An adaptive performance optimization air conditioning control system and method

Abstract

This invention discloses an adaptive performance optimization air conditioning control system and method, relating to the technical field of air conditioning control in energy storage stations. The air conditioning control system includes a data acquisition module, a data analysis and computation module, a hybrid predictive model control module, and a power control module. The hybrid predictive model control module is used to establish a hybrid predictive model based on the dynamic analysis results and parameters of thermal characteristics and to predict the temperature. The power control module is used to control the power of the air conditioner according to the temperature prediction results of the predictive model. This system and method collect and analyze thermal characteristic data, build a hybrid model, dynamically switch the fusion model to accurately predict the temperature, and dynamically adjust the prediction step size in combination with the thermal time constant to match it with the thermal inertia characteristics of the battery pack, further optimizing thermal inertia compensation and extending battery life.

Description

Technical Field

[0001] This invention relates to the technical field of air conditioning control for energy storage stations, and in particular to an adaptive performance optimization air conditioning control system and method. Background Technology

[0002] In energy storage stations, batteries, as the core energy storage unit, are significantly affected by operating temperature in terms of performance and lifespan. During charging, excessively high temperatures can trigger a series of adverse reactions. For example, excessively high temperatures accelerate the rate of internal chemical reactions, leading to increased self-discharge and a reduction in the battery's actual usable capacity. Simultaneously, high temperatures can cause the electrolyte to decompose, generating gas, increasing internal pressure, and even posing safety hazards such as fire or explosion. On the other hand, excessively low temperatures significantly increase the battery's internal resistance, leading to increased energy loss during charging and discharging, a substantial decrease in charging efficiency, and a shortened cycle life.

[0003] Therefore, precise control of battery operating temperature is crucial to ensure efficient, safe, and long-lasting battery operation. This requires the air conditioning control system to possess extremely high control precision to maintain the battery operating within its optimal temperature range and meet the battery's stringent requirements for temperature control accuracy.

[0004] Existing air conditioning control systems for energy storage stations have revealed many critical issues that urgently need to be addressed when dealing with complex working environments, especially low-temperature environments.

[0005] Because existing air conditioning control systems for energy storage stations mostly use linear control models to control temperature, it is difficult to capture nonlinear temperature rise characteristics. At low temperatures, the electrochemical reactions inside the battery change, leading to a significant increase in internal resistance. This increased internal resistance makes the heat generation mechanism during charging and discharging more complex, exhibiting obvious nonlinear characteristics. Traditional temperature prediction models are often based on simple linear assumptions or fixed parameter models, failing to accurately capture this nonlinear trend in battery pack temperature changes caused by low temperatures. This results in significant deviations in battery temperature predictions, making it difficult to meet the precise temperature control requirements of practical applications.

[0006] Based on the aforementioned issues, battery packs possess a certain thermal inertia (in this invention, thermal inertia refers to a lag characteristic exhibited by the battery pack during heat transfer and temperature changes), and its thermal time constant reflects the rate of temperature change within the battery pack. However, in existing air conditioning control systems, even those employing nonlinear predictive models, the prediction step size of these models is often fixed, failing to fully consider the dynamic changes in the battery pack's thermal time constant. This leads to a mismatch between the prediction step size and the actual thermal inertia characteristics of the battery pack, as the thermal inertia is reflected by the thermal time constant, which describes the time required for the battery pack to reach its final steady-state temperature after being subjected to thermal disturbance. When the prediction step size is too long, the control strategy may lag behind the actual changes in battery temperature, failing to adjust the air conditioning in a timely manner to maintain stable battery temperature; conversely, when the prediction step size is too short, it may lead to over-adjustment of the control strategy, resulting in frequent start-ups and shutdowns of the air conditioning, increasing energy consumption and potentially causing additional damage to the air conditioning equipment.

[0007] The above problems are collectively reflected in the insufficient energy efficiency optimization of the air conditioning control system of the energy storage station, the lack of adaptive capability, and the inability to flexibly adjust the operating power of the air conditioner according to the real-time status of the battery and environmental conditions. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides an adaptive performance optimization air conditioning control system and method, which solves the control problem of existing energy storage station air conditioning control systems in dealing with complex working environments.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] On one hand, this invention discloses an adaptive performance optimization air conditioning control system, comprising:

[0011] The data acquisition module is used to dynamically acquire data from multiple sensors;

[0012] The data analysis and calculation module is used for dynamic analysis of thermal properties and parameter calculation.

[0013] The hybrid prediction model control module is used to establish a hybrid prediction model based on the results and parameters of dynamic thermal property analysis and to predict the temperature.

[0014] The power control module is used to control the power of the air conditioner based on the temperature prediction results from the prediction model.

[0015] The hybrid prediction model control module includes:

[0016] The neural network model submodule is used to build and manage LSTM neural network models;

[0017] The Physical Heat Conduction Model submodule is used to create and manage physical heat conduction models;

[0018] The predictive model control submodule is used to control:

[0019] When the ambient temperature is greater than or equal to the first temperature threshold: the temperature is predicted by the LSTM model and the predicted value is corrected by the physical heat conduction model;

[0020] When the ambient temperature is below a first temperature threshold: the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model;

[0021] The prediction model control submodule is also used to control the adaptive adjustment of the prediction step size in temperature prediction of LSTM neural network models or physical heat conduction models.

[0022] On the other hand, this invention discloses an adaptive performance optimization air conditioning control method, including the following steps:

[0023] Dynamically acquire data from multiple sensors;

[0024] Perform dynamic analysis of thermal properties and parameter calculations;

[0025] A hybrid prediction model is established based on the results and parameters of dynamic thermal property analysis to predict temperature.

[0026] Establish an LSTM neural network model;

[0027] Establish a physical heat conduction model;

[0028] When the ambient temperature is greater than or equal to the first temperature threshold: the temperature is predicted by the LSTM model and the predicted value is corrected by the physical heat conduction model;

[0029] When the ambient temperature is below a first temperature threshold: the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model;

[0030] Adaptive adjustment of prediction step size in temperature prediction using LSTM neural network models or physical heat conduction models;

[0031] The power of the air conditioner is controlled based on the temperature prediction results from the prediction model.

[0032] In another aspect, this invention discloses an electronic device, comprising:

[0033] At least one processor; and

[0034] A memory communicatively connected to the at least one processor; wherein,

[0035] The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform the aforementioned adaptive performance optimization air conditioning control method.

[0036] This invention provides an adaptive performance optimization air conditioning control system and method, which has the following beneficial effects: This invention calculates the thermal time constant of the battery pack at regular intervals using different time thresholds to reflect thermal inertia characteristics, calculates the rate of change of internal resistance in real time to identify abnormal battery conditions, and uses heat flow sensor data to calculate the heat exchange coefficient to provide boundary conditions for the prediction model. Through this comprehensive and real-time thermal characteristic data acquisition and analysis, it can accurately grasp the changes in the thermal characteristics of the battery under different operating conditions and establish a hybrid prediction model. This not only effectively captures the nonlinear characteristics of battery temperature changes, but also predicts the theoretical temperature change trend of the battery pack under low-temperature conditions from the perspective of physical principles. In low-temperature environments, the physical heat conduction model can fully consider the heat transfer process between the battery pack and the environment. Importantly, through model switching and fusion, it fully leverages the advantages of the LSTM model in capturing nonlinear characteristics, while cleverly utilizing the advantages of the physical heat conduction model under low-temperature conditions. Through this dynamic model switching and fusion mechanism, it can achieve accurate prediction of battery temperature under different operating conditions. Based on the above effects, it can also dynamically adjust the prediction step size by combining the thermal time constant to match the thermal inertia characteristics of the battery pack, further optimizing thermal inertia compensation and extending battery life. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the modules of the adaptive performance optimization air conditioning control system of the present invention;

[0038] Figure 2 This is a flowchart illustrating the adaptive performance optimization air conditioning control method of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] like Figure 1 This invention discloses an adaptive performance optimization air conditioning control system, which can be implemented in hardware and / or software. The preferred software application of this adaptive performance optimization air conditioning control system can be installed in electronic devices such as terminals or servers.

[0041] This invention discloses an adaptive performance optimization air conditioning control system comprising:

[0042] Data acquisition module 100 is used to dynamically acquire data from multiple sensors;

[0043] Data analysis and calculation module 101 is used for dynamic analysis of thermal properties and parameter calculation;

[0044] The hybrid prediction model control module 102 is used to establish a hybrid prediction model based on the dynamic analysis results and parameters of thermal characteristics and to predict the temperature.

[0045] The power control module 103 is used to control the power of the air conditioner based on the temperature prediction results of the prediction model.

[0046] The hybrid prediction model control module includes:

[0047] Neural network model submodule 1020 is used to build and manage LSTM neural network models;

[0048] The Physical Heat Conduction Model Submodule 1021 is used to establish and manage physical heat conduction models.

[0049] Predictive model control submodule 1022 is used to control:

[0050] When the ambient temperature is greater than or equal to the first temperature threshold: the temperature is predicted by the LSTM model and the predicted value is corrected by the physical heat conduction model;

[0051] When the ambient temperature is below a first temperature threshold: the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model;

[0052] The prediction model control submodule 1022 is also used to control the adaptive adjustment of the prediction step size in temperature prediction of LSTM neural network models or physical heat conduction models.

[0053] Optionally, dynamically acquiring multi-sensor data includes:

[0054] Dynamically acquire data from temperature sensor, humidity sensor, battery internal resistance monitoring, and heat flow sensor;

[0055] It also includes: synchronously collecting real-time battery power at fixed intervals, and recording air conditioner operating parameters at fixed intervals.

[0056] During implementation, dynamic acquisition of multi-sensor data includes sensor array deployment: High-precision temperature sensors (such as Pt100, accuracy ±0.3℃) are installed at the top, middle, bottom and low-temperature areas of the battery pack, and environmental humidity sensors (accuracy ±0.5℃ / ±3% RH), battery internal resistance monitoring modules (accuracy 0.1mΩ) and heat flow sensors (accuracy ±5%) are deployed simultaneously.

[0057] During implementation, when the system initiates a data acquisition task, it sequentially reads data from the temperature sensor, humidity sensor, battery internal resistance monitoring module, and heat flow sensor according to the set acquisition frequency. Before reading the data, the system first checks the status of the sensors to ensure they are in normal working order. For the data read from each sensor, preliminary format conversion and verification are performed, converting it into a data format that the system can recognize and process, and checking the integrity and validity of the data. If data anomalies are found (such as missing data, data exceeding the normal range, etc.), the system automatically marks the data and records the time of the anomaly and the sensor number, among other information. After preliminary processing, the data is stored in the system's database according to different sensor types and time sequence. The database adopts a distributed storage structure to improve the reliability and scalability of data storage. Simultaneously, to facilitate subsequent data querying and analysis, the stored data is indexed and categorized, establishing corresponding index tables and metadata information.

[0058] During implementation, real-time battery power (current × voltage, accuracy 0.1% FS) is synchronously collected at fixed intervals.

[0059] Record the air conditioner operating parameters (including compressor frequency, air supply temperature, and fan speed level) at fixed intervals.

[0060] Physical parameter initialization: Fixed recording of physical parameters such as battery pack thermal capacity and heat dissipation area.

[0061] In practice, after the system starts, a fixed data acquisition interval is set, for example, 500ms. When the set acquisition time is reached, the system automatically sends a data request command to the battery current monitoring device and the voltage monitoring device. Upon receiving the command, the current monitoring device and the voltage monitoring device quickly read the current battery current value and voltage value. The read current and voltage values ​​are transmitted back to the system to quickly calculate the real-time battery power.

[0062] During implementation, a fixed recording interval is set, such as recording once every 2 seconds. The system sends a query command to the relevant data interface of the air conditioning control system to request information on compressor frequency, air supply temperature and fan speed level.

[0063] Upon receiving a command, the air conditioning control system reads the compressor's current operating frequency data stored internally in real time. This frequency data is precisely measured by the frequency monitoring module in the compressor drive circuit and fed back to the control system. Simultaneously, it reads the supply air temperature data measured by a high-precision temperature sensor installed at the air outlet, as well as the wind speed level data detected by a wind speed sensor. The accuracy of these sensors meets the system requirements.

[0064] The acquired compressor frequency, supply air temperature, and fan speed level data are transmitted back to the main system. The system then converts this data to conform to its internal data storage format requirements.

[0065] Optionally, dynamic analysis and parameter calculation of thermal properties include:

[0066] The thermal time constant of the battery pack is calculated periodically at a first time threshold to reflect its thermal inertia characteristics.

[0067] The internal resistance change rate is calculated in real time, and the data is updated at a second time threshold to identify abnormal battery status.

[0068] The heat exchange coefficient is calculated using heat flow sensor data at a third time threshold interval, providing boundary conditions for the prediction model.

[0069] During implementation, dynamic analysis of thermal properties and parameter calculations include:

[0070] The thermal time constant is calculated at intervals of a first time threshold (e.g., if the first time threshold is 5 minutes, then the interval is 5 minutes) to reflect the thermal inertia characteristics of the battery pack.

[0071] The thermal time constant reflects the thermal inertia characteristics of a battery pack, describing the time required for the battery pack to reach its final steady-state temperature after being subjected to thermal disturbance.

[0072] In this invention, a first-order heat transfer model is used to calculate the thermal time constant. The battery pack can be regarded as a heat capacity body with a uniform temperature distribution. Its heat transfer process can be described by a first-order linear differential equation. Based on the power balance equation at steady state, the steady-state temperature is calculated. The collected temperature data is fitted with the theoretical formula, and the thermal time constant can be calculated and determined by the least squares method. The unit of the thermal time constant is time unit.

[0073] The system monitors the rate of change of internal resistance, calculates the rate of change of internal resistance in real time, and updates the data at intervals with a second time threshold (e.g., if the second time threshold is 2 seconds, the data is updated every 2 seconds) to identify abnormal battery status.

[0074] During implementation, real-time data of the battery's internal resistance is obtained from the battery internal resistance monitoring module. The rate of change of internal resistance is calculated by comparing the internal resistance measurements at two adjacent time points.

[0075] Battery Status Anomaly Detection: A pre-set threshold for the rate of change of internal resistance is used to determine whether the battery status is abnormal. This threshold is determined based on the battery's characteristics and historical data; different types of batteries may have different thresholds. When the calculated rate of change of internal resistance exceeds the set threshold, the system will determine that the battery status is abnormal.

[0076] Determining the heat exchange coefficient: The heat exchange coefficient is calculated using heat flow sensor data at intervals of a third time threshold (e.g., if the third time threshold is 10 minutes, the heat exchange coefficient is calculated every 10 minutes) to provide boundary conditions for the prediction model.

[0077] In practice, heat flow data is read from a heat flow sensor, and the heat exchange coefficient is calculated as follows: According to the law of thermal conductivity, the heat exchange coefficient h can be calculated using the formula h = q / (AΔT), where q is the heat flux density (which can be measured by the heat flow sensor), A is the heat dissipation area of ​​the battery pack (already recorded in the physical parameter initialization), and ΔT is the temperature difference between the surface temperature of the battery pack and the ambient temperature. The heat exchange coefficient can be calculated using the measured heat flow data, the known heat dissipation area, and the real-time temperature difference.

[0078] Optionally, a hybrid prediction model is established and temperature prediction includes:

[0079] An LSTM neural network model is established. By inputting the current battery temperature, power, environmental parameters, historical temperature change rate, real-time internal resistance, and thermal characteristic parameters into the LSTM neural network model, the predicted temperature value is output for each time period (e.g., when 30 minutes is 6 time periods and each time period is 5 minutes) within a certain number of future time periods (e.g., within 30 minutes, 6 time periods and each time period is 5 minutes).

[0080] In practice, establishing the LSTM neural network model includes: first, denoising the raw temperature data collected from the temperature sensor by using a moving average filter to remove high-frequency noise and make the temperature data smoother; then, normalization is performed to map the temperature data to the [0,1] interval for easier processing by the LSTM neural network model.

[0081] For the real-time power data calculated from battery current and voltage, an accuracy check is first performed to ensure that the power data is within a reasonable range. Then, normalization is also performed, mapping the maximum and minimum power values ​​to the target numerical range.

[0082] Environmental parameters such as humidity are first checked for validity, and obviously erroneous or abnormal data are removed. Then, they are normalized according to their own value range to make them similar in scale to other input data.

[0083] By recording temperature data at different time points, calculating the temperature difference between adjacent time points, and then dividing by the time interval, the rate of temperature change is obtained. To make the rate of temperature change more reflective of the overall trend, the average rate of temperature change within a certain time window is used as the input feature.

[0084] Mean squared error (MSE) is used as the loss function for the LSTM neural network model to measure the difference between the temperature value predicted by the model and the actual temperature value. The model parameters are adjusted by minimizing the loss function.

[0085] The Adam optimization algorithm is used to update the model's weight parameters. This algorithm can adaptively adjust the learning rate, speed up the model's convergence, and improve training efficiency.

[0086] The hyperparameters of the LSTM neural network model were adjusted using cross-validation: the number of hidden layer neurons, the number of layers, and the time step.

[0087] Optionally, a hybrid prediction model is established and temperature prediction includes:

[0088] A physical heat conduction model is established based on the battery pack's heat capacity, heat dissipation area, heat exchange coefficient, ambient temperature, and thermal time constant. This model is then used to predict the theoretical temperature change trend under low-temperature conditions.

[0089] The specific steps involved in establishing a physical heat conduction model include:

[0090] The physical heat conduction model is based on Fourier's law of heat conduction and the law of conservation of energy. Fourier's law describes the relationship between heat flux and temperature gradient, that is, heat flux density is proportional to temperature gradient; the law of conservation of energy ensures energy balance during heat conduction. For the battery pack as a thermal system, it can be regarded as a heat-conducting object with an internal heat source (heat generated by battery operation), and its heat conduction process can be described by a partial differential equation.

[0091] During the data acquisition phase, the battery pack thermal capacity has been recorded and stored. This is a physical parameter that reflects the battery pack's ability to store heat, and its unit is joules per Kelvin (J / K).

[0092] For heat dissipation area (A): The heat dissipation area recorded during data acquisition is the surface area of ​​the battery pack that exchanges heat with the surrounding environment, and the unit is square meters.

[0093] For heat transfer coefficient (h): The heat transfer coefficient is calculated using heat flow sensor data at a third time threshold (e.g., 10 minutes). It represents the ability of the battery pack to transfer heat to the surrounding environment, and the unit is watts per square kelvin (W / (m^2*K)).

[0094] For ambient temperature: The temperature value of the surrounding environment is collected in real time by an ambient temperature sensor, and the unit is degrees Celsius (°C).

[0095] Thermal time constant: The thermal time constant of the battery pack is calculated at intervals of a first time threshold (e.g., 5 minutes). It reflects the thermal inertia characteristics of the battery pack and is measured in minutes (min).

[0096] Then, considering the heat conduction process of the battery pack, the following heat conduction equation is established:

[0097] C is the battery pack heat capacity, T is the battery pack temperature, t is time (s), h is the heat transfer coefficient, A is the heat dissipation area, and T env Where is the ambient temperature, and P is the heat generation power (W) of the battery pack, which can be obtained from the real-time power data of the battery.

[0098] When starting temperature prediction, the actual temperature of the battery pack at the current moment is used as the initial temperature T0.

[0099] Boundary conditions: The heat exchange coefficient h is used to describe the heat exchange between the battery pack and the surrounding environment, i.e., the heat flux density q = h(T - T) env ).

[0100] Numerical methods, such as the Euler method or the Runge-Kutta method, are used to solve the heat conduction equation. The specific steps are as follows:

[0101] Discretize the time by dividing the prediction period into several small time steps ∆t;

[0102] Within each time step, based on the current temperature T n The temperature T at the next moment is calculated using the heat conduction equation. n + 1 : ;

[0103] Repeat the above steps until the predicted time period is reached.

[0104] Under low-temperature conditions, the battery's internal resistance increases, causing a change in the heat generation power P. To accurately predict the theoretical temperature change trend under low-temperature conditions, the heat generation power needs to be corrected. Based on the battery's internal resistance monitoring data, the battery's internal resistance R is calculated in real time, and the heat generation power is calculated according to Ohm's law P = I^2*R, where I is the battery current.

[0105] Optionally, controlling the temperature prediction by the LSTM model and correcting the prediction by the physical heat conduction model includes:

[0106] When the ambient temperature is greater than or equal to the first temperature threshold (e.g., the first temperature threshold is 5℃, and the ambient temperature is ≥5℃): the temperature is predicted by the LSTM model and the predicted value is corrected by the physical heat conduction model (correction coefficient k = 0.9~1.1).

[0107] Specifically, the predicted values ​​are corrected by the physical heat conduction model by combining the temperature values ​​predicted by the LSTM model with the theoretical temperature values ​​calculated by the physical heat conduction model.

[0108] The specific calculation method is as follows: Corrected temperature prediction value = k × LSTM model prediction value + (1 - k) × physical heat conduction model theoretical temperature value. The corrected temperature prediction value will more accurately reflect the actual temperature change trend of the battery pack.

[0109] Optionally, controlling the temperature prediction by the physical heat conduction model and providing boundary conditions by the LSTM model includes:

[0110] When the ambient temperature is below a first temperature threshold (e.g., the first temperature threshold is 5℃, when the ambient temperature is <5℃), the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model.

[0111] The boundary conditions provided by the LSTM model are specifically: the temperature predictions over several future time periods output by the LSTM model provide boundary references for the physical heat conduction model. For example, during the solution process of the physical heat conduction model, the temperature values ​​predicted by the LSTM model are used as boundary constraints, making the prediction results of the physical heat conduction model more consistent with reality.

[0112] Optionally, the prediction model control submodule 1022 is also used to control:

[0113] When the temperature prediction error is greater than the second temperature threshold (e.g., the second temperature threshold is 1.2℃) when the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model, the temperature prediction is switched to the LSTM model and the predicted value is corrected by the physical heat conduction model.

[0114] Optionally, adaptive adjustment of the prediction step size in temperature prediction for LSTM neural network models or physical heat conduction models includes:

[0115] First, the prediction step size is initially set, and then the prediction step size is adaptively adjusted. That is, the prediction step size is dynamically adjusted according to the thermal time constant τ (for example, if the thermal time constant τ ≤ 15 minutes, the prediction step size is 20 minutes; if 15 minutes < thermal time constant τ ≤ 25 minutes, the prediction step size is 30 minutes; if the thermal time constant τ > 25 minutes, the prediction step size is 40 minutes) to match the thermal inertia characteristics.

[0116] In practice, dynamically adjusting the prediction step size based on the thermal time constant includes:

[0117] Each time a new thermal time constant is calculated, the value is updated immediately to ensure the accuracy of subsequent prediction step size adjustments.

[0118] The system has preset prediction step sizes for different thermal time constant ranges. When the thermal time constant meets the corresponding conditions, it automatically adjusts to the corresponding prediction step size.

[0119] The adjusted prediction step size will be applied to the temperature prediction of subsequent LSTM neural network models or physical heat conduction models to ensure that the model can make more accurate temperature predictions based on the thermal inertia characteristics of the battery pack.

[0120] Optionally, the power control module 103, used to control the power of the air conditioner based on the temperature prediction results from the prediction model, includes:

[0121] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature (the ideal temperature refers to the optimal temperature for battery charging) within a certain period of time, and the time of deviation is greater than the fourth time threshold (e.g., the fourth time threshold is 2 minutes), then the air conditioner's power will be controlled to continuously operate at maximum cooling / heating power.

[0122] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature within a certain period of time, and the time of deviation is less than or equal to the fourth time threshold and greater than the fifth time threshold (e.g., the fifth time threshold is 1 minute), then the air conditioner will be controlled to operate at normal power (e.g., normal power is 80% of the maximum cooling / heating power) and run continuously.

[0123] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature within a certain period of time, and the time of deviation is less than or equal to the fifth time threshold, then the air conditioner will be controlled to operate at a low efficiency power (e.g., the low efficiency power is 50% of the maximum cooling / heating power) for continuous operation.

[0124] In practice, controlling the power of the air conditioner involves controlling the air conditioner hardware. The air conditioner hardware control system sends a query command to read and transmit data such as compressor frequency, air supply temperature and fan speed level to the main system of this invention. At the same time, it receives the power control command sent by the main system of this invention. The sent command reaches the compressor, and the compressor adjusts its operating frequency to achieve different cooling / heating power.

[0125] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0126] like Figure 2 The present invention also discloses an adaptive performance optimization air conditioning control method, which includes the following steps:

[0127] S100. Dynamically acquires data from multiple sensors;

[0128] S101. Perform dynamic analysis and parameter calculation of thermal properties:

[0129] S102. Based on the results and parameters of dynamic thermal property analysis, a hybrid prediction model is established to predict temperature:

[0130] Establish an LSTM neural network model;

[0131] Establish a physical heat conduction model;

[0132] When the ambient temperature is greater than or equal to the first temperature threshold: the temperature is predicted by the LSTM model and the predicted value is corrected by the physical heat conduction model;

[0133] When the ambient temperature is below a first temperature threshold: the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model;

[0134] Adaptive adjustment of prediction step size in temperature prediction using LSTM neural network models or physical heat conduction models;

[0135] S103. Control the power of the air conditioner based on the temperature prediction results from the prediction model.

[0136] In some optional implementations, dynamic analysis of thermal properties and parameter calculations include:

[0137] The thermal time constant of the battery pack is calculated periodically at a first time threshold to reflect its thermal inertia characteristics; the rate of change of internal resistance is calculated in real time, and the data is updated at a second time threshold to identify abnormal battery conditions; the heat exchange coefficient is calculated using heat flow sensor data at a third time threshold to provide boundary conditions for the prediction model.

[0138] In some alternative implementations, an LSTM neural network model is used to output a predicted temperature value for each time period within a future number of time periods by taking the current battery temperature, power, environmental parameters, historical temperature change rate, real-time internal resistance, and thermal characteristic parameters as input to the LSTM neural network model.

[0139] In some optional implementations, a physical heat conduction model is established, including: establishing a physical heat conduction model based on the battery pack's heat capacity, heat dissipation area, heat exchange coefficient, ambient temperature, and thermal time constant, and using it to predict the theoretical temperature change trend under low-temperature conditions.

[0140] In some alternative implementations, when the temperature prediction error exceeds a second temperature threshold when the temperature is predicted by the physical heat conduction model and the boundary conditions are provided by the LSTM model, the temperature prediction is switched to the LSTM model, and the predicted value is corrected by the physical heat conduction model.

[0141] In some alternative implementations, adaptive adjustment of the prediction step size in temperature prediction for LSTM neural network models or physical heat conduction models includes: first, setting an initial prediction step size, and then adaptively adjusting the prediction step size, i.e., dynamically adjusting the prediction step size according to the thermal time constant τ to match the thermal inertia characteristics.

[0142] In some alternative implementations, controlling the power of the air conditioner based on temperature predictions from a predictive model includes:

[0143] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature within a certain period of time, and the time of deviation is greater than the fourth time threshold, then the power of the air conditioner will be controlled to run continuously at the maximum cooling / heating power.

[0144] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature within a certain period of time, and the time of deviation is less than or equal to the fourth time threshold and greater than the fifth time threshold, then the air conditioner will be controlled to run at normal power continuously.

[0145] If the temperature prediction results show that the predicted temperature deviates from the ideal temperature within a certain period in the future, and the time of deviation is less than or equal to the fifth time threshold, then the air conditioner will be controlled to operate at low-efficiency power continuously.

[0146] The adaptive performance optimization air conditioning control system provided in the embodiments of the present invention can execute the adaptive performance optimization air conditioning control method provided in any embodiment of the present invention, which has the corresponding method and beneficial effects of the system.

[0147] It is evident that the method of the present invention can be implemented by a computer program, and the computer program for implementing the method of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer program causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer program can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0148] Therefore, it can be understood that this invention discloses an electronic device, comprising:

[0149] At least one processor; and

[0150] A memory communicatively connected to the at least one processor; wherein,

[0151] The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform the aforementioned adaptive performance optimization air conditioning control method.

[0152] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An adaptive performance optimized air conditioning control system, characterized by: The application relates to a temperature prediction method and device for a battery pack. The application comprises: a data acquisition module for dynamically acquiring multi-sensor data; a data analysis and operation module for performing thermal characteristic dynamic analysis and parameter calculation; a hybrid prediction model control module for establishing a hybrid prediction model based on the thermal characteristic dynamic analysis result and the thermal characteristic dynamic analysis parameter and predicting temperature; a power control module for controlling the power of the air conditioner according to the result of temperature prediction of the prediction model; the hybrid prediction model control module comprises: a neural network model submodule for establishing and managing an LSTM neural network model; a physical heat conduction model submodule for establishing and managing a physical heat conduction model; a prediction model control submodule for controlling: when the working environment temperature is greater than or equal to a first temperature threshold value, controlling temperature prediction by the LSTM model and correcting the prediction value by the physical heat conduction model; when the working environment temperature is less than the first temperature threshold value, controlling temperature prediction by the physical heat conduction model and providing a boundary condition by the LSTM model; the prediction model control submodule is further used for controlling adaptive adjustment of a prediction step in temperature prediction of the LSTM neural network model or the physical heat conduction model; when the result of temperature prediction shows that the predicted temperature in a future period deviates from the ideal temperature, and the deviation time is greater than a fourth time threshold value, controlling the power of the air conditioner to continuously operate at the maximum refrigeration / heating power; when the result of temperature prediction shows that the predicted temperature in a future period deviates from the ideal temperature, and the deviation time is less than or equal to the fourth time threshold value and greater than a fifth time threshold value, controlling the power of the air conditioner to continuously operate at the normal power; 2. An adaptive performance optimization air conditioning control system according to claim 1, wherein: when the result of temperature prediction shows that the predicted temperature in a future period deviates from the ideal temperature, and the deviation time is less than or equal to the fifth time threshold value, controlling the power of the air conditioner to continuously operate at the low-efficiency power. The dynamic acquisition of multi-sensor data comprises:

3. An adaptive performance optimization air conditioning control system according to claim 1, wherein: dynamically acquiring temperature sensor data, humidity sensor data, battery internal resistance monitoring data and heat flow sensor data; further comprising: synchronously acquiring the real-time power of the battery at a fixed interval, and recording the air conditioner operation parameters at a fixed interval. The thermal characteristic dynamic analysis and parameter calculation comprises: calculating the thermal time constant of the battery pack at intervals of a first time threshold value to reflect the thermal inertia characteristic; calculating the internal resistance change rate in real time, updating the data at intervals of a second time threshold value, and identifying the abnormal state of the battery; 4. An adaptive performance optimization air conditioning control system according to claim 1, wherein: the heat exchange coefficient is calculated by using the heat flow sensor data at intervals of a third time threshold value to provide a boundary condition for the prediction model.

5. An adaptive performance optimization air conditioning control system according to claim 1, wherein: The LSTM neural network model is used for outputting the temperature prediction value of each time period in a future period by inputting the current battery temperature, power, environmental parameter, historical temperature change rate, real-time internal resistance and thermal characteristic parameter into the LSTM neural network model.

6. An adaptive performance optimization air conditioning control system according to claim 1, wherein: The establishment and management of the physical heat conduction model comprises: establishing the physical heat conduction model based on the thermal capacity of the battery pack, the heat dissipation area, the heat exchange coefficient, the environmental temperature and the thermal time constant, and using the physical heat conduction model to predict the theoretical temperature change trend under a low-temperature working condition. the prediction model control submodule is further used for controlling: When the temperature is predicted by the physical heat conduction model and the boundary condition is provided by the LSTM model, if the prediction temperature error is greater than the second temperature threshold, the temperature is predicted by the LSTM model, and the prediction value is corrected by the physical heat conduction model.

7. An adaptive performance optimization air conditioning control system according to claim 1, wherein: The adaptive adjustment of the prediction step in the temperature prediction of the LSTM neural network model or the physical heat conduction model comprises: firstly setting the prediction step, and then adaptively adjusting the prediction step, and dynamically adjusting the prediction step according to the thermal time constant to match the thermal inertia characteristics.

8. An adaptive performance optimization air conditioning control method, characterized by: The method comprises the steps of: Collecting multi-sensor data dynamically; Performing thermal characteristic dynamic analysis and parameter calculation; Establishing a hybrid prediction model based on the results of the thermal characteristic dynamic analysis and the thermal characteristic dynamic analysis parameters, and predicting the temperature; Establishing an LSTM neural network model; Establishing a physical heat conduction model; When the working environment temperature is greater than or equal to the first temperature threshold: controlling the temperature prediction by the LSTM model, and correcting the prediction value by the physical heat conduction model; When the working environment temperature is less than the first temperature threshold: controlling the temperature prediction by the physical heat conduction model, and providing the boundary condition by the LSTM model; Adaptive adjustment of the prediction step in the temperature prediction of the LSTM neural network model or the physical heat conduction model; Controlling the power of the air conditioner according to the result of the temperature prediction by the prediction model; When the result of the temperature prediction shows that the predicted temperature and the ideal temperature exist deviation in a certain period in the future, and the deviation time is greater than the fourth time threshold: controlling the power of the air conditioner to run at the maximum cooling / heating power; When the result of the temperature prediction shows that the predicted temperature and the ideal temperature exist deviation in a certain period in the future, and the deviation time is less than or equal to the fourth time threshold and greater than the fifth time threshold: controlling the power of the air conditioner to run at the normal power; When the result of the temperature prediction shows that the predicted temperature and the ideal temperature exist deviation in a certain period in the future, and the deviation time is less than or equal to the fifth time threshold: controlling the power of the air conditioner to run at the low-power.

9. An electronic device comprising: At least one processor; And a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of claim 8. At least one processor; And a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of claim 8.

Images