Intelligent control method and system for gas source of electric vehicle

Abstract

The present application provides an intelligent control method and system for an electric vehicle gas source. The method is applied to an intelligent control system for an electric vehicle gas source. In the scheme, an intelligent control module obtains a user's target object adjustment instruction, and obtains instruction text data according to the target object adjustment instruction conversion; performs semantic analysis processing on the instruction text data to obtain the user's intention analysis result, performs sentiment analysis processing on the instruction text data to obtain the user's sentiment analysis result; inputs the intention analysis result, the sentiment analysis result and the user's user portrait into a first gas source control model to obtain the gas source pressure value and gas source flow value corresponding to the target object adjustment instruction; controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value. In the scheme, a mapping relationship between the target object adjustment instruction and the gas source pressure value and the gas source flow value is learned based on a deep learning training model to achieve intelligent, personalized and efficient adjustment of the target object.

Description

Technical Field

[0001] The present application relates to the technical field of electric vehicle gas source management, and in particular to an intelligent control method and system for electric vehicle gas source. Background Art

[0002] With the rapid development of electric vehicle technology, the demand for on-board auxiliary systems is also growing. On-board auxiliary systems can ensure the comfort of the car during driving. For example, massage chairs can relieve the fatigue of drivers or passengers.

[0003] Taking into consideration the energy consumption and safety of electric vehicles, massage chairs with airbag massage function are mostly used. In the existing method, an electric air compressor system is used to provide an air source for the massage chair, and a pressure monitor is received by a motor control unit to monitor the air pressure value in the main airway and the airbag; the air pressure value in the airbag is controlled by controlling the action of the switching element and the air source, and the air pressure of different airbags is adjusted to achieve the massage function.

[0004] However, the existing gas source control method has the problems of poor adjustment effect and inconvenient operation when adjusting the gas source. Summary of the invention

[0005] The present application provides an intelligent control method and system for an electric vehicle gas source, which is used to solve the problems of poor adjustment effect and inconvenient operation in the existing gas source control method when adjusting the gas source.

[0006] In a first aspect, the present application provides an electric vehicle gas source intelligent control method, which is applied to an electric vehicle gas source intelligent control system. The electric vehicle gas source intelligent control system includes an air compressor, an intelligent control module, an application interface module, and an integrated pipeline connecting the air compressor and the application interface module. The method includes:

[0007] The intelligent control module obtains the target object adjustment instruction of the user, and converts and obtains instruction text data according to the target object adjustment instruction;

[0008] The intelligent control module performs semantic analysis on the instruction text data to obtain the user's intention analysis result, which includes the adjustment method for the target object;

[0009] The intelligent control module performs sentiment analysis on the instruction text data to obtain the user's sentiment analysis results, which are used to determine the user's satisfaction with the current operating status of the target object;

[0010] The intelligent control module inputs the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction. The user portrait is determined according to the user's historical interaction data. The first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of the neural network model;

[0011] The intelligent control module controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value.

[0012] In a specific implementation, the intelligent control module controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value, including:

[0013] The intelligent control module generates an air source adjustment instruction corresponding to the target object adjustment instruction according to the air source pressure value and the air source flow value, and sends the air source adjustment instruction to the air compressor;

[0014] The air compressor compresses the air according to the air source adjustment instruction to obtain the target air source corresponding to the air source adjustment instruction, and transmits the target air source to the application interface module through the integrated pipeline, so that the target air source enters the target object through the application interface module and adjusts the operating state of the target object.

[0015] In a specific implementation, the electric vehicle gas source intelligent control system further includes: a server;

[0016] Before inputting the intention analysis result, the sentiment analysis result and the user portrait of the user into the first gas source control model to obtain the gas source pressure value and the gas source flow value corresponding to the target object adjustment instruction, the method further includes:

[0017] The server obtains the user's basic information data, historical target object adjustment instructions within a preset historical time period, historical gas source pressure values, and historical gas source flow values;

[0018] Determine the user's adjustment information data according to the historical target object adjustment instructions, the historical gas source pressure value and the historical gas source flow value, wherein the adjustment information data includes the user's usage habits and preference settings;

[0019] Extract features from user basic information data and adjustment information data respectively to obtain user identity features and user behavior features;

[0020] Build a user profile based on user identity characteristics and user behavior characteristics.

[0021] In a specific implementation, the electric vehicle gas source intelligent control system further includes: a server;

[0022] Before inputting the intention analysis result, the sentiment analysis result and the user portrait of the user into the first gas source control model to obtain the gas source pressure value and the gas source flow value corresponding to the target object adjustment instruction, the method further includes:

[0023] The server obtains target object adjustment instructions, user portraits, and gas source pressure values ​​and gas source flow values ​​corresponding to each target object adjustment instruction of multiple users within a preset historical time period;

[0024] For each user, the server determines the intent parsing result and sentiment analysis result corresponding to each target object adjustment instruction of the user according to each target object adjustment instruction of the user;

[0025] Based on the user portraits of multiple users in the historical time period, the intent analysis results, sentiment analysis results, gas source pressure values, and gas source flow values ​​corresponding to each target object adjustment instruction, a first sample set is constructed, wherein each sample in the first sample set includes: the user portrait, intent analysis results, sentiment analysis results during a target object adjustment, and the gas source pressure value and gas source flow value corresponding to this target object adjustment;

[0026] Based on the first sample set, a pre-constructed first initial gas source control model is trained until a preset loss function converges to obtain a first gas source control model, where the first initial gas source control model is constructed based on a deep learning neural network model.

[0027] In a specific implementation, before training the pre-constructed first initial gas source control model based on the first sample set until the preset loss function converges to obtain the first gas source control model, the method further includes:

[0028] According to the preset batch processing requirements, the first sample set is divided into sample batches to obtain a sample set for multi-batch training.

[0029] In a specific embodiment, the method further comprises:

[0030] If the adjustment method in the user's intention analysis result is empty, the intelligent control module obtains the user's historical command text data within a preset time period, the system data of the electric vehicle gas source intelligent control system, and the vehicle data; wherein the system data includes the gas source pressure, gas source temperature, compressor status, pipeline status, and the vehicle data includes the vehicle speed and acceleration;

[0031] The intelligent control module extracts features from system data, vehicle data, and instruction text data, and based on the time series and the preset feature fusion strategy, fuses the data features of system data, vehicle data, and historical instruction text data at each time point to obtain the data fusion features at each time point;

[0032] The intelligent control module inputs the data fusion features of each time point into the second air source control model in chronological order to obtain the air source pressure value and air source flow value corresponding to the target object adjustment instruction. The second air source control model is a model for determining the air source pressure value and air source flow value obtained by training the time series model.

[0033] In a specific implementation, the electric vehicle gas source intelligent control system further includes: a server;

[0034] Before inputting the data fusion features at each time point into the second gas source control model in chronological order to obtain the gas source pressure value and the gas source flow value corresponding to the target object adjustment instruction, the method further includes:

[0035] The server obtains historical instruction text data, system data, vehicle data, and gas source pressure value and gas source flow value corresponding to each historical instruction text data of multiple users within a preset historical time period;

[0036] For each user, the server performs feature fusion on each historical instruction text data, system data, and vehicle data of the user within a preset historical time period based on the time series and the preset feature fusion strategy, and determines the data fusion features at each time point;

[0037] Based on the data fusion features of multiple users at each time point in the historical time period, the gas source pressure value and the gas source flow value corresponding to the historical instruction text data at each time point, a second sample set is constructed, and each sample in the second sample set includes: the data fusion features when a target object is adjusted once, and the gas source pressure value and the gas source flow value corresponding to this target object adjustment;

[0038] Based on the second sample set, the pre-constructed second initial gas source control model is trained until the preset loss function converges to obtain the second gas source control model, and the second initial gas source control model is constructed based on a neural network model of a time series.

[0039] In a specific implementation, the preset feature fusion strategy is a strategy constructed based on the attention mechanism.

[0040] In a specific implementation, the electric vehicle gas source intelligent control system further includes: a server;

[0041] The method further includes:

[0042] The server determines the operating parameters of the air compressor according to the compressor status in the system data;

[0043] If the operating parameter is not within the preset parameter operating range, determining a parameter difference between the operating parameter of the air compressor and a parameter that satisfies the preset parameter operating range;

[0044] The actual operating state of the air compressor is adjusted according to the parameter difference, vehicle data and the air source pressure value and air source flow value output by the second air source control model.

[0045] A second aspect of the present application provides an indoor positioning system, comprising: an air compressor, an intelligent control module, an application interface module, and an integrated pipeline connecting the air compressor and the application interface module;

[0046] The intelligent control module is used to obtain the user's target object adjustment instruction and convert the instruction text data according to the target object adjustment instruction;

[0047] The intelligent control module is also used to perform semantic analysis on the instruction text data to obtain the user's intention analysis result, and the intention analysis result includes the adjustment method of the target object;

[0048] The intelligent control module is also used to perform sentiment analysis on the instruction text data to obtain the user's sentiment analysis results, and the sentiment analysis results are used to determine the user's satisfaction with the current operating status of the target object;

[0049] The intelligent control module is also used to input the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction. The user portrait is determined according to the user's historical interaction data. The first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of the neural network model;

[0050] The intelligent control module is also used to control and adjust the operating state of the target object according to the gas source pressure value and the gas source flow value.

[0051] The third aspect of the present application further provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed, they are used to implement the above first aspect and / or various possible implementation methods of the first aspect.

[0052] The fourth aspect of the present application also provides a computer program product, including a computer program, which, when executed, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0053] The present application provides an intelligent control method and system for an electric vehicle gas source. The method is applied to an intelligent control system for an electric vehicle gas source. In the scheme, the intelligent control module obtains the user's target object adjustment instruction, and obtains instruction text data according to the target object adjustment instruction conversion; performs semantic analysis on the instruction text data to obtain the user's intention analysis result, performs sentiment analysis on the instruction text data to obtain the user's sentiment analysis result; inputs the intention analysis result, the sentiment analysis result and the user's user portrait into the first gas source control model to obtain the gas source pressure value and the gas source flow value corresponding to the target object adjustment instruction; controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value. In the present scheme, the target object adjustment instruction is subjected to semantic analysis and sentiment analysis, and the mapping relationship between the target object adjustment instruction and the gas source pressure value and the gas source flow value is learned in combination with the user portrait, so that the system automatically adjusts the operating state of the target object according to the user's needs, realizes the intelligent, personalized and efficient adjustment of the target object, and improves the user's satisfaction during use. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0055] Figure 1 A schematic diagram of an electric vehicle gas source intelligent control system provided in this application;

[0056] Figure 2 A schematic diagram of an intelligent control module of an electric vehicle gas source intelligent control system provided in this application;

[0057] Figure 3 A schematic diagram of a flow chart of a first embodiment of an intelligent control method for an electric vehicle gas source provided in an embodiment of the present application;

[0058] Figure 4 A flow chart of a second embodiment of the intelligent control method for an electric vehicle gas source provided in the embodiment of the present application;

[0059] Figure 5 A schematic diagram of a multi-type application module interface provided for this application;

[0060] Figure 6 A schematic diagram of the structure of the vehicle-mounted central gas source intelligent control device provided in an embodiment of the present application;

[0061] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0062] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0063] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0064] It should be noted that the data information (including but not limited to the user's basic information data, adjustment information data, other data information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0065] In the prior art, an electric air compressor system is usually used to provide an air source for the target object, and the air pressure value in the main trachea and the air bag is monitored by a pressure monitor, and the real-time air pressure data is transmitted to the motor control unit, and then the motor control unit receives and processes the air pressure value monitored by the pressure monitor; according to the preset control logic, the action of the switching element and the air source is controlled to adjust the air pressure value in the air bag; by switching the air source channels of different air bags and adjusting the air pressure, the system realizes the air pressure adjustment of different air bags to achieve the massage function. However, the existing air source control method mainly relies on the preset control logic, and cannot be intelligently adjusted according to the user's real-time needs and preferences; the system relies on simple pressure monitoring and control logic, and cannot respond to the user's complex instructions quickly and accurately, so the adjustment effect is not good, resulting in an unsatisfactory massage effect; due to the preset control logic, the user may need to frequently adjust the instructions during the operation to meet the massage needs, which increases the inconvenience of operation.

[0066] In order to solve the above problems, the embodiment of the present application provides an intelligent control method and system for the gas source of an electric vehicle, which performs semantic analysis and sentiment analysis on the target object adjustment instruction, and combines the user portrait to learn the mapping relationship between the target object adjustment instruction and the gas source pressure value and the gas source flow value, so that the system can automatically adjust the operating state of the target object according to the user's needs. Through semantic analysis and sentiment analysis, the system can more accurately understand the user's needs and emotional state. The introduction of user portraits enables the system to adjust according to the user's historical interaction data, thereby providing a more personalized and accurate target object adjustment plan and improving the convenience of operation; using a neural network model for deep learning training, the gas source control model can quickly and accurately determine the gas source pressure value and gas source flow value, thereby improving the system's response speed and adjustment accuracy, realizing intelligent, personalized and efficient adjustment of the target object, and improving user satisfaction during use.

[0067] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0068] Figure 1 A schematic diagram of an electric vehicle gas source intelligent control system provided in this application, such as Figure 1 As shown, the electric vehicle air source intelligent control system of the present application includes an air compressor, an intelligent control module, an application interface module, and an integrated pipeline connecting the air compressor and the application interface module. Among them, the intelligent control module is used to obtain and analyze the user's target object adjustment instructions, determine the air source pressure value and air source flow value to be adjusted, and generate corresponding air source adjustment instructions; the air compressor is used to compress the air according to the air source adjustment instructions to obtain the corresponding target air source, and transmit the target air source to the application interface module through the integrated pipeline. Optionally, the air compressor can refer to an oil-free vortex air compressor; the application interface module is used to receive the target air source transmitted by the air compressor, and provide the target air source to the target object to adjust the operating state of the target object.

[0069] Furthermore, in the actual application of the electric vehicle gas source intelligent control system, the intelligent control module may include: Figure 2The data acquisition module, data processing module, data analysis module, and instruction generation module are shown. For example, the data acquisition module can be used to acquire and convert the data or information required by the intelligent control module, for example, to acquire and convert user instructions (such as voice recognition, touch screen input) into text data; the data processing module can be used to store and preliminarily process the data or information of the data acquisition module, for example, to perform preprocessing operations such as noise reduction, echo cancellation and gain control on the collected voice signal to improve the accuracy of voice recognition, and for example, to perform semantic analysis on the text data after voice recognition; the data analysis module can be used to comprehensively analyze the information of the data processing module to determine the air source pressure value and air source flow value corresponding to the user instruction, for example, using the air source control model through the deep learning algorithm to calculate that the air source pressure that needs to be adjusted to the current situation is 120KPa and the flow rate is 2L / min; the instruction generation module can be used to generate specific control instructions according to the air source pressure value and flow value output by the data analysis module, control the air compressor and related components to adjust the operating state of the target object, for example, instruct the air compressor to provide an air pressure of 120KPa, and instruct the switching component to switch to the corresponding air bag channel to ensure that the air flow is accurately delivered to the massage part that needs to be adjusted.

[0070] Figure 3 The flowchart of the first embodiment of the intelligent control method of the gas source of an electric vehicle provided in the embodiment of the present application is as follows: Figure 3 As shown, the electric vehicle gas source intelligent control method specifically includes the following steps:

[0071] S301. The intelligent control module obtains a target object adjustment instruction from a user, and obtains instruction text data according to the target object adjustment instruction conversion.

[0072] The target object adjustment instruction may refer to an instruction input by a user through voice, touch screen operation, button click, knob adjustment, etc. to adjust the target object, wherein the target object may be a massage chair. The intelligent control module may obtain the target object adjustment instruction through a user interface or instruction input device pre-installed in the vehicle.

[0073] It should be noted that the user here refers to the driver or passenger who issues the command. The relevant data stored in the system includes the command of the driver or passenger corresponding to the seat, as well as the command text data and the target object adjustment data; it can also refer to the personal account registered in the user interface or command input device in the car. Each account is associated with specific user preferences and setting data to facilitate system data management. At the same time, the system can also make personalized settings for passengers in each seat in the car, and store and memorize data according to the adjustment instructions of each seat. In this way, no matter which seat the passenger sits in, the system can provide a customized comfort experience according to the previous settings. For example, you can log in through the account, the system identifies the user's identity, and automatically adjusts the gas source pressure and flow according to the user's historical preferences; the system manages the user data of each account or seat in a unified manner, including massage preferences, gas source usage history, etc., to facilitate personalized services and system optimization.

[0074] Instruction text data may refer to data that converts target object adjustment instructions into a predefined text format, so that the data is structured, standardized, and consistent, which is convenient for semantic parsing and sentiment analysis processing, and is easy to store and retrieve.

[0075] Furthermore, the system can store the user's historical instruction text data, analyze the user's usage habits and preferences, and provide personalized services.

[0076] In one example, when the instruction is a voice instruction, the time-frequency features in the voice are denoised and then encoded to obtain a voice sequence; the voice sequence is converted into text using a speech recognition model (such as the Whisper model) to obtain instruction text data. For example, when the user says "increase the intensity of the massage", the system converts it into the instruction text data "increase the intensity of the massage".

[0077] When the instruction is a touch screen input or selection instruction, capture the touch event (such as clicking a button or screen position); determine the user instruction based on the coordinates and type of the touch event; convert the user instruction into instruction text data, for example, when the user selects the "strong massage" mode on the touch screen, the system converts it into the instruction text data "strong massage".

[0078] When the instruction is a knob adjustment instruction, the rotation direction and angle of the knob are captured; the user instruction is determined based on the degree and direction of rotation; the user instruction is converted into instruction text data, for example, when the user rotates to the "relaxation mode" button, the system converts it into the instruction text data "relaxation mode".

[0079] Therefore, the intelligent control module can obtain the user's target object adjustment instructions in a variety of ways and convert them into instruction text data. Each method has its specific application scenarios and advantages, which can meet the needs of different users and ensure that the system can accurately obtain and process user instructions.

[0080] S302, the intelligent control module performs semantic analysis on the instruction text data to obtain the user's intention analysis result, and the intention analysis result includes an adjustment method for the target object.

[0081] The semantic parsing process may refer to natural language processing of the instruction text data to understand its meaning and the user's intention. The adjustment method of the massage chair may refer to the specific adjustment that the user wants to make to the massage chair, such as mode, intensity, angle, etc.

[0082] The implementation of natural language processing includes but is not limited to:

[0083] Optionally, the command text data is parsed using predefined rules and dictionaries to match the user's intention and adjustment method. For example, the command text data is "Please adjust the massage chair to comfort mode", and "adjust to" + "comfort mode" are matched. The parsing result is that the intention is "adjustment" and the adjustment method is "comfort mode".

[0084] Optionally, the instruction text data is parsed according to a predefined template using template matching technology. For example, the instruction text data is "I want a strong massage", and the matching template is "I want [mode]", and the parsing result is that the intention is "adjustment" and the adjustment method is "strong massage".

[0085] Optionally, a machine learning model (such as a support vector machine, a decision tree, etc.) is used to train and parse the instruction text data. For example, the instruction text data is "Please start gentle massage", and the intention corresponding to "start" + "gentle massage" is learned through the training data. The parsing result is that the intention is "start" and the adjustment method is "gentle massage".

[0086] Optionally, a deep learning model (such as LSTM, BERT, etc.) is used to perform semantic analysis on the instruction text data to extract the user's intention and adjustment method. For example, if the instruction text data is "I want a full body massage", through context understanding and semantic analysis, the analysis result is that the intention is "adjustment" and the adjustment method is "full body massage"; if the instruction text data is "have the massage just now again", through context understanding and historical semantic analysis, the analysis result is that the intention is "repeat" and the adjustment method is "the massage mode just now".

[0087] In this step, the intelligent control module can perform semantic analysis on the instruction text data in a variety of ways to obtain the user's intention analysis results and specific target object adjustment methods. These methods can help the system accurately understand the user's intentions and provide more intelligent and personalized services.

[0088] S303, the intelligent control module performs sentiment analysis on the instruction text data to obtain the user's sentiment analysis result, and the sentiment analysis result is used to determine the user's satisfaction with the current operating status of the target object.

[0089] Among them, sentiment analysis processing may refer to sentiment tendency analysis of instruction text data to judge the user's emotional state. The sentiment analysis result may represent the user's emotional state, such as satisfaction, dissatisfaction, happiness, anger, etc. Satisfaction is determined based on the sentiment analysis result, which is the user's subjective evaluation of the current operating state of the target object. For example, satisfaction may be divided into satisfaction and dissatisfaction, or divided into multiple levels, such as 1-5 levels, where a higher level represents greater satisfaction.

[0090] The implementation methods of sentiment analysis processing include but are not limited to:

[0091] Optionally, sentiment analysis based on keywords. A keyword library is pre-established, containing keywords of positive and negative emotions, such as positive keywords: satisfied, comfortable, pleasant, relaxed, etc.; negative keywords: dissatisfied, uncomfortable, painful, nervous, etc. Then, the intelligent control module is used to match the command text data with keywords, and the frequency of occurrence of positive and negative keywords is counted; the sentiment score is calculated based on the frequency of keywords to determine the user's sentiment analysis result. For example, the user inputs "the massage chair is too hard and uncomfortable", the system recognizes two negative keywords "too hard" and "uncomfortable", and the sentiment analysis result is "unsatisfied".

[0092] Optionally, sentiment analysis based on machine learning or deep learning. Use a text data set annotated with sentiment tags to train a sentiment analysis model (the model may be a model using a machine learning algorithm or deep learning); input the user's instruction text data into the trained sentiment analysis model, and output the sentiment analysis result. For example, the user inputs "This massage strength is just right, I like it very much", and the model predicts the sentiment as "satisfied".

[0093] Optionally, sentiment analysis based on sentiment dictionary and rules. First, it is necessary to build a dictionary containing sentiment words and their sentiment polarity; then, define a set of rules for processing the combination and contextual relationship of sentiment words; according to the sentiment dictionary and rule engine, perform sentiment calculation on the instruction text data to obtain the sentiment analysis result. For example, the user inputs "the massage chair has good functions, but it is a bit noisy, and the massage intensity can be lowered a bit", the system recognizes "good" as a positive sentiment word, "a bit noisy" as a negative sentiment word, and comprehensively judges the sentiment as "neutral and satisfied".

[0094] Through the above-mentioned multiple implementation methods, the intelligent control module can effectively perform sentiment analysis on the user's command text data and obtain the user's sentiment analysis results, so as to better determine the user's satisfaction with the current operating status of the massage chair and make corresponding adjustments. For example, if the intention recognition result is "lower the intensity of the whole body massage" and the sentiment analysis result is "neutral to satisfied", then in the actual adjustment, the adjustment plan determined by the intelligent control module is to reduce the air source pressure by 10KPa. When the intention recognition result is "lower the intensity of the whole body massage" and the sentiment analysis result is "unsatisfied", the adjustment plan determined by the intelligent control module may be to reduce the air source pressure by 30KPa, thereby meeting the user's personalized needs.

[0095] S304. The intelligent control module inputs the intention analysis results, the sentiment analysis results and the user portrait into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction. The user portrait is determined based on the user's historical interaction data. The first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of the neural network model.

[0096] The user portrait may represent the user's usage preferences and habits. The first gas source control model may refer to a neural network model that learns the mapping relationship between the intention analysis result, the sentiment analysis result, the user portrait of the user, and the gas source pressure and flow rate through training.

[0097] For example, the first gas source control model can be constructed and trained based on a multilayer perceptron (MLP), a convolutional neural network (CNN) or a long short-term memory network (LSTM). It can also be constructed and trained in combination using multiple different types of neural network models (such as MLP, CNN, LSTM) and an integrated learning method (such as random forest, gradient boosting, etc.).

[0098] In this step, the output parameters of the model are not limited to the gas source pressure value and the gas source flow value, but may also include other parameters that affect the adjustment method or mode of the target object, such as the duration of the gas source, which can be set according to the specific application scenario.

[0099] By analyzing the intention analysis results, sentiment analysis results, and user portraits to determine the air source pressure value and air source flow value corresponding to the target object adjustment instruction, user needs can be considered from multiple aspects, so that the adjustment plan can meet the user's personalized needs. At the same time, by using the intention analysis results and sentiment analysis results, the user does not need to perform multiple operations to determine the final massage method, thereby improving the convenience of operation.

[0100] S305. The intelligent control module controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value.

[0101] Optionally, the intelligent control module generates a corresponding control signal according to the air source pressure value and the air source flow value to drive the air compressor to adjust the pressure and flow of the air source; the control signal controls the control valves of each branch pipeline in the integrated pipeline, and inputs the adjusted air source into the airbag of the massage chair to control the operating state of the massage chair.

[0102] The electric vehicle gas source intelligent control method provided in the embodiment of the present application has the following advantages:

[0103] (1) The intelligent control module can perform semantic parsing and sentiment analysis on user commands to more accurately understand the user's intention and emotional state, thereby providing a more personalized target object adjustment plan and improving user satisfaction;

[0104] (2) The first gas source control model obtained by deep learning training can accurately determine the gas source pressure value and gas source flow value according to the user's intention and emotion analysis results, thereby achieving precise control of the operating state of the target object;

[0105] (3) The introduction of user portraits can be used to make personalized adjustments based on the user's historical interaction data, so that the system can adapt to the needs and preferences of different users and further improve the user experience;

[0106] (4) The intelligent control module can obtain and process user instructions in real time and quickly adjust the operating status of the target object based on the processing results to ensure the user's comfort and satisfaction during use;

[0107] (5) By integrating the air compressor, intelligent control module, application interface module and integrated pipeline, the efficient integration of the vehicle-mounted central air source system is achieved, the system structure is simplified, and the reliability and stability of the system are improved.

[0108] Among them, in the embodiment of the present application, the method for the intelligent control module to control and adjust the operating state of the target object according to the gas source pressure value and the gas source flow value may include:

[0109] The intelligent control module generates an air source adjustment instruction corresponding to the target object adjustment instruction according to the air source pressure value and the air source flow value, and sends the air source adjustment instruction to the air compressor;

[0110] The air compressor compresses the air according to the air source adjustment instruction to obtain the target air source corresponding to the air source adjustment instruction, and transmits the target air source to the application interface module through the integrated pipeline, so that the target air source enters the target object through the application interface module and adjusts the operating state of the target object.

[0111] In this step, the air source adjustment instruction may refer to an instruction generated by the intelligent control module, which is used to control the operating state of the air compressor. The method for generating the air source adjustment instruction includes but is not limited to: generating the air source adjustment instruction based on a PID controller, a fuzzy control rule or a predefined rule. For example, the intelligent control module generates a corresponding air source adjustment instruction according to the air source pressure value of 90KPa and the air source flow value of 6L / min; sends the air source adjustment instruction to the air compressor; the air compressor compresses the air to 90KPa and adjusts the flow to 6L / min according to the air source adjustment instruction; transmits the target air source to the application interface module through the integrated pipeline and enters the target object; after the target air source enters the target object, adjusts the strength and speed of the target object to achieve the user's expected operating state.

[0112] The system can accurately control and adjust the gas source according to actual needs to avoid unnecessary waste of resources. The intelligent control module can automatically generate and send gas source adjustment instructions to achieve automatic control of the target object and reduce manual intervention.

[0113] Based on the above embodiments, it can be known that in order to achieve accurate control and supply of the gas source of the on-board target object in the electric vehicle scenario, it is necessary to collect some data in this scenario in advance to build a user portrait of the user and train the gas source control model based on the user portrait and the collected data, and then use the gas source control model in the actual positioning scenario. Among them, the gas source control model described in this embodiment can include a first gas source control model and a second gas source control model, and the intelligent control module configuration gas source control model can include at least the following two implementation methods:

[0114] The first implementation method is to build user portraits and train models on other servers, and after training, configure the obtained gas source control model into the intelligent control module for use.

[0115] In one example, a user uses a massage chair in an electric car, and the user's usage data is collected by installing sensors and user interaction interfaces in the vehicle-mounted equipment, such as basic information data, massage chair adjustment instructions at a certain moment, air source pressure value and air source flow value; the collected data is transmitted to a remote server via a wireless network for processing and storage; on the remote server, the collected data is used to build a user profile and analyze the user's usage habits and preferences; on the remote server, the constructed user profile and the collected data are used to train the air source control model; and the trained air source control model is deployed to the intelligent control module for use in the electric car.

[0116] Among them, remote servers usually have more powerful computing power and storage resources, and can handle larger-scale data and more complex model training, but the need to transmit data to remote servers may introduce certain delays and be greatly affected by the network.

[0117] In the second implementation method, the electric vehicle gas source intelligent control system also includes a server. The server in the system performs data collection, user portrait construction and model training to obtain a gas source control model, which is finally applied in the intelligent control module.

[0118] Specifically, the collected data is processed and analyzed by the server in the intelligent control system of the gas source of electric vehicles to construct a user portrait; the user portrait and the collected data are used to train the gas source control model to obtain an optimized gas source control model; the trained gas source control model is directly applied to the intelligent control module to adjust the operating state of the target object in real time. The data collection, user portrait construction and model training of this method are all carried out in the intelligent control system of the gas source of electric vehicles, which can realize real-time adjustment and optimization, and does not rely on external network connections. The system can operate independently under any circumstances.

[0119] The following takes the server in the electric vehicle gas source intelligent control system for data collection, user portrait construction and model training as an example (the process of model training on other servers is similar) to illustrate the training process of the gas source control model.

[0120] Among them, in the embodiment of the present application, the method for constructing a user portrait in the server of the electric vehicle gas source intelligent control system may include:

[0121] The server obtains the user's basic information data, historical target object adjustment instructions within a preset historical time period, historical gas source pressure values, and historical gas source flow values;

[0122] Determine the user's adjustment information data according to the historical target object adjustment instructions, the historical gas source pressure value and the historical gas source flow value, wherein the adjustment information data includes the user's usage habits and preference settings;

[0123] Extract features from user basic information data and adjustment information data respectively to obtain user identity features and user behavior features;

[0124] Build a user profile based on user identity characteristics and user behavior characteristics.

[0125] The basic information data may refer to the user registration information in the personal account, or may refer to the user personal information collected and stored at a certain seat, such as age, gender, height, weight, etc. The adjustment information data may be data obtained by statistically analyzing the historical target object adjustment instructions, historical gas source pressure values, and historical gas source flow values ​​based on semantic analysis processing, sentiment analysis processing, etc.

[0126] In this step, you can set the historical time period based on expert experience, such as the periodicity of user behavior and the changing trend of user behavior. You can also set it dynamically based on automatic analysis of changes in user behavior based on machine learning technology, while considering resource usage during server processing. You can also set a fixed time period or a rolling time window, such as the last month, the last three months, or the last six months.

[0127] Feature extraction can be to extract representative features from the user's basic information data and adjustment information data to facilitate a better description of the user portrait, so that the user portrait can help the system better understand the user's usage habits and preferences, thereby providing a more personalized and accurate massage experience in actual use.

[0128] For example, the basic information data of user A includes age 30, male, height 175cm, weight 70kg, historical massage chair adjustment instructions include multiple adjustments of strength and angle, and historical air source pressure values ​​and flow values ​​record the specific parameters of each adjustment; through analysis, it is determined that the adjustment information data is: the user prefers stronger massage, and often adjusts the angle of the massage chair to 45 degrees; then the constructed user portrait may be: User A is a 30-year-old male, prefers stronger massage, and often adjusts the angle of the massage chair to 45 degrees.

[0129] Among them, in the embodiment of the present application, the method for training the first air source control model in the server of the electric vehicle air source intelligent control system may include:

[0130] The server obtains target object adjustment instructions, user portraits, and gas source pressure values ​​and gas source flow values ​​corresponding to each target object adjustment instruction of multiple users within a preset historical time period;

[0131] For each user, the server determines the intent parsing result and sentiment analysis result corresponding to each massage chair adjustment instruction of the user;

[0132] Based on the user portraits of multiple users in the historical time period, the intent analysis results, sentiment analysis results, air source pressure value and air source flow value corresponding to each massage chair adjustment instruction, a first sample set is constructed, wherein each sample in the first sample set includes: the user portrait, intent analysis result, sentiment analysis result during one massage chair adjustment, and the air source pressure value and air source flow value corresponding to this massage chair adjustment;

[0133] Based on the first sample set, a pre-constructed first initial gas source control model is trained until a preset loss function converges to obtain a first gas source control model, where the first initial gas source control model is constructed based on a deep learning neural network model.

[0134] Among them, the first initial gas source control model can be an MLP model, a CNN model, an LSTM model, or a combination model of multiple different types of neural network models combined using an integrated learning method.

[0135] Taking the MLP model as an example, first, build the model architecture of the MLP model, including the number of nodes in the input layer, hidden layer, and output layer. For example, the output layer is determined to have two nodes based on the gas source pressure value and the gas source flow value; the activation function can use the ReLU activation function; the preset loss function can be set to the mean square error (MSE); select the Adam optimizer to optimize the model.

[0136] Secondly, in this step, based on the first sample set, the pre-constructed first initial gas source control model is trained until the preset loss function converges. The method for obtaining the first gas source control model may include:

[0137] The user portrait, intention analysis result, and sentiment analysis result of a massage chair adjustment in the first sample set are used as model input features, and the air source pressure value and air source flow value corresponding to this massage chair adjustment are used as model output labels, so that the MLP model obtains the air source pressure estimation value and the air source flow estimation value according to the model input features;

[0138] Determine a first error value between the estimated value of the gas source pressure and the historical value of the gas source pressure, and a second error value between the estimated value of the gas source flow and the historical value of the gas source flow;

[0139] Based on the first error value, the second error value and the preset loss function, a loss value is calculated, and the model parameters of the MLP model are adjusted according to the loss value until the preset loss function converges to obtain a first gas source control model, wherein the model parameters include the number of nodes and the learning rate of the hidden layer in the model architecture.

[0140] In this step, through the training of the deep learning neural network model, the system can continuously optimize the gas source control strategy and improve the accuracy and robustness of the model, so that the trained first gas source control model can be applied to the intelligent control module in real time, quickly respond to the user's adjustment instructions, and improve the system's response speed and user experience.

[0141] In the embodiment of the present application, before training the pre-constructed first initial gas source control model based on the first sample set until the preset loss function converges and the first gas source control model is obtained, the method may further include:

[0142] According to the preset batch processing requirements, the first sample set is divided into sample batches to obtain a sample set for multi-batch training.

[0143] The preset batch processing requirements may include the number of samples in each batch, which can be determined according to training requirements and computing resources, such as setting the number of samples in each batch to 32. Further, according to the batch processing requirements, the first sample set is divided into multiple batches, each batch contains a certain number of samples, and then the divided multiple batch sample sets are used for batch training of the model. For example, the first sample set contains 320 samples, and 10 batches will be obtained by dividing it into 32 samples per batch.

[0144] Based on this step, the model parameters of the MLP model can also include the batch size.

[0145] By reasonably setting the batch processing requirements and dividing the first sample set into multiple batches, you can reduce memory usage, improve training efficiency, accelerate model convergence, enhance model generalization capabilities, optimize the use of computing resources, and improve the stability and efficiency of the training process.

[0146] However, although the above method solves the problems of poor adjustment effect and inconvenient operation when adjusting the massage chair through air source control, during the use of this method, if the user's massage chair adjustment instructions are not clear, the intention analysis results of the instruction text data analysis will not contain specific ways to adjust the massage chair (such as mode, intensity, angle, etc.), that is, when the adjustment method in the user's intention analysis results is empty, the air source pressure value and air source flow value obtained by the first air source control model have the problem of low accuracy, resulting in the operating state of the massage chair after adjustment unable to meet user needs, and there are still problems of poor adjustment effect and inconvenient operation.

[0147] In view of the above technical problems, in some embodiments, such as Figure 4 As shown, when the adjustment method in the user's intention analysis result is empty, the method for determining the air source pressure value and the air source flow value of the electric vehicle air source intelligent control system specifically includes the following steps:

[0148] S401. If the adjustment method in the user's intention analysis result is empty, the intelligent control module obtains the user's historical instruction text data within a preset time period, the system data of the electric vehicle gas source intelligent control system, and the vehicle data; wherein the system data includes the gas source pressure, gas source temperature, compressor status, pipeline status, and the vehicle data includes the vehicle speed and acceleration.

[0149] The preset time period may be a time period before the current time, for example, a time period of thirty minutes in the past corresponding to the current time, or a time period of one hour in the past corresponding to the current time. The historical instruction text data may refer to all instruction text data of the user within the preset time period. The system data and vehicle data may be data stored in the system within the preset time period.

[0150] For example, user A issues an ambiguous instruction at a certain moment, for example, the intention analysis result is "adjust", and the intelligent control module cannot directly analyze the specific adjustment method. At this time, the module will obtain the following data:

[0151] The historical command text data includes the command records of user A in the past hour, such as "increase strength", "adjust angle", etc.; the system data includes the gas source pressure (e.g. 70-90KPa), gas source temperature (e.g. 25-30℃), compressor status (e.g. running / stopped), pipeline status (e.g. normal / abnormal) in the past hour; the vehicle data includes the vehicle speed (e.g. 60-80km / h) and acceleration (e.g. 0.5-1.0m / s 2 ).

[0152] Through the user's historical command text data, system data and vehicle data within a preset time period, the intelligent control module can obtain the user's contextual information about the massage chair adjustment, thereby inferring the user's needs.

[0153] S402, the intelligent control module extracts features from the system data, vehicle data, and instruction text data, and based on the time series and a preset feature fusion strategy, performs feature fusion on the data features of the system data, vehicle data, and historical instruction text data at each time point to obtain the data fusion features at each time point.

[0154] In this step, the intelligent control module needs to extract features from the acquired system data, vehicle data, and instruction text data, thereby converting the raw data into features that can better represent the characteristics of the data. For example, features such as the average value, maximum value, and minimum value are extracted from the gas source pressure; features such as the average value and rate of change are extracted from the vehicle speed and acceleration; and features such as intent keywords and sentiment analysis results are extracted from the instruction text.

[0155] Then, based on the time series and the preset feature fusion strategy, the system data, vehicle data and historical instruction text data at each time point are fused to obtain the data fusion features at each time point. Feature fusion can be the splicing or weighted averaging of the features of the system data, vehicle data and instruction text data at the same time point.

[0156] S403, the intelligent control module inputs the data fusion features of each time point into the second air source control model in chronological order to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction. The second air source control model is a model for determining the air source pressure value and the air source flow value obtained by training the time series model.

[0157] In this step, the second gas source control model is a trained time series model, which is used to determine the gas source pressure value and the gas source flow value according to the input data fusion characteristics. In this way, the system can intelligently infer the gas source pressure and flow suitable for the current situation based on the current state and the data in the preset time period before the state.

[0158] Optionally, the second gas source control model can be constructed and trained according to time series models such as autoregressive integrated moving average model (ARIMA), seasonal ARIMA (SARIMA), long short-term memory network (LSTM), etc.

[0159] The electric vehicle gas source intelligent control method provided in the embodiment of the present application is capable of determining the gas source pressure value and gas source flow value suitable for the current situation when the user intention analysis result is unclear, thereby improving the intelligent adjustment effect of the system and enhancing the personalization and accuracy of the user experience.

[0160] Among them, in the embodiment of the present application, the method for training the second gas source control model in the server of the electric vehicle gas source intelligent control system may include:

[0161] The server obtains historical instruction text data, system data, vehicle data, and gas source pressure value and gas source flow value corresponding to each historical instruction text data of multiple users within a preset historical time period;

[0162] For each user, the server performs feature fusion on each historical instruction text data, system data, and vehicle data of the user within a preset historical time period based on the time series and the preset feature fusion strategy, and determines the data fusion features at each time point;

[0163] Based on the data fusion features of multiple users at each time point in the historical time period, the gas source pressure value and the gas source flow value corresponding to the historical instruction text data at each time point, a second sample set is constructed, and each sample in the second sample set includes: the data fusion features of a massage chair adjustment, and the gas source pressure value and the gas source flow value corresponding to this massage chair adjustment;

[0164] Based on the second sample set, the pre-constructed second initial gas source control model is trained until the preset loss function converges to obtain the second gas source control model, and the second initial gas source control model is constructed based on a neural network model of a time series.

[0165] In this step, the preset historical time period refers to a historical time period before a certain historical moment when the gas source pressure value and the gas source flow value are adjusted at that historical moment. For example, if the user adjusts the gas source pressure value and the gas source flow value at 15:00, then if the preset historical time period is 1 hour before the historical moment, the server obtains the historical instruction text data, system data, vehicle data stored by the user between 14:00-15:00, and the gas source pressure value and gas source flow value corresponding to each historical instruction text data.

[0166] Taking the LSTM model as an example, based on the second sample set, the pre-constructed second initial gas source control model is trained until the preset loss function converges. The method for obtaining the second gas source control model may include:

[0167] Dividing the second sample set into a plurality of time windows, each window containing data points within a certain time range;

[0168] Build the model architecture of the LSTM model, including the dimension of the input layer, the number of LSTM layers and the number of units in each layer, and the parameters of the output layer. For example, define the dimension of the input layer according to the number of features in the time window; the preset loss function can be set to mean square error (MSE) and root mean square error (RMSE);

[0169] Use the divided time window data to train the LSTM model until the preset loss function (mean square error MSE and root mean square error RMSE) converges; after training, use the validation set or test set to evaluate the performance of the model to ensure that the model has good generalization ability on unseen data.

[0170] For example, the second sample set contains 1000 samples, each sample includes timestamp, gas source pressure value, gas source flow value, data fusion feature, and the time window size is selected as 50 samples, then time window 1: sample 1 to sample 50; time window 2: sample 51 to sample 100; time window 3: sample 101 to sample 150; and so on for samples in each time window. If each sample has 1 feature and the time window size is 50 data points, the input layer dimension can be: 50 (time step) x1 (number of features); the LSTM layer can be: 2 layers, each with 64 units; the output layer can be: 2 outputs (gas source pressure value and gas source flow value).

[0171] The above-mentioned model training method enables the system model to process time series data with long-term dependencies. By analyzing multi-source information such as system data, vehicle data, and historical instruction text data, it learns the complex mapping relationship between user needs, system data, and air source pressure and flow within a preset time period, thereby determining the appropriate air source pressure value and air source flow value, and realizing intelligent adjustment of the massage chair air source.

[0172] Among them, in the embodiment of the present application, the preset feature fusion strategy is a strategy constructed based on the attention mechanism.

[0173] Furthermore, the feature fusion method may include:

[0174] Perform feature extraction on each historical instruction text data, system data, and vehicle data to obtain user intention features, system features, and vehicle features;

[0175] Align each user intention feature, system feature, and vehicle feature in the time series within a preset historical time period to obtain the aligned user intention features, system features, and vehicle features at each time point;

[0176] Based on the preset feature fusion strategy, determine the feature weights of system features, vehicle features, and user behavior features;

[0177] According to the feature weights, the system features, vehicle features and user intention features are fused to obtain data fusion features.

[0178] The feature weights are calculated by setting a separate input layer for each type of feature. In the attention layer, the weight of each feature is calculated by a trainable weight vector, which is multiplied by the representation of the feature and then normalized by an activation function (such as softmax).

[0179] The attention mechanism automatically assigns higher weights to important parts of the model input data, thereby more accurately extracting key features. At the same time, when fusing multiple types of data such as historical instruction text data, system data, vehicle data, etc., these multimodal data can be effectively processed and integrated to improve the effect of feature fusion, thereby improving the training efficiency of the model.

[0180] On the basis of the above embodiment, in the server of the electric vehicle air source intelligent control system, the operating state of the air compressor can also be adjusted in real time, and the specific steps may include:

[0181] The server determines the operating parameters of the air compressor according to the compressor status in the system data;

[0182] If the operating parameter is not within the preset parameter operating range, determining a parameter difference between the operating parameter of the air compressor and a parameter that satisfies the preset parameter operating range;

[0183] The actual operating state of the air compressor is adjusted according to the parameter difference, vehicle data and the air source pressure value and air source flow value output by the second air source control model.

[0184] In this step, the server obtains the current operating status of the air compressor from the system data, including but not limited to the current pressure, flow, temperature, speed and other parameters of the compressor; the server can determine the current operating parameters of the air compressor based on the obtained compressor status data, and the current operating parameters may include the output pressure, flow, power consumption, etc. of the compressor; then, the server compares the current operating parameters with the preset parameter operating range to determine whether the current operating parameters are within the preset parameter operating range; if the current operating parameters are not within the preset parameter operating range, the server calculates the parameter difference between the current operating parameters and the target parameters that meet the preset parameter operating range. For example, if the current pressure is P1 and the target pressure is P2, the pressure difference ΔP = P2-P1.

[0185] Combining the parameter difference, vehicle data, and the output of the second air source control model, the server needs to comprehensively analyze the current operating status and target status to determine the adjustment strategy of the air compressor. For example, if the vehicle is in a high-load state and the current pressure is lower than the target pressure, the server may need to increase the output pressure and flow of the compressor. In some specific scenarios, when the system determines the operating status of the air compressor, the air source pressure value and air source flow value output by the second air source control model can be replaced by the air source pressure value and air source flow value output by the first air source control model to meet the actual needs of the system during operation.

[0186] By real-time monitoring and adjusting the operating status of the air compressor, it can be ensured that its operating parameters are always within the preset parameter operating range, avoiding system instability or failure caused by parameter deviation; combined with vehicle data and the output of the air source control model, the output pressure and flow of the air compressor can be dynamically adjusted according to actual needs, ensuring the accuracy and timeliness of the air source supply and improving the response speed and efficiency of the system.

[0187] Optionally, the electric vehicle gas source intelligent control system can also monitor the operating status of each key module of the system (such as integrated pipelines, oxygen generators, etc.) according to other data in the system data, to ensure that when problems occur in each module, the system can promptly notify the driver and take corresponding measures. For example, the system data is stored in the form of system logs; the system logs are analyzed regularly to determine whether the parameters of each module exceed the preset threshold, or the collected data are analyzed using anomaly detection algorithms (statistical methods, machine learning algorithms) to detect abnormal conditions, for example, using time series analysis to detect abnormal fluctuations in parameters; based on preset rules and thresholds, the specific fault type is identified, for example, if the oxygen concentration is lower than the preset threshold, it is identified as an oxygen generator fault; different levels of alarms are set according to the fault type and fault severity, for example, a warning signal is issued for a minor fault, and an emergency alarm is issued for a serious fault.

[0188] The above-mentioned electric vehicle gas source intelligent control method is a situation where the application module interface is connected to the massage chair to provide a stable gas source for the massage chair, but in actual applications, the application module interface can also provide gas sources for equipment or needs such as oxygen generators, sensor cleaning, air suspensions, and external vehicle interfaces. When there are multiple types of needs, the electric vehicle gas source intelligent control system can use the electric vehicle gas source intelligent control method to train the gas source control model corresponding to each type, and adjust the input parameters when using the above-mentioned electric vehicle gas source intelligent control method and gas source control model. Through the dynamic adjustment of the gas source of each application module interface by the electric vehicle gas source intelligent control system, it is possible to provide gas sources for a variety of on-board auxiliary applications, and rationally manage and distribute the gas sources, improve the energy efficiency and reliability of the system, and provide users with a better user experience.

[0189] In one example, Figure 5 As shown in the figure, the application module interface of the electric vehicle gas source intelligent control system can provide gas source for multiple devices or needs through the main gas supply pipeline and branch pipeline in the integrated pipeline. The gas source between the branch pipelines can be controlled by the control valve, for example:

[0190] ①Massage chair interface: Through specific pipes and valves, a suitable air source is provided to the massage chair to realize the massage function of the chair.

[0191] ② Oxygen concentrator interface: Connect to the oxygen concentrator to provide it with necessary compressed air for passengers to use when needed.

[0192] For the oxygen concentrator, when the adjustment method in the user's intention analysis result is empty, the collected vehicle data also needs to include the current vehicle altitude, the current oxygen concentration outside the vehicle, and the current oxygen concentration inside the vehicle. The oxygen concentration can be collected by installing an oxygen concentration monitor in the vehicle.

[0193] ③Sensor cleaning interface: Design a special air blowing cleaning system to regularly provide clean airflow to the key sensors of electric vehicles to maintain their performance and accuracy.

[0194] For the air blowing cleaning system, when the adjustment method in the user's intention analysis result is empty, you can also use the timed cleaning method to preset the air source pressure value and air source flow value range for sensor cleaning to meet the cleaning needs.

[0195] ④Air suspension interface: provides a constant air source for the air suspension system to ensure the comfort and stability of the suspension system.

[0196] For the air suspension system, when the adjustment method in the user's intention analysis result is empty, the collected vehicle data also needs to include the current vehicle body height, so that the air source control model can determine the appropriate air source pressure value and air source flow value according to the vehicle body changes within a preset time period, where the vehicle body height can be monitored by a height sensor.

[0197] ⑤ External vehicle interface: can be used to inflate tires / tents / air mattresses.

[0198] In actual applications, there may be situations where multiple interfaces are used at the same time. Optionally, the air supply priority of each application module interface is set in the electric vehicle air source intelligent control system to give priority to the supply of important interface air source needs. For example, the air source supply priority of the air suspension interface is higher than the external vehicle interface. When a user's adjustment instruction causes the air source to be unable to be stably supplied to the high-priority interface, the system can show the user preset prompt problems such as excessive adjustment force or adjustment instruction not meeting the requirements.

[0199] Figure 6 The schematic diagram of the structure of the vehicle-mounted central gas source intelligent control device provided in the embodiment of the present application is as follows: Figure 6 As shown, the vehicle-mounted central gas source intelligent control device 60 provided in this embodiment includes:

[0200] The acquisition module 601 is used to acquire the user's target object adjustment instruction and convert the acquired instruction text data according to the target object adjustment instruction;

[0201] The data analysis module 602 is used to perform semantic analysis on the instruction text data to obtain the user's intention analysis result, and the intention analysis result includes the adjustment method of the target object;

[0202] The data analysis module 603 is used to perform sentiment analysis on the instruction text data to obtain the sentiment analysis result of the user, and the sentiment analysis result is used to determine the user's satisfaction with the current operating status of the target object;

[0203] Obtaining module 604, for inputting the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model, and obtaining the air source pressure value and the air source flow value corresponding to the target object adjustment instruction, the user portrait is determined according to the user's historical interaction data, and the first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of the neural network model;

[0204] The control module 605 is used to control and adjust the operating state of the target object according to the gas source pressure value and the gas source flow value.

[0205] In a possible implementation, the obtaining module 604 may also be used to:

[0206] The server obtains the user's basic information data, historical target object adjustment instructions within a preset historical time period, historical gas source pressure values, and historical gas source flow values;

[0207] Determine the user's adjustment information data according to the historical target object adjustment instructions, the historical gas source pressure value and the historical gas source flow value, wherein the adjustment information data includes the user's usage habits and preference settings;

[0208] Extract features from user basic information data and adjustment information data respectively to obtain user identity features and user behavior features;

[0209] Build a user profile based on user identity characteristics and user behavior characteristics.

[0210] In a possible implementation, the obtaining module 604 may also be used to:

[0211] The server obtains target object adjustment instructions, user portraits, and gas source pressure values ​​and gas source flow values ​​corresponding to each target object adjustment instruction of multiple users within a preset historical time period;

[0212] For each user, the server determines the intent parsing result and sentiment analysis result corresponding to each target object adjustment instruction of the user according to each target object adjustment instruction of the user;

[0213] Based on the user portraits of multiple users in the historical time period, the intent analysis results, sentiment analysis results, gas source pressure values, and gas source flow values ​​corresponding to each target object adjustment instruction, a first sample set is constructed, wherein each sample in the first sample set includes: the user portrait, intent analysis results, sentiment analysis results during a target object adjustment, and the gas source pressure value and gas source flow value corresponding to this target object adjustment;

[0214] Based on the first sample set, a pre-constructed first initial gas source control model is trained until a preset loss function converges to obtain a first gas source control model, where the first initial gas source control model is constructed based on a deep learning neural network model.

[0215] In a possible implementation, the obtaining module 604 may also be used to:

[0216] According to the preset batch processing requirements, the first sample set is divided into sample batches to obtain a sample set for multi-batch training.

[0217] In a possible implementation, the obtaining module 604 may also be used to:

[0218] If the adjustment method in the user's intention analysis result is empty, the intelligent control module obtains the user's historical command text data within a preset time period, the system data of the electric vehicle gas source intelligent control system, and the vehicle data; wherein the system data includes the gas source pressure, gas source temperature, compressor status, pipeline status, and the vehicle data includes the vehicle speed and acceleration;

[0219] The intelligent control module extracts features from system data, vehicle data, and instruction text data, and based on the time series and the preset feature fusion strategy, fuses the data features of system data, vehicle data, and historical instruction text data at each time point to obtain the data fusion features at each time point;

[0220] The intelligent control module inputs the data fusion features of each time point into the second air source control model in chronological order to obtain the air source pressure value and air source flow value corresponding to the target object adjustment instruction. The second air source control model is a model for determining the air source pressure value and air source flow value obtained by training the time series model.

[0221] In a possible implementation, the obtaining module 604 may also be used to:

[0222] The server obtains historical instruction text data, system data, vehicle data, and gas source pressure value and gas source flow value corresponding to each historical instruction text data of multiple users within a preset historical time period;

[0223] For each user, the server performs feature fusion on each historical instruction text data, system data, and vehicle data of the user within a preset historical time period based on the time series and the preset feature fusion strategy, and determines the data fusion features at each time point;

[0224] Based on the data fusion features of multiple users at each time point in the historical time period, the gas source pressure value and the gas source flow value corresponding to the historical instruction text data at each time point, a second sample set is constructed, and each sample in the second sample set includes: the data fusion features when a target object is adjusted once, and the gas source pressure value and the gas source flow value corresponding to this target object adjustment;

[0225] Based on the second sample set, the pre-constructed second initial gas source control model is trained until the preset loss function converges to obtain the second gas source control model, and the second initial gas source control model is constructed based on a neural network model of a time series.

[0226] In a possible implementation, the obtaining module 604 may also be used to:

[0227] The server determines the operating parameters of the air compressor according to the compressor status in the system data;

[0228] If the operating parameter is not within the preset parameter operating range, determining a parameter difference between the operating parameter of the air compressor and a parameter that satisfies the preset parameter operating range;

[0229] The actual operating state of the air compressor is adjusted according to the parameter difference, vehicle data and the air source pressure value and air source flow value output by the second air source control model.

[0230] In a possible implementation, the control module 605 may also be used to:

[0231] The intelligent control module generates an air source adjustment instruction corresponding to the target object adjustment instruction according to the air source pressure value and the air source flow value, and sends the air source adjustment instruction to the air compressor;

[0232] The air compressor compresses the air according to the air source adjustment instruction to obtain the target air source corresponding to the air source adjustment instruction, and transmits the target air source to the application interface module through the integrated pipeline, so that the target air source enters the target object through the application interface module and adjusts the operating state of the target object.

[0233] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application, such as Figure 7 As shown, the electronic device 70 provided in this embodiment includes: at least one processor 701 and a memory 702 .

[0234] Optionally, the device 70 further includes a communication component 703 . The processor 701 , the memory 702 and the communication component 703 are connected via a bus 704 .

[0235] In a specific implementation process, at least one processor 701 executes the computer-executable instructions stored in the memory 702, so that at least one processor 701 executes the above method.

[0236] The specific implementation process of the processor 701 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0237] This embodiment provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the intelligent control method of the gas source of the electric vehicle in the above embodiment.

[0238] This embodiment also provides a computer program product, including a computer program, which, when executed by a processor, implements the intelligent control method for the gas source of an electric vehicle provided in any one of the above embodiments.

[0239] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0240] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An intelligent control method for an electric vehicle gas source, characterized in that: Applied to an intelligent control system for an electric vehicle gas source, the intelligent control system for an electric vehicle gas source includes an air compressor, an intelligent control module, an application interface module, and an integrated pipeline connecting the air compressor and the application interface module. The method includes: The intelligent control module obtains a target object adjustment instruction from a user, and converts and obtains instruction text data according to the target object adjustment instruction; The intelligent control module performs semantic analysis on the instruction text data to obtain the user's intention analysis result, wherein the intention analysis result includes an adjustment method for the target object; The intelligent control module performs sentiment analysis on the instruction text data to obtain the sentiment analysis result of the user, and the sentiment analysis result is used to determine the user's satisfaction with the current operating status of the target object; The intelligent control module inputs the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction, the user portrait is determined according to the historical interaction data of the user, and the first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of a neural network model; The intelligent control module controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value.

2. The method according to claim 1, characterized in that The intelligent control module controls and adjusts the operating state of the target object according to the gas source pressure value and the gas source flow value, including: The intelligent control module generates an air source adjustment instruction corresponding to the target object adjustment instruction according to the air source pressure value and the air source flow value, and sends the air source adjustment instruction to the air compressor; The air compressor compresses the air according to the air source adjustment instruction to obtain a target air source corresponding to the air source adjustment instruction, and transmits the target air source to the application interface module through the integrated pipeline, so that the target air source enters the target object through the application interface module and adjusts the operating state of the target object.

3. The method according to claim 1, characterized in that The electric vehicle gas source intelligent control system further includes: a server; Before inputting the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction, the method further includes: The server obtains the basic information data of the user, historical target object adjustment instructions within a preset historical time period, historical gas source pressure values, and historical gas source flow values; Determining the user's adjustment information data according to the historical target object adjustment instruction, the historical gas source pressure value, and the historical gas source flow value, wherein the adjustment information data includes the user's usage habits and preference settings; Extracting features from the user basic information data and the adjustment information data to obtain user identity features and user behavior features; A user profile of the user is constructed based on the user identity characteristics and the user behavior characteristics.

4. The method according to claim 1, characterized in that: The electric vehicle gas source intelligent control system further includes: a server; Before inputting the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction, the method further includes: The server obtains target object adjustment instructions, user portraits, and gas source pressure values ​​and gas source flow values ​​corresponding to each target object adjustment instruction of multiple users within a preset historical time period; For each user, the server determines, according to each target object adjustment instruction of the user, an intent parsing result and a sentiment analysis result corresponding to the target object adjustment instruction; Based on the user portraits of the multiple users in the historical time period, the intent analysis result, the sentiment analysis result, the gas source pressure value and the gas source flow value corresponding to each target object adjustment instruction, a first sample set is constructed, wherein each sample in the first sample set includes: the user portrait, the intent analysis result, the sentiment analysis result during a target object adjustment, and the gas source pressure value and the gas source flow value corresponding to this target object adjustment; Based on the first sample set, a pre-constructed first initial gas source control model is trained until a preset loss function converges to obtain the first gas source control model, where the first initial gas source control model is constructed based on a deep learning neural network model.

5. The method according to claim 4, characterized in that Before training the pre-constructed first initial gas source control model based on the first sample set until a preset loss function converges to obtain the first gas source control model, the method further includes: According to preset batch processing requirements, the first sample set is divided into sample batches to obtain sample sets for multi-batch training.

6. The method according to claim 1, characterized in that The method further comprises: If the adjustment method in the intention analysis result of the user is empty, the intelligent control module obtains the historical instruction text data of the user within a preset time period, the system data of the electric vehicle gas source intelligent control system and the vehicle data; wherein the system data includes gas source pressure, gas source temperature, compressor status, pipeline status, and the vehicle data includes vehicle speed and acceleration; The intelligent control module extracts features from the system data, the vehicle data, and the instruction text data, and based on the time series and a preset feature fusion strategy, performs feature fusion on the data features of the system data, the vehicle data, and the historical instruction text data at each time point to obtain the data fusion features at each time point; The intelligent control module inputs the data fusion features of each time point into the second air source control model in chronological order to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction. The second air source control model is a model for determining the air source pressure value and the air source flow value obtained by training the time series model.

7. The method according to claim 6, characterized in that The electric vehicle gas source intelligent control system further includes: a server; Before inputting the data fusion features of each time point into the second gas source control model in chronological order to obtain the gas source pressure value and the gas source flow value corresponding to the target object adjustment instruction, the method further includes: The server obtains historical instruction text data, system data, vehicle data, and gas source pressure value and gas source flow value corresponding to each historical instruction text data of multiple users within a preset historical time period; For each user, the server performs feature fusion on each historical instruction text data, system data, and vehicle data of the user within a preset historical time period based on the time series and a preset feature fusion strategy, and determines the data fusion feature at each time point; Based on the data fusion features of the multiple users at each time point in the historical time period, the gas source pressure value and the gas source flow value corresponding to the historical instruction text data at each time point, a second sample set is constructed, wherein each sample in the second sample set includes: the data fusion features when a target object is adjusted once, and the gas source pressure value and the gas source flow value corresponding to this target object adjustment; Based on the second sample set, the pre-constructed second initial gas source control model is trained until the preset loss function converges to obtain the second gas source control model, where the second initial gas source control model is constructed based on a neural network model of a time series.

8. The method according to claim 7, characterized in that The preset feature fusion strategy is a strategy constructed based on the attention mechanism.

9. The method according to claim 6, characterized in that The electric vehicle gas source intelligent control system further includes: a server; The method further comprises: The server determines the operating parameters of the air compressor according to the compressor status in the system data; If the operating parameter is not within the preset parameter operating range, determining a parameter difference between the operating parameter of the air compressor and a parameter that satisfies the preset parameter operating range; The actual operating state of the air compressor is adjusted according to the parameter difference, the vehicle data, and the air source pressure value and the air source flow value output by the second air source control model.

10. An intelligent control system for the gas source of an electric vehicle, characterized in that: include: An air compressor, an intelligent control module, an application interface module, and an integrated pipeline connecting the air compressor and the application interface module; Wherein, the intelligent control module is used to obtain the user's target object adjustment instruction, and convert and obtain instruction text data according to the target object adjustment instruction; The intelligent control module is also used to perform semantic analysis on the instruction text data to obtain the user's intention analysis result, and the intention analysis result includes an adjustment method for the target object; The intelligent control module is further used to perform sentiment analysis on the instruction text data to obtain the sentiment analysis result of the user, and the sentiment analysis result is used to determine the user's satisfaction with the current operating status of the target object; The intelligent control module is also used to input the intention analysis result, the sentiment analysis result and the user portrait of the user into the first air source control model to obtain the air source pressure value and the air source flow value corresponding to the target object adjustment instruction, the user portrait is determined according to the historical interaction data of the user, and the first air source control model is a model for determining the air source pressure value and the air source flow value obtained by deep learning training of the neural network model; The intelligent control module is also used to control and adjust the operating state of the target object according to the gas source pressure value and the gas source flow value.

Images