Battery pack fault diagnosis system of energy storage power station based on network model

Abstract

The invention relates to the technical field of energy storage, and provides a battery pack fault diagnosis system of an energy storage power station based on a network model, comprising a first acquisition unit used for acquiring an initial performance parameter and a structural design drawing of a battery pack; the second acquisition unit is used for constructing an initial three-dimensional battery pack model; the third acquisition unit is used for acquiring real-time operation data through a plurality of sensors, including voltage, current, temperature, internal resistance, charging and discharging states and residual electric quantity percentage; the first generation unit is used for generating battery operation state analysis information; the second generation unit is used for generating a battery fault diagnosis model; and the execution unit is used for executing at least one fault diagnosis task by utilizing the model, such as voltage anomaly detection, temperature anomaly detection, real-time monitoring and internal resistance change simulation. The method can improve the fault diagnosis precision and efficiency of the energy storage power station battery pack, prevents potential faults, and guarantees the safe and stable operation of the battery pack.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage, and more specifically, the present invention relates to a battery pack fault diagnosis system for an energy storage power station based on a network model. Background Art

[0002] With the continuous growth of energy demand and the widespread application of renewable energy, the importance of energy storage technology in the power system has become increasingly prominent. As an efficient energy storage and conversion device, an energy storage power station can effectively solve the problems of intermittency and instability of renewable energy power generation, and improve the flexibility and reliability of power grid operation. However, the reliability and stability of the battery pack, which is the core component of the energy storage power station, are directly related to the performance and lifespan of the entire energy storage system. During long-term operation, the battery pack may be affected by various factors, such as changes in environmental temperature, an increase in the number of charge-discharge cycles, battery aging, etc., resulting in a decline in battery performance or even a failure. Once a failure occurs in the battery pack, it will not only affect the normal operation of the energy storage power station but may also trigger safety accidents, causing huge economic losses.

[0003] The existing battery pack fault diagnosis technologies for energy storage power stations mainly rely on traditional monitoring means, such as the real-time monitoring of basic parameters such as voltage, current, and temperature. Although these monitoring means can reflect the operating state of the battery pack to a certain extent, due to the lack of in-depth analysis of the internal structure and performance changes of the battery pack, it is difficult to accurately and timely diagnose potential faults in the battery pack. In addition, traditional fault diagnosis methods usually require manual data analysis and judgment, with low efficiency and being easily affected by subjective factors, resulting in insufficient accuracy and reliability of the diagnosis results.

[0004] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: First, the prior art lacks a comprehensive analysis of the initial performance parameters and structural design drawings of the battery pack, and cannot construct an accurate battery pack model, thus affecting the accuracy of fault diagnosis; Second, the prior art does not comprehensively collect the real-time operation data of the battery pack, and cannot cover key parameters such as battery voltage, current, temperature, internal resistance, charge-discharge state, and remaining percentage of battery power, resulting in incomplete input information for the fault diagnosis model; Third, the prior art lacks intelligent analysis means in the process of fault diagnosis, cannot effectively utilize advanced machine learning algorithms and deep learning models, and is difficult to quickly and accurately identify and locate complex battery faults; Finally, the prior art lacks an effective abnormal handling mechanism after detecting an abnormality in the battery pack, and cannot timely perform in-depth detection and repair operations on abnormal battery monomers, thus affecting the safe operation and service life of the energy storage power station. Summary of the Invention

[0005] The present invention provides a battery pack fault diagnosis system for an energy storage power station based on a network model, including: A first acquisition unit, configured to acquire initial performance parameters and a structural design diagram of a battery pack of a target energy storage power station; A second acquisition unit, configured to construct an initial three-dimensional battery pack model according to the initial performance parameters and the structural design diagram; A third acquisition unit, configured to acquire real-time operation data corresponding to a plurality of sensors deployed in the battery pack of the target energy storage power station, where the real-time operation data includes battery voltage, battery current, battery temperature, battery internal resistance, battery charge and discharge state, and remaining battery power percentage; A first generation unit, configured to generate battery operation state analysis information of the battery pack of the target energy storage power station according to the real-time operation data; A second generation unit, configured to generate a battery fault diagnosis model of the battery pack of the target energy storage power station according to the real-time operation data and the battery operation state analysis information; An execution unit, configured to execute at least one fault diagnosis processing task of the battery pack of the target energy storage power station by using the battery fault diagnosis model.

[0006] Further, the at least one fault diagnosis processing task includes a voltage anomaly detection task, a temperature anomaly detection task, a first real-time monitoring task based on an intelligent monitor, an internal resistance change simulation task, and a second real-time monitoring task based on a remote monitoring platform; the execution unit executing the at least one fault diagnosis processing task includes: In response to receiving a voltage anomaly detection instruction for the battery pack of the target energy storage power station, starting at least one voltage monitoring device in the battery pack of the target energy storage power station by using the battery fault diagnosis model, and executing the voltage anomaly detection task; In response to receiving a temperature anomaly detection instruction for the battery pack of the target energy storage power station, starting at least one temperature monitoring device in the battery pack of the target energy storage power station by using the battery fault diagnosis model, and executing the temperature anomaly detection task; In response to receiving a first real-time monitoring instruction for the battery pack of the target energy storage power station, starting the intelligent monitor by using the battery fault diagnosis model, and executing the first real-time monitoring task; In response to receiving a second real-time monitoring instruction for the battery pack of the target energy storage power station, executing the second real-time monitoring task by using the battery fault diagnosis model and the remote monitoring platform; In response to receiving an internal resistance change simulation instruction for the battery pack of the target energy storage power station, starting a simulation analysis program by using the battery fault diagnosis model, and executing the internal resistance change simulation task.

[0007] Further, the system further includes: An exception handling unit, configured to perform the following operations in response to the battery operation data exception information output by the battery fault diagnosis model: Obtain the detailed operation parameter records of each battery cell in the target energy storage power station battery pack collected by a high-precision sensor within a predetermined time period; Obtain the conventional operation parameter records of each battery cell within the predetermined time period collected by a conventional monitoring device; Perform periodic sampling on the detailed operation parameter records and the conventional operation parameter records to generate a detailed operation parameter sample set and a conventional operation parameter sample set; Input each conventional operation parameter sample in the conventional operation parameter sample set into a first battery identification information generation model to generate a battery position information set and a battery type information set of each battery cell, and obtain a battery position information set sequence and a battery type information set sequence; Determine the position association relationship information between the battery cells and the battery type information of each battery cell according to the battery position information set sequence and the battery type information set sequence; Input each detailed operation parameter sample in the detailed operation parameter sample set and the corresponding position association relationship information into a second battery identification information generation model to generate the detailed parameter battery position information and the detailed parameter battery type information of each battery cell; Input the detailed operation parameter sample, the corresponding detailed parameter battery position information, and the detailed parameter battery type information into a battery anomaly detection model to generate the battery anomaly judgment information of each battery cell; Generate the battery anomaly information of each battery cell according to the battery anomaly judgment information.

[0008] Further, the exception handling unit is further configured to: Send the battery anomaly information to the battery fault diagnosis model to trigger the battery fault diagnosis model to instruct the target maintenance device to perform in-depth detection and repair operations on the abnormal battery cells.

[0009] Further, the generation of the battery anomaly information of each battery cell includes: Perform sequence adjustment on the battery anomaly judgment information to generate a battery anomaly judgment information sequence for each battery cell; Extract target anomaly judgment information indicating abnormal battery voltage, abnormal battery temperature, abnormal battery internal resistance, or abnormal battery charge and discharge state from each battery anomaly judgment information sequence; Calculate the ratio value between the number of the target anomaly judgment information and the total number of the battery anomaly judgment information sequences; If the ratio value is less than a preset threshold, generate battery anomaly information indicating that the battery cell is normal; If the ratio value is greater than or equal to the preset threshold, generate battery anomaly information indicating that the battery cell has an anomaly.

[0010] Further, the determining the positional association relationship information between the battery cells and the battery type information of each battery cell includes: Obtain the actual battery position information of each battery cell according to the battery position information set sequence; Based on the overall battery position information of the target energy storage power station battery pack, filter out the invalid position information in the actual battery position information that exceeds the overall battery position range; Generate the positional association relationship information and the battery type information of each battery cell according to the filtered actual battery position information and the corresponding battery type information.

[0011] Further, the obtaining the actual battery position information includes: Input the conventional operation parameter sample into the parameter sample - actual position corresponding transformation model; Through the parameter sample - actual position corresponding transformation model, convert the coordinate position in the conventional operation parameter sample into an actual position coordinate to generate the actual battery position information.

[0012] Further, the first battery identification information generation model is a convolutional neural network model, and the structure of the convolutional neural network model includes: An input layer for receiving a conventional operation parameter sample image; A convolutional layer for extracting features from the conventional operation parameter sample image through multiple convolutional kernels to generate multiple feature maps; A pooling layer for downsampling the feature maps to reduce the dimension and retain key features; A fully - connected layer for mapping the dimensionality - reduced feature vectors to the output spaces of the battery position information set and the battery type information set; An output layer for outputting the battery position information set and the battery type information set.

[0013] Further, the second battery identification information generation model is a hybrid model combining a convolutional neural network and a long - short - term memory network, and the structure of the hybrid model includes: An input layer for receiving detailed operation parameter sample data; A convolutional neural network layer for extracting spatial features from the detailed operation parameter sample data; A long - short - term memory network layer for extracting time - series features from the detailed operation parameter sample data; A fully connected layer for fusing the spatial features and time series features and mapping them to the output space of detailed parameter battery position information and detailed parameter battery type information; An output layer for outputting the detailed parameter battery position information and the detailed parameter battery type information.

[0014] Further, the parameter sample and actual position correspondence transformation model is a geometric transformation model, and the working steps of the geometric transformation model include: Receiving a target normal operation parameter sample and its corresponding battery position information; According to a predefined coordinate transformation function, converting the battery position information from the parameter sample coordinate system to the actual position coordinate system; Outputting the converted actual battery position information for determining the position association relationship information and battery type information of the battery cell.

[0015] According to the above embodiments of the present invention, it has at least the following beneficial effects: The present invention can improve the fault diagnosis accuracy and efficiency of the energy storage power station battery pack. Through multi-dimensional data acquisition and intelligent analysis models, key parameters such as voltage, current, temperature, and internal resistance of the battery pack are monitored in real time, abnormal states are quickly identified, and the possibility of missed detection and misjudgment of faults is reduced. At the same time, the system can generate battery operation status analysis information and fault diagnosis models, providing a scientific basis for operation and maintenance personnel, optimizing the management and maintenance strategies of the battery pack, and extending the service life of the battery pack.

[0016] In addition, the present invention can support a variety of fault diagnosis tasks, including voltage anomaly detection, temperature anomaly detection, internal resistance change simulation, and real-time monitoring based on intelligent monitors and remote monitoring platforms, comprehensively covering potential risks during the operation of the battery pack. Through the combination of the anomaly processing unit and high-precision sensors, the system can accurately locate abnormal battery cells and trigger in-depth detection and repair operations, further improving the operation safety and stability of the energy storage power station and reducing operation and maintenance costs and risks. Description of the Drawings

[0017] By referring to the accompanying drawings and reading the following detailed description, the above and other objects, features, and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, wherein: Figure 1 It is a structural schematic diagram of a battery pack fault diagnosis system for an energy storage power station provided by an embodiment of the present invention. Detailed Embodiments

[0018] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and then implement the present invention, rather than limiting the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to be able to fully convey the scope of the present invention to those skilled in the art.

[0019] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, a device, an equipment, a method, or a computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0020] It should be noted that the quantity of any element in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0021] The following refers to Figure 1 , Figure 1 which is a schematic structural diagram of a battery pack fault diagnosis system for an energy storage power station provided in an embodiment of the present invention. As Figure 1 shown, a battery pack fault diagnosis system for an energy storage power station based on a network model includes: A first acquisition unit 101, configured to acquire the initial performance parameters and the structural design diagram of the battery pack of the target energy storage power station; A second acquisition unit 102, configured to construct an initial three-dimensional battery pack model according to the initial performance parameters and the structural design diagram; A third acquisition unit 103, configured to acquire the real-time operation data corresponding to the multiple sensors through the multiple sensors deployed in the battery pack of the target energy storage power station, where the real-time operation data includes battery voltage, battery current, battery temperature, battery internal resistance, battery charge and discharge state, and the remaining percentage of battery power; A first generation unit 104, configured to generate battery operation state analysis information of the battery pack of the target energy storage power station according to the real-time operation data; A second generation unit 105, configured to generate a battery fault diagnosis model of the battery pack of the target energy storage power station according to the real-time operation data and the battery operation state analysis information; An execution unit 106, configured to execute at least one fault diagnosis processing task of the battery pack of the target energy storage power station by using the battery fault diagnosis model.

[0022] It should be noted that the battery pack fault diagnosis system of the present invention first obtains the initial performance parameters and structural design drawings of the target energy storage power station battery pack through the first acquisition unit. The initial performance parameters refer to the basic performance indicators that the battery pack possesses when leaving the factory, such as rated voltage, rated capacity, internal resistance, etc. These parameters reflect the working performance of the battery pack under ideal conditions. The structural design drawing refers to the physical structure layout of the battery pack, including the arrangement method, connection method of battery cells, and the design of the heat dissipation system, etc. These information are crucial for constructing an accurate three-dimensional battery pack model. By obtaining this information, the system can provide basic data support for subsequent fault diagnosis, ensuring the accuracy and reliability of the diagnosis model.

[0023] Specifically, the initial performance parameters are determined during the design and manufacturing stages of the battery pack and are used to describe the performance indicators of the battery pack under normal usage conditions. For example, the rated voltage refers to the standard voltage value that the battery pack can provide when fully charged, usually in volts (V); the rated capacity refers to the amount of electricity that the battery pack can output under standard discharge conditions, usually in ampere-hours (Ah). The structural design drawing details the physical structure inside the battery pack. For example, the arrangement method of battery cells can be in series or parallel, and the connection method may involve welding or bolt connection, etc. The design of the heat dissipation system includes the layout of heat sinks, the flow direction of the coolant, etc. These design details are crucial for the thermal management of the battery pack. By obtaining these detailed initial performance parameters and structural design drawings, the system can construct an accurate initial three-dimensional battery pack model, which can simulate the physical and electrical characteristics of the battery pack during actual operation and provide an important reference basis for subsequent fault diagnosis.

[0024] Preferably, in order to further improve the accuracy of the initial three-dimensional battery pack model, the system can calibrate and verify the obtained initial performance parameters. For example, by comparing the actual measured values with the design values, fine-tuning parameters such as rated voltage and rated capacity to ensure the accuracy of the model. At the same time, for the heat dissipation system design in the structural design drawing, the system can evaluate and optimize the efficiency of the heat dissipation system by combining the temperature data of the actual operating environment. For example, if it is found that the temperature of the battery pack during actual operation is too high, the system can adjust the layout of the heat sinks or increase the flow rate of the coolant to improve the heat dissipation effect. In addition, the system can also introduce virtual reality (VR) or augmented reality (AR) technology to visually display the constructed initial three-dimensional battery pack model, enabling technicians to more intuitively understand the internal structure and operating status of the battery pack, thus providing a more intuitive reference for fault diagnosis.

[0025] In some embodiments, the at least one fault diagnosis processing task includes a voltage anomaly detection task, a temperature anomaly detection task, a first real-time monitoring task based on an intelligent monitor, an internal resistance change simulation task, and a second real-time monitoring task based on a remote monitoring platform; the execution unit executes the at least one fault diagnosis processing task, including: In response to receiving the voltage anomaly detection instruction for the target energy storage power station battery pack, the battery fault diagnosis model is used to activate at least one voltage monitoring device within the target energy storage power station battery pack, and the voltage anomaly detection task is executed; In response to receiving the temperature anomaly detection instruction for the target energy storage power station battery pack, the battery fault diagnosis model is used to activate at least one temperature monitoring device within the target energy storage power station battery pack, and the temperature anomaly detection task is executed; In response to receiving the first real-time monitoring instruction for the target energy storage power station battery pack, the battery fault diagnosis model is used to activate the intelligent monitor, and the first real-time monitoring task is executed; In response to receiving the second real-time monitoring instruction for the target energy storage power station battery pack, the battery fault diagnosis model and the remote monitoring platform are used to execute the second real-time monitoring task; In response to receiving the internal resistance change simulation instruction for the target energy storage power station battery pack, the battery fault diagnosis model is used to activate the simulation analysis program, and the internal resistance change simulation task is executed.

[0026] It should be noted that in the energy storage power station battery pack fault diagnosis system of the present invention, the at least one fault diagnosis processing task includes a voltage anomaly detection task, a temperature anomaly detection task, a first real-time monitoring task based on an intelligent monitor, an internal resistance change simulation task, and a second real-time monitoring task based on a remote monitoring platform. These tasks respectively monitor and diagnose different types of faults that may occur during the operation of the battery pack. The voltage anomaly detection task is used to identify whether the battery voltage deviates from the normal range, the temperature anomaly detection task is used to monitor whether the battery temperature is too high or too low, and the internal resistance change simulation task is used to analyze the change trend of the battery internal resistance to judge the degree of battery aging. The intelligent monitor and the remote monitoring platform are respectively used to monitor the operation status of the battery pack in real time to ensure that faults can be detected and processed in a timely manner.

[0027] Specifically, for the voltage anomaly detection task, a reasonable threshold range is set according to the rated voltage range of the battery pack. For example, for a battery pack with a rated voltage of 12V, the normal operating voltage range may be set between 11.5V and 12.5V. When the battery voltage exceeds this range, the system will trigger the voltage anomaly detection task and use the battery fault diagnosis model to activate the voltage monitoring device for detailed detection. The temperature anomaly detection task will set corresponding temperature thresholds in combination with the rated operating temperature range of the battery pack, usually -20°C to 50°C. If the battery temperature exceeds this range, the system will activate the temperature monitoring device for detection. The internal resistance change simulation task will predict the health status of the battery through a simulation analysis program based on the initial internal resistance value of the battery and the internal resistance change data during actual operation. The intelligent monitor and the remote monitoring platform continuously monitor the operating status of the battery pack by receiving sensor data in real time to ensure that any abnormal conditions can be detected in a timely manner.

[0028] Preferably, to improve the accuracy and efficiency of fault diagnosis, the system can further optimize these tasks. For example, in the voltage anomaly detection task, in addition to setting a fixed threshold range, a dynamic threshold adjustment mechanism can be introduced. According to the charge and discharge state of the battery pack and the ambient temperature change, the voltage threshold range is dynamically adjusted to more accurately identify voltage anomalies. In the temperature anomaly detection task, a multi-level temperature warning mechanism can be set in combination with the heat dissipation design of the battery pack and the actual operating environment. For example, a first-level warning is issued when the temperature reaches 40°C, a second-level warning is issued when the temperature reaches 45°C, and the heat dissipation system is activated. In the internal resistance change simulation task, machine learning algorithms can be used to model the internal resistance change data, and the model is trained through historical data to more accurately predict the health status of the battery. In addition, the intelligent monitor and the remote monitoring platform can integrate more types of sensor data, such as the remaining percentage of battery power, charge and discharge status, etc., to provide more comprehensive real-time monitoring information.

[0029] In some embodiments, the system further includes: An anomaly handling unit, configured to perform the following operations in response to the abnormal information of the battery operation data output by the battery fault diagnosis model: Obtain the detailed operation parameter records of the battery cells in the target energy storage power station battery pack collected by the high-precision sensor within a predetermined time period; Obtain the regular operation parameter records of the battery cells within the predetermined time period collected by the regular monitoring device; Perform periodic sampling on the detailed operation parameter records and the regular operation parameter records to generate a detailed operation parameter sample set and a regular operation parameter sample set; Input each conventional operation parameter sample in the set of conventional operation parameter samples into the first battery identification information generation model to generate a battery position information set and a battery type information set for the battery cell, and obtain a battery position information set sequence and a battery type information set sequence; Determine the position association relationship information between the battery cells and the battery type information of each battery cell according to the battery position information set sequence and the battery type information set sequence; Input each detailed operation parameter sample in the set of detailed operation parameter samples and the corresponding position association relationship information into the second battery identification information generation model to generate detailed parameter battery position information and detailed parameter battery type information for each battery cell; Input the detailed operation parameter sample, the corresponding detailed parameter battery position information and detailed parameter battery type information into the battery anomaly detection model to generate battery anomaly judgment information for each battery cell; Generate battery anomaly information for the battery cell according to the battery anomaly judgment information.

[0030] It should be noted that the anomaly processing unit in the present invention is an important part of the energy storage power station battery pack fault diagnosis system. Its main function is to respond to the battery operation data anomaly information output by the battery fault diagnosis model and perform a series of operations to determine the anomaly situation of the battery cell. High-precision sensors and conventional monitoring devices are respectively used to collect the detailed operation parameter records and conventional operation parameter records of the battery cell. The detailed operation parameter record refers to the battery operation data collected at a high frequency in a short time, such as voltage, current, temperature and other data collected once per second; while the conventional operation parameter record is the data collected at a low frequency, such as data collected once per minute. Periodic sampling is to sample the collected operation parameter records at a certain time interval to generate a sample set for subsequent analysis and processing. The battery position information set and the battery type information set are generated by the first battery identification information generation model and are used to determine the position and type of the battery cell. The position association relationship information refers to the relative position relationship of the battery cells in the battery pack, and the battery anomaly detection model is used to generate the anomaly judgment information of each battery cell, and finally generate the battery anomaly information.

[0031] Specifically, after receiving the abnormal information output by the battery fault diagnosis model, the abnormal handling unit first collects the detailed operation parameter records of the battery cells in the target energy storage power station battery pack within a predetermined time period through high-precision sensors. For example, data such as voltage, current, and temperature are collected once per second within 1 hour. At the same time, the conventional operation parameter records within the same time period are collected through conventional monitoring devices. For example, data such as voltage, current, and temperature are collected once per minute. Then, these data are sampled periodically. For example, one sample is extracted from the detailed operation parameter records every 10 seconds, and one sample is extracted from the conventional operation parameter records every 5 minutes, generating a detailed operation parameter sample set and a conventional operation parameter sample set. Next, each sample in the conventional operation parameter sample set is input into the first battery identification information generation model to generate a battery position information set and a battery type information set, obtaining a battery position information set sequence and a battery type information set sequence. Based on these sequences, the position association relationship information between the battery cells and the battery type information of each battery cell are determined. After that, each sample in the detailed operation parameter sample set and the corresponding position association relationship information are input into the second battery identification information generation model to generate the detailed parameter battery position information and the detailed parameter battery type information of each battery cell. Finally, the detailed operation parameter samples, the corresponding detailed parameter battery position information, and the detailed parameter battery type information are input into the battery abnormal detection model to generate the battery abnormal judgment information of each battery cell, and the battery abnormal information is generated based on this information.

[0032] Preferably, to improve the accuracy and efficiency of anomaly handling, refinement or alternative solutions can be provided in the following aspects. First, when collecting detailed operating parameter records and regular operating parameter records, the collection frequency can be adjusted according to the actual operating status and fault type of the battery pack. For example, for a battery pack that may experience voltage fluctuations, the voltage collection frequency can be increased; for a battery pack that may experience thermal runaway, the temperature collection frequency can be increased. Second, during periodic sampling, the sampling interval can be dynamically adjusted according to the operating status of the battery pack. For example, when the battery pack is operating smoothly, the sampling interval can be appropriately extended; when abnormal signs appear in the battery pack, the sampling interval can be shortened to obtain more detailed operating data. In addition, different algorithms or model structures can be used to optimize the first battery identification information generation model and the second battery identification information generation model. For example, the first battery identification information generation model can adopt an improved convolutional neural network (CNN) structure to improve the recognition accuracy of battery position and type information; the second battery identification information generation model can combine long short-term memory network (LSTM) and attention mechanism to better process the time series characteristics of detailed operating parameters. Finally, when generating battery anomaly information, multi-dimensional anomaly judgment criteria can be introduced, such as combining multiple indicators such as voltage anomaly, temperature anomaly, and internal resistance anomaly to comprehensively judge the anomaly situation of battery cells, thereby improving the accuracy and reliability of anomaly judgment.

[0033] In some embodiments, the anomaly handling unit is further configured to: Send the battery anomaly information to the battery fault diagnosis model to trigger the battery fault diagnosis model to instruct the target maintenance equipment to perform in-depth detection and repair operations on the abnormal battery cell.

[0034] It should be noted that after the anomaly handling unit in the present invention generates the anomaly information of the battery cell, it can send this anomaly information to the battery fault diagnosis model to trigger subsequent in-depth detection and repair operations. This process is to ensure that the abnormal battery cell can be processed in a timely manner, thereby ensuring the safe operation of the entire energy storage power station. The target maintenance equipment here refers to the equipment specifically used for detecting and repairing battery cells, such as battery testers, charge and discharge equipment, welding tools, etc. These equipment can perform detailed detection on the abnormal battery cell and perform corresponding repair operations according to the detection results.

[0035] Specifically, after the exception handling unit generates the exception information of the battery cell, it will send this information to the battery fault diagnosis model through the communication interface. After receiving the exception information, the battery fault diagnosis model will, according to the preset rules and logic, determine the type of in-depth detection and repair operations to be performed on the abnormal battery cell. For example, if the exception information indicates that there is a voltage abnormality in the battery cell, the battery fault diagnosis model may instruct the target repair device to perform a charge and discharge test on the battery cell to determine whether it can work properly; if the exception information indicates that there is a temperature abnormality in the battery cell, the battery fault diagnosis model may instruct the target repair device to check and repair the heat dissipation system of the battery cell. The parameter settings of the target repair device need to be adjusted according to the specific abnormal conditions of the battery cell. For example, when performing a charge and discharge test, appropriate current and voltage ranges need to be set to ensure the safety and accuracy of the test; when checking the heat dissipation system, appropriate temperature thresholds and coolant flow parameters need to be set to ensure the normal operation of the heat dissipation system.

[0036] Preferably, to improve the efficiency and accuracy of exception handling, refinement or alternative solutions can be provided in the following aspects. First, when sending exception information to the battery fault diagnosis model, an encrypted communication protocol can be adopted to ensure the security of data transmission. Second, the battery fault diagnosis model can prioritize the handling of abnormal battery cells with high risks according to the severity of the exception information. For example, if the exception information indicates that there is a serious voltage abnormality or temperature abnormality in the battery cell, the battery fault diagnosis model can give priority to instructing the target repair device to perform in-depth detection and repair operations on the battery cell.

[0037] Furthermore, the target repair device can be equipped with a variety of detection and repair tools to handle different types of abnormal conditions. For example, in addition to the conventional charge and discharge test equipment, an infrared thermal imager can be equipped to detect the temperature distribution of the battery cell, or an ultrasonic detection device can be equipped to detect internal structural damage of the battery cell. Finally, to further improve the repair efficiency, the target repair device can interact with the battery fault diagnosis model in real time and dynamically adjust the repair strategy according to the real-time data during the repair process. For example, if it is found during the repair process that the abnormal condition of the battery cell is more serious than expected, the target repair device can promptly feedback to the battery fault diagnosis model for re-evaluating the repair plan.

[0038] In some embodiments, the generation of the battery exception information of the battery cell includes: Performing sequence adjustment on the battery exception judgment information to generate a battery exception judgment information sequence for each battery cell; Extracting target exception judgment information characterizing battery voltage abnormality, battery temperature abnormality, battery internal resistance abnormality, or battery charge and discharge state abnormality from each battery exception judgment information sequence; Calculate the ratio value between the number of the target abnormal judgment information and the total number of the battery abnormal judgment information sequences; If the ratio value is less than a preset threshold, generate battery abnormal information indicating that there is no abnormality in the battery cell; If the ratio value is greater than or equal to the preset threshold, generate battery abnormal information indicating that there is an abnormality in the battery cell.

[0039] It should be noted that in the present invention, the process of generating battery abnormal information of a battery cell is completed by analyzing and processing battery abnormal judgment information. The battery abnormal judgment information refers to the preliminary judgment results on whether there is an abnormality in a battery cell generated by a battery abnormal detection model, and these information are usually presented in the form of data sequences. Sequence adjustment refers to reordering or formatting these data sequences to more clearly display the abnormal conditions of each battery cell. The target abnormal judgment information refers to the specific information related to battery voltage abnormality, temperature abnormality, internal resistance abnormality or charge and discharge state abnormality extracted from the sequences. The ratio value refers to the ratio between the number of the target abnormal judgment information and the total number of the battery abnormal judgment information sequences, and is used to evaluate the abnormal degree of the battery cell. The preset threshold is a preset reference value used to judge whether there is really an abnormality in the battery cell.

[0040] Specifically, the process of generating battery abnormal information of a battery cell first involves sequence adjustment of the battery abnormal judgment information. For example, if the battery abnormal judgment information is a sequence including multiple time points, the system will rearrange these information in chronological order to form a clear sequence. Next, the target abnormal judgment information is extracted from each battery abnormal judgment information sequence. For example, the system will identify which time points the battery voltage exceeds the normal range and which time points the battery temperature is abnormally increased, etc. Then calculate the ratio value between the number of the target abnormal judgment information and the total number of the battery abnormal judgment information sequences. For example, if a battery cell is judged to have voltage abnormality at 20 time points out of 100 time points, then the ratio value is 20%. Finally, the system will compare this ratio value with the preset threshold. If the ratio value is less than the preset threshold, for example, the preset threshold is 30%, then the system will judge that there is no abnormality in this battery cell; if the ratio value is greater than or equal to the preset threshold, then the system will judge that there is an abnormality in this battery cell.

[0041] Preferably, in order to improve the accuracy and flexibility of anomaly judgment, refinement or alternative solutions can be carried out in the following aspects. First, the preset threshold can be dynamically adjusted according to the actual operating environment of the battery pack and the battery type. For example, for some high-precision energy storage battery packs, the preset threshold can be set lower to increase the sensitivity to abnormal situations; while for some battery packs for general purposes, the preset threshold can be appropriately increased to reduce false alarms. Secondly, the extraction of target anomaly judgment information can be comprehensively analyzed by combining multiple anomaly features. For example, in addition to abnormal voltage, temperature, internal resistance, and charge and discharge status, abnormal conditions of other parameters such as the self-discharge rate and cycle life of the battery can also be considered.

[0042] Furthermore, the calculation of the proportional value can adopt the method of weighted average, and different weights are assigned according to the influence degree of different types of anomalies on the battery performance. For example, voltage anomaly may have a greater impact on battery performance and can be assigned a higher weight; while the impact of internal resistance anomaly is relatively small and can be assigned a lower weight. Finally, the system can introduce machine learning algorithms to automatically adjust the preset threshold and weight parameters through learning and analysis of a large amount of historical data, so as to further improve the accuracy and adaptability of anomaly judgment.

[0043] In some embodiments, the determining the positional association relationship information between the battery cells and the battery type information of each battery cell includes: Obtaining the actual battery position information of each battery cell according to the battery position information set sequence; Based on the overall battery position information of the target energy storage power station battery pack, filtering out the invalid position information in the actual battery position information that exceeds the overall battery position range; Generating the positional association relationship information and the battery type information of each battery cell according to the filtered actual battery position information and the corresponding battery type information.

[0044] It should be noted that the process of determining the positional association relationship information between the battery cells and the battery type information of each battery cell in the present invention is implemented based on the battery position information set sequence and the overall battery position information of the target energy storage power station battery pack. The battery position information set sequence refers to the serialized information about the position of each battery cell generated by the model, which records the specific position of the battery cell in the battery pack. The overall battery position information refers to the layout range of the battery pack in space, including the physical size and installation position of the battery pack, etc. By comparing the actual battery position information of the battery cells with the overall battery position information, the invalid position information can be filtered out, so as to accurately determine the positional association relationship between the battery cells and the battery type information.

[0045] Specifically, the process of determining the position association relationship information and the battery type information first involves obtaining the actual battery position information of each battery cell. This can be achieved by inputting the sample of the normal operation parameters into the transformation model corresponding to the sample and the actual position, which can convert the coordinate position in the sample into the actual position coordinate. For example, if the coordinate position in the sample of the normal operation parameters is represented by the relative coordinate inside the battery pack, it can be converted into the actual spatial position coordinate through the transformation model. Then, based on the overall position information of the battery pack, the system filters out the invalid position information beyond the overall position range of the battery pack. For example, if the position information of a certain battery cell shows that it is located outside the battery pack, then this position information is obviously invalid and needs to be filtered out. Finally, according to the filtered actual battery position information and the corresponding battery type information, the system can generate the position association relationship information, that is, the relative position relationship between the battery cells, and the battery type information of each battery cell, such as the model and specification of the battery.

[0046] Preferably, in order to improve the accuracy of determining the position association relationship and the battery type information, it can be refined or alternative solutions can be provided in the following aspects. First, when obtaining the actual battery position information, multiple sensor data can be fused. For example, in addition to the coordinate position in the sample of the normal operation parameters, the data of temperature sensors, pressure sensors, etc. can also be combined, and the accuracy of the position information can be improved through the multi-sensor data fusion algorithm. Second, when filtering out the invalid position information, a fault tolerance mechanism can be introduced. For example, a certain range of position deviation is allowed, rather than filtering it completely according to the overall position information of the battery pack, to avoid mis-filtering the valid position information caused by measurement errors.

[0047] Furthermore, for the determination of the battery type information, the identification information such as the bar code or two-dimensional code of the battery can be combined for auxiliary judgment. For example, the model and specification information can be obtained by scanning the bar code on the battery and compared with the battery type information generated by the model to further improve the accuracy of the information. Finally, the system can regularly update the overall position information of the battery pack and the battery position information set sequence to adapt to the possible physical position changes of the battery pack or the replacement of battery cells, etc.

[0048] In some embodiments, the obtaining of the actual battery position information includes: Inputting the sample of the normal operation parameters into the transformation model corresponding to the sample and the actual position; Through the transformation model corresponding to the sample and the actual position, converting the coordinate position in the sample of the normal operation parameters into the actual position coordinate to generate the actual battery position information.

[0049] It should be noted that in the present invention, the process of obtaining the actual position information of the battery is achieved by inputting the conventional operation parameter samples into the corresponding transformation model between the parameter samples and the actual position. Here, the conventional operation parameter samples refer to the data samples containing the battery operation state information collected from the conventional monitoring devices of the battery pack, such as parameters like the voltage, current, and temperature of the battery. The corresponding transformation model between the parameter samples and the actual position is a mathematical model or algorithm, whose function is to convert the coordinate position information in these operation parameter samples into the position coordinates of the battery cell in the actual physical space. This process is crucial for determining the specific position of the battery cell in the battery pack, thereby providing accurate spatial information for subsequent fault diagnosis and anomaly handling.

[0050] Specifically, the working principle of the corresponding transformation model between the parameter samples and the actual position is based on a predefined coordinate transformation function. For example, assume that the coordinates in the conventional operation parameter samples are represented in the local coordinate system inside the battery pack, while the actual position coordinates are represented in the global coordinate system of the entire energy storage power station. The role of the coordinate transformation function is to convert the local coordinates into global coordinates. In practical applications, this transformation function may be a linear transformation, such as achieving coordinate conversion through matrix multiplication; it may also be a non - linear transformation, such as considering the bending or irregular layout inside the battery pack. After receiving the target conventional operation parameter samples and their corresponding battery position information, the transformation model will perform calculations according to the predefined transformation rules and output the converted actual position information of the battery. For example, if the position of a battery cell in the local coordinate system is (x, y), after transformation, its position in the global coordinate system may be (X, Y).

[0051] Preferably, in order to improve the accuracy and reliability of position information acquisition, it can be refined or alternative solutions can be provided in the following aspects. First, the coordinate transformation function can be optimized according to the actual layout of the battery pack. For example, if the layout of the battery pack is a regular rectangular arrangement, a simple linear transformation can be used; if the layout is irregular, a more complex non - linear transformation function, such as a transformation model based on polynomial fitting or neural network, can be adopted. Second, in order to improve the accuracy of the transformation, an error correction mechanism can be introduced into the transformation model. For example, by comparing the actual position information after transformation with the known reference position information, calculating the error, and dynamically adjusting the transformation function.

[0052] Furthermore, the method of multi - sensor data fusion can be adopted, combining the operation parameter samples from different sensors to reduce the error of single - sensor data. For example, simultaneously using the data of voltage sensors and temperature sensors, and calculating more accurate battery position information through a fusion algorithm. Finally, in order to cope with the possible changes in the layout of the battery pack, the coordinate transformation function can be updated regularly to ensure the accuracy and timeliness of the position information.

[0053] In some embodiments, the first battery identification information generation model is a convolutional neural network model, and the structure of the convolutional neural network model includes: An input layer for receiving sample images of normal operating parameters; A convolutional layer for extracting features from the sample images of normal operating parameters through multiple convolutional kernels to generate multiple feature maps; A pooling layer for downsampling the feature maps to reduce the dimension and retain key features; A fully connected layer for mapping the feature vectors after dimensionality reduction to the output spaces of the battery position information set and the battery type information set; An output layer for outputting the battery position information set and the battery type information set.

[0054] It should be noted that the first battery identification information generation model mentioned in the present invention is a convolutional neural network model, which is a deep learning model widely used in image recognition and feature extraction tasks. The structure of this model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The input layer receives sample images of normal operating parameters, which are obtained by converting operating parameter data into a visual image form so that the model can process them. The convolutional layer extracts features from the images through multiple convolutional kernels to generate multiple feature maps, which can capture local features in the images. The pooling layer downsamples the feature maps, reducing the dimension while retaining key features and reducing the computational amount. The fully connected layer maps the feature vectors after dimensionality reduction to the output spaces of the battery position information set and the battery type information set, and the output layer outputs the final battery position information set and battery type information set. This model structure can effectively extract useful information from complex operating parameter images and provide support for the identification of the positions and types of battery cells.

[0055] Specifically, each layer of the convolutional neural network model has its specific functions and parameter settings. In the input layer, the size and resolution of the sample image of the regular operating parameters need to be set according to the actual application. For example, if the resolution of the operating parameter image is 256×256 pixels, the input layer will receive images of this size. The number and size of the convolutional kernels in the convolutional layer are key parameters, which determine the complexity and accuracy of feature extraction. For example, 32 convolutional kernels of size 3×3 can be set, and these convolutional kernels will slide on the image to extract local features. The pooling layer usually adopts max-pooling or average-pooling operations, and the size and stride of the pooling window also need to be set. For example, a 2×2 pooling window and a stride of 2 are used. The number of neurons in the fully connected layer determines the complexity and fitting ability of the model, and can be adjusted according to the scale and complexity of the training data. For example, 128 neurons are set. The output dimension of the output layer should match the dimensions of the battery position information set and the battery type information set. For example, the two-dimensional coordinates of the output position information and the classification labels of the type information are output.

[0056] Preferably, in order to further optimize the performance of the first battery identification information generation model, it can be refined or alternative solutions can be provided in the following aspects. First, in the convolutional layer, different combinations of the size and number of convolutional kernels can be adopted to extract richer features. For example, convolutional kernels of size 3×3 and 5×5 are used simultaneously to capture local features of different scales. Second, in the pooling layer, in addition to the traditional max-pooling and average-pooling, other pooling methods such as stochastic pooling or adaptive pooling can be tried to improve the robustness of feature extraction. In addition, regularization techniques such as Dropout or L2 regularization can be introduced into the fully connected layer to prevent the model from overfitting. During the model training process, data augmentation techniques such as image rotation, scaling, and translation can be adopted to expand the training data set and improve the generalization ability of the model. Finally, in order to improve the running efficiency of the model, a lightweight design can be introduced into the model structure. For example, depthwise separable convolution is used instead of the traditional convolution operation to reduce the computational amount and the size of the model.

[0057] In some embodiments, the second battery identification information generation model is a hybrid model combining a convolutional neural network and a long short-term memory network. The structure of the hybrid model includes: An input layer for receiving detailed operating parameter sample data; A convolutional neural network layer for extracting spatial features from the detailed operating parameter sample data; A long short-term memory network layer for extracting time series features from the detailed operating parameter sample data; A fully connected layer, which is used to fuse the spatial features and time series features and map them to the output space of detailed parameter battery position information and detailed parameter battery type information; An output layer, which is used to output the detailed parameter battery position information and the detailed parameter battery type information.

[0058] It should be noted that the second battery identification information generation model mentioned in the present invention is a hybrid model that combines a convolutional neural network (CNN) and a long short-term memory network (LSTM). The design of this model is to extract both spatial features and time series features in the detailed operation parameter sample data simultaneously. The convolutional neural network layer is mainly used to process the spatial relationships of the data, such as the mutual correlations between battery parameters; while the long short-term memory network layer focuses on capturing the time dynamic characteristics of the data, such as the change trends of battery parameters over time. The role of the fully connected layer is to fuse these two types of features and map them to the output space of detailed parameter battery position information and detailed parameter battery type information, and finally the output layer outputs the results. This hybrid model can analyze battery operation data more comprehensively, thereby improving the accuracy and reliability of identifying the position and type information of battery cells.

[0059] Specifically, the input layer of the hybrid model receives detailed operation parameter sample data, which usually includes the change sequences of multiple parameters such as battery voltage, current, temperature, and internal resistance over time. The convolutional neural network layer performs sliding operations on this data through convolutional kernels to extract spatial features. For example, multiple convolutional kernels can be set, and each convolutional kernel is responsible for extracting different types of spatial features. The long short-term memory network layer uses its unique gating mechanism to memorize and update key information in the time series, thereby capturing time features. The fully connected layer fuses the spatial features extracted by the convolutional layer and the time features extracted by the LSTM layer, and through weighted summation of neurons and processing by activation functions, generates the final feature vector. The output layer generates detailed parameter battery position information and detailed parameter battery type information based on the output of the fully connected layer. For example, the position information can be the specific coordinates of the battery in the battery pack, and the type information can be the model or specification of the battery.

[0060] Preferably, in order to further optimize the performance of the second battery identification information generation model, it can be refined or alternative solutions can be provided in the following aspects. First, in the convolutional neural network layer, a deeper network structure can be adopted or residual connections can be introduced to improve the model's ability to extract complex spatial features. For example, using a convolutional layer with a ResNet architecture can effectively alleviate the problem of gradient disappearance in the training of deep networks. Second, in the long short-term memory network layer, the number and structure of LSTM units can be adjusted to better adapt to the length and complexity of different time series. For example, for long sequence data, the number of LSTM units can be increased to improve the model's memory ability.

[0061] Furthermore, the fully connected layer can introduce the Batch Normalization technique to accelerate the model training process and improve stability. During the model training process, a multi-task learning method can be adopted to optimize the recognition tasks of battery position information and type information simultaneously to improve the overall performance of the model. Finally, to improve the generalization ability of the model, noise can be introduced into the training data or data augmentation can be performed, such as randomly cropping or time warping the time series data.

[0062] In some embodiments, the parameter sample and actual position corresponding transformation model is a geometric transformation model, and the working steps of the geometric transformation model include: Receiving a target normal operation parameter sample and its corresponding battery position information; According to a predefined coordinate transformation function, transforming the battery position information from the parameter sample coordinate system to the actual position coordinate system; Outputting the transformed actual battery position information for determining the position association relationship information and battery type information of the battery cell.

[0063] It should be noted that the parameter sample and actual position corresponding transformation model mentioned in the present invention is a geometric transformation model for transforming the coordinate position in the normal operation parameter sample into the actual position coordinate of the battery cell. The core function of this model is to transform the position information in the local coordinate system inside the battery pack into the position information in the global coordinate system through a predefined coordinate transformation function. This transformation is crucial for determining the precise position of the battery cell in the battery pack, especially when the battery pack layout is complex or there are irregular arrangements. The working steps of the geometric transformation model include receiving a target normal operation parameter sample and its corresponding battery position information, performing the transformation according to the coordinate transformation function, and outputting the transformed actual battery position information, so as to provide accurate spatial positioning for subsequent fault diagnosis and anomaly handling.

[0064] Specifically, the input of the geometric transformation model is the target regular operation parameter samples and their corresponding battery position information, which is usually represented in a local coordinate system. The predefined coordinate transformation function is designed according to the actual layout and installation method of the battery pack. It can be a simple linear transformation, such as translation and scaling, or a complex non-linear transformation, such as considering the bending or irregular arrangement inside the battery pack. The output battery actual position information is represented in a global coordinate system, which can accurately reflect the position of the battery cells in physical space. For example, if the battery pack is installed in a large energy storage power station, the local coordinate system may only cover the inside of the battery pack, while the global coordinate system covers the entire energy storage power station. Through the geometric transformation model, the position information in the local coordinate system can be converted into the position information in the global coordinate system, thus achieving precise positioning of the battery cells.

[0065] Preferably, in order to improve the accuracy and applicability of the geometric transformation model, it can be refined or alternative solutions can be provided in the following aspects. First, when defining the coordinate transformation function, more geometric parameters can be considered, such as rotation angle and translation vector, to more precisely describe the layout of the battery pack. For example, if there is a certain rotation angle during the installation of the battery pack, the coordinate transformation function can be adjusted through a rotation matrix. Second, in order to handle complex non-linear transformations, a neural network-based transformation model can be adopted, such as using a multi-layer perceptron (MLP) or a convolutional neural network (CNN) to learn the mapping relationship between local coordinates and global coordinates. This data-driven method can automatically adapt to different layouts and installation methods of the battery pack.

[0066] Furthermore, in order to improve the robustness of the transformation, an error correction mechanism can be introduced into the model, such as adjusting the transformation parameters through the least squares method or other optimization algorithms to minimize the error after transformation. Finally, in order to cope with the possible changes in the layout of the battery pack, the coordinate transformation function can be updated regularly, or an adaptive transformation model can be designed that can dynamically adjust the transformation parameters according to real-time monitoring data.

[0067] The above-mentioned embodiments of the present invention have the following beneficial effects: The energy storage power station battery pack fault diagnosis system of the present invention can achieve comprehensive monitoring and accurate diagnosis of the battery pack through the collaborative work of multiple units. The system uses the first acquisition unit and the second acquisition unit to obtain the initial performance parameters and structural design drawings of the battery pack and construct an initial three-dimensional battery pack model. Combining the real-time operation data collected by the third acquisition unit, it can provide rich and accurate basic information for fault diagnosis. The battery operation state analysis information and the battery fault diagnosis model generated by the first generation unit and the second generation unit further improve the scientificity and reliability of the diagnosis. The execution unit can flexibly start the corresponding monitoring devices or programs according to different fault diagnosis task instructions, and implement various fault diagnosis processing tasks such as voltage anomaly detection, temperature anomaly detection, real-time monitoring, and internal resistance change simulation, so as to meet the diagnosis requirements in different scenarios. When the abnormal processing unit detects abnormal battery operation data, it can obtain the detailed and regular operation parameter records collected by high-precision sensors and conventional monitoring devices, generate a sample set through periodic sampling, and use the battery identification information generation model to determine the position correlation relationship and type information of the battery cells, and then generate battery anomaly judgment information and anomaly information by means of the battery anomaly detection model, so as to achieve accurate positioning of the abnormal battery cells and accurate judgment of the anomaly type. In addition, the system can also feedback the battery anomaly information to the battery fault diagnosis model, trigger in-depth detection and repair operations on the abnormal battery cells, and improve the operation and maintenance efficiency and safety of the energy storage power station.

[0068] The system demonstrates innovation and practicality in multiple aspects. In terms of model construction and data processing, by constructing an initial three-dimensional battery pack model and combining comprehensive real-time operation data, it can provide a more accurate reference basis for fault diagnosis and improve the accuracy of diagnosis results. In terms of the diversity and flexibility of fault diagnosis tasks, the system can execute a variety of fault diagnosis processing tasks, meet the diagnosis requirements in different scenarios, and improve the applicability of the system. In terms of the abnormal processing mechanism, the system can not only accurately locate the abnormal battery cells and judge the anomaly type, but also trigger in-depth detection and repair operations through the feedback mechanism, improving the operation and maintenance efficiency and safety of the energy storage power station. In addition, the system adopts advanced algorithms such as a hybrid model combining convolutional neural network and long short-term memory network, which can better extract the spatial and temporal features of battery operation parameters, and further improve the intelligent level and accuracy of fault diagnosis.

[0069] Further, the storage medium of the embodiments of the present application stores program instructions capable of implementing all the above methods. Among them, the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, or terminal devices such as computers, servers, mobile phones, and tablets.

[0070] The above description is only some preferred embodiments of the present invention and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features having similar functions disclosed in the embodiments of the present invention.

Claims

1. A battery pack fault diagnosis system for an energy storage power station based on a network model, characterized in that, The system includes: A first acquisition unit, configured to acquire the initial performance parameters and the structural design diagram of the battery pack of the target energy storage power station; A second acquisition unit, configured to construct an initial three-dimensional battery pack model according to the initial performance parameters and the structural design diagram; A third acquisition unit, configured to acquire the real-time operation data corresponding to the plurality of sensors through the plurality of sensors deployed in the battery pack of the target energy storage power station, where the real-time operation data includes battery voltage, battery current, battery temperature, battery internal resistance, battery charge and discharge state, and the remaining percentage of battery power; A first generation unit, configured to generate battery operation state analysis information of the battery pack of the target energy storage power station according to the real-time operation data; A second generation unit, configured to generate a battery fault diagnosis model of the battery pack of the target energy storage power station according to the real-time operation data and the battery operation state analysis information; An execution unit, configured to perform at least one fault diagnosis processing task on the battery pack of the target energy storage power station by using the battery fault diagnosis model.

2. The system according to claim 1, characterized in that, The at least one fault diagnosis processing task includes a voltage anomaly detection task, a temperature anomaly detection task, a first real-time monitoring task based on an intelligent monitor, an internal resistance change simulation task, and a second real-time monitoring task based on a remote monitoring platform; The execution unit performing the at least one fault diagnosis processing task includes: In response to receiving a voltage anomaly detection instruction for the battery pack of the target energy storage power station, starting at least one voltage monitoring device in the battery pack of the target energy storage power station by using the battery fault diagnosis model to perform the voltage anomaly detection task; In response to receiving a temperature anomaly detection instruction for the battery pack of the target energy storage power station, starting at least one temperature monitoring device in the battery pack of the target energy storage power station by using the battery fault diagnosis model to perform the temperature anomaly detection task; In response to receiving a first real-time monitoring instruction for the battery pack of the target energy storage power station, starting the intelligent monitor by using the battery fault diagnosis model to perform the first real-time monitoring task; In response to receiving a second real-time monitoring instruction for the battery pack of the target energy storage power station, performing the second real-time monitoring task by using the battery fault diagnosis model and the remote monitoring platform; In response to receiving an internal resistance change simulation instruction for the battery pack of the target energy storage power station, starting a simulation analysis program by using the battery fault diagnosis model to perform the internal resistance change simulation task.

3. The system according to claim 1, characterized in that The system further includes: An anomaly handling unit, configured to perform the following operations in response to the battery operation data anomaly information output by the battery fault diagnosis model: Acquiring the detailed operation parameter records of the battery cells in the battery pack of the target energy storage power station collected by high-precision sensors within a predetermined time period; Acquiring the conventional operation parameter records of the battery cells within the predetermined time period collected by conventional monitoring devices; Performing periodic sampling on the detailed operation parameter records and the conventional operation parameter records to generate a detailed operation parameter sample set and a conventional operation parameter sample set; Input each of the conventional operation parameter samples in the conventional operation parameter sample set into the first battery identification information generation model to generate a battery position information set and a battery type information set for the battery cell, and obtain a battery position information set sequence and a battery type information set sequence; Determine the position association relationship information between the battery cells and the battery type information of each battery cell according to the battery position information set sequence and the battery type information set sequence; Input each detailed operation parameter sample in the detailed operation parameter sample set and the corresponding position association relationship information into the second battery identification information generation model to generate detailed parameter battery position information and detailed parameter battery type information for each battery cell; Input the detailed operation parameter sample, the corresponding detailed parameter battery position information and detailed parameter battery type information into the battery anomaly detection model to generate battery anomaly judgment information for each battery cell; Generate battery anomaly information for the battery cell according to the battery anomaly judgment information.

4. The system according to claim 3, wherein The anomaly processing unit is further configured to: Send the battery anomaly information to the battery fault diagnosis model to trigger the battery fault diagnosis model to instruct the target maintenance device to perform in-depth detection and repair operations on the abnormal battery cell.

5. The system according to claim 3, wherein The generation of the battery anomaly information for the battery cell includes: Perform sequence adjustment on the battery anomaly judgment information to generate a battery anomaly judgment information sequence for each battery cell; Extract target anomaly judgment information indicating abnormal battery voltage, abnormal battery temperature, abnormal battery internal resistance, or abnormal battery charge and discharge state from each battery anomaly judgment information sequence; Calculate the ratio value between the number of the target anomaly judgment information and the total number of the battery anomaly judgment information sequences; If the ratio value is less than a preset threshold, generate battery anomaly information indicating that the battery cell has no anomaly; If the ratio value is greater than or equal to the preset threshold, generate battery anomaly information indicating that the battery cell has an anomaly.

6. The system according to claim 3, wherein The determination of the position association relationship information between the battery cells and the battery type information of each battery cell includes: Obtain the actual battery position information of each battery cell according to the battery position information set sequence; Based on the overall battery position information of the target energy storage power station battery pack, filter out the invalid position information that exceeds the overall battery position range in the actual battery position information; Generate the position association relationship information and the battery type information of each battery cell according to the filtered actual battery position information and the corresponding battery type information.

7. The system according to claim 6, wherein The obtaining of the actual battery position information includes: Input the conventional operation parameter sample into the parameter sample and actual position corresponding transformation model; Through the parameter sample and actual position corresponding transformation model, convert the coordinate position in the conventional operation parameter sample into an actual position coordinate to generate the actual battery position information.

8. The system according to claim 3, wherein The first battery identification information generation model is a convolutional neural network model, and the structure of the convolutional neural network model includes: An input layer for receiving a conventional operation parameter sample image; A convolutional layer for performing feature extraction on the conventional operation parameter sample image through a plurality of convolutional kernels to generate a plurality of feature maps; Pooling layer, used to downsample the feature map to reduce the dimension and retain key features; Fully connected layer, used to map the feature vector after dimensionality reduction to the output spaces of the battery position information set and the battery type information set; Output layer, used to output the battery position information set and the battery type information set.

9. The system according to claim 3, characterized in that, The second battery identification information generation model is a hybrid model combining a convolutional neural network and a long short-term memory network. The structure of the hybrid model includes: Input layer, used to receive detailed operation parameter sample data; Convolutional neural network layer, used to extract spatial features in the detailed operation parameter sample data; Long short-term memory network layer, used to extract time series features in the detailed operation parameter sample data; Fully connected layer, used to fuse the spatial features and time series features and map them to the output spaces of the detailed parameter battery position information and the detailed parameter battery type information; Output layer, used to output the detailed parameter battery position information and the detailed parameter battery type information.

10. The system according to claim 7, characterized in that, The parameter sample and actual position correspondence transformation model is a geometric transformation model. The working steps of the geometric transformation model include: Receiving a target conventional operation parameter sample and its corresponding battery position information; According to a predefined coordinate transformation function, converting the battery position information from the parameter sample coordinate system to the actual position coordinate system; Outputting the converted actual battery position information, which is used to determine the position association relationship information and battery type information of the battery cell.

Images