5. The method according to claim 2, wherein the step of extracting input sample data from the monitoring data, inputting the input sample data into the self-organizing map neural network, and determining a winning neuron corresponding to the input sample data comprises:
extracting the input sample data from the historical monitoring data, and carrying out normalized processing on the input sample data by using a Euclidean norm method to obtain the Euclidean norm corresponding to the input sample data;
obtaining the results of performing dot product calculation on the Euclidean norm and each output layer neuron of the self-organizing mapping neural network respectively to obtain dot product vectors;
and determining a dot product maximum value in the dot product vector, and determining a winning neuron corresponding to the input sample data according to the dot product maximum value.
6. The method for monitoring the state of health of a satellite lithium-ion battery according to claim 2, wherein in the process of updating the connection weights between the winning neuron and each input neuron, the connection weights are updated according to an update formula as follows:
Wwin *=Wwin+η(t)(xw′-Wwin)
wherein, Wwin *Representing the updated connecting weight between the winning neuron and each input neuron, WwinRepresents a connection weight between the winning neuron and each input neuron, η (t) represents the learning rate, and xw' represents a euclidean norm of the input sample data.
7. The method for monitoring the state of health of a satellite lithium-ion battery according to any one of claims 2 to 6, wherein the step of initializing the weights of the weight network between the output layer neurons and the input layer neurons of the self-organizing map neural network is preceded by:
acquiring historical monitoring data of the satellite lithium ion battery;
determining the total number m of battery parameters according to the historical monitoring data, and clustering the historical monitoring data to obtain a clustering number r;
determining that the number of input layer neurons and the number of output layer neurons of the structure of the self-organizing map neural network are m and r according to the total number m of the battery parameters and the cluster number r;
setting a weight network between neurons of an output layer and an input layer of the self-organizing map neural network; the weight network is:
wherein, wijRepresenting the weight value between the ith input neuron and the jth output neuron.
8. A health status monitoring device of a satellite lithium ion battery is characterized by comprising:
the data acquisition module is used for acquiring real-time monitoring data of the satellite lithium ion battery;
the winning determining module is used for inputting the real-time monitoring data into a self-organizing mapping neural network obtained by training and determining winning neurons corresponding to the real-time monitoring data; the self-organizing mapping neural network is obtained by training based on historical monitoring data of the satellite lithium ion battery;
the index calculation module is used for calculating to obtain CV virtual indexes according to dot products between input neurons and winning neurons of the self-organizing mapping neural network and set parameters; the CV virtual index is used for indicating the health state of the satellite lithium ion battery.
9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor when executing the computer program implements the steps of the method for monitoring state of health of a satellite lithium ion battery according to any one of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method for monitoring the state of health of a satellite lithium-ion battery according to any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110225533.XA CN112816884B (en) | 2021-03-01 | 2021-03-01 | Method, device and equipment for monitoring health state of satellite lithium ion battery |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110225533.XA CN112816884B (en) | 2021-03-01 | 2021-03-01 | Method, device and equipment for monitoring health state of satellite lithium ion battery |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112816884A true CN112816884A (en) | 2021-05-18 |
CN112816884B CN112816884B (en) | 2022-08-02 |
Family
ID=75862585
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110225533.XA Active CN112816884B (en) | 2021-03-01 | 2021-03-01 | Method, device and equipment for monitoring health state of satellite lithium ion battery |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112816884B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113780352A (en) * | 2021-08-10 | 2021-12-10 | 北京自动化控制设备研究所 | Satellite receiver health management method based on neural network |
CN117592383A (en) * | 2024-01-19 | 2024-02-23 | 四川晟蔚智能科技有限公司 | Method, system, equipment and medium for predicting equipment health life |
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 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080288493A1 (en) * | 2005-03-16 | 2008-11-20 | Imperial Innovations Limited | Spatio-Temporal Self Organising Map |
US20180322393A1 (en) * | 2017-05-02 | 2018-11-08 | Stmicroelectronics S.R.L. | Neural network, corresponding device, apparatus and method |
CN108960424A (en) * | 2018-06-29 | 2018-12-07 | 深圳市汇沣世纪数据工程有限公司 | Determination method, apparatus, equipment and the storage medium of triumph neuron |
CN109165472A (en) * | 2018-10-11 | 2019-01-08 | 北京航空航天大学 | A kind of power supply health evaluating method based on variable topological self-organizing network |
CN111239685A (en) * | 2020-01-09 | 2020-06-05 | 辽宁工程技术大学 | Sound source positioning method based on uniform design and self-organizing feature mapping neural network |
CN112015854A (en) * | 2020-07-17 | 2020-12-01 | 河海大学常州校区 | Heterogeneous data attribute association algorithm based on self-organizing mapping neural network |
-
2021
- 2021-03-01 CN CN202110225533.XA patent/CN112816884B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080288493A1 (en) * | 2005-03-16 | 2008-11-20 | Imperial Innovations Limited | Spatio-Temporal Self Organising Map |
US20180322393A1 (en) * | 2017-05-02 | 2018-11-08 | Stmicroelectronics S.R.L. | Neural network, corresponding device, apparatus and method |
CN108960424A (en) * | 2018-06-29 | 2018-12-07 | 深圳市汇沣世纪数据工程有限公司 | Determination method, apparatus, equipment and the storage medium of triumph neuron |
CN109165472A (en) * | 2018-10-11 | 2019-01-08 | 北京航空航天大学 | A kind of power supply health evaluating method based on variable topological self-organizing network |
CN111239685A (en) * | 2020-01-09 | 2020-06-05 | 辽宁工程技术大学 | Sound source positioning method based on uniform design and self-organizing feature mapping neural network |
CN112015854A (en) * | 2020-07-17 | 2020-12-01 | 河海大学常州校区 | Heterogeneous data attribute association algorithm based on self-organizing mapping neural network |
Non-Patent Citations (2)
Title |
---|
孙扩 等: "干扰条件下基于改进SOM故障诊断研究", 《电子测量技术》, vol. 42, no. 9, 31 May 2019 (2019-05-31), pages 131 - 136 * |
谷吉海等: "SOM神经网络在卫星电源***故障诊断中的应用", 《强度与环境》, no. 02, 15 June 2002 (2002-06-15), pages 38 - 41 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113780352A (en) * | 2021-08-10 | 2021-12-10 | 北京自动化控制设备研究所 | Satellite receiver health management method based on neural network |
CN117592383A (en) * | 2024-01-19 | 2024-02-23 | 四川晟蔚智能科技有限公司 | Method, system, equipment and medium for predicting equipment health life |
CN117592383B (en) * | 2024-01-19 | 2024-03-26 | 四川晟蔚智能科技有限公司 | Method, system, equipment and medium for predicting equipment health life |
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 |
Also Published As
Publication number | Publication date |
---|---|
CN112816884B (en) | 2022-08-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112816884B (en) | Method, device and equipment for monitoring health state of satellite lithium ion battery | |
CN112731159B (en) | Method for pre-judging and positioning battery faults of battery compartment of energy storage power station | |
CN110221225B (en) | Spacecraft lithium ion battery cycle life prediction method | |
CN113805064B (en) | Lithium ion battery pack health state prediction method based on deep learning | |
CN109001640B (en) | Data processing method and device for power battery | |
EP3542173A1 (en) | Determining a state of health of a battery and providing an alert | |
CN113065283A (en) | Battery life prediction method, system, electronic device and storage medium | |
CN113109715B (en) | Battery health condition prediction method based on feature selection and support vector regression | |
CN112834927A (en) | Lithium battery residual life prediction method, system, device and medium | |
CN115201686B (en) | Lithium ion battery health state assessment method under incomplete charge and discharge data | |
US20230059529A1 (en) | Characterization of Rechargeable Batteries Using Machine-Learned Algorithms | |
CN114545274A (en) | Lithium battery residual life prediction method | |
CN115598557B (en) | Lithium battery SOH estimation method based on constant-voltage charging current | |
CN115409263A (en) | Method for predicting remaining life of lithium battery based on gating and attention mechanism | |
CN115902642A (en) | Battery state of charge estimation method and device, electronic equipment and storage medium | |
CN109921462B (en) | New energy consumption capability assessment method and system based on LSTM | |
CN115032540A (en) | Lithium ion battery health state estimation method, device, equipment and medium | |
CN113312807B (en) | Electrolyte formula recommendation method based on lithium battery performance simulation environment | |
CN116577686B (en) | Multi-working condition SOH estimation method and system based on local stage charging data | |
CN116908727A (en) | Multi-feature combination-based energy storage power station lithium battery capacity estimation method | |
CN117169743A (en) | Battery health state estimation method and device based on partial data and model fusion | |
CN114325393B (en) | Self-adaptive estimation method for SOH (self-adaptive state of charge) of lithium ion battery pack based on PF (power factor) and GPR (power factor) | |
CN116224085A (en) | Lithium battery health state assessment method based on data driving | |
CN116643177A (en) | Online battery health degree prediction method, device, equipment and medium | |
CN114896865B (en) | Digital twinning-oriented self-adaptive evolutionary neural network health state online prediction method |
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 |