Intelligent home control method and system based on AI voice interaction

Abstract

The invention discloses a smart home control method based on AI voice interaction, and the method comprises the steps: collecting environment data through a sensor, constructing a state vector, updating a value function through a Bellman equation through a reinforcement learning algorithm based on a deep Q network, and carrying out the autonomous learning of an optimal control strategy according to a reward function. And establishing a fault diagnosis model fusing an expert system and a naive Bayesian algorithm, and realizing rapid positioning and automatic recovery of an anomaly. And the AES and RSA mixed encryption technology is adopted to guarantee the data transmission and storage security. And an equipment energy consumption model is constructed, and energy consumption optimization scheduling is realized by combining real-time electricity price and comfort constraints and applying a dynamic programming algorithm. The system comprises a voice data analysis module, a data acquisition and preprocessing module, an intelligent decision and execution module and an abnormity monitoring and recovery module. According to the invention, the self-adaptive control precision, the operation reliability, the data security and the energy utilization efficiency of the smart home system are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of smart home control technology, specifically to a smart home control method and system based on AI voice interaction. Background Technology

[0002] In the current field of smart home control, some systems employ simple voice recognition technology, converting user voice commands into text and then controlling devices through a preset command set. The user speaks a specific command, the speaker recognizes the command, and sends an on signal to the connected smart light fixture. This method achieves voice interaction control to a certain extent, but it has many limitations.

[0003] Simple speech recognition technology has limited ability to understand voice commands and is easily affected by factors such as accents and environmental noise, leading to recognition errors or an inability to accurately understand user intentions, resulting in control failures or incorrect operations. For example, in noisy environments, smart devices may fail to correctly recognize user commands or misunderstand commands with similar pronunciations.

[0004] Obtaining only basic equipment status information is insufficient to provide adequate data support for optimized system control. Without the ability to intelligently adjust based on actual equipment operating conditions and energy consumption, it's difficult to achieve efficient energy utilization and precise equipment maintenance. For example, because it's impossible to monitor real-time energy consumption changes in smart home appliances, it may be impossible to adjust equipment operation appropriately during peak electricity consumption periods, resulting in energy waste.

[0005] Weakly encrypted or unencrypted data transmission and storage methods pose serious security threats to smart home systems. Users' voice commands, device control records, and personal configuration information can be stolen or tampered with, violating user privacy and potentially leading to home security risks, such as malicious control of door locks, cameras, and other devices.

[0006] The lack of an effective fault diagnosis and recovery mechanism means that equipment malfunctions cannot be resolved automatically and in a timely manner. This may lead to further damage to the equipment due to prolonged downtime, increasing maintenance costs and system maintenance difficulty. Summary of the Invention

[0007] The purpose of this invention is to provide a smart home control method and system based on AI voice interaction, in order to solve the problems of insufficient self-learning and optimization capabilities, weak ability to cope with abnormal situations, data security and privacy protection, and energy consumption management.

[0008] To achieve the above objectives, the technical solution provided by this invention is: a smart home control method based on AI voice interaction, comprising the following steps: S1. Receive user voice data and parse it into executable device control voice commands using voice recognition technology; S2. Collect smart home environment data in real time through multiple sensors, and combine the collected smart home environment data from each sensor into a state vector after preprocessing and normalization. S3. Introduce a value function and update the value function based on the Bellman equation and the state vector; use a deep Q-network algorithm to learn the value function, estimate the Q value of the value function through experience replay and target network mechanism, and optimize the smart home control strategy based on device control voice command according to the Q value of the value function. S4. Monitor the execution status of device control voice commands in real time. When a device is not responding, an abnormal status code is detected, or network communication is interrupted, trigger the fault diagnosis process. Use the expert system to make a preliminary judgment based on a predefined rule base, and combine machine learning algorithms to calculate the probability of occurrence of the fault type based on historical fault data. According to the diagnosis type, start the corresponding recovery strategy in the smart home control strategy.

[0009] To optimize the above technical solution, the specific measures also include: In step S1, after receiving user voice data, during the transmission of voice data and control commands, a symmetric encryption algorithm is used to encrypt the data, and an asymmetric encryption algorithm is used to encrypt the symmetric encryption key during transmission. When storing data, encrypted storage technology is used to protect local and cloud data. The symmetric encryption algorithm uses the Advanced Encryption Standard (AES), while the asymmetric encryption algorithm uses the RSA algorithm. The public key is used to encrypt the symmetric encryption key, and the private key is used for decryption.

[0010] In step S3, the process of introducing the value function and updating the value function based on the state vector and the Bellman equation is expressed as follows:

[0011] in, It is in state The control actions taken below It is the immediate reward obtained after performing the action. For learning rate, This is the discount factor.

[0012] In step S3, the predefined reward function includes: The reward function related to satisfaction rewards is:

[0013] in, The user satisfaction reward coefficient Indicates satisfaction rating. This represents the baseline satisfaction value; The reward function related to the rationality of equipment energy consumption is:

[0014] in, This represents the energy consumption reward coefficient. This indicates the rating of the equipment's energy consumption rationality. The reward function related to system stability rewards is:

[0015] in, The system stability reward coefficient and system stability score are used to calculate the system stability score. .

[0016] In step S3, the smart home control strategy based on device control voice commands is optimized according to the Q value of the value function. The specific process is as follows: A deep Q-network is used to calculate the Q-value for all possible control actions in the current state, and the action with the largest Q-value is selected as the optimal control action. The training of the deep Q-network is based on the instantaneous reward calculated by a predefined reward function and updated through the Bellman equation, wherein the reward function comprehensively considers user satisfaction, reasonable device energy consumption and system stability. In step S4, the probability of occurrence of a fault type is calculated based on historical fault data using a machine learning algorithm. The specific formula is as follows:

[0017] in, This represents the probability of occurrence of each fault type given fault characteristics; Fault type The prior probability; In the fault type Under the conditions that occur, various characteristics related to the fault appear. The probability of; It is the probability of feature combinations occurring; Based on the diagnosis type, the corresponding recovery strategy is initiated, which specifically includes: device reset, pushing fault information, switching to a backup network, or conducting network fault troubleshooting.

[0018] Based on historical energy consumption data, an energy consumption model is established for each smart home device. The specific process is as follows: The energy consumption of temperature-sensitive smart furniture is estimated using the following multiple linear regression model:

[0019] in, The coefficients are to be determined. This is the error term; Indoor temperature; Outdoor temperature; Set the temperature for the user; This refers to the equipment's uptime.

[0020] The formula for estimating the energy consumption of smart office furniture is:

[0021] in, Power during operation. This refers to standby power. The total working hours in a day This is the cumulative standby time.

[0022] An energy-saving control strategy is generated using an optimization algorithm. The specific formula is as follows:

[0023] in, For time intervals; The power of the smart furniture at time t; Real-time electricity price information; and Each represents a specific point in time.

[0024] As another important technical solution, the present invention also provides a smart home control system based on AI voice interaction, comprising: The voice data parsing module is used to receive user voice data and parse it into executable device control voice commands through voice recognition technology; The data acquisition and preprocessing module is used to collect smart home environment data in real time through multiple sensors, and to combine the smart home environment data collected by each sensor into a state vector after preprocessing and normalization. The intelligent decision-making and execution module is used to introduce a value function and update the value function based on the Bellman equation and the state vector; it uses a deep Q-network algorithm to learn the value function, estimates the Q value of the value function through experience replay and target network mechanism, and optimizes the smart home control strategy based on device control voice command according to the Q value of the value function. The anomaly monitoring and recovery module is used to monitor the execution status of device control voice commands in real time. When a device is not responding, an abnormal status code is detected, or network communication is interrupted, the fault diagnosis process is triggered. The expert system makes a preliminary judgment based on a predefined rule base and combines machine learning algorithms to calculate the probability of the occurrence of the fault type based on historical fault data. According to the diagnosis type, the corresponding recovery strategy is launched in the smart home control strategy.

[0025] The present invention also proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a smart home control method based on AI voice interaction as described above.

[0026] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute a smart home control method based on AI voice interaction as described above.

[0027] Compared with the prior art, the beneficial effects of the present invention are: This invention utilizes a reinforcement learning framework based on Deep Q-Networks (DQN) to continuously learn from user interactions and environmental feedback. It not only accurately executes voice commands but also understands the user's implicit intentions and dynamically adapts to different user habits and changing environmental conditions. Through value functions and reward mechanisms, it drives continuous optimization of control strategies, thereby greatly improving the accuracy and efficiency of control and providing users with a highly personalized intelligent experience.

[0028] This invention constructs an integrated proactive defense and self-healing system that combines monitoring, diagnosis, and recovery. Through the close integration of expert systems and machine learning algorithms, it enables rapid location and automated recovery from anomalies such as device unresponsiveness and network interruptions, significantly reducing system downtime and manual intervention, and ensuring the continuity and seamlessness of smart home services.

[0029] This invention employs a hybrid encryption system combining symmetric and asymmetric encryption, and implements encryption and strict access control on stored data, constructing a multi-layered security defense covering the entire process of data transmission and storage, effectively resisting external attacks and the risk of internal data leakage.

[0030] This invention establishes a precise equipment energy consumption model and combines it with real-time electricity prices, using intelligent optimization algorithms to achieve refined scheduling and energy-saving control of equipment. This not only directly reduces household energy consumption and electricity costs, but also improves energy utilization efficiency. Attached Figure Description

[0031] Figure 1 : A flowchart of the smart home control system in this embodiment of the invention. Detailed Implementation The present invention will be further described in detail below through specific embodiments, but it should not be construed as limiting the scope of the subject matter of the present invention to the following embodiments. All technologies implemented based on the above content of the present invention fall within the scope of the present invention.

[0032] The following is an explanation of some of the terms used in this invention: Speech recognition technology, also known as Automatic Speech Recognition (ASR) or Speech-to-Text Recognition (STT), is a technology that enables computers to automatically convert human speech into corresponding text. Its core purpose is to allow machines to understand and respond to human voice commands or dialogues, thereby achieving more natural human-computer interaction.

[0033] Deep Q-Networks (DQNs) are algorithms that combine deep learning and reinforcement learning, primarily used to solve problems with high-dimensional state spaces. DQNs are an upgrade to Q-Learning by using deep neural networks to approximate the Q-value function in traditional Q-Learning, thus enabling the processing of complex inputs such as images and videos.

[0034] Symmetric encryption algorithms are algorithms that use the same key for both encryption and decryption. They are characterized by fast encryption and decryption speeds, but their security depends on the protection of the key.

[0035] Asymmetric encryption algorithms are cryptographic algorithms that use different keys for encryption and decryption. They are characterized by high security and slow encryption speed, and have important applications in fields such as digital signatures and digital certificates.

[0036] Naive Bayes is a classification method based on Bayes' theorem and the assumption of conditional independence of features. It has a solid mathematical foundation, a simple algorithm, and a wide range of applications.

[0037] The Bellman equation, also known as the dynamic programming equation, was discovered by Richard Bellman. It is a core concept in dynamic programming and reinforcement learning, used to describe the recursive relationship of the value function in optimal decision problems.

[0038] This invention provides a smart home control method based on AI voice interaction, comprising the following steps: S1. Receive user voice data and parse it into executable device control voice commands through voice recognition technology.

[0039] In step S1, after receiving user voice data, during the transmission of voice data and control commands, a symmetric encryption algorithm is used to encrypt the data, and an asymmetric encryption algorithm is used to encrypt the symmetric encryption key during transmission. When storing data, encrypted storage technology is used to protect local and cloud data. In some implementations, symmetric encryption algorithms (such as AES) rapidly encrypt data, transforming the original data into ciphertext for transmission, effectively preventing data theft or tampering during transmission. Simultaneously, to ensure the secure distribution and management of the symmetric encryption key, an asymmetric encryption algorithm (such as RSA) is used to encrypt the symmetric encryption key during transmission. The receiver uses its private key to decrypt and obtain the symmetric encryption key, then decrypts and restores the transmitted data. This hybrid encryption method ensures the confidentiality, integrity, and authenticity of data transmission.

[0040] Preferably, both local storage devices and cloud servers employ encrypted storage technology to encrypt and store user voice command history, device control logs, and user configuration information, preventing data leakage risks. Furthermore, a strict identity authentication and access control mechanism is established, ensuring that only authorized users can access and operate specific smart home data resources, further strengthening data security.

[0041] The jointly constructed encryption system effectively resists various security threats that may be faced in the smart home environment. Whether it is malicious attacks from external networks or potential internal privacy leaks, it can effectively protect users' voice commands, device control information, and other private data from being leaked or tampered with. This not only safeguards users' personal privacy but also maintains the overall stability and reliability of the smart home system, working in synergy with fault diagnosis and recovery mechanisms and reinforcement learning algorithms.

[0042] S2. Collect smart home environment data in real time through multiple sensors, including: indoor temperature and humidity data collected by temperature and humidity sensors, light intensity data collected by light sensors, and device operating mode, energy consumption data and running time parameters collected by device status sensors; combine the smart home environment data collected by each sensor into a state vector after preprocessing and normalization.

[0043] In some implementations, taking a temperature and humidity sensor as an example, it collects indoor temperature and humidity data at fixed time intervals, denoted as... and ,in Indicating timestamps, this data reflects the comfort level of the indoor environment and is crucial for control decisions of devices such as air conditioners and humidifiers. Light intensity values ​​acquired by the light sensor... This affects the operational status of devices such as smart curtains and smart lights. Device status sensors record parameters such as the operating mode, energy consumption level, and runtime for each smart home device. For example, the washing progress of a smart washing machine can be represented as... The value ranges from 0 (not started) to 1 (washing completed), along with its energy consumption rate. .

[0044] Preferably, a state-space description of the system is constructed based on this abundant sensor data. Assume the system has n sensors of different types, and each sensor collects data with a data dimension of... Then the state of the system at time t It can be represented as a vector:

[0045] in, It represents the value of a certain dimension of data collected by a specific sensor after preprocessing and normalization.

[0046] S3. To quantify the impact of different control actions under different states on system performance, a value function is introduced and updated based on the Bellman equation and the state vector. A deep Q-network algorithm is used to learn the value function, and the Q value of the value function is estimated through experience replay and target network mechanism. Based on the Q value of the value function, the smart home control strategy based on device control voice commands is optimized.

[0047] Introducing a value function The expression is:

[0048] in, It is in state The control actions taken below It is the immediate reward obtained after performing the action. For learning rate, This is the discount factor.

[0049] Preferably, by continuously updating the state space description based on newly collected data and using this value function to evaluate system performance under different control strategies, the smart home system can gradually learn the optimal control strategy under various complex environmental conditions and device operating states, achieving precise device control. For example, based on factors such as indoor temperature, occupant activity, and device energy consumption, it can intelligently adjust the temperature, fan speed, and operating mode of the air conditioner. At the same time, it can work in conjunction with other smart home devices to improve the energy efficiency and user comfort of the entire home system. This, together with encryption technology to ensure data security and fault diagnosis and recovery mechanisms to ensure stable system operation, provides users with an efficient, intelligent, safe, and comfortable smart home experience, enhancing the overall efficiency and adaptability of the system to cope with various dynamically changing situations and diverse user needs in the smart home environment.

[0050] In some implementations, a reward function is designed that takes into account user satisfaction, reasonable device energy consumption, and system stability. When a user successfully controls the device to achieve the expected effect through voice commands, the system receives a positive reward; if there is excessive device energy consumption, frequent malfunctions, or delays in the execution of control commands, negative reward feedback is given.

[0051] When designing this reward function, multiple key factors need to be considered to accurately quantify the system's performance under different operating conditions.

[0052] User satisfaction is measured by analyzing user feedback after executing voice commands on the device. For example, if a user gives the voice command "Turn the living room lights to 50% brightness," and the lights quickly and accurately reach the specified brightness and remain stable for a period of time without the user issuing another adjustment command, then the operation can be considered to have met the user's needs, and a positive satisfaction score can be given. Assuming that a baseline satisfaction value is determined through statistical analysis of user historical operation data and feedback information. ,when Give positive rewards at that time. Its value can be calculated based on the degree to which it exceeds the benchmark, such as... ,in The user satisfaction reward coefficient aims to incentivize the system to respond to user commands more accurately, and this coefficient can be dynamically adjusted based on actual conditions and user preferences.

[0053] To ensure the rationality of equipment energy consumption, the energy consumption of each device is monitored in real time and compared with its historical energy consumption data and the average energy consumption standard of similar devices. Taking smart air conditioners as an example, over a period of time... The actual energy consumption is The theoretical energy consumption range calculated based on factors such as ambient temperature, set temperature, and operating time is: ,like If so, a certain energy consumption rationality score will be given. Positive rewards will be given when equipment energy consumption is within a reasonable range. ,For example , This is an energy consumption incentive coefficient used to encourage the system to optimize equipment operation and reduce unnecessary energy consumption. This aligns with the goal of smart home systems to achieve high efficiency and energy saving, while also taking into account overall energy management and cost control. It is interconnected with the previously mentioned energy consumption monitoring and optimization model to jointly ensure the sustainable operation of smart home systems.

[0054] For system stability, the focus is on the number of equipment failures. and the average execution delay time of control instructions If, within a certain time period, the equipment does not malfunction and the instruction execution delay is within an acceptable range (e.g., a maximum delay threshold is set), then... If the system stability score is given, then a stability score will be assigned. When the system is running stably, a positive reward will be given. ,for example , The system stability reward coefficient motivates the system to continuously optimize its control strategies and fault diagnosis and recovery mechanisms, reducing the probability of failures and instruction execution delays. This ensures that the smart home system can provide continuous and stable services, consistent with the overall reliability and stability requirements of the smart home system. It also prevents equipment failures or delays, improves the overall system performance and user satisfaction, and enables the system to maintain an efficient, stable, energy-saving, and user-friendly operating state in complex and ever-changing smart home environments. Guided by the reward function, the control strategy is continuously optimized and improved to adapt to the diverse needs of different users and changes in various environmental conditions.

[0055] Preferably, a Deep Q-Network (DQN) or its improved algorithm is used as the basis for reinforcement learning. Its powerful function approximation ability is utilized to estimate the Q-values ​​of various control actions under different states, i.e., the expected long-term rewards. During operation, the system continuously explores and tries different control strategies, updates the parameters of the Q-value network based on the obtained reward feedback, selects the action with the largest Q-value as the optimal control action, and gradually learns the optimal control strategy in different situations. This achieves precise, efficient, and adaptive control of smart home devices, improves system performance, and enables the system to continuously evolve and optimize control decisions as user habits and environmental changes occur.

[0056] S4. Monitor the execution status of device control voice commands in real time. When a device does not respond, an abnormal status code is detected, or network communication is interrupted, trigger the fault diagnosis process. Use the expert system to make a preliminary judgment based on a predefined rule base, and combine machine learning algorithms to calculate the probability of the occurrence of the fault type based on historical fault data. According to the diagnosis type, start the corresponding recovery strategy in the smart home control strategy. In some implementations, the expert system stores a rich knowledge rule base covering various potential fault scenarios for numerous smart home devices. For example, if a smart light bulb flickers, expert experience suggests it might be due to unstable power supply voltage or the bulb nearing the end of its lifespan. These rules are presented in the form of "IF-THEN," meaning "IF bulb flickers, THEN check power supply voltage or consider replacing the bulb." In this way, the expert system can quickly make preliminary judgments about common faults based on pre-set rules, providing a reliable starting point for fault diagnosis, reducing the scope and time of troubleshooting, and ensuring that the downtime of the smart home system is minimized. This is closely linked to the overall goal of system stability and reliability, and also corresponds to the consideration of system stability in the reward function. A stable system receives positive rewards, thereby incentivizing the system to continuously optimize its fault diagnosis mechanism.

[0057] Preferably, machine learning algorithms utilize historical fault data for in-depth analysis. Taking the Naive Bayes algorithm as an example, it is assumed that multiple fault types exist. (For example, a smart door lock might not open due to a dead battery, mechanical failure, or damaged electronic components, etc.), as well as various characteristics related to the malfunction. (Such as low battery alarm messages, abnormal unlocking sounds, error code displays, etc.). Through statistical learning of a large amount of historical fault data, the probability of occurrence of each fault type under given fault characteristics is calculated, i.e. Its calculation formula is based on Bayes' theorem:

[0058] in, Fault type The prior probability can be obtained by statistically analyzing the frequency of various faults in historical data; In the fault type Under the conditions that occur, characteristics appear The probability is also learned from historical data; This represents the probability of feature combinations occurring, used for normalization. When the system detects the current fault feature set, this algorithm calculates the posterior probability of each fault type, and the fault type with the highest probability value is the most likely cause of the fault.

[0059] In this way, machine learning algorithms can uncover fault patterns and regularities hidden behind complex data, compensating for the potential limitations of expert systems. For rare or complex fault situations, they can also make relatively accurate diagnoses, further improving the accuracy and comprehensiveness of fault diagnosis. This not only helps to quickly locate problems and take effective recovery measures to restore smart home devices to normal operation as soon as possible, but also provides data support for continuous system optimization, constantly improving fault diagnosis models, enhancing the system's ability to cope with various anomalies, and working synergistically with encryption technology to ensure data security and reinforcement learning algorithms to optimize control strategies, together building an efficient, intelligent, stable, and secure smart home environment, improving the overall competitiveness and practicality of the system.

[0060] In some implementations, appropriate recovery strategies are quickly initiated based on the diagnostic results. If the problem is a hardware failure, the system attempts to remotely reset the device; if the reset fails, detailed fault information and possible solutions are promptly sent to the user, such as contacting the device manufacturer's customer support. For network failures, the system automatically switches to a backup network connection and attempts to troubleshoot and repair the original network, such as checking network settings and restarting network devices, to ensure that voice commands can be smoothly transmitted to the target device, maintaining the stable operation of the smart home system and minimizing service interruptions due to abnormal situations.

[0061] As a preferred approach, an energy consumption model is established for each smart home device based on historical energy consumption data; and an energy-saving control strategy is generated by combining the energy consumption model with real-time electricity price information using an optimization algorithm.

[0062] The energy consumption of temperature-sensitive smart furniture is estimated using the following multiple linear regression model:

[0063] in, The coefficients to be determined were obtained through regression analysis of a large amount of historical energy consumption data; This is the error term; Indoor temperature; Outdoor temperature; Set the temperature for the user; This refers to the device's operating time. The model can estimate the energy consumption of temperature-sensitive smart furniture based on current indoor and outdoor temperatures, set temperatures, and estimated operating time.

[0064] For smart office furniture, its energy consumption at different times of the day is calculated based on its usage duration distribution and standby power consumption characteristics. The energy consumption estimation formula is:

[0065] in, Power during operation. This refers to standby power. The total working hours in a day This is the cumulative standby time.

[0066] Based on this data and models, intelligent optimization algorithms are used to formulate energy-saving strategies. For example, dynamic programming algorithms are used to optimize the operation mode of temperature-controlled smart furniture, with the goal of minimizing the total daily energy cost. The day is divided into multiple time intervals, and within each interval, the cooling and heating power and operating time are determined based on real-time electricity prices, indoor and outdoor temperatures, and user comfort requirements.

[0067] Assuming a certain time interval Inside, the optimal power setting for temperature-controlled smart furniture is obtained by solving a dynamic programming algorithm. and runtime This ensures that user comfort constraints (such as maintaining indoor temperature within a comfortable range) are met. Under the premise of ), the energy cost is the lowest, that is:

[0068]

[0069] For smart office furniture, if the idle time of the smart furniture exceeds a set threshold (such as 30 minutes) and the current electricity price is during peak hours, a shutdown command will be automatically sent; if it is during off-peak hours, a standby command will be sent to reduce energy consumption.

[0070] In another embodiment of the present invention, a smart home control system based on AI voice interaction is proposed, comprising: The voice data parsing module is used to receive user voice data and parse it into executable device control voice commands through voice recognition technology; The data acquisition and preprocessing module is used to collect smart home environment data in real time through multiple sensors, and to combine the smart home environment data collected by each sensor into a state vector after preprocessing and normalization. The intelligent decision-making and execution module is used to introduce a value function and update the value function based on the Bellman equation and the state vector; it uses a deep Q-network algorithm to learn the value function, estimates the Q value of the value function through experience replay and target network mechanism, and optimizes the smart home control strategy based on device control voice command according to the Q value of the value function. The anomaly monitoring and recovery module is used to monitor the execution status of device control voice commands in real time. When a device is not responding, an abnormal status code is detected, or network communication is interrupted, the fault diagnosis process is triggered. The expert system makes a preliminary judgment based on a predefined rule base and combines machine learning algorithms to calculate the probability of the occurrence of the fault type based on historical fault data. According to the diagnosis type, the corresponding recovery strategy is launched in the smart home control strategy.

[0071] In another embodiment of the present invention, an electronic device is proposed, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a smart home control method based on AI voice interaction as described above.

[0072] In another embodiment of the present invention, a computer-readable storage medium is provided, storing a computer program that causes a computer to execute a smart home control method based on AI voice interaction as described above.

[0073] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent substitutions, and improvements made by those skilled in the art to the above embodiments without departing from the scope of the technical solution of the present invention, based on the technical essence of the present invention, shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A smart home control method based on AI voice interaction, characterized in that, Includes the following steps: S1. Receive user voice data and parse it into executable device control voice commands using voice recognition technology; S2. Collect smart home environment data in real time through multiple sensors, and combine the collected smart home environment data from each sensor into a state vector after preprocessing and normalization. S3. Introduce a value function and update the value function based on the Bellman equation and the state vector; use a deep Q-network algorithm to learn the value function, estimate the Q value of the value function through experience replay and target network mechanism, and optimize the smart home control strategy based on device control voice command according to the Q value of the value function. S4. Monitor the execution status of device control voice commands in real time. When a device is not responding, an abnormal status code is detected, or network communication is interrupted, trigger the fault diagnosis process. Use the expert system to make a preliminary judgment based on a predefined rule base, and combine machine learning algorithms to calculate the probability of occurrence of the fault type based on historical fault data. According to the diagnosis type, start the corresponding recovery strategy in the smart home control strategy.

2. The smart home control method based on AI voice interaction according to claim 1, characterized in that: In step S1, after receiving user voice data, during the transmission of voice data and control commands, a symmetric encryption algorithm is used to encrypt the data, and an asymmetric encryption algorithm is used to encrypt the symmetric encryption key during transmission. When storing data, encrypted storage technology is used to protect local and cloud data. The symmetric encryption algorithm uses the Advanced Encryption Standard (AES), while the asymmetric encryption algorithm uses the RSA algorithm. The public key is used to encrypt the symmetric encryption key, and the private key is used for decryption.

3. The smart home control method based on AI voice interaction according to claim 1, characterized in that: In step S3, the introduction of the value function The process of updating the value function based on the state vector and the Bellman equation is expressed as follows: in, It is in state The control actions taken below It is the immediate reward obtained after performing the action. For learning rate, This is the discount factor.

4. The smart home control method based on AI voice interaction according to claim 1, characterized in that: In step S3, the predefined reward function includes: The reward function related to satisfaction rewards is: in, The user satisfaction reward coefficient Indicates satisfaction rating. This represents the baseline satisfaction value; The reward function related to the rationality of equipment energy consumption is: in, This represents the energy consumption reward coefficient. This indicates the rating of the equipment's energy consumption rationality. The reward function related to system stability rewards is: in, The system stability reward coefficient and system stability score are used to calculate the system stability score. .

5. The smart home control method based on AI voice interaction according to claim 4, characterized in that: In step S3, the process of optimizing the smart home control strategy based on device control voice commands according to the Q value of the value function is as follows: A deep Q-network is used to calculate the Q-value for all possible control actions in the current state, and the action with the largest Q-value is selected as the optimal control action. The training of the deep Q-network is based on the instantaneous reward calculated by a predefined reward function and updated through the Bellman equation, wherein the reward function comprehensively considers user satisfaction, reasonable device energy consumption and system stability.

6. The smart home control method based on AI voice interaction according to claim 1, characterized in that: In step S4, the probability of occurrence of a fault type is calculated based on historical fault data using a machine learning algorithm. The specific formula is as follows: in, This represents the probability of occurrence of each fault type given fault characteristics; It is a fault type The prior probability; In the fault type Under the conditions that occur, various characteristics related to the fault appear. The probability of; It is the probability of feature combinations occurring; Based on the diagnosis type, the corresponding recovery strategy is initiated, which specifically includes: device reset, pushing fault information, switching to a backup network, or conducting network fault troubleshooting.

7. The smart home control method based on AI voice interaction according to claim 1, characterized in that: This also includes establishing energy consumption models for various smart home devices based on historical energy consumption data. The specific process is as follows: The energy consumption of temperature-sensitive smart furniture is estimated using the following multiple linear regression model: in, The coefficients are to be determined. This is the error term; Indoor temperature; Outdoor temperature; Set the temperature for the user; For equipment uptime; The formula for estimating the energy consumption of smart office furniture is: in, Power during operation. This refers to standby power. The total working hours in a day This refers to the cumulative standby time. Combining the energy consumption model and real-time electricity price information, an optimization algorithm is used to generate an energy-saving control strategy for smart home control. The specific formula is as follows: in, For time intervals; The power of the smart furniture at time t; Real-time electricity price information; and Each represents a specific point in time.

8. A smart home control system based on AI voice interaction, characterized in that, include: The voice data parsing module is used to receive user voice data and parse it into executable device control voice commands through voice recognition technology; The data acquisition and preprocessing module is used to collect smart home environment data in real time through multiple sensors, and to combine the smart home environment data collected by each sensor into a state vector after preprocessing and normalization. The intelligent decision-making and execution module is used to introduce a value function and update the value function based on the Bellman equation and the state vector; it uses a deep Q-network algorithm to learn the value function, estimates the Q value of the value function through experience replay and target network mechanism, and optimizes the smart home control strategy based on device control voice command according to the Q value of the value function. The anomaly monitoring and recovery module is used to monitor the execution status of device control voice commands in real time. When a device is not responding, an abnormal status code is detected, or network communication is interrupted, the fault diagnosis process is triggered. The expert system makes a preliminary judgment based on a predefined rule base and combines machine learning algorithms to calculate the probability of the occurrence of the fault type based on historical fault data. According to the diagnosis type, the corresponding recovery strategy is launched in the smart home control strategy.

9. An electronic device, characterized in that, include: The invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements a smart home control method based on AI voice interaction as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program causes the computer to execute a smart home control method based on AI voice interaction as described in any one of claims 1 to 7.

Images