A smart thermal management method and system for energy storage cabinets combining moisture and heat coupling control

Abstract

This invention discloses an intelligent thermal management method and system for energy storage cabinets that combines moisture-heat coupling control. It addresses the problems of delayed response, neglect of moisture-heat coupling safety hazards, high energy consumption, and poor system robustness in energy storage cabinet heat dissipation. By real-time acquisition of the measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current, the future temperature rise of the battery pack is predicted. Furthermore, a moisture-heat coupling algorithm is used for proactive intervention in heat dissipation. This invention provides an intelligent thermal management method and system for energy storage cabinets that combines moisture-heat coupling control. Through multimodal perception and edge AI computing, it achieves moisture-heat coupling control for energy storage cabinet thermal management, ensuring the insulation safety of the battery pack in humid and hot environments, eliminating the risk of short circuits caused by battery pack condensation, effectively solving the thermal safety problem of energy storage systems in complex humid and hot environments, and balancing efficient heat dissipation, anti-condensation safety, and system reliability. It has significant engineering application value.

Description

Technical Field

[0001] This invention discloses a thermal management method and system for energy storage cabinets that combines moisture and heat coupling control, belonging to the field of thermal management technology for energy storage cabinets. Background Technology

[0002] The thermal management system within the energy storage cabinet controls the operating temperature of the battery pack. Existing thermal management systems for energy storage cabinets include natural cooling and passive forced cooling. Natural cooling refers to installing air inlet and outlet channels on the energy storage cabinet, utilizing natural air convection to dissipate heat from the battery pack. However, natural cooling alone has low efficiency and is greatly affected by environmental conditions, failing to meet the rapid heat dissipation requirements within the energy storage cabinet during summer or under high grid loads. Passive forced cooling typically uses air cooling or liquid cooling, employing simple temperature sensor threshold control; that is, the cooling mode is activated once the set temperature is reached. This method exhibits significant lag; when cooling is initiated, the battery pack is often already overheated.

[0003] Chinese patent application CN119806240A discloses a CNN_LSTM-based temperature control method and system for energy storage devices. The method involves using a temperature sensor to collect the load temperature of the energy storage device, which is then fed into a trained CNN_LSTM model for temperature prediction. The error between the predicted temperature and the target temperature of the energy storage device is used as feedforward input data for a PID control unit. The PID control unit then uses this error feedback to adjust the load temperature of the energy storage device's temperature control system. This patented technology utilizes an artificial intelligence model to predict the future battery pack temperature and proactively adjusts the thermal management system of the energy storage cabinet to regulate the temperature inside the cabinet. This addresses the lag inherent in passive forced cooling. However, existing energy storage cabinet thermal management control logic typically only focuses on temperature, neglecting the influence of humidity. In high-temperature and high-humidity environments, rapid cooling can easily cause condensation on the battery pack surface due to temperatures falling below the dew point, posing a short-circuit risk. Furthermore, starting cooling after the battery has overheated significantly reduces its lifespan.

[0004] In summary, the existing energy storage cabinet thermal management system has the following problems:

[0005] (1) Response lag: Traditional threshold control is passive and cannot cope with sudden high power loads, which can easily lead to instantaneous thermal shock.

[0006] (2) Ignoring the safety hazards of humidity and heat coupling: The lack of linkage control of environmental humidity makes it easy to generate condensation under extreme weather conditions.

[0007] (3) High energy consumption: uneven airflow organization and lack of fine adjustment (usually only on / off or a few simple speed settings) result in energy waste.

[0008] (4) Poor system robustness: It relies on a single sensor, and the thermal management system will fail once the sensor fails. Summary of the Invention

[0009] The technical problem solved by this invention is to provide an intelligent thermal management method and system for energy storage cabinets that combines moisture and heat coupling control, addressing the aforementioned technical problems existing in the thermal management methods of battery packs in existing energy storage cabinets.

[0010] This invention is achieved using the following technical solution:

[0011] This invention first discloses an intelligent thermal management method for energy storage cabinets that combines moisture and heat coupling control.

[0012] Real-time data collection of the measured temperature of the battery pack in the energy storage cabinet, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current are used to predict the future temperature rise of the battery pack.

[0013] Thermal management of the energy storage cabinet is carried out through the following methods;

[0014] When the future temperature rise of the battery pack does not exceed the first threshold, the energy storage cabinet will use natural convection cooling mode.

[0015] When the future temperature rise of the battery pack is between the first threshold and the second threshold, the energy storage cabinet adopts the natural convection heat dissipation mode. The wind speed of natural convection is controlled by PID feedback, and a humidity correction factor based on the ambient humidity of the energy storage cabinet is introduced to dynamically adjust the proportional gain of the PID controller.

[0016] When the battery pack temperature rises to the second threshold temperature, the energy storage cabinet switches to forced cooling mode, activating the strong cooling flow channel for the battery pack. In forced cooling mode...

[0017] When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, the controlled cooling channel is set to the first controlled cooling channel value under the condensation safety limit.

[0018] When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, the control value of the strong cooling channel is the second strong cooling channel control value under the full-speed cooling requirement.

[0019] In a smart thermal management method for energy storage cabinets combining moisture and heat coupling control according to the present invention, the measured temperature of the battery pack is further collected using a two-stage sensor. The first-stage sensor serves as the main channel sensor to collect the measured temperature of the battery pack in real time, while the second-stage sensor serves as a backup channel sensor. When the temperature data collected by the main channel sensor is abnormal, the data is corrected through the following fault-tolerance mechanism:

[0020] The first-level fault-tolerance mechanism switches to the backup channel sensor to collect the real-time temperature of the battery pack. If the temperature data collected by the backup channel sensor is within the fluctuation range of the actual measured temperature of the battery pack before the main channel sensor malfunctions, the backup channel sensor is directly used to replace the main channel sensor, and the corresponding main channel sensor is marked as faulty. If the temperature data collected by the backup channel sensor is abnormal, the second-level fault-tolerance mechanism is activated. Combining the actual measured temperature data of adjacent battery packs, a virtual sensor is constructed using Bayesian probability to predict the actual measured temperature of the battery pack, and the corresponding backup channel sensor is marked as faulty.

[0021] In a smart thermal management method for energy storage cabinets that combines humidity and heat coupling control according to the present invention, an LSTM time-series prediction model is further used to predict the future temperature rise of the battery pack. The measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current are used to construct an input vector. The input vector is then subjected to moving average filtering and feature normalization to form the tensor input of the LSTM time-series prediction model. The LSTM time-series prediction model includes an input layer, two stacked LSTM hidden layers, and a fully connected output layer. The fully connected output layer maps the final hidden state of the tensor input of the LSTM hidden layer to the predicted future temperature rise of the battery pack at future times.

[0022] In a smart thermal management method for energy storage cabinets that combines damp-heat coupling control according to the present invention, further, after the forced cooling mode ends, the parameters of the LSTM time-series prediction model are fine-tuned using the predicted future temperature rise of the battery pack at the same time as the measured temperature of the battery pack.

[0023] In a smart thermal management method for energy storage cabinets combining humidity and heat coupling control according to the present invention, the humidity correction factor is further expressed as a dynamic function of humidity in the PID control equation:

[0024] ,

[0025] in, The proportional gain of the PID control equation is dynamically adjusted based on the humidity correction factor. The ambient humidity of the energy storage cabinet is collected at time k. Humidity threshold Based on the proportional gain, Humidity gain coefficient This is the activation function.

[0026] In a smart thermal management method for energy storage cabinets combining humidity and heat coupling control according to the present invention, further, in the forced cooling mode, the first strong cooling channel control quantity or the second strong cooling channel control quantity is selected by minimum value. When the difference between the measured temperature of the battery pack and the dew point temperature under the current humidity of the energy storage cabinet is less than a safety threshold, the first strong cooling channel control quantity is set to a low limit value. When the difference between the measured temperature of the battery pack and the dew point temperature under the current humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current humidity of the energy storage cabinet is greater than the high humidity threshold, the first strong cooling channel control quantity is set to a medium limit value. When the difference between the measured temperature of the battery pack and the dew point temperature under the current humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current humidity of the energy storage cabinet is less than or equal to the high humidity threshold, the first strong cooling channel control quantity is set to a maximum limit value. When the first strong cooling channel control quantity is the low limit value and the medium limit value, it is less than the second strong cooling channel control quantity. When the first strong cooling channel control quantity is the maximum limit value, it is greater than the second strong cooling channel control quantity.

[0027] In a smart thermal management method for energy storage cabinets that combines humidity and heat coupling control according to the present invention, the control quantity of the first strong cooling channel is further output by the measured temperature of the battery pack and the ambient humidity of the energy storage cabinet through a fuzzy logic control algorithm:

[0028] ,

[0029] in, For fuzzy inference functions, Actual temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet The difference between them The ambient humidity of the energy storage cabinet.

[0030] In a smart thermal management method for energy storage cabinets that combines humidity and heat coupling control according to the present invention, the dew point temperature is further calculated in real time using the Magnus-Tetens empirical formula to determine the ambient humidity of the energy storage cabinet.

[0031] In a smart thermal management method for energy storage cabinets that combines damp heat coupling control according to the present invention, the forced cooling mode further includes maintaining the current heat dissipation mode for 20-40 seconds after the measured temperature of the battery pack rises or falls to a first threshold between the measured temperature of the battery pack and the ambient temperature of the energy storage cabinet, and then exiting.

[0032] This invention also discloses an intelligent thermal management system for energy storage cabinets that combines moisture and heat coupling control, comprising:

[0033] The data acquisition module collects real-time data on the measured temperature of the battery pack in the energy storage cabinet, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current.

[0034] The calculation and control module implements the intelligent thermal management method for energy storage cabinets described above in this invention to perform thermal management on the energy storage cabinets.

[0035] The natural convection channel connects the inside and outside of the energy storage cabinet. It is equipped with a natural convection fan, and the wind speed of natural convection entering and exiting the energy storage cabinet is controlled by the computing control module.

[0036] The forced cooling channel is a liquid-cooled or air-cooled channel that connects to the battery pack inside the energy storage cabinet. The forced cooling channel control quantity is controlled by the computing control module to control the cooling medium driving component in the corresponding channel.

[0037] The active thermal management method provided by this invention is not limited to a specific hardware form. It collects data through multimodal sensors, uses an LSTM time-series model to predict future temperature rise trends, and combines a humidity-heat coupling algorithm for proactive intervention. Specifically, it has the following beneficial effects:

[0038] (1) The measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current data of the energy storage cabinet are collected as historical data to train the LSTM time-series prediction model. The LSTM time-series prediction model predicts the battery temperature rise trend within a preset time period in the future, evolving from "passive response" to "active defense", avoiding the lag adjustment of traditional thermal management schemes that rely on a single temperature threshold for regulation. By using the LSTM time-series prediction model to predict the future temperature rise of the battery pack, intervention is made in advance before the thermal shock reaches the battery surface, eliminating thermal lag and reducing the lifespan degradation caused by battery overheating. After the strong cooling event, the parameters of the LSTM model are fine-tuned online using the measured data, evolving from "fixed parameters" to "lifelong learning". By comparing the error (Loss) between the predicted future temperature rise of the battery pack and the actual temperature rise, the backpropagation algorithm is used to fine-tune the weights of the LSTM network in real time, so that it can adapt to the changes in the entire battery life cycle and seasonal environmental drift, realizing the continuous evolution of the control strategy.

[0039] (2) Based on the real-time collected intake air temperature and relative humidity, the dew point temperature of the current environment is calculated in real time using the Magnus correction model to establish a dynamic anti-condensation safety boundary.

[0040] (3) The controller automatically switches between different working modes based on the measured temperature of the battery pack, the predicted temperature rise trend, and the ambient humidity. A humidity-heat coupling control mechanism is introduced, and a humidity correction factor is introduced into the PID algorithm to dynamically adjust the proportional gain of the PID controller. As the humidity increases, the air volume is automatically increased to compensate for the fluctuation of heat exchange efficiency in humid air, thus overcoming the heat exchange efficiency decay in humid and hot environments and improving the robustness of the system under complex climates. In the forced cooling mode, the strategy of implementing the minimum value optimization to limit condensation safety and meet the full-speed cooling requirement changes the traditional practice of directly shutting down in high humidity environments. In the case of high humidity but no condensation in the energy storage cabinet, by limiting the forced cooling operation, both battery overheating and instantaneous condensation caused by aggressive cooling are avoided, achieving the best balance between safety and performance. It successfully solves the natural contradiction between cooling and anti-condensation in the "high temperature and high humidity" working conditions, ensuring maximum heat dissipation performance while physically eliminating the risk of condensation.

[0041] (4) When sensor data is missing, this invention employs a dual fault-tolerance mechanism of "physical redundancy + algorithm inference". It adopts a two-level sensor physical redundancy design with a main channel and a backup channel. In the fault handling logic, the system does not simply report an error and shut down, but prioritizes switching the backup channel sensor. When all two-level physical channel sensors are unavailable, Bayesian probabilistic inference is further initiated to generate virtual battery pack temperature readings. This dual fault-tolerance mechanism of "physical + algorithm" ensures that the thermal management system can still maintain closed-loop operation when the sensor fails. It greatly improves the robustness of the system by addressing the pain point of sensor failure in harsh industrial scenarios.

[0042] In summary, the present invention provides an intelligent thermal management method and system for energy storage cabinets that combines humid-thermal coupling control. By using multimodal perception and edge AI computing, the method achieves humid-thermal coupling control of energy storage cabinet thermal management, ensuring the insulation safety of the battery pack in humid and hot environments, eliminating the short-circuit risk caused by condensation in the battery pack, effectively solving the thermal safety problem of energy storage systems in complex humid and hot environments, and taking into account efficient heat dissipation, anti-condensation safety, and system reliability, thus having significant engineering application value.

[0043] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating an intelligent thermal management method for an energy storage cabinet that combines moisture and heat coupling control, as shown in the embodiment.

[0045] Figure 2 This is a control flowchart for the forced cooling mode in the embodiment.

[0046] Figure 3 This is a schematic diagram of an intelligent thermal management system for an energy storage cabinet that incorporates moisture and heat coupling control, as shown in the embodiment. Detailed Implementation

[0047] Example

[0048] See Figure 1 The intelligent thermal management method for the energy storage cabinet illustrated is a specific implementation of the present invention. The specific process is as follows: real-time acquisition of multimodal data of the energy storage cabinet; after preprocessing the multimodal data, LSTM is used to predict the future temperature rise of the battery pack; and then, thermal management in different modes is determined by combining the measured temperature of the battery pack and the ambient humidity of the energy storage cabinet.

[0049] Specifically, the multimodal data collected by the energy storage cabinet includes the measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current. In this embodiment, an LSTM time-series prediction model is used to predict the future temperature rise of the battery pack. The above multimodal data in the energy storage cabinet is collected every 5 seconds, and the LSTM model is used to predict the temperature rise in the next minute. Trend. The measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current are used to construct an input vector. The input vector is then filtered by moving average and normalized to form the tensor input of the LSTM time series prediction model. The LSTM time series prediction model includes an input layer, two stacked LSTM hidden layers, and a fully connected output layer. The fully connected output layer maps the final hidden state of the tensor input of the LSTM hidden layers to the predicted future temperature rise of the battery pack at future times.

[0050] The specific process by which the LSTM time-series prediction model predicts the future temperature rise of the battery pack using multimodal data from the energy storage cabinet is as follows:

[0051] Set sampling period T s =5s. At any discrete time point k, the system synchronously acquires environmental and battery status data through a sensor array to construct the original state vector of multimodal data. :

[0052] .

[0053] in, The measured temperature of the battery pack surface at time k represents the temperature of a battery pack with 1 to N layers, where N is the number of battery layers. Let K be the ambient temperature of the energy storage cabinet at time k. Let be the ambient humidity of the energy storage cabinet at time k, using the relative humidity inside the energy storage cabinet. Let k be the real-time load current of the energy storage cabinet. Calculate the maximum measured temperature rise of the battery pack at this time. , This ultimately forms the input vector for the algorithm. .

[0054] To eliminate high-frequency electromagnetic interference and airflow disturbance noise from the sensor, a moving average filter is applied to the input vector. The moving window length is set to W=3, and the filtered state values ​​are... for:

[0055] .

[0056] This represents the state value from the previous time step. To eliminate the influence of different physical dimensions of multimodal data on the convergence of the LSTM time series prediction model weights, a Min-Max normalization method is used to map all features to the [0,1] interval:

[0057] .

[0058] Normalized sequence data L represents the historical time step, such as the past 60 points, which will be used as the tensor input of the LSTM time series prediction model.

[0059] This embodiment utilizes an LSTM timing prediction model to predict the future temperature rise of the battery pack inside the energy storage cabinet based on the nonlinear hysteresis relationship between load current and temperature changes. The LSTM timing prediction model consists of an input layer, two stacked LSTM hidden layers, and a fully connected dense output layer.

[0060] In the memory cell operations of the LSTM time-series prediction model, for each time step t, the input x... t ′, Forgotten Gate f t Input gate i t and output gate o t The update formula is as follows:

[0061] ,

[0062] ,

[0063] ,

[0064] ,

[0065] ,

[0066] .

[0067] in, It is the Sigmoid activation function. Let be the forget gate activation value at time step t. It is the Sigmoid activation function. Here is the weight matrix for the forget gate. Let be the concatenated vector of the hidden state at time step (t-1) and the preprocessed input at time step t. This represents the hidden state at time step t-1. The preprocessed input data at time step t For the bias term of the forget gate, Let be the input gate activation value at time step t. Here is the weight matrix of the input gate. For the bias term of the input gate, Candidate cell states at time step t The hyperbolic tangent activation function is used. This is the weight matrix for the candidate cell states. This is a bias term for the candidate cell state. Let t be the cell state at time step t. Cell state at time step t-1 The output gate activation value at time step t. The weight matrix of the output gate, The bias term of the output gate. The hidden state at time step t. Hyperbolic tangent transform of cell state.

[0068] The above process uses the fully connected layer of the LSTM time series prediction model to process the input x. t The final hidden state is mapped to the future. Time (e.g.) The predicted temperature rise (=60s) is used to output the future temperature rise of the battery pack in the energy storage cabinet. :

[0069] .

[0070] in, Fully connected layer weight matrix, Let k be the LSTM hidden state at time k. Fully connected layer bias terms, At the current discrete sampling time, Predicted time step. Predicted future temperature rise of the battery pack. This reflects the internal heat that has not yet been conducted to the surface of the battery pack due to the impact of the large current.

[0071] In this embodiment, the measured temperature of the battery pack is collected using a two-stage sensor system. The primary sensor serves as the main channel sensor, collecting the measured temperature of the battery pack in the energy storage cabinet in real time, while the secondary sensor serves as a backup channel sensor. A multi-point redundant sensor layout is adopted, with two environmental sensors and two sets of heat source sensors arranged on each layer of the energy storage cabinet. The environmental sensors, consisting of temperature and humidity sensors, are placed at the air inlet and outlet of each layer of the energy storage cabinet to collect the ambient temperature and humidity in real time. The two sets of heat source sensors include a primary channel sensor and a backup channel sensor installed on the surface of the battery pack on each layer of the energy storage cabinet. The primary channel sensor preferably uses K-type thermocouples deployed at the center and edge of the battery pack, while the backup channel sensor preferably uses a non-contact infrared (IR) sensor. The dual-redundant sensors are used to collect the measured temperature of the battery pack, forming a comprehensive sensing network within the energy storage cabinet.

[0072] Under normal conditions, the primary sensor is used as the main channel sensor to collect the measured temperature of the battery pack. When the temperature data collected by the primary channel sensor is abnormal, the following fault tolerance mechanism is used to correct the data: The first-level fault tolerance mechanism switches to the backup channel sensor to collect the real-time temperature of the battery pack. If the temperature data collected by the backup channel sensor is within the fluctuation range of the measured temperature of the battery pack before the primary channel sensor malfunctions, the backup channel sensor is used directly to replace the primary channel sensor, and the corresponding primary channel sensor is marked as faulty. If the temperature data collected by the backup channel sensor is abnormal, the second-level fault tolerance mechanism is activated. Combining the measured temperature data of adjacent battery packs, a virtual sensor is constructed using Bayesian probability to predict the measured temperature of the battery pack, and the corresponding backup channel sensor is marked as faulty.

[0073] This embodiment employs a dual fault-tolerance mechanism of "physical redundancy + algorithm inference". When the controller detects the main channel sensor data x of the i-th layer of the energy storage cabinet... i In the event of loss or anomalies such as open circuit, short circuit, or numerical jumps exceeding physical limits, the system performs self-healing according to the following sub-steps:

[0074] Level 1 fault tolerance: Physical channel switching with infrared intervention. Immediately activate the corresponding backup channel sensor for this layer—a non-contact infrared (IR) sensor—to read the battery pack surface temperature x. i_IR If x i_IR If the data is valid and within a reasonable margin of error, then use x directly. i_IR Replace x i It participates in subsequent control and marks the main sensor of this layer as faulty. This step utilizes the non-contact and fast-response characteristics of infrared sensors as the first line of physical defense.

[0075] Level 2 Fault Tolerance: Bayesian Probabilistic Inference. If the primary channel sensor fails and backup channel sensor data is also unavailable, or to further calibrate the accuracy of backup channel sensor data, the following Bayesian probabilistic inference process is initiated.

[0076] Prior probability construction: based on the current load current of the energy storage cabinet The theoretical temperature rise probability distribution under this current can be obtained by looking up the table. .

[0077] Likelihood function construction: using temperature data x from adjacent layers (i−1,i+1) i-1 x i+1 Construct the spatial correlation likelihood function:

[0078] .

[0079] in, Let Variance be the spatial variation of temperature. Posterior estimation outputs virtual readings:

[0080] .

[0081] Take the value with the highest posterior probability The measured temperature of the battery pack, as predicted by a virtual sensor, ensures uninterrupted closed-loop control of multimodal data sampling.

[0082] After predicting the future temperature rise of the battery pack using the LSTM time-series prediction model, the energy storage cabinet is thermally managed using the following method.

[0083] Mode I: When the future temperature rise of the battery pack does not exceed the first threshold, the energy storage cabinet adopts natural convection heat dissipation mode. This mode is a zero-power mode, which uses natural convection to achieve heat dissipation.

[0084] Mode II: When the future temperature rise of the battery pack is between the first and second thresholds, the energy storage cabinet adopts a natural convection cooling mode. The natural convection wind speed is controlled by PID feedback, and a humidity correction factor based on the ambient humidity of the energy storage cabinet is introduced to dynamically adjust the proportional gain of the PID controller. This mode is a PID humidity-heat coupling regulation mode. As the ambient humidity of the energy storage cabinet increases, the natural convection air volume is automatically increased to compensate for the fluctuation of the heat exchange efficiency of the humid air.

[0085] Mode III: When the battery pack temperature rises to the second threshold temperature, the energy storage cabinet enters forced cooling mode, activating the forced cooling channel for the battery pack. This mode combines forced cooling intervention with anti-condensation. In this mode, a minimum-value selection strategy is used to calculate the final control value of the forced cooling channel, taking the smaller value between the "full-speed cooling demand value" and the "condensation safety limit value." In forced cooling mode, when the difference between the measured battery pack temperature and the dew point temperature under the current energy storage cabinet ambient humidity is less than the safety threshold, the forced cooling channel is controlled at the first forced cooling channel control value under the condensation safety limit. When the difference between the measured battery pack temperature and the dew point temperature under the current energy storage cabinet ambient humidity is greater than or equal to the safety threshold, the forced cooling channel is controlled at the second forced cooling channel control value under the full-speed cooling demand.

[0086] In practical applications, each layer of the energy storage cabinet is equipped with natural convection channels for natural air circulation between the inside and outside of the cabinet. Natural convection fans are installed on the cabinet to regulate the airflow speed. The forced cooling channels for the battery pack inside the energy storage cabinet generally use liquid cooling or air cooling channels. In this embodiment, the forced cooling channel control quantity refers to the duty cycle of the cooling medium drive component within the corresponding channel. This is based on the predicted future temperature rise of the battery pack. and the ambient humidity of the energy storage cabinet The duty cycle of the final fan and the medium drive component of the forced cooling channel is calculated using the following piecewise function. .

[0087] The first threshold is set at 20℃, the second threshold at 35℃, and the safety threshold at 2℃.

[0088] In Mode I, when the battery pack temperature rises in the future... Temperature <20℃ and ambient humidity of the energy storage cabinet When the power consumption is less than 20%, the natural convection cooling mode enters zero-power mode. At this time, neither the natural convection fan on the cabinet nor the medium drive component on the forced cooling channel is activated. The duty cycle of the output natural convection fan is:

[0089] .

[0090] The natural convection fan has a 0% duty cycle, and the battery pack inside the energy storage cabinet relies solely on natural convection for heat dissipation.

[0091] In Mode II, when the future temperature rise of the battery pack meets the requirement of 20℃≤ At temperatures below 35℃, the natural convection fan on the energy storage cabinet is activated based on the natural convection mode of Mode I. The wind speed of natural convection is controlled by PID feedback, and a humidity correction factor based on the ambient humidity of the energy storage cabinet is introduced to dynamically adjust the proportional gain of the PID controller.

[0092] Traditional PID control algorithms rely solely on temperature error as a single input, neglecting changes in air medium properties. At the same temperature rise, the enthalpy of high-humidity air differs from that of dry air, and condensation easily occurs in the low-velocity zone at the air inlet under high humidity conditions. This embodiment employs a humidity-adaptive PID algorithm, introducing a humidity correction factor based on the ambient humidity of the energy storage cabinet, in addition to the traditional PID algorithm. Its PID control equation is:

[0093] .

[0094] in, The duty cycle of a natural convection fan. Actual temperature rise of the battery pack The proportional gain of the PID control equation is dynamically adjusted based on the humidity correction factor. The integral coefficients of the PID control equation are... These are the differential coefficients of the PID control equation.

[0095] The humidity correction factor is expressed as a dynamic function of humidity in the PID control equation:

[0096] .

[0097] in, The proportional gain of the PID control equation is dynamically adjusted based on the humidity correction factor. The ambient humidity of the energy storage cabinet is collected at time k. Humidity threshold Based on the base proportional gain, it is pre-calibrated according to the heat dissipation requirements of the energy storage cabinet and the performance of the fan. This is the humidity gain coefficient (adjustment factor). This is the activation function, which only takes effect when the input is a positive number.

[0098] When the ambient humidity of the energy storage cabinet When the humidity is greater than or equal to the humidity threshold, the system automatically increases the proportional gain coefficient of the PID control to adjust the ambient humidity of the energy storage cabinet. For every 10% increase Gain boost ×0.1 times, utilizing the high wind speed of natural convection inside the energy storage cabinet to compensate for the decrease in heat exchange efficiency of humid air and prevent local moisture accumulation.

[0099] In Mode III, when the battery pack temperature rises in the future... >35℃ or measured temperature rise of the battery pack When the temperature reaches >35℃, the energy storage cabinet enters the forced cooling mode and activates the strong cooling flow channel for the battery pack. The control objective in the forced cooling mode is to provide the maximum cooling capacity to suppress temperature rise while ensuring that physical condensation does not occur.

[0100] In the forced cooling mode, the minimum value is used to preferentially select either the first or second forced cooling flow channel control quantity. .

[0101] in, To minimize the optimal selection of the strong cooling flow channel control quantity, The control quantity for the second-strong cooling channel under full-speed cooling requirements is initially set to the maximum flow rate of the strong cooling channel, meaning the cooling medium drive component of the strong cooling channel operates at 100% duty cycle. This is to cope with the extreme heat dissipation requirements of the high-temperature emergency zone. The first strong cooling flow channel control quantity under the condensation safety limit is a safety upper limit dynamically calculated by a fuzzy logic control algorithm. If there is a risk of condensation in the ambient humidity of the energy storage cabinet, this value will forcibly reduce the duty cycle output of the strong cooling flow channel cooling medium drive component. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, the first strong cooling flow channel control quantity is set to the lower limit. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current ambient humidity of the energy storage cabinet is greater than the high humidity threshold, the first strong cooling flow channel control quantity is set to the middle limit. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current ambient humidity of the energy storage cabinet is less than or equal to the high humidity threshold, the first strong cooling flow channel control quantity is set to the highest limit. The lower and middle limits of the first strong cooling flow channel control quantity are lower than the second strong cooling flow channel control quantity, and the highest limit of the first strong cooling flow channel control quantity is greater than the second strong cooling flow channel control quantity.

[0102] Before implementing the forced cooling mode, this embodiment must calculate the dew point temperature under the current ambient humidity of the energy storage cabinet to establish a safe lower limit for thermal management. This embodiment uses the Magnus-Tetens empirical formula to calculate the dew point temperature under the current ambient humidity of the energy storage cabinet in real time:

[0103] ,

[0104] .

[0105] in, Where is the dew point temperature, and γ(T,H) is an intermediate variable. The ambient temperature of the energy storage cabinet. The ambient humidity of the energy storage cabinet is constant b=17.625, c=243.04℃.

[0106] In forced cooling mode, all output quantities The following anti-condensation safety constraints must be met to prevent condensation from forming on the battery surface:

[0107] .

[0108] in, The measured temperature of the battery pack surface at time k. The forced cooling output u corresponds to the decrease in battery surface temperature.

[0109] in =2℃ is the safety threshold.

[0110] Specifically, such as Figure 2 As shown, in forced cooling mode, the system's operating limits are determined based on the current intake air humidity and the difference between the battery pack temperature and the dew point temperature. The specific control strategy includes two levels: the future temperature rise of the battery pack... >35℃ or measured temperature rise of the battery pack At temperatures above 35℃, the forced cooling mode activates anti-condensation fuzzy logic, and the control quantity of the forced cooling flow channel is set in stages according to the severity of the environment. The specific strategy is as follows:

[0111] When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, i.e., the above-mentioned anti-condensation safety constraints are not met, the battery pack is determined to be in an "immediate condensation risk" state, indicating an extremely high risk of condensation. At this time, the first-level restriction strategy is implemented, adjusting the control amount of the first strong cooling flow channel. The operating limit value is set to Low, such as the duty cycle of the forced cooling flow channel drive component being 0%-10%. The final output is a low level of the first forced cooling flow channel control quantity to suppress the cooling output to the greatest extent and prevent condensation. At this time, a physical phase change is imminent, and the highest priority action must be executed to prevent the battery surface temperature from falling further below the dew point by almost cutting off the forced cooling output.

[0112] When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, but the current ambient humidity of the energy storage cabinet is... Greater than the high humidity threshold, When the air humidity is >50%, the intake air humidity is extremely high, creating a "high-risk environment for potential condensation." In this case, the second-level limiting strategy is implemented, and the first-level strong cooling flow channel control quantity is set. The operating limit value is Medium, such as when the duty cycle of the forced cooling flow channel cooling drive is 40%-50%, at which point the control quantity of the first forced cooling flow channel is... The moderate limit is still lower than the second strongest cold runner control value under full-speed cooling demand. Ultimately, the output is a moderately limited first-stage cooling channel control quantity, appropriately reducing the upper limit of the cooling channel power. This satisfies cooling requirements while reducing the potential risk of condensation in high-humidity environments. By limiting the upper limit of power, the rate of temperature decrease (cooling slope) of the battery surface is reduced, preventing localized temperature spikes below the dew point due to excessive cooling, thus achieving "defensive derating." This ensures that the battery surface temperature always meets the following requirements: > +2℃.

[0113] The first-level restriction strategy is more restrictive than the second-level restriction strategy, reflecting the system's priority response to the risk of direct condensation.

[0114] When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current ambient humidity of the energy storage cabinet is... Less than or equal to the high humidity threshold, ≤50%, at this point the internal environment of the energy storage cabinet is dry, and the first strong cooling flow channel control quantity is set. This is the maximum limit; at this point, the control quantity of the first strong cooling runner is... The maximum limit is greater than the control value of the second strongest cold runner under the full-speed cooling requirement. Ultimately, the second strongest cold runner control quantity is output. The strong cooling channel cooling drive component has a duty cycle of 100%, which rapidly suppresses temperature rise.

[0115] The control quantity for the first strong cooling channel is output by a fuzzy logic control algorithm based on the measured temperature of the battery pack and the ambient humidity of the energy storage cabinet.

[0116] .

[0117] in, For fuzzy inference functions, Actual temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet The difference between them The ambient humidity of the energy storage cabinet.

[0118] Fuzzy inference function The output solution process includes three steps: fuzzification, fuzzy inference, and defuzzification. In fuzzification, the real-time collected data of "battery pack measured temperature - dew point temperature difference" and "energy storage cabinet ambient humidity" are mapped to fuzzy linguistic variables using preset Gaussian or triangular membership functions, and their corresponding membership values ​​are calculated. Fuzzy inference is performed based on the Mamdani inference model, calling the pre-stored IF-THEN rule base according to the rules set in Mode III. The "Min-Max" operation rule is used to calculate the trigger strength of each rule based on the input membership degree, and the output membership function is truncated. Defuzzification uses the Center of Gravity (CoG) method to aggregate and defuzzify the output fuzzy sets of all activated rules, calculating the centroid abscissa of the aggregated geometric figure and converting it into a precise percentage value, such as... = 42%, which serves as the final safety limit value and is used in the minimum value selection decision for either the first or second strong cold runner control quantity.

[0119] The forced cooling mode reduces the measured temperature of the battery pack to a first threshold value when the measured temperature of the battery pack rises or falls between the measured temperature of the battery pack and the ambient temperature of the energy storage cabinet. If the temperature is below 20℃, maintain the current cooling mode for a 20-40 second delay and then exit by executing the following decay function:

[0120] .

[0121] in, Let be the forced cooling output at time t. To reduce the cooling output at startup, The decay time constant, This is the base of the exponential decay, approximately equal to 2.71828.

[0122] Because individual battery cells have significant thermal inertia, the internal temperature of the cells may remain high even as the surface temperature of the battery pack decreases. If cooling is stopped immediately, the outward conduction of internal heat will cause a rebound in surface temperature, inducing frequent start-stop cycles of the cooling medium drive components in the strong cooling channel. Delaying the operation with low airflow before stopping cooling removes residual heat conducted from the cells, eliminating thermal rebound caused by thermal hysteresis within the battery pack and ensuring a stable temperature decrease.

[0123] To address battery aging, increased internal resistance leading to greater heat generation, and seasonal changes, after each forced cooling mode, the parameters of the LSTM time-series prediction model are fine-tuned using the measured temperature of the battery pack at the same time as the predicted future temperature rise. The specific process is as follows:

[0124] Loss function calculation: Calculate the mean square error (MSE) between the future temperature rise of the battery pack predicted by the LSTM time-series prediction model and the measured temperature of the battery pack at the same moment under forced cooling mode.

[0125] .

[0126] in, This refers to the number of sampling points within the forced cooling mode. For sampling point index, For the first LSTM at 1 sampling point predicts future temperature rise. For the first The measured temperature rise of the battery pack at each sampling point.

[0127] Model parameter update (gradient descent): Update the LSTM model weights using the backpropagation time-propagation (BPTT) algorithm. :

[0128] .

[0129] in, For the updated LSTM model parameter set, The set of parameters for the LSTM model before the update. For learning rate, For the loss function with respect to model parameters The gradient.

[0130] in For a very small learning rate, such as 10 −4 This ensures that the model is only fine-tuned, avoiding the forgetting of old knowledge.

[0131] See Figure 3 This embodiment also discloses an intelligent thermal management system for energy storage cabinets that combines moisture and heat coupling control, including a data acquisition module, a calculation and control module, and a heat dissipation execution module inside the energy storage cabinet. The heat dissipation execution module includes a natural convection fan and a cooling medium driving component in a strong cooling channel, including a strong cooling water pump that drives the liquid cooling medium and a strong cooling fan that drives the air cooling medium.

[0132] The data acquisition module collects real-time data on the measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current. The calculation and control module, employing an edge computing controller and integrating an AI acceleration engine, runs the LSTM timing prediction model and fuzzy logic control algorithm that implement the thermal management method described in this embodiment. This predicts the future temperature rise of the battery pack and manages the energy storage cabinet's thermal performance through the following modes: when the future temperature rise of the battery pack does not exceed a first threshold, the energy storage cabinet operates in natural convection cooling mode; when the future temperature rise of the battery pack is between the first and second thresholds, the energy storage cabinet operates in natural convection cooling mode, with PID feedback controlling the natural convection airflow speed and incorporating humidity correction based on the ambient humidity of the energy storage cabinet. The positive factor dynamically adjusts the proportional gain of the PID controller; when the future temperature rise of the battery pack reaches the second threshold temperature, the energy storage cabinet enters forced cooling mode and activates the strong cooling channel for the battery pack. In forced cooling mode, when the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, the strong cooling channel is controlled as the first strong cooling channel control quantity under the condensation safety limit; when the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet exceeds the safety threshold, the strong cooling channel is controlled as the second strong cooling channel control quantity under the full-speed cooling requirement.

[0133] The energy storage cabinet is equipped with a natural convection channel connecting the inside and outside of the cabinet, and a natural convection fan is installed on it. The wind speed of the natural convection entering and exiting the energy storage cabinet is controlled by the calculation control module to handle the conventional heat load in Mode II.

[0134] The forced cooling channel of the energy storage cabinet is a liquid cooling channel or air cooling channel connected to the battery pack inside the energy storage cabinet. The forced cooling channel control quantity of the corresponding cooling medium driving component in the channel is controlled by the computing control module to cope with the peak heat load in Mode III and prevent thermal runaway.

[0135] After completing the above-mentioned closed-loop control of thermal management, the intelligent thermal management system of the energy storage cabinet in this embodiment uploads the thermal data and fault logs generated during the thermal management process to the cloud through an asynchronous thread, and pulls global model updates as needed to achieve collaborative optimization of the system.

[0136] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the embodiments of the present invention have been clearly and completely described above in conjunction with the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0137] Therefore, the above detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0138] It should be noted that, unless otherwise specified, the embodiments and features described in this invention can be combined with each other. It should also be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments; similar or identical parts between embodiments can be referred to interchangeably.

[0139] The above specific embodiments further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made 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 intelligent thermal management of an energy storage cabinet combining moisture and heat coupling control, characterized in that: Real-time data collection of the measured temperature of the battery pack in the energy storage cabinet, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current are used to predict the future temperature rise of the battery pack. Thermal management of the energy storage cabinet is carried out through the following methods; When the future temperature rise of the battery pack does not exceed the first threshold, the energy storage cabinet will use natural convection cooling mode. When the future temperature rise of the battery pack is between the first threshold and the second threshold, the energy storage cabinet adopts the natural convection heat dissipation mode. The wind speed of natural convection is controlled by PID feedback, and a humidity correction factor based on the ambient humidity of the energy storage cabinet is introduced to dynamically adjust the proportional gain of the PID controller. When the battery pack temperature rises to the second threshold temperature, the energy storage cabinet will switch to forced cooling mode, activating the strong cooling flow channel for the battery pack. In forced cooling mode... When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, the controlled cooling channel is set to the first controlled cooling channel value under the condensation safety limit. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, the control value of the strong cooling channel is set to the second strong cooling channel control value under the full-speed cooling requirement.

2. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: The measured temperature of the battery pack is acquired using a two-stage sensor system. The primary sensor serves as the main channel sensor, collecting the measured temperature of the battery pack in real time. The secondary sensor acts as a backup channel sensor. When the temperature data collected by the primary channel sensor is abnormal, the following fault-tolerant mechanism is used to correct the data: The first-level fault-tolerance mechanism switches to the backup channel sensor to collect the real-time temperature of the battery pack. If the temperature data collected by the backup channel sensor is within the fluctuation range of the actual measured temperature of the battery pack before the main channel sensor malfunctions, the backup channel sensor is directly used to replace the main channel sensor, and the corresponding main channel sensor is marked as faulty. If the temperature data collected by the backup channel sensor is abnormal, the second-level fault-tolerance mechanism is activated. Combining the actual measured temperature data of adjacent battery packs, a virtual sensor is constructed using Bayesian probability to predict the actual measured temperature of the battery pack, and the corresponding backup channel sensor is marked as faulty.

3. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: An LSTM time-series prediction model is used to predict the future temperature rise of the battery pack. The measured temperature of the battery pack, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current are used to construct an input vector. The input vector is then filtered by moving average and normalized by feature to form the tensor input of the LSTM time-series prediction model. The LSTM time-series prediction model includes an input layer, two stacked LSTM hidden layers, and a fully connected output layer. The fully connected output layer maps the final hidden state of the tensor input of the LSTM hidden layers to the predicted future temperature rise of the battery pack at future time.

4. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 2, characterized in that: After the forced cooling mode ends, the parameters of the LSTM time series prediction model are fine-tuned using the predicted future temperature rise of the battery pack at the same time as the measured temperature of the battery pack.

5. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: The humidity correction factor is expressed as a dynamic function of humidity in the PID control equation: ， in, The proportional gain of the PID control equation is dynamically adjusted based on the humidity correction factor. The ambient humidity of the energy storage cabinet is collected at time k. Humidity threshold Based on the proportional gain, Humidity gain coefficient This is the activation function.

6. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: In the forced cooling mode, the first or second strong cooling channel control quantity is selected by minimizing the value. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is less than the safety threshold, the first strong cooling channel control quantity is set to the lower limit. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current ambient humidity of the energy storage cabinet is greater than the high humidity threshold, the first strong cooling channel control quantity is set to the middle limit. When the difference between the measured temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet is greater than or equal to the safety threshold, and the current ambient humidity of the energy storage cabinet is less than or equal to the high humidity threshold, the first strong cooling channel control quantity is set to the highest limit. When used as the lower and middle limits, the first strong cooling channel control quantity is less than the second strong cooling channel control quantity. When used as the highest limit, the first strong cooling channel control quantity is greater than the second strong cooling channel control quantity.

7. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 6, characterized in that: The control quantity for the first strong cooling channel is output by a fuzzy logic control algorithm based on the measured temperature of the battery pack and the ambient humidity of the energy storage cabinet. ， in, For fuzzy inference functions, Actual temperature of the battery pack and the dew point temperature under the current ambient humidity of the energy storage cabinet The difference between them The ambient humidity of the energy storage cabinet.

8. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: The dew point temperature is calculated in real time using the Magnus-Tetens empirical formula to determine the ambient humidity of the energy storage cabinet.

9. The intelligent thermal management method for energy storage cabinets combining moisture and heat coupling control according to claim 1, characterized in that: The forced cooling mode maintains the current heat dissipation mode for 20-40 seconds after the measured temperature of the battery pack rises or falls to the first threshold between the measured temperature of the battery pack and the ambient temperature of the energy storage cabinet, and then exits.

10. An intelligent thermal management system for energy storage cabinets that combines moisture and heat coupling control, characterized in that... include: The data acquisition module collects real-time data on the measured temperature of the battery pack in the energy storage cabinet, the ambient temperature of the energy storage cabinet, the ambient humidity of the energy storage cabinet, and the load current. The calculation and control module implements the intelligent thermal management method for energy storage cabinets according to any one of claims 1-9 to perform thermal management on the energy storage cabinet; The natural convection channel connects the inside and outside of the energy storage cabinet. It is equipped with a natural convection fan, and the wind speed of natural convection entering and exiting the energy storage cabinet is controlled by the computing control module. The forced cooling channel is a liquid-cooled or air-cooled channel that connects to the battery pack inside the energy storage cabinet. The forced cooling channel control quantity is controlled by the computing control module to control the cooling medium driving component in the corresponding channel.

Images