Method and system for monitoring of submerged user-side energy storage device

Abstract

The application discloses a kind of monitoring method and system of submerged user side energy storage equipment, it is related to submerged energy storage equipment technical field, the training of model is carried out by the historical monitoring data of submerged energy storage equipment, obtains feedback verification model, after the acquisition of real-time monitoring data is carried out by data monitoring unit, if according to the set threshold value determining that the real-time monitoring data of acquisition exists anomaly, then according to real-time monitoring data, generate predicted three-dimensional temperature field by feedback verification model, generate actual three-dimensional temperature field by temperature field construction unit, and by subsequent digitized processing and comparison analysis, verify and judge that the real-time monitoring data of acquisition appears abnormal is data monitoring unit abnormal or submerged energy storage equipment itself abnormal, can carry out further abnormal diagnosis when real-time monitoring data appears abnormal, determine abnormal reason, avoid the redundant design of using data monitoring unit, save hardware cost, improve the precision of submerged energy storage equipment monitoring.

Description

Technical Field

[0001] This invention relates to the field of submerged energy storage equipment technology, specifically to a monitoring method and system for submerged user-side energy storage equipment. Background Technology

[0002] Immersion user-side energy storage equipment is an energy storage system that completely immerses energy storage batteries or other energy storage media in insulating coolant. It is mainly deployed in industrial and commercial parks, shopping malls, residences, data centers, etc., to realize the storage and flexible dispatch of electrical energy. Its core feature is that the energy storage unit is directly in contact with the liquid, so as to achieve efficient heat dissipation, safety protection and system compactness.

[0003] Chinese invention patent (CN119153819A) discloses an "immersion-type energy storage battery box and its monitoring method, electronic device and storage medium", specifically disclosing: insulating oil is introduced into the battery box through an immersion system; the average temperature of the battery module is obtained through a first temperature sensor; when the average temperature is greater than a first preset temperature, the battery management system controls the immersion system to adjust the temperature of the insulating oil to cool the battery module; when the average temperature is greater than a second preset temperature, the battery management system controls the immersion system to adjust the temperature of the insulating oil and controls the cooling system to introduce cooling gas into the hollow flow channel to cool the battery module; the conductivity of the insulating oil is detected by a built-in sensor, and the insulating oil is replaced when the conductivity of the insulating oil exceeds a first preset value; the oil leakage status of the battery box is detected by an external sensor, and an alarm is triggered when oil leakage occurs in the battery box.

[0004] In the aforementioned publicly available technical solutions, the monitoring of submerged energy storage devices is all done through hardware devices to monitor the data of the energy storage devices. When the monitoring data does not meet the system's set conditions, it is determined that the energy storage device is abnormal and corresponding remedial measures are taken. However, this method lacks effective verification and mutual feedback, because the abnormal monitoring data may also be caused by the hardware monitoring device malfunctioning. Conventional redundancy design will lead to incomplete coverage and increased hardware costs, and is prone to misdiagnosis of faults and increased hardware failure rate, affecting the normal use of submerged energy storage devices.

[0005] Therefore, there is an urgent need for a new monitoring method and system for submerged user-side energy storage devices to solve the above-mentioned technical problems. Summary of the Invention

[0006] The purpose of this invention is to provide a monitoring method and system for submerged user-side energy storage devices to solve the problems raised in the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a monitoring method for submerged user-side energy storage devices, comprising the following steps:

[0008] S1. Use the data retrieval unit to retrieve historical monitoring data from the storage database of the submerged energy storage device, and train the model through the model training module based on the historical monitoring data to obtain a feedback verification model.

[0009] S2. The data monitoring unit collects real-time monitoring data of the submerged energy storage device. In the event of abnormal real-time monitoring data, the real-time monitoring data is input into the feedback verification model. The feedback verification model generates a predicted three-dimensional temperature field, and the actual three-dimensional temperature field is generated by the temperature field construction unit based on the real-time monitoring data.

[0010] S3. The generated actual three-dimensional temperature field and the predicted three-dimensional temperature field are digitally processed by the digital processing module to obtain the actual isothermal region function and the predicted isothermal region function.

[0011] S4. Use the comparison analysis module to compare the similarity between the actual isothermal region function and the predicted isothermal region function, and verify the abnormal real-time monitoring data based on the similarity comparison results. If the similarity comparison results meet the set threshold, it is determined that the data monitoring unit is abnormal; otherwise, it is determined that the submerged energy storage device is abnormal.

[0012] According to the above technical solution, in step S1, the feedback verification model is constructed based on a deep neural network model, the retrieved historical monitoring data is divided into datasets, and the divided datasets are preprocessed, including normalization and standardization.

[0013] The coolant inlet temperature data in the input features are respectively... Coolant outlet temperature data and ambient temperature data Normalization is performed, and the specific processing method is as follows:

[0014] ;

[0015] in, This represents the temperature data after normalization. , This indicates a specific temperature data point in the second monitoring dataset. and These represent the minimum and maximum values ​​of a certain type of temperature data in the second monitoring data, respectively;

[0016] The coolant flow rate data in the input features are analyzed separately. and coolant flow rate data The standardization process is as follows: ;

[0017] in, This represents the standardized flow rate or velocity data, where X represents a specific flow rate or velocity data in the second monitoring data, and μ and σ represent the mean and standard deviation of a certain type of data in the second monitoring data, respectively.

[0018] The processing of the output features includes expanding the three-dimensional temperature field distribution data into a one-dimensional vector, and then performing the same normalization processing as the second monitoring data. The details will not be elaborated here.

[0019] Any set of historical monitoring data includes the first monitoring data and the second monitoring data;

[0020] The second monitoring data is used as the input feature of the feedback validation model, and the first monitoring data is used as the output feature of the feedback validation model. A loss function is defined. ;

[0021] The model training module is based on input features, output features, and a loss function. The deep neural network model is trained to obtain a feedback verification model.

[0022] According to the above technical solution, the historical monitoring data includes three-dimensional temperature field distribution data of the submerged energy storage device, coolant inlet temperature data, coolant outlet temperature data, coolant flow rate data, coolant velocity data, and ambient temperature data.

[0023] The three-dimensional temperature field distribution data of the submerged energy storage device is the primary monitoring data;

[0024] Coolant inlet temperature data, coolant outlet temperature data, coolant flow rate data, coolant velocity data, and ambient temperature data are the second monitoring data;

[0025] loss function ;

[0026] in, This indicates the total number of samples in the second monitoring data. Let represent the loss function for the i-th sample;

[0027] Wherein, the loss function of the i-th sample Represented as:

[0028] ;

[0029] in, This represents the number of grid points in the three-dimensional temperature field distribution data, where j represents the j-th grid point in the i-th sample. This represents the weight value of the j-th grid point in the first monitoring data. This represents the predicted value of the number of the j-th grid point. This represents the actual value of the number of the j-th grid point;

[0030] The retrieved historical monitoring data is divided into a training set, a validation set, and a test set;

[0031] The model training module trains the deep neural network model using a training set, then validates the trained model using a validation set, tests the model using a test set, and finally obtains a feedback validation model.

[0032] According to the above technical solution, in step S2, the temperature data inside the submerged energy storage device collected by the data monitoring unit is defined as the third monitoring data, and the coolant inlet temperature data, coolant outlet temperature data, coolant flow rate data, coolant velocity data and ambient temperature data collected by the data monitoring unit are defined as the fourth monitoring data.

[0033] Determine if there are any anomalies in the real-time monitoring data by setting threshold conditions;

[0034] If no anomalies are found, load the timestamp of the collected real-time monitoring data and store it in the storage database;

[0035] If an anomaly is found, the fourth monitoring data will be transmitted to the feedback verification model, which will then generate a predicted three-dimensional temperature field based on the fourth monitoring data.

[0036] The third monitoring data is transmitted to the temperature field construction unit, which then generates the actual three-dimensional temperature field.

[0037] According to the above technical solution, the temperature data inside the submersible energy storage device is collected through distributed optical fibers;

[0038] The construction of the actual three-dimensional temperature field includes the following steps:

[0039] Based on the physical layout path of the distributed optical fiber, a mapping relationship between the location of the distributed optical fiber and the three-dimensional coordinates of the energy storage device is established, and the physical location of each measurement point of the distributed optical fiber is recorded. , where k represents the k-th measurement point of the distributed optical fiber;

[0040] Extracting discrete temperature points from temperature data acquired via distributed optical fibers. The temperature of uncovered areas is inferred through interpolation algorithms, such as Kriging interpolation, radial basis function interpolation, or inverse distance weighting, and the optimal interpolation parameters are selected through cross-validation.

[0041] By combining the heat conduction equations of submerged energy storage devices, such as Fourier's law, and using the Kalman filter algorithm with temperature data from distributed optical fibers to correct model parameters in real time, the interpolation results are corrected to ensure they conform to actual heat diffusion laws. Finally, the three-dimensional temperature field is solved using the finite element method or finite volume method to generate the actual three-dimensional temperature field data. .

[0042] According to the above technical solution, in step S3, the digital processing of the actual three-dimensional temperature field and the predicted three-dimensional temperature field includes the following steps:

[0043] S3-1. Establish three-dimensional rectangular coordinate systems on the actual three-dimensional temperature field and the predicted three-dimensional temperature field model respectively, and assign coordinate values ​​to each coordinate point;

[0044] S3-2. Slice the actual and predicted three-dimensional temperature fields, i.e., select a plane in three-dimensional space, for example: In a plane, establish a corresponding Cartesian coordinate system, project the actual three-dimensional temperature field and the predicted three-dimensional temperature field onto the same selected plane, and obtain the actual two-dimensional temperature distribution function. and predicting two-dimensional temperature distribution function ;

[0045] S3-3. Based on the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function, solve for the target temperature, that is: for the target temperature C, solve the equations respectively. and The solution set of the equation is An isotherm with temperature C on a plane is expressed as a region function, obtained through... This indicates that in the actual three-dimensional temperature field, The actual isothermal region function on the plane, through This indicates that in the predicted three-dimensional temperature field, The function for predicting isothermal regions on a plane, where n represents the region in the plane. The function of the nth isothermal region on the plane;

[0046] The functions of the actual isothermal regions are summarized and organized to obtain the sets of functions of the actual isothermal regions. and the set of functions for the actual isothermal region ;

[0047] According to the above technical solution, in step S4, the area and perimeter of the actual isothermal region function and the predicted isothermal region function are calculated based on the basic geometric properties of the isothermal region function.

[0048] If the difference between the area or perimeter data of a certain region in the actual three-dimensional temperature field and the corresponding area or perimeter data in the predicted three-dimensional temperature field exceeds a set threshold, it is determined that the data monitoring unit is abnormal. If it does not exist, it is determined that the submerged energy storage device is abnormal.

[0049] A monitoring system for an immersion user-side energy storage device includes a storage database, a data retrieval unit, a model training module, a data monitoring unit, an anomaly detection unit, a temperature field construction unit, a digital processing module, and a comparison and analysis module.

[0050] The storage database is used to store monitoring data of the submerged energy storage device; the data retrieval unit is used to retrieve historical monitoring data from the storage database; the model training module is used to train the model based on the retrieved historical monitoring data to obtain a feedback verification model; the data monitoring unit is used to collect real-time monitoring data of the submerged energy storage device; the anomaly determination unit is used to determine whether there are anomalies in the real-time monitoring data; the temperature field construction unit is used to construct the actual three-dimensional temperature field of the submerged energy storage device based on the real-time monitoring data; the feedback verification model generates the predicted three-dimensional temperature field of the submerged energy storage device based on the real-time monitoring data; the digitization processing module is used to digitize the actual three-dimensional temperature field and the predicted three-dimensional temperature field to obtain the actual isothermal region function and the predicted isothermal region function; the comparison and analysis module is used to compare the similarity between the actual isothermal region function and the predicted isothermal region function and output the comparison results.

[0051] According to the above technical solution, the digital processing module includes a coordinate system establishment unit, a temperature field slicing unit, and a function generation unit;

[0052] The coordinate system establishment unit is used to establish a three-dimensional rectangular coordinate system on the actual three-dimensional temperature field and the predicted three-dimensional temperature field; the temperature field slicing unit is used to slice the actual three-dimensional temperature field and the predicted three-dimensional temperature field, and to map them according to the slicing plane to obtain the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function; the function generation unit solves for the target temperature based on the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function to obtain the actual isothermal region function and the predicted isothermal region function.

[0053] According to the above technical solution, the system also includes an anomaly warning unit, which is used to determine, based on the comparison results of the comparison and analysis module, whether the anomaly in the real-time monitoring data of the submerged energy storage device is due to an anomaly in the submerged energy storage device itself or to an anomaly in the data monitoring unit, and then issue an early warning.

[0054] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention trains a model using historical monitoring data from submerged energy storage devices to obtain a feedback verification model. After real-time monitoring data is collected through a data monitoring unit, if an anomaly is detected in the collected real-time monitoring data based on a set threshold, a predicted three-dimensional temperature field is generated through the feedback verification model based on the real-time monitoring data. An actual three-dimensional temperature field is generated through a temperature field construction unit. Through subsequent digital processing and comparative analysis, it is verified and determined whether the anomaly in the collected real-time monitoring data is due to an anomaly in the data monitoring unit or an anomaly in the submerged energy storage device itself. This allows for further anomaly diagnosis when anomalies occur in the real-time monitoring data, determining the cause of the anomaly. This avoids redundant design of the data monitoring unit, saves hardware costs, and improves the accuracy of monitoring submerged energy storage devices. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the logic flow of the monitoring method for the submerged user-side energy storage device of the present invention;

[0056] Figure 2 This is a schematic diagram of the steps in the monitoring method for the submerged user-side energy storage device of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Example: Figures 1-2 As shown, the present invention provides a technical solution for a monitoring method of an immersion user-side energy storage device, comprising the following steps:

[0059] S1. Use the data retrieval unit to retrieve historical monitoring data from the storage database of the submerged energy storage device. Divide the retrieved historical monitoring data into datasets. For example, retrieve the most recent 1000 sets of historical monitoring data, divide 70% of the historical monitoring data into the training set, 20% into the validation set, and 10% into the test set. Specifically, the retrieved historical monitoring data is the normal monitoring data of the energy storage device, because abnormal monitoring data will affect the output results after model training.

[0060] In this embodiment, any set of historical monitoring data includes first monitoring data and second monitoring data, wherein the first monitoring data is the three-dimensional temperature field distribution data of the submerged energy storage device. Specifically, temperature data is collected through distributed optical fibers. Based on the physical layout of the distributed optical fibers, a mapping relationship is established between the location of the distributed optical fibers and the three-dimensional coordinates of the energy storage device. Using one-dimensional optical fiber data points, an interpolation algorithm is used to estimate the temperature of uncovered areas. Combined with the heat conduction equations of the energy storage device, such as Fourier's law, the interpolation results are corrected to ensure that they conform to the actual heat diffusion law. The second monitoring data is the coolant inlet temperature data of the submersible energy storage device. Coolant outlet temperature data Coolant flow rate data Coolant flow rate data and ambient temperature data Data is collected through sensors.

[0061] Based on the segmented historical monitoring data training set, the model is trained using the model training module to obtain a feedback validation model. ;

[0062] Specifically, the feedback verification model is constructed based on a multilayer perceptron. The feedback verification model consists of an input layer, a hidden layer, and an output layer. The input layer is used to receive the original feature data, the hidden layer is used to perform nonlinear transformations, and the output layer is used to generate the final prediction result.

[0063] In this embodiment, the second monitoring data in the training set is used as the input feature for training the feedback verification model, and the first monitoring data is used as the output feature for training the feedback verification model.

[0064] The coolant inlet temperature data in the input features are respectively... Coolant outlet temperature data and ambient temperature data Normalization is performed, and the specific processing method is as follows:

[0065] ;

[0066] in, This represents the temperature data after normalization. , This indicates a specific temperature data point in the second monitoring dataset. and These represent the minimum and maximum values ​​of a certain type of temperature data in the second monitoring data, respectively;

[0067] The coolant flow rate data in the input features are analyzed separately. and coolant flow rate data The standardization process is as follows:

[0068] ;

[0069] in, This represents the standardized flow rate or velocity data. This refers to a specific flow rate or velocity data point within the second monitoring data set. and These represent the mean and standard deviation of a certain type of data in the second monitoring data, respectively.

[0070] For example, a set of second monitoring data after normalization and standardization is represented as: [0.3, 0.6, -0.2, 1.1, 0.4], corresponding to the normalized and standardized coolant inlet temperature data, respectively. Coolant outlet temperature data Coolant flow rate data Coolant flow rate data and ambient temperature data ;

[0071] The processing of the output features includes expanding the three-dimensional temperature field distribution data into a one-dimensional vector, and then performing the same normalization processing as the second monitoring data. The details will not be elaborated here.

[0072] For example: three-dimensional temperature field distribution data The grid is 30*30*30. It is expanded into a one-dimensional vector, with a total of 27,000 dimensions. After normalization, it forms a 27,000-dimensional vector, with each value between [0,1], corresponding to the temperature data of any grid point after normalization.

[0073] In feedback validation model During the training process, after preprocessing the first and second monitoring data, a loss function is defined. Specifically: ;

[0074] in, This indicates the total number of samples in the second monitoring data. Let represent the loss function for the i-th sample;

[0075] Wherein, the loss function of the i-th sample Represented as:

[0076] ;

[0077] in, This represents the number of grid points in the three-dimensional temperature field distribution data. For example, M=27000, and j represents the j-th grid point in the i-th sample. This represents the weight value of the j-th grid point in the first monitoring data, with higher weights assigned to grid points closer to the coolant channel. This represents the predicted value of the number of the j-th grid point. This represents the actual value of the number of grid points j.

[0078] Based on the above, a multilayer perceptron is trained. The trained model is then validated using a validation set and tested using a test set, ultimately yielding a feedback validation model. .

[0079] This invention trains the model using historical monitoring data to obtain a feedback verification model. The purpose is to verify the monitoring data and determine the cause of the anomaly when anomalies occur, avoiding the judgment of anomalies based solely on the submerged energy storage device. This makes the monitoring of submerged energy storage devices more comprehensive and ensures their normal use.

[0080] S2. Use the data monitoring unit to collect real-time monitoring data of the submerged energy storage device;

[0081] Specifically, the real-time monitoring data collected includes: temperature data inside the submerged energy storage device collected via distributed optical fiber; coolant inlet temperature data, coolant outlet temperature data, and ambient temperature data collected via temperature sensors; and coolant flow rate data and coolant velocity data collected via flow sensors. The temperature data inside the submerged energy storage device collected via distributed optical fiber is defined as the third monitoring data, and the coolant inlet temperature data, coolant outlet temperature data, and ambient temperature data collected via temperature sensors, as well as the coolant flow rate data and coolant velocity data collected via flow sensors, are defined as the fourth monitoring data.

[0082] Determine if there are any anomalies in the real-time monitoring data by setting threshold conditions;

[0083] If no anomalies are found, load the timestamp of the collected real-time monitoring data and store it in the storage database;

[0084] If an anomaly is detected, the fourth monitoring data will be transmitted to the feedback verification model. Validate the model through feedback A predicted three-dimensional temperature field is generated based on the fourth monitoring data, and the predicted three-dimensional temperature field data is obtained. ;

[0085] The third monitoring data is transmitted to the temperature field construction unit, which then generates the actual three-dimensional temperature field.

[0086] Specifically: Based on the physical layout path of the distributed optical fiber, establish a mapping relationship between the location of the distributed optical fiber and the three-dimensional coordinates of the energy storage device, and record the physical location of each measurement point of the distributed optical fiber. Where k represents the k-th measurement point of the distributed optical fiber; discrete temperature points are extracted from the temperature data collected by the distributed optical fiber. The temperature of uncovered areas is inferred through interpolation algorithms, such as Kriging interpolation, radial basis function interpolation, or inverse distance weighting, and the optimal interpolation parameters are selected through cross-validation.

[0087] By combining the heat conduction equations of submerged energy storage devices, such as Fourier's law, and using the Kalman filter algorithm with temperature data from distributed optical fibers to correct model parameters in real time, the interpolation results are corrected to ensure they conform to actual heat diffusion laws. Finally, the three-dimensional temperature field is solved using the finite element method or finite volume method to generate the actual three-dimensional temperature field data. .

[0088] S3. The generated actual three-dimensional temperature field and the predicted three-dimensional temperature field are digitally processed by the digital processing module respectively.

[0089] Specifically, it includes the following steps:

[0090] S3-1. Establish three-dimensional rectangular coordinate systems on the actual three-dimensional temperature field and the predicted three-dimensional temperature field model respectively, and assign coordinate values ​​to each coordinate point;

[0091] S3-2. Slice the actual and predicted three-dimensional temperature fields, i.e., select a plane in three-dimensional space, for example: In a plane, establish a corresponding Cartesian coordinate system, project the actual three-dimensional temperature field and the predicted three-dimensional temperature field onto the same selected plane, and obtain the actual two-dimensional temperature distribution function. and predicting two-dimensional temperature distribution function ;

[0092] S3-3. Based on the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function, solve for the target temperature, that is: for the target temperature C, solve the equations respectively. and The solution set of the equation is An isotherm with temperature C on a plane is expressed as a region function, obtained through... This indicates that in the actual three-dimensional temperature field, The actual isothermal region function on the plane, through This indicates that in the predicted three-dimensional temperature field, The function for predicting isothermal regions on a plane, where n represents the region in the plane. The function of the nth isothermal region on a plane; for example: if the predicted two-dimensional temperature distribution function on a certain plane is... Then the equation for the predicted isotherm is: The function for predicting the isothermal region is then: The functions of the actual isothermal regions are summarized and organized to obtain the sets of functions of the actual isothermal regions. and the set of functions for the actual isothermal region ;

[0093] Through the above technical solutions, the actual three-dimensional temperature field and the predicted three-dimensional temperature field are further digitally processed. The isothermal region is processed by expressing it through a region function, which enables precise comparative analysis and makes the comparative analysis results more visual and traceable, thus making the comparative analysis results more valuable for reference and learning.

[0094] S4. Use the comparison analysis module to compare the similarity between the actual isothermal region function and the predicted isothermal region function;

[0095] Furthermore, based on the fundamental geometric properties of the isothermal region function, further data calculations are performed on the actual isothermal region function and the predicted isothermal region function.

[0096] Specifically, the area and perimeter of the actual isothermal region function and the predicted isothermal region function can be calculated using integral operations.

[0097] The set of regions that form the actual isothermal region function Set with perimeter ;

[0098] And the set of regions with areas that predict the isothermal region function Set with perimeter ;

[0099] Where s represents the area data of the region, and d represents the perimeter data. Indicates in On a plane, m represents the position of the plane. The area or perimeter data of the m-th region on the plane;

[0100] If the difference between the area or perimeter data of a certain region in the actual three-dimensional temperature field and the corresponding area or perimeter data in the predicted three-dimensional temperature field exceeds a set threshold, it is determined that the data monitoring unit is abnormal; if it is not, it is determined that the submerged energy storage device is abnormal.

[0101] The above technical solutions enable comparative analysis based on digitalization, allowing for more accurate identification of the cause of anomalies based on the comparison results. This facilitates timely adjustments and avoids misdiagnosis of submerged energy storage equipment anomalies caused by abnormalities in the data monitoring unit.

[0102] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method of monitoring an immersed user-side energy storage device, characterized in that, The method comprises the following steps: S1, using a data retrieval unit to retrieve historical monitoring data from the storage database of the submerged energy storage device, and training a model through a model training module based on the historical monitoring data to obtain a feedback verification model; S2, using a data monitoring unit to collect real-time monitoring data of the submerged energy storage device, and in the case of abnormal real-time monitoring data, inputting the real-time monitoring data into the feedback verification model to generate a predicted three-dimensional temperature field through the feedback verification model, and generating an actual three-dimensional temperature field through a temperature field construction unit based on the real-time monitoring data; S3, performing digital processing on the generated actual three-dimensional temperature field and predicted three-dimensional temperature field through a digital processing module to obtain an actual isothermal region function and a predicted isothermal region function; S4, using a comparison and analysis module to compare the similarity of the actual isothermal region function and the predicted isothermal region function, and based on the similarity comparison result, verifying the abnormal real-time monitoring data, if the similarity comparison result meets the set threshold, determining that the data monitoring unit is abnormal, otherwise, determining that the submerged energy storage device is abnormal.

2. The monitoring method of an immersed user-side energy storage device according to claim 1, characterized in that, In step S1, the feedback verification model is constructed based on a deep neural network model, the historical monitoring data is divided into data sets, and the data sets after division are preprocessed, including normalization and standardization; Any one group of historical monitoring data includes first monitoring data and second monitoring data; The second monitoring data is taken as an input feature of the feedback verification model, the first monitoring data is taken as an output feature of the feedback verification model, and a loss function is defined ; The model training module is based on input features, output features and loss function The deep neural network model is trained to obtain a feedback verification model.

3. The method of claim 2, wherein: The historical monitoring data includes three-dimensional temperature field distribution data, cooling liquid inlet temperature data, cooling liquid outlet temperature data, cooling liquid flow data, cooling liquid flow rate data, and environmental temperature data of the submerged energy storage device; The three-dimensional temperature field distribution data of the submerged energy storage device is the first monitoring data; The cooling liquid inlet temperature data, cooling liquid outlet temperature data, cooling liquid flow data, cooling liquid flow rate data, and environmental temperature data are the second monitoring data; Loss function ; wherein, denotes the total number of samples in the second monitoring data, denotes the loss function of the i-th sample; where the loss function of the ith sample is is represented as: ； wherein, represents the number of grid points in the three-dimensional temperature field distribution data, j represents the jth grid point in the ith sample, represents the weight value of the jth grid point number in the first monitoring data, represents the predicted value of the jth grid point number, represents the true value of the jth grid point number; The retrieved historical monitoring data is divided into a training set, a verification set, and a test set; The model training module trains the deep neural network model through the training set, then verifies the trained model through the verification set, tests the model through the test set, and finally obtains the feedback verification model.

4. The method of claim 1, wherein: In step S2, the temperature data inside the submerged energy storage device collected by the data monitoring unit is defined as third monitoring data, and the cooling liquid inlet temperature data, cooling liquid outlet temperature data, cooling liquid flow data, cooling liquid flow rate data, and environmental temperature data collected by the data monitoring unit are defined as fourth monitoring data; Determine whether the real-time monitoring data is abnormal by setting a threshold condition; If there is no abnormality, load the time stamp collected for the real-time monitoring data and store it in the storage database; If there is an abnormality, the fourth monitoring data is transmitted to the feedback verification model to generate a predicted three-dimensional temperature field based on the fourth monitoring data through the feedback verification model; The third monitoring data is transmitted to the temperature field construction unit to generate an actual three-dimensional temperature field through the temperature field construction unit.

5. The method of claim 4, wherein: The temperature data inside the submerged energy storage device is collected through a distributed optical fiber; The construction of the actual three-dimensional temperature field comprises the following steps: S2-1, according to the physical arrangement path of the distributed optical fiber, a mapping relationship between the distributed optical fiber position and the three-dimensional coordinates of the energy storage device is established, and the physical position of each measurement point of the distributed optical fiber is recorded; S2-2, extract discrete temperature points from the temperature data collected by the distributed optical fiber, predict the temperature of the uncovered area by interpolation algorithm, and select the best interpolation parameter by cross-validation; S2-3, combine the heat conduction equation of the immersed energy storage device, and use the temperature data of the distributed optical fiber to correct the model parameters in real time through Kalman filtering algorithm to modify the interpolation result; S2-4, combine the finite element or finite volume method to solve the three-dimensional temperature field, and generate the actual three-dimensional temperature field. 6.The method of claim 1, wherein, In step S3, for the digital processing of the actual three-dimensional temperature field and the predicted three-dimensional temperature field, the following steps are included: S3-1, respectively, in the actual three-dimensional temperature field and the predicted three-dimensional temperature field model, a three-dimensional rectangular coordinate system is established, and each coordinate point is given a coordinate value; S3-2, slice the actual three-dimensional temperature field and the predicted three-dimensional temperature field, establish the corresponding plane rectangular coordinate system, project the actual three-dimensional temperature field and the predicted three-dimensional temperature field onto the same selected plane, and respectively obtain the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function; S3-3, according to the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function, the target temperature is solved to obtain the actual isothermal region function and the predicted isothermal region function.

7. The method of claim 1, wherein: In step S4, according to the basic geometric properties of the isothermal region function, the area and perimeter of the actual isothermal region function and the predicted isothermal region function are calculated; When the difference between the area data or perimeter data of a certain region in the actual three-dimensional temperature field and the corresponding area data or perimeter data in the predicted three-dimensional temperature field exceeds the set threshold, it is determined that the data monitoring unit is abnormal, otherwise, it is determined that the immersed energy storage device is abnormal.

8. A monitoring system for an immersed user-side energy storage device for performing the monitoring method of any one of claims 1-7, characterized in that: The storage database is used to store the monitoring data of the immersed energy storage device; the data retrieval unit is used to retrieve the historical monitoring data from the storage database; the model training module is used to train the model according to the retrieved historical monitoring data to obtain the feedback verification model; the data monitoring unit is used to collect real-time monitoring data of the immersed energy storage device; the abnormality determination unit is used to determine whether the real-time monitoring data is abnormal; the temperature field construction unit is used to construct the actual three-dimensional temperature field of the immersed energy storage device according to the real-time monitoring data; the feedback verification model generates the predicted three-dimensional temperature field of the immersed energy storage device according to the real-time monitoring data, the digital processing module is used to digitally process the actual three-dimensional temperature field and the predicted three-dimensional temperature field to obtain the actual isothermal region function and the predicted isothermal region function; the comparison and analysis module is used to compare the similarity of the actual isothermal region function and the predicted isothermal region function, and output the comparison result. ​ 9. The monitoring system of an immersed user-side energy storage device according to claim 8, characterized in that: The digital processing module comprises a coordinate system establishing unit, a temperature field slicing unit and a function generating unit; The coordinate system establishing unit is configured to establish a three-dimensional rectangular coordinate system on the actual three-dimensional temperature field and the predicted three-dimensional temperature field; the temperature field slicing unit is configured to perform slicing processing on the actual three-dimensional temperature field and the predicted three-dimensional temperature field, and perform mapping according to a slicing plane to obtain an actual two-dimensional temperature distribution function and a predicted two-dimensional temperature distribution function; and the function generating unit is configured to solve a target temperature according to the actual two-dimensional temperature distribution function and the predicted two-dimensional temperature distribution function to obtain an actual isothermal region function and a predicted isothermal region function.

10. The monitoring system of an immersed user-side energy storage device according to claim 8, characterized in that: The system further comprises an abnormality early warning unit configured to determine, according to the comparison result of the comparison and analysis module, whether the real-time monitoring data of the submerged energy storage device is abnormal because of an abnormality of the submerged energy storage device itself or because of an abnormality of the data monitoring unit, and issue a warning.

Images