Wind-light-storage battery immersion liquid cooling intelligent prediction control method and system

Abstract

The application discloses a wind-solar energy storage battery immersion liquid cooling intelligent prediction control method and system, relates to the battery liquid cooling prediction technical field and comprises the following steps: S1, real-time collection of wind storage liquid cooling multi-source data, data preprocessing operation is carried out, and a heating trend prediction model is constructed;S2, working substance pressure regulation and control, adjust the speed of the two-way pump and the working pressure of the condenser;S3, quantifying the cooling heat exchange demand of the energy storage battery, adjusting the cooling equipment, and feeding forward response of the heat load;S4, monitoring the wind storage liquid cooling multi-source data in the adjustment process, judging the opening degree of the electric control valve, adjusting the cooling liquid flow, and optimizing and updating the heating trend prediction model parameters according to the deviation of the pressure regulation and control, the cooling heat exchange demand and the opening degree of the electric control valve.The problem that the existing energy storage battery two-phase immersion liquid cooling system cannot timely respond to the intermittent power mutation of wind-solar power generation due to passive temperature control response lag, leading to easy overheating of the battery and energy waste, affecting the safety of the system is solved.

Description

Technical Field

[0001] This invention relates to the field of battery liquid cooling prediction technology, specifically to a method and system for intelligent prediction and control of immersion liquid cooling of wind and solar energy storage batteries. Background Technology

[0002] With the continuous large-scale grid connection of new energy sources, energy storage power stations, as a key supporting technology for improving the absorption capacity of new energy, smoothing grid fluctuations, and ensuring the safe and stable operation of the system, are facing increasing application demands and technological advancements. Two-phase immersion liquid cooling technology, due to its efficient and uniform heat dissipation capabilities, is gradually becoming an important direction for the thermal management of next-generation energy storage batteries.

[0003] For example, invention patent CN117371304A discloses a method for optimizing the structure of a liquid-cooled lithium-ion battery pack, belonging to the field of liquid cooling heat dissipation technology for electric vehicle power batteries. The method includes: establishing a geometric model of the liquid-cooled lithium-ion battery pack, setting design variables and analysis objectives, and performing mesh generation; the design variables include: inlet flow velocity, inlet temperature, ambient temperature, channel thickness, and channel width; the analysis objectives are the highest temperature and the maximum temperature difference; using a regression model to describe the relationship between the design variables and the analysis objectives, and using response surface methodology to analyze and obtain the regression equation; substituting the regression equation into a particle swarm optimization algorithm to find the optimal structure, and selecting the optimal structural parameters based on the Pareto optimal solution. This invention utilizes response surface methodology to establish a model, which facilitates and quickly finds the optimal comprehensive value to reduce analysis time. The optimized parameters can achieve the best cooling performance, reduce the overall battery temperature, and improve the uniformity of temperature distribution.

[0004] For example, invention patent CN119475763A discloses a liquid cooling method and apparatus for a battery pack, relating to the field of liquid cooling. The method includes constructing a simulation model based on the battery cells and coolant in the battery pack; performing discharge simulation on the simulation model to obtain the temperature distribution of the battery cells and the flow rate and temperature of the coolant; determining the flow rate and temperature of the coolant based on the influence relationship between the maximum temperature and maximum temperature difference and the flow rate and temperature of the coolant when both are at their minimum values; and liquid cooling of the battery pack based on the flow rate and temperature of the coolant. Through simulation of the battery pack, the simulation model simulates the power generation state of the actual battery pack. Adjusting the flow rate and temperature of the coolant in the simulation model to minimize both the maximum temperature and maximum temperature difference of the battery pack, liquid cooling of the battery pack is achieved based on the flow rate and temperature values ​​of the simulation model. The entire process has a short cycle and does not require verification with an actual battery pack, thus reducing costs.

[0005] However, the power output of wind and solar power is significantly intermittent and fluctuating, directly causing frequent and sudden changes in the charging and discharging power of batteries in energy storage power stations, and large fluctuations in the heating rate within a short period of time. Existing passive heat dissipation control results in a lag in response. When the power increases suddenly, the heat dissipation system cannot prepare in advance, which can easily lead to local overheating of the battery due to insufficient heat dissipation capacity, affecting battery life and even causing safety risks. At the same time, there is also energy waste. When the power drops suddenly, the heat dissipation system still maintains high load operation, resulting in excessive heat dissipation and increasing unnecessary energy consumption.

[0006] Therefore, in order to address the above problems, there is an urgent need for intelligent predictive control methods and systems for immersion liquid cooling of wind and solar energy storage batteries. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides an intelligent predictive control method and system for immersion liquid cooling of wind and solar energy storage batteries. This solves the problem that existing two-phase immersion liquid cooling systems for energy storage batteries suffer from delayed passive temperature control response, making them unable to respond promptly to intermittent power surges in wind and solar power generation, leading to battery overheating, energy waste, and compromised system safety.

[0009] Technical solution

[0010] To achieve the above objectives, the present invention provides the following technical solution: an intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries, comprising the following steps: S1, real-time acquisition of multi-source data on wind and solar energy storage liquid cooling, data preprocessing of the multi-source data, and construction of a heat generation trend prediction model based on the multi-source data; S2, based on the preprocessed multi-source data and the heat generation trend prediction model, regulation of the working fluid pressure through nonlinear calculation and temperature compensation, and real-time adjustment of the bidirectional pump speed and condenser working pressure; S3, quantification of the cooling heat exchange requirements of the energy storage battery based on the preprocessed multi-source data and the heat generation trend prediction model, adjustment of the cooling equipment according to the cooling heat exchange requirements, and feedforward response of the heat load; S4, real-time monitoring of the multi-source data on wind and solar energy storage liquid cooling during the adjustment process, determination of the opening degree of the electronic control valve, adjustment of the coolant flow rate according to the opening degree of the electronic control valve, and optimization and updating of the heat generation trend prediction model parameters based on the deviation between pressure regulation, cooling heat exchange requirements, and electronic control valve opening.

[0011] Furthermore, the specific process of real-time acquisition of multi-source data on wind-storage-liquid cooling and the preprocessing of this data is as follows: Real-time acquisition of multi-source data on wind-storage-liquid cooling involves the following steps: Powered by the photovoltaic and wind power systems, smart meters and various sensors are used to collect real-time data on photovoltaic power generation, wind power generation, battery charge / discharge power and temperature of the energy storage battery, cooling circuit outlet temperature, cooling circuit inlet temperature, coolant mass flow rate, sealed tank pressure, buffer tank pressure, working fluid level, and real-time working fluid density. Simultaneously, the specific heat capacity of the coolant and the standard density of the working fluid are also collected. Simultaneously, the battery charge / discharge power and temperature data are used to perform battery energy efficiency analysis. The battery heating power is calculated using a quantity balance algorithm, and the rate of change of battery heating power is calculated based on a sliding time window. Satellite remote sensing data is used to collect light intensity, wind speed, wind direction, ambient temperature, and ambient humidity for the next ten minutes. Missing values ​​are detected and linearly interpolated to complete the multi-source data of wind-storage-liquid cooling, and time alignment and resampling are performed to standardize the sampling interval. Sliding window normalization and standard score normalization methods are used to standardize and dimensionlessly normalize the multi-source data of wind-storage-liquid cooling, and a median filtering algorithm is used to remove abnormal peaks and noise from sensors. A multi-source data database of wind-storage-liquid cooling is established, storing the original and preprocessed multi-source data.

[0012] Furthermore, the specific process of constructing a heat generation trend prediction model based on multi-source data of wind, storage, and liquid cooling is as follows: historical photovoltaic power generation, wind power generation, battery charging and discharging power, battery temperature, battery heat generation power change rate, and corresponding light intensity, wind speed, wind direction, ambient temperature, and ambient humidity are combined to construct a heat generation feature dataset. The heat generation feature dataset is trained using a long short-term memory network deep time series modeling algorithm to construct a heat generation trend prediction model. The model outputs the predicted change rate of battery heat generation power of the energy storage battery in real time, and stores the original and normalized predicted change rates of battery heat generation power with time labels into the wind, storage, and liquid cooling multi-source data database.

[0013] Furthermore, based on the preprocessed multi-source data of air-cooled and liquid-cooled systems and the heat generation trend prediction model, the specific process of regulating the working fluid pressure through nonlinear calculation and temperature compensation is as follows: Real-time acquisition of multi-source data of air-cooled and liquid-cooled systems and the predicted change rate of battery heat generation power; statistical analysis of battery heat generation power and calculation of the average value based on a sliding time window to obtain the battery heat generation power benchmark; simultaneous use of atmospheric pressure as the initial reference pressure value and ambient temperature as the target set value for battery temperature; addition of the ratio of the predicted change rate of battery heat generation power to the battery heat generation power benchmark to a constant, and performing natural logarithmic calculation to obtain the heat generation prediction correction value; subtraction of the target set value for battery temperature from the current battery temperature to obtain the temperature deviation value; and subtraction of the product of the heat generation prediction correction value and the heat generation prediction sensitivity weighting factor from the initial reference pressure value, and subtraction of the product of the temperature deviation value and the temperature compensation weighting factor to obtain the boiling heat dissipation pressure regulation value.

[0014] Furthermore, the specific process of real-time adjustment of the bidirectional pump speed and condenser working pressure is as follows: The control center calculates the boiling heat dissipation pressure control value in real time, writes it into the air-cooled liquid-cooled multi-source data database, and controls the speed of the bidirectional pump and the working pressure of the condenser according to the boiling heat dissipation pressure control value, adjusts the working fluid pressure in the sealed box, and coordinates with the gas collection tank for gas supply and recovery. At the same time, the actual sealed box pressure is fed back. When the deviation between the actual sealed box pressure and the boiling heat dissipation pressure control value exceeds the deviation threshold, the backup pressure regulating unit is switched. When the standard deviation of the working fluid level height is detected to be greater than the fluctuation threshold, the boiling heat dissipation pressure control value is increased. The initial reference pressure value and the target battery temperature setting value are corrected periodically based on the deviation between the actual sealed box pressure and the boiling heat dissipation pressure control value using sliding window data statistics and an adaptive minimum mean square error algorithm.

[0015] Furthermore, based on the preprocessed multi-source data of wind-cooled and liquid-cooled systems and the heat generation trend prediction model, the specific process for quantifying the cooling and heat exchange requirements of energy storage batteries is as follows: Obtain the predicted rate of change of battery heat generation power, current sealed enclosure pressure, battery temperature, and temperature deviation; combine historical battery temperature and temperature deviation values, and obtain battery charging / discharging power and battery heat generation power data; use the least squares method of a first-order inertial model to fit the first-order delay response curve of battery temperature to calculate the thermal inertia time constant; based on historical sliding time windows, statistically analyze battery temperature and battery charging / discharging power, select intervals where the battery temperature change amplitude is continuously greater than a time threshold and the change direction is consistent, accumulate the battery charging / discharging power within the interval, and calculate... The battery heat capacity is obtained by calculating the interval length and the ratio of the cumulative value of battery charging and discharging power to the interval length. The heat capacity inertia coefficient is obtained by calculating the ratio of battery heat capacity to thermal inertia time constant. The heat capacity inertia coefficient is multiplied by the temperature deviation value to obtain the temperature difference driven heat correction amount. The absolute value of the temperature deviation value is divided by the temperature response adjustment weight factor and the negative value is taken as the exponent to obtain the exponential correction value. The nonlinear response correction factor is obtained by subtracting the exponential correction value from the constant. The temperature difference driven heat correction amount is multiplied by the nonlinear response correction factor and added to the predicted change rate of battery heating power to obtain the cooling demand prediction value. The cooling demand prediction value is multiplied by the comprehensive correction weight factor to obtain the cooling heat exchange load control value.

[0016] Furthermore, the specific process of adjusting the cooling equipment according to the cooling heat exchange demand and performing feedforward response to the heat load is as follows: The control center adjusts the power of the cooling coil according to the cooling heat exchange load control value, and at the same time, it acquires the inlet temperature of the cooling loop, the outlet temperature of the cooling loop, the mass flow rate of the coolant, and the specific heat capacity of the coolant in real time to calculate the actual heat exchange. If there is a deviation between the actual heat exchange and the cooling heat exchange load control value, the speed of the bidirectional pump and the power of the cooling coil are corrected according to the deviation. When the rate of change of battery heating power is detected to be greater than the sudden change threshold, the emergency cooling strategy is activated: the speed of the maximum bidirectional pump is increased, the power output of the cooling coil is increased, and the redundant cooling loop is enabled. The cooling heat exchange load control value and the actual heat exchange are written into the air-cooled liquid cooling multi-source data database for real-time statistical analysis. The cooling heat exchange load control value parameters are optimized through the minimum mean square error and sliding window self-learning algorithm.

[0017] Furthermore, the specific process for determining the opening degree of the electronically controlled valve by real-time monitoring of multi-source data of air-cooled liquid storage during the adjustment process is as follows: During the adjustment of boiling heat dissipation pressure and cooling heat exchange load, the pressure of the sealing box and the pressure of the buffer tank are acquired in real time, and the difference between the pressure of the sealing box and the pressure of the buffer tank is calculated to obtain the pressure deviation value. The pressure deviation value is multiplied by the pressure deviation weighting factor to obtain the pressure adjustment term; the real-time density of the working fluid liquid phase and the standard density of the working fluid are acquired, and the difference between the real-time density of the working fluid liquid phase and the standard density of the working fluid is calculated to obtain the density deviation value. The density deviation value is multiplied by the density deviation weighting factor to obtain the density adjustment term; the pressure adjustment term and the density adjustment term are summed, and the negative value of the summation result is taken as the exponent. Natural exponent calculation is performed to obtain the comprehensive adjustment index factor. The comprehensive adjustment index factor is added to the constant and the reciprocal is taken to obtain the valve opening control value.

[0018] Furthermore, the specific process of adjusting the coolant flow rate according to the opening of the electrically controlled valve, and optimizing and updating the parameters of the heat generation trend prediction model based on the deviation between pressure regulation, cooling heat exchange requirements, and the opening of the electrically controlled valve, is as follows: The control center generates valve regulation commands based on the valve opening regulation values ​​and sends them to the electrically controlled valve in real time to adjust the valve opening and regulate the coolant flow rate; if the continuously collected pressure deviation and density deviation values ​​are detected to exceed the pressure deviation threshold and density deviation threshold respectively, the maximum valve opening limit is triggered, and an audible and visual alarm is triggered; the valve opening regulation values, valve regulation commands, pressure deviation values, and density deviation values ​​are written into the multi-source data database of air-cooled liquid cooling system in real time, and the regulation sensitivity, response speed, and energy efficiency are evaluated in conjunction with the electrically controlled valve opening regulation process, and the standard density of the working fluid is adjusted periodically; the multi-source data of air-cooled liquid cooling system, boiling heat dissipation pressure regulation values, cooling heat exchange load regulation values, and valve opening regulation values ​​during operation are continuously acquired, and the deviation between the regulation values ​​of each link and the corresponding actual feedback data is monitored. The parameters of the heat generation trend prediction model are corrected using the recursive least squares method and the minimum mean square error adaptive algorithm.

[0019] The second aspect of this invention provides an intelligent predictive control system for immersion liquid cooling of wind and solar energy storage batteries, comprising: a multi-source data acquisition and intelligent prediction module, used to acquire multi-source data on wind and solar energy storage liquid cooling in real time, perform data preprocessing on the multi-source data, and construct a heat generation trend prediction model based on the multi-source data; a boiling heat dissipation pressure active regulation module, used to regulate the working fluid pressure through nonlinear calculation and temperature compensation based on the preprocessed multi-source data and the heat generation trend prediction model, and to adjust the bidirectional pump speed and condenser working pressure in real time; a cooling power adaptive regulation module, used to quantify the cooling heat exchange demand of the energy storage battery based on the preprocessed multi-source data and the heat generation trend prediction model, and to adjust the cooling equipment according to the cooling heat exchange demand, performing a feedforward response to the heat load; and a working fluid gas-liquid mixing and self-learning fine-tuning module, used to monitor the multi-source data on wind and solar energy storage liquid cooling in real time during the regulation process, determine the opening degree of the electronic control valve, adjust the coolant flow rate according to the opening degree of the electronic control valve, and simultaneously optimize and update the parameters of the heat generation trend prediction model based on the deviation between the pressure regulation, the cooling heat exchange demand, and the opening degree of the electronic control valve.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) This invention uses real-time acquisition and intelligent prediction of multi-source data and LSTM deep time series model to predict the trend of battery heat generation power change in advance, upgrades liquid cooling heat dissipation regulation from passive response to active and advanced control, effectively improves the adaptability of energy storage battery to new energy power change, significantly reduces the risk of battery overheating and thermal runaway, and enhances system safety and operational reliability.

[0023] (2) This invention constructs a boiling heat dissipation pressure control value through heat generation trend prediction, nonlinear calculation and temperature compensation mechanism, and dynamically adjusts key parameters such as working fluid pressure, cooling coil power and pump speed to ensure that the boiling heat dissipation intensity is accurately coupled with the actual heat load of the battery, thereby avoiding the response lag and energy waste of traditional heat dissipation systems and greatly improving energy efficiency and thermal management flexibility.

[0024] (3) This invention uses first-order inertial modeling and sliding window statistical methods to identify the physical properties of battery heat capacity and thermal inertia time constant in real time. Combined with temperature difference-driven thermal correction and nonlinear response factor, it realizes the scientific quantification and adaptive adjustment of cooling heat exchange demand, effectively suppresses sudden temperature rise and ensures that the battery pack is always in a safe and efficient range.

[0025] (4) This invention, by real-time monitoring of the deviation between the control values ​​and actual feedback at each stage, and combining recursive least squares method and minimum mean square error adaptive algorithm, dynamically corrects the heating trend prediction model and control parameters, realizing data closure, online self-learning and self-optimization throughout the entire process. It has the ability to adapt to long-term operating conditions and working fluid aging drift, continuously optimizes thermal management strategies, and improves long-term stability and intelligent operation and maintenance level.

[0026] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0027] Figure 1 Flowchart of intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries;

[0028] Figure 2 Schematic diagram of the intelligent predictive control system for immersion liquid cooling of wind and solar energy storage batteries;

[0029] Figure 3 Schematic diagram of an intelligent predictive control system for immersion liquid cooling of wind and solar energy storage batteries;

[0030] Figure 4 This is a bar chart showing the distribution of cooling and heat exchange load control values ​​under various operating conditions.

[0031] In the diagram, 1. Buffer tank; 2. Electrically controlled valve; 3. Gas collection tank; 4. Condenser; 5. Sealed enclosure; 6. Energy storage battery; 7. Cooling coil; 8. Control center; 9. Photovoltaic power generation system; 10. Wind power generation system; 11. Bidirectional pump. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. As those skilled in the art will understand, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Please see Figures 1-4This invention provides a technical solution: a method and system for intelligent predictive control of immersion liquid cooling for wind and solar energy storage batteries, comprising the following steps: S1, real-time acquisition of multi-source data on wind and solar energy storage liquid cooling, data preprocessing of the multi-source data, and construction of a heat generation trend prediction model based on the multi-source data; S2, based on the preprocessed multi-source data and the heat generation trend prediction model, regulation of the working fluid pressure through nonlinear calculation and temperature compensation, and real-time adjustment of the rotational speed of the bidirectional pump 11 and the working pressure of the condenser 4; S3, quantification of the cooling heat exchange demand of the energy storage battery 6 based on the preprocessed multi-source data and the heat generation trend prediction model, adjustment of the cooling equipment according to the cooling heat exchange demand, and feedforward response of the heat load; S4, real-time monitoring of the multi-source data on wind and solar energy storage liquid cooling during the adjustment process, determination of the opening degree of the electronically controlled valve 2, adjustment of the coolant flow rate according to the opening degree of the electronically controlled valve 2, and optimization and updating of the heat generation trend prediction model parameters based on the deviation between the pressure regulation, cooling heat exchange demand, and the opening degree of the electronically controlled valve 2.

[0034] Specifically, the real-time acquisition of multi-source data from wind-cooled storage and liquid cooling systems, and the data preprocessing operation of this multi-source data, are as follows: Real-time acquisition of multi-source data from wind-cooled storage and liquid cooling systems involves the following acquisition process: Powered by the photovoltaic power generation system 9 and the wind power generation system 10, the data is collected in real time using smart meters and various types of sensors, including photovoltaic power generation, wind power generation, battery charging and discharging power and battery temperature of the energy storage battery 6, cooling circuit outlet temperature, cooling circuit inlet temperature, coolant mass flow rate, sealed tank pressure, buffer tank pressure, working fluid level height, and real-time working fluid liquid phase density. Among these, smart meters are used to accurately acquire current and voltage from each circuit to achieve automatic power statistics. Various types of sensors include thermocouples, PT100 platinum resistance thermometers, pressure transmitters, flow meters, ultrasonic level gauges, and high-precision density meters, enabling real-time sensing of multiple parameters under all operating conditions of the liquid cooling system. Simultaneously, the specific heat capacity of the coolant and the standard density of the working fluid are collected; the battery charging and discharging power and battery temperature data are used concurrently, and a battery energy balance algorithm based on battery efficiency is employed to calculate the battery's heat generation power. This calculation, based on the difference between the input / output power and heat loss during battery charging and discharging, and referencing coulombic efficiency and energy efficiency, reflects the real-time rate at which battery electrochemical reactions and internal resistance losses are converted into heat energy. Furthermore, the rate of change of battery heat generation power is calculated based on a sliding time window, reflecting the dynamic characteristics of heat load changes. Satellite remote sensing data for the next ten minutes is collected, including light intensity, wind speed, wind direction, ambient temperature, and ambient humidity. This satellite remote sensing data is accessed via a meteorological satellite data API interface, enabling prediction of the impact of external weather changes on power generation. The impact of cooling conditions is assessed to improve the foresight of prediction and control. Sequence integrity verification and extreme anomaly identification are used to detect missing values ​​and perform linear interpolation to complete multi-source data of wind-cooled liquid cooling systems. Time alignment and resampling are then performed to standardize the sampling interval. Sliding window normalization and standard score normalization methods are used to standardize and dimensionlessly normalize the multi-source data of wind-cooled liquid cooling systems, transforming it into dimensionless variables. Median filtering is then used to remove abnormal peaks and noise from sensors. A multi-source data database for wind-cooled liquid cooling systems is established, storing both raw and preprocessed data. This database supports high-concurrency writing and multi-condition index queries, ensuring system data traceability, analyzability, and modeling efficiency.

[0035] like Figure 3The diagram shows the schematic of an intelligent predictive control system for immersion liquid cooling in a wind-solar energy storage battery. It illustrates the main structure of the system and the collaborative relationships between its core components. The system integrates various new energy sources and energy storage equipment, and constructs a high-efficiency thermal management loop within a sealed enclosure 5, achieving deep integration of wind-solar energy storage and intelligent liquid cooling control. The photovoltaic power generation system 9 and the wind power generation system 10 are the main energy supply units for renewable energy, collecting electricity from solar and wind power respectively and feeding it into the energy storage battery 6. The energy storage battery 6 is entirely housed within the sealed enclosure 5, and thermal management is achieved through liquid cooling. The coolant flows through the cooling coil 7, which is responsible for uniform heat exchange around the battery. The circulation control system consists of a buffer tank 1, an electrically controlled valve 2, a gas collection tank 3, and a bidirectional pump 11. The buffer tank 1 balances the pressure, the electrically controlled valve 2 regulates the coolant flow rate, and the bidirectional pump 11 adjusts the flow rate according to system predictions and real-time operating conditions. The condenser 4 condenses the gas in the working fluid, enabling working fluid condensation and recovery, ensuring closed-loop operation of thermal management, and further improving liquid cooling efficiency. The gas collecting tank 3 achieves gas-liquid separation, preventing air bubbles from affecting heat exchange performance. The control center 8 is responsible for collecting and processing data from various sensors, and realizing real-time predictive control based on AI algorithms, dynamically adjusting pump speed, valve opening, pressure setting, condenser 4 working pressure, and various control parameters. In the entire system, the signal channels drawn by dashed lines and the energy channels drawn by solid lines are closely linked to ensure data interaction and command transmission between various links, achieving efficient collaboration.

[0036] In this implementation plan, high-precision, end-to-end acquisition and preprocessing of multi-source data from wind-cooled liquid storage systems is achieved through multiple types of intelligent sensors and energy balance algorithms, improving data integrity and consistency. After normalization, standardization, and noise removal, data quality and adaptability to abnormal operating conditions are enhanced, providing solid and reliable data support for subsequent intelligent prediction and adaptive control.

[0037] Specifically, the process of constructing a heat generation trend prediction model based on multi-source data of wind-storage liquid cooling is as follows: Historical photovoltaic power generation, wind power generation, battery charging and discharging power, battery temperature, battery heat generation power change rate, and corresponding light intensity, wind speed, wind direction, ambient temperature, and ambient humidity are combined to construct a heat generation feature dataset. This dataset uses ordered time series of various operating condition variables and environmental variables as model input features to form a high-dimensional dynamic feature matrix, enhancing the model's ability to perceive the overall operational situation. The heat generation feature dataset is then trained using a long short-term memory (LSTM) network deep time series modeling algorithm. LTM is a type of recurrent neural network suitable for nonlinear, time-varying, and multivariate time series prediction, possessing forget gate, input gate, and output gate structures. This method effectively captures long-term dependencies and abrupt changes in historical characteristics, improving prediction accuracy under complex operating conditions. The training process employs a sliding window input method, combined with forward and backward propagation algorithms, to iteratively optimize model parameters and minimize prediction errors. A heat generation trend prediction model is constructed, outputting the predicted change rate of battery heat generation power for future energy storage battery 6 in real time, enabling forward-looking and continuous judgment of battery heat generation conditions. The original and normalized predicted change rates of battery heat generation power, along with time labels, are stored in a multi-source data database of wind-cooled and liquid-cooled systems. The normalized predicted change rate of battery heat generation power facilitates horizontal comparison between different operating conditions and models, while the time labels ensure traceability and accountability, guaranteeing the operability of data management and subsequent optimization and control.

[0038] In this implementation plan, by integrating multi-source operating condition and environmental data and employing advanced time-series modeling technology using long short-term memory networks, accurate and dynamic prediction of the heating trend of the energy storage battery 6 is achieved, significantly improving the model's perception and response capabilities to heating changes under complex operating conditions. The real-time output of the predicted change rate of battery heating power not only provides a reliable feedforward basis for pressure regulation and heat exchange optimization but also supports the forward-looking and adaptive capabilities of intelligent temperature control. Normalization and time-stamping further enhance the comparability and traceability of the data, laying a solid data foundation for subsequent system self-learning optimization and full-process intelligent operation and maintenance.

[0039] Specifically, based on the preprocessed multi-source data of wind-storage liquid cooling and the heat generation trend prediction model, the specific process of regulating the working fluid pressure through nonlinear calculation and temperature compensation is as follows: real-time acquisition of multi-source data of wind-storage liquid cooling and the predicted change rate of battery heat generation power; based on the sliding time window, statistical analysis of battery heat generation power and calculation of the average value to obtain the battery heat generation power benchmark; the sliding time window refers to setting a fixed length window on a continuous data sequence, sliding forward one sampling point each time, dynamically calculating the average value of battery heat generation power within the window, which is used to eliminate instantaneous fluctuations and obtain the steady-state benchmark of heat generation power. Simultaneously, atmospheric pressure is used as the initial reference pressure value, and room temperature is used as the target setpoint for battery temperature. The ratio of the predicted change rate of battery heat generation power to the baseline battery heat generation power is added to a constant, and a natural logarithmic operation is performed to obtain the heat generation prediction correction value. The natural logarithmic operation can effectively amplify abnormal responses with large deviations while suppressing small disturbances, thus improving the sensitivity to changes in heat generation trends. The temperature deviation value is obtained by subtracting the target setpoint for battery temperature from the current battery temperature. The temperature deviation is the direct reference for temperature control adjustment, reflecting the direct impact of real-time heat load on heat dissipation requirements. The boiling heat dissipation pressure control value is obtained by subtracting the product of the heat generation prediction correction value and the heat generation prediction sensitivity weighting factor from the initial reference pressure value, and by subtracting the product of the temperature deviation value and the temperature compensation weighting factor. The heat generation prediction sensitivity weighting factor and the temperature compensation weighting factor are both dynamically fitted by an adaptive algorithm, which can continuously optimize the control sensitivity based on historical operating conditions. The final output boiling heat dissipation pressure control value serves as the basis for setting the working fluid pressure, guiding the adjustment of the cooling equipment's operation, and achieving dynamic and precise adjustment of the boiling heat exchange capacity.

[0040] The specific formula for the boiling heat dissipation pressure control value is as follows:

[0041] ;

[0042] In the formula, This represents the boiling heat dissipation pressure control value, which is used to calculate and adjust the working fluid pressure setpoint at the next moment based on the future heating trend of the battery and the current temperature state, so as to achieve dynamic and precise matching between the boiling heat dissipation intensity of the immersion liquid cooling system and the battery heat load, thereby preventing the battery from overheating, improving heat dissipation efficiency and optimizing energy consumption. This represents the initial reference pressure value, which is 1.0 × 10⁻⁶ at atmospheric pressure. 5 Pa, as the starting point for pressure regulation; It represents the predicted rate of change of battery heating power, reflecting the trend of battery load change in the short term, and serves as a feedforward signal for early regulation. This represents the battery heating power benchmark, used to normalize the current predicted heating power, so that the adjustment amount is adapted to the actual heat load characteristics and to eliminate the interference caused by absolute values. This indicates the current battery temperature, reflecting the current heat dissipation status and load level. This indicates the target setpoint for battery temperature, which is 25°C at room temperature. It is the ideal temperature value set in the design and serves as a control target for balancing safety and energy efficiency. This indicates the heating prediction correction value. Based on the ratio of the predicted future heating power to the baseline value, the pressure is actively reduced to enhance boiling intensity in response to the upcoming increase in heating. The logarithmic form can suppress the nonlinear effects of extreme fluctuations, making the adjustment smoother and the response more sensitive. This indicates the temperature deviation value, which monitors the difference between the current battery temperature and the target battery temperature setting in real time. If the actual battery temperature is higher than the target battery temperature setting, the pressure is further reduced to achieve rapid response and risk compensation; conversely, the pressure adjustment range is appropriately reduced to prevent excessive heat dissipation. The heat generation prediction sensitivity weighting factor is determined by collecting historical battery heat generation power prediction change rate and actual battery temperature response data. The goal is to minimize the actual battery temperature deviation, achieve the fastest heat exchange response, and avoid overshoot. The minimum mean square error regression algorithm is used to fit the heat generation prediction sensitivity weighting factor, with a value range between 0.5 and 2. The temperature compensation weighting factor is determined by fitting the data of historical battery temperature and target battery temperature setting values ​​using the least squares method, with the objective of minimizing the residual of temperature deviation. The value ranges from 0.1 to 1.

[0043] In this implementation scheme, a dual control mechanism of heat generation prediction correction and temperature deviation compensation is constructed to achieve dynamic and precise adjustment of the working fluid pressure. This mechanism can sensitively respond to changes in battery heat load and temperature deviation, effectively improving the adaptive capability to sudden heat loads. Nonlinear calculations enhance the identification and control of abnormal operating conditions, while adaptive optimization of weighting factors ensures a balance between adjustment sensitivity and robustness, thus improving the overall intelligence level of boiling heat exchange pressure control.

[0044] Specifically, the process of real-time adjustment of the bidirectional pump 11 speed and the condenser 4 working pressure is as follows: The control center 8 calculates the boiling heat dissipation pressure control value in real time, writes it into the air-cooled liquid-cooled multi-source data database, and controls the speed of the bidirectional pump 11 and the working pressure of the condenser 4 according to the boiling heat dissipation pressure control value, adjusting the working fluid pressure in the sealed box 5, and cooperating with the gas collection tank 3 to replenish and recover gas. The gas collection tank 3 can automatically replenish and absorb excess gas to maintain the stability of the box pressure and liquid level, avoiding drastic pressure fluctuations that affect the boiling heat exchange effect; at the same time, the actual sealed box pressure is fed back, and when the actual sealed box pressure and the boiling heat dissipation pressure are adjusted, the working fluid pressure in the sealed box 5 is adjusted. When the deviation of the thermal pressure control value exceeds the deviation threshold, the backup pressure regulating unit is switched on. The backup pressure regulating unit can be a backup pressure regulating valve to ensure safe operation even if the main pressure regulating unit fails or the regulation fails. When the standard deviation of the working fluid level is detected to be greater than the fluctuation threshold, the boiling heat dissipation pressure control value is increased. The standard deviation of the liquid level reflects the fluctuation of the gas-liquid interface. Increasing the pressure can quickly suppress abnormal fluctuations and restore a stable boiling heat exchange state. The initial reference pressure value and the target battery temperature setting value are corrected periodically based on the deviation between the actual sealed box pressure and the boiling heat dissipation pressure control value, using sliding window data statistics and an adaptive minimum mean square error algorithm.

[0045] In this implementation scheme, by real-time intelligent adjustment of the rotational speed of the bidirectional pump 11 and the pressure of the condenser 4, combined with the gas supply and pressure feedback of the gas collecting tank 3, precise dynamic control of the working fluid pressure is achieved, effectively improving the cooling circuit's adaptability to load fluctuations and abnormal disturbances. Through real-time monitoring and adaptive parameter optimization of the liquid level standard deviation and pressure deviation, the system ensures efficient and stable boiling heat exchange under different operating conditions, enhancing the system's safety, reliability, and intelligent level of long-term operation.

[0046] Specifically, based on the preprocessed multi-source data of wind-cooled and liquid-cooled storage and the heat generation trend prediction model, the specific process for quantifying the cooling and heat exchange requirements of energy storage battery 6 is as follows: Obtain the predicted change rate of battery heat generation power, current sealed enclosure pressure, battery temperature, and temperature deviation value; combine historical battery temperature and temperature deviation values, and obtain battery charging / discharging power and battery heat generation power data; use the first-order inertial model least squares method to fit the first-order delay response curve of battery temperature to calculate the thermal inertia time constant. The first-order inertial model describes the response of temperature to power disturbances, and the least squares method automatically identifies the optimal thermal inertia time constant by fitting historical battery temperature and battery heat generation power data curves, reflecting the hysteresis characteristics of temperature response; based on historical sliding time windows, statistically analyze battery temperature and battery charging / discharging power, select intervals where the battery temperature change amplitude is continuously greater than the time threshold and the change direction is consistent, accumulate the battery charging / discharging power within the interval, calculate the interval length, and calculate the battery charging / discharging power. The ratio of the accumulated power to the interval length yields the battery heat capacity, which represents the heat absorbed per unit temperature difference. The accumulated power reflects the input energy, and the interval length is the time span. The ratio of the battery heat capacity to the thermal inertia time constant yields the thermal inertia coefficient. Multiplying the thermal inertia coefficient by the temperature deviation value yields the temperature difference-driven thermal correction. The thermal inertia coefficient reflects the temperature response speed, and multiplying it by the temperature deviation value quantifies the corrected energy demand under the current temperature difference. The absolute value of the temperature deviation value is divided by the temperature response adjustment weight factor, and the negative value is used as the exponent to perform natural exponential calculation to obtain the exponential correction value. Subtracting the exponential correction value from the constant value yields the nonlinear response correction factor, achieving nonlinear adjustment with rapid response to large temperature differences and smooth decay to small temperature differences. Multiplying the temperature difference-driven thermal correction by the nonlinear response correction factor and adding it to the predicted change rate of battery heating power yields the predicted cooling demand value. Multiplying the predicted cooling demand by the comprehensive correction weight factor yields the cooling heat exchange load control value.

[0047] The specific formula for the cooling heat exchange load control value is as follows:

[0048] ;

[0049] In the formula, This represents the cooling heat exchange load control value, which is used to comprehensively consider the future heating trend of the battery and the current temperature deviation, and dynamically set the heat flow that the cooling system needs to remove. This not only realizes the feedforward response to the upcoming heat load, but also compensates for the current temperature deviation and thermal inertia in real time, thereby ensuring that the battery pack temperature is always in a safe and efficient range. It represents the predicted rate of change of battery heat generation power, predicting the future rate of change of battery heat generation power, serving as an advanced reference for allocating cooling power in advance, and enabling proactive response to the risk of impending temperature rise; Indicates the battery's heat capacity; Represents the thermal inertia time constant. It represents the thermal capacity inertia coefficient, quantifies the thermal buffering capacity and response speed, and reflects the actual impact of the overall heat absorption or release of the battery pack on temperature changes; This indicates the current battery temperature, reflecting the current heat dissipation status and load level. This indicates the target setpoint for battery temperature, which is 25°C at room temperature. It is the ideal temperature value set in the design and serves as a control target for balancing safety and energy efficiency. This represents the temperature deviation value, the difference between the current actual battery temperature and the set target temperature. It is an important feedback signal for driving cooling adjustment, reflecting whether cooling needs to be strengthened or reduced. This represents a nonlinear response correction factor, which nonlinearly weights the temperature deviation value, so that the cooling response is enhanced when the temperature difference is larger and automatically weakened when the temperature difference is smaller, achieving a smoother and more sensitive adaptive adjustment. This represents the temperature response adjustment weight factor, which is based on historical data of actual battery temperature, target setpoint of battery temperature, and historical data of battery charging and discharging power and temperature control response. Using the minimum mean square error method, the temperature response adjustment weight factor with the smallest mean square error is taken as the current optimal value, with a value range between 0.1 and 2.0. The comprehensive correction weighting factor is determined by collecting actual heat exchange volume and cooling heat exchange load control values, performing a univariate linear regression over the historical operating range, and the slope of the regression is the comprehensive correction weighting factor, with a value range between 0.8 and 1.2.

[0050] In this example, Table 1 shows the cooling heat exchange load control values. The comprehensive correction weighting factor was set to 0.95, the temperature response adjustment weighting factor to 0.2, the target battery temperature setpoint to 25, and the thermal inertia time constant to 550. The table details the predicted change rate of battery heating power, battery heat capacity, current battery temperature, and cooling heat exchange load control values ​​under five different operating conditions. Specifically, for condition I, the predicted change rate of battery heating power was 180, the battery heat capacity was 6500, the current battery temperature was 36, and the cooling heat exchange load control value was 294.50; for condition II, the predicted change rate of battery heating power was 120, and the battery heat capacity was 7000. The current battery temperature is 32°C, and the cooling heat exchange load control value is 198.64. For Condition III, the predicted change rate of battery heat generation power is 60%, the battery heat capacity is 6700 Nm³ / h, the current battery temperature is 30°C, and the cooling heat exchange load control value is 114.86. For Condition IV, the predicted change rate of battery heat generation power is 100%, the battery heat capacity is 7100 Nm³ / h, the current battery temperature is 29°C, and the cooling heat exchange load control value is 144.05. For Condition V, the predicted change rate of battery heat generation power is 200%, the battery heat capacity is 6600 Nm³ / h, the current battery temperature is 38°C, and the cooling heat exchange load control value is 338.20.

[0051] Table 1. Cooling and heat exchange load control values.

[0052]

[0053] like Figure 4 The image shows a bar chart illustrating the distribution of cooling and heat exchange load control values ​​under five different operating conditions. The horizontal axis represents the different operating condition numbers, and the vertical axis represents the cooling and heat exchange load control values ​​under each condition. This is based on Table 1 and... Figure 4 It can be seen that there are significant differences in the cooling heat exchange load control values ​​under different operating conditions. The cooling heat exchange load control values ​​for conditions I and V are much higher than those for other conditions, indicating that the battery's heat load demand is higher under these two conditions, requiring stronger cooling capacity. Conditions III and IV have lower cooling demands, resulting in the lowest cooling heat exchange load control values. Overall, the cooling load control values ​​are influenced by multiple coupled factors, exhibiting a clear dynamic distribution, which is beneficial for achieving differentiated and precise thermal management control.

[0054] This implementation scheme achieves dynamic and precise quantification of cooling and heat transfer requirements by integrating real-time prediction and historical identification, and comprehensively utilizing multi-dimensional information such as battery heat generation power change rate, temperature deviation, thermal inertia time constant, thermal capacity inertia coefficient, and battery thermal capacity. Employing a first-order inertial model and nonlinear response correction, it accurately reflects the actual impact of battery temperature changes on cooling requirements, improving adaptability to sudden heat loads and complex operating conditions. The adaptive weight correction mechanism enhances the flexibility and robustness of the control strategy, providing a more scientific and intelligent adjustment basis for liquid cooling.

[0055] Specifically, the process of adjusting the cooling equipment according to the cooling and heat exchange requirements and making a feedforward response to the heat load is as follows: the control center 8 adjusts the power of the cooling coil 7 according to the cooling and heat exchange load control value, thereby affecting the overall heat exchange capacity and ensuring the flexibility of response to different heat load conditions. Simultaneously, the inlet and outlet temperatures of the cooling circuit, the mass flow rate of the coolant, and the specific heat capacity of the coolant are acquired in real time to calculate the actual heat exchange. If there is a deviation between the actual heat exchange and the cooling heat exchange load control value, the speed of the bidirectional pump 11 and the power of the cooling coil 7 are corrected according to the deviation. That is, through closed-loop control, the measured values ​​are compared with the set target. If the deviation is greater than the tolerance threshold, the speed of the bidirectional pump 11 and the power of the cooling coil 7 are automatically adjusted to achieve real-time adaptive adjustment. When the rate of change of battery heating power is detected to be greater than the sudden change threshold, an emergency cooling strategy is activated: the speed of the maximum bidirectional pump 11 is increased, the power output of the cooling coil 7 is increased, and redundant cooling circuits are enabled. The cooling heat exchange load control value and the actual heat exchange are written into the multi-source data database of air-cooled storage and liquid cooling for real-time statistical analysis. The cooling heat exchange load control value parameters are optimized through minimum mean square error and sliding window self-learning algorithms. The minimum mean square error algorithm minimizes the error between the prediction and the actual data and adaptively fine-tunes the control parameters, while the sliding window self-learning algorithm continuously updates the parameter weights to improve the model generalization and the intelligence level of system operation.

[0056] In this implementation scheme, the power of the cooling coil 7 and the speed of the bidirectional pump 11 are driven by the cooling heat exchange load regulation value, achieving flexible and dynamic response to different heat load conditions and accurately adapting to real-time changes in heat load. Through closed-loop control and adaptive parameter optimization, the deviation between the predicted value and the actual heat exchange volume is effectively reduced, improving the accuracy and robustness of the regulation. The construction of an emergency cooling strategy further enhances the ability to cope with sudden high heat loads, ensuring the safety and reliability of the battery thermal management system. Overall, it provides solid technical support for energy storage to achieve efficient, intelligent, and self-learning cooling regulation.

[0057] Specifically, the process of real-time monitoring and adjustment of multi-source data on air-cooled liquid storage to determine the opening degree of the electronically controlled valve 2 is as follows: During the adjustment of boiling heat dissipation pressure and cooling heat exchange load, the pressure of the sealed box and the pressure of the buffer tank are acquired in real time, and the difference between the pressure of the sealed box and the pressure of the buffer tank is calculated to obtain the pressure deviation value, which reflects the pressure gradient and can sensitively reflect the flow state and distribution of the cooling circuit. The pressure deviation value is multiplied by the pressure deviation weighting factor to obtain the pressure adjustment term; the real-time density of the working fluid liquid phase and the standard density of the working fluid are acquired, and the difference between the real-time density of the working fluid liquid phase and the standard density of the working fluid is calculated to obtain the density deviation value. The real-time liquid density is acquired from an online density meter. The standard density of the working fluid serves as a reference value under normal operating conditions. The density deviation reflects the gas-liquid phase changes and mixing degree of the working fluid, significantly impacting boiling stability. The density deviation is multiplied by a density deviation weighting factor to obtain the density adjustment term. The pressure and density adjustment terms are summed, and the negative value of the sum is used as the exponent for natural exponential calculation to obtain the comprehensive adjustment index factor. This achieves a nonlinear normalized response, enhancing the valve's adaptability and robustness to dynamic changes. The comprehensive adjustment index factor is added to a constant and its reciprocal is taken to obtain the valve opening control value. The valve opening control value is a continuously adjustable dimensionless parameter, directly used as the target opening signal input for the solenoid valve 2, enabling high-precision intelligent control of coolant flow and pressure.

[0058] The specific formula for the valve opening control value is as follows:

[0059] ;

[0060] In the formula, This represents the valve opening control value, used for the adaptive adjustment of the opening of the cooling solenoid valve 2. By comprehensively feeding back the dual deviations of the sealing box pressure and buffer tank pressure, and the actual density and standard density of the working fluid, and performing Sigmoid calculation after weight modulation, it outputs a continuous opening command from 0 to 1, realizing the engineering closed-loop characteristics of automatic convergence when the deviation is too large and linear adjustment when the deviation is small, thus achieving closed-loop precise control of cooling flow, pressure and medium state. It represents the pressure inside the sealed enclosure, reflecting the actual pressure inside the enclosure at present, and is an important physical quantity for overall safety and cooling performance; This indicates the pressure in the buffer tank, reflecting the target pressure level of the entire liquid cooling circuit; This indicates the real-time density of the working fluid liquid phase, and represents the actual density of the current coolant, affecting cooling capacity and phase stability. It represents the standard density of the working fluid, the required target density of the working fluid, and is used to quantify density deviation. It is a fixed constant obtained according to the type of working fluid and the working conditions. This represents the pressure regulation term, quantifying the degree of deviation between the actual pressure and the target pressure. This represents the density adjustment term, quantifying the degree of deviation between the actual working fluid density and the target density. This is used to combine the two adjustment components as a power of the negative natural exponent, reflecting the dual-channel attenuation effect on pressure and density over-limits. The pressure deviation weighting factor is determined by using the recursive least squares method to dynamically minimize the residual between the actual opening of the electric valve 2 and the valve opening control value, based on the historical pressure deviation value and the actual opening of the electric valve 2. The optimal pressure deviation weighting factor is fitted, and its value ranges from 0.1 to 10. The density deviation weighting factor is determined by using the recursive least squares method to dynamically minimize the residual between the actual opening of the electric valve 2 and the valve opening control value, based on the historical density deviation value and the actual opening of the electric valve 2. The optimal density deviation weighting factor is fitted, and its value ranges from 0.1 to 10.

[0061] In this implementation scheme, through real-time comprehensive analysis and nonlinear normalization of pressure and density deviations, high-precision, adaptive adjustment of the opening degree of the electronically controlled valve 2 is achieved. This allows for sensitive detection of the flow state and working fluid gas-liquid phase changes in the cooling circuit, dynamically matching the coolant flow rate and pressure. The continuously adjustable valve opening parameters effectively enhance the adaptability to complex operating conditions and abnormal disturbances, ensuring the accuracy, stability, and operational safety of the cooling system adjustment. This provides a solid guarantee for the efficient heat exchange and long-term reliable operation of the energy storage battery liquid cooling system.

[0062] Specifically, the process of adjusting the coolant flow rate based on the opening degree of the electrically controlled valve 2, and simultaneously optimizing and updating the parameters of the heat generation trend prediction model based on the deviation between pressure regulation, cooling heat exchange requirements, and the opening degree of the electrically controlled valve 2, is as follows: The control center 8 generates a valve regulation command based on the valve opening regulation value and sends it to the electrically controlled valve 2 in real time to adjust the opening degree of the electrically controlled valve 2 and regulate the coolant flow rate; if the continuously collected pressure deviation value and density deviation value are detected to exceed the pressure deviation threshold and density deviation threshold respectively, the maximum valve opening limit is triggered, and an audible and visual alarm is activated. The maximum opening limit can prevent abnormal overpressure and uncontrolled gas-liquid separation, while the audible and visual alarm realizes the function of the valve. The system provides synchronized early warning both locally and remotely, improving the timeliness of fault response. It writes valve opening control values, valve control commands, pressure deviation values, and density deviation values ​​into the multi-source data database for air-cooled liquid cooling in real time. All control commands and feedback parameters are fully stored, supporting post-event traceability, trend analysis, and data-driven optimization. Combined with the opening adjustment process of the electrically controlled valve 2, it evaluates the adjustment sensitivity, response speed, and energy efficiency, and periodically adjusts the standard density of the working fluid. Adjustment sensitivity and response speed are achieved by analyzing the time delay and magnitude of change between the actual valve action and the coolant mass flow rate, while energy efficiency is evaluated based on the heat exchange per unit time and energy consumption ratio. The system continuously acquires multi-source data on air-cooled liquid cooling during operation, including boiling heat dissipation pressure control values, cooling heat exchange load control values, and valve opening control values. It monitors the deviation between the control values ​​at each stage and the corresponding actual feedback data, and uses recursive least squares and minimum mean square error adaptive algorithms to correct the parameters of the heating trend prediction model. Recursive least squares is an online parameter estimation algorithm that can quickly update the parameters of the fever trend prediction model when new data arrives. The minimum mean square error algorithm continuously iterates to minimize the error between the model output and the actual observation, ensuring that the fever trend prediction model maintains high accuracy and self-learning ability under long-term operation and external disturbances.

[0063] In this implementation scheme, by continuously and adaptively adjusting the opening of the electronically controlled valve 2, combined with real-time monitoring and intelligent alarm of pressure and density deviations, the accuracy, sensitivity, and safety of coolant flow control are greatly improved. Simultaneously, multi-source data and feedback results from the control process are dynamically written into the database, and the parameters of the heating trend prediction model are continuously optimized using recursive least squares and minimum mean square error algorithms. This gives the system excellent self-learning and adaptive capabilities, enabling it to cope with complex operating conditions and environmental changes, and achieving efficient, stable, and intelligent closed-loop management of the entire cooling system process.

[0064] Reference Figure 2As shown, the second aspect of the present invention provides an intelligent predictive control system for immersion liquid cooling of wind and solar energy storage batteries, applied to the aforementioned intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries. The system includes: a multi-source data acquisition and intelligent prediction module, used to acquire multi-source data on wind and solar energy storage liquid cooling in real time, perform data preprocessing on the multi-source data, and construct a heating trend prediction model based on the multi-source data; and a boiling heat dissipation pressure active regulation module, used to regulate the working fluid pressure based on the preprocessed multi-source data and the heating trend prediction model through nonlinear calculations and temperature compensation, and to adjust the bidirectional pressure in real time. Pump 11 speed and condenser 4 working pressure; cooling power adaptive adjustment module, used to quantify the cooling heat exchange demand of energy storage battery 6 based on pre-processed air-storage liquid cooling multi-source data and heat generation trend prediction model, adjust the cooling equipment according to the cooling heat exchange demand, and perform feedforward response of heat load; working fluid gas-liquid mixing and self-learning fine-tuning module, used to monitor the air-storage liquid cooling multi-source data in real time during the adjustment process, determine the opening degree of electric control valve 2, adjust the coolant flow rate according to the opening degree of electric control valve 2, and optimize and update the heat generation trend prediction model parameters according to the deviation between pressure regulation, cooling heat exchange demand and the opening degree of electric control valve 2.

[0065] This implementation scheme achieves data-driven, dynamic, adaptive, and closed-loop intelligent control throughout the entire process through the organic synergy of multi-source data acquisition, intelligent prediction, active pressure regulation, adaptive cooling power adjustment, and working fluid gas-liquid blending and self-learning fine-tuning. The tightly integrated modules can accurately sense and predict changes in battery thermal load, optimize cooling pressure and power regulation in real time, dynamically adjust the working fluid gas-liquid state, and continuously iterate and optimize the prediction model parameters. This comprehensively improves the operational intelligence, safety, and energy efficiency of the energy storage battery liquid cooling system, effectively ensuring efficient cooling and long-term stable operation under complex conditions.

[0066] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0067] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. As those skilled in the art will understand, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for intelligent predictive control of immersion liquid cooling for wind and solar energy storage batteries, characterized in that, Includes the following steps: S1, collect multi-source data of wind-cooled liquid storage in real time, perform data preprocessing on the multi-source data of wind-cooled liquid storage, and build a heat generation trend prediction model based on the multi-source data of wind-cooled liquid storage. S2, based on the pre-processed multi-source data of air-cooled liquid storage and the heat generation trend prediction model, the working fluid pressure is regulated by nonlinear calculation and temperature compensation, and the speed of the bidirectional pump (11) and the working pressure of the condenser (4) are adjusted in real time. The specific process of regulating the working fluid pressure through nonlinear calculations and temperature compensation based on preprocessed multi-source data of air-cooled liquid storage and a heat generation trend prediction model is as follows: Real-time acquisition of multi-source data on wind-cooled and liquid-cooled systems and prediction of battery heating power change rate; based on a sliding time window, statistical analysis of battery heating power and calculation of average value to obtain battery heating power benchmark; simultaneously using atmospheric pressure as initial reference pressure value and ambient temperature as target setpoint for battery temperature; The ratio of the predicted change rate of battery heating power to the baseline of battery heating power is added to a constant and the natural logarithm is used to obtain the heating prediction correction value; the temperature deviation value is obtained by subtracting the target battery temperature value from the current battery temperature; the boiling heat dissipation pressure control value is obtained by subtracting the product of the heating prediction correction value and the heating prediction sensitivity weighting factor from the initial reference pressure value, and by subtracting the product of the temperature deviation value and the temperature compensation weighting factor. S3, based on the pre-processed wind-cooled liquid-cooled multi-source data and the heat generation trend prediction model, quantify the cooling heat exchange requirements of the energy storage battery (6), adjust the cooling equipment according to the cooling heat exchange requirements, and carry out the feedforward response of the heat load. The specific process of quantifying the cooling and heat exchange requirements of the energy storage battery (6) based on the pre-processed multi-source data of wind-cooled liquid storage and the heat generation trend prediction model is as follows: The predicted rate of change of battery heat generation power, current sealed box pressure, battery temperature and temperature deviation value are obtained; combined with historical battery temperature and temperature deviation value, and battery charging and discharging power and battery heat generation power data, the thermal inertia time constant is calculated by fitting the first-order delay response curve of battery temperature using the least squares method of the first-order inertial model. Based on historical sliding time windows, the battery temperature and battery charging and discharging power are statistically analyzed. Intervals in which the change in battery temperature is greater than the time threshold for a continuous period of time and the change direction is consistent are selected. The battery charging and discharging power within the interval is accumulated, and the interval length is calculated. The ratio of the accumulated value of battery charging and discharging power to the interval length is used to obtain the battery heat capacity. The ratio of battery heat capacity to thermal inertia time constant is calculated to obtain the heat capacity inertia coefficient. The heat capacity inertia coefficient is multiplied by the temperature deviation value to obtain the temperature difference driven thermal correction amount. The absolute value of the temperature deviation is divided by the temperature response adjustment weighting factor and the negative value is taken as the exponent. The natural exponent calculation is performed to obtain the exponent correction value. The nonlinear response correction factor is obtained by subtracting the exponent correction value from the constant. Multiply the temperature difference-driven thermal correction amount by the nonlinear response correction factor and add it to the predicted rate of change of battery heating power to obtain the cooling demand prediction value. Multiply the cooling demand prediction value by the comprehensive correction weight factor to obtain the cooling heat exchange load regulation value. S4, real-time monitoring of multi-source data of air-cooled liquid cooling during the adjustment process, judging the opening degree of the electric control valve (2), adjusting the coolant flow rate according to the opening degree of the electric control valve (2), and optimizing and updating the parameters of the heat generation trend prediction model according to the deviation between pressure regulation, cooling heat exchange requirements and the opening degree of the electric control valve (2).

2. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process of real-time acquisition of multi-source data for wind-cooled storage and liquid cooling, and the data preprocessing operation of the multi-source data for wind-cooled storage and liquid cooling is as follows: Real-time acquisition of multi-source data for wind-storage liquid cooling: The acquisition process is as follows: Powered by the photovoltaic power generation system (9) and the wind power generation system (10), the photovoltaic power generation, wind power generation, battery charging and discharging power and battery temperature of the energy storage battery (6), cooling circuit outlet temperature, cooling circuit inlet temperature, coolant mass flow rate, sealed box pressure, buffer tank pressure, working fluid level height and working fluid liquid phase real-time density are collected simultaneously; the specific heat capacity of the coolant and the standard density of the working fluid are also collected. Simultaneously utilizing battery charging and discharging power and battery temperature data, a battery energy balance algorithm based on battery efficiency is used to calculate the battery heating power, and the rate of change of battery heating power is calculated based on a sliding time window; and satellite remote sensing data is used to collect light intensity, wind speed, wind direction, ambient temperature and ambient humidity for the next ten minutes. Missing value detection and linear interpolation were performed on the multi-source data of wind-cooled storage and liquid cooling systems. Time alignment and resampling were also performed to ensure a consistent sampling interval. Sliding window normalization and standard score normalization were used to standardize and dimensionlessly normalize the multi-source data of wind-cooled storage and liquid cooling systems. Median filtering algorithm was used to remove abnormal peaks and noise from the sensors. A multi-source data database of wind-cooled storage and liquid cooling systems was established, and the original and preprocessed multi-source data of wind-cooled storage and liquid cooling systems were stored in the database.

3. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process of constructing a heat generation trend prediction model based on multi-source data of air-cooled liquid storage is as follows: By combining historical photovoltaic power generation, wind power generation, battery charging and discharging power, battery temperature, battery heat generation power change rate, and corresponding light intensity, wind speed, wind direction, ambient temperature and ambient humidity, a heat generation feature dataset is constructed. The heat generation feature dataset is trained by a long short-term memory network deep time series modeling algorithm to construct a heat generation trend prediction model. The predicted change rate of battery heat generation power of the energy storage battery (6) at future time is output in real time. The original and normalized predicted change rates of battery heat generation power with time tags are stored in the wind-storage-liquid cooling multi-source data database.

4. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process of real-time adjustment of the speed of the bidirectional pump (11) and the working pressure of the condenser (4) is as follows: The control center (8) calculates the boiling heat dissipation pressure control value in real time, writes it into the air-cooled liquid multi-source data database, and controls the speed of the bidirectional pump (11) and the working pressure of the condenser (4) according to the boiling heat dissipation pressure control value. It also adjusts the working fluid pressure in the sealed box (5) and cooperates with the gas collection tank (3) to supply and recover gas. At the same time, it feeds back the actual sealed box pressure. When the deviation between the actual sealed box pressure and the boiling heat dissipation pressure control value exceeds the deviation threshold, it switches the standby pressure regulating unit. When the standard deviation of the working fluid level height is detected to be greater than the fluctuation threshold, it raises the boiling heat dissipation pressure control value. The initial reference pressure value and the target battery temperature setting value are periodically corrected based on the deviation between the actual sealed chamber pressure and the boiling heat dissipation pressure adjustment value, using sliding window data statistics and an adaptive minimum mean square error algorithm.

5. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process of adjusting the cooling equipment according to the cooling heat exchange demand and performing a feedforward response to the heat load is as follows: The control center (8) adjusts the power of the cooling coil (7) according to the cooling heat exchange load control value, and at the same time obtains the inlet temperature of the cooling circuit, the outlet temperature of the cooling circuit, the mass flow rate of the coolant and the specific heat capacity of the coolant in real time to calculate the actual heat exchange amount. If there is a deviation between the actual heat exchange amount and the cooling heat exchange load control value, the speed of the bidirectional pump (11) and the power of the cooling coil (7) are corrected according to the deviation. When the rate of change of battery heat generation power is detected to be greater than the sudden change threshold, an emergency cooling strategy is activated: the speed of the maximum bidirectional pump (11) is increased, the power output of the cooling coil (7) is increased, and the redundant cooling circuit is enabled. The cooling heat exchange load control value and the actual heat exchange volume are written into the air-cooled liquid cooling multi-source data database for real-time statistical analysis. The cooling heat exchange load control value parameters are optimized through minimum mean square error and sliding window self-learning algorithms.

6. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process for determining the opening degree of the electronically controlled valve (2) based on the multi-source data of the air-cooled liquid storage system during the real-time monitoring and adjustment process is as follows: During the adjustment of boiling heat dissipation pressure and cooling heat exchange load, the pressure of the sealing box and the pressure of the buffer tank are acquired in real time, and the difference between the pressure of the sealing box and the pressure of the buffer tank is calculated to obtain the pressure deviation value. The pressure deviation value is multiplied by the pressure deviation weighting factor to obtain the pressure adjustment term. The real-time density of the working fluid liquid phase and the standard density of the working fluid are acquired, and the difference between the real-time density of the working fluid liquid phase and the standard density of the working fluid is calculated to obtain the density deviation value. The density deviation value is multiplied by the density deviation weighting factor to obtain the density adjustment term. The pressure adjustment term and the density adjustment term are summed, and the negative value of the summation result is taken as the exponent. Natural exponent calculation is performed to obtain the comprehensive adjustment index factor. The comprehensive adjustment index factor is added to the constant and the reciprocal is taken to obtain the valve opening control value.

7. The intelligent predictive control method for immersion liquid cooling of wind and solar energy storage batteries according to claim 1, characterized in that, The specific process of adjusting the coolant flow rate according to the opening degree of the electronically controlled valve (2), and simultaneously optimizing and updating the parameters of the heat generation trend prediction model based on the deviation between the pressure regulation, cooling heat exchange requirements, and the opening degree of the electronically controlled valve (2) is as follows: The control center (8) generates valve control commands based on the valve opening control value and sends them to the electric control valve (2) in real time to adjust the opening of the electric control valve (2) and regulate the coolant flow rate. If the continuously collected pressure deviation value and density deviation value are detected to exceed the pressure deviation threshold and density deviation threshold, the maximum valve opening limit is triggered and accompanied by an audible and visual alarm. The valve opening control value, valve control command, pressure deviation value and density deviation value are written into the air-cooled liquid cooling multi-source data database in real time. Combined with the opening adjustment process of the electric control valve (2), the adjustment sensitivity, response speed and energy efficiency are evaluated, and the standard density of the working fluid is adjusted periodically. Continuously acquire multi-source data on air-cooled liquid cooling, boiling heat dissipation pressure control value, cooling heat exchange load control value and valve opening control value during operation, monitor the deviation between the control value of each link and the corresponding actual feedback data, and use recursive least squares method and minimum mean square error adaptive algorithm to correct the parameters of the heating trend prediction model.

8. A smart predictive control system for immersion liquid cooling of wind and solar energy storage batteries, employing the smart predictive control method for immersion liquid cooling of solar energy storage batteries as described in any one of claims 1-7, characterized in that, include: The multi-source data acquisition and intelligent prediction module is used to acquire multi-source data of wind-cooled liquid storage in real time, perform data preprocessing operations on the multi-source data of wind-cooled liquid storage, and build a heating trend prediction model based on the multi-source data of wind-cooled liquid storage. The boiling heat dissipation pressure active control module is used to control the working fluid pressure based on the pre-processed air-cooled liquid multi-source data and the heat generation trend prediction model, through nonlinear calculation and temperature compensation, and to adjust the speed of the bidirectional pump (11) and the working pressure of the condenser (4) in real time. The cooling power adaptive adjustment module is used to quantify the cooling heat exchange requirements of the energy storage battery (6) based on the pre-processed wind-cooled liquid-cooled multi-source data and the heat generation trend prediction model, adjust the cooling equipment according to the cooling heat exchange requirements, and perform a feedforward response to the heat load. The working fluid gas-liquid mixing and self-learning fine-tuning module is used to monitor the multi-source data of air-storage liquid cooling in real time during the adjustment process, determine the opening degree of the electric control valve (2), adjust the coolant flow rate according to the opening degree of the electric control valve (2), and optimize and update the parameters of the heat generation trend prediction model according to the deviation between pressure regulation, cooling heat exchange requirements and the opening degree of the electric control valve (2).

Images