Automatic regulation and control method and system for smart home

Abstract

The invention relates to an automatic regulation and control method and system for a smart home. The automatic regulation and control method of the smart home comprises the steps of obtaining behavior data of a user; analyzing the behavior data to obtain a behavior pattern of the user; generating an operation strategy of the equipment based on the behavior pattern; performing linkage control on the equipment based on the operation strategy; continuously monitoring the behavior change of the user; updating an operation strategy of the equipment based on the behavior change of the user; and performing linkage control on the equipment based on the updated operation strategy. By deeply analyzing the behavior data of the user, a personalized equipment operation strategy can be constructed, accurate energy-saving control on the equipment can be realized, energy consumption can be effectively reduced, the use experience of the user can be remarkably improved, and a new breakthrough is brought to the development of an intelligent home system.

Description

Technical Field

[0001] This application relates to the field of smart home technology, and in particular to an automatic control method and system for smart homes. Background Technology

[0002] With the rapid development of IoT and AI technologies, smart home systems have gradually become an indispensable part of modern homes. Currently, existing smart home systems mainly rely on preset rules or simple sensor feedback to control device operation. For example, some systems automatically adjust light brightness based on light intensity, turning on the lights when the outside light dims; or they automatically adjust the air conditioner temperature based on indoor temperature, starting to cool when the indoor temperature exceeds the set value.

[0003] However, these systems currently have significant shortcomings; they often lack a deep understanding of users' daily habits. Take lighting control as an example: in some situations, even with sufficient indoor lighting, users may prefer brighter lighting because they are accustomed to reading in specific areas. Existing systems cannot personalize their settings to accommodate this, leading to a mismatch between device operation strategies and actual user needs, resulting in energy waste. Summary of the Invention

[0004] Therefore, it is necessary to provide an automatic control method and system for smart homes to address the problems in related technologies.

[0005] To achieve the above objectives, in a first aspect, this application provides an automatic control method for a smart home, wherein the smart home is used to automatically control devices; the automatic control method for the smart home includes:

[0006] Obtain user behavior data;

[0007] The behavioral data is analyzed to obtain the user's behavioral patterns;

[0008] The operating strategy of the device is generated based on the behavioral pattern;

[0009] Based on the aforementioned operating strategy, the equipment is controlled in a coordinated manner.

[0010] Continuously monitor changes in user behavior;

[0011] Update device operation strategies based on changes in user behavior;

[0012] The updated operating strategy is used to perform coordinated control of the equipment.

[0013] In some embodiments, acquiring user behavior data includes:

[0014] The user's behavioral data is obtained based on multiple sensors and device operation records. The behavioral data is multi-dimensional data including device usage, user activity intensity, and indoor environment data.

[0015] In some embodiments, the behavioral data is analyzed to obtain user behavior patterns, including:

[0016] The behavioral data is preprocessed;

[0017] Machine learning models are used to classify preprocessed behavioral data in order to obtain user behavior patterns in different scenarios.

[0018] In some embodiments, analyzing the behavioral data to obtain user behavior patterns further includes:

[0019] Use time series analysis methods to predict user behavior patterns over future time periods.

[0020] In some embodiments, generating the device's operating strategy based on the behavioral pattern includes:

[0021] The behavioral patterns are matched with device operation records to identify user device usage needs in different scenarios;

[0022] The decision tree algorithm is used to generate the operating strategy for the device.

[0023] In some embodiments, the device is controlled in a coordinated manner based on the operating strategy, including:

[0024] The operating status of the device is automatically controlled based on the operating strategy.

[0025] The device is controlled in a coordinated manner using a fuzzy logic control algorithm.

[0026] In some embodiments, updating the device's operating strategy based on changes in user behavior includes:

[0027] Based on changes in user behavior, reinforcement learning algorithms are used to update the device's operating strategy.

[0028] The updated operating strategy was optimized and verified.

[0029] In some embodiments, a machine learning model is used to analyze the behavioral data to obtain user behavior patterns; the automatic control method for smart homes further includes: optimizing the machine learning model, including:

[0030] The parameters of the machine learning model are adjusted based on the Bayesian optimization algorithm;

[0031] Introduce the behavioral data of new users using an online learning algorithm to gradually update the parameters of the machine learning model;

[0032] Evaluate the performance of the updated machine learning model;

[0033] Use the model integration method to integrate and fuse the machine learning models.

[0034] In some embodiments, the automatic control method for a smart home further includes:

[0035] The user makes a personalized adjustment to the operation strategy of the device and feeds the operation strategy of the personalized adjustment back to the machine learning model as user feedback data;

[0036] Based on the user feedback data, optimize the machine learning model using an incremental learning method;

[0037] Optimize the machine learning model based on the feedback mechanism of reinforcement learning.

[0038] In a second aspect, the present application further provides an automatic control system for a smart home, and the automatic control system for the smart home includes:

[0039] An acquisition module that acquires the behavioral data and behavioral changes of the user;

[0040] An analysis module for analyzing the behavioral data to obtain the behavioral pattern of the user;

[0041] An operation strategy generation module for generating the operation strategy of the device based on the behavioral pattern and updating the operation strategy of the device based on the behavioral changes of the user;

[0042] A linkage control module for performing linkage control on the device based on the operation strategy generated based on the behavioral pattern or the operation strategy updated based on the behavioral changes of the user.

[0043] In the above automatic control method and system for a smart home, acquire the behavioral data of the user; analyze the behavioral data to obtain the behavioral pattern of the user; generate the operation strategy of the device based on the behavioral pattern; perform linkage control on the device based on the operation strategy; continuously monitor the behavioral changes of the user; update the operation strategy of the device based on the behavioral changes of the user; perform linkage control on the device for use based on the updated operation strategy. By deeply analyzing the behavioral data of the user, a personalized device operation strategy can be constructed to achieve precise energy-saving control of the device, which can not only effectively reduce energy consumption but also significantly improve the user experience, bringing new breakthroughs to the development of the smart home system. Description of the Drawings

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of an automatic control method for smart homes provided in one embodiment of this application;

[0046] Figure 2 This is a structural block diagram of an automatic control system for smart homes provided in different embodiments of this application.

[0047] Explanation of reference numerals in the attached diagram: 10, Acquisition module; 20, Analysis module; 30, Operation strategy generation module; 40, Linkage control module. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] With the rapid development of IoT and AI technologies, smart home systems have gradually become an indispensable part of modern homes. Smart home systems primarily control device operation through preset rules or simple sensor feedback. For example, some systems automatically adjust light brightness based on ambient light intensity, brightening when the outside light dims; or automatically adjust air conditioning temperature based on indoor temperature, starting to cool when the indoor temperature exceeds the set value.

[0050] However, these smart home systems have significant drawbacks; they often lack a deep understanding of users' daily habits. Take lighting control as an example: in some situations, even with sufficient indoor lighting, users may prefer brighter lights because they habitually read in specific areas. The aforementioned smart home systems fail to personalize their settings based on this habit, leading to a mismatch between device operation strategies and actual user needs, ultimately resulting in energy waste.

[0051] The main challenge facing the aforementioned smart home systems lies in extracting meaningful patterns from complex user behavior data and transforming them into effective device operation strategies. While some systems have begun to utilize machine learning algorithms to analyze user behavior, these methods typically rely on limited data sources. For example, some systems analyze user air conditioning habits solely through indoor temperature sensors and time recordings, ignoring the impact of factors such as user activity levels and clothing on air conditioning demand. Furthermore, these methods struggle to adapt to dynamic changes in user habits. For instance, users' weekday and weekend routines may differ significantly; weekday users typically wake up early for work, while weekend users may wake up late. However, existing systems often fail to adjust device operation strategies in a timely manner, continuing to operate lights and air conditioning according to weekday schedules even when users wake up late on weekends, leading to energy waste or a degraded user experience.

[0052] Furthermore, the aforementioned smart home systems also have shortcomings in terms of device linkage control. While some systems can achieve simple device linkage, such as automatically turning off lights and appliances when a user leaves home, these linkages are usually based on fixed schedules or simple sensor feedback, lacking a deep understanding of user behavior patterns. For example, if a user only briefly leaves the room to retrieve something from the kitchen, the system might turn off all devices because it detects no one in the room, only to have them turned back on when the user returns. This not only wastes energy but also degrades the user experience. Therefore, these smart home systems still have significant room for improvement in terms of energy efficiency and user experience.

[0053] In one embodiment, see Figure 1 This application provides an automatic control method for smart homes, which includes the following steps: S10 to S70.

[0054] S10: Obtain user behavior data.

[0055] S20: Analyze the behavioral data to obtain the user's behavioral patterns.

[0056] S30: Generate the operating strategy of the device based on the behavior pattern.

[0057] S40: Perform linkage control on the device based on the operating strategy.

[0058] S50: Continuously monitor changes in user behavior.

[0059] S60: Updates device operating policies based on changes in user behavior.

[0060] S70: Perform linkage control on the device based on the updated operating strategy.

[0061] In the automatic control method for smart homes disclosed in this application, personalized device operation strategies can be constructed by deeply analyzing user behavior data, enabling precise energy-saving control of devices. This not only effectively reduces energy consumption but also significantly improves the user experience, bringing a new breakthrough to the development of smart home systems.

[0062] In step S10, please refer to step S10 in 1 to obtain user behavior data.

[0063] As an example, user behavior data can be acquired based on various sensors and device operation records. This behavior data is multi-dimensional, including device usage, user activity intensity, and indoor environmental data. Specifically, the various sensors can include temperature sensors, humidity sensors, light sensors, and motion sensors, etc. The device operation records can include light switching times, appliance usage periods, air conditioning adjustment records, etc. The behavior data not only includes the user's device usage at different times but also covers multi-dimensional information such as the user's activity intensity and environmental changes; for example, it can record when the user turns on the lights upon waking up in the morning, the frequency of using the rice cooker during dinner, and the user's habits of adjusting the air conditioning temperature while resting at night, etc.

[0064] In step S20, please refer to Figure 1 In step S20, the behavioral data is analyzed to obtain the user's behavioral pattern.

[0065] As an example, in order to extract meaningful behavioral patterns from complex behavioral data, machine learning algorithms can be used for deep analysis.

[0066] Specifically, step S20 may include the following steps: S201 to S202.

[0067] S201: Preprocess the behavioral data.

[0068] S202: Use machine learning models to classify preprocessed behavioral data to obtain user behavior patterns in different scenarios.

[0069] As an example, in step S201, the behavioral data can be cleaned and normalized to ensure the quality and consistency of the behavioral data.

[0070] As an example, in step S202, the machine learning model can use a clustering algorithm (e.g., K-means clustering) to classify the preprocessed behavioral data to identify user behavior patterns in different scenarios. For instance, by analyzing a user's device usage at different times such as morning, noon, and evening, a user's fixed daily routine can be identified.

[0071] As an example, step S20 may also include step S203.

[0072] S203: Use time series analysis methods (such as the ARIMA model) to predict user behavior patterns over future time periods.

[0073] As an example, the ARIMA model (Autoregressive Integral Moving Average) is a commonly used time series forecasting method, and its corresponding formula can be expressed as follows:

[0074] ARIMA(p,d,q)=φ p (B)(1-B) d X t =θ q (B)∈ t

[0075] Where p is the autoregressive order, d is the difference order, q is the moving average order, and φ p (V) and θ q (B) are the coefficient polynomials for autoregression and moving average, respectively, where B is the lag operator, and X... t For time series data, ∈ t This represents the white noise error term. The ARIMA model can predict future device usage patterns, allowing for proactive adjustments to device operation strategies.

[0076] In a specific example, sensors can collect data on user behavior between 7:00 and 8:00 AM, including light switching, air conditioning adjustments, and rice cooker usage. First, this data is cleaned and normalized to remove outliers and noise. Then, a K-means clustering algorithm is used to categorize the data, identifying consistent user behavior patterns in the morning, such as turning on lights at 7:30, adjusting the air conditioning at 7:40, and using the rice cooker at 7:50. Next, an ARIMA model is used to predict future user behavior patterns, such as predicting that the user will turn on lights at 7:35, adjust the air conditioning at 7:45, and use the rice cooker at 7:55 the next morning. Based on these predictions, personalized user behavior patterns are constructed, providing accurate data support for subsequent device operation strategies. Through this process, a deep understanding of users' daily habits can be achieved, providing solid data support for subsequent device operation strategies and enabling precise energy-saving control.

[0077] In step S30, please refer to Figure 1 In step S30, the operating strategy of the device is generated based on the behavior pattern.

[0078] As an example, in step S30, generating the operating strategy of the device based on the behavior pattern may include the following steps: S301 to S302.

[0079] S301: Match the behavior pattern with the device operation record to identify the user's device usage needs in different scenarios.

[0080] S302: Use a decision tree algorithm to generate the operating strategy for the device.

[0081] As an example, in step S301, it can be identified that the user has a fixed behavioral pattern such as turning on the lights at 7:30 am, adjusting the air conditioner temperature at 7:40 am, and using the rice cooker at 7:50 am.

[0082] As an example, in step S302, a decision tree algorithm (such as the CART algorithm) can be used to generate the device's operating strategy. The decision tree algorithm recursively divides the dataset into smaller subsets, ultimately generating a tree structure where each node represents a decision condition and each leaf node represents an operating strategy for the device. The mathematical formula for a decision tree can be as follows:

[0083]

[0084] Where D is the dataset, p i Let be the proportion of samples of class i in the dataset, Gini(D) be the Gini impurity of the dataset, and n be the number of samples in the dataset. By minimizing the Gini impurity, the optimal operating strategy for the device can be generated.

[0085] In step S40, please refer to Figure 1 In step S40, the device is controlled in conjunction with the operating strategy.

[0086] As an example, step S40, which involves performing linkage control on the device based on the operating strategy, may include the following steps: S401 to S402.

[0087] S401: Automatically control the operating status of the device based on the operating strategy.

[0088] S402: Use a fuzzy logic control algorithm to perform linkage control on the device.

[0089] As an example, in step S40, the operating status of the equipment can be automatically controlled according to the generated operating strategy. For instance, during the user's long-term away-from-home period, smart lights, TVs, and other devices can be automatically turned off to avoid unnecessary energy consumption. During lunch breaks or nighttime sleep, the air conditioner can be automatically adjusted to the energy-saving temperature to reduce the operating power of devices such as humidifiers and air purifiers, which can both protect the user's comfort experience and achieve energy-saving goals. During the user's usual mealtimes, smart rice cookers, ovens, and other home appliances can be automatically preheated to reduce standby energy consumption and improve energy utilization efficiency.

[0090] As an example, in step S40, a fuzzy logic control algorithm is used to implement the linkage control of the devices. The fuzzy logic control algorithm applies fuzzy rules to the input variables to generate fuzzy values ​​of the input variables, and then obtains the specific linkage control control signal through defuzzification.

[0091] As an example, the mathematical formula for a fuzzy logic control algorithm can be as follows:

[0092] μA∩B(x)=min(μA(x),μB(x))

[0093] Where, μ A (x) represents the membership degree of the input variable x in the fuzzy set A, μ B Let (x) be the membership degree of the input variable x in the fuzzy set B, and let μA∩B(x) be the membership degree of the input variable x in the fuzzy set A∩B. Through fuzzy logic control algorithms, the operation of equipment can be precisely controlled according to the user's actual needs, achieving energy-saving goals.

[0094] As an example, input variables may include time information (such as whether it is during the time away from home or during mealtime), environmental parameters (such as temperature, humidity, and light intensity), user behavior data (such as whether human activity is detected and the frequency of device use), and the current status of the device (such as the current temperature of the air conditioner and the current brightness of the lights), etc.

[0095] As an example, input variables can be mapped to the membership function of a fuzzy set to obtain fuzzy values ​​for the input variables. For instance, the input variable is "temperature," and the fuzzy set A is "comfortable temperature." The membership function is defined as: when the temperature is between 22℃ and 26℃, the membership degree μ... A (x) gradually increases from 0 to 1, exceeding 26℃, μ A (x) will gradually become 0. Then the fuzzy value μ of 25℃ in fuzzy set A. A (25) = 0.8.

[0096] As an example, defuzzification can be performed using the centroid method and the maximum membership method to obtain the control signal. Specifically, the centroid value of the fuzzy set C can be calculated and used as the control signal. For example, if the membership distribution of the fuzzy set "energy-saving temperature" in the range of 26℃ to 28℃ is: 26℃→0.3, 27℃→0.5, 28℃→0.2, then the centroid value is 27℃; the temperature value with the highest membership (i.e., the centroid value) is used as the control signal. Finally, the obtained precise value (27℃) is converted into a command that the device can recognize (such as an air conditioner temperature control signal) to achieve automatic adjustment.

[0097] As an example, the automatic regulation method of the smart home may further include the following steps: continuously monitor the operating status of the device and adjust the operating strategy of the device according to the actual operating conditions. For example, if it is detected that there are still devices running during the period when the user is away from home, these devices will be automatically turned off, and the abnormal behavior of the user will be recorded for subsequent analysis. Specifically, the state monitoring and feedback of the device can be implemented based on the Bayesian network algorithm. Through the Bayesian network algorithm, the operating strategy of the device can be dynamically adjusted according to the actual operating status of the device to ensure the effectiveness of the device operating strategy.

[0098] In a specific example, through the analysis results of user habits, identify fixed behavior patterns such as the user being accustomed to turning on the lights at 7:30 in the morning, adjusting the air conditioner temperature at 7:40, and using the rice cooker at 7:50. Use the decision tree algorithm to generate the operating strategy of the device, such as automatically turning on the lights at 7:30 in the morning, automatically adjusting the air conditioner temperature at 7:40, and automatically preheating the rice cooker at 7:50. Then, use the fuzzy logic control algorithm to implement device linkage control, such as automatically adjusting the light brightness according to the indoor light intensity and automatically adjusting the air conditioner temperature according to the indoor temperature. Continuously monitor the operating status of the device. For example, if it is detected that there are still devices running during the period when the user is away from home, these devices will be automatically turned off, and the abnormal behavior of the user will be recorded for subsequent analysis. In this way, the device operation can be accurately controlled according to the actual needs of the user to achieve the energy-saving goal.

[0099] In step S50, please refer to Figure 1 step S50 in

[0100] As an example, the behavior changes of the user can be continuously monitored through a variety of sensors (such as infrared sensors, motion sensors or smart device operation records, etc.); for example, it can be detected that the user wakes up later on weekends than on weekdays, or the user's work and rest time during holidays is different from usual.

[0101] As an example, a time series analysis algorithm (such as the ARIMA model) can be used to identify the change trend of user behavior. The mathematical formula of the ARIMA model can be as follows:

[0102]

[0103] where y t is the time series data, φ(B) and θ(B) are the lag operators of autoregression and moving average respectively, is the difference operator, ∈ t is the error term. Through the ARIMA model, the change trend of user behavior can be predicted, providing a basis for timely adjusting the operating strategy of the device subsequently.

[0104] In step S60, please refer to Figure 1 The S60 step updates the device's operating strategy based on changes in user behavior.

[0105] As an example, step S60, updating the device's operating strategy based on changes in user behavior, may include the following steps: S601 to S602.

[0106] S601: Update the device's operating strategy using reinforcement learning algorithms based on changes in user behavior.

[0107] S602: Optimize and verify the updated operating strategy.

[0108] As an example, in step S601, if it is detected that the user wakes up later on weekends than on weekdays, the on-times of smart lights and air conditioning are automatically adjusted to avoid wasting energy when the user is not awake. Specifically, reinforcement learning algorithms (such as Q-learning algorithms) can be used to update the device's operating strategy. Through Q-learning algorithms, the device's operating strategy can be dynamically adjusted based on changes in user behavior, ensuring the effectiveness of the operating strategy.

[0109] As an example, in step S602, the updated operating strategy can be periodically optimized and verified. The optimization process may include adjusting strategy parameters, introducing new data, and evaluating strategy performance. Specifically, a genetic algorithm can be used to optimize the device's operating strategy. The formula for the genetic algorithm can be as follows:

[0110]

[0111] Where f(x) represents the fitness function, w i f represents the weight. i (x) represents the sub-objective function. The genetic algorithm can find the optimal operating strategy, ensuring its accuracy and energy efficiency.

[0112] In a specific example, time series analysis algorithms (such as the ARIMA model) can detect that users wake up later on weekends than on weekdays, automatically adjusting the on-times of smart lights and air conditioning to avoid wasting energy while the user is still asleep. Next, reinforcement learning algorithms (such as Q-learning) are used to update the device control strategy, such as adjusting the on-times of lights and air conditioning. The updated device control strategy is periodically optimized and validated, and a genetic algorithm is used to find the optimal device control strategy. In this way, the operating strategy of the devices can be dynamically adjusted according to the user's actual needs, achieving precise energy saving.

[0113] In step S70, please refer to Figure 1In step S70, the device is controlled in conjunction with the updated operating strategy.

[0114] As an example, the automatic control method for smart homes in this application may further include the following steps: S80: Optimize the machine learning model.

[0115] As an example, step S80 may include the following steps: S801 to S804.

[0116] S801: Adjust the parameters of the machine learning model based on the Bayesian optimization algorithm.

[0117] S802: Introduce new user behavior data using an online learning algorithm to progressively update the parameters of the machine learning model.

[0118] S803: Perform performance evaluation on the updated machine learning model.

[0119] S804: Use model ensemble methods to integrate and fuse machine learning models.

[0120] As an example, in step S801, the parameters of the machine learning model can be fine-tuned periodically to improve the accuracy of user behavior pattern recognition. Specifically, a Bayesian optimization algorithm can be used to adjust the parameters of the machine learning model. The core idea of ​​the Bayesian optimization algorithm is to approximate the objective function by constructing a surrogate model (such as a Gaussian process) and to select the next evaluation point by sampling a function (such as the desired improvement in EI). The formula for the Bayesian optimization algorithm can be as follows:

[0121]

[0122] Where, x t+1 For the next evaluation point, For acquisition functions, Given existing observation data, Bayesian optimization can efficiently find the optimal parameters for a machine learning model, thereby improving its predictive performance.

[0123] As an example, in step S802, new user behavior data can be periodically introduced to update and optimize the machine learning model. Specifically, online learning algorithms (such as stochastic gradient descent (SGD)) can be used to progressively update the parameters of the machine learning model. Through online learning algorithms, new data can be used to update the machine learning model in a timely manner, maintaining its timeliness and accuracy.

[0124] As an example, in step S803, the optimized machine learning model can be periodically evaluated to ensure its effectiveness in practical applications. Specifically, cross-validation can be used to evaluate the performance of the machine learning model. Cross-validation allows for a comprehensive evaluation of the model's performance, ensuring its stability and reliability across different datasets.

[0125] As an example, in step S804, model ensemble methods (such as stacking) can be used to further improve the accuracy of user behavior pattern recognition. The core idea of ​​the stacking method is to use the prediction results of multiple base models as input to train a meta-model to generate the final prediction result. The formula for the stacking method can be as follows:

[0126]

[0127] Among them, h i (x) represents the prediction result of the i-th base model, f meta This is a meta-model. By using model ensemble methods, the advantages of multiple models can be combined to improve the accuracy and robustness of predictions.

[0128] In a specific example, Bayesian optimization algorithms can be used to adjust the parameters of a machine learning model, such as the learning rate and regularization coefficient, to improve its predictive performance. Next, new user behavior data, such as users' daily routines during holidays, is introduced, and stochastic gradient descent is used to progressively update the model parameters. The optimized machine learning model is then periodically evaluated, and cross-validation is used to ensure its stability and reliability across different datasets. Finally, a stacking method is used to integrate the predictions from multiple base models, training a meta-model to generate the final prediction. This approach continuously improves the accuracy of user behavior pattern recognition, ensuring the precision of device operation strategies and energy-saving performance.

[0129] As an example, the automatic control method for smart homes in this application may also include the following steps:

[0130] S901~S903.

[0131] S901: The user makes personalized adjustments to the operating strategy of the device and feeds the personalized operating strategy back to the machine learning model as user feedback data.

[0132] S902: Based on the user feedback data, the machine learning model is optimized using an incremental learning method.

[0133] S903: Optimize the machine learning model based on the feedback mechanism of reinforcement learning.

[0134] As an example, in step S901, the user is allowed to personalize the device's operating strategy, and the personalized operating strategy is fed back to the machine learning model as user feedback data. Specifically, the user can personalize the device's operating strategy through various channels (such as mobile applications, voice assistants, and smart home control panels). For example, the user can manually adjust the air conditioner's temperature setting, the brightness of lights, and the on / off times of appliances through a mobile application. These adjustments are recorded in real time and input into the machine learning model as new training data.

[0135] As an example, in step S902, based on the user feedback data, an incremental learning method can be used to process the user feedback data. Incremental learning is a learning method that can progressively update the parameters of a machine learning model, allowing the machine learning model to be optimized using new data without retraining the entire model. Specifically, an online learning algorithm (such as stochastic gradient descent) is used to progressively update the parameters of the machine learning model.

[0136] As an example, step S903 can be based on a reinforcement learning feedback mechanism. Reinforcement learning is a learning method that optimizes decisions through reward and punishment mechanisms. In smart homes, user adjustments to device operating strategies can be viewed as a "reward" or "punishment" signal. For example, if a user lowers the air conditioner temperature, it can be seen as a "punishment" signal, indicating that the current temperature setting does not meet the user's needs; conversely, if the user keeps the air conditioner temperature unchanged, it can be seen as a "reward" signal, indicating that the current temperature setting meets the user's needs. Based on these signals, the device's operating strategy can be adjusted to better meet user needs.

[0137] In a specific example, a user adjusts the air conditioner temperature from 24℃ to 22℃ via a mobile app. This adjustment is recorded in real time and used as new training data to feed into a machine learning model. Incremental learning is used to progressively update the model parameters to adapt to the user's new needs. Simultaneously, based on a reinforcement learning mechanism, the user's temperature adjustment is treated as a "penalty" signal, adjusting the air conditioner control strategy to prevent similar temperature settings from failing to meet user needs in the future. In this way, the accuracy of user behavior pattern recognition can be continuously improved, ensuring the precision of the device's operating strategy and energy-saving performance.

[0138] It should be understood that, although Figure 1The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0139] In another embodiment, please refer to Figure 2 This application also provides an automatic control system for smart homes, which may include: an acquisition module 10, an analysis module 20, an operation strategy generation module 30, and a linkage control module 40; wherein, the acquisition module 10 acquires user behavior data and user behavior changes; the analysis module 20 analyzes the behavior data to obtain user behavior patterns; the operation strategy generation module 30 generates an operation strategy for the device based on the behavior patterns and updates the device's operation strategy based on user behavior changes; and the linkage control module 40 performs linkage control on the device based on the operation strategy generated from the behavior patterns or the operation strategy updated based on user behavior changes.

[0140] As an example, the automatic control system of the smart home in this embodiment can be used to perform actions such as... Figure 1 And the automatic control method for smart homes described in the relevant embodiments.

[0141] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features of the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0142] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An automatic control method for smart homes, characterized in that, The smart home system is used for automatic control of devices; the automatic control method of the smart home system includes: Obtain user behavior data; The behavioral data is analyzed to obtain the user's behavioral patterns; The operating strategy of the device is generated based on the behavioral pattern; Based on the aforementioned operating strategy, the equipment is controlled in a coordinated manner. Continuously monitor changes in user behavior; Update device operating strategies based on changes in user behavior; The updated operating strategy is used to perform coordinated control of the equipment.

2. The method according to claim 1, characterized in that, Obtain user behavior data, including: The user's behavioral data is obtained based on multiple sensors and device operation records. The behavioral data is multi-dimensional data including device usage, user activity intensity, and indoor environment data.

3. The method according to claim 1, characterized in that, Analyzing the behavioral data to obtain user behavior patterns includes: The behavioral data is preprocessed; Machine learning models are used to classify preprocessed behavioral data in order to obtain user behavior patterns in different scenarios.

4. The method according to claim 3, characterized in that, Analyzing the behavioral data to obtain user behavior patterns also includes: Use time series analysis methods to predict user behavior patterns over future time periods.

5. The method according to claim 1, characterized in that, Generate the device's operating strategy based on the behavioral pattern, including: The behavioral patterns are matched with device operation records to identify user device usage needs in different scenarios; The decision tree algorithm is used to generate the operating strategy for the device.

6. The method according to claim 1, characterized in that, Based on the aforementioned operating strategy, the device is controlled in a coordinated manner, including: The operating status of the device is automatically controlled based on the operating strategy. The device is controlled in a coordinated manner using a fuzzy logic control algorithm.

7. The method according to claim 1, characterized in that, Update device operating policies based on changes in user behavior, including: Based on changes in user behavior, reinforcement learning algorithms are used to update the device's operating strategy. The updated operating strategy was optimized and verified.

8. The method according to any one of claims 1 to 7, characterized in that, The behavioral data is analyzed using machine learning models to obtain user behavior patterns. The automatic control method for smart homes further includes: optimizing the machine learning model, including: The parameters of the machine learning model are adjusted based on the Bayesian optimization algorithm; New user behavior data is introduced using online learning algorithms to progressively update the parameters of the machine learning model; The performance of the updated machine learning model is evaluated; Model ensemble methods are used to integrate and fuse machine learning models.

9. The method according to claim 8, characterized in that, The automatic control methods for smart homes also include: Users can personalize the operating strategy of the device and feed the personalized operating strategy back to the machine learning model as user feedback data. Based on the user feedback data, the machine learning model is optimized using an incremental learning method; The machine learning model is optimized based on the feedback mechanism of reinforcement learning.

10. An automatic control system for smart homes, characterized in that, The smart home's automatic control system includes: The acquisition module allows users to obtain user behavior data and changes in user behavior. The analysis module is used to analyze the behavioral data to obtain the user's behavioral patterns; The operation strategy generation module is used to generate the operation strategy of the device based on the behavior pattern, and update the operation strategy of the device based on changes in user behavior. The linkage control module is used to perform linkage control on the device based on the operating strategy generated based on the behavior pattern or the operating strategy updated based on changes in user behavior.

Images