Temperature control method and system based on deep learning

Abstract

The invention discloses a temperature control method and system based on deep learning, and the method comprises the steps: 1, employing a deep neural network model, combining real-time data and historical temperature records, and predicting the temperature change in a phase change refrigeration system through regression analysis; and 2, the deep neural network model dynamically adjusts the working parameters of the phase change refrigeration system based on the real-time data. By integrating the deep learning technology and the existing overcharge charging gun phase change refrigeration system, an innovative solution is provided for improving the charging efficiency and safety of the electric vehicle, the adaptability and response speed of the phase change refrigeration system to environmental changes are enhanced by introducing the deep learning algorithm, and the charging efficiency and safety of the electric vehicle are improved. In addition, energy consumption and operation cost can be reduced through implementation of the electric vehicle charging system, and therefore a more sustainable and environment-friendly solution is provided for charging of the electric vehicle.

Description

Technical Field

[0001] The present invention relates to the field of charging, and in particular to a temperature control method and system based on deep learning. Background Art

[0002] Fuel vehicles have become an indispensable part of people's work and life. Once hailed as machines that changed the world, vehicle technology has brought us convenient and fast travel methods, but it has also caused many problems such as environmental pollution and energy security.

[0003] Therefore, energy conservation and emission reduction has become one of the top priorities in the development of vehicle technology today. In this context, electric vehicles, as an important product in energy conservation and emission reduction and new energy development, are gradually replacing traditional power vehicles, leading the entire vehicle industry to a new development direction. With the continuous recognition of electric vehicles by users, the demand for electric vehicles is also increasing, so the market requirements for electric vehicle charging guns are also getting higher and higher.

[0004] When charging an electric vehicle, it is necessary to use a charging gun. After the charging gun is connected to the charging port of the electric vehicle, the temperature of the charging gun changes with the change of charging current. When the temperature is too high, if it is not adjusted in time, it is easy to scald the user, increase energy loss and thus reduce charging efficiency, and damage the charging interface of the electric vehicle.

[0005] Therefore, it is necessary to propose a temperature regulation method for a charging gun to improve the efficiency and safety of charging of electric vehicles. Summary of the invention

[0006] Purpose of the invention: To provide a temperature control method and system based on deep learning to solve the above-mentioned problems existing in the prior art.

[0007] Technical solution: A temperature control method based on deep learning, including: Step 1: Use a deep neural network model to combine real-time data and historical temperature records to predict temperature changes in the phase change refrigeration system through regression analysis; Step 2: The deep neural network model dynamically adjusts the operating parameters of the phase change refrigeration system based on real-time data.

[0008] The present invention provides an innovative solution to improve the efficiency and safety of electric vehicle charging by integrating deep learning technology and the existing supercharging gun phase change refrigeration system. The introduction of deep learning algorithms not only enhances the adaptability and response speed of the phase change refrigeration system to environmental changes, but also improves the overall energy efficiency and convenience of maintenance. In addition, the implementation of the present invention can also reduce energy consumption and operating costs, thereby providing a more sustainable and environmentally friendly solution for electric vehicle charging.

[0009] In a further embodiment, step 1 further comprises: Step 11: Analyze the temperature sensor distribution data using a convolutional neural network (CNN); Step 12: Use the convolutional neural network structure to analyze the spatial distribution data of the temperature sensor and predict the future temperature trend through time series analysis; in, is the predicted temperature, and They are historical temperature and pressure data, It is a deep learning model; Step 13: predict the future temperature changes at various locations inside the charging gun; By adopting a multi-layer perceptron structure and combining external environmental factors, the network parameters of the deep neural network model are continuously adjusted through optimization algorithms to improve prediction accuracy.

[0010] In a further embodiment, the step 11 further comprises: Step 111, data collection, installing temperature sensors at key locations to collect spatial distribution data, the location coordinates of each sensor and the temperature reading at the corresponding time point; Step 112: data preprocessing, standardizing the collected temperature data to eliminate the influence of different scales and ensure the consistency of data input; Step 113, construct a convolutional neural network model, and design a neural network model containing multiple convolutional layers to extract the spatial features of temperature distribution. Each convolutional layer automatically learns local patterns in the data by applying different filters.

[0011] Step 114, feature extraction and analysis, the convolution layer extracts key spatial features by performing multiple transformations and pooling operations on the input data. These features capture the spatial correlation of temperature changes and provide a basis for subsequent temperature trend prediction; Step 115, time series prediction, combining the extracted spatial features and historical temperature data, using convolutional neural network to perform time series analysis, the network predicts future temperature trends by learning how temperature changes over time and space.

[0012] Step 116, training and optimization, using the historical temperature data set to train the network model, by adjusting the network parameters (such as learning rate, filter size, number of layers, etc.), using the gradient descent optimization algorithm, minimizing the prediction error algorithm, to improve the accuracy and generalization ability of the model; Step 117, model deployment and real-time application, deploying the trained model to the actual system, processing the data from the temperature sensor in real time, dynamically adjusting the working parameters of the phase change refrigeration system, and ensuring that the system operates in the optimal state; In a further embodiment, step 13 further comprises: The optimization algorithm includes the gradient descent algorithm, including: Step 131, parameter initialization, the weight and bias parameters of the network are initialized to small random numbers, which helps to break the symmetry and ensure effective gradient descent; Step 132, forward propagation, input data (such as temperature data and external environmental factor data) is forward propagated through the neural network, and each layer uses an activation function to process its output until the last layer generates a prediction result; Step 133, loss function calculation, use the loss function (such as mean square error) to calculate the difference between the actual output and the predicted output; this step is key to evaluating the accuracy of the model's prediction.

[0013] Step 134, back propagate, calculate the gradient according to the loss function; this involves calculating partial derivatives with respect to the weights and biases of each layer in the network. Through the chain rule, these gradients indicate the sensitivity of the loss function with respect to each parameter.

[0014] Step 135: Parameter update: Use the gradient descent algorithm to update the network parameters. For each parameter, its update can be expressed as: parameter = parameter − learning rate × gradient The learning rate is a key hyperparameter that controls the size of the step in the loss landscape. Too large a learning rate may lead to overshoot, while too small a learning rate may lead to slow convergence.

[0015] Step 136, iterative optimization, repeating the forward propagation, loss calculation, back propagation and parameter update steps until the performance of the model reaches a predetermined threshold or a preset number of iterations is completed; Step 137: Model validation and adjustment: Test the model performance on an independent validation dataset to verify the prediction accuracy and generalization ability. Adjust the learning rate, network architecture or training strategy based on the validation results to optimize the model.

[0016] In a further embodiment, step 2 further comprises: Step 21: Adjust based on the principle of minimizing the prediction error to ensure that the system operates in the optimal state. The error minimization is expressed by the loss function L: L= , the goal is to minimize L; Step 22: The deep neural network model works in conjunction with the four-way microchannel heat exchanger to monitor the efficiency of the four-way microchannel heat exchanger in real time through data fusion, and adjusts the operating parameters based on the efficiency of the four-way microchannel heat exchanger.

[0017] In a further embodiment, the step 22 further comprises: Step 221, data collection, collecting data such as temperature, flow rate and pressure from multiple detection points of the four-way microchannel heat exchanger. At the same time, collecting data on external environmental factors such as ambient temperature and humidity; Step 222, data preprocessing, cleaning and standardizing the collected data to ensure data quality and consistency, and provide accurate input for subsequent analysis; Step 223, feature fusion, using feature engineering methods, such as principal component analysis (PCA) or autoencoders, to extract key features from multi-source data and fuse them to reduce data dimensions and improve processing efficiency; Step 224, applying a deep learning model, such as a convolutional neural network (CNN), to use the fused features to predict the operating efficiency of the heat exchanger in real time; the model is able to identify efficiency patterns implicit in the data and maintain accurate predictions even under changing operating conditions.

[0018] Step 225, parameter adjustment decision, based on the predicted heat exchanger efficiency, dynamically adjust the working parameters of the heat exchanger, such as flow rate, inlet and outlet water temperature difference, etc., through an optimization algorithm (such as gradient descent or genetic algorithm) to optimize system performance.

[0019] Step 226, feedback loop, establish a feedback mechanism, compare the actual operation data with the predicted data, evaluate the effect of parameter adjustment, and retrain the model and fine-tune the parameters according to the deviation to ensure the system adaptability and long-term stability.

[0020] A temperature control system based on deep learning, comprising: A phase change refrigeration system, comprising a variable frequency compressor, and a charging gun connected to an outlet of the variable frequency compressor; The charging gun has a built-in four-way microchannel heat exchanger connected to the outlet of the variable frequency compressor; A plurality of temperature sensors are designed and installed on the charging gun to monitor the temperature change of the charging gun in real time and transmit the data to the deep neural network model for analysis in real time.

[0021] In a further embodiment, the four-way microchannel heat exchanger comprises: Four groups of microchannels are installed inside the charging gun. A contact layer is provided on the four groups of microchannels. The contact layer is in contact with the charging gun pins to absorb and conduct the heat of the charging gun pins. There is a cooling medium in the microchannels.

[0022] Beneficial effects: The present invention discloses a temperature control method and system based on deep learning. The present invention provides an innovative solution to improve the efficiency and safety of electric vehicle charging by integrating deep learning technology and the existing supercharging gun phase change refrigeration system. The introduction of deep learning algorithm not only enhances the adaptability and response speed of the phase change refrigeration system to environmental changes, but also improves the overall energy efficiency and convenience of maintenance. In addition, the implementation of the present invention can also reduce energy consumption and operating costs, thereby providing a more sustainable and environmentally friendly solution for electric vehicle charging. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a schematic flow chart of the method of the present invention.

[0024] Figure 2 It is a schematic flow chart of step 1 of the present invention.

[0025] Figure 3 It is a schematic flow chart of step 2 of the present invention. DETAILED DESCRIPTION

[0026] The present application relates to a temperature control method and system based on deep learning, which is explained in detail below through specific implementation methods.

[0027] A temperature control method based on deep learning, comprising: Step 1: Use a deep neural network model to combine real-time data and historical temperature records to predict temperature changes in the phase change refrigeration system through regression analysis; Step 2: The deep neural network model dynamically adjusts the operating parameters of the phase change refrigeration system based on real-time data.

[0028] The present invention provides an innovative solution to improve the efficiency and safety of electric vehicle charging by integrating deep learning technology and the existing supercharging gun phase change refrigeration system. The introduction of deep learning algorithms not only enhances the adaptability and response speed of the phase change refrigeration system to environmental changes, but also improves the overall energy efficiency and convenience of maintenance. In addition, the implementation of the present invention can also reduce energy consumption and operating costs, thereby providing a more sustainable and environmentally friendly solution for electric vehicle charging.

[0029] The step 1 also includes: Step 11: Analyze the temperature sensor distribution data using a convolutional neural network (CNN); Step 12: Use the convolutional neural network structure to analyze the spatial distribution data of the temperature sensor and predict the future temperature trend through time series analysis; in, is the predicted temperature, and They are historical temperature and pressure data, It is a deep learning model; Step 13: predict the future temperature changes at various locations inside the charging gun; By adopting a multi-layer perceptron structure and combining external environmental factors, the network parameters of the deep neural network model are continuously adjusted through optimization algorithms to improve prediction accuracy.

[0030] The step 11 further comprises: Step 111, data collection, installing temperature sensors at key locations to collect spatial distribution data, the location coordinates of each sensor and the temperature reading at the corresponding time point; Step 112: data preprocessing, standardizing the collected temperature data to eliminate the influence of different scales and ensure the consistency of data input; Step 113, construct a convolutional neural network model, and design a neural network model containing multiple convolutional layers to extract the spatial features of temperature distribution. Each convolutional layer automatically learns local patterns in the data by applying different filters.

[0031] Step 114, feature extraction and analysis, the convolution layer extracts key spatial features by performing multiple transformations and pooling operations on the input data. These features capture the spatial correlation of temperature changes and provide a basis for subsequent temperature trend prediction; Step 115, time series prediction, combining the extracted spatial features and historical temperature data, using convolutional neural network to perform time series analysis, the network predicts future temperature trends by learning how temperature changes over time and space.

[0032] Step 116, training and optimization, using the historical temperature data set to train the network model, by adjusting the network parameters (such as learning rate, filter size, number of layers, etc.), using the gradient descent optimization algorithm, minimizing the prediction error algorithm, to improve the accuracy and generalization ability of the model; Step 117, model deployment and real-time application, deploying the trained model to the actual system, processing the data from the temperature sensor in real time, dynamically adjusting the working parameters of the phase change refrigeration system, and ensuring that the system operates in the optimal state; The step 13 further comprises: The optimization algorithm includes the gradient descent algorithm, including: Step 131, parameter initialization, the weight and bias parameters of the network are initialized to small random numbers, which helps to break the symmetry and ensure effective gradient descent; Step 132, forward propagation, input data (such as temperature data and external environmental factor data) is forward propagated through the neural network, and each layer uses an activation function to process its output until the last layer generates a prediction result; Step 133, loss function calculation, use the loss function (such as mean square error) to calculate the difference between the actual output and the predicted output; this step is key to evaluating the accuracy of the model's prediction.

[0033] Step 134, back propagate, calculate the gradient according to the loss function; this involves calculating partial derivatives with respect to the weights and biases of each layer in the network. Through the chain rule, these gradients indicate the sensitivity of the loss function with respect to each parameter.

[0034] Step 135: Parameter update: Use the gradient descent algorithm to update the network parameters. For each parameter, its update can be expressed as: parameter = parameter − learning rate × gradient The learning rate is a key hyperparameter that controls the size of the step in the loss landscape. Too large a learning rate may lead to overshoot, while too small a learning rate may lead to slow convergence.

[0035] Step 136, iterative optimization, repeating the forward propagation, loss calculation, back propagation and parameter update steps until the performance of the model reaches a predetermined threshold or a preset number of iterations is completed; Step 137: Model validation and adjustment: Test the model performance on an independent validation dataset to verify the prediction accuracy and generalization ability. Adjust the learning rate, network architecture or training strategy based on the validation results to optimize the model.

[0036] The step 2 also includes: Step 21: Adjust based on the principle of minimizing the prediction error to ensure that the system operates in the optimal state. The error minimization is expressed by the loss function L: L= , the goal is to minimize L; Step 22: The deep neural network model works in conjunction with the four-way microchannel heat exchanger to monitor the efficiency of the four-way microchannel heat exchanger in real time through data fusion, and adjusts the operating parameters based on the efficiency of the four-way microchannel heat exchanger.

[0037] The step 22 further comprises: Step 221, data collection, collecting data such as temperature, flow rate and pressure from multiple detection points of the four-way microchannel heat exchanger. At the same time, collecting data on external environmental factors such as ambient temperature and humidity; Step 222, data preprocessing, cleaning and standardizing the collected data to ensure data quality and consistency, and provide accurate input for subsequent analysis; Step 223, feature fusion, using feature engineering methods, such as principal component analysis (PCA) or autoencoders, to extract key features from multi-source data and fuse them to reduce data dimensions and improve processing efficiency; Step 224, applying a deep learning model, such as a convolutional neural network (CNN), to use the fused features to predict the operating efficiency of the heat exchanger in real time; the model is able to identify efficiency patterns implicit in the data and maintain accurate predictions even under changing operating conditions.

[0038] Step 225, parameter adjustment decision, based on the predicted heat exchanger efficiency, dynamically adjust the working parameters of the heat exchanger, such as flow rate, inlet and outlet water temperature difference, etc., through an optimization algorithm (such as gradient descent or genetic algorithm) to optimize system performance.

[0039] Step 226, feedback loop, establish a feedback mechanism, compare the actual operation data with the predicted data, evaluate the effect of parameter adjustment, and retrain the model and fine-tune the parameters according to the deviation to ensure the system adaptability and long-term stability.

[0040] A temperature control system based on deep learning, comprising: A phase change refrigeration system, comprising a variable frequency compressor, and a charging gun connected to an outlet of the variable frequency compressor; The charging gun has a built-in four-way microchannel heat exchanger connected to the outlet of the variable frequency compressor; A plurality of temperature sensors are designed and installed on the charging gun to monitor the temperature change of the charging gun in real time and transmit the data to the deep neural network model for analysis in real time.

[0041] The four-way microchannel heat exchanger comprises: Four groups of microchannels are installed inside the charging gun. A contact layer is provided on the four groups of microchannels. The contact layer is in contact with the charging gun pins to absorb and conduct the heat of the charging gun pins. There is a cooling medium in the microchannels.

[0042] Working principle description: Using a deep neural network model, combined with real-time data and historical temperature records, the temperature changes in the phase change refrigeration system are predicted through regression analysis; Use convolutional neural network (CNN) structure to analyze the spatial distribution data of temperature sensors and predict future temperature trends through time series analysis; in, is the predicted temperature, and They are historical temperature and pressure data, It is a deep learning model.

[0043] Predict future temperature changes at various locations inside the charging gun; Install temperature sensors at the predetermined locations of the supercharger guns to monitor temperature changes in real time and transmit the data to the deep neural network model for analysis in real time; Adopting the multi-layer perceptron structure (MLP), combined with external environmental factors, and through optimization algorithms, the network parameters of the deep neural network model are continuously adjusted to improve the prediction accuracy; The multi-layer perceptron structure includes installing a temperature sensor at a predetermined location on the supercharger gun to monitor temperature changes in real time; The optimization algorithm includes a gradient descent algorithm; External environmental factors include ambient temperature and humidity.

[0044] The deep neural network model dynamically adjusts the operating parameters of the phase change refrigeration system based on real-time data; The adjustment is based on the principle of minimizing the prediction error to ensure that the system operates in the optimal state. The error minimization is expressed by the loss function L: L= , the goal is to minimize L; The deep neural network model works together with the four-way microchannel heat exchanger to monitor the efficiency of the four-way microchannel heat exchanger in real time through data fusion, and adjust the working parameters based on the efficiency of the four-way microchannel heat exchanger.

[0045] Data fusion methods include weighted average method; The working parameters include the rotation speed and outlet pressure of the variable frequency compressor.

[0046] The preferred specific embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to the specific details in the above specific embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical scheme of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A temperature control method based on deep learning, characterized in that: include: Step 1: Use a deep neural network model to combine real-time data and historical temperature records to predict temperature changes in the phase change refrigeration system through regression analysis; Step 2: The deep neural network model dynamically adjusts the operating parameters of the phase change refrigeration system based on real-time data.

2. The temperature control method based on deep learning according to claim 1 is characterized in that: The step 1 also includes: Step 11: Analyze the temperature sensor distribution data using a convolutional neural network; Step 12: Use the convolutional neural network structure to analyze the spatial distribution data of the temperature sensor and predict the future temperature trend through time series analysis; in, is the predicted temperature, and They are historical temperature and pressure data, It is a deep learning model; Step 13: predict the future temperature changes at various locations inside the charging gun; By adopting a multi-layer perceptron structure and combining external environmental factors, the network parameters of the deep neural network model are continuously adjusted through optimization algorithms to improve prediction accuracy.

3. The temperature control method based on deep learning according to claim 2 is characterized in that: The step 11 further comprises: Step 111, data collection, installing temperature sensors at key locations to collect spatial distribution data, the location coordinates of each sensor and the temperature reading at the corresponding time point; Step 112: data preprocessing, standardizing the collected temperature data to eliminate the influence of different scales and ensure the consistency of data input; Step 113: construct a convolutional neural network model, designing a neural network model including multiple convolutional layers to extract spatial features of temperature distribution, each convolutional layer automatically learning local patterns in the data by applying different filters; Step 114, feature extraction and analysis, the convolution layer extracts key spatial features by performing multiple transformations and pooling operations on the input data. These features capture the spatial correlation of temperature changes and provide a basis for subsequent temperature trend prediction; Step 115, time series prediction, combining the extracted spatial features and historical temperature data, using a convolutional neural network to perform time series analysis, the network predicts future temperature trends by learning how temperature changes over time and space; Step 116, training and optimization, using the historical temperature data set to train the network model, by adjusting the network parameters, using the gradient descent optimization algorithm, minimizing the prediction error algorithm, to improve the accuracy and generalization ability of the model; Step 117, model deployment and real-time application, deploy the trained model to the actual system, process the data from the temperature sensor in real time, dynamically adjust the working parameters of the phase change refrigeration system, and ensure that the system operates in the optimal state.

4. The temperature control method based on deep learning according to claim 2 is characterized in that: The step 13 further comprises: The optimization algorithm includes the gradient descent algorithm, including: Step 131, parameter initialization, the weight and bias parameters of the network are initialized to small random numbers; Step 132, forward propagation, the input data is forward propagated through the neural network, and each layer uses an activation function to process its output until the last layer generates a prediction result; Step 133, loss function calculation, using the loss function to calculate the difference between the actual output and the predicted output; Step 134, back propagation, calculating the gradient according to the loss function; Step 135: parameter update, using gradient descent algorithm to update network parameters; Step 136, iterative optimization, repeating the forward propagation, loss calculation, back propagation and parameter update steps until the performance of the model reaches a predetermined threshold or a preset number of iterations is completed; Step 137: Model validation and adjustment: Test the model performance on an independent validation dataset to verify the prediction accuracy and generalization ability.

5. The temperature control method based on deep learning according to claim 1 is characterized in that: The step 2 also includes: Step 21: Adjust based on the principle of minimizing the prediction error to ensure that the system operates in the optimal state. The error minimization is expressed by the loss function L: L= , the goal is to minimize L; Step 22: The deep neural network model works in conjunction with the four-way microchannel heat exchanger to monitor the efficiency of the four-way microchannel heat exchanger in real time through data fusion, and adjusts the operating parameters based on the efficiency of the four-way microchannel heat exchanger.

6. The temperature control method based on deep learning according to claim 5 is characterized in that: The step 22 further comprises: Step 221, data collection, collecting data such as temperature, flow rate and pressure from multiple detection points of the four-way microchannel heat exchanger. At the same time, collecting data on external environmental factors such as ambient temperature and humidity; Step 222, data preprocessing, cleaning and standardizing the collected data to ensure data quality and consistency, and provide accurate input for subsequent analysis; Step 223: feature fusion, using feature engineering to extract key features from multi-source data and fuse them to reduce data dimensions and improve processing efficiency; Step 224: Apply the deep learning model to predict the working efficiency of the heat exchanger in real time using the fused features; Step 225, parameter adjustment decision, dynamically adjusting the working parameters of the heat exchanger through an optimization algorithm according to the predicted heat exchanger efficiency; Step 226, feedback loop, establish a feedback mechanism, compare the actual operation data with the predicted data, evaluate the effect of parameter adjustment, and retrain the model and fine-tune the parameters according to the deviation to ensure the system adaptability and long-term stability.

7. A temperature control system based on deep learning, characterized in that: include: A phase change refrigeration system, comprising a variable frequency compressor, and a charging gun connected to an outlet of the variable frequency compressor; The charging gun has a built-in four-way microchannel heat exchanger connected to the outlet of the variable frequency compressor; A plurality of temperature sensors are designed and installed on the charging gun to monitor the temperature change of the charging gun in real time and transmit the data to the deep neural network model for analysis in real time.

8. The temperature control system based on deep learning according to claim 7 is characterized in that: The four-way microchannel heat exchanger comprises: Four groups of microchannels are installed inside the charging gun. A contact layer is provided on the four groups of microchannels. The contact layer is in contact with the charging gun pins to absorb and conduct the heat of the charging gun pins. There is a cooling medium in the microchannels.

Images