Intelligent household electrical appliance interaction control method and system

Abstract

The invention discloses an intelligent household electrical appliance interaction control method and system, and the method comprises the following steps: collecting the operation state parameters, environment perception data and user behavior characteristics of each household electrical appliance in a living space in real time, and generating an environment perception data packet through a multi-modal data fusion model; constructing a user habit preference model library, generating a household appliance control strategy decision tree in combination with historical interaction records, and matching an optimal control strategy set according to a real-time scene type; and when a user active intervention signal or an environment sudden change event is detected, a dynamic strategy reconstruction mechanism is triggered, an instruction updating request carrying a priority identifier is broadcasted to the associated household appliance group, and a redistribution control instruction set is generated. The method has the following advantages and effects: multi-source environment data can be deeply fused, the user habit model is constructed in real time, and the method has a dynamic strategy reconstruction capability so as to solve the problems of response lag, strategy stiffness and energy efficiency imbalance in a complex home environment in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of household appliance interaction, and in particular to an intelligent household appliance interaction control method and system. BACKGROUND

[0002] With the rapid development of Internet of Things technology, smart home systems have gradually become popular, and various types of electrical equipment in the home can be interconnected through the network and achieve remote control and automatic management. Existing smart home control systems mostly use timing control, scene mode switching or simple response mechanisms based on single sensor feedback, such as automatically starting and stopping air conditioners according to environmental temperature, adjusting light brightness according to light intensity, etc. These methods have improved the convenience and energy efficiency of home life to some extent, but still have obvious limitations. The most core problem is that the existing system lacks comprehensive perception and dynamic decision-making ability for multi-dimensional state of the living environment, especially when facing user-initiated intervention, environmental sudden changes or multi-device coordination needs, the system cannot quickly generate an optimized control strategy that takes into account comfort, energy efficiency and safety. Specifically, the system often relies on predefined static rules and is difficult to adapt to the randomness of user behavior, the volatility of environmental parameters and the time-varying nature of device state in the home environment; when multiple events occur concurrently, the system lacks effective priority arbitration and resource allocation mechanisms, which can easily lead to command conflicts, response delays or energy consumption surges, and cannot achieve truly intelligent adaptive control. Therefore, there is an urgent need for an intelligent home appliance interaction control method that can deeply integrate multi-source environmental data, real-time construct user habit models, and have dynamic strategy reconstruction capability, to solve the response lag, strategy rigidity and energy imbalance problems faced by existing technologies in complex home environments. SUMMARY

[0003] The purpose of the present application is to provide an intelligent household appliance interaction control method and system to solve the problems raised in the background art.

[0004] The above technical purpose of the present application is achieved by the following technical solution: The present application provides an intelligent household appliance interaction control method, comprising the following steps: S100, real-time collection of running state parameters, environmental perception data and user behavior characteristics of each electrical equipment in the living space, generation of an environmental perception data package through a multi-modal data fusion model for comprehensive quantification of the comfort index of the home environment and the energy consumption characteristics of the equipment; S200, constructing a user habit preference model library based on the environment perception data packet, generating a household appliance control strategy decision tree in combination with historical interaction records, and matching an optimal control strategy set according to a real-time scene type; wherein the household appliance control strategy decision tree comprises a time period dependent strategy, an event triggered strategy and an energy saving optimization strategy branch, and the optimal control strategy set is used to initialize a collaborative control instruction sequence of the household appliance equipment; S300, when a user active intervention signal or an environment mutation event is detected, triggering a dynamic strategy reconstruction mechanism, broadcasting an instruction update request carrying a priority identifier to an associated household appliance group, and generating a reallocation control instruction set according to a device response delay threshold and an energy consumption constraint condition; wherein the active intervention signal includes a voice control instruction, a gesture recognition instruction and an emergency button trigger signal, and the environment mutation event includes a sudden change in temperature and humidity, an abnormal power fluctuation and a safety risk alarm.

[0005] By adopting the above technical solution, the environment perception data packet integrates the device state, the environment parameter and the user behavior into a unified index system, which not only accurately reflects the comprehensive comfort of the home space, but also quantifies the energy consumption characteristics of each device, so that the system has the dual ability of simultaneously optimizing the user experience and the energy efficiency; the user habit preference model library constructed by the machine learning technology analyzes the internal correlation between the historical interaction records and the real-time data, dynamically generates a multi-branch strategy decision tree, so that the control strategy has scene adaptability; the time period dependent strategy can automatically adjust the device working mode based on the living habits, the event triggered strategy can quickly respond to unexpected events, and the energy saving optimization strategy continuously taps the energy efficiency potential of the collaborative operation of the devices; the three-dimensional architecture of the three strategy branches ensures full coverage from regular scenes to unexpected scenes; the matching mechanism of the optimal control strategy set relies on the scene type identification model to improve the strategy switching efficiency, and the collaborative control instruction sequence generated by the matching mechanism significantly improves the accuracy and response speed of multi-device linkage; more importantly, the dynamic strategy reconstruction mechanism solves the rigidity problem of the smart home system in the case of user intervention or environment mutation, and through the broadcast mechanism of the priority identifier and the instruction reallocation under the constraint condition, the dynamic adjustment of the strategy is realized on the premise of ensuring the stability of the system, so that the system has the flexibility of automation and manual intervention; especially in the case of sudden changes in temperature and humidity and other environmental mutations or safety alarms and other emergency situations, the instruction reallocation can be performed based on the device response delay threshold and the energy consumption constraint condition, effectively avoiding the safety risks or energy waste caused by strategy lag.

[0006] Further, the S100 specifically comprises the following steps: The running state parameters of each household appliance equipment in the living space are collected in real time by the embedded power metering module, and the running state parameters include an instantaneous power consumption value, a current working mode code and a remaining life prediction value calculated based on the equipment running time; A distributed environmental sensor network is deployed to synchronously collect multi-dimensional environmental perception data; wherein a three-dimensional temperature distribution matrix is generated by a temperature sensor array, an indoor humidity gradient value data is constructed by a humidity sensor group, and dynamic light intensity spatio-temporal distribution data is generated by a light sensor matrix; User behavior characteristics are captured by a millimeter wave radar monitoring system and a non-contact infrared sensing system; wherein human heart rate data is analyzed by biological reflection signals of the millimeter wave radar, and body surface temperature distribution data is obtained by the non-contact infrared sensing system, and a user activity trajectory heat map is generated based on a multi-target trajectory tracking algorithm; A multi-modal data fusion model is used for fusion operation; wherein the running state parameters, environmental perception data and user behavior characteristics are time-stamped and aligned and mapped to spatial coordinates; a comfort index set of the home environment is generated by using a feature weighted fusion algorithm, wherein the comfort index set includes a thermal comfort sub-index, a visual comfort sub-index and a behavior adaptation sub-index; and a device energy consumption feature vector is generated based on running state parameter analysis, which includes a peak energy consumption identifier, a steady-state energy consumption curve and an energy efficiency deviation coefficient; Based on the fusion result, an environmental perception data packet is generated, which includes a time-stamped comfort index set, a device energy consumption feature vector and an original data check code.

[0007] By using the above technical solutions, the deployment of the distributed environmental sensor network generates a three-dimensional temperature distribution by a temperature sensor array, a humidity gradient model is constructed by a humidity sensor, and dynamic light spatio-temporal distribution is captured by a light sensor, so that environmental monitoring breaks through the limitations of traditional single-point measurement and realizes dynamic monitoring; millimeter wave radar and non-contact infrared sensing include user physiological indicators and behavior trajectories into the monitoring system, heart rate data and body surface temperature distribution provide biological feature basis for environmental comfort evaluation, and the activity trajectory heat map based on the multi-target tracking algorithm accurately captures the behavior rules; the multi-modal data fusion model solves the problem of inconsistent time and space reference of multi-source data through time stamp alignment and spatial coordinate mapping, and the feature weighted fusion algorithm generates three sub-indices of thermal comfort, visual comfort and behavior adaptation, which scientifically quantifies the living experience from three aspects of human physiological adaptability, environmental optical suitability and behavior interaction convenience; the construction of the device energy consumption feature vector combines instantaneous power and steady-state curve characteristics, especially the introduction of the energy efficiency deviation coefficient, which provides a quantitative basis for energy efficiency anomaly diagnosis; and the finally generated environmental perception data packet ensures data integrity and traceability through time stamp and check code, and establishes a data foundation with breadth and depth for upper-level decision-making.

[0008] Further, the generation of the thermal comfort sub-index, the visual comfort sub-index and the behavior adaptation sub-index specifically includes the following steps: Based on the three-dimensional space temperature distribution matrix, the body surface temperature distribution data and the human heart rate data, a thermal comfort sub-index is generated through a heat balance model, and a humidity influence correction is made in combination with the humidity gradient value data; based on the dynamic light intensity spatio-temporal distribution data, a visual comfort sub-index is generated through a light adaptation model, and a spatial position correlation analysis is made by integrating the user activity trajectory heat map; based on the user activity trajectory heat map, a behavior adaptation degree sub-index is generated through a behavior pattern matching algorithm, and a device interaction adaptation degree evaluation is made in combination with the current working mode code in the operating state parameter; the thermal comfort sub-index, the visual comfort sub-index and the behavior adaptation degree sub-index are normalized.

[0009] By adopting the above technical solution, the visual comfort sub-index relies on the light adaptation model to analyze the spatio-temporal distribution of light data, and corrects the position correlation in combination with the actual activity trajectory of the user, solving the defect that the uniform light index cannot reflect the real visual experience, such as automatically improving the uniformity of illumination in the reading area; the behavior adaptation degree sub-index couples and analyzes the activity trajectory and the device working mode through the behavior pattern matching algorithm, quantitatively evaluates the degree of fit between the device state and the user behavior, such as predicting the demand for adjusting the air supply direction of the air conditioner according to the user movement trajectory; the normalization processing of the three sub-indices constructs a unified comfort evaluation scale, making the comfort experiences of different dimensions comparable and having the ability of weighted calculation, providing a basis for multi-objective optimization control.

[0010] Further, the generation of the device energy consumption feature vector specifically includes the steps of: The power consumption value sequence of each household appliance in a continuous preset period is extracted, a steady-state energy consumption curve is generated through a steady-state feature extraction algorithm, and a peak energy consumption identifier exceeding a preset threshold is labeled; based on the current working mode code in the operating state parameter, the devices are grouped, and the deviation of the instantaneous power consumption value of the device from the rated power or the historical average power under the same working mode is calculated to generate an energy efficiency deviation coefficient; based on the remaining life prediction value, the energy efficiency warning threshold for determining whether the energy efficiency deviation coefficient is abnormal is dynamically adjusted; the peak energy consumption identifier, the steady-state energy consumption curve and the energy efficiency deviation coefficient are integrated to form a device energy consumption feature vector.

[0011] By adopting the technical scheme, the labeling mechanism of the peak energy consumption identifier provides early warning basis for overload protection, and the dynamic adjustment capability of the preset threshold can adapt to the operation characteristics of different devices; the grouping energy consumption analysis technology based on the working mode, in combination with the calculation model of the energy efficiency deviation degree coefficient, realizes real-time comparison of the actual energy consumption of the device and the theoretical benchmark value, breaks through the limitation of traditional absolute power consumption monitoring, and can directly locate the abnormally high energy consumption device; the remaining life prediction value is innovatively associated with the energy efficiency alarm threshold, for example, the deviation tolerance of the aging device is relaxed, the real energy efficiency degradation phenomenon is accurately identified while avoiding false positives; and finally, the device energy consumption feature vector forms a three-dimensional energy efficiency portrait of "basic energy consumption-peak feature-health correlation", which provides fine data support for energy saving strategy making.

[0012] Further, the S200 specifically includes the following steps: Based on the comfort index set in the environment perception data packet, the device energy consumption feature vector and the historical interaction record, a user habit preference model library is constructed through a clustering algorithm; wherein, the historical interaction record includes user active intervention signal execution frequency, device use time period distribution data and energy saving preference score, the clustering algorithm adopts a K-means optimization model to generate a user habit preference clustering cluster, each clustering cluster corresponds to a habit preference type, and a habit preference weight value; In combination with the habit preference clustering cluster and the habit preference weight value in the user habit preference model library, a home appliance control strategy decision tree is generated; wherein, the home appliance control strategy decision tree includes a time period dependent strategy branch, an event triggered strategy branch and an energy saving optimization strategy branch, each branch is divided based on a decision tree splitting algorithm; Based on the real-time scene type, an optimal control strategy set in the home appliance control strategy decision tree is matched; wherein, the real-time scene type includes a daily mode, an energy saving mode and an emergency mode, the matching operation adopts a scene similarity calculation model, calculates the Euclidean distance between the current environment perception data packet and each scene type, and selects the optimal control strategy set of the scene type with the smallest distance to activate the corresponding strategy branch.

[0013] By adopting the technical scheme, in the process of constructing the household appliance control strategy decision tree, the time period dependent strategy branch is divided according to the clustering of the equipment use time period, and the optimal operation mode combination is recommended combining the comfort requirement and the energy consumption characteristic, for example, the energy saving mode is automatically started in the power consumption peak time period; the event triggered strategy branch predicts the intervention signal occurrence law through the probability model, and realizes the second level response combining the event action rule library. The multi-objective optimization algorithm of the energy saving optimization strategy branch excavates the energy efficiency potential under the premise of ensuring the comfort critical value, such as adjusting the linkage parameters of the air conditioner set temperature and the fresh air system; the scene similarity calculation model matches the real-time scene type based on the Euclidean distance, so that the strategy switching has scene adaptability, the daily mode focuses on comfort, the energy saving mode strengthens energy management, and the emergency mode prioritizes safety, which solves the pain point that the fixed strategy cannot adapt to the variable scene.

[0014] Further, after the step of matching the optimal control strategy set in the household appliance control strategy decision tree based on the real-time scene type, the method further comprises the steps of: initializing a cooperative control instruction sequence of the household appliance equipment; wherein the cooperative control instruction sequence is generated based on the optimal control strategy set, the instruction sequence contains the equipment identifier, the control action parameter and the time sequence scheduling information, and the initialization operation includes instruction priority sorting and resource conflict detection, which ensures that the instruction sequence is compatible with the peak energy consumption identifier and the steady state energy consumption curve in the equipment energy consumption characteristic vector.

[0015] By adopting the technical scheme, the cooperative control instruction sequence converted from the optimal control strategy set of the multi-device instruction cooperative execution mechanism is optimized, the strategy executability is ensured through the accurate binding of the equipment identifier and the control action parameter, the time coupling problem of multi-device action is solved through the embedding of the time sequence scheduling information, such as avoiding the tripping caused by the simultaneous start of multiple high-power devices; the instruction priority sorting mechanism ensures that the key operation is executed preferentially, and the resource conflict detection function actively avoids the conflict risk in the execution process by predicting the compatibility of the instruction and the real-time energy consumption curve, such as automatically delaying the start request of the unnecessary equipment when it is detected that the current circuit load has approached the safety threshold.

[0016] Further, the household appliance control strategy decision tree is generated combining the habit preference clustering cluster and the habit preference weight value in the user habit preference model library, and specifically comprising the steps of: According to the equipment use time period distribution data in the user habit preference model library, and combining the comfort index set and the equipment energy consumption characteristic vector in the environment perception data packet, a plurality of time period intervals are divided through a time period clustering algorithm, and a recommended operation mode set of the household appliance equipment in each time period interval is generated, so as to construct a time period dependent strategy branch; wherein the time period dependent strategy branch comprises a plurality of time period strategy nodes, each time period strategy node corresponds to a time period interval and a recommended operation mode set of the time period interval. An event-triggered probability model is established based on the frequency of user active intervention signals in the user habit preference model library, and combined with the type of active intervention signals and the type of environmental mutation events collected in real time, an event-triggered strategy branch is constructed through an event-action association rule library; wherein the event-triggered strategy branch includes multiple event strategy nodes, each event strategy node corresponds to an intervention event or environmental event type and the response action sequence of the household appliance when the event is triggered; The energy-saving preference score in the user habit preference model library and the energy efficiency deviation coefficient in the device energy consumption feature vector are integrated, and a energy-saving optimization strategy branch is generated through a multi-objective optimization algorithm with energy efficiency optimization as the target; wherein the energy-saving optimization strategy branch includes multiple energy efficiency strategy nodes, each energy efficiency strategy node corresponds to an energy efficiency optimization target and an energy efficiency adjustment instruction set that the household appliance needs to execute to achieve the target.

[0017] By adopting the above technical solutions, the branch construction logic of the strategy decision tree is deepened. The time period dependent strategy branch is divided into time intervals with similar behavior characteristics by a time period clustering algorithm, and a set of recommended operation modes is customized for each interval, such as the air conditioner preheating and lighting gradual brightening strategy automatically executed in the morning of weekdays for energy saving. The event-triggered strategy branch predicts the triggering probability of different events according to the probability model, and matches the response action sequence in the association rule library. In the energy-saving optimization strategy branch, each energy efficiency strategy node corresponds to a specific optimization target and its adjustment instruction set, such as setting a temperature difference dead zone to reduce the frequent start-stop of the air conditioner compressor, so that energy-saving control is upgraded from a rough on-off to precise parameter adjustment. The three-branch structure covers three types of scenes: periodic rules, random events, and continuous optimization, forming a complete control strategy map.

[0018] Further settings are that the S300 specifically includes the following steps: Real-time monitoring of active intervention signals and environmental mutation events; wherein the active intervention signals include voice control instructions, gesture recognition instructions and emergency button trigger signals, and the environmental mutation events include sudden changes in temperature and humidity, abnormal power consumption fluctuations and safety risk alarms; when any signal or event is detected, a dynamic strategy reconstruction mechanism is activated; the dynamic strategy reconstruction mechanism performs the following operations: Broadcasting an instruction update request carrying a priority identifier to an associated group of household appliances; and Generating a reallocation control instruction set according to the device response delay threshold and the energy consumption constraint condition; Otherwise, maintaining the current cooperative control instruction sequence.

[0019] By adopting the technical scheme, the emergency response mechanism covers natural interaction modes such as voice gestures, the monitoring range of the environmental mutation event contains environmental abnormalities such as sudden changes in temperature and humidity and risk signals such as safety alarms, the system has all-weather abnormal sensing capability; the trigger condition of the dynamic strategy reconstruction mechanism ensures stable operation of the system while maintaining sensitive response to key events; the broadcast mechanism sends an update request with priority to the associated device group, which avoids resource waste caused by full system refresh and ensures that critical devices receive instructions first; the redistribution control instruction set is generated by comprehensively considering the device response delay threshold and the energy consumption constraint condition, for example, delaying the response speed of low-priority devices in the critical state of circuit load, effectively preventing system collapse risk.

[0020] Further, the dynamic strategy reconstruction mechanism specifically includes the steps of: broadcasting an instruction update request carrying a priority identifier to the associated group of home appliances; wherein the priority identifier is generated based on the event type: the safety risk alarm and the emergency button trigger signal are assigned the highest priority identifier, the sudden change in temperature and humidity and the abnormal power consumption fluctuation are assigned the intermediate priority identifier, and the regular voice control instruction and the gesture recognition instruction are assigned the basic priority identifier; generating a redistribution control instruction set according to the device response delay threshold and the energy consumption constraint condition, including: obtaining real-time response delay parameters and device energy consumption feature vectors of each home appliance device in the associated group of home appliances; wherein the device response delay threshold is pre-set according to the device type, and the device energy consumption feature vector contains a peak energy consumption identifier, a steady-state energy consumption curve and an energy efficiency deviation coefficient; establishing a redistribution optimization model for multi-objective optimization calculation with the priority identifier as the main constraint, the device response delay threshold as the secondary constraint, and the energy consumption constraint condition as the boundary condition; wherein the energy consumption constraint condition integrates the steady-state energy consumption curve and the peak energy consumption identifier to limit the total power consumption after redistribution to be less than the pre-set safety threshold; based on the optimization calculation result, generating a redistribution control instruction set containing device identifiers, redistribution control action parameters and timing scheduling information; execute the dynamic strategy reconstruction mechanism: freeze the instructions in the current cooperative control instruction sequence that conflict with the redistribution control instruction set, and insert the redistribution control instruction set in descending order of priority identifier to generate a reconstructed cooperative control instruction sequence; If the compatibility between the reconstructed instruction sequence and energy consumption constraints is verified by a resource conflict detection algorithm, the reconstructed collaborative control instruction sequence will be sent to the associated home appliance group for execution. If the verification fails, a degradation strategy will be initiated: based on the priority identifier, the sub-instruction set with the highest priority identifier will be selected from the redistribution control instruction set; the compatibility between this sub-instruction set and the energy consumption constraints will be verified again. If the verification passes, the sub-instruction set will be sent for execution, and a system exception log will be generated. If the verification of this sub-instruction set still fails, the current collaborative control instruction sequence will remain unchanged, and the highest level system alarm signal will be triggered.

[0021] By adopting the above technical solutions, the redistribution optimization model performs multi-objective optimization with priority as the primary constraint, response delay as the secondary constraint, and energy consumption conditions as the boundary. The generation of the reconstructed instruction set includes three information: device identification, execution parameters, and time-series scheduling, ensuring the integrity and executability of the instructions. The instruction freezing and priority interpolation mechanism in the dynamic reconstruction operation solves the conflict problem during the transition between old and new strategies. The two-level verification process of the resource conflict detection algorithm automatically downgrades the execution of a subset of instructions when energy consumption exceeds the limit, and the triggering of the highest-level alarm signal provides the final guarantee for manual intervention. A three-level fault protection system of "conflict prevention - local degradation - emergency alarm" is constructed, enabling the system to maintain its core functions even under extreme anomalies.

[0022] The present invention also provides an intelligent home appliance interactive control system, comprising the following modules: The data acquisition module is configured to collect real-time operating status parameters of various household appliances in the living space, environmental perception data, and user behavior characteristics. The data processing module is configured to generate environmental perception data packets through a multimodal data fusion model, which are used to comprehensively quantify the comfort index of the home environment and the energy consumption characteristics of the equipment. The user habit modeling module is configured to build a user habit preference model library based on the environmental perception data package, and generate a home appliance control strategy decision tree by combining historical interaction records; The strategy matching module is configured to match the optimal set of control strategies based on the real-time scenario type and initialize the collaborative control instruction sequence of home appliances; The dynamic reconfiguration module is configured to trigger a dynamic policy reconfiguration mechanism and generate a set of reallocation control instructions when a user-initiated intervention signal or a sudden environmental event is detected. The communication control module is configured to broadcast control commands to associated home appliance groups and receive device status feedback. The data storage module is used to store environment-aware data packets, user habit and preference model library, control strategy decision tree, and historical interaction records.

[0023] In summary, the present application has the following beneficial effects: it can deeply integrate multi-source environmental data, construct a user habit model in real time, and has dynamic strategy reconstruction capability, to solve the problems of response lag, strategy rigidity and energy efficiency imbalance in the prior art under complex home environment. BRIEF DESCRIPTION OF DRAWINGS

[0024] Fig. 1 a schematic diagram of the main flow of the embodiment; Fig. 2 a schematic diagram of S100 in the embodiment; Fig. 3 a schematic diagram of S200 in the embodiment; Fig. 4 a schematic diagram of S300 in the embodiment; Fig. 5 a system block diagram of the embodiment. DETAILED DESCRIPTION

[0025] The present application will be further described in detail below in combination with the drawings.

[0026] As shown in the accompanying Figs. 1 to 5 ; The embodiment discloses an intelligent household appliance interaction control method, comprising the following steps: S100, real-time collection of running state parameters, environmental perception data and user behavior characteristics of each household appliance in the living space, generation of an environmental perception data package through a multi-modal data fusion model, for comprehensive quantification of the comfort index and device energy consumption characteristics of the home environment. Specifically, it comprises the following steps: The system collects the running state parameters of each electrical equipment in the residential space in real time through the embedded power metering module, including the instantaneous power consumption value, the current working mode code and the remaining life prediction value calculated based on the equipment running time. These parameters reflect the current energy consumption status, working mode and predicted remaining service life of the equipment, providing basic data for subsequent energy efficiency evaluation and strategy formulation. At the same time, a distributed environmental sensor network is deployed to synchronously collect multi-dimensional environmental perception data, including a three-dimensional space temperature distribution matrix generated by a temperature sensor array, indoor humidity gradient value data constructed by a humidity sensor group, and dynamic light intensity spatio-temporal distribution data generated by a light sensor matrix. These environmental data are used to evaluate the thermal environment, humidity distribution and lighting conditions of the residential space, providing input for the calculation of the comfort index. In addition, user behavior characteristics are captured using a millimeter wave radar monitoring system and a non-contact infrared sensing system, human heart rate data are analyzed through the biological reflection signal of the millimeter wave radar, body surface temperature distribution data are obtained through the non-contact infrared sensing system, and user activity trajectory heat maps are generated based on a multi-target trajectory tracking algorithm. These behavior characteristic data are used to analyze the activity mode, physiological state and position distribution of the user in the space, providing the basis for behavior adaptation degree evaluation. Based on the multi-modal data fusion model, the system timestamps aligns and spatial coordinate maps the running state parameters, environmental perception data and user behavior characteristics, generates a comfort index set of the home environment using a feature weighted fusion algorithm, including thermal comfort sub-index, visual comfort sub-index and behavior adaptation degree sub-index. At the same time, based on the analysis of the running state parameters, a device energy consumption feature vector is generated, including peak energy consumption identification, steady-state energy consumption curve and energy efficiency deviation coefficient. Finally, the system generates an environmental perception data package containing the timestamped comfort index set, device energy consumption feature vector and original data check code, which is used for subsequent control strategy formulation and optimization.

[0027] In this embodiment, the embedded power metering module is a hardware unit integrated within each household appliance, responsible for periodically collecting electrical parameters of the device, including instantaneous power consumption value, current working mode code, and remaining life prediction value calculated based on the cumulative running time of the device and the life model provided by the manufacturer. The instantaneous power consumption value reflects the actual energy consumption level of the device at a certain time, which is a key indicator for evaluating the energy efficiency of the device; the current working mode code is used to identify the working state of the device (such as refrigeration, heating, standby, etc.), providing context information for subsequent strategy matching; the remaining life prediction value is estimated based on the historical running data of the device and the preset decay model, used to warn potential device failure and optimize maintenance strategies. The distributed environmental sensor network is composed of temperature sensor arrays, humidity sensor groups, and illumination sensor matrices deployed at key locations in the living space. The temperature sensor array is arranged in a three-dimensional grid form, generating a three-dimensional matrix data reflecting the temperature distribution of the space through synchronous sampling; the humidity sensor group is deployed by region, constructing a scalar field representing the indoor humidity gradient; the illumination sensor matrix dynamically captures the intensity and spatio-temporal distribution of natural light and artificial light sources, generating a light intensity spatio-temporal distribution dataset. These environmental perception data provide objective physical environment representation for comfort assessment. The capture of user behavior features relies on non-intrusive sensing technology: the millimeter wave radar monitoring system transmits and receives millimeter wave signals to analyze human reflection signals and extract heart rate data; the non-contact infrared sensing system generates a body surface temperature distribution map by detecting human body surface infrared radiation; combined with multi-target trajectory tracking algorithms (such as target tracking models based on Kalman filtering or deep learning), the system further generates user activity trajectory heat maps reflecting user movement patterns and activity hot zones. These behavior feature data not only reveal the real-time state of the user, but also provide the basis for personalized comfort adjustment. In the multi-modal data fusion stage, the system first aligns the timestamps and maps the spatial coordinates of data from different sources to ensure consistency in the time and space dimensions. Then, a feature weighting fusion algorithm is used to integrate the processed data: for the generation of the comfort index set, the system calculates the thermal comfort sub-index (based on the three-dimensional space temperature distribution matrix, body surface temperature distribution data, and heart rate data, combined with the heat balance model and humidity correction), visual comfort sub-index (based on the spatial correlation analysis of dynamic light intensity spatio-temporal distribution data and user activity trajectory heat map), and behavior adaptation sub-index (based on the behavior pattern matching of user activity trajectory heat map and device working mode). After normalization, each sub-index is combined into a unified comfort index set.The construction of the device energy consumption feature vector is achieved through in-depth analysis of the operating state parameters: the power consumption sequence in the continuous time window is extracted, and the steady-state energy consumption curve is identified and the peak energy consumption event is labeled by applying a steady-state feature extraction algorithm (such as moving average or change point detection); in combination with the current working mode code, the deviation of the actual power from the rated power in the same mode is calculated to generate the energy efficiency deviation coefficient; the remaining life prediction value is used to dynamically adjust the energy efficiency alarm threshold for determining whether the energy efficiency deviation coefficient is abnormal, thereby realizing life-aware energy efficiency monitoring. Finally, the system encapsulates the timestamped comfort index set, the device energy consumption feature vector, and the original data check code for data integrity verification into an environment-aware data packet, providing standardized and multi-dimensional environment representation inputs for the upper layer decision-making.

[0028] In the present embodiment, the generation of the thermal comfort sub-index is based on the PMV-PPD thermal balance model framework, and a three-dimensional space temperature distribution matrix is introduced as the environmental temperature input, the body surface temperature distribution data is introduced as the individual thermal state representation, and the human heart rate data is introduced as the auxiliary index for metabolic rate estimation. The model first calculates the operating temperature of the micro-environment where the user is located, and modifies the evaporative heat dissipation efficiency in combination with the humidity gradient value data to quantify the influence of environmental thermal stress on comfort. The output result is a scalar value within the range of 0 to 1, the closer to 1 , the higher the thermal comfort. The visual comfort sub-index is calculated by a light adaptation model, which considers the illuminance level, color temperature distribution, and glare index provided by the dynamic light intensity spatio-temporal distribution data, and identifies the visual demand of the user's resident area in combination with the user activity trajectory heat map. By calculating the deviation of the current lighting conditions from the ideal visual environment (such as the recommended illuminance value based on CIE standards), and weighting the user position weight, the index value reflecting the visual comfort degree is generated. Specifically, the visual comfort sub-index ; wherein represents the number of light sensors or virtual grid points in the area where the user's activities are frequent; represents the index; represents the measured illuminance value at position in the dynamic light intensity spatio-temporal distribution data; represents the ideal illuminance value recommended for the activity type (such as reading, resting) at position based on CIE standards, etc., and the activity type is determined by the user's main behavior mode inferred from the user activity trajectory heat map; represents the weight of position , which is proportional to the "heat" value of the area in the user activity trajectory heat map. The longer the user stays in an area, the greater the influence of the visual comfort of that area on the total index; the denominator of the formula is used for normalization, so that the output value remains near . The final result can also be strictly constrained to 0 to 1 by a Sigmoid function The behavior adaptation sub-index depends on a behavior pattern matching algorithm that matches the real-time user activity trajectory heat map with a preset behavior pattern library (such as "rest", "activity", "work", etc.) to identify the current user behavior pattern. Specifically, the input real-time user activity trajectory heat map is calculated with the similarity (such as cosine similarity, Jaccard index, or direct calculation of the proportion of stay time in a specific functional area such as sofa, desk) of each template heat map in the preset behavior pattern library , , , and the output is the most likely current behavior pattern. At the same time, combined with the current working mode code of the device, the compatibility of the device running state and the user behavior is evaluated (such as whether the air conditioner is in silent mode when the user is resting), and the behavior adaptation score is output. Specifically, the input of all home appliances current working mode code , , , …, is then queried to the preset "behavior pattern-ideal state of device" rule library, and the output is: the proportion of the number of devices that meet the rules to the total number of devices, or the weighted compatibility score according to the importance of the device (range ), and the behavior adaptation sub-index is , the closer the value is to , the more the running state of the home appliance conforms to the user's current behavior habit. Finally, the three sub-indices are respectively processed by Min-Max normalization to eliminate the dimensional difference, and an interface is reserved to allow weight adjustment according to user's individualized preferences to form a set of comprehensive comfort index.

[0029] In this embodiment, the construction of the device energy consumption feature vector is a multi-step analysis process: the system first extracts the instantaneous power consumption value sequence of each device in the recent complete cycle (such as hours) (the power consumption value sequence is composed of continuously collected instantaneous power consumption values in chronological order) applies a steady-state feature extraction algorithm based on sliding window and change point detection to identify the period when the device is in stable running state, and fits the average power consumption of the period to form a steady-state energy consumption curve. At the same time, the power point of short-time surge in the sequence is detected, and if its value exceeds a certain proportion of the rated power of the device (such as a peak energy consumption flag is marked. The energy efficiency deviation coefficient is calculated based on the device grouping: the system divides the devices into different groups according to the current working mode code (e.g., air conditioner refrigeration group, washing machine washing group, etc.), and calculates the relative deviation of the instantaneous power consumption value of the devices in the same group from the rated power or the historical average power in the group. The coefficient reflects the abnormal degree of energy efficiency of the device in the current mode, and a positive value indicates that the energy consumption is high, and a negative value indicates that the energy consumption is low. Specifically, the energy efficiency deviation coefficient ; wherein, represents the steady-state power measured in the current mode; represents the expected power in the mode, which can be the rated power or the average power in the group based on historical data. The remaining life prediction value is introduced as an adaptive adjustment factor: the system presets a baseline energy efficiency warning threshold, but as the remaining life prediction value of the device decreases (indicating that the device is aging), the threshold is gradually relaxed to avoid excessive warning of old devices. Specifically, ; wherein represents the adjusted energy efficiency warning threshold, represents the baseline energy efficiency warning threshold; is the preset maximum relaxation amplitude; represents the normalized remaining life prediction value. The lower the device remaining life prediction value, the more relaxed the corresponding energy efficiency warning threshold; finally, the system encapsulates the peak energy consumption flag (Boolean type or level identifier), steady-state energy consumption curve (time series data or fitting parameters) and energy efficiency deviation coefficient (scalar value) into a structured device energy consumption feature vector, which not only describes the real-time energy consumption state of the device, but also integrates the life-aware energy efficiency evaluation logic, providing a quantitative basis for subsequent energy-saving strategy formulation.

[0030] S200, based on the environment perception data packet, a user habit preference model library is constructed, a household appliance control strategy decision tree is generated combined with historical interaction records, and an optimal control strategy set is matched according to the real-time scene type; wherein, the household appliance control strategy decision tree contains time period dependent strategy, event triggered strategy and energy saving optimization strategy branch, and the optimal control strategy set is used to initialize the cooperative control instruction sequence of the household appliance device. Specifically, it includes the following steps: The system first constructs a user habit preference model library based on the time-stamped comfort index set (including thermal comfort sub-index, visual comfort sub-index and behavior adaptation sub-index) contained in the environment perception data packet, the device energy consumption feature vector (including peak energy consumption identifier, steady-state energy consumption curve and energy efficiency deviation coefficient) and the historical interaction record (including user active intervention signal execution frequency, device usage period distribution data and energy saving preference score) through the improved K-means clustering algorithm. In this process, the historical interaction record provides a quantitative representation of the user's long-term behavior pattern, such as the user's operation frequency on a specific device, the adjustment record of environmental parameters such as temperature, humidity and light, and the energy saving tendency score. The clustering algorithm takes these multi-dimensional data as input, measures the similarity between samples by calculating the Euclidean distance or cosine similarity, and determines the optimal number of clusters according to the elbow rule or silhouette coefficient , and finally generates a user habit preference clustering cluster. Each clustering cluster represents a typical user behavior pattern, such as "morning active energy-saving type", "night comfort priority type" or "home office balanced type", etc., and assigns a habit preference weight value to each clustering cluster. The weight value is dynamically updated according to the number of historical data samples contained in the cluster and the matching degree with the current user real-time behavior, to reflect the influence degree of different habit patterns on the current strategy formulation.

[0031] On the basis of successfully constructing the user habit preference model library, the system further combines the obtained habit preference clustering cluster and its corresponding habit preference weight value to generate a structured home appliance control strategy decision tree. The decision tree is constructed by using a decision tree splitting algorithm based on information gain or Gini impurity, and contains three main strategy branches: a time period dependent strategy branch, an event triggered strategy branch, and an energy saving optimization strategy branch. The generation of the time period dependent strategy branch is as follows: according to the device use time period distribution data extracted from the user habit preference model library, and by fusing the comfort index set in the environment perception data packet and the time period performance consumption and comfort degree law reflected by the device energy consumption feature vector, a time series clustering algorithm (such as K-means for time series data or clustering based on dynamic time warping) is used to divide a day into multiple time period intervals with significant feature differences (for example, “morning low temperature and low activity time period”, “midday high light period”, “evening high comfort demand time period”, “midnight super low power consumption time period”, etc.). For each divided time period interval, the system comprehensively analyzes the historical optimal comfort index value interval, the typical steady-state energy consumption level of the device, and the user habit preference in the time period, to generate a recommended running mode set of each home appliance device in the time period (for example, in the “evening high comfort demand time period”, the air conditioner is recommended to be set to a comfortable temperature, the light is recommended to be set to a warm color temperature with medium brightness, and the curtain is recommended to be kept open; in the “midnight super low power consumption time period”, the air conditioner is recommended to enter sleep mode, the light is turned off, and all unnecessary devices are recommended to enter standby or off state). Each time period interval and its corresponding recommended running mode set constitute a time period strategy node in the time period dependent strategy branch. The construction of the event triggered strategy branch depends on the user active intervention signal execution frequency in the user habit preference model library, and the system establishes an event triggering probability model to predict the possibility of occurrence of various intervention events (such as voice instructions, gesture control). At the same time, a pre-defined event-action association rule library (which contains rules such as “detecting “leave home” voice instruction triggers the sequence of turning off all unnecessary devices”, “environment sensor reports that the air quality index exceeds the standard triggers the air purifier to enter strong wind mode and turns on the fresh air system”, etc.) is used to define the response action sequence of the home appliance device to be executed when each known intervention event type (voice control instruction, gesture recognition instruction) and environmental mutation event type (sudden change of temperature and humidity, abnormal power consumption fluctuation, safety risk alarm) is triggered. Each event type and its corresponding response action sequence constitute an event strategy node in the event triggered strategy branch. The user active intervention signal execution frequency can be expressed by the following formula: ; wherein represents the user active intervention signal execution frequency, which comprehensively reflects the overall frequency of user active control in unit time, and the higher the value, the more frequently the user manually intervenes; Total number of active intervention signal types, including voice control instructions, gesture recognition instructions, emergency button trigger signals, etc. Signal type index; Class information; Weight coefficient of class information, used to distinguish the importance of different types of signals, for example, the weight of the emergency button trigger signal should be much higher than that of the regular voice control instruction, indicating that the emergency event, although occurring rarely, is extremely important once it occurs. The specific setting value can be pre-set or learned from historical data; Number of times of occurrence of the first class signal within the statistical time window ; Length of the statistical time window, for example, the past 24 hours, the past week, etc., used to normalize the count to frequency. The generation of the energy-saving optimization strategy branch is to integrate the energy-saving preference score (quantifying the user's acceptance of energy saving) in the user habit preference model library and the energy efficiency deviation coefficient (identifying the current energy efficiency state of the device) in the device energy consumption feature vector, to build a multi-objective optimization model (for example, linear weighting, constraint method or evolutionary algorithm, etc.) with comprehensive energy efficiency optimization as the primary goal, while considering comfort constraints. After solving the model, a series of energy efficiency optimization goals (such as "total power consumption minimization", "peak demand reduction", "high energy efficiency deviation device priority load reduction", etc.) and detailed energy efficiency adjustment instruction sets for each device to implement each goal (for example, to achieve the "total power consumption minimization" goal, the instruction set may include appropriately increasing the air conditioner set temperature by 2 degrees Celsius, reducing the light brightness under the premise of meeting visual comfort, delaying the start of high-power washing tasks, etc.). The energy-saving preference score in this embodiment can be represented by the formula: ; wherein Energy-saving preference score, which will eventually be converted into the energy-saving preference weight of the multi-objective optimization function ; User historical average energy consumption, the actual average total power consumption in the user's home within the statistical period; Baseline energy consumption, which can be the average energy consumption of similar families, the estimated energy consumption according to the house area and the number of appliances, or the energy consumption budget set by the user. The ratio reflects the user's relative energy consumption level; User response rate to energy-saving suggestions, when the system recommends energy-saving strategies (such as increasing the air conditioner temperature), the proportion of the number of times the user accepts the suggestion to the total number of recommendations; The number of times that the user actively intervenes into the energy saving mode. For example, the user manually adjusts the air conditioner from low temperature mode to high temperature mode, or actively turns off unnecessary lights; The total number of times that the user actively intervenes; A very small positive number, used to prevent the denominator from being zero. 、 and are preset weight coefficients, used to adjust the importance of the three factors, energy consumption level, system interaction response, and active behavior, in the final score, which can be preset or optimized through machine learning. The multi-objective optimization model in this embodiment can be expressed by the formula: total target function The energy consumption target and the comfort target are weighted and summed to obtain, The optimization target is to minimize the total target function ; wherein represents the energy saving preference weight, represents complete priority to energy saving, represents complete priority to comfort, and the value is directly derived from the energy saving preference score in the user habit preference model library The system maps the user's score to this weight interval. Each energy efficiency optimization target and its corresponding energy efficiency adjustment instruction set constitutes an energy efficiency strategy node in the energy saving optimization type strategy branch.

[0032] After generating the home appliance control strategy decision tree, the system matches the optimal control strategy set based on the real-time scene type. The real-time scene type is divided into three categories: daily mode, energy-saving mode and emergency mode. The matching operation is realized through a scene similarity calculation model: the model extracts the key features (such as the values of the current comfort index set, the instantaneous state of the device energy consumption feature vector) in the environment perception data packet collected and generated in the current period, and compares them with the predefined feature templates of each scene type. Calculate the Euclidean distance between the current environment perception data packet feature vector and each scene template feature vector, and select the scene type with the smallest Euclidean distance as the currently activated scene. For example, if the current calculated comfort index is generally high and the device energy consumption is within the normal range, match the daily mode; if the device energy consumption feature vector shows that the peak energy consumption identifier frequently appears or the total energy efficiency deviation coefficient is high, and the user's energy-saving preference score is also high, match the energy-saving mode; if the environment perception data packet contains a safety risk alarm or an emergency signal in an environmental mutation event, immediately match the emergency mode. After matching the scene type, the system activates the optimal control strategy set of the corresponding strategy branch (the daily mode focuses on the time period-dependent branch, the energy-saving mode focuses on the energy-saving optimization branch, and the emergency mode prioritizes the corresponding high-priority strategy node in the event-triggered branch) in the home appliance control strategy decision tree. The optimal control strategy set is selected from the activated strategy branch according to the currently matched scene type, and the combination of the strategy node that best fits the current environmental state and user habit preference.

[0033] After successfully matching and obtaining the optimal control policy set, the system immediately performs the initialization operation of the household appliance cooperative control instruction sequence. The cooperative control instruction sequence is an ordered instruction list, which is generated based on the strategy content contained in the optimal control policy set. Each instruction contains a clear device identifier (specifying the controlled device), a control action parameter (such as setting temperature, brightness level, switch state, running mode, etc.), and accurate timing scheduling information (specifying the start time, duration or relative trigger time of instruction execution). The initialization process is not simply a list of instructions, but includes key instruction priority sorting logic and resource conflict detection mechanism. Instruction priority sorting is based on the importance of strategy source (such as emergency mode event triggered strategy instruction usually has the highest priority, followed by energy saving optimization type instruction, and finally time period dependent type instruction) and internal constraints of the strategy. Resource conflict detection focuses on whether the execution of the instruction sequence will conflict with the device energy consumption characteristics described in the device energy consumption feature vector, for example, checking whether there are instructions that will cause multiple high-power devices to run simultaneously in the peak energy consumption period, thereby possibly causing circuit overload, or whether there are instructions that require the device running mode to deviate significantly from the best energy efficiency interval represented by its steady-state energy consumption curve. The system adjusts the timing or parameters of the instructions iteratively to ensure that the finally generated cooperative control instruction sequence meets the intention of the optimal control policy set and is compatible with the current energy consumption capability and state of the device, ensuring the feasibility of control and the stability of the system. The initialized cooperative control instruction sequence will be loaded into the execution queue, thereby realizing intelligent and collaborative fine control of household appliances in the home environment.

[0034] S300, when a user active intervention signal or an environmental mutation event is detected, a dynamic policy reconstruction mechanism is triggered, an instruction update request carrying a priority identifier is broadcast to the associated household appliance group, and a reallocation control instruction set is generated according to the device response delay threshold and the energy consumption constraint condition; wherein the active intervention signal includes voice control instruction, gesture recognition instruction and emergency button trigger signal, and the environmental mutation event includes sudden change of temperature and humidity, abnormal power fluctuation and safety risk alarm. Specifically, it includes the following steps: The system continuously and sensitively monitors and analyzes multiple input signal streams through multi-signal acquisition and event listening modules deployed on the central controller and edge devices. These signal streams mainly include three categories: first, user-initiated intervention signals, including voice control instructions obtained by microphone arrays deployed in key areas such as living rooms and bedrooms and analyzed by voice recognition engines, gesture recognition instructions captured by millimeter wave radars or depth cameras and identified by computer vision algorithms, and emergency button trigger signals generated when emergency buttons installed on walls or mobile terminals are triggered; second, environmental mutation events, including temperature and humidity sudden change events triggered when the change amplitude of the values reported by the humidity sensor group within a unit time exceeds the preset safety threshold (e.g., temperature change per minute exceeds Celsius or humidity change per minute exceeds ), abnormal power consumption fluctuation events triggered when the real-time monitoring and judgment of the embedded power metering module shows that the instantaneous power consumption value deviates from the device steady-state energy consumption curve benchmark range learned based on historical data by a certain proportion (e.g., more than ), and safety risk alarms triggered and uploaded by security subsystems (such as smoke sensors, door magnetic sensors, and carbon monoxide alarms). The monitoring process uses multi-level filtering technology and event relevance analysis algorithms to effectively distinguish between environmental noise, device fluctuations, and other incidental disturbances and important events that truly require system response. Once any valid user-initiated intervention signal or environmental mutation event is detected and confirmed by confidence verification, the system immediately triggers the built-in dynamic strategy reconstruction mechanism, interrupting the normal flow process of the currently executing collaborative control instruction sequence.

[0035] After the dynamic policy reconfiguration mechanism is activated, its internal logic unit first carries out accurate classification and priority determination of the event: the system automatically identifies the specific type of the triggering event according to the pre-defined and continuously learning-updated event classification and priority rule library, and gives it the corresponding priority identification. Among them, the signals directly related to life and property safety, such as safety risk alarms indicating the occurrence of fire, gas leakage or illegal intrusion, and user's explicit emergency button triggering signals, are given the highest priority identification; the signals indicating potential risks of rapid deterioration of environmental state or possible device failure, such as the aforementioned sudden change of temperature and humidity and abnormal power consumption fluctuation events, are given the intermediate priority identification; and the regular voice control instructions and gesture recognition instructions reflecting the user's normal comfort or convenience adjustment requirements are given the basic priority identification. Then, the system dynamically determines the associated group of home appliances directly affected by the event (for example, the living room temperature sudden drop event is mainly associated with the air conditioner, heating radiator, fresh air system and smart curtain in the living room; the kitchen detects abnormal high power consumption, mainly associated with the oven, induction cooker, refrigerator and other high-power devices in the kitchen) according to the type of the event and its location (determined by sensor ID or event source positioning), and uses a reliable broadcast protocol with redundant retransmission mechanism to broadcast an instruction update request message carrying the event type code and priority identification to all device control nodes in the group, to ensure that all associated devices can reliably and timely receive and analyze the request in the complex home wireless network environment.

[0036] In the meantime of broadcasting the instruction update request, the system initiates a parallel computing process for generating the redistribution control instruction set. The core of this computing process is a redistribution optimization model based on multi-objective constraint satisfaction problem. The model first acquires the latest state data of each appliance in the associated appliance group through a device state query interface, including the real-time response delay parameter read from the built-in controller or agent program of the device (this parameter represents the theoretical or measured maximum time required for the device to start executing hardware actions after receiving the instruction, which is pre-set according to the device type and model and stored in the device capability database, for example, the response delay of a smart curtain motor is usually hundreds of milliseconds to several seconds, while the switching delay of a smart light bulb can be as low as tens of milliseconds) and the device energy consumption feature vector of the device extracted from the latest period generated environmental perception data packet (this vector contains the peak energy consumption identifier, the steady-state energy consumption curve representing the typical power consumption level of the device in stable operation state, and the energy efficiency deviation coefficient reflecting the deviation degree of the current actual energy consumption from the expected energy efficiency level). The redistribution optimization model takes the priority identifier assigned to the event as the primary optimization objective (ensuring that the instruction demand with the highest priority is given the highest priority and the fastest satisfaction), takes the response delay threshold of each device as the hard time constraint (requiring the task to be redistributed to be completed within the response time range allowed by the device's own capability), and takes the global and local energy consumption constraints as strict boundary conditions (this condition considers the steady-state energy consumption curve of each device in the group and whether it is currently in the peak energy consumption identifier state, ensuring that the total instantaneous power consumption or the predicted power consumption peak of the entire associated appliance group after redistribution does not exceed the pre-set circuit safety threshold or the rated capacity of the smart meter, thereby absolutely preventing the occurrence of overload risk). In this embodiment, heuristic algorithms such as genetic algorithm and particle swarm optimization algorithm or special online constraint solvers are used for fast approximate solution, and the final output is a series of specific and executable control command sets, i.e. the redistribution control instruction set. This instruction set specifies the unique identifier of each device that needs to be adjusted (device identifier), the specific control action and its parameters (redistribution control action parameters, such as setting the target temperature value, switching to a specific operating mode, adjusting the brightness percentage, setting the switch state, etc.), and the precise execution timing requirements (timing scheduling information, which takes into account the response delay difference between devices and the logical order and coordination relationship between device actions).

[0037] Next, the system performs the substantive dynamic strategy reconstruction operation: first, freeze all original instructions in the current cooperative control instruction sequence that have any form of conflict with the newly generated redistribution control instruction set in terms of target device, execution time window, or desired final state (for example, if the redistribution instruction requires the living room air conditioner to immediately switch to the powerful heating mode minutes, there may be minutes, the instruction of setting the temperature of the living room air conditioner lower or switching to the ventilation mode will be identified and frozen). Then, all instruction units in the control instruction set are re-distributed, sorted in descending order according to the priority identifiers carried therein, and inserted into the appropriate time position in the current collaborative control instruction sequence after the conflict avoidance calculation according to the sequence and the timing scheduling information of each instruction (usually, instructions with extremely high priority will try to preempt the earliest executable time slot, even interrupting the execution of low-priority instructions). In this way, a new collaborative control instruction sequence is generated after reconstruction.

[0038] Before the reconstructed collaborative control instruction sequence is issued to the actuators, the system must start the resource conflict detection algorithm for final comprehensive feasibility verification. This algorithm is a key link to ensure the stable and safe operation of the system, and its details are as follows: The resource conflict detection algorithm is a multi-dimensional, hierarchical verification process, which mainly detects the following three types of conflicts: Energy consumption conflict detection: This is the most core safety detection. First, extract the power consumption requirements of all device instructions that are scheduled to be executed concurrently in the same time period (usually with a granularity of one main power supply cycle or a short time window, such as seconds) in the reconstructed instruction sequence. The algorithm queries the device energy consumption feature vector library to obtain the typical power consumption value or power consumption range (from the steady-state energy consumption curve) of each device when executing a specific instruction (such as "air conditioner starts strong heating" or "oven preheats to degrees") and whether it is a peak power consumption identifier state. Then, calculate the total power consumption estimate of all concurrent action devices in this time window. Then compare this estimate with the system's preset multiple levels of safety thresholds (for example, the , , ) of the line rated capacity. If the total power consumption estimate exceeds the highest safety threshold (such as ), it is determined that there is an energy consumption conflict, and the sequence is marked as "dangerous". If it is above the high warning threshold (such as ), it is marked as "warning" and may need to be optimized. This process also considers the energy efficiency deviation coefficient. If a device coefficient is abnormally high, indicating that its current energy efficiency is low, the actual power consumption when executing the same instruction may be much higher than the typical value. The algorithm will use a penalty coefficient to amplify its power consumption estimate, thus more conservatively assessing the risk.

[0039] Timing conflict detection: This detection ensures the rationality of the instructions in time arrangement. The algorithm checks whether there are the following situations in the sequence: first, there is overlap in the execution time of two instructions assigned to the same device; second, there are instructions with logical dependency relationship (such as "turn off the curtain before turning on the projector") arranged in the wrong time sequence; third, the start time of the instruction has expired or is far later than the trigger time of the associated event, losing the response meaning; fourth, the duration of the instruction does not match the response delay threshold or the minimum stable running time requirement of the device. The algorithm maintains a virtual timeline of all devices to simulate the execution process of the instructions to detect these timing contradictions.

[0040] Device state conflict detection: This detection ensures that the state required by the instruction is reachable and does not conflict with the current state of the device or the target state required by other instructions. The algorithm accesses the current state of the device (such as "off", "idle", "running - cooling mode") and checks whether the state transformation required by the instruction is allowed by the device capability (for example, whether the device supports the mode, whether the time required to switch from the current state to the target state exceeds the response delay threshold). At the same time, it also checks whether multiple instructions in the sequence try to set the same device to different, mutually exclusive states at similar times (for example, one instruction requires the air conditioner to heat, and another instruction later requires it to cool).

[0041] The resource conflict detection algorithm traverses the entire reconstructed instruction sequence and performs the above checks for all potential time windows and device combinations. If all checks pass and no "danger" level conflicts are found, the reconstructed collaborative control instruction sequence is safely issued to the devices in the associated group of home appliances for execution. If any conflicts are detected, the algorithm initiates different handling strategies depending on the conflict level and type: for "warning" level conflicts (e.g., power consumption close to the threshold), the algorithm may attempt to automatically fine-tune (e.g., slightly delay the start time of a non-critical instruction); for "danger" level conflicts or conflicts that cannot be automatically resolved, the verification fails. When the verification fails, the system initiates the degradation strategy as described earlier: from the re-allocated control instruction set, the sub-instruction set with the highest priority identifier is selected. The system runs the resource conflict detection algorithm again to attempt to verify the compatibility of this reduced highest priority sub-instruction set with the energy consumption constraints. If this verification passes, only the sub-instruction set is issued and executed, and a system exception log is generated, recording the failure to fully execute all re-allocated instructions and the reason. If even this highest priority sub-instruction set fails the verification (indicating an extreme resource bottleneck or an unresolvable conflict), the system maintains the current collaborative control instruction sequence without executing any reconstruction operations triggered by this event, but simultaneously triggers a high-level system alarm signal (e.g., pushes an urgent notification to the mobile app, starts a sound-light alarm), notifying the user or system administrator of a major conflict or system resource bottleneck that cannot be automatically handled and requires manual intervention. If no active intervention signals or environmental mutation events requiring handling are detected from the beginning to the end, the dynamic strategy reconstruction mechanism remains in standby sleep state, and the system continues to stably execute the current collaborative control instruction sequence, maintaining the established operating state of the home environment.

[0042] The embodiment also discloses an intelligent home appliance interactive control system, comprising the following modules: A data acquisition module configured to acquire real-time running state parameters, environmental perception data, and user behavior characteristics of each home appliance in the living space; A data processing module configured to generate an environmental perception data package through a multi-modal data fusion model, for comprehensively quantifying the comfort index and device energy consumption characteristics of the home environment; A user habit modeling module configured to build a user habit preference model library based on the environmental perception data package, and generate a home appliance control strategy decision tree in combination with historical interaction records; A strategy matching module configured to match an optimal control strategy set according to a real-time scene type, and initialize a collaborative control instruction sequence of the home appliances; The dynamic reconstruction module is configured to trigger a dynamic policy reconstruction mechanism to generate a redistribution control instruction set when a user active intervention signal or an environmental mutation event is detected. The communication control module is configured to broadcast control instructions to the associated home appliance group and receive device state feedback. The data storage module is used to store environmental perception data packets, user habit preference model libraries, control policy decision trees and historical interaction records.

[0043] The specific embodiments are only an explanation of the present application, which is not a limitation of the present application. Those skilled in the art can make modifications to the embodiments without creative contribution after reading the specification, as long as the modifications are within the scope of the claims of the present application.

Claims

1. A method for interactive control of intelligent home appliances, characterized in that, Includes the following steps: S100: Real-time collection of operating status parameters, environmental perception data and user behavior characteristics of various home appliances in the living space; generates environmental perception data packages through a multimodal data fusion model; used to comprehensively quantify the comfort index of the home environment and the energy consumption characteristics of the equipment. S200. Based on the environmental perception data packet, a user habit preference model library is constructed, and a home appliance control strategy decision tree is generated by combining historical interaction records. The optimal control strategy set is matched according to the real-time scenario type. The home appliance control strategy decision tree includes time-dependent strategy, event-triggered strategy and energy-saving optimization strategy branches. The optimal control strategy set is used to initialize the collaborative control instruction sequence of home appliances. S300: When a user-initiated intervention signal or an environmental abrupt event is detected, a dynamic strategy reconfiguration mechanism is triggered to broadcast an instruction update request carrying a priority identifier to the associated home appliance group, and a redistribution control instruction set is generated based on the device response delay threshold and energy consumption constraints; wherein, the active intervention signal includes voice control instructions, gesture recognition instructions and emergency button trigger signals, and the environmental abrupt event includes sudden changes in temperature and humidity, abnormal power consumption fluctuations and safety risk alarms.

2. The intelligent home appliance interactive control method according to claim 1, characterized in that: S100 specifically includes the following steps: The operating status parameters of each household appliance in the living space are collected in real time through an embedded power metering module. The operating status parameters include instantaneous power consumption value, current working mode code and remaining life prediction value calculated based on the device's running time. Deploy a distributed environmental sensor network to synchronously collect multi-dimensional environmental sensing data; among them, generate a three-dimensional spatial temperature distribution matrix through a temperature sensor array, construct indoor humidity gradient data through a humidity sensor group, and generate dynamic spatiotemporal distribution data of light intensity using a light sensor matrix. The system uses millimeter-wave radar monitoring system and non-contact infrared sensing system to capture user behavior characteristics. Specifically, it analyzes human heart rate data through bio-reflection signals from millimeter-wave radar, obtains body surface temperature distribution data through non-contact infrared sensing system, and generates user activity trajectory heat map based on multi-target trajectory tracking algorithm. The fusion operation is performed based on a multimodal data fusion model. Specifically, the operating status parameters, environmental perception data, and user behavior characteristics are time-stamped and mapped to spatial coordinates. A feature-weighted fusion algorithm is used to generate a set of comfort indices for the home environment, which includes thermal comfort sub-indices, visual comfort sub-indices, and behavioral adaptability sub-indices. Furthermore, an energy consumption feature vector is generated based on the analysis of operating status parameters, and this feature vector includes peak energy consumption indicators, steady-state energy consumption curves, and energy efficiency deviation coefficients. An environmental perception data packet is generated based on the fusion results. The environmental perception data packet includes a set of comfort indices with timestamps, a device energy consumption feature vector, and a raw data check code.

3. The intelligent home appliance interactive control method according to claim 2, characterized in that: The generation of the thermal comfort sub-index, visual comfort sub-index, and behavioral fit sub-index specifically includes the following steps: Based on a three-dimensional spatial temperature distribution matrix, body surface temperature distribution data, and human heart rate data, a thermal comfort sub-index is generated through a thermal balance model, and humidity gradient data is used to correct for humidity effects. Based on dynamic spatiotemporal distribution data of light intensity, a visual comfort sub-index is generated through a light adaptation model, and a user activity trajectory heatmap is integrated for spatial location correlation analysis. Based on the user activity trajectory heatmap, a behavior adaptation sub-index is generated through a behavior pattern matching algorithm, and the device interaction adaptation is evaluated by combining the current working mode code in the operating status parameters. The thermal comfort sub-index, visual comfort sub-index, and behavior adaptation sub-index are then normalized.

4. The intelligent home appliance interactive control method according to claim 2, characterized in that: The generation of the device energy consumption feature vector specifically includes the following steps: Extract the power consumption value sequence of each household appliance within a continuous preset period, generate a steady-state energy consumption curve through a steady-state feature extraction algorithm, and mark the peak energy consumption exceeding the preset threshold. Based on the current operating mode code in the operating status parameters, the devices are grouped, and the deviation between the instantaneous power consumption value of the device under the same operating mode and the rated power or historical average power is calculated to generate an energy efficiency deviation coefficient; based on the remaining life prediction value, the energy efficiency alarm threshold used to determine whether the energy efficiency deviation coefficient is abnormal is dynamically adjusted; the peak energy consumption identifier, steady-state energy consumption curve and energy efficiency deviation coefficient are integrated to form a device energy consumption feature vector.

5. The intelligent home appliance interactive control method according to claim 1, characterized in that: S200 specifically includes the following steps: Based on the comfort index set, device energy consumption feature vector, and historical interaction records in the environmental perception data packet, a user habit preference model library is constructed using a clustering algorithm. The historical interaction records include the frequency of user active intervention signals, device usage time distribution data, and energy-saving preference scores. The clustering algorithm uses a K-means optimization model to generate user habit preference clusters, with each cluster corresponding to a habit preference type and a corresponding habit preference weight value. By combining the habit preference clusters and habit preference weights in the user habit preference model library, a home appliance control strategy decision tree is generated; wherein, the home appliance control strategy decision tree includes time-dependent strategy branches, event-triggered strategy branches, and energy-saving optimization strategy branches, and each branch is divided based on the decision tree splitting algorithm; The optimal control strategy set in the home appliance control strategy decision tree is matched based on the real-time scene type. The real-time scene type includes daily mode, energy-saving mode and emergency mode. The matching operation adopts a scene similarity calculation model to calculate the Euclidean distance between the current environmental perception data packet and each scene type, and selects the scene type with the smallest distance to activate the optimal control strategy set of the corresponding strategy branch.

6. The intelligent home appliance interactive control method according to claim 5, characterized in that: Following the step of matching the optimal set of control strategies in the home appliance control strategy decision tree based on the real-time scenario type, the following step is also included: Initialize the collaborative control instruction sequence of home appliances; wherein, the collaborative control instruction sequence is generated based on the optimal control strategy set, and the instruction sequence includes device identifiers, control action parameters and timing scheduling information, and the initialization operation includes instruction priority sorting and resource conflict detection to ensure that the instruction sequence is compatible with the peak energy consumption identifier and steady-state energy consumption curve in the device energy consumption feature vector.

7. The intelligent home appliance interactive control method according to claim 5, characterized in that: By combining the habit preference clusters and habit preference weights in the user habit preference model library, a home appliance control strategy decision tree is generated, specifically including the following steps: Based on the device usage time distribution data in the user habit preference model library, and combined with the comfort index set and device energy consumption feature vector in the environmental perception data package, multiple time intervals are divided through a time interval clustering algorithm, and a set of recommended operating modes for home appliances in each time interval is generated to construct a time-dependent strategy branch; wherein, the time-dependent strategy branch includes multiple time interval strategy nodes, and each time interval strategy node corresponds to a time interval and the set of recommended operating modes for that time interval; Based on the frequency of user-initiated intervention signals in the user habit preference model library, an event triggering probability model is established. Combined with the real-time collected active intervention signal types and environmental change event types, an event-triggered strategy branch is constructed through an event-action association rule library. The event-triggered strategy branch includes multiple event strategy nodes, each corresponding to an intervention event or environmental event type and the response action sequence of the home appliance when the event is triggered. The energy-saving preference score in the user habit preference model library and the energy efficiency deviation coefficient in the device energy consumption feature vector are integrated, and an energy-saving optimization strategy branch is generated by a multi-objective optimization algorithm with optimal energy efficiency as the goal. The energy-saving optimization strategy branch includes multiple energy efficiency strategy nodes, each corresponding to an energy efficiency optimization goal and a set of energy efficiency adjustment instructions that the home appliance needs to execute to achieve the goal.

8. The intelligent home appliance interactive control method according to claim 1, characterized in that: S300 specifically includes the following steps: Real-time monitoring of proactive intervention signals and sudden environmental events; wherein, the proactive intervention signals include voice control commands, gesture recognition commands, and emergency button trigger signals, and the sudden environmental events include sudden changes in temperature and humidity, abnormal power consumption fluctuations, and safety risk alarms; when any signal or event is detected, a dynamic policy reconstruction mechanism is activated; the dynamic policy reconstruction mechanism performs the following operations: Broadcast a request for an updated instruction with a priority identifier to the associated appliance group; and A set of redistribution control instructions is generated based on the equipment response delay threshold and energy consumption constraints. Otherwise, maintain the current sequence of coordinated control instructions.

9. The intelligent home appliance interactive control method according to claim 8, characterized in that, The dynamic strategy reconstruction mechanism specifically includes the following steps: Broadcast an instruction update request carrying a priority identifier to the associated home appliance group; wherein, the priority identifier is generated based on the event type: the highest priority identifier is assigned to the security risk alarm and emergency button trigger signal, the medium priority identifier is assigned to the sudden change in temperature and humidity and abnormal power consumption fluctuation, and the basic priority identifier is assigned to the regular voice control instructions and gesture recognition instructions. A set of redistribution control instructions is generated based on the equipment response delay threshold and energy consumption constraints, including: Obtain the real-time response delay parameters and device energy consumption feature vectors of each home appliance in the associated home appliance group; wherein, the device response delay threshold is preset according to the device type, and the device energy consumption feature vector includes peak energy consumption identifier, steady-state energy consumption curve and energy efficiency deviation coefficient; A redistribution optimization model is established, with priority identifier as the primary constraint, device response delay threshold as the secondary constraint, and energy consumption constraint as the boundary condition for multi-objective optimization calculation; wherein, the energy consumption constraint integrates steady-state energy consumption curve and peak energy consumption identifier, limiting the total power consumption after redistribution to not exceed a preset safety threshold. Based on the optimization calculation results, a set of reallocation control instructions is generated, which includes device identifiers, reallocation control action parameters and timing scheduling information. Execute a dynamic strategy reconstruction mechanism: freeze the instructions in the current cooperative control instruction sequence that conflict with the reallocation control instruction set, and insert the reallocation control instruction set in descending order of priority identifier to generate a reconstructed cooperative control instruction sequence; If the compatibility between the reconstructed instruction sequence and energy consumption constraints is verified by a resource conflict detection algorithm, the reconstructed collaborative control instruction sequence will be sent to the associated home appliance group for execution. If the verification fails, a degradation strategy will be initiated: based on the priority identifier, the sub-instruction set with the highest priority identifier will be selected from the redistribution control instruction set; the compatibility between this sub-instruction set and the energy consumption constraints will be verified again. If the verification passes, the sub-instruction set will be sent for execution, and a system exception log will be generated. If the verification of this sub-instruction set still fails, the current collaborative control instruction sequence will remain unchanged, and the highest level system alarm signal will be triggered.

10. An intelligent home appliance interactive control system, applied to the intelligent home appliance interactive control method according to any one of claims 1-9, characterized in that, It includes the following modules: The data acquisition module is configured to collect real-time operating status parameters of various household appliances in the living space, environmental perception data, and user behavior characteristics. The data processing module is configured to generate environmental perception data packets through a multimodal data fusion model, which are used to comprehensively quantify the comfort index of the home environment and the energy consumption characteristics of the equipment. The user habit modeling module is configured to build a user habit preference model library based on the environmental perception data package, and generate a home appliance control strategy decision tree by combining historical interaction records; The strategy matching module is configured to match the optimal set of control strategies based on the real-time scenario type and initialize the collaborative control instruction sequence of home appliances; The dynamic reconfiguration module is configured to trigger a dynamic policy reconfiguration mechanism and generate a set of reallocation control instructions when a user-initiated intervention signal or a sudden environmental event is detected. The communication control module is configured to broadcast control commands to associated home appliance groups and receive device status feedback. The data storage module is used to store environment-aware data packets, user habit and preference model library, control strategy decision tree, and historical interaction records.

Images