Battery fault prediction method, device and equipment based on big data and AI algorithm

Abstract

The present invention belongs to the technical field of battery fault detection, and discloses a battery fault prediction method, device and equipment based on big data and AI algorithm. By taking the negative feedback data set, cell parameter set, historical operation data set and historical fault data of the battery as training sets, model training and verification are performed based on the training sets to obtain a first AI prediction model and a second AI prediction model, collect actual operation data during the battery operation process, input the actual operation data into the first AI prediction model and the second AI prediction model respectively, obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model, determine the target prediction result according to the first prediction result and the second prediction result, and through the above method, the fault of the energy storage battery pack can be predicted in advance, the risk hidden dangers can be eliminated, and the probability of faults and accidents can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault detection, and in particular to a battery fault prediction method, device and equipment based on big data and AI algorithm. Background Art

[0002] Thermal runaway and life prediction of batteries are difficult problems in the industry. Possible failures and risks in the operation of energy storage battery packs cannot be predicted in advance. Problems can only be known after they are exposed or failures or accidents occur. How to predict failures of energy storage battery packs in advance has become a technical problem that needs to be solved urgently.

[0003] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] The main purpose of the present invention is to provide a battery fault prediction method, device and equipment based on big data and AI algorithm, aiming to solve the technical problem that the prior art cannot predict the fault of energy storage battery pack in advance.

[0005] To achieve the above object, the present invention provides a battery fault prediction method based on big data and AI algorithm, and the battery fault prediction method based on big data and AI algorithm comprises the following steps:

[0006] Acquire a negative feedback data set, a cell parameter set, a historical operation data set, and historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set, and the historical fault data of the battery as a training set;

[0007] Performing model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions;

[0008] Collect actual operating data during battery operation;

[0009] Inputting the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model;

[0010] A target prediction result is determined according to the first prediction result and the second prediction result.

[0011] In some embodiments, the first AI prediction model is used to predict the type of battery failure, and the battery failure prediction method based on big data and AI algorithm includes:

[0012] Identifying the data point corresponding to the maximum resistance in the training set;

[0013] Taking the data point as the midpoint, smoothing the training set by a preset function to obtain a processed training set;

[0014] Initialize the population size, maximum evolutionary generations, chromosome length, probability of crossover, probability of mutation, and model penalty parameter range, and after completing the initialization, train the first preset model based on the processed training set to obtain a first AI prediction model.

[0015] In some embodiments, the training of the first preset model based on the processed training set to obtain the first AI prediction model includes:

[0016] Converting the model parameters into binary chromosome gene strings of the chromosome length in the form of binary coding, wherein one chromosome gene string corresponds to one individual;

[0017] Use the accuracy of support vector machine classification as fitness and calculate the accuracy of support vector machine classification corresponding to all individuals in the population;

[0018] Selecting a number of individuals from the population based on the accuracy rate and forming a new population;

[0019] Performing mutation and crossover operations on the new population based on the probability of the crossover and the probability of the mutation to update the evolution of the new population;

[0020] After reaching the maximum number of evolutionary generations, the iterative evolution is stopped, and the individual with the highest accuracy in the iterative process is taken as the target individual;

[0021] The target model parameters corresponding to the target individual are used to replace the current model parameters of the first preset model to obtain a first AI prediction model.

[0022] In some embodiments, the second preset model is used to predict the life of the battery, and the battery failure prediction method based on big data and AI algorithm includes:

[0023] Preprocessing the training set to obtain a processed training set;

[0024] Determining a correspondence between battery capacity and battery voltage based on the processed training set, and extracting capacity difference features and discharge features associated with battery life from the processed training set according to the correspondence;

[0025] Extracting other features associated with battery life from the processed training set, the other features at least including charging temperature and battery internal resistance within a unit charge and discharge cycle;

[0026] Determine the historical battery life corresponding to each group of capacity difference features, discharge features and other features according to the processed training set;

[0027] The second preset model is iteratively trained multiple times according to each set of capacity difference characteristics, discharge characteristics, other characteristics and corresponding historical battery life to obtain a second AI prediction model.

[0028] In some embodiments, preprocessing the training set to obtain a processed training set includes:

[0029] Identify a first data position corresponding to a maximum battery capacity from a data header of the training set, and remove data before the first data position;

[0030] Identify from the tail of the data of the training set the second data position corresponding to the first occurrence of a battery voltage greater than a preset voltage, and remove the data after the second data position;

[0031] Based on the training set after removing the first and last data, the charge and discharge data of each cycle are obtained;

[0032] The charge and discharge data are interpolated to obtain a processed training set.

[0033] In some embodiments, determining a target prediction result according to the first prediction result and the second prediction result includes:

[0034] Determining a predicted fault type of the battery according to the first prediction result;

[0035] Determining a predicted remaining life of the battery according to the second prediction result;

[0036] Constructing a first feature vector corresponding to the predicted fault type and a second feature vector corresponding to the predicted remaining life respectively;

[0037] Cross-merge the first feature vector and the second feature vector to obtain a target feature vector;

[0038] A target prediction result is obtained based on the target feature vector, and the target prediction result is used to represent the remaining life of the battery corresponding to different fault types.

[0039] In some embodiments, cross-merging the first feature vector and the second feature vector to obtain a target feature vector includes:

[0040] determining a similarity between the first feature vector and the second feature vector;

[0041] Determine a fusion method according to the similarity;

[0042] The first feature vector and the second feature vector are cross-fused according to the fusion method to obtain a target feature vector.

[0043] In addition, to achieve the above purpose, the present invention also proposes a battery fault prediction device based on big data and AI algorithm, and the battery fault prediction device based on big data and AI algorithm includes:

[0044] An acquisition module, used to acquire a negative feedback data set, a cell parameter set, a historical operation data set, and historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set, and the historical fault data of the battery as a training set;

[0045] A construction module is used to perform model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions;

[0046] The acquisition module is used to collect actual operation data during the operation of the battery;

[0047] A prediction module, used to input the actual operation data into the first AI prediction model and the second AI prediction model respectively, to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model;

[0048] A fusion module is used to determine a target prediction result based on the first prediction result and the second prediction result.

[0049] In some embodiments, the construction module is used to identify the data point corresponding to the maximum resistance in the training set;

[0050] Taking the data point as the midpoint, smoothing the training set by a preset function to obtain a processed training set;

[0051] Initialize the population size, maximum evolutionary generations, chromosome length, probability of crossover, probability of mutation, and model penalty parameter range, and after completing the initialization, train the first preset model based on the processed training set to obtain a first AI prediction model.

[0052] In addition, to achieve the above-mentioned purpose, the present invention also proposes a battery fault prediction device based on big data and AI algorithm, and the battery fault prediction device based on big data and AI algorithm includes: a memory, a processor, and a battery fault prediction program based on big data and AI algorithm stored in the memory and executable on the processor, and the battery fault prediction program based on big data and AI algorithm is configured to implement the steps of the battery fault prediction method based on big data and AI algorithm as described above.

[0053] The present invention uses the negative feedback data set, cell parameter set, historical operation data set and historical fault data of the battery as training sets, performs model training and verification based on the training sets, obtains a first AI prediction model and a second AI prediction model, collects actual operation data during the operation of the battery, inputs the actual operation data into the first AI prediction model and the second AI prediction model respectively, obtains a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model, determines a target prediction result based on the first prediction result and the second prediction result, and through the above method, can predict the failure of the energy storage battery pack in advance, eliminate potential risks, and reduce the probability of failure and accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of a first embodiment of a battery failure prediction method based on big data and AI algorithm of the present invention;

[0055] Figure 2 This is a flow chart of a second embodiment of a battery failure prediction method based on big data and AI algorithm of the present invention;

[0056] Figure 3 This is a structural block diagram of the first embodiment of the battery fault prediction device based on big data and AI algorithm of the present invention.

[0057] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0058] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0059] The embodiment of the present invention provides a battery failure prediction method based on big data and AI algorithm, referring to Figure 1 , Figure 1 This is a flow chart of a first embodiment of a battery fault prediction method based on big data and AI algorithm of the present invention.

[0060] In this embodiment, the battery failure prediction method based on big data and AI algorithm includes the following steps:

[0061] Step S10: Acquire a negative feedback data set, a cell parameter set, a historical operation data set, and historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set, and the historical fault data of the battery as a training set.

[0062] In this embodiment, the executor of this embodiment is a battery fault prediction device based on big data and AI algorithm, wherein the battery fault prediction device based on big data and AI algorithm has functions such as data processing, data communication and program running. The battery fault prediction device based on big data and AI algorithm can be a computer terminal device or other network device, and of course it can also be other devices with similar functions, and this embodiment does not limit this.

[0063] It should be noted that thermal runaway and life prediction of batteries are difficult problems in the industry. The failures and risks that may occur during the operation of energy storage battery packs cannot be predicted in advance. Problems can only be known after they are exposed or failures or accidents occur. How to predict failures of energy storage battery packs in advance has become a technical problem that needs to be solved urgently.

[0064] In order to solve the above technical problems, in this embodiment, the negative feedback data set, cell parameter set, historical operation data set and historical fault data of the battery are used as training sets, and model training and verification are performed based on the training sets to obtain a first AI prediction model and a second AI prediction model, collect actual operation data during battery operation, and input the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model, and determine the target prediction result based on the first prediction result and the second prediction result. In this way, the faults of the energy storage battery pack can be predicted in advance, risks and hidden dangers can be eliminated, and the probability of faults and accidents can be reduced. Specifically, it can be achieved as follows.

[0065] In the specific implementation, in this embodiment, it is necessary to first obtain a training set to train the AI ​​prediction model. The training set consists of four parts, including a negative feedback data set, a battery cell parameter set, a historical operation data set, and historical fault data. The negative feedback data set is a data set formed by the model's judgment error results and the manual correction content. The basic parameters of the battery cell of a certain type of battery pack include the manufacturer's information, size, weight, design capacity, design life, discharge rate, internal resistance range, positive electrode material element ratio, negative electrode material, electrolyte material ratio, electrolyte density, diaphragm material, shell material and other parameters of the battery cell. The historical operation data set includes historical data during operation such as single cell capacity change, number of cycles, charge and discharge temperature change, charge and discharge power change, group voltage change, group capacity change, voltage difference within the group, internal resistance difference within the group, temperature difference within the group, and ambient temperature. The historical fault data is a training set manually extracted from the recorded data and a training set generated by the self-learning of the later model.

[0066] Step S20: Perform model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions.

[0067] It should be noted that in this embodiment, the first AI prediction model and the second AI prediction model are obtained by training different preset models through the above-mentioned training set, and the prediction dimensions of the two different prediction models are different. For example, the first AI prediction model is used to predict the fault type of the battery, and the second AI prediction model is used to predict the remaining life of the battery. The first AI prediction model and the second AI prediction model can predict the life of the battery under different fault types.

[0068] Furthermore, the training process of the first AI prediction model is to identify the data point corresponding to the maximum resistance in the training set; smooth the training set with the data point as the midpoint through a preset function to obtain a processed training set; initialize the population size, maximum evolutionary generation, chromosome length, probability of crossover, probability of mutation and model penalty parameter range, and after completing the initialization, train the first preset model based on the processed training set to obtain the first AI prediction model.

[0069] When constructing the first AI prediction model, the training set needs to be smoothed. The data point corresponding to the maximum resistance can be determined according to the parameters of the battery cell. The preset function adopted in this embodiment can be a two-dimensional Gaussian filter function, for example Among them, Ω represents the maximum resistance value, x represents the data point corresponding to the maximum resistance, and y represents any other data point.

[0070] Furthermore, the first preset model is trained according to the processed training set, and the process of obtaining the first AI prediction model is specifically to convert the model parameters into a binary chromosome gene string of the chromosome length in the form of binary coding, wherein one chromosome gene string corresponds to one individual. Use the accuracy of the support vector machine classification as the fitness, calculate the accuracy of the support vector machine classification corresponding to all individuals in the population; select a number of individuals from the population based on the accuracy, and form a new population; perform mutation and crossover operations on the new population based on the probability of the crossover and the probability of the mutation, so as to update the evolution of the new population; after reaching the maximum evolutionary generation, stop the iterative evolution, and use the individual with the highest accuracy in the iterative process as the target individual; use the target model parameters corresponding to the target individual to replace the current model parameters of the first preset model, and obtain the first AI prediction model. It should be noted that, assuming that the population size after initialization is set to 20, the maximum evolutionary generation is set to 100, the chromosome length is set to 40, the probability of crossover is set to 0.7, the probability of mutation is set to 0.0175, the range of the model coefficient c is set to [0, 100], and the model parameters are converted into a binary chromosome gene string with a length of 40 in the form of binary coding. The accuracy calculation formula is: (number of correctly classified samples / total number of samples in the test set)*100%. The accuracy of the support vector machine classification corresponding to all individuals in the entire population can be obtained through this calculation formula. Whether the sample classification is correct can be known by comparing the classification predicted by the model with the classification in the historical data. The accuracy of the support vector machine classification corresponds to the fitness of the individual, and the higher the fitness, the easier it is for the corresponding individual to be screened. Therefore, the individuals screened out in this embodiment are excellent individuals with a higher accuracy of the corresponding support vector machine classification, and the size of the new population composed of these excellent individuals is also 20. The crossover probability is set to 0.7, indicating that about 70% of the individuals in the population will participate in the crossover. The two individuals participating in the crossover randomly exchange part of the binary sequence to generate two new individuals. The new individuals and the individuals that did not participate in the crossover form a new population, and the population size remains 20. The mutation probability is set to 0.0175. It can be considered that about 1.75% of all binary bits of the individuals in the population will mutate. The mutated bits will be reversed. The mutated individuals and the individuals that did not participate in the mutation form the next generation population, and the population size remains unchanged. The new population is updated and evolved through the above operations. After reaching the maximum evolutionary generation, the iterative evolution is stopped, and the individual with the highest accuracy in the iterative process is used as the target individual, that is, the individual with the highest accuracy in all iterated populations. The individual is obtained by converting the model parameters, so the target individual corresponds to a target model parameter. The target model parameter corresponding to the target individual is used to replace the current model parameter of the first preset model to obtain the first AI prediction model.

[0071] Furthermore, the training process of the second AI prediction model is to preprocess the training set to obtain a processed training set; determine the correspondence between the battery capacity and the battery voltage based on the processed training set, and extract the capacity difference characteristics and discharge characteristics associated with the battery life from the processed training set according to the correspondence; extract other characteristics associated with the battery life from the processed training set, and the other characteristics include at least the charging temperature and the battery internal resistance within a unit charge and discharge cycle; determine the historical battery life corresponding to each group of capacity difference characteristics, discharge characteristics and other characteristics according to the processed training set; and perform multiple iterative training on the second preset model according to each group of capacity difference characteristics, discharge characteristics, other characteristics and corresponding historical battery life to obtain a second AI prediction model.

[0072] It should be noted that the capacity difference, such as △Q 100-10 (V) = Q 10 (V)-Q 100 (V), where Q 10 (V) represents the corresponding relationship between battery capacity and battery voltage in the tenth charge and discharge cycle, Q 100 (V) represents the correspondence between the battery capacity and the battery voltage in the 100th charge and discharge cycle, and the discharge characteristics such as the maximum discharge capacity under multiple charge and discharge cycles and the average charging time per unit charge and discharge cycle. In this embodiment, it is necessary to further extract other features associated with the battery life, such as the charging temperature and the internal resistance of the battery per unit charge and discharge cycle. There is a correlation between the above features and the battery life. The historical battery life corresponding to each group of capacity difference features, discharge features and other features can be determined through historical data, and then each group of capacity difference features, discharge features and other features are placed in a preset model for training, and the training results are compared with the actual historical battery life. The model parameters are updated based on each comparison result to achieve multiple iterative training, thereby obtaining a second AI prediction model.

[0073] In this embodiment, the preprocessing process of the training set corresponding to the second AI prediction model is to identify the first data position corresponding to the maximum battery capacity from the data header of the training set, and remove the data before the first data position; identify the second data position corresponding to the first occurrence of the battery voltage greater than the preset voltage from the data tail of the training set, and remove the data after the second data position; obtain the charging and discharging data under each cycle based on the training set after the head and tail data are removed; and interpolate the charging and discharging data to obtain the processed training set.

[0074] It should be noted that in this embodiment, the head data, intermediate data and tail data of the training set are divided according to the time node of obtaining the data. For example, the data obtained first is used as the head data, and the data obtained last is used as the tail data. Then, part of the head data and the tail data are removed in the above manner. The preset voltage can be set to 2v, or it can be adjusted accordingly according to the actual data processing requirements. This is not limited in this embodiment. After the above-mentioned head and tail data are removed, each segment of data after the head and tail data are removed corresponds to a charge and discharge cycle, and the charge and discharge data under each cycle can be accurately obtained. The charge and discharge data is essentially a collection of charge and discharge data of each data point in the charge and discharge process. The method adopted in this embodiment is interpolation processing. The specific methods that can be adopted for interpolation processing include nearest neighbor interpolation, linear difference and cubic spline interpolation.

[0075] Step S30: Collecting actual operating data during the battery operation process.

[0076] The above model is constructed based on a large amount of historical data and parameters. In practical applications, it is necessary to collect the actual operating data of the battery in real time. The actual operating data includes changes in single cell capacity, number of cycles, charge and discharge temperature changes, charge and discharge power changes, group voltage changes, group capacity changes, voltage difference within the group, internal resistance difference within the group, temperature difference within the group, ambient temperature and basic parameters related to the battery cell.

[0077] Step S40: Input the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model.

[0078] In practical applications, the collected actual operating data can be input into two AI prediction models respectively to obtain prediction results. The first prediction result is the predicted fault type of the battery pack, that is, the possible fault type of the battery pack in the future is predicted based on the current operating data. The second prediction result is the remaining life of the battery pack, that is, the possible remaining life of the battery in the future is predicted based on the current operating data.

[0079] Step S50: determining a target prediction result according to the first prediction result and the second prediction result.

[0080] After obtaining the prediction results in two different dimensions, the two prediction results are further integrated in this embodiment to form a corresponding relationship, so as to obtain the target prediction result, for example, simultaneously obtaining the possible failure type of the battery and the estimated remaining life, thereby preventing risks from two aspects.

[0081] In this embodiment, the negative feedback data set, cell parameter set, historical operation data set and historical fault data of the battery are used as training sets, and model training and verification are performed based on the training sets to obtain a first AI prediction model and a second AI prediction model, collect actual operation data during the operation of the battery, and input the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model, and determine a target prediction result based on the first prediction result and the second prediction result. In this way, the faults of the energy storage battery pack can be predicted in advance, risks and hidden dangers can be eliminated, and the probability of faults and accidents can be reduced.

[0082] refer to Figure 2 , Figure 2 This is a flow chart of a second embodiment of a battery fault prediction method based on big data and AI algorithm of the present invention.

[0083] Based on the first embodiment above, in the battery fault prediction method based on big data and AI algorithm in this embodiment, step S50 further includes:

[0084] Step S501: determining a predicted fault type of the battery according to the first prediction result.

[0085] Step S502: Determine the predicted remaining life of the battery according to the second prediction result.

[0086] Step S503: constructing a first feature vector corresponding to the predicted fault type and a second feature vector corresponding to the predicted remaining life respectively.

[0087] After obtaining the predicted fault type and predicted remaining life of the battery respectively through the first AI prediction model and the second AI prediction model, the corresponding feature vector is further constructed in this embodiment, the predicted fault type corresponds to the first feature vector, and the predicted remaining life corresponds to the second feature vector. The feature vector is, for example, [feature1, feature2, feature3], feature1, feature2, feature3 correspond to the predicted fault type or the predicted remaining life.

[0088] Step S504: cross-merge the first feature vector and the second feature vector to obtain a target feature vector.

[0089] When fusing the first eigenvector with the second eigenvector, it is necessary to first determine the similarity between the first eigenvector and the second eigenvector. Specifically, the cosine similarity or Euclidean distance can be used to calculate the similarity. The similarity determines the fusion method. Feature splicing is used for low similarity to retain more information. Weighted fusion is used for medium similarity to balance the contributions of different vectors. Selective fusion is used for high similarity to avoid redundant information. The specific similarity calculation process and fusion method belong to conventional technology and will not be described in detail here. The specific similarity division can be based on a preset similarity range. Low, medium and high similarities have their corresponding preset ranges respectively. If they are within the corresponding preset ranges, the specific similarity level is determined.

[0090] Step S505: obtaining a target prediction result based on the target feature vector.

[0091] The elements in the target feature vector are specific prediction results. After obtaining the target feature vector, the target prediction result can be obtained. The target prediction result in this embodiment includes different battery failure types and corresponding remaining lifespans.

[0092] This embodiment determines the predicted fault type of the battery according to the first prediction result; determines the predicted remaining life of the battery according to the second prediction result; constructs a first feature vector corresponding to the predicted fault type and a second feature vector corresponding to the predicted remaining life respectively; determines the similarity between the first feature vector and the second feature vector; determines a fusion method according to the similarity; cross-fuses the first feature vector and the second feature vector according to the fusion method to obtain a target feature vector; obtains a target prediction result based on the target feature vector, and the target prediction result is used to represent the remaining life of the battery corresponding to different fault types. Fusion of the prediction results by different fusion methods can further improve the accuracy of battery fault prediction.

[0093] Reference Figure 3 , Figure 3 This is a structural block diagram of the first embodiment of the battery fault prediction device based on big data and AI algorithm of the present invention.

[0094] like Figure 3 As shown, the battery fault prediction device based on big data and AI algorithm proposed in the embodiment of the present invention includes:

[0095] The acquisition module 10 is used to acquire a negative feedback data set, a cell parameter set, a historical operation data set and a historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set and the historical fault data of the battery as a training set.

[0096] The construction module 20 is used to perform model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions.

[0097] The acquisition module 10 is used to collect actual operation data of the battery during operation.

[0098] The prediction module 30 is used to input the actual operation data into the first AI prediction model and the second AI prediction model respectively, to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model.

[0099] The fusion module 40 is used to determine a target prediction result according to the first prediction result and the second prediction result.

[0100] In this embodiment, the negative feedback data set, cell parameter set, historical operation data set and historical fault data of the battery are used as training sets, and model training and verification are performed based on the training sets to obtain a first AI prediction model and a second AI prediction model, collect actual operation data during the operation of the battery, and input the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model, and determine a target prediction result based on the first prediction result and the second prediction result. In this way, the faults of the energy storage battery pack can be predicted in advance, risks and hidden dangers can be eliminated, and the probability of faults and accidents can be reduced.

[0101] In some embodiments, the first AI prediction model is used to predict the failure type of the battery, and the construction module 20 is used to identify the data point corresponding to the maximum resistance in the training set; the training set is smoothed by a preset function with the data point as the midpoint to obtain a processed training set; the population size, maximum evolutionary generation, chromosome length, probability of crossover, probability of mutation and model penalty parameter range are initialized, and after completing the initialization, the first preset model is trained based on the processed training set to obtain a first AI prediction model.

[0102] In some embodiments, the construction module 20 is used to convert the model parameters into a binary chromosome gene string of the chromosome length in the form of binary coding, wherein one chromosome gene string corresponds to one individual; use the accuracy of support vector machine classification as the fitness to calculate the accuracy of support vector machine classification corresponding to all individuals in the population; select a number of individuals from the population based on the accuracy to form a new population; perform mutation and crossover operations on the new population based on the probability of crossover and the probability of mutation to update the evolution of the new population; after reaching the maximum evolutionary generation, stop the iterative evolution, and take the individual with the highest accuracy in the iterative process as the target individual; use the target model parameters corresponding to the target individual to replace the current model parameters of the first preset model to obtain the first AI prediction model.

[0103] In some embodiments, the second preset model is used to predict the life of the battery, and the construction module 20 is used to pre-process the training set to obtain a processed training set; determine the correspondence between the battery capacity and the battery voltage based on the processed training set, and extract the capacity difference characteristics and discharge characteristics associated with the battery life from the processed training set according to the correspondence; extract other characteristics associated with the battery life from the processed training set, and the other characteristics at least include the charging temperature and the battery internal resistance within a unit charge and discharge cycle; determine the historical battery life corresponding to each group of capacity difference characteristics, discharge characteristics and other characteristics according to the processed training set; and perform multiple iterative training on the second preset model according to each group of capacity difference characteristics, discharge characteristics, other characteristics and corresponding historical battery life to obtain a second AI prediction model.

[0104] In some embodiments, the construction module 20 is used to identify the first data position corresponding to the maximum battery capacity from the data header of the training set, and remove the data before the first data position; identify the second data position corresponding to the first occurrence of a battery voltage greater than a preset voltage from the data tail of the training set, and remove the data after the second data position; obtain the charge and discharge data for each cycle based on the training set after the head and tail data are removed; and interpolate the charge and discharge data to obtain a processed training set.

[0105] In some embodiments, the fusion module 40 is used to determine the predicted fault type of the battery according to the first prediction result; determine the predicted remaining life of the battery according to the second prediction result; respectively construct a first feature vector corresponding to the predicted fault type and a second feature vector corresponding to the predicted remaining life; cross-fuse the first feature vector with the second feature vector to obtain a target feature vector; obtain a target prediction result based on the target feature vector, and the target prediction result is used to represent the remaining life of the battery corresponding to different fault types.

[0106] In some embodiments, the fusion module 40 is used to determine the similarity between the first feature vector and the second feature vector; determine a fusion method according to the similarity; and cross-fuse the first feature vector and the second feature vector according to the fusion method to obtain a target feature vector.

[0107] An embodiment of the present application also provides a battery fault prediction device based on big data and AI algorithm, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus, and the memory is used to store computer programs; the processor is used to implement the above-mentioned battery fault prediction method based on big data and AI algorithm when executing the program stored in the memory.

[0108] The communication bus mentioned in the battery fault prediction device based on big data and AI algorithm can be a peripheral component interconnect standard (English: Peripheral Component Interconnect, referred to as: PCI) bus or an extended industrial standard architecture (English: Extended Industry Standard Architecture, referred to as: EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0109] The communication interface is used for communication between the above-mentioned battery fault prediction device based on big data and AI algorithm and other devices.

[0110] The memory may include a random access memory (RAM) or a non-volatile memory (NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0111] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, and discrete hardware components.

[0112] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.

[0113] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0114] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0115] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0116] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present invention. In specific applications, technicians in this field can make settings as needed, and the present invention does not limit this.

[0117] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0118] In addition, for technical details not described in detail in this embodiment, reference can be made to the battery fault prediction method based on big data and AI algorithm provided in any embodiment of the present invention, which will not be repeated here.

[0119] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0120] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0121] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0122] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

[0123] It is understandable that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts in the above method.

Claims

1. A battery failure prediction method based on big data and AI algorithm, characterized in that: The battery failure prediction method based on big data and AI algorithm includes: Acquire a negative feedback data set, a cell parameter set, a historical operation data set, and historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set, and the historical fault data of the battery as a training set; Performing model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions, wherein the second AI preset model is used to predict the life of the battery, and preprocessing the training set to obtain a processed training set; Determining a correspondence between battery capacity and battery voltage based on the processed training set, and extracting capacity difference features and discharge features associated with battery life from the processed training set according to the correspondence; Extracting other features associated with battery life from the processed training set, the other features at least including charging temperature and battery internal resistance within a unit charge and discharge cycle; Determine the historical battery life corresponding to each group of capacity difference features, discharge features and other features according to the processed training set; Iteratively training the second preset model multiple times according to each group of capacity difference characteristics, discharge characteristics, other characteristics, and corresponding historical battery life to obtain a second AI prediction model; Collect actual operating data during battery operation; Inputting the actual operation data into the first AI prediction model and the second AI prediction model respectively to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model; A target prediction result is determined according to the first prediction result and the second prediction result.

2. The battery failure prediction method based on big data and AI algorithm according to claim 1, characterized in that: The first AI prediction model is used to predict the fault type of the battery. The battery fault prediction method based on big data and AI algorithm includes: Identifying the data point corresponding to the maximum resistance in the training set; Taking the data point as the midpoint, smoothing the training set by a preset function to obtain a processed training set; Initialize the population size, maximum evolutionary generations, chromosome length, probability of crossover, probability of mutation, and model penalty parameter range, and after completing the initialization, train the first preset model based on the processed training set to obtain a first AI prediction model.

3. The battery failure prediction method based on big data and AI algorithm as claimed in claim 2, characterized in that: The step of training the first preset model based on the processed training set to obtain a first AI prediction model includes: Converting the model parameters into binary chromosome gene strings of the chromosome length in the form of binary coding, wherein one chromosome gene string corresponds to one individual; Use the accuracy of support vector machine classification as fitness and calculate the accuracy of support vector machine classification corresponding to all individuals in the population; Selecting a number of individuals from the population based on the accuracy rate and forming a new population; Performing mutation and crossover operations on the new population based on the probability of the crossover and the probability of the mutation to update the evolution of the new population; After reaching the maximum number of evolutionary generations, the iterative evolution is stopped, and the individual with the highest accuracy in the iterative process is taken as the target individual; The target model parameters corresponding to the target individual are used to replace the current model parameters of the first preset model to obtain a first AI prediction model.

4. The battery failure prediction method based on big data and AI algorithm according to claim 1, characterized in that: The preprocessing of the training set to obtain a processed training set includes: Identify a first data position corresponding to a maximum battery capacity from a data header of the training set, and remove data before the first data position; Identify from the tail of the data of the training set the second data position corresponding to the first occurrence of a battery voltage greater than a preset voltage, and remove the data after the second data position; Based on the training set after removing the first and last data, the charge and discharge data of each cycle are obtained; The charge and discharge data are interpolated to obtain a processed training set.

5. The battery failure prediction method based on big data and AI algorithm according to claim 1, characterized in that: The determining a target prediction result according to the first prediction result and the second prediction result includes: Determining a predicted fault type of the battery according to the first prediction result; Determining a predicted remaining life of the battery according to the second prediction result; Constructing a first feature vector corresponding to the predicted fault type and a second feature vector corresponding to the predicted remaining life respectively; Cross-merge the first feature vector and the second feature vector to obtain a target feature vector; A target prediction result is obtained based on the target feature vector, and the target prediction result is used to represent the remaining life of the battery corresponding to different fault types.

6. The battery failure prediction method based on big data and AI algorithm as claimed in claim 5, characterized in that: The cross-fusion of the first feature vector and the second feature vector to obtain a target feature vector includes: determining a similarity between the first feature vector and the second feature vector; Determine a fusion method according to the similarity; The first feature vector and the second feature vector are cross-fused according to the fusion method to obtain a target feature vector.

7. A battery failure prediction device based on big data and AI algorithm, characterized in that: The battery failure prediction device based on big data and AI algorithm includes: An acquisition module, used to acquire a negative feedback data set, a cell parameter set, a historical operation data set, and historical fault data of a battery, and use the negative feedback data set, the cell parameter set, the historical operation data set, and the historical fault data of the battery as a training set; A construction module is used to perform model training and verification based on the training set to obtain a first AI prediction model and a second AI prediction model, wherein the first AI prediction model and the second AI prediction model respectively perform predictions in different dimensions, wherein the second AI preset model is used to predict the life of the battery, and the training set is preprocessed to obtain a processed training set; Determining a correspondence between battery capacity and battery voltage based on the processed training set, and extracting capacity difference features and discharge features associated with battery life from the processed training set according to the correspondence; Extracting other features associated with battery life from the processed training set, the other features at least including charging temperature and battery internal resistance within a unit charge and discharge cycle; Determine the historical battery life corresponding to each group of capacity difference features, discharge features and other features according to the processed training set; Iteratively training the second preset model multiple times according to each group of capacity difference characteristics, discharge characteristics, other characteristics, and corresponding historical battery life to obtain a second AI prediction model; The acquisition module is used to collect actual operation data during the operation of the battery; A prediction module, used to input the actual operation data into the first AI prediction model and the second AI prediction model respectively, to obtain a first prediction result output by the first AI prediction model and a second prediction result output by the second AI prediction model; A fusion module is used to determine a target prediction result based on the first prediction result and the second prediction result.

8. The battery failure prediction device based on big data and AI algorithm according to claim 7, characterized in that: The building block is used to identify the data point corresponding to the maximum resistance in the training set; Taking the data point as the midpoint, smoothing the training set by a preset function to obtain a processed training set; Initialize the population size, maximum evolutionary generations, chromosome length, probability of crossover, probability of mutation, and model penalty parameter range, and after completing the initialization, train the first preset model based on the processed training set to obtain a first AI prediction model.

9. A battery failure prediction device based on big data and AI algorithm, characterized in that: The battery fault prediction device based on big data and AI algorithm includes: a memory, a processor, and a battery fault prediction program based on big data and AI algorithm stored in the memory and executable on the processor. The battery fault prediction program based on big data and AI algorithm is configured to implement the steps of the battery fault prediction method based on big data and AI algorithm as described in any one of claims 1 to 6.

Images