Household equipment control method based on artificial intelligence and related equipment

Abstract

The invention provides a home equipment control method based on artificial intelligence and related equipment. The method comprises the following steps: acquiring user environment data in real time; acquiring user habit data corresponding to the user environment data; calling a preset AI big data model to generate a control instruction based on the user habit data; and controlling the target home equipment based on the control instruction. According to the method, the user environment data is acquired in real time, so that the user habit data of the user in the environment condition can be accurately grasped. The personalized control instruction can be generated based on the user habit data and the preset AI big data model, so that personalized and automatic control of the home equipment is realized. Meanwhile, compared with a traditional manual control mode, automatic control is higher in response speed and more accurate in control.

Description

Technical Field

[0001] The present invention relates to the field of smart home technology, and in particular to a home appliance control method based on artificial intelligence and related equipment. Background Art

[0002] With the rapid development of smart home technology, the types of home appliances are becoming increasingly diverse, and people's demand for automated and intelligent control of home appliances is becoming stronger.

[0003] However, the control methods of most home appliances are relatively simple, usually relying on manual operation or simple timing settings by users, making it difficult to achieve personalized control. Summary of the Invention

[0004] In view of this, the present invention proposes a home appliance control method and related equipment based on artificial intelligence. The specific scheme is as follows: In the first part, the present invention proposes a household appliance control method based on artificial intelligence, the method comprising: Obtain user environment data in real time; Acquiring user habit data corresponding to the user environment data; Calling a preset AI big data model to generate control instructions based on the user habit data; The target home appliance is controlled based on the control instruction.

[0005] In an optional embodiment, after generating the control instruction, the method further includes: Converting the execution result corresponding to the control instruction into voice broadcast content through the AI ​​big data model; Perform voice broadcast on the voice broadcast content.

[0006] In an optional embodiment, the method further includes: Obtaining user feedback voice data on the execution result; Determining whether the user is satisfied with the execution result based on the feedback voice data using the AI ​​big data model; If the user is not satisfied with the execution result, the AI ​​big data model is called to regenerate the control instruction according to the feedback voice data; The target home device is controlled based on the regenerated control instruction.

[0007] In an optional embodiment, the method further includes: Record the number of negative feedbacks from users who are dissatisfied with the execution results; if the number of continuous negative feedbacks reaches the threshold of the number of negative feedbacks, the user habit data is reconstructed based on feedback voice data corresponding to the number of continuous negative feedbacks; the AI big data model is updated based on the reconstructed user habit data; wherein the threshold of the number of negative feedbacks is a dynamically adjustable parameter.

[0008] In an optional embodiment, the threshold of the number of negative feedbacks is adjusted in the following way: the AI big data model is called to perform emotional analysis based on the latest feedback voice data, and an emotional score is obtained; the AI big data model is called to perform acoustic feature analysis based on the latest feedback voice data, and an acoustic feature score is obtained; a first threshold adjustment weight factor is determined based on the emotional score; a second threshold adjustment weight factor is determined based on the acoustic feature score; a target threshold adjustment weight factor is obtained based on the first threshold adjustment weight factor and the second threshold adjustment weight factor; the threshold of the number of negative feedbacks is adjusted based on the target threshold adjustment weight factor and a preset reference threshold of the number of negative feedbacks, and a threshold of the number of negative feedbacks is obtained.

[0009] In an optional embodiment, the method further comprises: the user habit data is desensitized to obtain user habit desensitized data; the user habit data corresponding to the user environment data is obtained, including: the user habit desensitized data corresponding to the user environment data is obtained; the AI big data model is called to generate control instructions based on the user habit data, including: the AI big data model is called to generate control instructions based on the user habit desensitized data.

[0010] In an optional embodiment, the user habit data is desensitized to obtain user habit desensitized data, including: the remaining power of the electronic device is obtained; the device privacy level corresponding to the user environment data is read from a preset configuration file; the privacy protection strength is determined according to the remaining power and the device privacy level; the data desensitization mode is determined according to the privacy protection strength; the user habit data is desensitized according to the data desensitization mode to obtain user habit desensitized data.

[0011] Secondly, the application provides a home device control system based on artificial intelligence, which comprises: The acquisition module is configured to acquire user environment data in real time. The acquisition module is configured to acquire user habit data corresponding to the user environment data. The generation module is configured to call a preset AI big data model to generate a control instruction based on the user habit data. The control module is configured to control a target home device based on the control instruction.

[0012] In a third aspect, the present application provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the AI-based home device control method according to the above technical solution when executing the computer program.

[0013] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the AI-based home device control method according to the above technical solution when executed by a processor.

[0014] Advantages: The present application can accurately grasp the user habit data of the user in the environment by acquiring the user environment data in real time. Based on the user habit data, the preset AI big data model can generate personalized control instructions, thereby realizing personalized and automatic control of the home device. Meanwhile, the automatic control of the present application has faster response speed and more accurate control compared with the traditional manual control mode. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a flowchart of the AI-based home device control method of the embodiment of the present application; Figure 2 is a flowchart of the negative feedback number threshold adjustment method of the embodiment of the present application; Figure 3 is a flowchart of the desensitization of the user habit data of the embodiment of the present application; Figure 4 is a module diagram of the AI-based home device control system of the embodiment of the present application; Figure 5 is a structural diagram of the electronic device of the embodiment of the present application. DETAILED DESCRIPTION

[0016] Hereinafter, various embodiments of the present disclosure will be described more fully. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it is understood that there is no intention to limit various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to encompass all adjustments, equivalents, and / or alternatives falling within the spirit and scope of various embodiments of the present disclosure.

[0017] The terms used in various embodiments of the present disclosure are used only for the purpose of describing particular embodiments and are not intended to limit various embodiments of the present disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly dictates otherwise. Unless otherwise defined, all terms used herein (including technical terms and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which various embodiments of the present disclosure belong. The terms (such as terms defined in a generally used dictionary) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized or overly formal meaning, unless clearly defined in various embodiments of the present disclosure.

[0018] Embodiment 1 This embodiment proposes an artificial intelligence-based home device control method, as shown in the accompanying drawings Figure 1 The method comprises: 101, acquiring user environment data in real time.

[0019] Specifically, the user environment data can include, but is not limited to, temperature, humidity, light intensity, air quality, user location information, door and window status, system clock, etc.

[0020] In some embodiments, various high-precision sensors (such as temperature and humidity sensors, light intensity sensors, human infrared sensors, etc.) deployed in the home environment can be used to collect user environment data in real time, and the collected user environment data can be transmitted to the electronic device through a local area network. In other embodiments, various high-precision sensors can also be integrated in the electronic device, and the user environment data can be acquired in real time through the integrated various high-precision sensors.

[0021] 102, acquiring user habit data corresponding to the user environment data.

[0022] User habit data refers to regular data based on user behavior patterns and preferences, such as user daily wake-up time, device usage habits after coming home (such as air conditioner temperature setting, light switch time), device linkage preferences in specific scenarios, etc.

[0023] The user operation behavior can be recorded through device logs (such as air conditioner operation records, light switch time) to obtain user habit data.

[0024] The historical user environment data and the user habit data can be mapped, and the mapping relationship can be stored in the local database of the electronic device. After the user environment data is obtained, the user habit data corresponding to the user environment data can be matched based on the mapping relationship. For example, when the indoor temperature is greater than or equal to 28 degrees Celsius and the time is from 18:00 to 22:00, the user habit sets the air conditioner temperature to 24 degrees Celsius, and when the light intensity is less than 50 lux and there is human activity, the user habit turns on the light.

[0025] 103, calling a preset AI big data model to generate a control instruction based on the user habit data.

[0026] The AI big data model in the embodiment is a machine learning model that is pre-trained. The model learns deeply based on a large amount of historical user habit data and corresponding control instructions, and can understand the potential needs of users in different environments.

[0027] Specifically, the AI big data model training process of the embodiment is as follows: First, a large amount of historical user habit data and corresponding control instructions are collected. The user habit data includes but is not limited to the behavior patterns of users in different time periods (such as morning wake-up time, evening sleep time), behavior habits in different rooms (such as bedroom light usage habits, living room air conditioner temperature setting), usage frequency and preferences for specific devices (such as smart speaker playlist, smart curtain opening and closing time), etc.

[0028] Then, a recurrent neural network (RNN) or its variant long short-term memory network (LSTM) in deep learning is used as a basic model. The RNN / LSTM model can process sequence data and is suitable for capturing the time sequence characteristics of user habits.

[0029] Further, in the model training process, key features in the user habit data are automatically extracted. The key features include but are not limited to time features (such as specific time points, time periods), device usage frequency features, device linkage features, etc. These features are nonlinearly transformed and abstracted through the hidden layer of the model to form high-dimensional feature representations.

[0030] Next, based on the extracted key features, the model learns to set user habit pattern rules through training. For example, if the model finds that the user often turns on the living room lights around 7 pm every day, it sets the rule as "turn on living room lights at 7 pm." The rule setting process is achieved through the output layer of the model, which maps high-dimensional features to specific control instructions. In addition, during the model training process, a series of hyperparameters need to be set, such as learning rate, batch size, number of iterations, etc. These parameters can be obtained according to actual needs or experiments to optimize the convergence speed and generalization ability of the model.

[0031] Further, the model is trained using the backpropagation algorithm and the gradient descent optimizer. By continuously adjusting the model parameters, the loss function (such as cross-entropy loss function) between the predicted control instructions and the actual control instructions is minimized, thereby improving the accuracy and robustness of the model.

[0032] Finally, upon receiving real-time user habit data, the AI big data model uses the trained feature extractor and rule setter to extract the current key features from the user habit data and generate corresponding control instructions based on the set rules.

[0033] Specifically, upon receiving user habit data, the AI big data model extracts key features from the user habit data, such as: user behavior patterns at different time periods (such as morning, evening, weekend, etc.); user behavior habits in different rooms (such as bedroom light usage, living room air conditioning temperature, etc.); user usage frequency and preferences for specific devices (such as smart speaker playlists, smart window curtain opening and closing times, etc.).

[0034] Then, according to the user habit pattern setting rules, for example, if the user habit is to turn on the lights at 7 pm, the instruction "turn on living room lights at 7 pm" is generated; if the user habit is to set the air conditioning temperature to 24°C, the instruction "air conditioning temperature set to 24°C" is generated. Thus, control instructions that meet the user's individual preferences are generated, such as automatically adjusting the air conditioning to the user's preferred temperature, automatically adjusting the window curtain opening degree according to the indoor lighting, automatically playing the user's favorite music at a specific time period, etc.

[0035] 104, control the target home device based on the control instruction.

[0036] In some embodiments, efficient communication between devices can be achieved through wireless communication technologies (such as Wi-Fi, Bluetooth, etc.). After receiving the control instruction through wireless communication, the target home device immediately executes the operation corresponding to the control instruction, such as adjusting the indoor temperature, adjusting the lighting brightness, controlling the home appliance switch, etc., to ensure that the home environment meets the user's individual needs and provides a convenient and comfortable experience for the user.

[0037] In an optional embodiment, after generating the control instruction, in order to further improve the user experience and enhance the friendliness of human-computer interaction, the method further comprises: converting the execution result corresponding to the control instruction into voice broadcast content through an AI big data model; voice broadcasting the voice broadcast content.

[0038] The execution result refers to the feedback information generated after the home device completes the operation corresponding to the control instruction after receiving the control instruction. It is the direct output after the control instruction is executed, used to inform the user whether the operation is successful, the execution state or related information.

[0039] For example, if the control instruction is to adjust the indoor temperature to the user's preferred 26 degrees Celsius, the execution result is that the indoor temperature has been adjusted to 26 degrees Celsius.

[0040] The execution result can be directly used as voice broadcast content. It can also be based on the execution result to generate voice broadcast content according to a preset voice broadcast text template. The voice broadcast content converted from the execution result can be: "Dear user, the indoor temperature has been adjusted to your favorite 26 degrees, and you will enjoy this just-right comfort." In the process of converting the execution result corresponding to the control instruction into voice broadcast content, the AI big data model will not only consider the content of the instruction itself, but also combine the current home environment state, user historical behavior patterns, and preset personalized preference settings to accurately convert the execution result into natural, smooth and easy-to-understand voice broadcast content. The generated voice broadcast content is closer to the user's daily communication habits, which can not only accurately convey the information of the execution result, but also present it to the user in a warm and friendly way.

[0041] Specifically, after receiving the execution result of the control instruction, the AI big data model not only considers the content of the instruction itself, but also real-time acquires the current home environment state (such as temperature, humidity, light intensity, etc.), user historical behavior patterns (such as user's operation habits for devices at different time periods), and preset personalized preference settings (such as user's preferred voice style, volume size, etc.).

[0042] Then, using natural language processing (NLP) technology, the information of the execution result, home environment state, user historical behavior patterns, and personalized preference settings is converted into natural language text that conforms to the user's daily communication habits. The specific implementation includes: First, through the semantic analysis algorithm in NLP, the core information of the execution result is understood, and it is associated with the home environment state, user historical behavior patterns, and personalized preference settings.

[0043] Then, based on the results of semantic understanding, a pre-trained language model (such as the GPT series model) or template matching technology is used to generate natural language text close to the user's daily communication habits. For example, if the execution result is "the air conditioner has been set to 24°C", combined with the current time (7 pm) and the user's historical behavior pattern (usually at home at 7 pm), the generated voice broadcast content may be: "Dear user, good evening! The air conditioner has been set to 24°C according to your habits, and I hope you have a pleasant evening at home." Further, the generated natural language text is input into a speech synthesis engine to convert it into natural, fluent, and easy-to-understand voice broadcast content. The speech synthesis engine supports multiple voice styles and volume size settings to meet the user's individual needs.

[0044] In addition, the electronic device can broadcast the voice broadcast content through built-in or external speakers, sound systems, etc., allowing the user to immediately understand the execution result of the home device, improving the interaction experience between the user and the home system, and making the smart home more thoughtful and more humanized.

[0045] In an optional embodiment, the method further comprises: Obtaining feedback voice data of the user on the execution result; Determining whether the user is satisfied with the execution result based on the feedback voice data through an AI big data model; If the user is not satisfied with the execution result, re-generating the control instruction based on the feedback voice data by calling the AI big data model; Controlling the target home device based on the re-generated control instruction.

[0046] The feedback voice data can include but is not limited to the user's evaluation of the device execution result, the expected improvement direction or specific operation suggestion, and can be used as an important basis for subsequent analysis and decision-making. For example: after setting the air conditioner temperature to 26°C, the user feedbacks "I hope the air conditioner temperature can be set to 24°C in the future". In some embodiments, the user's feedback voice data can be collected through a voice recognition component (such as a microphone) located in the home environment.

[0047] After obtaining the user's feedback voice data, the AI big data model determines whether the user is satisfied with the execution result based on the feedback voice data. Specifically, the AI big data model in this embodiment has been trained a large amount of times and can accurately identify the emotional tendency, semantic content and potential needs in the voice, so as to intelligently judge the specific satisfaction level of the user on the current execution result based on this information.

[0048] If the analysis result shows that the user is not satisfied with the execution result, the system will call the AI big data model to regenerate more accurate control instructions that meet the user's expectations according to the specific dissatisfaction points in the feedback voice data. For example, according to the user's feedback voice data "I hope to set the air conditioner temperature to 24 degrees Celsius in the future", the system will regenerate the control instruction "adjust the indoor temperature to 24 degrees Celsius".

[0049] Then, the target home device is controlled based on the regenerated control instruction, realizing the intelligentization and adaptive optimization of the home device, and improving the interaction experience between the user and the home device.

[0050] The above optional implementation analyzes the feedback voice data through the AI big data model, judges the user's satisfaction in real time, generates new control instructions and controls the home device when the user is not satisfied, significantly improves the response speed and user experience, reduces manual intervention, and enhances the intelligent level of the home.

[0051] In an optional embodiment, in order to further improve the user experience and continuously optimize the system performance, the method further comprises: Recording the number of negative feedbacks of the user's dissatisfaction with the execution result; If the number of consecutive negative feedbacks reaches the negative feedback number threshold, the user habit data is reconstructed based on the feedback voice data corresponding to the number of consecutive negative feedbacks; The AI big data model is updated based on the reconstructed user habit data.

[0052] The number of negative feedbacks is the number of times the user expresses dissatisfaction with the execution result. After each interaction, the AI big data model judges whether the user is satisfied with the execution result based on the feedback voice data. If it is determined that the user is not satisfied, a negative feedback is recorded.

[0053] The negative feedback number threshold is a standard for judging whether the number of consecutive negative feedbacks needs to be reconstructed. For example, when the system detects that the number of consecutive negative feedbacks exceeds three times, the system will extract the voice content related to these negative feedbacks from the stored feedback voice data. Then, the voice recognition technology is used to convert these voice contents into text information. The converted text information is analyzed and mined to extract the specific reasons and needs of the user's dissatisfaction with the execution result. For example, if multiple negative feedbacks mention "too hot", it can be inferred that the user may want to lower the indoor temperature. Based on the analyzed information data, the user habit data is reconstructed or refined to form a more personalized model that meets the user's real needs.

[0054] Subsequently, the reconstructed user habit data is input as new training data into the AI big data model for training to realize iterative updating of the AI big data model, which can involve adjusting model parameters, optimizing algorithm logic, or introducing new feature dimensions, aiming to enable the AI big data model to better capture and predict user behavior, thereby providing more accurate and personalized services in future interactions. After training, the updated model parameters are saved to the system for use in subsequent home device control. The updated model can provide more accurate and personalized control recommendations, thereby improving user satisfaction.

[0055] Notably, the negative feedback frequency threshold is not fixed but a dynamically adjustable parameter. The system intelligently adjusts the negative feedback frequency threshold according to actual application scenarios, user characteristics, and historical data feedback to balance user experience and system efficiency, thereby improving user experience.

[0056] The above optional implementation can capture user dissatisfaction with the execution result by recording the number of negative feedbacks and, when the number of consecutive negative feedbacks reaches the threshold, reconstruct the user habit data using the corresponding feedback voice data to effectively mine the user's potential needs and preferences. After updating the AI big data model based on the reconstructed data, the model's understanding of user habits is more accurate, enabling it to anticipate user intent, optimize execution results, and reduce future interactions. In this way, a closed loop of user feedback and model optimization is achieved, significantly improving the system's intelligence level and user experience.

[0057] Specifically, as shown in the accompanying drawings, Figure 2 The negative feedback frequency threshold is adjusted as follows: Call the AI big data model to perform sentiment analysis based on the latest feedback voice data to obtain a sentiment score; Call the AI big data model to perform acoustic feature analysis based on the latest feedback voice data to obtain an acoustic feature score; Determine a first threshold adjustment weight factor based on the sentiment score; Determine a second threshold adjustment weight factor based on the acoustic feature score; Obtain a target threshold adjustment weight factor based on the first threshold adjustment weight factor and the second threshold adjustment weight factor; Adjust the preset negative feedback frequency baseline threshold based on the target threshold adjustment weight factor to obtain the negative feedback frequency threshold.

[0058] Emotion analysis is the process of identifying and analyzing emotions expressed in text or speech data. In this embodiment, the AI big data model performs emotion analysis on the feedback speech data to determine whether the user is satisfied, dissatisfied, or in another emotional state, and outputs a corresponding emotion score. The emotion score is a quantitative representation of the results of emotion analysis on the user's latest feedback speech data, reflecting the intensity and type of emotion expressed in the user's feedback speech (e.g., strong dissatisfaction, moderate dissatisfaction, slight dissatisfaction, etc.). The higher the score, the more intense the user's dissatisfaction. For example: the feedback speech data of the user's "too hot" with a rapid tone is calculated to have an emotion score of 85, indicating the user's strong dissatisfaction.

[0059] Specifically, the AI big data model uses its internal emotion analysis algorithm to process the latest feedback speech data, extracting emotion features from the speech, such as tone changes, speech speed, and tone strength. Based on the extracted emotion features, the emotion score is calculated and output. This score reflects the emotional tendency and intensity expressed in the user's feedback speech. For example, a pre-trained emotion analysis model (such as a deep learning-based emotion classifier) can be used to extract emotion features from the user's feedback speech data. This model is trained on a large amount of labeled speech data and can accurately identify the emotional tendency (such as positive, negative, neutral) and intensity in the speech. Based on the output of the emotion analysis model, an emotion score is assigned to each piece of feedback speech data. The emotion score is normalized and ranges from 0 to 1, where 0 represents extreme negativity and 1 represents extreme positivity. The specific calculation formula is: Emotion Score = (Model Output Positive Probability - Model Output Negative Probability) / 2 + 0.5.

[0060] Acoustic feature analysis is the process of analyzing the acoustic characteristics of speech signals. For example, acoustic features can be extracted from user feedback speech data, including but not limited to tone, volume, and speech speed. These features are extracted and quantified through signal processing techniques such as short-time Fourier transform, mel-frequency cepstral coefficients, etc. Acoustic features include but are not limited to tone, volume, and speech speed, and by analyzing these features, the emotional state and speech characteristics of the user's speech feedback can be understood. The acoustic feature score is a quantitative representation of the results of acoustic feature analysis on the user's latest feedback speech data, reflecting the degree of matching of the acoustic features of the feedback speech to a certain standard or pattern. The higher the score, the more negative emotional signals the user's speech may contain. For example: the user says "too hot" with a rapid tone and high pitch, and a large volume, and the feedback speech data is calculated to have an acoustic feature score of 95, indicating the user's strong dissatisfaction.

[0061] Specifically, the latest feedback voice data is input into the AI big data model, the AI big data model uses an acoustic feature analysis algorithm to perform spectral analysis, formant analysis and other operations on the voice signal, and extracts acoustic features such as tone, volume, and speech rate. According to the extracted acoustic features, the model calculates and outputs an acoustic feature score. This score represents the matching degree of the acoustic features of the feedback voice with a certain negative emotion related pattern. For example, a weight can be assigned to each acoustic feature, and the acoustic feature score is calculated according to the extracted acoustic feature value and its weight. The acoustic feature score is also normalized, with a range of 0 to 1. The specific calculation formula is:

[0062] wherein the maximum possible acoustic feature value is determined according to a predefined acoustic feature range.

[0063] Further, based on the emotion score and the acoustic feature score, the negative feedback frequency threshold can be dynamically adjusted to optimize the control strategy of the home device.

[0064] The first threshold adjustment weight factor (for the emotion score) and the second threshold adjustment weight factor (for the acoustic feature score) are set according to the relative importance of the emotion score and the acoustic feature score. For example, the first threshold adjustment weight factor can be set to 0.6 and the second threshold adjustment weight factor to 0.4, indicating that the emotion score has a greater impact on the negative feedback frequency threshold adjustment.

[0065] Specifically,

[0066] When the emotion score and the acoustic feature score are both high, the target threshold adjustment weight factor will also be high, but it will not directly cause the negative feedback frequency threshold to increase, but will be used for subsequent threshold adjustment calculation.

[0067] Then a reverse adjustment strategy is adopted, that is, when the emotion score and the acoustic feature score are high (indicating that the user is satisfied with the execution result), the negative feedback frequency threshold is reduced to reduce unnecessary regeneration of control instructions; when the emotion score and the acoustic feature score are low (indicating that the user is not satisfied with the execution result), the negative feedback frequency threshold is increased to allow more attempts to regenerate control instructions. The specific adjustment formula is: wherein the adjustment coefficient is a constant greater than 0, used to control the amplitude of the threshold adjustment.

[0068] The negative feedback frequency reference threshold is a pre-set basic threshold, which serves as a reference for adjusting the negative feedback frequency threshold. For example, the negative feedback frequency reference threshold is three times.

[0069] Since the sentiment score can reflect the emotional tendency in user feedback, such as positive, negative or neutral. The acoustic feature score provides information about the physical characteristics of the user feedback voice, such as tone, speed, volume, etc. These features can reveal the emotional intensity and expression of the user feedback. Combined with the sentiment score and acoustic feature score, the subtle differences in user feedback can be captured more comprehensively, reducing missed or false judgments, and thus more accurately assessing user satisfaction and potential problems. High sentiment score and high acoustic feature reduce the negative feedback frequency threshold, thereby responding quickly. Low sentiment score and low acoustic feature increase the negative feedback frequency threshold, thereby avoiding over-response. Considering the sentiment score and acoustic feature score and adjusting the negative feedback frequency threshold accordingly, this adjustment method makes the threshold more flexible to adapt to the feedback habits of different user groups, thereby improving the accuracy and reliability of the system.

[0070] In an optional embodiment, as shown in the description accompanying drawings Figure 3 The desensitization of the user habit data includes: Obtaining the remaining power of the electronic device; Reading the device privacy level corresponding to the user environment data from the preset configuration file; Determining the privacy protection strength according to the remaining power and the device privacy level; Determining the data desensitization mode according to the privacy protection strength; Desensitizing the user habit data by using the data desensitization mode to obtain the user habit desensitization data.

[0071] The remaining power of the electronic device refers to the percentage of the current battery capacity of the electronic device to the total capacity, usually represented by 0-100% of the electronic device. The remaining power of the electronic device can be obtained through the power query interface provided by the battery management system or the operating system of the electronic device. The battery status of the electronic device can indirectly affect the data transmission capability of the device and the flexibility of privacy protection measures (for example, the device has sufficient power to support high-intensity privacy protection measures; the device has moderate power, and needs to balance privacy protection and device endurance; the device has low power, and needs to prioritize device endurance, and the privacy protection strength needs to be reduced).

[0072] The preset configuration file is a file that has been pre-set and stored with device privacy level information corresponding to user environment data, which can be in the form of XML, JSON or database, etc.

[0073] The device privacy level is a privacy protection requirement level defined according to the environment data. Different privacy levels represent different degrees of strictness of user habit data protection. For example, the device privacy level can be divided into low, medium and high. If the device privacy level corresponding to the user environment data is low, it means that the user habit data is minimally protected. If the user environment data corresponds to a device privacy level of medium, it means that the user habit data is standardized. If the user environment data corresponds to a device privacy level of medium, it means that the user habit data is maximally protected.

[0074] For example, when the temperature is in the human comfort interval (generally considered to be 20-26°C), the device privacy level is set to a lower level. Because in this environment, the user is usually in a normal activity state, the risk perception of privacy leakage is relatively low, and the device can share some non-sensitive information more openly. If the temperature exceeds the comfort interval, for example, below 10°C or above 35°C, the device privacy level will be raised to a higher level. Extreme temperature may mean that the user is in a special environment (such as outdoor cold or high-temperature working environment), at which time the user's demand for privacy protection increases, and the device will limit the sharing of some data to protect the user's privacy and security.

[0075] When the humidity is in the appropriate range of 40%-60%, the device privacy level is the regular level. Under this humidity condition, the user's living and working environment is normal, and the device operates according to the regular privacy policy. When the humidity is less than 30% or more than 70%, the device privacy level is adjusted accordingly. Low humidity environment may occur in dry desert areas or indoor use of warm air in winter, and high humidity environment is common in humid coastal areas or summer rainy season. These special humidity environments may imply the particularity of the user's location, and in order to prevent potential risks caused by the leakage of location information, the device privacy level is increased.

[0076] Under normal light intensity during the day (for example, 1000lx-10000lx), the device privacy level is set to medium. Adequate light indicates that the user is in a daily activity scenario, and the device can normally collect and process environment-related data while following certain privacy protection rules. At night or in extremely low light intensity (less than 100lx), the device privacy level is raised to high. Low light environment may mean that the user is in a resting state or a more private space, at which time the device will strengthen the protection of user activity data, location information, etc., and reduce unnecessary data transmission and sharing.

[0077] When the air quality is good (Air Quality Index, AQI, is between 0-100), the device privacy level is basic. Good air quality represents a healthy and safe environment for the user, and the device can appropriately collect environmental data for analysis and optimization of services while ensuring basic privacy. When the air quality is poor (AQI is greater than 100), the device privacy level is increased. Polluted air can affect user health and may indicate specific environmental problems in the user's area. To protect user privacy and avoid unnecessary attention or risks due to environmental data leakage, the device will strengthen privacy protection measures.

[0078] If the user's location information shows that they are in a public place (such as a mall, park, station, etc.), the device privacy level is set according to the specific scenario. In general public places, the device privacy level is medium, allowing the device to collect and use location information under legal and compliant conditions to provide location-based services. When the user's location information points to sensitive areas (such as government offices, military bases, private residences, or unauthorized areas), the device privacy level immediately increases to the highest level. At this time, the device will strictly limit the collection, storage, and transmission of location information to prevent any unauthorized access and use, ensuring user privacy and safety.

[0079] When the doors and windows are closed, the device privacy level is higher. Closed doors and windows usually indicate that the user wants to maintain the privacy of the indoor space, so the device will reduce the collection and transmission of indoor environmental data, especially information that may leak privacy such as user activity tracks and sounds. When the doors and windows are open, the device privacy level is appropriately lowered, but still needs to follow certain privacy principles. Opening the doors and windows may mean that the user is in a normal ventilation or communication state with the outside world, so the device can collect some environmental data for improving indoor environment or providing related services while ensuring basic user privacy requirements, but will avoid collecting overly sensitive information.

[0080] When the system clock shows normal working hours on weekdays (such as Monday to Friday from 9:00 to 18:00), the device privacy level is regular. During this time period, the user is usually in a working or public activity state, and the device operates according to normal privacy policies, collecting and processing data related to work and life. During non-working hours (such as weekends, holidays, and from 18:00 on weekdays to 9:00 the next day), the device privacy level is increased to a higher level. During non-working hours, users pay more attention to personal privacy and rest, so the device will strengthen the protection of user activity data and device usage records, reduce unnecessary data collection and sharing, and avoid disturbing users or leaking privacy information.

[0081] Through the above correspondence, the preset configuration file will dynamically determine the privacy level of the device according to the real-time acquired user environment data, so as to reasonably use and manage the data collected by the device under the premise of protecting the user privacy.

[0082] The privacy protection strength represents a quantitative index of data protection strength. The local database of the electronic device stores a first data table, and the first data table stores a mapping relationship between the remaining power, the device privacy level, and the privacy protection strength. After obtaining the remaining power and the device privacy level of the electronic device, the privacy protection strength matched with the remaining power and the device privacy level is obtained from the data table. Different remaining power and device privacy levels correspond to different privacy protection strengths. For example, when the remaining power is sufficient and the device privacy level is high, a higher privacy protection strength is corresponded; when the remaining power is low and the device privacy level is low, a lower privacy protection strength is corresponded.

[0083] The data desensitization mode is a specific method for anonymizing or de-identifying data. The local database of the electronic device stores a second data table, and the second data table stores a mapping relationship between the privacy protection strength and the data desensitization mode. After determining the privacy protection strength, the data desensitization mode corresponding to the privacy protection strength can be matched according to the second data table. Different privacy protection strengths correspond to different data desensitization modes. For example, a high privacy protection strength corresponds to a first data desensitization mode, such as a strict data desensitization mode, i.e., complete encryption of sensitive information. A low privacy protection strength corresponds to a second data desensitization mode, such as a relaxed data desensitization mode, i.e., only partial sensitive information is blurred.

[0084] In an optional embodiment, the data desensitization mode can include the type of added noise, the encryption algorithm, the encryption level, whether the encryption can be delayed, the duration of the delayed encryption, etc. For example, for a strict data desensitization mode, an encryption algorithm can be used to encrypt sensitive information; for a relaxed data desensitization mode, a partial information hiding method can be used.

[0085] The user habit data is desensitized using the determined data desensitization mode, and the desensitized data is user habit desensitization data. Different desensitization techniques differ in protection effect and computational overhead.

[0086] Desensitization is a data protection technology, for example, by hiding, replacing, or encrypting sensitive information in user habit data, the processed data can retain a certain usability while not directly identifying the user's personal identity information, thereby protecting the user's privacy.

[0087] Specifically, sensitive information in the user habit data is identified. For example, the sensitive information can include information directly identifying the personal identity of the user, such as the name, address, and contact information of the user, and information indirectly inferring the personal identity, such as the association between a specific behavior pattern and the personal identity. For example, the real name of the user is replaced by a randomly generated code, and the middle four digits of the mobile phone number are replaced by “*”. After the sensitive information in the user habit data is desensitized according to the selected method, the user habit desensitized data is obtained.

[0088] Since high-strength privacy protection (such as encryption algorithms) will significantly increase the computing load of the electronic device, resulting in accelerated power consumption. In the home scenario, different areas (such as the bedroom and the living room) and time periods (such as night) have different privacy needs. By the residual power of the electronic device and the device privacy level corresponding to the user environment data, the privacy protection strength can be dynamically adjusted, and the data desensitization mode can be accurately determined based on the privacy protection strength, so as to perform desensitization processing. Both protect privacy and avoid resource waste. Automatically reduce the privacy protection strength when the power is low, avoid the device from shutting down due to power consumption, and reduce user distress.

[0089] It should be understood that, however, when the user habit data is desensitized, obtaining the user habit data corresponding to the user environment data includes obtaining the user habit desensitized data corresponding to the user environment data. Similarly, calling the preset AI big data model to generate the control instruction based on the user habit data includes calling the preset AI big data model to generate the control instruction based on the user habit desensitized data.

[0090] Embodiment 2 The embodiment provides an intelligent home device control system based on artificial intelligence.

[0091] As shown in the accompanying drawings Figure 4 The intelligent home device control system 40 based on artificial intelligence can include a plurality of function modules composed of program code segments. The program code of each program segment in the intelligent home device control system 40 based on artificial intelligence can be stored in the memory of the electronic device and executed by at least one processor to perform the functions of the intelligent home device control based on artificial intelligence (see Figure 1 Description).

[0092] In this embodiment, the home device control system 40 based on artificial intelligence can be divided into a plurality of function modules according to the functions performed by the system. The function modules can include a first acquisition module 401, a second acquisition module 402, a generation module 403, a control module 404, a broadcast module 405, a judgment module 406, a recording module 407, a reconstruction module 408, an update module 409, and a desensitization module 410. The module referred to in this application refers to a series of computer readable instruction segments that can be executed by at least one processor and can complete a fixed function, which is stored in the memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0093] The first acquisition module 401 is configured to acquire user environment data in real time. The second acquisition module 402 is configured to acquire user habit data corresponding to the user environment data. The generation module 403 is configured to call a preset AI big data model to generate a control instruction based on the user habit data. The control module 404 is configured to control a target home device based on the control instruction.

[0094] In an optional embodiment, after the control instruction is generated, the system further includes: The broadcast module 405 is configured to convert an execution result corresponding to the control instruction into voice broadcast content by the AI big data model, and broadcast the voice broadcast content.

[0095] In an optional embodiment, the system further includes: The judgment module 406 is configured to acquire feedback voice data of the user on the execution result, and determine whether the user is satisfied with the execution result based on the feedback voice data by the AI big data model. The generation module 403 is further configured to call the AI big data model to regenerate the control instruction based on the feedback voice data if the judgment module 406 determines that the user is not satisfied with the execution result.

[0096] The control module 404 is further configured to control the target home device based on the regenerated control instruction.

[0097] In an optional embodiment, the system further includes: The recording module 407 is configured to record a negative feedback frequency at which the user is not satisfied with the execution result. The reconstruction module 408 is configured to reconstruct the user habit data based on feedback voice data corresponding to a continuous negative feedback frequency if the continuous negative feedback frequency reaches a negative feedback frequency threshold. The update module 409 is configured to update the AI big data model based on the reconstructed user habit data. The negative feedback number threshold is a dynamically adjustable parameter.

[0098] In an optional embodiment, the negative feedback number threshold is adjusted in the following manner: The AI big data model is called to perform sentiment analysis on the latest feedback voice data to obtain a sentiment score. The AI big data model is called to perform acoustic feature analysis on the latest feedback voice data to obtain an acoustic feature score. A first threshold adjustment weight factor is determined based on the sentiment score. A second threshold adjustment weight factor is determined based on the acoustic feature score. A target threshold adjustment weight factor is obtained based on the first threshold adjustment weight factor and the second threshold adjustment weight factor. The preset negative feedback number reference threshold is adjusted based on the target threshold adjustment weight factor to obtain the negative feedback number threshold.

[0099] In an optional embodiment, the system further comprises: The desensitization module 410 is configured to perform desensitization processing on the user habit data to obtain user habit desensitization data. The second acquisition module 402 is further configured to acquire user habit desensitization data corresponding to the user environment data. The generation module 403 is further configured to call the preset AI big data model to generate the control instruction based on the user habit desensitization data.

[0100] In an optional embodiment, the desensitization processing on the user habit data to obtain user habit desensitization data comprises: The remaining power of the target home device is acquired. A device privacy level corresponding to the user environment data is read from a preset configuration file. A privacy protection strength is determined according to the remaining power and the device privacy level. A data desensitization mode is determined according to the privacy protection strength. The user habit data is desensitized according to the data desensitization mode to obtain the user habit desensitization data.

[0101] It should be understood that the various changes and specific embodiments of the artificial intelligence-based home device control method provided in the above embodiments are also applicable to the artificial intelligence-based home device control apparatus in the present embodiment. Those skilled in the art can clearly understand the implementation process of the artificial intelligence-based home device control apparatus in the present embodiment through the foregoing detailed description of the artificial intelligence-based home device control method. For the sake of brevity of the description, no further detailed description is given herein.

[0102] Embodiment 3 This embodiment proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements all or part of the steps of the artificial intelligence-based home appliance control method mentioned in the technical solution of Example 1.

[0103] This invention achieves intelligent, personalized control of home appliances by acquiring real-time user environment data and corresponding user habit data, and using AI big data models to generate control instructions. It can automatically adjust device operating status based on the user's real-time environment and habits, improving user comfort. It also reduces manual operation, improves device convenience, and makes home appliance control more precise and efficient, providing users with a better smart home experience.

[0104] Example 4 See Figure 5 FIG. 5 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. In a preferred embodiment of the present application, the electronic device 5 includes a memory 501 , at least one processor 502 , and at least one communication bus 503 .

[0105] Those skilled in the art should understand that Figure 5 The structure of the electronic device shown does not constitute a limitation of the embodiments of the present application. The electronic device 5 may also include more or less other hardware or software than shown in the figure, or a different arrangement of components.

[0106] In some embodiments, the electronic device 5 is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits, programmable gate arrays, digital processors, and embedded devices. The electronic device 5 may also include client devices, which include but are not limited to any electronic product capable of human-computer interaction with a client via a keyboard, mouse, remote control, touchpad, or voice-controlled device, such as a personal computer, tablet computer, smartphone, digital camera, etc.

[0107] It should be noted that the electronic device 5 is only an example. Other existing or future electronic products that are suitable for this application should also be included in the scope of protection of this application and included here by reference.

[0108] In some embodiments, the memory 501 stores a computer program and an operating system, and the computer program is executed by the at least one processor 502 to implement all or part of the steps of the dark watermark-based document provenance method as described. The memory 501 includes a Read-Only Memory (ROM), a Programmable Read-Only Memory (PROM), an Erasable Programmable Read-Only Memory (EPROM), a One-time Programmable Read-Only Memory (OTPROM), an Electrically-Erasable Programmable Read-Only Memory (EEPROM), a Compact Disc Read-Only Memory (CD-ROM) or other optical disk storage, a magnetic disk storage, a magnetic tape storage, or any other medium that can be used to carry or store computer-readable data. Further, the computer-readable storage medium can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, and the like.

[0109] In some embodiments, the at least one processor 502 is a control unit of the electronic device 5, which connects various components of the entire electronic device 5 through various interfaces and lines, and performs various functions of the electronic device 5 and processes data by running or executing programs or modules stored in the memory 501 and calling data stored in the memory 501. For example, the at least one processor 502 executes the computer program stored in the memory to implement all or part of the steps of the dark watermark-based document provenance method described in the embodiments of the present application, or to implement all or part of the functions of the dark watermark-based document provenance apparatus. The at least one processor 502 can be composed of integrated circuits, for example, can be composed of a single packaged integrated circuit, or can be composed of multiple packaged integrated circuits with the same function or different functions, including one or more combinations of a Central Processing Unit (CPU), a microprocessor, a digital processing chip, a graphics processor, and various control chips.

[0110] In some embodiments, the at least one communication bus 503 is configured to enable connection and communication between the memory 501 and the at least one processor 502. Although not shown, the electronic device 5 can further include a power supply (such as a battery) for powering the various components of the electronic device 5, and it is preferably connected to the at least one processor 502 logically through a power management apparatus, so as to realize the functions of managing charging, discharging, and power consumption management, etc. through the power management apparatus. The power supply can further include one or more DC or AC power sources, recharging means, power failure detection circuitry, power conversion or inverter circuitry, power status indicator, and the like. The electronic device 5 can further include various sensors, a Bluetooth module, a Wi-Fi module, an internal memory, a network interface, an input position and a display screen, etc., which are not described herein.

[0111] The integrated units in the form of software function modules described above can be stored in a computer readable storage medium. The software function modules described above are stored in a storage medium, and include a plurality of instructions for causing an electronic device (which can be a personal computer, an electronic device, or a network device, etc.) or a processor to execute part of the method described in each embodiment of the present application.

[0112] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the described apparatus embodiments are merely schematic, and the division of the modules is merely a logical function division, and there can be another division manner in actual implementation.

[0113] The modules described as separated components can or can not be physically separated, and the components displayed as modules can or can not be physical units, and can be located in one place, or distributed on a plurality of network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiments.

Claims

1. A home appliance control method based on artificial intelligence, characterized in that: The method comprises: Obtain user environment data in real time; Acquiring user habit data corresponding to the user environment data; Calling a preset AI big data model to generate control instructions based on the user habit data; The target home appliance is controlled based on the control instruction.

2. The artificial intelligence-based home appliance control method according to claim 1, characterized in that: After generating the control instruction, the method further includes: Converting the execution result corresponding to the control instruction into voice broadcast content through the AI ​​big data model; Perform voice broadcast on the voice broadcast content.

3. The artificial intelligence-based home appliance control method according to claim 2, characterized in that: The method further comprises: Obtaining user feedback voice data on the execution result; Determining whether the user is satisfied with the execution result based on the feedback voice data using the AI ​​big data model; If the user is not satisfied with the execution result, the AI ​​big data model is called to regenerate the control instruction according to the feedback voice data; The target home device is controlled based on the regenerated control instruction.

4. The artificial intelligence-based home appliance control method according to claim 3, characterized in that: The method further comprises: Record the number of negative feedbacks from users who are dissatisfied with the execution results; If the number of consecutive negative feedbacks reaches a negative feedback threshold, reconstructing the user habit data based on the feedback voice data corresponding to the number of consecutive negative feedbacks; Updating the AI ​​big data model based on the reconstructed user habit data; The negative feedback threshold is a dynamically adjustable parameter.

5. The artificial intelligence-based home appliance control method according to claim 4, characterized in that: The negative feedback threshold is adjusted as follows: Calling the AI ​​big data model to perform sentiment analysis based on the latest feedback voice data to obtain a sentiment score; Calling the AI ​​big data model to perform acoustic feature analysis based on the latest feedback voice data to obtain an acoustic feature score; determining a first threshold adjustment weighting factor based on the sentiment score; determining a second threshold adjustment weighting factor based on the acoustic feature score; Obtaining a target threshold adjustment weight factor based on the first threshold adjustment weight factor and the second threshold adjustment weight factor; The preset negative feedback times reference threshold is adjusted based on the target threshold adjustment weight factor to obtain a negative feedback times threshold.

6. The artificial intelligence-based home appliance control method according to claim 1, characterized in that: The method further comprises: Performing desensitization processing on the user habit data to obtain user habit desensitized data; The acquiring of user habit data corresponding to the user environment data includes: acquiring user habit desensitized data corresponding to the user environment data; The calling of the preset AI big data model to generate a control instruction based on the user habit data includes: calling the preset AI big data model to generate a control instruction based on the user habit desensitized data.

7. The artificial intelligence-based home appliance control method according to claim 6, characterized in that: The desensitizing the user habit data to obtain the desensitized user habit data includes: Get the remaining power of electronic devices; Reading a device privacy level corresponding to the user environment data from a preset configuration file; determining a privacy protection strength according to the remaining power and the device privacy level; Determining a data desensitization mode according to the privacy protection strength; The data desensitization mode is adopted to desensitize the user habit data to obtain user habit desensitized data.

8. A home appliance control system based on artificial intelligence, characterized in that: The system comprises: Collection module, used to obtain user environment data in real time; An acquisition module, configured to acquire user habit data corresponding to the user environment data; A generation module, configured to call a preset AI big data model to generate control instructions based on the user habit data; A control module is used to control the target home appliance based on the control instruction.

9. An electronic device, characterized in that: The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the artificial intelligence-based home appliance control method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the home appliance control method based on artificial intelligence according to any one of claims 1 to 7 are implemented.

Images