Energy storage device temperature control event state monitoring method and system

Abstract

The application discloses a kind of energy storage equipment temperature control event state monitoring method and system, comprising: in response to the temperature control event of energy storage equipment, determine monitoring time window and obtain temperature state data, operating load data, temperature control instruction state data and temperature control execution mechanism feedback state data;According to the temperature state data before event execution, generate pre-event temperature state estimation result, and generate predicted thermal response in combination with operating load data and temperature control instruction state data;After event execution, actual temperature response and temperature control path heat exchange result are obtained;Generate instruction-execution consistency difference amount, predicted-measured thermal response difference amount and heat closure difference amount, and generate abnormality determination result accordingly.The application can improve the state observability and abnormality determination accuracy of temperature control process.

Description

Technical Field

[0001] This invention relates to the field of temperature monitoring technology for energy storage devices, and in particular to a method and system for monitoring the status of temperature control events in energy storage devices. Background Technology

[0002] The application of residential stacked energy storage devices and industrial and commercial energy storage devices is increasing. These devices typically integrate battery modules, power conversion units, control units, and temperature control actuators such as fans, air conditioners, liquid-cooled pumps, valves, and heating modules within a relatively compact cabinet or rack space. Energy storage devices continuously generate heat during charging and discharging, and the heat load distribution among different battery modules, power devices, and air zones within the cabinet is uneven. If the temperature control process cannot reflect the actual internal thermal state of the device in a timely manner, it can easily lead to excessively high temperatures in localized areas, widened temperature differences between areas, and increased energy consumption for temperature control, thereby affecting the operational stability and lifespan of the energy storage device.

[0003] Existing energy storage devices typically collect temperature data from locations such as battery modules, internal air, the air inlet side, the air outlet side, or liquid cooling branches using temperature sensors, and trigger alarms based on preset temperature thresholds, temperature difference thresholds, or heating rate thresholds. Some solutions also control fan speed, air conditioning cooling power, liquid cooling pump frequency, valve opening, or heating power based on temperature data to maintain the internal temperature of the energy storage device within a set range. While these methods can meet general temperature monitoring and basic temperature control needs, their monitoring focus is usually on the temperature value itself, with less emphasis on correlation analysis of state changes before and after a specific temperature control action.

[0004] In actual operation, after the temperature control system issues a temperature control command, whether the temperature control actuator acts according to the command, whether the actuator produces the expected temperature response after action, and whether the temperature control path actually removes the corresponding heat all affect the temperature control effect. Existing monitoring methods mostly use whether the temperature exceeds the limit as the basis for judgment, making it difficult to distinguish whether the temperature anomaly is caused by increased operating load, insufficient response of the temperature control actuator, obstruction of heat transfer or exchange in the temperature control path, abnormal sensor data, or insufficient overall temperature control capability. When the temperature anomaly has reached the alarm threshold, the system usually can only provide a relatively general alarm result, making it difficult to pinpoint the abnormal link in a timely manner.

[0005] Furthermore, temperature changes within energy storage devices exhibit a certain degree of lag and coupling. There are mutual influences among the status of temperature control actuators, changes in operating load, the status of cooling media or airflow paths, and changes in the temperature field. Existing technologies typically lack comprehensive verification of the temperature control command status, actuator feedback status, predicted temperature response, actual temperature response, and heat balance, resulting in insufficient observability of the temperature control process. When temperature control actions have been executed but the actual temperature does not change as expected, the system struggles to promptly determine whether the issue stems from execution anomalies, abnormal thermal response, or insufficient temperature control capability, thus hindering subsequent control parameter adjustments and maintenance assessments. Therefore, it is necessary to further improve the status monitoring capabilities of the temperature control process in energy storage devices. Summary of the Invention

[0006] The purpose of this invention is to address the problem that existing energy storage equipment temperature monitoring mainly relies on temperature thresholds, temperature difference thresholds, or heating rate thresholds for alarms, making it difficult to comprehensively judge the execution of temperature control commands, the actual response of temperature control actuators, whether the temperature response meets expectations, and whether the heat exchange in the temperature control path is effective. This invention provides a method and system for monitoring the status of temperature control events in energy storage equipment, thereby improving the observability of the temperature control process and the accuracy of anomaly detection.

[0007] To achieve the above objectives, the present invention provides a method for monitoring the status of temperature control events in an energy storage device, comprising: In response to a temperature control event of an energy storage device, a corresponding monitoring time window is determined, and temperature status data, operating load data, temperature control command status data, and temperature control actuator feedback status data within the monitoring time window are acquired. Based on the temperature status data before the execution of the temperature control event in the monitoring time window, generate the pre-event temperature status estimation result, and generate the predicted thermal response based on the pre-event temperature status estimation result, operating load data, and temperature control command status data; After the temperature control event is executed, the actual temperature response and the heat exchange result of the temperature control path are obtained based on the temperature status data after the event is executed. The command-execution consistency difference is generated based on the temperature control command status data and the temperature control actuator feedback status data. The predicted-measured thermal response difference is generated based on the predicted thermal response and the actual temperature response. The heat closure difference is generated based on the operating load data, the heat exchange results of the temperature control path, and the change in temperature status data before and after the execution of the temperature control event. Based on the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference, an anomaly determination result corresponding to the temperature control event is generated.

[0008] As a further improvement to the above technical solution, determining the corresponding monitoring time window includes: using the execution start time of the temperature control event as the time reference, determining a first time interval before the execution start time and a second time interval after the execution start time; It also includes: acquiring temperature status data and operating load data within the first time interval and the second time interval, acquiring temperature control command status data corresponding to the temperature control event, and acquiring temperature control actuator feedback status data within the second time interval; associating the acquired data with the temperature control event according to the sampling time to form a monitoring status data group corresponding to the temperature control event.

[0009] As a further improvement to the above technical solution, the step of generating the pre-event temperature state estimation result includes: extracting the temperature of the measuring point corresponding to each temperature control area from the temperature state data in the first time interval according to the correspondence between the temperature control area and the temperature measuring point in the energy storage device; determining the temperature estimate and temperature distribution characteristics of each temperature control area before the execution of the temperature control event according to the measuring point temperature and the operating load data in the first time interval, and using them as the pre-event temperature state estimation result.

[0010] As a further improvement to the above technical solution, the step of generating the predicted thermal response includes: determining the initial temperature state of each temperature control zone before the execution of the temperature control event based on the pre-event temperature state estimation results; determining the heat input state corresponding to the temperature control event based on the operating load data; determining the target adjustment state of the corresponding temperature control path based on the temperature control command state data; inputting the initial temperature state, heat input state, and target adjustment state into the thermal response prediction model to obtain the predicted temperature response and the predicted heat exchange result of the temperature control path after the execution of the temperature control event, and using the predicted temperature response and the predicted heat exchange result of the temperature control path as the predicted thermal response.

[0011] As a further improvement to the above technical solution, the steps for obtaining the actual temperature response and the heat exchange result of the temperature control path include: determining the actual temperature response after the execution of the temperature control event based on the relationship between the temperature state data in the second time interval and the estimated temperature state before the event; determining the temperature control area associated with the temperature control path based on the temperature control path corresponding to the temperature control event; and generating the heat exchange result of the temperature control path based on the change in temperature state of the associated temperature control area before and after the execution of the temperature control event and the feedback state data of the temperature control actuator.

[0012] As a further improvement to the above technical solution, the steps for generating the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference include: calculating the difference between the target execution state represented by the temperature control instruction state data and the actual execution state represented by the temperature control actuator feedback state data to obtain the instruction-execution consistency difference; calculating the difference between the predicted temperature response and the actual temperature response in the predicted thermal response to obtain the prediction-measured thermal response difference; converting the heat input reference quantity corresponding to the temperature control event based on the operating load data, obtaining the actual heat exchange quantity of the temperature control path based on the heat exchange result of the temperature control path, and characterizing the cumulative heat change in the energy storage device by the change in temperature state data before and after the execution of the temperature control event; and generating the thermal closure difference based on the difference between the heat input reference quantity, the actual heat exchange quantity of the temperature control path, and the cumulative heat change.

[0013] As a further improvement to the above technical solution, the step of generating the anomaly determination result includes: matching the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference with the corresponding anomaly determination conditions respectively; when the instruction-execution consistency difference meets the corresponding anomaly determination conditions, generating an anomaly determination result for execution response anomaly; when the instruction-execution consistency difference does not meet the corresponding anomaly determination conditions, and the prediction-measured thermal response difference meets the corresponding anomaly determination conditions, generating an anomaly determination result for thermal response anomaly; when the thermal closure difference meets the corresponding anomaly determination conditions, generating an anomaly determination result for thermal closure anomaly or insufficient temperature control capability based on whether the prediction-measured thermal response difference meets the corresponding anomaly determination conditions.

[0014] As a further improvement to the above technical solution, the step of generating the anomaly determination result further includes: generating a temperature state estimation confidence level based on the deviation between the temperature state data within the first time interval and the temperature state estimation result before the event; generating a thermal response prediction confidence level based on the deviation between the predicted thermal response and the actual temperature response; configuring judgment weights for the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference based on the temperature state estimation confidence level and the thermal response prediction confidence level, and correcting the matching result of the anomaly determination condition according to the configured judgment weights to generate the anomaly determination result.

[0015] As a further improvement to the above technical solution, it also includes: selecting a target difference quantity that triggers the anomaly judgment condition from the instruction-execution consistency difference quantity, the prediction-measured thermal response difference quantity, and the thermal closure difference quantity according to the anomaly type corresponding to the anomaly judgment result; generating a status correction prompt based on the temperature control instruction status data, temperature control actuator feedback status data, actual temperature response, or temperature control path heat exchange result associated with the target difference quantity; the status correction prompt includes the object to be checked and the correction direction, and is output in association with the anomaly judgment result.

[0016] The present invention also provides a temperature control event status monitoring system for energy storage devices, comprising: a temperature acquisition device, an operating status acquisition device, a temperature control status acquisition device, a memory, a controller, and an output interface; The temperature acquisition device is used to acquire temperature status data of the energy storage device within the monitoring time window. The operation status acquisition device is used to acquire the operation load data within the monitoring time window; The temperature control status acquisition device is used to acquire temperature control command status data and temperature control actuator feedback status data; The memory is used to store the thermal response prediction model and the condition monitoring program; The controller is connected to the temperature acquisition device, the operating status acquisition device, the temperature control status acquisition device, the memory, and the output interface, and is configured to perform the following processes: in response to a temperature control event of the energy storage device, determine the corresponding monitoring time window; generate a pre-event temperature status estimation result based on the temperature status data before the temperature control event is executed, and generate a predicted thermal response by combining the operating load data and the temperature control command status data; after the temperature control event is executed, obtain the actual temperature response and the heat exchange result of the temperature control path based on the temperature status data after the event is executed; generate the command-execution consistency difference, the predicted-measured thermal response difference, and the heat closure difference, and generate an anomaly judgment result based on the command-execution consistency difference, the predicted-measured thermal response difference, and the heat closure difference; the output interface is used to output the anomaly judgment result.

[0017] The beneficial effects of the energy storage device temperature control event status monitoring method and system provided by the present invention are as follows: By using temperature control events as the monitoring object, the status of temperature control commands, the feedback status of temperature control actuators, predicted thermal response, actual temperature response, and heat exchange results along the temperature control path are incorporated into the same monitoring time window for correlation analysis. This avoids general alarms based solely on whether the temperature exceeds the limit. Through joint determination of command-execution consistency differences, predicted-measured thermal response differences, and thermal closure differences, different states such as abnormal execution response, abnormal thermal response, abnormal thermal closure, and insufficient temperature control capability can be distinguished, improving the accuracy and interpretability of anomaly identification in the temperature control process of energy storage equipment. Furthermore, by evaluating the reliability of temperature state estimation and thermal response prediction, and correcting the weighting of difference determination, the impact of model mismatch or data anomalies on state determination results can be reduced, improving the stability of temperature control state monitoring. Attached Figure Description

[0018] Figure 1 This is an operation flowchart of the energy storage device temperature control event status monitoring method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the energy storage device temperature control event status monitoring system in an embodiment of the present invention; Figure 3 This is a schematic diagram of the monitoring time window and monitoring status data group in an embodiment of the present invention; Figure 4 This is a schematic diagram of the pre-event temperature state estimation and predicted thermal response model structure in an embodiment of the present invention; Figure 5 This is a schematic diagram comparing the predicted temperature response and the actual temperature response in an embodiment of the present invention. Figure 6 This is a schematic diagram of the temperature field distribution before and after the event in an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the generation of anomaly determination results in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention. Equivalent substitutions, simple modifications or conventional adjustments made by those skilled in the art to the following embodiments without departing from the concept of the present invention should all fall within the scope of protection of the present invention.

[0020] In the description of this invention, unless otherwise expressly defined, the terms "first," "second," etc., are used only to distinguish different objects and do not indicate a sequence, degree of importance, or quantity limitation; the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion relationships. Those skilled in the art can understand the specific meaning of the relevant terms in this invention according to the specific application scenario.

[0021] In one specific embodiment, the energy storage device is a residential stacked energy storage device or an industrial / commercial energy storage cabinet. The cabinet contains several battery modules, a power conversion unit, a control unit, and a temperature control actuator. The temperature control actuator may include one or more of the following: a fan, an air conditioner, a liquid-cooled pump, a valve, a damper, or a heating unit. Temperature acquisition devices are distributed on the surface of the battery modules, in the cabinet's air channels, air inlets and outlets, and the inlet and outlet of the liquid-cooled branch lines. Operational status acquisition devices read operational load data from the energy management controller or power conversion unit of the energy storage device. Temperature control status acquisition devices read temperature control command status data and feedback status data from the temperature control controller and the temperature control actuator. The controller can be deployed within the local edge controller of the energy storage device or within a local monitoring gateway that is communicatively connected to the energy storage device.

[0022] This method uses temperature control events as the status monitoring unit. Temperature control events can be generated by changes in temperature control commands, changes in the operating status of temperature control actuators, switching of temperature control modes, temperature status entering warning conditions, or manual maintenance triggering. After an event occurs, the controller does not directly use a single point temperature exceeding the limit as the basis for an anomaly. Instead, it forms a monitoring time window around the event, uniformly collecting temperature status data, operating load data, temperature control command status data, and temperature control actuator feedback status data before and after the event. This collection result is stored internally by the controller in the form of a monitoring status data group, enabling a temperature control action, its execution feedback, and the subsequent thermal response to be correlated and processed within the same data object.

[0023] like Figure 1 As shown, the temperature control event status monitoring process of the present invention is executed in the following order: event response, pre-event status estimation, post-event response acquisition, generation of three types of difference quantities, and anomaly judgment output. Figure 1The first step corresponds to determining the monitoring time window and acquiring temperature status data, operating load data, temperature control command status data, and temperature control actuator feedback status data; the second step corresponds to generating a pre-event temperature status estimation result from the pre-event temperature status data, and generating a predicted thermal response by combining the operating load data and temperature control command status data; the third step corresponds to acquiring the actual temperature response and the heat exchange result of the temperature control path after the event is executed; the fourth step corresponds to generating the command-execution consistency difference, the predicted-measured thermal response difference, and the heat closure difference, respectively; the fifth step corresponds to generating anomaly judgment results based on the three types of differences.

[0024] Temperature status data can be recorded in engineering using key-value records or tabular records with timestamps. For example, a single temperature status record may include the sampling time, temperature control zone number, temperature measurement point number, measurement point temperature, data quality identifier, and the temperature control path number to which it belongs; the operating load record may include the sampling time, energy storage device charging and discharging power, battery cluster current, power conversion unit load, and equipment operating mode; temperature control command status data may include the command issuance time, target execution object, target speed, target frequency, target opening degree, or target power; temperature control actuator feedback status data may include actual speed, actual frequency, actual opening degree, actual power, operating status bit, and fault status bit. The above data can be implemented using JSON messages, CSV tables, relational database tables, or time-series database records.

[0025]

[0026] In one specific implementation, after receiving a temperature control event, the controller uses the start time of the temperature control event as a time reference, dividing the sampled data before that time into a first time interval and the sampled data after that time into a second time interval. The first time interval describes the thermal state and load state before the event execution, while the second time interval describes the feedback from the actuator and the temperature change after the event execution. The duration of the first and second time intervals can be set by the device's thermal inertia, sampling period, and actuator response time, or it can be determined by the controller calling a preset window template under different temperature control modes.

[0027] The controller reads temperature status data and operating load data within the first time interval, and also reads the temperature control command status data corresponding to the temperature control event. Within the second time interval, it continues to read temperature status data and operating load data, while simultaneously reading the feedback status data from the temperature control actuator. Subsequently, the controller associates the above data with the event identifier of the temperature control event according to the sampling time. For example, the event identifier can be generated using a combination of "device number + execution object number + start time." The associated monitoring status data group includes an event header, a pre-event data sequence, a post-event data sequence, a command data segment, and a feedback data segment.

[0028] In a practical system, the monitoring status data group can be stored using the following data structure: The event header records the event identifier, event type, execution start time, and target temperature control path; the pre-event data sequence stores temperature status data and operating load data arranged by sampling time within the first time interval; the post-event data sequence stores temperature status data and operating load data arranged by sampling time within the second time interval; the instruction data segment stores the target execution status; and the feedback data segment stores the actual execution status. The controller reads the corresponding data from this monitoring status data group when subsequently generating pre-event temperature status estimation results, predicted thermal response, actual temperature response, and the three types of difference quantities.

[0029] like Figure 3 As shown, the monitoring time window is divided by the start time of the temperature control event. t0 to t1 corresponds to the first time interval, used to carry pre-event state data; t2 to t3 corresponds to the second time interval, used to carry post-event response data; and the interval between t1 and t2 is the transition period for event triggering, command issuance, and execution initiation. In the first time interval, the controller primarily reads pre-event temperature state data and operating load data. In the second time interval, it continues to read post-event temperature state data, operating load data, and feedback state data from the temperature control actuator, and writes both sets of data into the monitoring state data group. The diagram illustrates that the monitoring state data group is not a simple collection of ordinary continuously sampled data, but rather a data object segmented before and after a single temperature control event, time-aligned, and associated with the event.

[0030] In one specific implementation, multiple temperature control zones are pre-defined within the energy storage device. These zones can be set according to the layer where the battery modules are located, the coverage area of ​​the air duct or liquid cooling path, the height of the cabinet, the area adjacent to power devices, or the area with concentrated heat sources. Each temperature control zone is associated with one or more temperature measurement points; when a temperature control zone does not have a sensor directly placed there, an indirect association can be established with adjacent or representative measurement points. The controller extracts the temperature of the measurement points corresponding to each temperature control zone from the temperature status data within a first time interval, and combines this with the operating load data within the first time interval to generate estimated temperature values ​​and temperature distribution characteristics for each temperature control zone before the event execution.

[0031] Temperature estimates can be directly obtained from the end-sampling values, mean values, or weighted mean values ​​of the corresponding measuring points in the temperature-controlled area within the first time interval. Temperature distribution characteristics can include the highest temperature, lowest temperature, regional temperature difference, temperature gradient, inter-regional temperature dispersion, and temperature rise trend. To accommodate situations with a limited number of sensors or missing measuring points, the controller can use a reduced-order temperature field reconstruction model to generate pre-event temperature state estimates. Specifically, in the offline phase, the controller uses simulation data and historical operating data to establish a temperature field sample matrix and extract the dominant temperature field mode; in the online phase, the temperatures of representative measuring points are input into the model to obtain estimated temperatures for multiple temperature-controlled areas.

[0032] One possible reconstruction method is Gappy-POD temperature field reconstruction. The controller constructs a temperature vector from representative measurement points within the first time interval, uses the pre-extracted temperature field mode matrix as the reconstruction basis, obtains the mode coefficients through regularized least squares solution, and reconstructs the pre-event temperature field from the mode coefficients. This method is suitable for scenarios with missing measurement points, sparse sensor layout, and multi-region temperature field estimation.

[0033]

[0034] In the formula, This represents the temperature vector corresponding to the pre-event temperature state estimation result. This represents the historical average temperature. This represents the retained dominant temperature mode matrix, where 'a' represents the modal coefficients and 'S' represents the measurement point selection matrix. This represents the temperature vector at a representative measuring point. This represents the regularization coefficient. To improve the adaptability of the pre-event temperature state estimation results to rapid load changes, the controller can also superimpose a set of time-series residual correction networks after the Gappy-POD reconstruction results. This network takes temperature state data, operating load data, and temperature control actuator feedback state data from the most recent multiple sampling periods as inputs, and outputs the residual compensation amount for the reconstructed temperature field.

[0035] In one specific implementation, the controller determines the initial temperature state of each temperature-controlled zone before the execution of the temperature control event based on the pre-event temperature state estimation results. The initial temperature state may include the estimated temperature vectors, temperature distribution characteristics, and temperature change trends of multiple temperature-controlled zones. The controller obtains the heat input state corresponding to the temperature control event based on the operating load data. The heat input state is used to characterize the heating trend of the energy storage device within the monitoring time window. The controller also obtains the target adjustment state of the corresponding temperature control path based on the temperature control command state data. The target adjustment state may include the target air volume, target cooling intensity, target liquid cooling flow rate, target valve opening, or target heating power.

[0036] The controller inputs the initial temperature state, heat input state, and target adjustment state into the thermal response prediction model to obtain the predicted temperature response after the temperature control event is executed and the predicted heat exchange result of the temperature control path. The predicted temperature response can be represented as the predicted temperature change sequence of each temperature control region within the second time interval; the predicted heat exchange result of the temperature control path can be represented as the predicted heat absorption or predicted heat dissipation corresponding to the temperature control path.

[0037] The thermal response prediction model can employ a graph convolutional layer, gated loop layer, and dual-output head structure for engineering deployment. The graph convolutional layer uses temperature-controlled regions as nodes and spatial adjacency, heat conduction relationships, air duct coverage relationships, or liquid cooling branch connection relationships as edges to extract spatial thermal coupling features between temperature-controlled regions. The gated loop layer receives thermal coupling features and temperature-controlled event features within continuous sampling periods to characterize the thermal inertia and time lag after temperature control actions. The dual output heads output the predicted temperature response and the predicted heat exchange results of the temperature control path, respectively.

[0038] like Figure 4 As shown, the pre-event temperature state estimation and predicted thermal response model can form a dual-branch collaborative structure: the upper branch takes representative measurement point temperature and operating load data as input, first obtains the pre-event basic temperature field through Gappy-POD basic temperature field reconstruction, and then encodes the spatial thermal coupling relationship between temperature control regions through a graph convolutional layer; the lower branch takes the temporal characteristics of temperature state data, operating load data, and temperature control path relationship as input, compensates for the time response error of the basic temperature field through a TCN residual correction network, and learns the thermal inertia change after the temperature control event is executed by a gated recurrent layer. The outputs of the two branches are input to a dual-output head, which respectively provides the predicted temperature response and heat exchange prediction results. The figure illustrates the continuous calling relationship between temperature state estimation, spatial thermal coupling encoding, time response compensation, and dual-objective output.

[0039]

[0040] In the formula, Indicates the predicted temperature response. This indicates the predicted results of heat exchange in the temperature control path. This represents a thermal response prediction model. Let L represent the initial temperature state, L represent the feature vector composed of operating load data, U represent the event feature vector composed of temperature control command status data, and A represent the thermal coupling adjacency matrix between temperature control zones.

[0041] In one specific implementation, after a temperature control event is executed, the controller obtains the actual temperature response based on the relationship between the temperature state data within a second time interval and the temperature state estimation result before the event. The actual temperature response can be represented by a temperature change vector of the temperature-controlled area, the temperature change of the terminal area, the change in regional temperature difference, or the change in temperature distribution characteristics. For multiple temperature-controlled areas, the actual temperature response can be represented as a sequence of differences between the temperature of each temperature-controlled area and the temperature state estimation result before the event within the second time interval.

[0042] like Figure 6 As shown, the pre-event temperature field can be generated from the pre-event temperature state estimation results, and the post-event temperature field can be generated from the temperature state data after the event is executed. Each grid cell corresponds to a temperature control region or equivalent temperature region, and the value within the grid represents the temperature of that region. By comparing the two temperature fields, the controller can identify the cooling magnitude, residual hot spot location, and temperature distribution change direction of each region after the temperature control event is executed. If a region located at the end of the temperature control path or with concentrated heat sources still maintains a high temperature after the event, the temperature state change of that region will be included in the calculation of the actual temperature response and the cumulative heat change.

[0043] The heat exchange result of the temperature control path is used to represent the degree to which the temperature control path actually participates in heat exchange after the execution of a temperature control event. The controller first locates the temperature control region associated with the temperature control path corresponding to the temperature control event; then, based on the temperature state change of the associated temperature control region before and after the event, as well as the feedback status data from the temperature control actuator, it generates the heat exchange result of the temperature control path. For liquid cooling paths, the actual heat exchange volume can be obtained based on the inlet temperature, outlet temperature, flow rate of the cooling medium, and pump feedback status; for air-cooled paths, the actual heat exchange volume can be obtained based on the inlet air temperature, return air temperature, air volume, and fan feedback status; for air conditioning paths, the actual heat exchange volume can be obtained by combining the supply air temperature, return air temperature, and air conditioning power feedback.

[0044]

[0045] In the formula, This represents the actual heat exchange volume in the temperature control path, where M represents the number of medium paths involved in the heat exchange. This represents the density of the heat transfer medium in the k-th path. Indicates specific heat capacity. Indicates volumetric flow rate, and These represent the outlet and inlet temperatures of the path, respectively. This indicates the length of the statistical period. For temperature control paths that lack complete flow measurement capabilities, the equivalent flow rate can be obtained by mapping the fan speed, pump frequency, or valve opening to a historical calibration table.

[0046] In one specific implementation, the controller calculates the difference between the target execution state represented by the temperature control command status data and the actual execution state represented by the temperature control actuator feedback status data to obtain a command-execution consistency difference. The target execution state may include target speed, target frequency, target opening degree, or target power, while the actual execution state may include feedback speed, feedback frequency, feedback opening degree, feedback power, and actuator operating status position. This difference reflects whether the command has been effectively implemented by the actuator.

[0047]

[0048] In the formula, This indicates the amount of instruction-execution consistency difference. This represents the target execution state vector composed of temperature control command status data. This represents the actual execution state vector composed of feedback status data from the temperature control actuator. To prevent stable terms with a denominator of zero.

[0049] The controller calculates the difference between the predicted temperature response and the actual temperature response in the predicted thermal response, obtaining the predicted-measured thermal response difference. This difference reflects whether the temperature change after the event is executed matches the expected result given by the thermal response prediction model. When the actuator feedback is normal but the difference is large, it usually indicates that heat transfer in the temperature control path is blocked, local heat load is abnormal, or the temperature field estimation deviates from the actual operating conditions.

[0050]

[0051] In the formula, This represents the difference between the predicted and measured thermal response. Indicates the predicted temperature response. This indicates the actual temperature response.

[0052] like Figure 5 As shown, the predicted temperature response and the actual temperature response can be compared and displayed using curves on the time axis after the event. The predicted thermal response curve represents the expected cooling trend given by the thermal response prediction model; the actual thermal response-normal curve gradually converges around the predicted trend even with sampling fluctuations and short-term disturbances; the actual thermal response-abnormal curve exhibits characteristics of response lag, plateau, or continuous deviation. When the abnormal curve is higher than the predicted trend for a long time after the temperature control event is executed, the controller can form a large difference between the predicted and measured thermal response, and further determine whether there is insufficient heat exchange or insufficient temperature control capability by combining the heat exchange results of the temperature control path.

[0053] The controller calculates the heat input reference value corresponding to the temperature control event based on the operating load data, obtains the actual heat exchange volume of the temperature control path based on the heat exchange results of the temperature control path, and characterizes the cumulative heat change in the energy storage device by the change in temperature state data before and after the execution of the temperature control event. Subsequently, the controller generates a heat closure difference value based on the difference between the heat input reference value, the actual heat exchange volume of the temperature control path, and the cumulative heat change.

[0054]

[0055] In the formula, This represents the cumulative change in heat within the energy storage device, where N represents the number of temperature-controlled zones. This represents the equivalent heat capacity corresponding to the i-th temperature control zone. Indicates the temperature of the area after the event is executed. This indicates the temperature of the area before the event was executed.

[0056]

[0057] In the formula, Indicates the difference in thermal closure. This represents the heat input reference value calculated from the operating load data. This indicates the actual heat exchange volume of the temperature control path. This represents the cumulative change in heat. The heat closure difference is used to verify whether there is a match between the heat input, the heat carried away or introduced by the temperature control path, and the changes in the internal temperature field of the energy storage device.

[0058] In one specific implementation, the controller matches the instruction-execution consistency difference, the predicted-measured thermal response difference, and the thermal closure difference with their respective anomaly determination conditions. Anomaly determination conditions may include threshold conditions, duration conditions, trend conditions, or combinations thereof. The controller employs a hierarchical determination order when generating anomaly determination results: first, it judges the instruction-execution consistency difference to distinguish whether the actuator operates according to the instruction; if execution consistency is not abnormal, it then judges the predicted-measured thermal response difference to identify thermal response anomalies; when the thermal closure difference meets the anomaly determination conditions, it combines the predicted-measured thermal response difference to generate an anomaly determination result indicating thermal closure anomaly or insufficient temperature control capability.

[0059] For example, when the target frequency of a liquid cooling pump has increased but the feedback frequency remains unchanged, the controller generates an anomaly judgment result indicating abnormal execution response; when the feedback frequency of the liquid cooling pump is normal, but the actual temperature response of the target temperature control area is significantly weaker than the predicted temperature response, the controller generates an anomaly judgment result indicating abnormal thermal response; when the heat exchange result of the temperature control path shows insufficient heat in the path, and the temperature status data after the event continues to show heat accumulation, the controller generates a heat closure anomaly; when the actuator feedback has reached or is close to the adjustable boundary, and the difference between the predicted and measured thermal response and the difference in heat closure remain consistently high, the controller generates an anomaly judgment result indicating insufficient temperature control capability.

[0060] like Figure 7 As shown, the anomaly determination result generation process uses three types of difference quantities as inputs: instruction-execution consistency difference quantity corresponds to whether the actuator implements the temperature control instruction; prediction-measured thermal response difference quantity corresponds to whether the temperature change after the event matches the prediction; and thermal closure difference quantity corresponds to whether the heat input, path heat exchange, and heat accumulation change match. After the three types of difference quantities are matched with the anomaly determination conditions and the determination weight is corrected, anomaly determination results such as execution response anomaly, thermal response anomaly, thermal closure anomaly, or insufficient temperature control capability can be output. Figure 7 The multiple input and multiple output relationships in the invention indicate that the invention does not only provide temperature over-limit alarms, but also locates abnormal states to multiple aspects such as execution links, thermal response links, heat exchange links, or temperature control capabilities through combinations of differential quantities.

[0061]

[0062] In one specific implementation, to reduce the impact of model mismatch, abnormal measurement points, or missing data on the anomaly determination results, the controller further generates temperature state estimation confidence and thermal response prediction confidence. The temperature state estimation confidence can be obtained based on the deviation between the temperature state data within a first time interval and the temperature state estimation result before the event; the thermal response prediction confidence can be obtained based on the historical or current deviation between the predicted thermal response and the actual temperature response.

[0063] When the reliability of the temperature state estimation or the reliability of the thermal response prediction are low, the controller no longer uses the difference between the predicted and measured thermal response as the sole primary criterion. Instead, it reduces the weight of this difference in anomaly detection while increasing the weight of the difference between instruction-execution consistency and thermal closure. In this way, even when the thermal response prediction model is temporarily unreliable or the data quality at some measurement points is poor, the system can still make robust decisions based on execution feedback and thermal closure relationships.

[0064]

[0065] In the formula, This represents the corrected decision weight vector. This represents the original decision weight vector. This represents the confidence vector obtained by combining the confidence of temperature state estimation and the confidence of thermal response prediction. This indicates element-wise multiplication. The revised decision weights are used to correct the matching results for anomaly detection conditions.

[0066] In one specific implementation, the controller selects a target difference quantity that triggers the anomaly determination condition from among the command-execution consistency difference quantity, the predicted-measured thermal response difference quantity, and the thermal closure difference quantity, based on the anomaly type corresponding to the anomaly determination result. The target difference quantity is used to indicate the main source of this anomaly determination. The controller then traces the data fields associated with the target difference quantity. If the target difference quantity is a command-execution consistency difference quantity, it reads the associated temperature control command status data and temperature control actuator feedback status data; if the target difference quantity is a predicted-measured thermal response difference quantity, it reads the predicted thermal response, the actual temperature response, and the corresponding temperature control zone; if the target difference quantity is a thermal closure difference quantity, it reads the heat exchange result of the temperature control path, the operating load data, and the temperature status data change.

[0067] Status correction prompts can include the object to be checked and the direction of correction. The object to be checked can be the temperature control actuator, temperature control path, temperature measuring point, target temperature control area, or operating load input; the direction of correction can be checking the actuator feedback, adjusting the target regulation state for the next control cycle, switching the temperature control path, reducing the operating load, verifying measuring point data, or recalibrating the thermal response prediction model. Status correction prompts are output in conjunction with anomaly determination results and can be sent to the local monitoring interface, remote operation and maintenance platform, or the temperature controller of the energy storage device.

[0068] In one specific embodiment, the energy storage device temperature control event status monitoring system includes a temperature acquisition device, an operating status acquisition device, a temperature control status acquisition device, a memory, a controller, and an output interface. The temperature acquisition device can consist of multiple temperature sensors and acquisition circuits; the acquisition circuits convert the temperature sensor signals into temperature status data and write it into the controller's buffer. The operating status acquisition device can read the energy storage device's operating load data via CAN, RS485, Ethernet, or an internal bus. The temperature control status acquisition device can read the command status issued by the temperature controller and the feedback status of the actuators.

[0069] The memory stores the thermal response prediction model, temperature state estimation model, anomaly detection conditions, historical event data, and status monitoring program. The controller connects to each acquisition device, the memory, and the output interface. Upon receiving a temperature control event, the controller invokes the status monitoring program in the memory to execute the following steps: constructing the monitoring time window, estimating the temperature state before the event, generating the predicted thermal response, acquiring the actual temperature response, generating the heat exchange results of the temperature control path, calculating the three types of differences, and generating the anomaly detection result. The output interface outputs the anomaly detection result and status correction prompts to a local display screen, the energy storage system's energy management platform, a cloud-based operation and maintenance platform, or a host computer.

[0070] like Figure 2 As shown, the energy storage device temperature control event status monitoring system is centered on the controller. The temperature acquisition device provides the controller with temperature status data within the monitoring time window, the operation status acquisition device provides the controller with operating load data, the temperature control status acquisition device provides the controller with temperature control command status data and temperature control actuator feedback status data, the memory is used to provide the controller with thermal response prediction models, anomaly judgment conditions and status monitoring programs, and the output interface is used to output anomaly judgment results and status correction prompts.

[0071] In one specific implementation, the model training data consists of historical and simulation data of the energy storage device under different operating loads, different temperature control modes, and different temperature control paths. The training data is sliced ​​according to temperature control events, and each sample includes pre-event temperature state data, operating load data, temperature control command state data, actuator feedback state data, post-event temperature state data, and heat exchange results of the temperature control path. The controller or training server converts each sample into a unified format to form a sample set used to train the temperature state estimation model and the thermal response prediction model.

[0072] For the temperature state estimation model, the training phase first constructs a temperature field matrix from complete temperature field samples, extracts the dominant modes, and selects representative measurement points. In the online phase, the pre-event temperature state estimation results for multiple temperature-controlled zones are reconstructed using the temperature data from representative measurement points and operating load data. For the thermal response prediction model, the training phase uses the pre-event temperature state estimation results, operating load data, temperature control command state data, and temperature-controlled zone topology as inputs, and the actual temperature response and heat exchange results along the temperature control path after the event as output labels. After model training is complete, the model parameters are deployed to memory, and the controller loads the model for prediction during real-time operation.

[0073] To verify the feasibility of this method in engineering, several types of temperature control events can be constructed in both the energy storage cabinet simulation platform and the actual test platform. These events include normal temperature control events, actuator feedback anomaly events, insufficient response events in the air-cooled path, insufficient heat exchange events in the liquid-cooled path, events with missing local measuring points, and events with insufficient temperature control capability. The simulation platform can output the temperature field before and after the event, the medium state of the temperature control path, and the actuator state; the actual test platform can output sensor data and actuator feedback. The system compares the anomaly judgment results output by this method with manually labeled results or simulation settings to obtain the anomaly identification accuracy, early identification time, and false positive rate.

[0074]

[0075] In the above verification process, the predicted temperature response and the actual temperature response can be plotted as time trend curves, the temperature state estimation results before and after the event can be plotted as a temperature field distribution map, and the three types of difference quantities can be plotted as residual change curves over time. If all three types of difference quantities remain outside the corresponding anomaly judgment conditions in normal temperature control events, while at least one target difference quantity continuously triggers the anomaly judgment conditions in abnormal events, it indicates that this method can provide interpretable state monitoring of temperature control events.

[0076] In the engineering implementation of this invention, the thermal response prediction model, temperature state estimation model, and anomaly judgment conditions can be configured according to the equipment model, temperature control actuator type, and temperature control path structure. For residential stacked energy storage devices with a small number of sensors, representative measurement points and reduced-order reconstruction methods can be used to generate pre-event temperature state estimation results. For industrial and commercial energy storage cabinets, the temperature control path heat exchange results can be calculated by combining the state data of liquid-cooled branches, air-cooled channels, and air conditioning paths. For systems with cloud-based operation and maintenance capabilities, temperature control events, three types of differential quantities, and anomaly judgment results can be uploaded to a remote platform for centralized monitoring of the temperature control status of multiple energy storage devices.

[0077] In this specific implementation, the computational models can be deployed using offline training and online inference. During the offline training phase, model parameters are obtained using historical event samples and simulation samples. During the online inference phase, the controller only performs matrix operations, network forward computation, and decision condition matching; complex fluid simulations do not need to be repeated locally on the energy storage device. For controllers with limited storage resources or computing power, the pre-event temperature state estimation results and predicted thermal response generation can be deployed in the edge gateway. The controller only receives the corresponding results and completes the calculation of three types of differences and anomaly determination.

[0078] To enable continuous execution of model training, online inference, and variance calculation within the controller, the controller can perform uniform preprocessing on the raw acquired data before each event processing. The data alignment, quality labeling, and scale normalization processes described below are used to convert data from different acquisition sources into the same sampling benchmark and the same state record format, thereby providing a consistent input basis for pre-event temperature state estimation, predicted thermal response generation, and anomaly detection.

[0079] In one specific implementation, after forming a monitoring status data group, the controller first preprocesses the data entering the model and calculating the difference. Preprocessing does not change the operational meaning of the temperature control event, but rather ensures that data from different sources can enter the subsequent calculation process according to a unified sampling time. Temperature status data comes from the temperature acquisition device, operating load data comes from the energy storage device controller or power conversion unit, temperature control command status data comes from the temperature control controller, and temperature control actuator feedback status data comes from each actuator or its driver. Since the sampling frequency and communication delay of the above data may differ, the controller uses the start time of the temperature control event as a reference to perform timestamp correction on the data in the first and second time intervals.

[0080] Timestamp correction can be achieved using nearest neighbor alignment, linear interpolation, or preserving previous values. For example, if temperature status data is uploaded at a lower frequency, while the temperature control actuator's feedback status data is uploaded at a higher frequency, the controller can aggregate the higher-frequency data into the mean, maximum, or last value based on the temperature sampling time, thus forming a state vector at the same sampling time. For single-point missing data caused by short-term communication packet loss, the controller can use interpolation of adjacent sampling points to fill in the gaps; for continuously missing or abnormally quality-identified data, the controller does not directly use it for model inference, but marks it as low-confidence data in the monitoring status data group and reflects this low-confidence state when generating the temperature status estimate confidence level.

[0081] Temperature state data can undergo unit standardization and scale normalization before entering the pre-event temperature state estimation step. For example, temperature values ​​output by different types of temperature sensors are converted to the same temperature scale; the rotational speed, frequency, opening degree, or power in the actuator feedback state are converted to a unified normalized execution state range; charging power, discharging power, battery cluster current, and power conversion unit load in the operating load data are mapped to thermal input-related features. After preprocessing, a state record is formed at each sampling time. The state record includes at least the temperature vector of the temperature control area, the operating load feature vector, the temperature control command feature vector, the execution feedback feature vector, and the data quality identifier vector. This state record serves as the unified input for subsequent pre-event temperature state estimation results, predicted thermal response, and differential quantity calculations.

[0082]

[0083] In the formula, This represents the state record corresponding to sampling time t. This represents the temperature vector of the temperature control zone. Represents the characteristic vector of operating load. This represents the feature vector of the temperature control command. This represents the feedback feature vector of the actuator. This represents the data quality identifier vector. The data quality identifier vector is used to indicate whether the measurement point is missing, whether it exceeds the effective range of the sensor, whether there is communication delay, and whether it has been interpolated.

[0084] In one specific implementation, the spatial thermal coupling relationship of the thermal response prediction model is represented by the temperature control region topology. The temperature control region topology can be established based on the actual structure of the energy storage device: spatial adjacency edges are established between temperature control regions corresponding to adjacent surfaces of the same battery module; temperature control path edges are established between temperature control regions arranged along the air duct direction or liquid cooling branch direction; and local heat source coupling edges are established between the power conversion unit and its adjacent air region. The weights of the topology edges can be given based on the spatial distance between regions, the direction of medium flow, the type of heat transfer path, or the correlation with historical temperatures.

[0085] During the model training phase, the controller or training server converts the temperature-controlled area topology into an adjacency matrix. The adjacency matrix does not need to reflect all mechanical details, but rather represents whether there are significant thermal coupling relationships between areas and whether temperature control actions can affect one area from another. For energy storage devices without liquid cooling structures, the adjacency matrix can be primarily constructed based on the air channels within the cabinet and the adjacency relationships between areas; for liquid-cooled energy storage devices, the adjacency matrix can further include directional edges at the liquid cooling branch inlets, middle sections, and ends; for energy storage cabinets with air conditioning and dampers, the adjacency matrix can encode the interaction relationships between the air supply vents, return air vents, and each target temperature-controlled area as path edges.

[0086]

[0087] In the formula, This represents the thermal coupling weight between the i-th temperature control zone and the j-th temperature control zone. This indicates the spatial distance between two temperature-controlled zones. This indicates the temperature correlation between the two in historical data. This indicates whether the two are on the same temperature control path. , , This represents the weighting coefficients. This adjacency matrix serves as one of the inputs to the thermal response prediction model, enabling the model to identify response differences between the front, middle, and end regions.

[0088] The network structure of the thermal response prediction model can be deployed in a lightweight form within an edge controller. The input layer receives the pre-event temperature state estimation results, operating load data, temperature control command state data, and an adjacency matrix. The graph convolutional layer aggregates the thermal states of adjacent regions and regions along the same path based on the adjacency matrix. The gated recurrent layer reads the graph convolutional output within consecutive sampling periods to learn the thermal inertia changes after the temperature control action. Dual output heads output the predicted temperature response and the predicted heat exchange results along the temperature control path, respectively. For ease of engineering deployment, the model can be trained on a training server first, and then the weight file can be exported as a model file that the controller can load.

[0089]

[0090] In one specific implementation, the heat exchange result of the temperature control path can be calculated directly from the inlet and outlet states of the medium, or it can be obtained by mapping from a calibration table. For liquid cooling paths with temperature, flow rate, and medium parameter acquisition conditions, the controller directly calculates the actual heat exchange volume of the path based on the inlet temperature, outlet temperature, and flow rate of the cooling medium. For liquid cooling paths where only pump frequency or valve opening can be obtained, the controller reads a pre-calibrated pump frequency-flow rate relationship table or valve opening-branch flow rate relationship table to obtain the equivalent flow rate before calculating the heat exchange volume.

[0091] For the air-cooled path, if the energy storage device is equipped with a wind speed or air volume sensor, the controller calculates the actual heat exchange volume of the air-cooled path based on the inlet air temperature, return air temperature, and air volume. If there is no air volume sensor, the controller queries the equivalent air volume table based on the fan speed, damper opening, and duct mode to obtain the equivalent air volume corresponding to the air-cooled path. For the air-conditioning path, the controller estimates the cooling or heating contribution based on the supply air temperature, return air temperature, fan status, and air conditioning operating status. The above mapping table can be established during the equipment factory testing phase or gradually updated during operation using stable operating condition data.

[0092] During online operation, the controller does not need to perform complex simulations for each path. It only needs to select the corresponding heat exchange calculation method based on the temperature control path associated with the current temperature control event. For example, if the temperature control event corresponds to an increase in the liquid cooling pump frequency, the heat exchange result of the temperature control path will be calculated primarily from the liquid cooling path; if the temperature control event corresponds to an increase in fan speed or a change in damper opening, the heat exchange result of the temperature control path will be calculated primarily from the air cooling path; if the temperature control event corresponds to a change in air conditioning cooling power, the heat exchange result of the temperature control path will be calculated primarily from the air conditioning supply and return air paths.

[0093] In one specific implementation, anomaly detection conditions can be configured according to the category of difference. The difference between instruction and execution consistency corresponds to the execution deviation condition, the difference between prediction and measured thermal response corresponds to the thermal response deviation condition, and the difference between thermal closure and thermal balance corresponds to the thermal budget deviation condition. Each type of anomaly detection condition can include an amplitude threshold, a number of continuous samplings, a trend direction, and anomaly resolution conditions. This avoids misjudgments caused by a single instantaneous sampling point and allows for timely restoration to normal operation after anomaly elimination.

[0094] The anomaly detection criteria can also be dynamically updated based on the type of temperature control event. For example, if the temperature control event is an increase in fan speed, the difference between the predicted and measured thermal response mainly focuses on the cooling response of the target air-cooled area; if the temperature control event is an adjustment of the liquid-cooled valve opening, the difference in heat closure mainly focuses on the difference between the actual heat exchange in the liquid-cooled branch and the reference heat input; if the temperature control event is a heating action, the actual temperature response should be in the direction of heating rather than cooling. The controller calls the corresponding judgment condition template according to the event type, thereby ensuring that different temperature control events can use judgment logic consistent with their physical effects.

[0095] When the reliability of temperature state estimation or thermal response prediction is deemed low, the anomaly detection criteria can be adjusted appropriately. Specifically, the controller can increase the duration requirement for anomalies triggered by the difference between predicted and measured thermal responses, while simultaneously increasing the priority of determining differences in command-execution consistency and thermal closure. Once data quality recovers, the controller restores the original decision weights. This dynamic adjustment process is used to avoid generating too many false alarms when the model is temporarily unstable.

[0096] In one specific implementation, a simulation test environment corresponding to the structure of the target energy storage device can be established first. The simulation test environment includes the geometric model of the energy storage cabinet, the temperature control zone division results, the temperature control path topology, the operating load input, the temperature control command input, and the actuator response model. Before and after each temperature control event, the simulation platform outputs the temperature field, the temperature control path medium state, and the actuator state. The controller forms a monitoring status data set based on these outputs and performs anomaly detection.

[0097] Simulation verification can be configured with scenarios such as normal events, abnormal actuator feedback, obstructed temperature control path, increased local heat sources, missing measurement points, and insufficient temperature control capability. Normal events are used to verify the stability of the three types of differential quantities within normal ranges; abnormal actuator feedback is used to verify whether instruction-execution consistency differential quantities can be triggered preferentially; obstructed temperature control path is used to verify whether predicted-measured thermal response differential quantities and thermal closure differential quantities can be jointly identified; missing measurement points are used to verify whether the pre-event temperature state estimation results and the reliability of the temperature state estimation can reflect data integrity; insufficient temperature control capability is used to verify whether the anomaly judgment results can be accurately output when the actuator approaches the adjustable boundary.

[0098] The actual test platform can deploy temperature acquisition devices, operational status acquisition devices, and temperature control status acquisition devices on the energy storage cabinet prototype, and run a status monitoring program through an edge controller. During the test, the controller records the monitoring status data set, predicted thermal response, actual temperature response, heat exchange results of the temperature control path, and three types of differences for each temperature control event, and compares the anomaly judgment results with manual observation results or preset fault labels. If the execution response anomaly is triggered first in the scenario of abnormal actuator feedback, the thermal response anomaly or heat closure anomaly is triggered first in the scenario of blocked thermal path, and the temperature control capability is insufficient in the scenario of excessive overall load, then it indicates that the status monitoring method is engineering feasible.

[0099] In one specific implementation, after generating an anomaly determination result, the controller writes the key process data corresponding to the temperature control event into the event log. The event log may include an event identifier, a monitoring time window, an estimated temperature state before the event, a predicted thermal response, an actual temperature response, heat exchange results along the temperature control path, three types of differential quantities, the anomaly determination result, and a status correction prompt. The event log does not need to save all the original sensor data, but it should retain intermediate results that can reproduce the anomaly determination process. In this way, when maintenance personnel replay the event, they can see from the same event log whether the temperature control command was executed, whether the temperature change after execution matches the prediction, and whether the heat input and path heat exchange are closed.

[0100] Event logs can be stored locally in chronological order or uploaded to a remote operation and maintenance platform when the energy storage device is connected to the network. For normal events, the system can save only summary information, such as the event type, statistical values ​​of the three types of differences, and the final judgment result. For abnormal events, the system can save more complete data segments, including temperature state sequences, operating load sequences, and execution feedback sequences within the first and second time intervals. This hierarchical storage method can reduce storage pressure while ensuring that abnormal events are available for subsequent analysis.

[0101] After a certain operating cycle, the controller or remote operation and maintenance platform can update the thermal response prediction model and anomaly judgment conditions based on event records. Specifically, temperature control events that are manually confirmed or confirmed through subsequent operation and maintenance are used as labeled samples; for normal events with large model prediction deviations, the residual correction part in the thermal response prediction model is updated; for repeated fluctuations in differential quantities that do not cause anomalies, the duration conditions of the corresponding anomaly judgment conditions are adjusted; for combinations of differential quantities that are repeatedly confirmed as anomalies, the judgment priority of the anomaly judgment result corresponding to that combination is increased. The updated parameters are verified before being sent to the local controller of the energy storage device.

[0102] When multiple energy storage devices use the same model or similar temperature control structure, the remote operation and maintenance platform can aggregate and analyze the event records of similar devices. This aggregation analysis does not change the local anomaly detection process for a single device; instead, it generates general initial model parameters, recommended results for representative measurement points, and anomaly detection condition templates. Newly installed energy storage devices can use this template as their initial configuration and then gradually correct it using local operating data to adapt to the specific installation environment and the state of the temperature control actuator.

[0103] In actual deployments, event logs can also be used to generate operation and maintenance reports. These reports are categorized by anomaly assessment results, displaying the frequency, duration, and associated objects of execution response anomalies, thermal response anomalies, thermal closure anomalies, and insufficient temperature control capabilities. If an actuator frequently triggers anomalies in instruction-execution consistency across different temperature control events, the operation and maintenance report prioritizes that actuator for investigation. If a temperature control path frequently triggers anomalies in thermal closure differences, the report identifies that path as a target for heat exchange capability verification. If the predicted-measured thermal response difference is consistently large while the thermal closure difference remains normal, the report prompts verification of the thermal response prediction model or measurement point mapping relationship.

[0104] During the above operation, if the same temperature control event involves multiple temperature control paths simultaneously, the controller can generate the heat exchange results corresponding to each temperature control path separately, and merge the heat exchange quantities of each path when generating the heat closure difference. If only a single path is involved, the actual heat exchange quantity of that path is directly used in the calculation. Thus, energy storage devices with different structures can all monitor event status according to a unified data organization method.

[0105] It should be noted that the technical features of the above-described embodiments can be combined with each other without contradicting each other. Those skilled in the art can make adaptive adjustments to the step sequence, execution subject, or implementation method of each embodiment according to actual application needs. Such adjustments should not be regarded as departing from the essence of the present invention.

[0106] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, or combinations made to the above embodiments within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for monitoring the status of temperature control events in an energy storage device, characterized in that, include: In response to a temperature control event of an energy storage device, a corresponding monitoring time window is determined, and temperature status data, operating load data, temperature control command status data, and temperature control actuator feedback status data within the monitoring time window are acquired. Based on the temperature status data before the execution of the temperature control event in the monitoring time window, generate the pre-event temperature status estimation result, and generate the predicted thermal response based on the pre-event temperature status estimation result, operating load data, and temperature control command status data; After the temperature control event is executed, the actual temperature response and the heat exchange result of the temperature control path are obtained based on the temperature status data after the event is executed. The command-execution consistency difference is generated based on the temperature control command status data and the temperature control actuator feedback status data. The predicted-measured thermal response difference is generated based on the predicted thermal response and the actual temperature response. The heat closure difference is generated based on the operating load data, the heat exchange results of the temperature control path, and the change in temperature status data before and after the execution of the temperature control event. Based on the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference, an anomaly determination result corresponding to the temperature control event is generated.

2. The method for monitoring the temperature control event status of an energy storage device according to claim 1, characterized in that, The determination of the corresponding monitoring time window includes: using the execution start time of the temperature control event as the time reference, determining a first time interval before the execution start time and a second time interval after the execution start time; It also includes: acquiring temperature status data and operating load data within the first time interval and the second time interval, acquiring temperature control command status data corresponding to the temperature control event, and acquiring temperature control actuator feedback status data within the second time interval; associating the acquired data with the temperature control event according to the sampling time to form a monitoring status data group corresponding to the temperature control event.

3. The method for monitoring the temperature control event status of an energy storage device according to claim 2, characterized in that, The steps for generating the pre-event temperature state estimation result include: extracting the temperature of the measuring point corresponding to each temperature control area from the temperature state data within the first time interval based on the correspondence between the temperature control area and the temperature measuring point within the energy storage device; determining the estimated temperature value and temperature distribution characteristics of each temperature control area before the execution of the temperature control event based on the measuring point temperature and the operating load data within the first time interval, and using them as the pre-event temperature state estimation result.

4. The method for monitoring the temperature control event status of an energy storage device according to claim 3, characterized in that, The steps for generating the predicted thermal response include: determining the initial temperature state of each temperature control zone before the execution of the temperature control event based on the pre-event temperature state estimation results; determining the heat input state corresponding to the temperature control event based on the operating load data; determining the target adjustment state of the corresponding temperature control path based on the temperature control command state data; inputting the initial temperature state, heat input state, and target adjustment state into the thermal response prediction model to obtain the predicted temperature response and the predicted heat exchange result of the temperature control path after the execution of the temperature control event, and using the predicted temperature response and the predicted heat exchange result of the temperature control path as the predicted thermal response.

5. The method for monitoring the temperature control event status of an energy storage device according to claim 4, characterized in that, The steps for obtaining the actual temperature response and the heat exchange result of the temperature control path include: determining the actual temperature response after the execution of the temperature control event based on the relationship between the temperature state data in the second time interval and the estimated temperature state before the event; determining the temperature control area associated with the temperature control path based on the temperature control path corresponding to the temperature control event; and generating the heat exchange result of the temperature control path based on the change in temperature state of the associated temperature control area before and after the execution of the temperature control event and the feedback status data of the temperature control actuator.

6. The method for monitoring the temperature control event status of an energy storage device according to claim 5, characterized in that, The steps for generating the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference include: calculating the difference between the target execution state represented by the temperature control instruction state data and the actual execution state represented by the temperature control actuator feedback state data to obtain the instruction-execution consistency difference; calculating the difference between the predicted temperature response and the actual temperature response in the predicted thermal response to obtain the prediction-measured thermal response difference; converting the heat input reference quantity corresponding to the temperature control event based on the operating load data, obtaining the actual heat exchange quantity of the temperature control path based on the heat exchange result of the temperature control path, and characterizing the cumulative heat change in the energy storage device by the change in temperature state data before and after the execution of the temperature control event; and generating the thermal closure difference based on the difference between the heat input reference quantity, the actual heat exchange quantity of the temperature control path, and the cumulative heat change.

7. The method for monitoring the temperature control event status of an energy storage device according to claim 6, characterized in that, The steps for generating the anomaly determination result include: matching the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference with their respective anomaly determination conditions; generating an anomaly determination result for execution response anomaly when the instruction-execution consistency difference meets the corresponding anomaly determination condition; generating an anomaly determination result for thermal response anomaly when the instruction-execution consistency difference does not meet the corresponding anomaly determination condition, but the prediction-measured thermal response difference meets the corresponding anomaly determination condition; and generating an anomaly determination result for thermal closure anomaly or insufficient temperature control capability based on whether the prediction-measured thermal response difference meets the corresponding anomaly determination condition when the thermal closure difference meets the corresponding anomaly determination condition.

8. The method for monitoring the temperature control event status of an energy storage device according to claim 7, characterized in that, The steps for generating the anomaly determination result further include: generating a temperature state estimation confidence level based on the deviation between the temperature state data within the first time interval and the pre-event temperature state estimation result; generating a thermal response prediction confidence level based on the deviation between the predicted thermal response and the actual temperature response; configuring judgment weights for the instruction-execution consistency difference, the prediction-measured thermal response difference, and the thermal closure difference based on the temperature state estimation confidence level and the thermal response prediction confidence level; and correcting the matching result of the anomaly determination condition according to the configured judgment weights to generate the anomaly determination result.

9. The method for monitoring the temperature control event status of an energy storage device according to claim 8, characterized in that, Also includes: Based on the anomaly type corresponding to the anomaly determination result, a target difference quantity that triggers the anomaly determination condition is selected from the instruction-execution consistency difference quantity, the prediction-measured thermal response difference quantity, and the thermal closure difference quantity; based on the temperature control instruction status data, temperature control actuator feedback status data, actual temperature response, or temperature control path heat exchange result associated with the target difference quantity, a status correction prompt is generated; the status correction prompt includes the object to be checked and the correction direction, and is output in association with the anomaly determination result.

10. A temperature control event status monitoring system for an energy storage device, characterized in that, include: Temperature acquisition device, operating status acquisition device, temperature control status acquisition device, memory, controller, and output interface; The temperature acquisition device is used to acquire temperature status data of the energy storage device within the monitoring time window. The operation status acquisition device is used to acquire the operation load data within the monitoring time window; The temperature control status acquisition device is used to acquire temperature control command status data and temperature control actuator feedback status data; The memory is used to store the thermal response prediction model and the condition monitoring program; The controller is connected to the temperature acquisition device, the operating status acquisition device, the temperature control status acquisition device, the memory, and the output interface, and is configured to perform the following processes: in response to a temperature control event of the energy storage device, determine the corresponding monitoring time window; generate a pre-event temperature status estimation result based on the temperature status data before the temperature control event is executed, and generate a predicted thermal response by combining the operating load data and the temperature control command status data; after the temperature control event is executed, obtain the actual temperature response and the heat exchange result of the temperature control path based on the temperature status data after the event is executed. Generate instruction-execution consistency difference, prediction-measured thermal response difference, and thermal closure difference, and generate anomaly determination results based on the instruction-execution consistency difference, prediction-measured thermal response difference, and thermal closure difference; The output interface is used to output the anomaly determination result.

Images