Battery management method, system and device based on neural network

Abstract

The invention discloses a neural network-based battery management method, system and device, and the method comprises the steps: obtaining the real-time battery data of a battery, and inputting the real-time battery data into a preset fault diagnosis model; receiving a fault probability output by a preset fault diagnosis model according to the real-time battery data; judging whether the fault probability is greater than a preset threshold; and if the fault probability is greater than a preset threshold value, generating detailed fault information according to the real-time battery data and the fault probability, and retrieving an optimal repair strategy according to the detailed fault information to repair the battery. According to the method, the preset fault diagnosis model is pre-trained on the neurons, the neural network detects the real-time battery data through the preset fault diagnosis model to determine whether the battery has a fault, and if the fault occurs, the corresponding repair strategy is retrieved for repair; the overall performance and the service life of the battery are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and particularly to a battery management method, system and device based on a neural network. Background Art

[0002] With the rapid development of photography and live streaming, various shooting devices have been widely used. As a key power supply component, the effectiveness of battery management is particularly important. Therefore, battery management methods have emerged. Battery management methods can play an active role in various aspects such as the life, safety, performance, and efficiency of the battery, and comprehensively improve the use value of the battery.

[0003] However, traditional battery management methods mainly rely on simple power monitoring and basic charge and discharge control strategies. Although they can ensure the basic use of the battery to a certain extent, there are many limitations. For example, they cannot accurately predict what faults will occur during the use of the battery, and cannot handle the faults in time when they occur, resulting in a decline in the overall performance of the battery and a shortening of its service life.

[0004] Therefore, there is still an urgent need for a battery management method that can improve the overall performance and service life of the battery. Summary of the Invention

[0005] The main object of the present invention is to propose a battery management method, system and device based on a neural network to solve the problems of the decline in the overall performance of existing defective batteries and the shortening of their service life.

[0006] To achieve the above object, the present invention proposes a battery management method based on a neural network. The battery management method based on a neural network includes:

[0007] Obtain real-time battery data of the battery, and input the real-time battery data into a preset fault diagnosis model;

[0008] Receive the fault probability output by the preset fault diagnosis model according to the real-time battery data;

[0009] Judge whether the fault probability is greater than a preset threshold;

[0010] If the fault probability is greater than the preset threshold, generate detailed fault information according to the real-time battery data and the fault probability, and retrieve the best repair strategy according to the detailed fault information to repair the battery.

[0011] In some embodiments, before obtaining the real-time battery data of the battery, it further includes:

[0012] Obtain historical battery data of the battery;

[0013] Filter the historical battery data to obtain preprocessed battery data;

[0014] According to the marking instruction, mark the preprocessed battery data to obtain a plurality of marked battery data, wherein the marked battery data includes normal battery data and abnormal battery data;

[0015] Train an initial fault diagnosis model according to the plurality of marked battery data to obtain the preset fault diagnosis model.

[0016] In some embodiments, after receiving the fault probability output by the preset fault diagnosis model according to the real-time battery data, it further includes:

[0017] Update the historical battery data according to the real-time battery data, and perform the step of filtering the historical battery data to obtain preprocessed battery data.

[0018] In some embodiments, the repairing the battery according to the best repair strategy retrieved according to the detailed fault information includes:

[0019] Analyze the detailed fault information to determine the fault type;

[0020] Retrieve a preset repair strategy library according to the fault type to obtain a plurality of repair strategies;

[0021] Obtain the success rate of each repair strategy, and sort the success rates of the repair strategies in descending order;

[0022] Determine the repair strategy corresponding to the highest success rate as the best repair strategy;

[0023] Repair the battery according to the best repair strategy.

[0024] In some embodiments, after repairing the battery according to the best repair strategy, it further includes:

[0025] Obtain the current battery state of the battery;

[0026] Compare the current battery state with a preset normal state to determine whether the battery is repaired successfully;

[0027] If the battery is repaired successfully, update the success rate of the best repair strategy and record the information that the battery is repaired successfully.

[0028] In some embodiments, after determining whether the battery is repaired successfully, it further includes:

[0029] If the battery is not repaired successfully, generate a maintenance suggestion according to the fault type and display the maintenance suggestion; or,

[0030] If the battery is not successfully repaired, obtain a second repair strategy and repair the battery according to the second repair strategy, where the second repair strategy is the repair strategy corresponding to the success rate ranked second in the descending order.

[0031] In some embodiments, the battery includes a plurality of single cells; the battery management method based on a neural network further includes:

[0032] Collect the current voltages of the single cells;

[0033] Perform an average operation on the current voltages of the single cells to obtain the current average voltage;

[0034] Compare the current voltages of the single cells with the current average voltage;

[0035] If the current voltage of the single cell is greater than the current average voltage, control the equalization circuit to discharge the single cell until the current voltage of the single cell is equal to the current average voltage;

[0036] If the current voltage of the single cell is less than the current average voltage, control the equalization circuit to charge the single cell until the current voltage of the single cell is equal to the current average voltage.

[0037] In some embodiments, the battery management method based on a neural network further includes:

[0038] Detect the current parameters of the charging current and obtain the current battery state of the battery;

[0039] Adjust the current parameters according to the current battery state to obtain the optimal current parameters;

[0040] Adjust the charging current according to the optimal current parameters and input the charging current into the battery.

[0041] The present invention also provides a battery management system based on a neural network. The battery management system based on a neural network includes a neural chip and a battery. The neural chip is used to configure and run a neural network; the battery management system based on a neural network can execute the battery management method based on a neural network described in any one of the above.

[0042] The present invention also provides a battery management device based on a neural network, including:

[0043] At least one processor; and,

[0044] A memory communicatively connected to the at least one processor; wherein,

[0045] The memory stores instructions that are executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the neuron network-based battery management method described in any one of the above.

[0046] In the present invention, a preset fault diagnosis model is pre-trained on neurons; the neuron network obtains real-time battery data of the battery in real time, and then detects the real-time battery data through the preset fault diagnosis model to determine the fault probability of the real-time battery data having a fault, and compares the fault probability with a preset threshold; thereby determining whether the battery has a fault. If a fault occurs, a corresponding repair strategy will also be retrieved for repair to improve the overall performance and service life of the battery; and by introducing a neuron network in battery management, the effectiveness and intelligence of battery management are also improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0048] Figure 2 is another schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0049] Figure 3 is another schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0050] Figure 4 is another schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0051] Figure 5 is another schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0052] Figure 6 is another schematic flowchart of the neuron network-based battery management method in an embodiment of the present invention;

[0053] Figure 7 is a schematic structural diagram of the neuron network-based battery management system according to the embodiment solution of the present invention;

[0054] Figure 8 is a schematic structural diagram of the neuron network-based battery management device according to the embodiment solution of the present invention.

[0055] The realization, functional characteristics and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] The following will clearly and completely describe the solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the protection scope of the present invention.

[0057] It should be noted that all directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0058] It should also be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or there may be an intermediate element at the same time. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time.

[0059] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0060] To achieve the above object, the present invention proposes a battery management method based on a neural network. The battery management method based on a neural network includes:

[0061] Step S110, obtaining real-time battery data of the battery and inputting the real-time battery data into a preset fault diagnosis model;

[0062] Step S120, receiving the fault probability output by the preset fault diagnosis model according to the real-time battery data;

[0063] Step S130, determining whether the fault probability is greater than a preset threshold;

[0064] Step S140, if the fault probability is greater than the preset threshold, generating detailed fault information according to the real-time battery data and the fault probability, and retrieving the best repair strategy according to the detailed fault information to repair the battery.

[0065] In this embodiment, referring to Figure 1 andFigure 7 , a battery management method based on a neural network is applied to a battery management system based on a neural network; the battery management system based on a neural network is used to manage a battery to improve the overall performance and service life of the battery. Among them, the neural network is a computational model that mimics the structure and function of a biological neural network. It is composed of a large number of neurons connected to each other and can achieve various tasks through learning and processing of data. The battery management system based on a neural network includes a neural chip and a battery. The neural chip is used to configure and run the neural network. The neural network learns by obtaining battery data of a large number of batteries, so as to realize the processing of real-time battery data to achieve various battery management tasks. In this embodiment, the execution subject of the method steps is the neural network configured on the neural chip.

[0066] It can be understood that the neural network may include neurons, synaptic connections, and network structures. A neuron is the basic processing unit of the neural network, similar to a biological neuron; each neuron receives multiple input signals, sums these signals after weighting, and processes them through an activation function to generate an output signal. The synaptic connection is the channel for information interaction between neurons. Neurons transmit information through synaptic connections, and each connection has a weight value, which is used to represent the importance of the connection for information transmission; the weights are continuously adjusted during the learning process of the neural network to optimize the performance of the network. The network structure is composed of multiple neurons in a certain hierarchical and connection manner; common structures include feedforward neural networks, recurrent neural networks, convolutional neural networks, etc.

[0067] The battery can be connected to the neural chip, and the neural chip can obtain the battery data of the battery. The neural network is configured on the neural chip, and the neural network can obtain the battery data through the neural chip. The neural network obtains the real-time battery data of the battery. When there is a pre-trained preset fault diagnosis model saved in the neural network; the neural network inputs the real-time battery data into the preset fault diagnosis model. At this time, the preset fault diagnosis model will process the real-time battery data.

[0068] After the preset fault diagnosis model processes the real-time battery data, it will output the processing result. Among them, what the preset fault diagnosis model outputs is the fault probability corresponding to the real-time battery data. After the preset fault diagnosis model outputs the fault probability, the neural network can receive the fault probability output by the preset fault diagnosis model according to the real-time battery data.

[0069] After the neural network receives the fault probability, it will judge the fault probability to determine whether the fault probability is greater than a preset threshold. If the fault probability is greater than the preset threshold, it means that the battery has a fault. If the fault probability is less than the preset threshold, it means that the battery has no fault.

[0070] If the neural network determines that the failure probability is greater than the preset threshold, it will generate detailed failure information based on the real-time battery data and the failure probability, and then display the detailed failure information; it will also retrieve the best repair strategy according to the detailed failure information, so as to repair the battery according to the best repair strategy.

[0071] If the neural network determines that the failure probability is less than the preset threshold, it will continue to monitor the failure probability. That is, it will continue to execute the steps of obtaining the real-time battery data of the battery and inputting the real-time battery data into the preset fault diagnosis model. It realizes that the battery can be detected in time when a failure occurs, thus improving the overall performance and service life of the battery.

[0072] Through this embodiment, a preset fault diagnosis model is pre-trained on the neuron; the neural network obtains the real-time battery data of the battery in real time, then detects the real-time battery data through the preset fault diagnosis model, determines the failure probability of the real-time battery data, and compares the failure probability with the preset threshold; thus determining whether the battery fails. If a failure occurs, it will also retrieve the corresponding repair strategy for repair to improve the overall performance and service life of the battery; and by introducing the neural network into the battery management, the effectiveness and intelligence of the battery management are also improved.

[0073] In some embodiments, before obtaining the real-time battery data of the battery as described above, it further includes:

[0074] Step S150, obtaining the historical battery data of the battery;

[0075] Step S151, performing filtering processing on the historical battery data to obtain preprocessed battery data;

[0076] Step S152, marking the preprocessed battery data according to the marking instruction to obtain a plurality of marked battery data, where the marked battery data includes normal battery data and abnormal battery data;

[0077] Step S153, training the initial fault diagnosis model according to the plurality of marked battery data to obtain the preset fault diagnosis model.

[0078] In this embodiment, referring to Figure 2 , before the neural network executes step S110, it is necessary to train the preset fault diagnosis model first. The neural network can continuously obtain the real-time battery data of the battery; when the continuous acquisition time is long, a large amount of battery data will be obtained, and the neural network can save these large amounts of battery data to form historical battery data; that is, the historical battery data includes a plurality of battery data. When the neural network trains the preset fault diagnosis model; the neural network can obtain the historical battery data of the battery.

[0079] After the neural network obtains the historical battery data, it will first perform filtering processing on the historical battery data. For example, to remove the abnormal noise in the historical battery data, the moving average filtering algorithm or the Kalman filtering algorithm can be used to operate on the historical battery data, so as to remove the abnormal noise in the historical battery data. After the neural network performs filtering processing on the historical battery data, the preprocessed battery data can be obtained.

[0080] After the neural network obtains the preprocessed battery data, it is also necessary to label the preprocessed battery data. The neural network labels the preprocessed battery data according to the labeling instruction, and obtains multiple labeled battery data. Among them, the labeled battery data includes normal battery data and abnormal battery data. The labeling instruction can be generated according to the labeling information input by the user. The labeling information input by the user can be classified and labeled according to different diagnostic requirements. For example, the battery data in the normal operating state is labeled as "normal battery data", and the battery data in abnormal situations such as overvoltage, undervoltage, overcurrent, overheating, and short circuit is labeled as "abnormal battery data". In particular, the abnormal battery data can be further subdivided, that is, it is labeled as the corresponding fault category according to abnormal situations such as overvoltage, undervoltage, overcurrent, overheating, and short circuit. That is, the abnormal battery data can include overvoltage battery data, undervoltage battery data, overcurrent battery data, overheating battery data, short circuit battery data, etc.

[0081] After the neural network obtains multiple labeled battery data, it can train the initial fault diagnosis model according to the multiple labeled battery data, so as to obtain the preset fault diagnosis model. For example, a large amount of normal battery data and abnormal battery data are input into the initial fault diagnosis model, and the initial fault diagnosis model can learn and train according to the large amount of normal battery data and abnormal battery data, so as to know which battery data is normal and which battery data is abnormal; finally, the preset fault diagnosis model is obtained.

[0082] In some embodiments, after receiving the fault probability output by the preset fault diagnosis model according to the real-time battery data, it further includes:

[0083] Updating the historical battery data according to the real-time battery data, and performing the step of filtering the historical battery data to obtain the preprocessed battery data.

[0084] In this embodiment, after the neural network executes step S120, it will also update the historical battery data. After the neural network receives the failure probability output by the preset failure diagnosis model based on the real-time battery data, it will also update the historical battery data. Among them, updating the historical battery data according to the real-time battery data can be to add the real-time battery data to the historical battery data; it can also be to replace the real-time battery data with the battery data with the earliest acquisition time in the historical battery data. After updating the historical battery data, the neural network will also perform the step of filtering the historical battery data to obtain the preprocessed battery data. That is, retrain the preset failure diagnosis model to update the preset failure diagnosis model. For example, as time goes by and new battery data accumulates continuously, the neural network will periodically retrain or online update the preset failure diagnosis model. This can adopt the method of incremental learning, that is, only train on the new battery data and fine-tune some weights of the original model at the same time; or use a combination of new battery data and old battery data for a complete retraining, while keeping the structure of the model unchanged or adjusting the structure according to the performance evaluation results.

[0085] In some embodiments, retrieving the best repair strategy to repair the battery according to the detailed failure information as described above includes:

[0086] Step S160, analyzing the detailed failure information to determine the failure type;

[0087] Step S161, retrieving the preset repair strategy library according to the failure type to obtain multiple repair strategies;

[0088] Step S162, obtaining the success rate of each repair strategy and sorting the success rates of each repair strategy in descending order;

[0089] Step S163, determining the repair strategy corresponding to the highest success rate as the best repair strategy;

[0090] Step S164, repairing the battery according to the best repair strategy.

[0091] In this embodiment, referring to Figure 3 , when the neural network executes step S140, it retrieves the best repair strategy from the preset repair strategy library. The preset repair strategy library can be pre-configured by the user on the neural network, and one failure type can correspond to multiple repair strategies. Each repair strategy can have the success rate of repairing this failure type. The neural network will first analyze the detailed failure information to determine the failure type. For example, the failure type can include overvoltage failure, undervoltage failure, overcurrent failure, overtemperature failure, short circuit failure, and so on.

[0092] After the neural network determines the fault type, it can retrieve the preset repair strategy library according to the fault type to obtain multiple repair strategies. For example, if the fault type is overvoltage fault, there may be many reasons for the overvoltage fault, such as overvoltage during charging, abnormal overvoltage of the load, overvoltage caused by environmental factors, etc. For each reason of the overvoltage fault, there can be one or more repair strategies.

[0093] After the neural network obtains multiple repair strategies, it can obtain the success rate of each repair strategy and sort the success rates of each repair strategy in descending order. Among them, the success rate of each repair strategy can be obtained according to the repair simulation. For example: the simulation shows that the battery has a total of 10 overvoltage faults. The preset repair strategy library is retrieved to obtain repair strategy a, repair strategy b, and repair strategy c; use repair strategy a, repair strategy b, and repair strategy c to repair the 10 overvoltage faults of the battery respectively; if repair strategy a successfully repairs 9 times, that is, the success rate of repair strategy a is 90%; similarly, if repair strategy b successfully repairs 10 times, that is, the success rate of repair strategy b is 100%, and repair strategy c successfully repairs 5 times, that is, the success rate of repair strategy c is 50%. Of course, the success rate of each repair strategy can be obtained from the number of times the repair strategy has been used historically and the number of times the repair strategy has been successfully repaired.

[0094] After the neural network sorts the success rates of each repair strategy in descending order, it can determine the repair strategy corresponding to the highest success rate as the best repair strategy. That is, the repair strategy with the highest success rate is determined as the best repair strategy. Then, the neural network will repair the fault type of the battery according to the best repair strategy.

[0095] In some embodiments, after repairing the battery according to the best repair strategy as described above, it further includes:

[0096] Step S170, obtaining the current battery state of the battery;

[0097] Step S171, comparing the current battery state with the preset normal state to determine whether the battery is repaired successfully;

[0098] Step S172, if the battery is repaired successfully, update the success rate of the best repair strategy and record the information that the battery is repaired successfully.

[0099] In this embodiment, referring to Figure 4 , after the neural network executes step S164, it will also determine whether the battery is repaired successfully. The neural network will obtain the current battery state of the battery; then compare the current battery state with the preset normal state to determine whether the battery is repaired successfully. If the current battery state is consistent with the preset normal state, it means that the battery is repaired successfully. If the current battery state is inconsistent with the preset normal state, it means that the battery repair fails.

[0100] If the neural network determines that the battery repair is successful, it will update the success rate of the best repair strategy, that is, increase the success rate of the best repair strategy, and record the information of successful battery repair; it can also display the information of successful battery repair.

[0101] In some embodiments, after determining whether the battery is repaired successfully as described above, it further includes:

[0102] If the battery is not repaired successfully, generate a maintenance suggestion according to the fault type and display the maintenance suggestion; or,

[0103] If the battery is not repaired successfully, obtain a second repair strategy and repair the battery according to the second repair strategy, where the second repair strategy is the repair strategy corresponding to the second highest success rate in the descending order.

[0104] In this embodiment, after the neural network executes step S171, when the battery is not repaired successfully, there are two processing methods. If the neural network determines that the battery is not repaired successfully, it can generate a maintenance suggestion according to the fault type and display the maintenance suggestion; or it can obtain a second repair strategy and repair the battery twice according to the second repair strategy, where the second repair strategy is the repair strategy corresponding to the second highest success rate in the descending order.

[0105] In a preferred embodiment, it is possible to determine which processing method to execute when the battery is not repaired successfully according to the number of multiple repair strategies. For example: when the battery is not repaired successfully, each repair strategy in the descending order can be used in turn for repair. That is, repair the battery according to the second repair strategy, and if the battery is not repaired successfully; then repair the battery according to the third repair strategy. If the battery is not repaired successfully according to the third repair strategy; then repair the battery according to the fourth repair strategy. And so on until the last repair strategy in the order. Among them, the third repair strategy is the repair strategy corresponding to the third highest success rate in the descending order; the fourth repair strategy is the repair strategy corresponding to the fourth highest success rate in the descending order. If the battery is still not repaired successfully according to the last repair strategy in the order; then a maintenance suggestion will be generated according to the fault type and displayed.

[0106] In some embodiments, the battery includes a plurality of single cells; the battery management method based on the neural network as described above further includes:

[0107] Step S180, collect the current voltage of each single cell;

[0108] Step S181, perform a mean operation on the current voltages of each single cell to obtain the current average voltage;

[0109] Step S182, compare the current voltage of each single battery with the current average voltage;

[0110] Step S183, if the current voltage of the single battery is greater than the current average voltage, control the equalization circuit to discharge the single battery until the current voltage of the single battery is equal to the current average voltage;

[0111] Step S184, if the current voltage of the single battery is less than the current average voltage, control the equalization circuit to charge the single battery until the current voltage of the single battery is equal to the current average voltage.

[0112] In this embodiment, referring to Figure 5 , when the neural network executes the battery management method based on the neural network, it also includes the equalization voltage. The battery can be composed of multiple battery cells, and the multiple battery cells are also connected to the equalization circuit. The neural network can control the equalization circuit to charge or discharge the battery cells. If the voltages of multiple battery cells are not balanced, it will cause a large fluctuation in the total voltage of the battery, thereby affecting the overall performance and service life of the battery. At this time, the neural network can equalize the voltages of each battery cell to ensure the overall performance and service life of the battery. The neural network first collects the current voltages of each single battery; then performs a mean operation on the current voltages of each single battery to obtain the current average voltage. For example: add the current voltages of each single battery, and then divide by the number of single batteries to obtain the current average voltage.

[0113] After the neural network obtains the current average voltage, it will compare the current voltage of each single battery with the current average voltage to compare the magnitudes of the current voltage of each single battery and the current average voltage.

[0114] If the current voltage of the single battery is greater than the current average voltage, the neural network will control the equalization circuit to discharge the single battery until the current voltage of the single battery is equal to the current average voltage, and then stop discharging. If the current voltage of the single battery is less than the current average voltage, the neural network will control the equalization circuit to charge the single battery until the current voltage of the single battery is equal to the current average voltage, and then stop charging.

[0115] In some embodiments, the aforementioned battery management method based on the neural network further includes:

[0116] Step S190, detect the current parameters of the charging current and obtain the current battery state of the battery;

[0117] Step S191, adjust the current parameters according to the current battery state to obtain the optimal current parameters;

[0118] Step S192, adjust the charging current according to the optimal current parameters and input the charging current into the battery.

[0119] In this embodiment, referring to Figure 6 , when the neural network executes the battery management method based on the neural network, it further includes charge control. When the battery is charging, the neural network can first detect the current parameters of the charging current and obtain the current battery state of the battery. Then, the current parameters are adjusted according to the current battery state to obtain the optimal current parameters. Finally, the charging current is adjusted according to the optimal current parameters, and only after the adjustment is the charging current input into the battery.

[0120] The present invention pre-trains a preset fault diagnosis model on the neurons in advance; the neural network obtains the real-time battery data of the battery in real time, and then detects the real-time battery data through the preset fault diagnosis model to determine the fault probability of the real-time battery data having a fault, and compares the fault probability with a preset threshold; thereby determining whether the battery has a fault. If a fault occurs, a corresponding repair strategy will also be retrieved for repair to improve the overall performance and service life of the battery; and by introducing the neural network into the battery management, the effectiveness and intelligence level of the battery management are also improved.

[0121] The present invention also proposes a battery management system based on a neural network. The battery management system based on the neural network includes a neural chip and a battery. The neural chip is used to configure and run the neural network; the battery management system based on the neural network can execute the battery management method based on the neural network described in any one of the above.

[0122] In this embodiment, referring to Figure 7 , the battery management system based on the neural network includes a neural chip and a battery. The neural chip is used to configure and run the neural network. The neural network learns by obtaining the battery data of a large number of batteries, so as to realize the processing of real-time battery data to realize various battery management tasks.

[0123] The present invention also proposes a battery management device based on a neural network. Referring to Figure 8 , Figure 8 is a schematic structural diagram of the battery management device based on the neural network in the hardware operating environment involved in the embodiment solution of the present invention.

[0124] The battery management device based on the neural network in the embodiment of the present invention can be a processor capable of running the battery management method based on the neural network; there is at least one processor. As Figure 8As shown in the figure, the neuron network-based battery management device may include: a processor 1001 (such as a CPU), a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen and an input unit, such as a keyboard. Optionally, the user interface 1003 may further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0125] Those skilled in the art can understand that Figure 8 the structure of the neuron network-based battery management device shown in the figure does not constitute a limitation on the neuron network-based battery management device, and it may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0126] As Figure 8 shown, the memory 1005, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a computer program.

[0127] In Figure 8 the neuron network-based battery management device shown in the figure, the network interface 1004 is mainly used to connect to a background server and perform data communication with the background server; the user interface 1003 is mainly used to connect to a client (user side) and perform data communication with the client; and the processor 1001 may be used to call the computer program stored in the memory 1005, and when the computer program is called and executed by the processor 1001, the steps of the aforementioned neuron network-based battery management method are implemented.

[0128] Based on the computer program proposed in the foregoing embodiments, the present invention also proposes a storage medium that stores a computer program, and when the computer program is executed by a controller, the neuron network-based battery management method recorded in the foregoing embodiments is implemented.

[0129] The present invention also proposes a storage medium that stores a computer program, and when the computer program is executed by a processor, the steps of the neuron network-based battery management method according to any one of the above technical solutions are implemented.

[0130] The above are only partial or preferred embodiments of the present invention. Neither the text nor the drawings can limit the scope of protection of the present invention. Any equivalent structural transformation made by using the content of the specification and drawings of the present invention under the overall concept of the present invention, or any direct / indirect application in other related technical fields is included in the scope of protection of the present invention.

Claims

1. A battery management method based on a neural network, characterized in that: The battery management method based on neural network includes: Acquire real-time battery data of the battery, and input the real-time battery data into a preset fault diagnosis model; Receiving the fault probability output by the preset fault diagnosis model according to the real-time battery data; Determining whether the failure probability is greater than a preset threshold; If the failure probability is greater than the preset threshold, detailed failure information is generated according to the real-time battery data and the failure probability, and an optimal repair strategy is retrieved according to the detailed failure information to repair the battery.

2. The battery management method based on neural network according to claim 1, characterized in that: Before acquiring the real-time battery data of the battery, the method further includes: Obtaining historical battery data of the battery; Performing filtering processing on the historical battery data to obtain pre-processed battery data; According to the marking instruction, the pre-processed battery data is marked to obtain a plurality of marked battery data, wherein the marked battery data includes normal battery data and abnormal battery data; The preset fault diagnosis model is obtained by training an initial fault diagnosis model according to the plurality of labeled battery data.

3. The battery management method based on neural network according to claim 2, characterized in that: After receiving the fault probability output by the preset fault diagnosis model according to the real-time battery data, the method further includes: The historical battery data is updated according to the real-time battery data, and filtering is performed on the historical battery data to obtain pre-processed battery data.

4. The battery management method based on neural network according to claim 1, characterized in that: The retrieving the best repair strategy according to the detailed fault information to repair the battery comprises: Analyze the detailed fault information to determine the fault type; Retrieving a preset repair strategy library according to the fault type to obtain multiple repair strategies; Obtaining the success rate of each of the repair strategies, and arranging the success rates of each of the repair strategies in descending order; Determine the best repair strategy by using the repair strategy corresponding to the success rate ranked first; The battery is repaired according to the optimal repair strategy.

5. The battery management method based on neural network according to claim 4, characterized in that: After repairing the battery according to the optimal repair strategy, the method further includes: Obtaining the current battery status of the battery; Comparing the current battery state with a preset normal state to determine whether the battery is repaired successfully; If the battery repair is successful, the success rate of the optimal repair strategy is updated, and the information of the successful battery repair is recorded.

6. The battery management method based on neural network according to claim 5, characterized in that: After determining whether the battery is repaired successfully, the method further includes: If the battery is not repaired successfully, generating a maintenance suggestion according to the fault type and displaying the maintenance suggestion; or, If the battery is not repaired successfully, a second repair strategy is obtained, and the battery is repaired according to the second repair strategy, wherein the second repair strategy is the repair strategy corresponding to the second success rate in the descending order.

7. The battery management method based on neural network according to claim 1, characterized in that: The battery includes a plurality of single cells; the battery management method based on the neural network also includes: Collecting the current voltage of each of the single cells; Performing an average calculation on the current voltage of each of the single cells to obtain a current average voltage; Comparing the current voltage of each of the single cells with the current average voltage; If the current voltage of the single battery is greater than the current average voltage, controlling the balancing circuit to discharge the single battery until the current voltage of the single battery is equal to the current average voltage; If the current voltage of the single battery is less than the current average voltage, the balancing circuit is controlled to charge the single battery until the current voltage of the single battery is equal to the current average voltage.

8. The battery management method based on neural network according to claim 1, characterized in that: The battery management method based on the neural network also includes: Detecting current parameters of the charging current and obtaining the current battery state of the battery; Adjust the current parameter according to the current battery state to obtain the optimal current parameter; The charging current is adjusted according to the optimal current parameter, and the charging current is input into the battery.

9. A battery management system based on a neural network, characterized in that: The battery management system based on the neural network includes a neural chip and a battery, wherein the neural chip is used to configure and run the neural network; the battery management system based on the neural network can execute the battery management method based on the neural network described in any one of claims 1 to 8.

10. A battery management device based on a neural network, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executed by the at least one processor, and the instructions are executed by the at least one processor so as to enable the at least one processor to execute the battery management method based on a neural network according to any one of claims 1 to 8.

Images