Battery health estimation methods, devices, electronic equipment and storage media
By optimizing the parameters of the support vector machine through the harmony search algorithm, a battery health estimation model is constructed, which solves the problems of long time consumption and inability to monitor in real time in the existing technology, and realizes real-time prediction of battery health and improves the reliability of the power supply system.
Patent Information
- Application Number
- CN202110565723.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-05-24
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2041-05-24
AI Technical Summary
Existing methods for estimating battery health are time-consuming, labor-intensive, and cannot achieve real-time monitoring, which prevents data centers from promptly identifying and eliminating potential problems and affects power supply reliability.
The parameters of the support vector machine are optimized using the harmony search algorithm, and a battery health estimation model is constructed. Real-time prediction is then performed by acquiring monitoring data.
It enables real-time monitoring of battery health, reduces the probability of accidents caused by untimely maintenance, and improves the reliability of the power supply system.
Smart Images

Figure CN115392103B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent detection technology, and in particular to a method, apparatus, electronic device, and storage medium for estimating the health of a battery. Background Technology
[0002] With the rapid development of cloud computing and big data technologies, much of the data storage and processing takes place in data centers, which requires data centers to provide sufficient and stable power for various storage and computing servers.
[0003] AC and DC uninterruptible power supply systems are the power guarantee for data center computer rooms, ensuring the continuity of power supply for various equipment in the data center. Batteries are the last line of defense in uninterruptible power supply systems, and their quality directly affects the operation of uninterruptible power supply systems. Therefore, there is an increasing amount of research on battery health prediction.
[0004] The state of health (SOH) of a battery refers to the ratio of the battery's capacity to its nominal capacity under certain conditions. It is an indicator for judging whether a battery can continue to provide reliable protection.
[0005] Current methods for estimating battery health primarily rely on raw comparison and measurement. This involves estimating through verification tests and full-capacity tests to establish a correlation between battery internal resistance and health status. The battery's internal resistance characteristics are then used to accurately reflect its current health. However, due to the long testing cycles, this approach is unsuitable for data centers with large numbers of batteries. Furthermore, this rudimentary method cannot provide real-time monitoring of battery health.
[0006] The battery capacity estimation methods mentioned above all require obtaining the actual capacity of the battery. The most commonly used method is the full capacity test method. However, it takes about 10 hours to perform a full capacity test on each battery. For a large data center with tens of thousands of batteries, performing a full capacity test on each battery is almost an impossible task.
[0007] In addition, according to relevant maintenance regulations, most data centers conduct capacity tests every three years. Such a low state of health (SOH) measurement frequency makes it impossible for maintenance personnel to keep track of the battery's working status in a timely manner, and even more impossible to detect and eliminate potential hazards in a timely manner. Summary of the Invention
[0008] Existing methods for estimating battery health generally suffer from problems such as long processing time, large workload, and inability to achieve real-time monitoring. This invention provides a method, apparatus, electronic device, and storage medium for estimating battery health.
[0009] This invention provides a method for estimating the health of a storage battery, comprising: obtaining a health value for each sample storage battery and obtaining monitoring data for each sample storage battery; wherein the health value is the ratio of the current capacity to the nominal capacity;
[0010] The monitoring data of each sample battery is used as a sample data, and the health value of the sample battery is used as the corresponding label to construct a sample dataset;
[0011] Based on the harmony search algorithm, the optimal parameters of the support vector machine are determined according to the sample dataset to optimize the support vector machine;
[0012] Using the sample dataset, the optimized support vector machine is pre-trained to obtain a health value estimation model;
[0013] Acquire monitoring data of the target battery and use the monitoring data as input to the health value estimation model to obtain the health value of the target battery output by the health value estimation model.
[0014] According to a battery health estimation method provided by the present invention, the step of determining the optimal parameters of a support vector machine based on the sample dataset using a harmony search algorithm to optimize the support vector machine includes:
[0015] Initialize the parameters of the support vector machine and the algorithm parameters of the harmony search algorithm, and determine the initial solution space related to the initial parameters of the support vector machine;
[0016] The initial solution space is used as the initial harmony vector of the harmony search algorithm, and the harmony search algorithm is iteratively calculated to obtain the optimal harmony vector.
[0017] Based on the optimal harmony vector, the optimal parameters of the support vector machine are determined, and the optimal parameters are used as the parameters of the support vector machine to complete the optimization of the support vector machine;
[0018] The initial parameters are the parameters of the support vector machine after initialization. The parameters of the support vector machine include the penalty factor c and the kernel function parameter g;
[0019] The algorithm parameters include the harmony memory size (HMS), the harmony memory value probability (HMCR), the pitch fine-tuning probability (PAR), the pitch fine-tuning bandwidth (BW), and the maximum number of compositions (T). max .
[0020] According to a battery health estimation method provided by the present invention, after initializing the algorithm parameters of the harmony search algorithm, the method further includes:
[0021] Let the expression of the fitness evaluation function f(X) optimized by the harmony search algorithm be f(X), X={x1,x2…,x n}∈R n ; where x n is the nth parameter to be optimized in the support vector machine, X is a band consisting of n members, R n is an n-dimensional set;
[0022] Randomly generate HMS harmonies from the solution space of X to form a harmony memory HM.
[0023] According to a storage battery health estimation method provided by the present invention, the step of using the initial solution space as the initial harmony vector of the harmony search algorithm and performing iterative calculation of the harmony search algorithm to obtain the optimal harmony vector comprises:
[0024] S1: using the initial solution space as the initial harmony vector of the harmony search algorithm, and performing harmony search algorithm calculation to determine the fitness value corresponding to the initial harmony vector;
[0025] S2: if the fitness value is not less than the minimum fitness value corresponding to all harmony vectors in HM, replacing the harmony variable corresponding to the minimum fitness value in HM with the initial harmony vector;
[0026] S3: generating a random number r1 between 0 and 1, and comparing r1 with HMCR;
[0027] If r1<HMCR, randomly take out one harmony variable from HM, and fine-tune the harmony variable by using PAR and BW to generate a new harmony variable
[0028] Otherwise, generate a new harmony variable from the solution space of X;
[0029] S4: determining the fitness value corresponding to the new harmony variable;
[0030] S5: iteratively executing steps S2 to S4 until the number of iterations reaches T max , and determining the harmony vector corresponding to the maximum fitness value in the current HM as the optimal harmony vector.
[0031] According to a storage battery health estimation method provided by the present invention, the monitoring data includes at least one of the following data: voltage, resistance and temperature.
[0032] According to a storage battery health estimation method provided by the present invention, the fitness evaluation function f(X) optimized by the harmony search algorithm is specifically a svmtrain function;
[0033] The kernel function of the support vector machine is specifically the radial basis function (RBF).
[0034] The present invention also provides a battery health estimation device, comprising:
[0035] The data acquisition unit is used to acquire the health value of each sample battery and to acquire the monitoring data of each sample battery; the health value is the ratio of the current capacity to the nominal capacity.
[0036] The dataset construction unit is used to construct a sample dataset by taking the monitoring data of each sample battery as a sample data and the health value of the sample battery as a corresponding label.
[0037] The parameter optimization unit is used to determine the optimal parameters of the support vector machine based on the harmony search algorithm and the sample dataset, so as to optimize the support vector machine.
[0038] The model building unit is used to pre-train the optimized support vector machine based on the sample dataset to obtain a health value estimation model.
[0039] The result calculation unit is used to obtain the health value of the target battery output by the health value estimation model by using the acquired monitoring data of the target battery as input to the health value estimation model.
[0040] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of any of the above-described battery health estimation methods.
[0041] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the battery health estimation method described above.
[0042] The battery health estimation method, device, electronic device, and storage medium provided by this invention utilize a harmony search algorithm to optimize support vector machine parameters, thereby enabling real-time monitoring of battery health and reducing the probability of accidents caused by untimely battery maintenance. Attached Figure Description
[0043] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0044] Figure 1 This is a flowchart illustrating the battery health estimation method provided by the present invention;
[0045] Figure 2 This is a flowchart illustrating the harmony search algorithm provided by the present invention;
[0046] Figure 3 This is a schematic diagram of the battery health estimation device provided by the present invention;
[0047] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0049] It should be noted that in the description of the embodiments of the present invention, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.
[0050] This invention provides a health estimation model that uses a harmony search algorithm to optimize a support vector machine, and then uses a pre-built sample dataset to train the optimized support vector machine to obtain a health value estimation model for real-time prediction of the state of harmonics (SOH) of a target battery.
[0051] The following is combined Figures 1-4 This invention describes the battery health estimation method, apparatus, electronic device, and storage medium provided in embodiments of the present invention.
[0052] Figure 1 This is a flowchart illustrating the battery health estimation method provided by the present invention, as shown below. Figure 1 As shown, including but not limited to the following steps:
[0053] Step 101: Obtain the SOH of each sample battery and obtain the monitoring data of each sample battery; the SOH is the ratio of the current capacity to the nominal capacity.
[0054] The sample batteries can be selected from different data centers. To ensure the robustness of the estimation results, batteries of different models, capacities, manufacturers and service lives can be collected as sample batteries.
[0055] Then, the monitoring data of each sample battery are measured, such as the battery's output voltage, internal resistance, and the temperature of the battery body, which can be uniformly represented as the average temperature of the battery body. At the same time, the health value of each sample battery is obtained, that is, the current capacity of each sample battery is measured using the existing battery SOH measurement method, and the ratio of this current capacity to its nominal capacity is calculated.
[0056] Step 102: Take the monitoring data of each sample battery as a sample data and the health value of the sample battery as the corresponding label to construct a sample dataset.
[0057] The monitoring data of each sample battery is used as a sample data point, and the SOH of the corresponding sample battery is used as a label to construct a sample dataset. This invention does not impose a specific limit on the number of sample batteries involved in the sample dataset; for example, the monitoring data measured from 150 sample batteries and their corresponding SOH can be combined to form a single sample dataset.
[0058] Step 103: Based on the harmony search algorithm, determine the optimal parameters of the support vector machine according to the sample dataset, so as to optimize the support vector machine.
[0059] Harmony search (HS) is an intelligent optimization algorithm that simulates the process by which musicians continuously adjust the pitch of each instrument based on their memory during a band performance, ultimately resulting in a beautiful harmony in the overall band performance.
[0060] Specifically, the HS algorithm completes the optimization process by repeatedly adjusting the solution variables in the harmony memory library, causing the function value to converge continuously with the increase of the number of iterations. The algorithm is simple in concept, has few adjustable parameters, and is easy to implement.
[0061] Support Vector Machine (SVM) is a classification model whose main purpose is to find the optimal hyperplane for classification. Specifically, the SVM model is a classification algorithm that uses statistical theory to seek the best learning ability based on a limited amount of sample data. The main advantage of the SVM model is that, with a limited number of samples, by introducing an appropriate kernel function, it can separate samples of different classes, transforming linearly inseparable problems into linearly separable ones.
[0062] Different parameters of the SVM model have a significant impact on its final output. In the battery health estimation method provided by this invention, the values of different parameters of the SVM model (such as the penalty factor c or the kernel function parameter g) greatly affect the generalization ability and estimation performance of the SVM.
[0063] The battery health estimation method provided by this invention mainly uses the HS algorithm to optimize the relevant parameters of the SVM model, so as to optimize the SVM model with the relevant parameters obtained after optimization.
[0064] Step 104: Using the sample dataset, pre-train the optimized support vector machine to obtain a health value estimation model;
[0065] After optimizing the SVM model, the optimized SVM model is pre-trained using a pre-built sample dataset so that the optimized SVM model can fully learn the correlation between SOH and monitoring data, and a health value estimation model (hereinafter referred to as the SOH estimation model) is constructed from the pre-trained SVM model.
[0066] Specifically, in the battery health estimation method provided by this invention, a supervised learning method can be used to pre-train the optimized SVM model, including:
[0067] The monitoring data corresponding to each sample in the sample dataset is used as the input to the model, and the health value corresponding to that sample is used as the output of the model. Through iterative methods, the model parameters of the optimized SVM model are adjusted appropriately according to the cross-entropy loss to obtain the final SOH estimation model.
[0068] Step 105: Obtain monitoring data of the target battery and use the monitoring data as input to the health value estimation model to obtain the health value of the target battery output by the health value estimation model.
[0069] Finally, if it is necessary to estimate the SOH of any target battery, it is only necessary to collect the monitoring data of the target battery and input the monitoring data into the SOH estimation model to obtain the SOH of the target battery output by the SOH estimation model.
[0070] Furthermore, the health of the target battery can be further assessed based on the state of harm (SOH) of the target battery, according to the levels classified by relevant maintenance procedures.
[0071] For example, based on the range of State of Health (SOH) values, and according to relevant maintenance procedures, the entire range can be divided into different levels (such as healthy, normal, poor, and scrap). In this way, after obtaining the SOH of any target battery, its corresponding health level can be automatically determined based on its range.
[0072] The battery health estimation method provided by this invention uses a harmony search algorithm to optimize the parameters of a support vector machine, so as to realize real-time monitoring of battery health and reduce the probability of accidents caused by untimely battery maintenance.
[0073] Based on the above embodiments, as an optional embodiment, the above-described method of determining the optimal parameters of a support vector machine (SVM) based on a harmony search algorithm and a sample dataset to optimize the SVM mainly includes: initializing the parameters of the SVM and the algorithm parameters of the harmony search algorithm, and determining the initial solution space related to the initial parameters of the SVM; using the initial solution space as the initial harmony vector of the harmony search algorithm, performing iterative calculations of the harmony search algorithm to obtain the optimal harmony vector; determining the optimal parameters of the SVM based on the optimal harmony vector, and using the optimal parameters as the parameters of the SVM to complete the optimization of the SVM; the initial parameters are the parameters of the SVM after initialization.
[0074] As an optional embodiment, the parameters of the aforementioned support vector machine mainly include, but are not limited to: the penalty factor c and the kernel function parameter g; the algorithm parameters mainly include, but are not limited to: the harmony memory size HMS, the harmony memory value probability HMCR, the pitch fine-tuning probability PAR, the pitch fine-tuning bandwidth BW, and the maximum number of compositions T. max .
[0075] Since the SVM model is greatly affected by the set parameters, and the harmony search algorithm has a good parameter optimization effect, this invention uses the harmony search algorithm to optimize the penalty factor C and kernel function parameter g of the SVM model, thereby optimizing the SVM model.
[0076] The main idea is to consider the penalty factor C and the kernel function parameter g as two instruments in a band, mimicking the process of instruments adjusting their pitch to achieve optimal harmony, and continuously adjusting the values of the two parameters to find the optimal solution (C). best ,g best This mainly includes, but is not limited to, the following steps:
[0077] Step 1) Retrieve the sample dataset, which contains sample data related to multiple sample batteries and their corresponding health values.
[0078] Step 2) Initialize the SVM model and the parameters of the HS algorithm, including: initializing the search range of the SVM model and initializing the parameters of the harmony search algorithm, such as: setting the maximum number of creations T. max The values of harmony memory size (HMS), harmony memory probability (HMCR), pitch fine-tuning bandwidth (BW), pitch fine-tuning probability (PAR), and initial fitness value of the SVM model are used to generate a set of random data (C0, g0) to serve as the initial solution space.
[0079] Step 3): Using the initial solution space (C0, g0) obtained in Step 2), the initial values of the SVM model optimized by the HS algorithm are used. The fitness values corresponding to the initial solution space (C0, g0) are then calculated, and the parameter penalty factor C and kernel function parameter g of the SVM model optimized by the HS algorithm are evaluated based on the obtained fitness values, including:
[0080] By selecting information from the current harmony memory and comparing it with the calculated harmony fitness value, the harmony memory is updated by selecting the harmony vector corresponding to the optimal harmony fitness.
[0081] If the termination condition is not met, the harmony memory will continue to be updated. The termination condition may refer to whether the maximum number of iterations T has been reached. max .
[0082] If the maximum number of iterations is satisfied, the search ends and the search result is given, which is the penalty factor C and kernel function parameter g of the SVM classifier.
[0083] Step 4), based on the optimal penalty factor C obtained in Step 3 best and the optimal kernel function parameter g best The parameters of the SVM model are set accordingly, thus obtaining the initial model (i.e. the optimal SVM model) suitable for estimating the health value of the battery's SOH.
[0084] Figure 2 This is a flowchart illustrating the harmony search algorithm provided by the present invention, which can be used as an optional viewership rating, such as... Figure 2As shown, after initializing the algorithm parameters of the harmony search algorithm, the method further includes:
[0085] Let the expression of the fitness evaluation function f(X) optimized by the harmony search algorithm be f(X), where X={x1, x2..., x n}∈R n ; where x n is the nth parameter to be optimized in the support vector machine, X is a band composed of n members, and R n is an n-dimensional set; randomly generate HMS harmonies from the solution space of X to form a harmony memory library HM.
[0086] Further, the above-mentioned step of taking the initial solution space as the initial harmony vector of the harmony search algorithm and performing iterative calculation of the harmony search algorithm to obtain the optimal harmony vector includes:
[0087] S1, taking the initial solution space (C0, g0) as the initial harmony vector of the HS algorithm, performing HS algorithm calculation to determine the fitness value corresponding to the initial harmony vector;
[0088] S2, if the fitness value is not less than the minimum fitness value corresponding to all harmony vectors in HM, replacing the harmony variable corresponding to the minimum fitness value in HM with the initial harmony vector;
[0089] S3, generating a random number r1 between 0 and 1, and comparing r1 with HMCR;
[0090] If r1<HMCR, randomly take out a harmony variable from HM, and fine-tune the harmony variable using PAR and BW to generate a new harmony variable
[0091] Otherwise, generate a new harmony variable from the solution space of X;
[0092] S4, determining the fitness value corresponding to the new harmony variable;
[0093] S5, iteratively performing steps S2 to S4 until the number of iterations reaches T max , and determining the harmony vector corresponding to the maximum fitness value in the current HM as the optimal harmony vector.
[0094] It should be emphasized that the parameter setting of the SVM model according to the present invention takes the following points into consideration:
[0095] Commonly used kernel functions for SVM models include: linear kernel function, polynomial kernel function, radial basis RBF kernel function and Sigmoid kernel function.
[0096] Since the results of the SVM algorithm are affected by its parameters, this invention employs a harmony search algorithm to optimize the parameters of the SVM model in order to obtain better parameters. Furthermore, it uses the Radial Basis Function (RBF) kernel function as the kernel function for the SVM model. The RBF kernel function can map sample data to a higher-dimensional space and has fewer parameters compared to the polynomial kernel function. In situations where it is unclear which kernel function to choose, the RBF kernel function is preferred, as shown below:
[0097]
[0098] Furthermore, the parameters for initializing the SVM model include: setting the initial penalty factor C = 2, the initial kernel function parameter g = 0.015, and using the svmtrain function as the fitness function to calculate the fitness value of each harmony vector during the optimization of the SVM model parameters by the HS algorithm.
[0099] As an optional implementation, the following points should be emphasized regarding the HS algorithm parameter settings:
[0100] The algorithm parameters for initializing the HS algorithm mentioned above include: setting the initial harmony memory bank size HMS to 200, the harmony memory bank value probability HMCR to 9.5, the pitch fine-tuning probability PAR to 0.3, the pitch fine-tuning bandwidth BW to 0.01, and the maximum number of compositions T. max It is 20.
[0101] The battery health estimation method provided by this invention can realize real-time monitoring of the SOH of each battery in the data center, which can significantly reduce the problems and accidents caused by untimely battery maintenance, and at the same time better ensure the reliability of batteries as the power guarantee for computer room equipment.
[0102] To further demonstrate the feasibility of the battery health estimation method provided by this invention, relevant verification was conducted on batteries at the China Mobile (Hohhot) Data Center, as follows:
[0103] 149 sets of sample data were randomly collected from the data center. The monitoring data mainly included battery voltage, internal resistance, and temperature information. These 149 sets of sample data were divided into two parts: training sample data and validation sample data. The training sample data consisted of 139 sets of sample data, and the validation sample data consisted of 10 sets of sample data. The specific test results are shown in Table 1.
[0104] Table 1: Comparison of Actual Measured SOH and Estimated SOH Values
[0105] Actual measured SOH value Estimated SOH value 58% 60% 56% 57% 58% 60% 57% 59% 57% 60% 59% 60% 60% 60% 59% 60% 59% 60% 60% 60%
[0106] As can be clearly seen from Table 1, the error between the estimated SOH value obtained by using the battery health estimation method provided by this invention and the actual measured SOH value for 149 batteries is very small, which fully meets the needs of actual work. Therefore, it fully proves that the battery health estimation method provided by this invention meets the expected requirements and the solution is effective.
[0107] Figure 3 This is a schematic diagram of the battery health estimation device provided by the present invention, as shown below. Figure 3 As shown, it mainly includes: a data acquisition unit 31, a dataset construction unit 32, a parameter optimization unit 33, a model construction unit 34, and a result calculation unit 35, wherein:
[0108] The data acquisition unit 31 is mainly used to obtain the health value of each sample battery and to obtain the monitoring data of each sample battery; the health value is the ratio of the current capacity to the nominal capacity.
[0109] The dataset construction unit 32 is mainly used to construct a sample dataset by taking the monitoring data of each sample battery as a sample data and the health value of the sample battery as the corresponding label.
[0110] The parameter optimization unit 33 is mainly used to determine the optimal parameters of the support vector machine based on the harmony search algorithm and the sample dataset, so as to optimize the support vector machine.
[0111] The model building unit 34 is mainly used to pre-train the optimized support vector machine based on the sample dataset to obtain a health value estimation model.
[0112] The result calculation unit 35 is mainly used to obtain the health value of the target battery output by the health value estimation model by using the acquired monitoring data of the target battery as input to the health value estimation model.
[0113] It should be noted that the battery health estimation device provided in this embodiment of the invention can be implemented based on the battery health estimation method described in any of the above embodiments, and this embodiment will not elaborate on this.
[0114] The battery health estimation device provided by this invention uses a harmony search algorithm to optimize the parameters of a support vector machine in order to achieve real-time monitoring of battery health, thereby reducing the probability of accidents caused by untimely battery maintenance.
[0115] Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logic instructions in the memory 430 to execute a battery health estimation method. This method includes: acquiring a health value for each sample battery and acquiring monitoring data for each sample battery; the health value being the ratio of current capacity to nominal capacity; constructing a sample dataset by using the monitoring data of each sample battery as sample data and the health value of the sample battery as a corresponding label; determining the optimal parameters of a support vector machine (SVM) based on a harmony search algorithm and the sample dataset to optimize the SVM; pre-training the optimized SVM using the sample dataset to obtain a health value estimation model; acquiring monitoring data for a target battery and using the monitoring data as input to the health value estimation model to obtain the health value of the target battery output by the health value estimation model.
[0116] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0117] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer is able to execute the battery health estimation method provided by the above methods, the method comprising: acquiring a health value for each sample battery and acquiring monitoring data for each sample battery; the health value being the ratio of current capacity to nominal capacity; constructing a sample dataset by using the monitoring data of each sample battery as sample data and the health value of the sample battery as a corresponding label; determining the optimal parameters of a support vector machine based on a harmony search algorithm and the sample dataset to optimize the support vector machine; pre-training the optimized support vector machine using the sample dataset to obtain a health value estimation model; acquiring monitoring data of a target battery and using the monitoring data as input to the health value estimation model to obtain the health value of the target battery output by the health value estimation model.
[0118] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the battery health estimation method provided in the above embodiments. The method includes: acquiring a health value for each sample battery and acquiring monitoring data for each sample battery; the health value being the ratio of current capacity to nominal capacity; constructing a sample dataset by using the monitoring data of each sample battery as sample data and the health value of the sample battery as a corresponding label; determining the optimal parameters of a support vector machine based on a harmony search algorithm and the sample dataset to optimize the support vector machine; pre-training the optimized support vector machine using the sample dataset to obtain a health value estimation model; acquiring monitoring data of a target battery and using the monitoring data as input to the health value estimation model to obtain the health value of the target battery output by the health value estimation model.
[0119] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0120] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for estimating the health of a storage battery, characterized in that, comprising: acquiring the state of health value of each sample storage battery, and acquiring the monitoring data of each said sample storage battery; said state of health value is the ratio of the current capacity to the nominal capacity; taking the monitoring data of each said sample storage battery as one sample data, and the state of health value of said sample storage battery as the corresponding label, to construct a sample data set; based on a harmony search algorithm, determining the optimal parameters of a support vector machine according to said sample data set, so as to optimize said support vector machine, comprising: initializing the parameters of said support vector machine and the algorithm parameters of said harmony search algorithm, and determining an initial solution space related to the initial parameters of said support vector machine; said initial parameters are the parameters of said support vector machine after initialization; the parameters of said support vector machine include a penalty factor c and a kernel function parameter g; said algorithm parameters include a harmony memory library size HMS, a harmony memory library taking value probability HMCR, a pitch adjustment probability PAR, a pitch adjustment bandwidth BW, and a maximum creation number Tmax; Let the fitness evaluation function for optimizing the harmony search algorithm be defined. The expression is ;in Let X be the nth parameter that needs to be optimized in a support vector machine, and let X be a band of n members. Let X be an n-dimensional set; HMS harmonies are randomly generated from the solution space of X to form a harmony memory HM; wherein, the svmtrain function is used as the fitness evaluation function to calculate the fitness value of the harmony vectors; taking said initial solution space as the initial harmony vector of said harmony search algorithm, and performing iterative calculation of the harmony search algorithm to obtain the optimal harmony vector, comprising: S1: taking said initial solution space as the initial harmony vector of said harmony search algorithm, and performing harmony search algorithm calculation to determine the fitness value corresponding to said initial harmony vector; S2: if said fitness value is not less than the minimum fitness value corresponding to all harmony vectors in HM, replacing the harmony variable corresponding to said minimum fitness value in HM with said initial harmony vector; S3: generating a random number r1 between [0,1], and comparing r1 with the harmony memory library taking value probability HMCR; if r1 < HMCR, randomly taking out one harmony variable from the harmony memory library HM, and fine-tuning said harmony variable by using the pitch adjustment probability PAR and the pitch adjustment bandwidth BW to generate a new harmony variable; otherwise, generating a new harmony variable from the solution space; S4: determining the fitness value corresponding to said new harmony variable; S5: iteratively performing steps S2 to S4 until the number of iterations reaches Tmax, and determining the harmony vector corresponding to the maximum fitness value in the current HM as the optimal harmony vector; determining the optimal parameters of said support vector machine according to said optimal harmony vector, and taking said optimal parameters as the parameters of said support vector machine to complete the optimization of said support vector machine; using said sample data set to pre-train the optimized said support vector machine, so as to obtain a state of health value estimation model; acquiring the monitoring data of a target storage battery, and taking said monitoring data as the input of said state of health value estimation model, so as to acquire the state of health value of said target storage battery output by said state of health value estimation model.
2. The battery health estimation method according to claim 1, characterized in that, said monitoring data includes at least one of the following data: voltage, resistance, and temperature.
3. The storage battery state of health estimation method according to claim 1, characterized in that, the kernel function of said support vector machine is specifically a radial basis RBF function.
4. A battery health estimation device, characterized in that, comprising: a data acquisition unit, configured to acquire the health degree value of each sample storage battery, and acquire the monitoring data of each said sample storage battery; The health degree value is the ratio of the current capacity to the nominal capacity; a data set construction unit, configured to take the monitoring data of each said sample storage battery as a piece of sample data, and take the health degree value of said sample storage battery as a corresponding label, so as to construct a sample data set; a parameter optimization unit, configured to determine the optimal parameters of the support vector machine based on the harmony search algorithm and according to the sample data set, so as to optimize the support vector machine, comprising: initializing the parameters of the support vector machine and the algorithm parameters of the harmony search algorithm, and determining an initial solution space related to the initial parameters of the support vector machine; the initial parameters are the parameters of the support vector machine after initialization; The parameters of the support vector machine include a penalty factor c and a kernel function parameter g; the algorithm parameters include a harmony memory library size HMS, a harmony memory library taking value probability HMCR, a pitch adjustment probability PAR, a pitch adjustment bandwidth BW, and a maximum creation number Tmax; Let the fitness evaluation function for optimizing the harmony search algorithm be defined. The expression is ;in Let X be the nth parameter that needs to be optimized in a support vector machine, and let X be a band of n members. Let X be an n-dimensional set; HMS harmonies are randomly generated from the solution space of X to form a harmony memory HM; wherein, the svmtrain function is used as the fitness evaluation function to calculate the fitness value of the harmony vectors; taking the initial solution space as the initial harmony vector of the harmony search algorithm, and performing iterative calculation of the harmony search algorithm to obtain the optimal harmony vector, comprising: S1, taking the initial solution space as the initial harmony vector of the harmony search algorithm, performing calculation of the harmony search algorithm to determine the fitness value corresponding to the initial harmony vector; S2, if the fitness value is not less than the minimum fitness value corresponding to all harmony vectors in HM, replacing the harmony variable corresponding to the minimum fitness value in HM with the initial harmony vector; S3, generating a random number r1 between 0 and 1, and comparing r1 with the harmony memory library taking value probability HMCR; if r1 is less than HMCR, randomly taking out a harmony variable from the harmony memory library HM, and performing fine adjustment on the harmony variable by using the pitch adjustment probability PAR and the pitch adjustment bandwidth BW to generate a new harmony variable; otherwise, generating a new harmony variable from the solution space; S4, determining the fitness value corresponding to the new harmony variable; S5, iteratively executing steps S2 to S4 until the number of iterations reaches Tmax, and determining the harmony vector corresponding to the maximum fitness value in the current HM as the optimal harmony vector; determining the optimal parameters of the support vector machine according to the optimal harmony vector, and taking the optimal parameters as the parameters of the support vector machine to complete the optimization of the support vector machine; a model construction unit, configured to pre-train the optimized support vector machine according to the sample data set to obtain a health degree value estimation model; a result operation unit, configured to take the acquired monitoring data of the target storage battery as the input of the health degree value estimation model according to the monitoring data, and acquire the health degree value of the target storage battery output by the health degree value estimation model.
5. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the steps of the storage battery health degree estimation method according to any one of claims 1 to 3 are implemented.
6. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the battery health estimation method as described in any one of claims 1 to 3.
Citation Information
Patent Citations
Power maximum load small-sample prediction method
CN105184398A
Lead-acid storage battery health state estimation method for starting internal combustion engine
CN112540318A
