CN116901650A - Automobile air conditioner control method and system based on environment, CAN signal and occupant habit - Google Patents

Abstract

The application provides an automobile air conditioner control method and system based on environment, CAN signals and occupant habits, comprising the following steps: step 1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup; step 2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory; step 3: and (3) providing personalized air conditioner use setting recommendation by optimizing the general adaptive scheme in the step 2 for learning the air conditioner use habit of the passenger. According to the application, the predicted wind mode, the wind quantity and the air outlet temperature are estimated by training the regression decision tree, and the recommendation probability of the setting scheme is increased or reduced according to whether the current setting recommendation is accepted by the passenger after the prediction is recommended to the passenger, so that the accurate recommendation of the air conditioner setting scheme of the passenger is realized, and the comfort of the passenger is improved.

Description

Automobile air conditioner control method and system based on environment, CAN signal and occupant habit

Technical Field

The application relates to the technical field of automatic air conditioner adjustment, in particular to an automobile air conditioner control method and system based on environment, CAN signals and occupant habits.

Background

At present, the traditional automatic air conditioner cannot meet the personalized requirements of specific users, the users frequently operate the air conditioner, driving safety can be influenced, and the user requirements cannot be accurately predicted, so that resource waste is easily caused.

In the prior art (China patent with application number 202010725871.5, which discloses a smart air conditioner control method and system), the current temperature and humidity of a body are calculated according to the current outdoor environment data and the current time information; reading historical outdoor environment data in a database; predicting the outdoor temperature and humidity of the next time sequence according to the current and historical outdoor environment data; according to the current indoor environment data, the current temperature and humidity and the outdoor temperature and humidity of the next time sequence, calculating to obtain a control instruction through a current control algorithm model; the control command is sent to the air conditioning system to enable the air conditioning system to adjust temperature setting according to the control command; and acquiring the energy consumption value and the comfort level of the air conditioning system after the temperature setting is regulated, and updating the control algorithm model according to the energy consumption value and the comfort level. Although the patent can predict the outdoor temperature and humidity in the future and calculate the air conditioner control command according to the algorithm model, the sunlight intensity, the geographic position information and the air conditioner using habit of a driver are not considered.

Disclosure of Invention

Aiming at the defects in the prior art, the application aims to provide an automobile air conditioner control method and system based on environment, CAN signals and occupant habits.

The automobile air conditioner control method based on the environment, the CAN signal and the passenger habit provided by the application comprises the following steps:

step 1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup;

step 2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory;

step 3: and (3) providing personalized air conditioner use setting recommendation by optimizing the general adaptive scheme in the step 2 for learning the air conditioner use habit of the passenger.

Preferably, the step 1 includes:

step 1.1: calibrating vehicle-end CAN signal parameters by a calibration engineer, wherein the parameters comprise vehicle interior environment parameters and an air conditioner parameter curve; the in-vehicle environmental parameters include in-vehicle temperature, PM2.5 value, and sunlight intensity; the air conditioner parameter curve comprises an air conditioner air outlet temperature curve, an air conditioner air outlet air quantity curve and an air conditioner air outlet mode;

step 1.2: generating a multi-dimensional table according to different weather parameters and geographic information, wherein each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and searching a table to find out a corresponding recommended air conditioning scheme according to the table when the vehicle is first used; the weather parameters include temperature, somatosensory temperature, relative humidity, air pressure and weather; the geographic information includes city, longitude, latitude, and altitude;

the step 2 comprises the following steps:

step 2.1: acquiring air conditioning scheme data used by other users through a cloud platform or a user database;

step 2.2: training the big data obtained in the step 2.1 into a general model, and combining the vehicle-end CAN signal parameters calibrated in the step 1.1 to prepare a general adaptive air conditioning scheme suitable for the vehicle, so as to replace the multidimensional table in the step 1.2 and recommend the general adaptive air conditioning scheme to a user.

Preferably, the specific process of the general model training comprises the following steps:

data input: converting the data into a type which can be read by the model;

data preprocessing: preprocessing input data, including data cleaning, repeated data removal and missing value processing;

feature selection: selecting proper characteristics for training a model according to the characteristics of input and output data;

model construction: constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and using a decision tree algorithm;

model evaluation: evaluating the model obtained through training by using an evaluation index so as to measure the performance of the model on the personalized recommended tasks of each passenger; wherein, the evaluation index comprises an accuracy rate, a recall rate and an F1 value; cross-validation techniques are used in the evaluation;

and (3) model tuning: and optimizing the model according to the evaluation result, including adjusting parameters of the decision tree, limiting the depth of the tree and pruning, and improving the performance and generalization capability of the model.

Preferably, the step 3 includes:

step 3.1: the habit of using the air conditioner by the user is collected and stored, wherein the habit comprises the air outlet temperature, the air quantity and the mode of the air conditioner, the predicted air mode, the air quantity and the temperature are estimated through training a decision regression tree under the conditions of different in-vehicle environment parameters, weather parameters and geographic information, the result is recommended to the user, the probability of the recommended scheme is increased or reduced later according to whether the recommended scheme is accepted by the user, and after the user accepts a plurality of recommended scheme settings, the frequency and the duration of using each air conditioner by the user are recorded;

step 3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition;

step 3.3: and (3) taking the optimal scheme in the step (3.2) as a substitute for the universal scheme under the same condition in the step (2) and recommending the optimal scheme to a user, continuously and circularly executing the step (3.1) and the step (3.2), and continuously updating the recommended scheme.

Preferably, the specific process of training the decision regression tree is as follows:

data preparation: collecting a dataset for training a model;

feature selection: selecting characteristics for training according to task requirements and data characteristics, wherein the characteristics comprise information gain and a base index;

model construction: constructing a regression tree model according to the selected characteristics and the target variable by utilizing a decision tree algorithm, wherein the decision tree algorithm divides the data set into different subsets according to the values of the characteristics so as to minimize the mean square error or other loss functions of the target variable;

and (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace corresponding to a decision node which divides the dataset based on a certain division point of a certain feature;

and (3) recursion construction: repeatedly executing model construction and feature division on each subspace, constructing deeper decision nodes until stopping conditions are met, wherein the maximum depth is reached and the number of samples contained in the nodes is smaller than a threshold value;

stop condition: defining a stopping condition to avoid overfitting in the process of constructing a regression decision tree, wherein the stopping condition comprises the steps of setting the maximum depth, setting the number of samples contained in the nodes to be smaller than a threshold value and setting the purity of the nodes to reach a preset threshold value;

model evaluation: and evaluating the model by using evaluation indexes comprising mean square error and average absolute error, measuring the fitting degree of the model on training data, and evaluating the generalization capability of the model on unseen data by using a cross-validation technology.

The automobile air conditioner control system based on the environment, the CAN signal and the habit of passengers provided by the application comprises:

module M1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup;

module M2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory;

module M3: personalized air conditioner use setting recommendation is provided through a generic scheme in the learning optimization module M2 for the air conditioner use habit of the passenger.

Preferably, the module M1 comprises:

module M1.1: calibrating vehicle-end CAN signal parameters by a calibration engineer, wherein the parameters comprise vehicle interior environment parameters and an air conditioner parameter curve; the in-vehicle environmental parameters include in-vehicle temperature, PM2.5 value, and sunlight intensity; the air conditioner parameter curve comprises an air conditioner air outlet temperature curve, an air conditioner air outlet air quantity curve and an air conditioner air outlet mode;

module M1.2: generating a multi-dimensional table according to different weather parameters and geographic information, wherein each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and searching a table to find out a corresponding recommended air conditioning scheme according to the table when the vehicle is first used; the weather parameters include temperature, somatosensory temperature, relative humidity, air pressure and weather; the geographic information includes city, longitude, latitude, and altitude;

the module M2 includes:

module M2.1: acquiring air conditioning scheme data used by other users through a cloud platform or a user database;

module M2.2: training the big data acquired in the module M2.1 into a general model, and combining the vehicle-end CAN signal parameters calibrated in the module M1.1 to prepare a general adaptive air conditioning scheme suitable for the vehicle, so as to replace the multidimensional form in the module M1.2 and recommend the general adaptive air conditioning scheme to a user.

Preferably, the specific process of the general model training comprises the following steps:

data input: converting the data into a type which can be read by the model;

data preprocessing: preprocessing input data, including data cleaning, repeated data removal and missing value processing;

feature selection: selecting proper characteristics for training a model according to the characteristics of input and output data;

model construction: constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and using a decision tree algorithm;

model evaluation: evaluating the model obtained through training by using an evaluation index so as to measure the performance of the model on the personalized recommended tasks of each passenger; wherein, the evaluation index comprises an accuracy rate, a recall rate and an F1 value; cross-validation techniques are used in the evaluation;

and (3) model tuning: and optimizing the model according to the evaluation result, including adjusting parameters of the decision tree, limiting the depth of the tree and pruning, and improving the performance and generalization capability of the model.

Preferably, the module M3 includes:

module M3.1: the habit of using the air conditioner by the user is collected and stored, wherein the habit comprises the air outlet temperature, the air quantity and the mode of the air conditioner, the predicted air mode, the air quantity and the temperature are estimated through training a decision regression tree under the conditions of different in-vehicle environment parameters, weather parameters and geographic information, the result is recommended to the user, the probability of the recommended scheme is increased or reduced later according to whether the recommended scheme is accepted by the user, and after the user accepts a plurality of recommended scheme settings, the frequency and the duration of using each air conditioner by the user are recorded;

module M3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition;

module M3.3: and taking the optimal scheme in the module M3.2 as a general adaptive scheme under the same condition in the module M2 and recommending the general adaptive scheme to a user, continuously and circularly calling the module M3.1 and the module M3.2, and continuously updating the recommended scheme.

Preferably, the specific process of training the decision regression tree is as follows:

data preparation: collecting a dataset for training a model;

feature selection: selecting characteristics for training according to task requirements and data characteristics, wherein the characteristics comprise information gain and a base index;

model construction: constructing a regression tree model according to the selected characteristics and the target variable by utilizing a decision tree algorithm, wherein the decision tree algorithm divides the data set into different subsets according to the values of the characteristics so as to minimize the mean square error or other loss functions of the target variable;

and (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace corresponding to a decision node which divides the dataset based on a certain division point of a certain feature;

and (3) recursion construction: repeatedly executing model construction and feature division on each subspace, constructing deeper decision nodes until stopping conditions are met, wherein the maximum depth is reached and the number of samples contained in the nodes is smaller than a threshold value;

stop condition: defining a stopping condition to avoid overfitting in the process of constructing a regression decision tree, wherein the stopping condition comprises the steps of setting the maximum depth, setting the number of samples contained in the nodes to be smaller than a threshold value and setting the purity of the nodes to reach a preset threshold value;

model evaluation: and evaluating the model by using evaluation indexes comprising mean square error and average absolute error, measuring the fitting degree of the model on training data, and evaluating the generalization capability of the model on unseen data by using a cross-validation technology.

Compared with the prior art, the application has the following beneficial effects:

in different use stages of the vehicle, the most suitable air conditioner use setting is finally recommended to the user by combining information such as environment, user big data and the like and utilizing a regression decision tree method; the predicted wind mode, the wind quantity and the air outlet temperature are estimated by training a regression decision tree, and the recommendation probability of the setting scheme is increased or reduced according to whether the passenger accepts the current setting recommendation after the current setting recommendation is recommended to the passenger, so that the accurate recommendation of the air conditioner setting scheme of the passenger is realized, and the comfort of the passenger is improved.

Drawings

Other features, objects and advantages of the present application will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:

FIG. 1 is a flow chart of the method of the present application;

FIG. 2 is a flow chart of a lookup recommended air conditioning scheme.

Detailed Description

The present application will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the present application, but are not intended to limit the application in any way. It should be noted that variations and modifications could be made by those skilled in the art without departing from the inventive concept. These are all within the scope of the present application.

Example 1:

as shown in fig. 1, the application provides a method for recommending an air outlet mode, air quantity and air outlet temperature of an air conditioner by using an artificial intelligent algorithm to predict the use setting of the air conditioner based on weather information, a vehicle-end CAN signal and occupant habits, which comprises the following specific steps:

step 1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup;

the step 1 comprises the following steps:

step 1.1: the vehicle end CAN signal parameters are calibrated by a calibration engineer, and the parameters comprise vehicle internal environment parameters (vehicle internal temperature, PM2.5 value and sunlight intensity) and air conditioner parameter curves (air conditioner air outlet temperature curve, air conditioner air outlet air volume curve and air conditioner air outlet mode);

step 1.2: a multi-dimensional table is generated according to different weather parameters (temperature, body sensing temperature, relative humidity, air pressure and weather) and geographic information (city, longitude, latitude and altitude), each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and then a corresponding recommended air conditioning scheme is found out according to the table by looking up a table when the vehicle is first used. An example of this step is shown in fig. 2, and the specific implementation steps include: the method comprises the steps that an in-vehicle sensor and a data acquisition device acquire in-vehicle parameters including in-vehicle temperature, body temperature, climate and various geographic information; searching output parameters corresponding to the input parameters, namely outputting air conditioner parameters according to the calibration table, wherein the output parameters comprise set temperature, set air quantity and set air outlet mode; and taking the air conditioning parameters obtained by the table lookup as a recommended air conditioning scheme when the vehicle is used for the first time.

Step 2: recommending the general adaptive scheme to the user based on big data training;

the step 2 comprises the following steps:

step 2.1: acquiring air conditioning scheme data used by other users, namely personalized air conditioning use settings of the other users, through a cloud platform or a user database;

step 2.2: training the big data obtained in the step 2.1 into a general model, and combining the parameters of the vehicle calibrated in the step 1.1 to make a general adaptive air conditioning scheme suitable for the vehicle to replace the multidimensional table in the step 1.2 for recommending to a user.

The specific process of model training comprises the following steps:

(1) Data input, converting the data into a type which can be read by the model;

(2) Preprocessing data, namely preprocessing input data, including data cleaning, removing repeated data, processing missing values and the like;

(3) Selecting proper characteristics for training a model according to the characteristics of input data and output data;

(4) Model construction, namely constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and a decision tree algorithm;

(5) Model evaluation, wherein an evaluation index (such as accuracy, recall rate, F1 value and the like) is used for evaluating the trained model so as to measure the performance of the model on individual passenger personalized recommendation tasks, and cross verification and other technologies can be used for evaluation so as to reduce the deviation of an evaluation result;

(6) And (3) model tuning, namely tuning the model according to the evaluation result, such as adjusting parameters of a decision tree, limiting the depth of the tree, pruning and the like, so as to improve the performance and generalization capability of the model.

Step 3: providing personalized air conditioner use setting recommendation by optimizing the general adaptive scheme in the step 2 for learning the air conditioner use habit of the passenger;

the step 3 comprises the following steps:

step 3.1: and collecting and storing habits of users using the air conditioner, including air-conditioning air-out temperature, air quantity and mode, estimating and predicting the air mode, the air quantity and the air temperature through training a decision regression tree under the conditions of different in-car environment parameters, weather parameters and geographic information, recommending the results to the users, and increasing or reducing the probability of the recommended proposal according to whether the users accept the recommended proposal. After the user receives the plurality of recommended scheme settings, recording the frequency and duration of the user using each air conditioner setting;

the specific process of training the decision regression tree is as follows:

(1) Data preparation: collecting a dataset for training a model;

(2) Feature selection: according to task requirements and data characteristics, characteristics for training are selected, and common characteristic selection methods comprise information gain, a base index and the like;

(3) Model construction: constructing a regression tree model based on the selected features and the target variables using a decision tree algorithm that divides the data set into different subsets based on the values of the features to minimize the mean square error (Mean Squared Error, MSE) or other loss function of the target variables;

(4) And (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace generally corresponding to a decision node which divides the dataset based on a certain division point of a certain feature;

(5) And (3) recursion construction: repeating the step (3) and the step (4) for each subspace, and constructing decision nodes with deeper layers until stopping conditions are met, such as maximum depth is reached, the number of samples contained in the nodes is smaller than a threshold value, and the like;

(6) Stop condition: in the process of constructing the regression decision tree, stop conditions need to be defined to avoid overfitting, for example, the maximum depth can be set, the number of samples contained in the node is smaller than a threshold value, the purity of the node reaches a certain threshold value, and the like;

(7) Model evaluation: evaluating the model by using evaluation indexes (such as mean square error, average absolute error and the like), measuring the fitting degree of the model on training data, and simultaneously, evaluating the generalization capability of the model on unseen data by using technologies such as cross verification and the like;

step 3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition;

step 3.3: and substituting the optimal scheme in the step 3.2 for the universal scheme under the same condition in the step 2, and recommending the universal scheme to the user. Step 3.1 and step 3.2 are continuously and circularly executed, and the recommended scheme is continuously updated.

Example 2:

the application also provides an automobile air conditioner control system based on the environment, the CAN signal and the passenger habit, which CAN be realized by executing the flow steps of the automobile air conditioner control method based on the environment, the CAN signal and the passenger habit, namely, the automobile air conditioner control method based on the environment, the CAN signal and the passenger habit CAN be understood as a preferred implementation mode of the automobile air conditioner control system based on the environment, the CAN signal and the passenger habit by a person skilled in the art.

The automobile air conditioner control system based on the environment, the CAN signal and the habit of passengers provided by the application comprises: module M1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup; module M2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory; module M3: personalized air conditioner use setting recommendation is provided through a generic scheme in the learning optimization module M2 for the air conditioner use habit of the passenger.

The module M1 includes: module M1.1: calibrating vehicle-end CAN signal parameters by a calibration engineer, wherein the parameters comprise vehicle interior environment parameters and an air conditioner parameter curve; the in-vehicle environmental parameters include in-vehicle temperature, PM2.5 value, and sunlight intensity; the air conditioner parameter curve comprises an air conditioner air outlet temperature curve, an air conditioner air outlet air quantity curve and an air conditioner air outlet mode; module M1.2: generating a multi-dimensional table according to different weather parameters and geographic information, wherein each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and searching a table to find out a corresponding recommended air conditioning scheme according to the table when the vehicle is first used; the weather parameters include temperature, somatosensory temperature, relative humidity, air pressure and weather; the geographic information includes city, longitude, latitude, and altitude;

the module M2 includes: module M2.1: acquiring air conditioning scheme data used by other users through a cloud platform or a user database; module M2.2: training the big data acquired in the module M2.1 into a general model, and combining the vehicle-end CAN signal parameters calibrated in the module M1.1 to prepare a general adaptive air conditioning scheme suitable for the vehicle, so as to replace the multidimensional form in the module M1.2 and recommend the general adaptive air conditioning scheme to a user.

The specific process of the general model training comprises the following steps: data input: converting the data into a type which can be read by the model; data preprocessing: preprocessing input data, including data cleaning, repeated data removal and missing value processing; feature selection: selecting proper characteristics for training a model according to the characteristics of input and output data; model construction: constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and using a decision tree algorithm; model evaluation: evaluating the model obtained through training by using an evaluation index so as to measure the performance of the model on the personalized recommended tasks of each passenger; wherein, the evaluation index comprises an accuracy rate, a recall rate and an F1 value; cross-validation techniques are used in the evaluation; and (3) model tuning: and optimizing the model according to the evaluation result, including adjusting parameters of the decision tree, limiting the depth of the tree and pruning, and improving the performance and generalization capability of the model.

The module M3 includes: module M3.1: the habit of using the air conditioner by the user is collected and stored, wherein the habit comprises the air outlet temperature, the air quantity and the mode of the air conditioner, the predicted air mode, the air quantity and the temperature are estimated through training a decision regression tree under the conditions of different in-vehicle environment parameters, weather parameters and geographic information, the result is recommended to the user, the probability of the recommended scheme is increased or reduced later according to whether the recommended scheme is accepted by the user, and after the user accepts a plurality of recommended scheme settings, the frequency and the duration of using each air conditioner by the user are recorded; module M3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition; module M3.3: and taking the optimal scheme in the module M3.2 as a general adaptive scheme under the same condition in the module M2 and recommending the general adaptive scheme to a user, continuously and circularly calling the module M3.1 and the module M3.2, and continuously updating the recommended scheme.

The specific process of training the decision regression tree is as follows: data preparation: collecting a dataset for training a model; feature selection: selecting characteristics for training according to task requirements and data characteristics, wherein the characteristics comprise information gain and a base index; model construction: constructing a regression tree model according to the selected characteristics and the target variable by utilizing a decision tree algorithm, wherein the decision tree algorithm divides the data set into different subsets according to the values of the characteristics so as to minimize the mean square error or other loss functions of the target variable; and (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace corresponding to a decision node which divides the dataset based on a certain division point of a certain feature; and (3) recursion construction: repeatedly executing model construction and feature division on each subspace, constructing deeper decision nodes until stopping conditions are met, wherein the maximum depth is reached and the number of samples contained in the nodes is smaller than a threshold value; stop condition: defining a stopping condition to avoid overfitting in the process of constructing a regression decision tree, wherein the stopping condition comprises the steps of setting the maximum depth, setting the number of samples contained in the nodes to be smaller than a threshold value and setting the purity of the nodes to reach a preset threshold value; model evaluation: and evaluating the model by using evaluation indexes comprising mean square error and average absolute error, measuring the fitting degree of the model on training data, and evaluating the generalization capability of the model on unseen data by using a cross-validation technology.

Those skilled in the art will appreciate that the systems, apparatus, and their respective modules provided herein may be implemented entirely by logic programming of method steps such that the systems, apparatus, and their respective modules are implemented as logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc., in addition to the systems, apparatus, and their respective modules being implemented as pure computer readable program code. Therefore, the system, the apparatus, and the respective modules thereof provided by the present application may be regarded as one hardware component, and the modules included therein for implementing various programs may also be regarded as structures within the hardware component; modules for implementing various functions may also be regarded as being either software programs for implementing the methods or structures within hardware components.

The foregoing describes specific embodiments of the present application. It is to be understood that the application is not limited to the particular embodiments described above, and that various changes or modifications may be made by those skilled in the art within the scope of the appended claims without affecting the spirit of the application. The embodiments of the application and the features of the embodiments may be combined with each other arbitrarily without conflict.

Claims (10)

1. An automobile air conditioner control method based on environment, CAN signals and occupant habits is characterized by comprising the following steps:

step 1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup;

step 2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory;

step 3: and (3) providing personalized air conditioner use setting recommendation by optimizing the general adaptive scheme in the step 2 for learning the air conditioner use habit of the passenger.

2. The method for controlling the air conditioner of the automobile based on the environment, the CAN signal and the occupant habit according to claim 1, wherein the step 1 comprises:

step 1.1: calibrating vehicle-end CAN signal parameters by a calibration engineer, wherein the parameters comprise vehicle interior environment parameters and an air conditioner parameter curve; the in-vehicle environmental parameters include in-vehicle temperature, PM2.5 value, and sunlight intensity; the air conditioner parameter curve comprises an air conditioner air outlet temperature curve, an air conditioner air outlet air quantity curve and an air conditioner air outlet mode;

step 1.2: generating a multi-dimensional table according to different weather parameters and geographic information, wherein each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and searching a table to find out a corresponding recommended air conditioning scheme according to the table when the vehicle is first used; the weather parameters include temperature, somatosensory temperature, relative humidity, air pressure and weather; the geographic information includes city, longitude, latitude, and altitude;

the step 2 comprises the following steps:

step 2.1: acquiring air conditioning scheme data used by other users through a cloud platform or a user database;

step 2.2: training the big data obtained in the step 2.1 into a general model, and combining the vehicle-end CAN signal parameters calibrated in the step 1.1 to prepare a general adaptive air conditioning scheme suitable for the vehicle, so as to replace the multidimensional table in the step 1.2 and recommend the general adaptive air conditioning scheme to a user.

3. The method for controlling the air conditioner of the automobile based on the environment, the CAN signal and the habit of the passenger according to claim 2, wherein the specific process of the general model training comprises the following steps:

data input: converting the data into a type which can be read by the model;

data preprocessing: preprocessing input data, including data cleaning, repeated data removal and missing value processing;

feature selection: selecting proper characteristics for training a model according to the characteristics of input and output data;

model construction: constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and using a decision tree algorithm;

model evaluation: evaluating the model obtained through training by using an evaluation index so as to measure the performance of the model on the personalized recommended tasks of each passenger; wherein, the evaluation index comprises an accuracy rate, a recall rate and an F1 value; cross-validation techniques are used in the evaluation;

and (3) model tuning: and optimizing the model according to the evaluation result, including adjusting parameters of the decision tree, limiting the depth of the tree and pruning, and improving the performance and generalization capability of the model.

4. The method for controlling the air conditioner of the automobile based on the environment, the CAN signal and the occupant habit according to claim 1, wherein the step 3 comprises:

step 3.1: the habit of using the air conditioner by the user is collected and stored, wherein the habit comprises the air outlet temperature, the air quantity and the mode of the air conditioner, the predicted air mode, the air quantity and the temperature are estimated through training a decision regression tree under the conditions of different in-vehicle environment parameters, weather parameters and geographic information, the result is recommended to the user, the probability of the recommended scheme is increased or reduced later according to whether the recommended scheme is accepted by the user, and after the user accepts a plurality of recommended scheme settings, the frequency and the duration of using each air conditioner by the user are recorded;

step 3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition;

step 3.3: and (3) taking the optimal scheme in the step (3.2) as a substitute for the universal scheme under the same condition in the step (2) and recommending the optimal scheme to a user, continuously and circularly executing the step (3.1) and the step (3.2), and continuously updating the recommended scheme.

5. The method for controlling the air conditioner of the automobile based on the environment, the CAN signal and the occupant habit according to claim 4, wherein the specific process of training the decision regression tree is as follows:

data preparation: collecting a dataset for training a model;

feature selection: selecting characteristics for training according to task requirements and data characteristics, wherein the characteristics comprise information gain and a base index;

model construction: constructing a regression tree model according to the selected characteristics and the target variable by utilizing a decision tree algorithm, wherein the decision tree algorithm divides the data set into different subsets according to the values of the characteristics so as to minimize the mean square error or other loss functions of the target variable;

and (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace corresponding to a decision node which divides the dataset based on a certain division point of a certain feature;

and (3) recursion construction: repeatedly executing model construction and feature division on each subspace, constructing deeper decision nodes until stopping conditions are met, wherein the maximum depth is reached and the number of samples contained in the nodes is smaller than a threshold value;

stop condition: defining a stopping condition to avoid overfitting in the process of constructing a regression decision tree, wherein the stopping condition comprises the steps of setting the maximum depth, setting the number of samples contained in the nodes to be smaller than a threshold value and setting the purity of the nodes to reach a preset threshold value;

model evaluation: and evaluating the model by using evaluation indexes comprising mean square error and average absolute error, measuring the fitting degree of the model on training data, and evaluating the generalization capability of the model on unseen data by using a cross-validation technology.

6. An automotive air conditioning control system based on environment, CAN signals and occupant habits, comprising:

module M1: before the delivery of the vehicle, a calibration engineer provides table data, and an expert system recommends an air conditioning scheme according to table data table lookup;

module M2: recommending the universal adaptive scheme to a user based on big data training after the vehicle leaves the factory;

module M3: personalized air conditioner use setting recommendation is provided through a generic scheme in the learning optimization module M2 for the air conditioner use habit of the passenger.

7. The vehicle air conditioning control system based on the environment, CAN signal and occupant habit of claim 6, wherein said module M1 comprises:

module M1.1: calibrating vehicle-end CAN signal parameters by a calibration engineer, wherein the parameters comprise vehicle interior environment parameters and an air conditioner parameter curve; the in-vehicle environmental parameters include in-vehicle temperature, PM2.5 value, and sunlight intensity; the air conditioner parameter curve comprises an air conditioner air outlet temperature curve, an air conditioner air outlet air quantity curve and an air conditioner air outlet mode;

module M1.2: generating a multi-dimensional table according to different weather parameters and geographic information, wherein each weather parameter, geographic information and in-vehicle environment parameter correspond to one air conditioning parameter, and searching a table to find out a corresponding recommended air conditioning scheme according to the table when the vehicle is first used; the weather parameters include temperature, somatosensory temperature, relative humidity, air pressure and weather; the geographic information includes city, longitude, latitude, and altitude;

the module M2 includes:

module M2.1: acquiring air conditioning scheme data used by other users through a cloud platform or a user database;

module M2.2: training the big data acquired in the module M2.1 into a general model, and combining the vehicle-end CAN signal parameters calibrated in the module M1.1 to prepare a general adaptive air conditioning scheme suitable for the vehicle, so as to replace the multidimensional form in the module M1.2 and recommend the general adaptive air conditioning scheme to a user.

8. The vehicle air conditioning control system based on environment, CAN signal and occupant habit of claim 7, wherein the specific process of the generic model training comprises:

data input: converting the data into a type which can be read by the model;

data preprocessing: preprocessing input data, including data cleaning, repeated data removal and missing value processing;

feature selection: selecting proper characteristics for training a model according to the characteristics of input and output data;

model construction: constructing a model capable of predicting according to user characteristics and historical behaviors by using training data and using a decision tree algorithm;

model evaluation: evaluating the model obtained through training by using an evaluation index so as to measure the performance of the model on the personalized recommended tasks of each passenger; wherein, the evaluation index comprises an accuracy rate, a recall rate and an F1 value; cross-validation techniques are used in the evaluation;

and (3) model tuning: and optimizing the model according to the evaluation result, including adjusting parameters of the decision tree, limiting the depth of the tree and pruning, and improving the performance and generalization capability of the model.

9. The vehicle air conditioning control system based on the environment, CAN signal and occupant habit of claim 6, wherein said module M3 comprises:

module M3.1: the habit of using the air conditioner by the user is collected and stored, wherein the habit comprises the air outlet temperature, the air quantity and the mode of the air conditioner, the predicted air mode, the air quantity and the temperature are estimated through training a decision regression tree under the conditions of different in-vehicle environment parameters, weather parameters and geographic information, the result is recommended to the user, the probability of the recommended scheme is increased or reduced later according to whether the recommended scheme is accepted by the user, and after the user accepts a plurality of recommended scheme settings, the frequency and the duration of using each air conditioner by the user are recorded;

module M3.2: calculating scores of various air conditioner settings used by a user according to weights of various frequency and duration settings, and selecting the setting of the highest score as an optimal scheme under the current condition;

module M3.3: and taking the optimal scheme in the module M3.2 as a general adaptive scheme under the same condition in the module M2 and recommending the general adaptive scheme to a user, continuously and circularly calling the module M3.1 and the module M3.2, and continuously updating the recommended scheme.

10. The vehicle air conditioning control system based on environment, CAN signal and occupant habit of claim 9, wherein the training decision regression tree is specifically:

data preparation: collecting a dataset for training a model;

feature selection: selecting characteristics for training according to task requirements and data characteristics, wherein the characteristics comprise information gain and a base index;

model construction: constructing a regression tree model according to the selected characteristics and the target variable by utilizing a decision tree algorithm, wherein the decision tree algorithm divides the data set into different subsets according to the values of the characteristics so as to minimize the mean square error or other loss functions of the target variable;

and (3) feature division: dividing the feature space into a plurality of subspaces according to a selected division criterion, each subspace corresponding to a decision node which divides the dataset based on a certain division point of a certain feature;

and (3) recursion construction: repeatedly executing model construction and feature division on each subspace, constructing deeper decision nodes until stopping conditions are met, wherein the maximum depth is reached and the number of samples contained in the nodes is smaller than a threshold value;

stop condition: defining a stopping condition to avoid overfitting in the process of constructing a regression decision tree, wherein the stopping condition comprises the steps of setting the maximum depth, setting the number of samples contained in the nodes to be smaller than a threshold value and setting the purity of the nodes to reach a preset threshold value;

model evaluation: and evaluating the model by using evaluation indexes comprising mean square error and average absolute error, measuring the fitting degree of the model on training data, and evaluating the generalization capability of the model on unseen data by using a cross-validation technology.