Automobile head temperature sensor replacement method and system

Abstract

The invention provides an automobile head temperature sensor replacement method and system, and relates to the technical field of automobile head temperature detection, and the method comprises the following steps: a basic model development step: before the development of a new automobile model, firstly developing a basic artificial intelligence algorithm model; and a transfer learning stage step: before the new vehicle model is delivered, based on the basic artificial intelligence algorithm model, performing fine adjustment and optimization of the artificial intelligence algorithm model on the new vehicle model through transfer learning, and realizing prediction of the head temperature data of the vehicle model. The head breathing point temperature data can be predicted through the artificial intelligence algorithm, traditional entity physical hardware is replaced with the artificial intelligence algorithm, and the cost reduction requirement of a vehicle enterprise is met.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile head temperature detection, and in particular to a method and system for replacing an automobile head temperature sensor. Background Art

[0002] A temperature sensor is a sensor that can sense temperature and convert it into a usable output signal. According to the measurement method, temperature sensors can be divided into contact and non-contact types; according to the characteristics of the sensor material and the electronic components, temperature sensors can be divided into thermal resistors and thermocouples. The functions of automotive temperature sensors are: 1) Detecting the ambient temperature outside the car, the control system will determine the control mode based on the difference between the temperature outside the car and the temperature inside the car; 2) Providing the temperature signal outside the car to the ECU, it will determine the temperature inside the car based on the comparison of this signal with the temperature signal inside the car to meet the needs of the people inside the car.

[0003] The invention patent with publication number CN115257303B discloses an intelligent automobile heated roof, control method, equipment and storage medium. The intelligent automobile heated roof includes a distance sensor, a heating unit and an ambient temperature sensor installed on the automobile roof; the distance sensor and the ambient temperature sensor are respectively connected to the heating unit; the distance sensor is used to detect the distance between the head and the ceiling, and send the detection result to the heating unit, the ambient temperature sensor is used to detect the temperature inside the car in real time and send the result to the heating unit, and the heating unit decides whether to start heating and the heating temperature based on the ambient temperature and the head distance.

[0004] Traditional car head temperature detection uses a lot of actual physical hardware. At the same time, multiple components need to cooperate with each other to obtain and process data, which results in high overall costs.

[0005] The current method of replacing the actual physical hardware sensors in the car with algorithm software is mainly to simulate the thermal physics model and the whole vehicle structure model of a specific car model. This replacement method will cause the fitted data to deviate greatly from the actual data as the working conditions become more complicated, which will bring greater control risks to the subsequent thermal management control. In addition, due to the structural differences between different car models, this method requires the thermal physics model simulation and the whole vehicle structure model simulation for each car model, which involves a lot of repetitive work and low development efficiency. Summary of the invention

[0006] In view of the defects in the prior art, the present invention provides a method and system for replacing an automobile head temperature sensor.

[0007] According to a method and system for replacing a head temperature sensor of an automobile provided by the present invention, the scheme is as follows:

[0008] In a first aspect, a method for replacing a vehicle head temperature sensor is provided, the method comprising:

[0009] Basic model development steps: Before developing new models, develop the basic artificial intelligence algorithm model first;

[0010] Steps of the transfer learning phase: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

[0011] Preferably, the basic model development step comprises:

[0012] Step S1.1: Collect training data, where the training data includes vehicle internal and external environment data and vehicle number data that may affect the change of head breathing point temperature in the vehicle;

[0013] Step S1.2: Encode, clean, and normalize the collected training data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed data into training set, validation set, and test set;

[0014] Step S1.3: Build a basic artificial intelligence algorithm model;

[0015] Step S1.4: training the basic artificial intelligence algorithm model;

[0016] Step S1.5: Test and evaluate the basic artificial intelligence algorithm model.

[0017] Preferably, the step S1.2 comprises:

[0018] Perform humidity / 10 operation on the humidity data in the car, perform (sun_level / 1000)*1.5 operation on the light intensity data sun_level, and sum the front left, front right, rear left, rear center, and rear right of the vehicle occupancy signal to obtain the final data on the number of people in the car;

[0019] Clean all collected temperature data and keep all temperature data to 0.1 accuracy;

[0020] Clean the light data and humidity data and convert them into integers;

[0021] All temperature data are processed for outliers, and the table is looked up sequentially at every 3s interval, and data with a sudden change of more than 0.5 degrees before and after is deleted;

[0022] The processed data is divided into training set, validation set and test set in a ratio of 2:1:1.

[0023] Preferably, the step S1.3 comprises:

[0024] Use supervised learning to build a basic artificial intelligence algorithm model;

[0025] The outside temperature, the inside temperature, the air-conditioning level, the air-conditioning mode, the air-conditioning internal and external circulation status, the vehicle speed, the evaporator temperature, the AC status, the number of people in the vehicle, the humidity in the vehicle, and the light intensity are used as input features of the basic artificial intelligence algorithm model, and the left breathing point temperature and the right breathing point temperature in the vehicle are used as output labels;

[0026] Use decision trees to build the structure of the basic artificial intelligence algorithm model, and control the size of the decision tree by the number of trees, tree depth, and number of leaf nodes;

[0027] Using loss function Measure the deviation between the model's output value and the true label value;

[0028] Among them, y represents the true label value, Represents the predicted value of the model during training, and n represents the total number of samples of training data;

[0029] use Controls the complexity of the model, k represents the number of trees in the model;

[0030] by As the overall optimization index of the model, ω(f k ) represents the complexity control function of each tree.

[0031] Preferably, the step S1.4 comprises:

[0032] Divide 50% of the independent data as training samples and 25% of the independent data as verification samples during the training process, set the maximum number of training rounds, adopt an early stopping strategy to avoid overfitting of the basic artificial intelligence algorithm model, and save the optimal weight of the basic artificial intelligence algorithm model during the training process.

[0033] Preferably, the step S1.5 comprises:

[0034] Divide 25% of the independent data as test samples for the basic artificial intelligence algorithm model, load the optimal weights trained in step S1.4 for prediction, and compare the prediction error distribution of the basic artificial intelligence algorithm model. Take the absolute value of the prediction of the basic artificial intelligence algorithm model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 are the data that meet the requirements;

[0035] If, after test evaluation, the sample size with a prediction error between 0 and 2 accounts for 90% of the total test sample size, the development of the basic artificial intelligence algorithm model is considered to meet the requirements, and the best weight model is saved as best_model.

[0036] Preferably, the transfer learning stage steps include:

[0037] Step S2.1: Collect data of the target vehicle model, including data of the internal and external environment of the vehicle and data of the number of people in the vehicle that may affect the change of the temperature at the head breathing point in the vehicle of the target vehicle model;

[0038] Step S2.2: Encode, clean, and normalize the collected target vehicle model data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed target vehicle model data into a training set, a validation set, and a test set;

[0039] Step S2.3: Perform AI algorithm model transfer learning for the target vehicle model;

[0040] Step S2.4: Perform migration test and evaluation on the artificial intelligence algorithm model of the target vehicle model.

[0041] Preferably, the step S2.3 comprises:

[0042] Based on the model best_model, transfer learning is performed, the left breathing point temperature and the right breathing point temperature of the target vehicle model are used as output labels, and other data except these two data are used as input features, the best model best_model is loaded, and the parameters of the best model best_model including the number of trees, tree depth, number of leaf nodes, and learning rate are fine-tuned to obtain the best model finetune_best_model;

[0043] Using the objective function As an optimization target for transfer learning models;

[0044] Divide 50% of the target vehicle data into an independent training set for transfer learning, and 25% of the target vehicle data as a validation set. Set the maximum number of training rounds, adopt an early stopping strategy to prevent the best model finetune_best_model from overfitting, and save the optimal weight of the best model finetune_best_model in the transfer training process.

[0045] Preferably, the step S2.4 comprises:

[0046] Divide 25% of the independent target vehicle data as test samples for the transfer learning model, load the optimal weights trained in step 2.3 for prediction, and compare the prediction error distribution of the best model finetune_best_model. Take the absolute value of the prediction of the best model finetune_best_model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 is the data that meets the requirements;

[0047] If after test evaluation, the sample size with prediction errors between 0-2 accounts for 90% of the total test sample size, it is considered that the development of the transfer learning model for the target vehicle model meets the requirements, and the optimal weight model finetune_best_model is saved. The obtained optimal weight model finetune_best_model is used as the final target vehicle model head breathing point temperature prediction model, realizing the use of artificial intelligence algorithm to fit the head breathing point temperature of the target vehicle model.

[0048] In a second aspect, a vehicle head temperature sensor replacement system is provided, the system comprising:

[0049] Basic model development module: before developing new models, develop the basic artificial intelligence algorithm model first;

[0050] Transfer learning stage module: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] The present invention uses an artificial intelligence algorithm to train the whole vehicle data related to the head breathing point temperature in the vehicle and the collected head breathing point temperature data, and successfully uses the software algorithm to replace the actual physical sensor, achieving the effect of reducing costs and increasing efficiency.

[0053] Other beneficial effects of the present invention will be explained in the specific implementation manner through the introduction of specific technical features and technical solutions. Through the introduction of these technical features and technical solutions, those skilled in the art should be able to understand the beneficial technical effects brought about by the technical features and technical solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0055] Figure 1 It is a schematic diagram of the overall process of the present invention. DETAILED DESCRIPTION

[0056] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0057] The embodiment of the present invention provides a method for replacing a vehicle head temperature sensor, referring to Figure 1 As shown, the method specifically includes the following contents:

[0058] Basic model development steps: Before developing new models, develop the basic artificial intelligence algorithm model first;

[0059] Steps of the transfer learning phase: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

[0060] Specifically, the basic model development steps include:

[0061] Step S1.1: Collect training data, where the training data includes vehicle internal and external environment data and vehicle number data that may affect the change of head breathing point temperature in the vehicle;

[0062] Step S1.2: Encode, clean, and normalize the collected training data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed data into training set, verification set, and test set.

[0063] Perform humidity / 10 operation on the humidity data in the car, perform (sun_level / 1000)*1.5 operation on the light intensity data sun_level, and sum the front left, front right, rear left, rear center, and rear right of the vehicle occupancy signal to obtain the final data on the number of people in the car;

[0064] Clean all collected temperature data and keep all temperature data to 0.1 accuracy;

[0065] Clean the light data and humidity data and convert them into integers;

[0066] All temperature data are processed for outliers, and the table is looked up sequentially at every 3s interval, and data with a sudden change of more than 0.5 degrees before and after is deleted;

[0067] The processed data is divided into training set, validation set and test set in a ratio of 2:1:1.

[0068] Step S1.3: Build a basic artificial intelligence algorithm model.

[0069] Use supervised learning to build a basic artificial intelligence algorithm model;

[0070] The outside temperature, the inside temperature, the air-conditioning level, the air-conditioning mode, the air-conditioning internal and external circulation status, the vehicle speed, the evaporator temperature, the AC status, the number of people in the vehicle, the humidity in the vehicle, and the light intensity are used as input features of the basic artificial intelligence algorithm model, and the left breathing point temperature and the right breathing point temperature in the vehicle are used as output labels;

[0071] Use decision trees to build the structure of the basic artificial intelligence algorithm model, and control the size of the decision tree by the number of trees, tree depth, and number of leaf nodes;

[0072] Using loss function Measure the deviation between the model's output value and the true label value;

[0073] Among them, y represents the true label value, Represents the predicted value of the model during training, and n represents the total number of samples of training data;

[0074] use Controls the complexity of the model, k represents the number of trees in the model;

[0075] by As the overall optimization index of the model, ω(f k ) represents the complexity control function of each tree.

[0076] Step S1.4: Train the basic artificial intelligence algorithm model.

[0077] Divide 50% of the independent data as training samples and 25% of the independent data as verification samples during the training process, set the maximum number of training rounds, adopt an early stopping strategy to avoid overfitting of the basic artificial intelligence algorithm model, and save the optimal weight of the basic artificial intelligence algorithm model during the training process.

[0078] Step S1.5: Test and evaluate the basic artificial intelligence algorithm model.

[0079] Divide 25% of the independent data as test samples for the basic artificial intelligence algorithm model, load the optimal weights trained in step S1.4 for prediction, and compare the prediction error distribution of the basic artificial intelligence algorithm model. Take the absolute value of the prediction of the basic artificial intelligence algorithm model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 are the data that meet the requirements;

[0080] If, after test evaluation, the sample size with a prediction error between 0 and 2 accounts for 90% of the total test sample size, the development of the basic artificial intelligence algorithm model is considered to meet the requirements, and the best weight model is saved as best_model.

[0081] Specifically, the steps of the transfer learning phase include:

[0082] Step S2.1: Collect data of the target vehicle model, including data of the internal and external environment of the vehicle and data of the number of people in the vehicle that may affect the change of the temperature at the head breathing point in the vehicle of the target vehicle model;

[0083] Step S2.2: Encode, clean, and normalize the collected target vehicle model data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed target vehicle model data into a training set, a validation set, and a test set;

[0084] Step S2.3: Perform AI algorithm model transfer learning for the target vehicle model.

[0085] Based on the model best_model, transfer learning is performed, the left breathing point temperature and the right breathing point temperature of the target vehicle model are used as output labels, and other data except these two data are used as input features, the best model best_model is loaded, and the parameters of the best model best_model including the number of trees, tree depth, number of leaf nodes, and learning rate are fine-tuned to obtain the best model finetune_best_model;

[0086] Using the objective function As an optimization target for transfer learning models;

[0087] Divide 50% of the target vehicle data into an independent training set for transfer learning, and 25% of the target vehicle data as a validation set. Set the maximum number of training rounds, adopt an early stopping strategy to prevent the best model finetune_best_model from overfitting, and save the optimal weight of the best model finetune_best_model in the transfer training process.

[0088] Step S2.4: Perform migration test and evaluation on the artificial intelligence algorithm model of the target vehicle model.

[0089] Divide 25% of the independent target vehicle data as test samples for the transfer learning model, load the optimal weights trained in step 2.3 for prediction, and compare the prediction error distribution of the best model finetune_best_model. Take the absolute value of the prediction of the best model finetune_best_model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 is the data that meets the requirements;

[0090] If after test evaluation, the sample size with prediction errors between 0-2 accounts for 90% of the total test sample size, it is considered that the development of the transfer learning model for the target vehicle model meets the requirements, and the optimal weight model finetune_best_model is saved, and the obtained optimal weight model finetune_best_model is used as the final target vehicle model head breathing point temperature prediction model, realizing the use of artificial intelligence algorithm to fit the head breathing point temperature of the target vehicle model.

[0091] The present invention also provides a vehicle head temperature sensor replacement system, which can be implemented by executing the process steps of the vehicle head temperature sensor replacement method, that is, those skilled in the art can understand the vehicle head temperature sensor replacement method as a preferred implementation of the vehicle head temperature sensor replacement system. The system specifically includes:

[0092] Basic model development module: before developing new models, develop the basic artificial intelligence algorithm model first;

[0093] Transfer learning stage module: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

[0094] Next, the present invention will be described in more detail.

[0095] Step 1: Basic model development phase: Before developing new models, develop a basic artificial intelligence algorithm model with the ability to predict the head temperature of existing models, which can predict the head temperature data of existing models.

[0096] Step 2: Transfer learning stage. Before the delivery of new models, the basic artificial intelligence algorithm model developed in step 1 is used to fine-tune and optimize the artificial intelligence algorithm model for new models, so that the head temperature data of existing models can be predicted.

[0097] Step 1 includes the following steps:

[0098] Step 1.1: Collect the training data required for basic model development. The training data should include the in-vehicle and out-vehicle environment data that can affect the temperature change of the head breathing point in the vehicle and the number of people in the vehicle;

[0099] The specific implementation steps of step 1.1 include recording the vehicle data at each moment, the left and right head breathing point temperature data collected by the temperature sensor, the vehicle light intensity data, and the vehicle humidity data. The required data are shown in Table 1-1 below.

[0100] Table 1-1

[0101]

[0102]

[0103] Step 1.2: Data encoding, cleaning, normalization, and division into training set and validation set. Encode, clean, and normalize the collected data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed data into training set, validation set, and test set;

[0104] The specific implementation of step 1.2 includes performing humidity / 10 operation on the humidity data humidity in the car, performing (Sun_level / 1000)*1.5 operation on the light intensity data sun_level, and summing the front left, front right, rear left, rear center, and rear right of the vehicle occupancy signal as the final data on the number of people in the car. Clean all the collected temperature data and retain all temperature data to 0.1 accuracy; clean the light data and humidity data and convert them into integers. Perform outlier processing on all temperature data, perform sequential table lookup at every 3s time interval, and delete the data with a mutation of more than 0.5 degrees before and after. Divide the processed data into training set, validation set, and test set in a ratio of 2:1:1.

[0105] Step 1.3: Build an artificial intelligence algorithm model for predicting head breathing point temperature.

[0106] The specific implementation of step 1.3 includes building an artificial intelligence algorithm model for predicting the head breathing point temperature by means of supervised learning. The outside temperature, inside temperature, air conditioning outlet level, air conditioning outlet mode, air conditioning internal and external circulation status, vehicle speed, evaporator temperature, air conditioning AC status, number of people in the car, humidity in the car, and light intensity are used as input features of the artificial intelligence algorithm model for predicting the head breathing point temperature. The left breathing point temperature in the car and the right breathing point temperature in the car are used as output labels. The decision tree is used to construct the structure of the artificial intelligence algorithm model for predicting the head breathing point temperature. The size of the decision tree is controlled by the number of trees, tree depth, and number of leaf nodes. Using the loss function Where y represents the true label value, It represents the predicted value of the model during the training process, and measures the deviation between the output value of the model and the true label value. n represents the total number of samples of the training data. Controls the complexity of the model, and k represents the number of trees in the model. As the overall optimization indicator of the model.

[0107] Step 1.4: Model training;

[0108] The specific implementation of step 1.4 includes dividing 50% of the independent data as training samples, 25% of the independent data as validation samples during the training process, setting the maximum number of training rounds, adopting an early stopping strategy to avoid model overfitting, and saving the optimal weights of the model during the training process.

[0109] Step 1.5: Model testing and evaluation;

[0110] The specific implementation of step 1.5 includes dividing 25% of the independent data as the test samples of the model, loading the optimal weights trained in step 1.4 for prediction, and comparing the prediction error distribution of the model, taking the absolute value of the model's prediction and the true label value as the prediction error, and counting the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 are the data that meet the requirements. If, after test evaluation, the number of samples with prediction errors between 0-2 accounts for 90% of the total test samples, it is considered that the development of the basic model meets the requirements, and the best weight model is saved as best_model1.

[0111] During transfer learning, step 2 includes the following steps:

[0112] Step 2.1: Collect data on the target vehicle model, including data on the internal and external environment of the vehicle and data on the number of people in the vehicle that may affect the change in the temperature of the head breathing point in the target vehicle model.

[0113] The specific implementation steps of step 2.1 include recording the whole vehicle data at each moment, the left and right head breathing point temperature data collected by the temperature sensor, the interior light intensity data, and the interior humidity data. The required data are consistent with the data required in step 1.1, and the target vehicle model is allowed to miss individual data due to differences in vehicle configuration.

[0114] Step 2.2: Data encoding, cleaning, normalization, and division into training set and validation set. Encode, clean, and normalize the collected target vehicle data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed data into training set, validation set, and test set.

[0115] The specific implementation of step 2.2 includes performing humidity / 10 operation on the humidity data of the target vehicle model, performing (sun_level / 1000)*1.5 operation on the light intensity data sun_level, and summing the front left, front right, rear left, rear center, and rear right of the vehicle occupancy signal as the final data on the number of people in the vehicle. Clean all the collected temperature data and retain all temperature data to 0.1 accuracy; clean the light data and humidity data and convert them into integers. Perform outlier processing on all temperature data, perform sequential table lookup at every 3s time interval, and delete the data with a mutation of more than 0.5 degrees before and after. Divide the processed data into training set, validation set, and test set in a ratio of 2:1:1.

[0116] Step 2.3: Transfer learning of the artificial intelligence algorithm model for predicting the head breathing point temperature of the target vehicle model.

[0117] The specific implementation of step 2.3 includes transfer learning based on the best model best_model1 selected after training, verification, and test evaluation in 1.5, taking the left breathing point temperature and the right breathing point temperature of the target vehicle model as output labels, and other data other than these two data as input features, loading the best model best_model1, and fine-tuning the parameters of the model including the number of trees, tree depth, number of leaf nodes, and learning rate. Using the objective function As the optimization target of the transfer learning model. Separate 50% of the target vehicle data as the training set for transfer learning, and 25% of the target vehicle data as the validation set. Set the maximum number of training rounds, adopt an early stopping strategy to avoid model overfitting, and save the optimal weight of the model during the transfer training process.

[0118] Step 2.4: Migration test evaluation.

[0119] The specific implementation of step 2.4 includes dividing 25% of the independent target vehicle data as the test samples of the transfer learning model, loading the optimal weights trained in step 2.3 for prediction, and comparing the prediction error distribution of the model, taking the absolute value of the model's prediction and the true label value as the prediction error, and counting the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 are the data that meet the requirements. If, after test evaluation, the number of samples with prediction errors between 0-2 accounts for 90% of the total test samples, it is considered that the development of the transfer learning model for the target vehicle meets the requirements, and the best weight model is saved as best_model2.

[0120] The embodiments of the present invention provide a method and system for replacing an automobile head temperature sensor. By adopting the data fitting capability of an artificial intelligence algorithm, it solves the cost reduction needs of many automobile companies to replace the traditional method of collecting head breathing point temperature by actual physical hardware with algorithm software.

[0121] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0122] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for replacing a car head temperature sensor, characterized in that: include: Basic model development steps: Before developing new models, develop the basic artificial intelligence algorithm model first; Steps of the transfer learning phase: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

2. The method for replacing the automobile head temperature sensor according to claim 1, characterized in that: The basic model development steps include: Step S1.1: Collect training data, where the training data includes vehicle internal and external environment data and vehicle number data that may affect the change of head breathing point temperature in the vehicle; Step S1.2: Encode, clean, and normalize the collected training data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed data into training set, validation set, and test set; Step S1.3: Build a basic artificial intelligence algorithm model; Step S1.4: training the basic artificial intelligence algorithm model; Step S1.5: Test and evaluate the basic artificial intelligence algorithm model.

3. The method for replacing the automobile head temperature sensor according to claim 2, characterized in that: The step S1.2 comprises: Perform humidity / 10 operation on the humidity data in the car, perform (sun_level / 1000)*1.5 operation on the light intensity data sun_level, and sum the front left, front right, rear left, rear center, and rear right of the vehicle occupancy signal to obtain the final data on the number of people in the car; Clean all collected temperature data and keep all temperature data to 0.1 accuracy; Clean the light data and humidity data and convert them into integers; All temperature data are processed for outliers, and the table is looked up sequentially at every 3s interval, and data with a sudden change of more than 0.5 degrees before and after is deleted; The processed data is divided into training set, validation set and test set in a ratio of 2:1:

1.

4. The method for replacing the automobile head temperature sensor according to claim 2, characterized in that: The step S1.3 comprises: Use supervised learning to build a basic artificial intelligence algorithm model; The outside temperature, the inside temperature, the air-conditioning level, the air-conditioning mode, the air-conditioning internal and external circulation status, the vehicle speed, the evaporator temperature, the AC status, the number of people in the vehicle, the humidity in the vehicle, and the light intensity are used as input features of the basic artificial intelligence algorithm model, and the left breathing point temperature and the right breathing point temperature in the vehicle are used as output labels; Use decision trees to build the structure of the basic artificial intelligence algorithm model, and control the size of the decision tree by the number of trees, tree depth, and number of leaf nodes; Using loss function Measure the deviation between the model's output value and the true label value; Among them, y represents the true label value, Represents the predicted value of the model during training, and n represents the total number of samples of training data; use Controls the complexity of the model, k represents the number of trees in the model; by As the overall optimization index of the model, ω(f k ) represents the complexity control function of each tree.

5. The method for replacing the automobile head temperature sensor according to claim 2, characterized in that: The step S1.4 comprises: Divide 50% of the independent data as training samples and 25% of the independent data as verification samples during the training process, set the maximum number of training rounds, adopt an early stopping strategy to avoid overfitting of the basic artificial intelligence algorithm model, and save the optimal weight of the basic artificial intelligence algorithm model during the training process.

6. The method for replacing the automobile head temperature sensor according to claim 5, characterized in that: The step S1.5 comprises: Divide 25% of the independent data as test samples for the basic artificial intelligence algorithm model, load the optimal weights trained in step S1.4 for prediction, and compare the prediction error distribution of the basic artificial intelligence algorithm model. Take the absolute value of the prediction of the basic artificial intelligence algorithm model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 are the data that meet the requirements; If, after test evaluation, the sample size with a prediction error between 0 and 2 accounts for 90% of the total test sample size, the development of the basic artificial intelligence algorithm model is considered to meet the requirements, and the best weight model is saved as best_model.

7. The method for replacing the automobile head temperature sensor according to claim 6, characterized in that: The steps of the transfer learning phase include: Step S2.1: Collect data of the target vehicle model, including data of the internal and external environment of the vehicle and data of the number of people in the vehicle that may affect the change of the temperature at the head breathing point in the vehicle of the target vehicle model; Step S2.2: Encode, clean, and normalize the collected target vehicle model data to convert it into data that can be read by the artificial intelligence algorithm, and divide the processed target vehicle model data into a training set, a validation set, and a test set; Step S2.3: Perform AI algorithm model transfer learning for the target vehicle model; Step S2.4: Perform migration test and evaluation on the artificial intelligence algorithm model of the target vehicle model.

8. The method for replacing the automobile head temperature sensor according to claim 7, characterized in that: The step S2.3 comprises: Based on the model best_model, transfer learning is performed, the left breathing point temperature and the right breathing point temperature of the target vehicle model are used as output labels, and other data except these two data are used as input features, the best model best_model is loaded, and the parameters of the best model best_model including the number of trees, tree depth, number of leaf nodes, and learning rate are fine-tuned to obtain the best model finetune_best_model; Using the objective function As an optimization target for transfer learning models; Divide 50% of the target vehicle data into an independent training set for transfer learning, and 25% of the target vehicle data as a validation set. Set the maximum number of training rounds, adopt an early stopping strategy to prevent the best model finetune_best_model from overfitting, and save the optimal weight of the best model finetune_best_model in the transfer training process.

9. The method for replacing the automobile head temperature sensor according to claim 8, characterized in that: The step S2.4 comprises: Divide 25% of the independent target vehicle data as test samples for the transfer learning model, load the optimal weights trained in step 2.3 for prediction, and compare the prediction error distribution of the best model finetune_best_model. Take the absolute value of the prediction of the best model finetune_best_model and the true label value as the prediction error, and count the number of samples with prediction errors between 0-2 and greater than 2. The data with prediction errors between 0-2 is the data that meets the requirements; If after test evaluation, the sample size with prediction errors between 0-2 accounts for 90% of the total test sample size, it is considered that the development of the transfer learning model for the target vehicle model meets the requirements, and the optimal weight model finetune_best_model is saved. The obtained optimal weight model finetune_best_model is used as the final target vehicle model head breathing point temperature prediction model, realizing the use of artificial intelligence algorithm to fit the head breathing point temperature of the target vehicle model.

10. A car head temperature sensor replacement system, characterized in that: include: Basic model development module: before developing new models, develop the basic artificial intelligence algorithm model first; Transfer learning stage module: Before the delivery of a new model, based on the basic artificial intelligence algorithm model, the artificial intelligence algorithm model of the new model is fine-tuned and optimized through transfer learning to realize the prediction of the head temperature data of the model.

Images