Smart home generation strategy optimization system based on adaptive learning

Abstract

The invention belongs to the technical field of smart home control, and particularly relates to a smart home generation strategy optimization system based on adaptive learning, and the system comprises the steps: collecting the operation history, sensor data and equipment use modes of a user in a smart home system, and determining an evaluation standard and an optimization direction; extracting features according to user behaviors, environment changes and equipment states, and constructing a feature matrix; selecting a reinforcement learning method for modeling according to an optimization problem, designing an adaptive learning mechanism, and enabling the model to be continuously adjusted according to real-time data in an operation process; according to the method, historical data is used for training the model, a reward mechanism in reinforcement learning is used for guiding the implementation of an optimization target, a control strategy adapting to the current environment and demand is generated based on model prediction, and the method has the effects of better meeting the use demand of a user for the smart home and improving the intelligent level of the system and the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of smart home control, and specifically to a smart home generation strategy optimization system based on adaptive learning. Background Art

[0002] As a part of modern technological life, smart homes aim to provide users with a more convenient and intelligent living experience by integrating various smart devices. With the continuous development of artificial intelligence technology, adaptive learning methods have been widely applied in smart home systems, which can automatically generate and optimize control strategies for the home environment according to users' behavior habits and preferences. However, existing adaptive learning methods still have the problem of low learning efficiency when facing complex and changeable home environments and diverse user needs. Specifically, current adaptive learning systems often require a large amount of time to collect specific user behavior data to generate relatively accurate optimization strategies, which not only increases the time cost of users but also limits the real-time performance and flexibility of the system. In addition, due to insufficient training data or low data quality, adaptive learning systems often suffer from misjudgment or overfitting, thus affecting user experience and the normal operation of the home system.

[0003] To address the above challenges, researchers have proposed a series of optimization methods. Among them, algorithm optimization based on machine learning is one of the main directions, and such methods attempt to improve the performance of adaptive learning systems by improving training algorithms or introducing new optimization techniques. Although these methods have improved the system performance to a certain extent, they still face problems such as strong data dependence and high overfitting risk in practical applications. Therefore, there is an urgent need to develop a more efficient and robust adaptive learning algorithm and corresponding generation strategy optimization method to better meet users' usage requirements for smart homes and improve the intelligence level and user experience of the system.

[0004] In view of the above technical deficiencies, a solution for a smart home generation strategy optimization system based on adaptive learning is now proposed. Summary of the Invention

[0005] To solve the above problems, the present invention provides the following technical solutions:

[0006] A smart home generation strategy optimization method based on adaptive learning, comprising:

[0007] Collecting the operation history, sensor data, and device usage patterns of users in the smart home system to determine evaluation criteria and optimization directions;

[0008] Extracting features according to user behavior, environmental changes, and device status to construct a feature matrix;

[0009] Select a reinforcement learning method for modeling according to the optimization problem, design an adaptive learning mechanism, and let the model continuously adjust according to real-time data during operation;

[0010] Use historical data to train the model, utilize the reward mechanism in reinforcement learning to guide the realization of the optimization goal, and generate a control strategy that adapts to the current environment and requirements based on the model prediction.

[0011] Furthermore, the collection of user information includes recording the direct operation behaviors of users in the smart home system, counting the device usage frequency, time period distribution, and user preference settings, collecting indoor and outdoor environmental parameters in real time through environmental sensors, monitoring the device status through the built-in sensors of the device, extracting the linkage relationships between devices, identifying user habits, removing sensor noise and filling in missing data for the collected information data, standardizing the multi-source data format, extracting time series features, constructing a user behavior portrait, and quantifying the relevance between the environment and the device status.

[0012] Furthermore, the construction of the feature matrix is carried out by recording the user behavior characteristics, including:

[0013] Statistical features, daily average operation times, peak periods, device usage duration; sequence patterns, operation sequences, modeling the state transition probability through a Markov chain; preference labels, classifying users into categories based on clustering; time series features, sliding window statistics, periodic features; event triggers, abrupt change detection; device status features, device start-stop times, continuous working duration, energy consumption efficiency, device association graph and device dependency relationship.

[0014] Furthermore, the adaptive learning includes defining a reinforcement learning framework, including a state space, an action space, and a composite reward function; constructing a state vector based on real-time environmental data, device status, and user behavior; selecting a reinforcement learning algorithm to model the policy network; designing an adaptive learning mechanism to dynamically adjust the reward function weights, exploration rate, and learning rate; updating the policy network parameters through online learning and experience replay; introducing a safety constraint mechanism to ensure the stability and compliance of device operation;

[0015] The calculation of the reward composite function is as follows:

[0016] R = α·Renergy + β·Rcomfort - γ·Rwear,

[0017] where α, β, γ are weight parameters, R energy is the energy-saving reward, R comfort is the comfort reward, R wear is the device loss penalty;

[0018] The reward policy update formula is as follows:

[0019]

[0020] Among them, E π is the policy network parameter, G t is the cumulative return, T and is the time.

[0021] Furthermore, the training of the model using historical data includes statistically analyzing the power consumption of the device per unit time, establishing a benchmark energy consumption model, comprehensively scoring through sensor data and user manual adjustment of frequency, analyzing the correlation between the continuous working duration of the device, the start-stop times and the loss, dynamically adjusting the index weights according to the user's historical operation data, supporting the user to customize the scoring rules, converting the evaluation criteria into an objective function, adding constraint conditions, using reinforcement learning to balance multi-objective conflicts through a reward function, using an evolutionary algorithm to generate a Pareto optimal solution set for the user or the system to automatically select, triggering policy re-optimization based on real-time data, updating and optimizing the model parameters through a feedback loop, using historical data replay to verify the effectiveness of the optimized policy, simulating extreme scenarios, collecting the user's satisfaction score for the automatic policy, monitoring the user's manual intervention frequency, judging the matching degree between the policy and the user's needs, re-training the adaptive learning module according to the feedback data, and regularly adjusting the weights of the evaluation criteria to adapt to seasonal or user habit changes.

[0022] Furthermore, the policy optimization identifies the user preference type based on the device usage pattern, defines the quantitative indicators of energy consumption, comfort, and device life through a multi-objective optimization algorithm, dynamically adjusts the weight allocation of the evaluation indicators in combination with user feedback, constructs a set of constraint conditions for the device control policy, including the device power upper limit and the user activity time window, uses a reinforcement learning algorithm to generate a candidate policy set, and screens the Pareto optimal solution through simulation testing.

[0023] According to one aspect of the present invention, there is provided a smart home generation policy optimization system based on adaptive learning, including:

[0024] A data acquisition module for real-time obtaining the status data, user behavior data, and environmental data of smart home devices;

[0025] An adaptive learning module, connected to the data acquisition module, for training a machine learning model according to the user behavior data and environmental data to generate an initial policy set;

[0026] A policy generation module for dynamically generating a device control policy based on the initial policy set and real-time data;

[0027] A policy optimization module for adjusting the control policy through a multi-objective optimization algorithm, and the optimization objectives include energy consumption efficiency, user comfort, and device life;

[0028] A feedback execution module, configured to execute the optimized policy and collect user feedback data, and input the feedback data into the adaptive learning module to iteratively update the model.

[0029] Furthermore, the adaptive learning includes establishing a personalized behavior pattern library based on user historical behavior data, dynamically evaluating and calculating rewards for the device control policy using a reinforcement learning algorithm, generating a prediction model that matches user preferences by combining the time series features in the environmental data, establishing weight functions for energy consumption, comfort level, and device wear, dynamically adjusting the weights according to user preferences, using a genetic algorithm or a particle swarm optimization algorithm to solve the Pareto optimal solution set, locally fine-tuning the policy according to real-time environmental data, and recording the device usage frequency, time period preferences, temperature / brightness preference settings, and device interlock triggers.

[0030] According to one aspect of the present invention, there is provided a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned method for optimizing a smart home generation policy based on adaptive learning are implemented.

[0031] According to one aspect of the present invention, there is provided a computer-readable storage medium, on which the steps of the above-mentioned method for optimizing a smart home generation policy based on adaptive learning according to any one of claims 1 to 6 are stored.

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] 1. In the method for optimizing a smart home generation policy based on adaptive learning of the present invention, by collecting the operation history, sensor data, and device usage patterns of users in the smart home system, the evaluation criteria and optimization directions are determined; features are extracted according to user behavior, environmental changes, and device status to construct a feature matrix; a reinforcement learning method is selected for modeling according to the optimization problem, and an adaptive learning mechanism is designed to enable the model to continuously adjust according to real-time data during operation; the model is trained using historical data, and the reward mechanism in reinforcement learning is used to guide the achievement of the optimization goal. Based on the model prediction, a control policy suitable for the current environment and requirements is generated, which can better meet the user's usage requirements for the smart home, improve the intelligent level of the system, and enhance the user experience.

[0034] 2. In an intelligent home generation strategy optimization system based on adaptive learning according to the present invention, a data acquisition module is used to obtain the status data, user behavior data, and environmental data of intelligent home devices in real time; an adaptive learning module is connected to the data acquisition module and is used to train a machine learning model according to the user behavior data and environmental data to generate an initial strategy set; a strategy generation module is used to dynamically generate device control strategies based on the initial strategy set and real-time data; a strategy optimization module adjusts the control strategy through a multi-objective optimization algorithm, and the optimization objectives include energy consumption efficiency, user comfort, and device lifespan; a feedback execution module is used to execute the optimized strategy and collect user feedback data, and input the feedback data into the adaptive learning module to iteratively update the model, achieving the effect of generating relatively accurate optimized strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings;

[0036] Figure 1 It is a schematic diagram of the overall intelligent home generation strategy optimization method based on adaptive learning according to the present invention;

[0037] Figure 2 It is a schematic flow diagram of an intelligent home generation strategy optimization system based on adaptive learning according to the present invention;

[0038] Figure 3 It is a schematic diagram of a computer device in an intelligent home generation strategy optimization system based on adaptive learning according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0040] As Figures 1-3 shown, the present application provides an intelligent home generation strategy optimization method based on adaptive learning, including:

[0041] S1: Collect the operation history, sensor data, and device usage patterns of users in the intelligent home system, and determine the evaluation criteria and optimization directions;

[0042] S2: Extract features according to user behavior, environmental changes, and device status, and construct a feature matrix;

[0043] S3: Select a reinforcement learning method for modeling according to the optimization problem, design an adaptive learning mechanism, and enable the model to continuously adjust according to real-time data during operation;

[0044] S4: Use historical data to train the model, utilize the reward mechanism in reinforcement learning to guide the achievement of the optimization goal, and generate a control strategy that adapts to the current environment and requirements based on model prediction.

[0045] In one embodiment, the home lighting system is optimized through adaptive learning to achieve a balance between energy conservation and user comfort.

[0046] Devices:

[0047] Smart bulb: Supports brightness adjustment (0%-100%).

[0048] Light sensor: Monitors the indoor light intensity in real time.

[0049] Human body sensor: Detects whether the user is in the room.

[0050] User requirements:

[0051] Daytime: Dynamically adjust the light brightness according to the natural light intensity to reduce energy consumption.

[0052] Nighttime: Provide appropriate brightness according to the user's location and activity status to improve comfort.

[0053] Historical data collection and preprocessing

[0054] Data collection:

[0055] The light sensor records the indoor light intensity every 5 minutes.

[0056] The human body sensor detects whether the user is in the room and records the change in the user's location.

[0057] The smart bulb records the brightness adjustment history.

[0058] Data preprocessing:

[0059] Clean abnormal data (such as abnormal values caused by light sensor failures).

[0060] Extract features: Calculate the average light intensity per hour, the duration of the user at home, etc.

[0061] Specifically, the collection of user information includes recording the direct operation behaviors of users in the smart home system, counting the device usage frequency, time period distribution, and user preference settings, collecting indoor and outdoor environmental parameters in real time through environmental sensors, monitoring the device status through built-in device sensors, extracting the linkage relationships between devices, identifying user habits, removing sensor noise and filling in missing data from the collected information data, standardizing the multi-source data format, extracting time series features, constructing user behavior portraits, and quantifying the correlation between the environment and the device status.

[0062] Specifically, the construction of the feature matrix is carried out by recording user behavior characteristics, including:

[0063] Statistical features, such as the average daily operation times, peak hours, and device usage duration; sequence patterns, such as operation sequences, and modeling the state transition probability through Markov chains; preference labels, classifying users into categories based on clustering; time series features, such as sliding window statistics and periodic features; event triggers, such as detecting abrupt changes; device status features, such as the number of device startups and stops, continuous working duration, and energy consumption efficiency, as well as the device association graph and device dependency relationships.

[0064] Specifically, the adaptive learning includes defining a reinforcement learning framework, including a state space, an action space, and a composite reward function; constructing a state vector based on real-time environmental data, device status, and user behavior; selecting a reinforcement learning algorithm to model the policy network; designing an adaptive learning mechanism to dynamically adjust the weights of the reward function, exploration rate, and learning rate; updating the policy network parameters through online learning and experience replay; introducing a safety constraint mechanism to ensure the stability and compliance of device operation;

[0065] The calculation of the composite reward function is as follows:

[0066] R = α·Renergy + β·Rcomfort - γ·Rwear,

[0067] where α, β, and γ are weight parameters, R energy is the energy-saving reward, R comfort is the comfort reward, R wear is the device wear penalty;

[0068] The update formula for the reward policy is as follows:

[0069]

[0070] where E π are the policy network parameters, G t is the cumulative return, T is the time.

[0071] Specifically, training the model using historical data includes: statistically analyzing the power consumption of the device per unit time to establish a benchmark energy consumption model; comprehensively scoring based on sensor data and user manual frequency adjustment; analyzing the correlation between the continuous working duration, start-stop times, and losses of the device; dynamically adjusting the index weights according to the user's historical operation data; supporting the user to customize the scoring rules; converting the evaluation criteria into an objective function; adding constraint conditions; using reinforcement learning to balance multi-objective conflicts through a reward function; using an evolutionary algorithm to generate a Pareto optimal solution set for the user or the system to automatically select; triggering policy re-optimization based on real-time data; updating and optimizing the model parameters through a feedback loop; using historical data replay to verify the effectiveness of the optimized policy; simulating extreme scenarios; collecting the user's satisfaction score for the automatic policy; monitoring the user's manual intervention frequency to judge the matching degree between the policy and the user's needs; re-training the adaptive learning module according to the feedback data; and regularly adjusting the weights of the evaluation criteria to adapt to seasonal or user habit changes.

[0072] Specifically, the policy optimization includes: identifying the user preference type based on the device usage pattern; defining the quantitative indicators of energy consumption, comfort, and device life through a multi-objective optimization algorithm; dynamically adjusting the weight allocation of the evaluation indicators in combination with the user feedback; constructing a set of constraint conditions for the device control policy, including the device power upper limit and the user activity time window; using a reinforcement learning algorithm to generate a candidate policy set, and screening the Pareto optimal solution through simulation testing.

[0073] In one embodiment, algorithm selection:

[0074] DQN (Deep Q-Network): Suitable for discrete action spaces (such as switch control).

[0075] PPO (Proximal Policy Optimization): Suitable for continuous action spaces (such as temperature regulation).

[0076] Training process:

[0077] Extract the state-action-reward sequence from historical data and randomly sample data through Experience Replay to alleviate temporal correlation.

[0078] Use the gradient descent method to update the policy network parameters, minimizing the mean square error between the predicted Q value and the target Q value (DQN) or the policy gradient loss (PPO).

[0079] Verify the model convergence: Stop training when the training loss function fluctuates less than 5% within 10 consecutive epochs.

[0080] Environmental perception and state input:

[0081] Obtain sensor data (such as the current room temperature, user location) and device status in real time.

[0082] Dynamically correct the user behavior characteristics in the state vector according to the user's schedule (such as work and rest time).

[0083] Model prediction and action generation:

[0084] Input the current state into the trained policy network to output the action probability distribution

[0085] Balance exploration and exploitation through the ε-greedy policy:

[0086] Randomly select actions with probability ε (exploration).

[0087] Select the action with the highest Q value with probability (1 - ε) (exploitation).

[0088] Apply safety constraints to high-risk actions (such as turning on multiple high-power devices simultaneously):

[0089] Use action masking to directly block operations beyond the circuit load capacity.

[0090] Online learning and parameter update:

[0091] Collect real-time interaction data and store it in an experience replay pool with rolling updates.

[0092] Every fixed time interval (such as every 30 minutes), sample data from the replay pool to fine-tune the model parameters.

[0093] If the user frequently manually overrides the policy (such as lowering the air conditioner temperature), increase the comfort weight. If extreme weather is detected (such as a high temperature warning), temporarily increase the energy-saving weight

[0094] Offline evaluation: Simulate the execution of the policy in historical data and calculate key metrics (such as the total energy consumption reduction rate, user satisfaction score). Compare with baseline methods (such as rule-based control, traditional PID algorithms) to verify the superiority of the policy.

[0095] Online testing:

[0096] Shadow Mode: Run the model policy and the actual control system in parallel, compare the results but do not execute actions. Statistically calculate the model prediction accuracy (such as the matching degree between actions and actual operations).

[0097] Progressive deployment: Initially, enable model control only for some devices (such as a single-room air conditioner), and gradually expand to the entire system.

[0098] Continuous monitoring and optimization: Record the long-term impacts after the execution of the policy (such as changes in device lifespan, migration of user habits).

[0099] Periodically (e.g., monthly) retrain the model to adapt to data distribution drift (e.g., seasonal changes).

[0100] Multi-objective adaptation: Balance the conflicting requirements of energy conservation, comfort, and device lifespan through dynamic reward weights and online learning. Safety and robustness: Incorporate action masks and real-time constraints to avoid system overload or failure caused by policy generation. Enhanced user participation: Optimize the model directly using user feedback (manual operations) to improve the personalized experience. Lightweight deployment: Support local operation on edge devices (such as smart gateways) to reduce cloud dependence and communication latency.

[0101] According to one aspect of the present invention, there is provided an intelligent home generation strategy optimization system based on adaptive learning, including:

[0102] A data acquisition module for real-time acquisition of status data, user behavior data, and environmental data of intelligent home devices;

[0103] An adaptive learning module, connected to the data acquisition module, for training a machine learning model based on user behavior data and environmental data to generate an initial policy set;

[0104] A policy generation module for dynamically generating device control policies based on the initial policy set and real-time data;

[0105] A policy optimization module for adjusting the control policy through a multi-objective optimization algorithm, and the optimization objectives include energy consumption efficiency, user comfort, and device lifespan;

[0106] A feedback execution module for executing the optimized policy and collecting user feedback data, and inputting the feedback data into the adaptive learning module to iteratively update the model.

[0107] Specifically, the adaptive learning includes establishing a personalized behavior pattern library based on user historical behavior data, dynamically evaluating and calculating rewards for device control policies using reinforcement learning algorithms, combining time series features in environmental data to generate a prediction model that matches user preferences, establishing weight functions for energy consumption, comfort, and device wear, dynamically adjusting weights according to user preferences, using genetic algorithms or particle swarm optimization algorithms to solve the Pareto optimal solution set, locally fine-tuning the policy according to real-time environmental data, and recording device usage frequency, time period preferences, temperature / brightness preferences, and device interlock triggers.

[0108] In one embodiment, the system composition includes:

[0109] Sensor module:

[0110] Temperature sensor: Real-time monitoring of indoor and outdoor temperatures.

[0111] Humidity sensor: Monitoring of indoor humidity.

[0112] Light sensor: Detects the indoor light intensity.

[0113] Human body sensor: Detects whether the user is in the room.

[0114] Actuating device:

[0115] Air conditioner: Supports on / off, temperature setting, and wind speed adjustment.

[0116] Smart curtain: Supports opening and closing control.

[0117] Smart light: Supports brightness adjustment.

[0118] Control center:

[0119] Edge computing device (such as a smart gateway): Runs a reinforcement learning model to generate control policies.

[0120] Cloud server: Stores historical data and supports model training and parameter updates.

[0121] User terminal:

[0122] Mobile APP: Displays the current environmental status, device operation status, and policy execution results, and supports manual adjustment.

[0123] System functions include:

[0124] Environmental perception: Real-time collection of temperature, humidity, light, and user location data.

[0125] Policy generation: Generates energy-saving, comfortable, and safe control policies based on a reinforcement learning model.

[0126] Dynamic adjustment: Adaptively optimizes policy parameters according to real-time data and user feedback.

[0127] User interaction: Provides a visual interface and supports users to view policy details and manual override.

[0128] Security guarantee: Prevents policies from causing device overload or failure through action constraints and anomaly detection.

[0129] The present invention relates to an intelligent home generation strategy optimization system based on adaptive learning. By collecting the operation history, sensor data, and device usage patterns of users in the intelligent home system, evaluation criteria and optimization directions are determined; features are extracted based on user behavior, environmental changes, and device status to construct a feature matrix; a reinforcement learning method is selected according to the optimization problem for modeling, and an adaptive learning mechanism is designed to enable the model to continuously adjust according to real-time data during operation; historical data is used to train the model, and the reward mechanism in reinforcement learning is utilized to guide the achievement of the optimization goal. Based on model prediction, a control strategy suitable for the current environment and requirements is generated, which can better meet the user's usage requirements for intelligent home, improve the intelligent level of the system and the user experience. It includes a data acquisition module for real-time obtaining the status data of intelligent home devices, user behavior data, and environmental data; an adaptive learning module connected to the data acquisition module for training a machine learning model according to user behavior data and environmental data to generate an initial policy set; a policy generation module for dynamically generating a device control policy based on the initial policy set and real-time data; a policy optimization module for adjusting the control policy through a multi-objective optimization algorithm, and the optimization goals include energy consumption efficiency, user comfort, and device lifespan; a feedback execution module for executing the optimized policy and collecting user feedback data, and inputting the feedback data into the adaptive learning module for iterative model update, which has the effect of generating a relatively accurate optimization policy.

[0130] The present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned intelligent home generation strategy optimization method based on adaptive learning are implemented.

[0131] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned intelligent home generation strategy optimization method based on adaptive learning are implemented.

[0132] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided in this application and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0133] It should be noted that in this document, the terms "including", "comprising", or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that includes a series of elements includes not only those elements but also other elements that are not explicitly listed, or elements that are inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, apparatus, article, or method that includes the element.

[0134] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific embodiments. Obviously, according to the content of this specification, many modifications and variations can be made. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A smart home generation strategy optimization method based on adaptive learning, characterized in that: include: Collect user operation history, sensor data, and device usage patterns in smart home systems to determine evaluation criteria and optimization directions; Extract features based on user behavior, environmental changes, and device status, and construct a feature matrix; Select reinforcement learning methods to model the optimization problem and design an adaptive learning mechanism so that the model can be continuously adjusted according to real-time data during operation; Use historical data to train the model, use the reward mechanism in reinforcement learning to guide the realization of optimization goals, and generate control strategies that adapt to the current environment and needs based on model predictions.

2. According to claim 1, a smart home generation strategy optimization method based on adaptive learning is characterized in that: The collection of user information includes recording the user's direct operation behavior in the smart home system, counting the frequency of device use, time period distribution, and user preference settings, collecting indoor and outdoor environmental parameters in real time through environmental sensors, monitoring device status through built-in sensors in the device, extracting the linkage relationship between devices, identifying user habits, removing sensor noise from the collected information data, filling in missing data, standardizing multi-source data formats, extracting time series features, building user behavior portraits, and quantifying the correlation between the environment and device status.

3. According to claim 2, a smart home generation strategy optimization method based on adaptive learning is characterized in that: The feature matrix is ​​constructed by recording the user behavior features, including: Statistical features, average number of operations per day, peak hours, device usage time; Sequential patterns, operation sequences, state transition probabilities modeled through Markov chains; Preference labels, users are divided into categories based on clustering; Time series features, sliding window statistics, periodic features; Event triggering, mutation detection; equipment status characteristics, equipment start and stop times, continuous working time, energy efficiency, equipment association diagram and equipment dependency.

4. The method for optimizing smart home generation strategy based on adaptive learning according to claim 3, characterized in that: The adaptive learning includes defining a reinforcement learning framework, including a state space, an action space, and a composite reward function; constructing a state vector based on real-time environmental data, device status, and user behavior; selecting a reinforcement learning algorithm to model a policy network; designing an adaptive learning mechanism to dynamically adjust the reward function weight, exploration rate, and learning rate; updating policy network parameters through online learning and experience playback; and introducing a safety constraint mechanism to ensure the stability and compliance of device operation. The reward compound function is calculated as follows: R=α·Renergy+β·Rcomfort-γ·Rwear, Among them, α, β, γ are weight parameters, R energy For energy saving rewards, R comfort For comfort bonus, R wear Penalty for equipment wear and tear; The reward strategy update formula is as follows: Among them, E π is the policy network parameter, G t For cumulative returns, T For time.

5. The method for optimizing smart home generation strategy based on adaptive learning according to claim 4, characterized in that: The use of historical data to train the model includes counting the power consumption of the equipment per unit time, establishing a baseline energy consumption model, comprehensively scoring through sensor data and user manual adjustment frequency, analyzing the correlation between the continuous working time of the equipment, the number of start-stop times and the loss, dynamically adjusting the indicator weights according to the user's historical operation data, supporting user-defined scoring rules, converting the evaluation criteria into an objective function, adding constraints, using reinforcement learning to balance multi-objective conflicts through a reward function, using an evolutionary algorithm to generate a Pareto optimal solution set for automatic selection by the user or the system, triggering strategy re-optimization according to real-time data, updating optimization model parameters through a feedback loop, using historical data playback to verify the effectiveness of the optimization strategy, simulating extreme scenarios, collecting user satisfaction scores for automatic strategies, monitoring the frequency of user manual intervention, judging the matching degree between strategies and user needs, retraining the adaptive learning module according to feedback data, and regularly adjusting the evaluation criteria weights to adapt to seasonal or user habit changes.

6. The method for optimizing smart home generation strategy based on adaptive learning according to claim 5, characterized in that: The strategy optimization identifies user preference types based on device usage patterns, defines quantitative indicators of energy consumption, comfort and device life through a multi-objective optimization algorithm, dynamically adjusts the weight distribution of evaluation indicators based on user feedback, constructs a set of constraint conditions for the device control strategy, including the device power upper limit and the user activity time window, uses a reinforcement learning algorithm to generate a candidate strategy set, and screens the Pareto optimal solution through simulation tests.

7. A smart home generation strategy optimization system based on adaptive learning, characterized in that: include: Data collection module, used to obtain status data, user behavior data and environmental data of smart home devices in real time; An adaptive learning module, connected to the data acquisition module, for training a machine learning model based on user behavior data and environmental data to generate an initial strategy set; A strategy generation module, used for dynamically generating device control strategies based on the initial strategy set and real-time data; A strategy optimization module adjusts the control strategy through a multi-objective optimization algorithm, wherein the optimization objectives include energy efficiency, user comfort and equipment life; The feedback execution module is used to execute the optimized strategy and collect user feedback data, and input the feedback data into the adaptive learning module to iteratively update the model.

8. The smart home generation strategy optimization system based on adaptive learning according to claim 7, characterized in that: The adaptive learning includes establishing a personalized behavior pattern library based on user historical behavior data, using a reinforcement learning algorithm to dynamically evaluate and calculate rewards for device control strategies, combining time series features in environmental data to generate a prediction model that matches user preferences, establishing weight functions for energy consumption, comfort, and equipment loss, dynamically adjusting weights based on user preferences, using a genetic algorithm or a particle swarm optimization algorithm to solve the Pareto optimal solution set, locally fine-tuning strategies based on real-time environmental data, and setting device usage frequency, time period preferences, temperature / brightness preferences, and triggering records for linkages between devices.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the smart home generation strategy optimization method based on adaptive learning as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a smart home generation strategy optimization method based on adaptive learning as described in any one of claims 1 to 6 are implemented.

Images