Sofa control method based on mobile terminal and intelligent sofa
By using multi-dimensional data acquisition and graph neural networks to build an intent reasoning engine, personalized control of the smart sofa and collaborative linkage with home appliances are realized, solving the problems of explicit operation and resource waste in existing technologies, and improving user experience and control accuracy.
Patent Information
- Application Number
- CN202512043433.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-17
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing smart sofa control methods rely on explicit operation, lack user status awareness and personalization, and cannot coordinate with other home devices, resulting in poor user experience and wasted resources.
By collecting multi-dimensional data to build a data fusion model, and using graph neural networks to build an intent reasoning engine, personalized control strategies are generated by combining user physiological, environmental and device data, and coordinated with other home devices to optimize control strategies in real time.
It enables seamless and personalized sofa control, enhancing user comfort and convenience, reducing resource waste, and improving the overall comfort and control accuracy of the home environment.
Smart Images

Figure CN121879169A_ABST
Abstract
Description
Technical Field
[0001] This invention proposes a sofa control method and a smart sofa based on a mobile terminal, belonging to the field of smart home technology. Background Technology
[0002] With the rapid development of smart homes, sofas, as an important piece of furniture in home life, have received much attention for their intelligent control. However, existing smart sofa control methods have many fundamental flaws.
[0003] Currently, most smart sofas rely on mobile apps or remote controls to send commands via Bluetooth or Wi-Fi, such as adjusting the backrest angle or activating the massage function. This interaction method is explicit and cumbersome, requiring users to actively open the app, find function buttons, and manually adjust them, severely disrupting the immersive experience. For example, the operation process can be distracting when watching movies or reading.
[0004] Meanwhile, the existing control mode is static and lacks awareness; the system cannot perceive the user's current state, such as fatigue, cold, or concentration, and can only mechanically execute commands, lacking intelligent judgment capabilities. Moreover, the sofa is isolated from other home appliances, such as air conditioners, lights, and audio systems, and cannot work together to create a cohesive and comfortable environment.
[0005] Furthermore, existing controls lack personalization, with all users using the same control logic without considering individual differences, such as elderly people preferring hard seats or young people preferring massage. Manufacturers' marketing also largely focuses on listing basic functions, with no one reconstructing sofa controls into a "seamless service engine." Essentially, they remain "motorized remote-controlled furniture," failing to meet users' demands for a high-quality smart home lifestyle. Summary of the Invention
[0006] This invention provides a sofa control method and a smart sofa based on a mobile terminal, to solve the problems mentioned in the background section above: The present invention proposes a sofa control method based on a mobile terminal, the method comprising: S1. Collect multi-dimensional data on user physiology, environment, and device status to generate user physiological data, environmental data, and device data; construct a data fusion model based on the multi-dimensional data. S2. Based on the data fusion model, an intent reasoning engine is built using graph neural networks. User physiological data, environmental data, and device data are input into the intent reasoning engine for deep analysis and learning to infer the user's current needs and intentions, and generate user needs and intention data. S3. Based on user demand and intent data, combined with user personalized preference information, formulate sofa control strategies and generate personalized sofa control strategy data. S4. Based on the personalized sofa control strategy data, control the sofa to generate sofa control command data; at the same time, coordinate the sofa control command data with other home appliances to generate home appliance coordinated control command data. S5. Receive sofa control command data and home appliance collaborative control command data through the mobile terminal, and provide real-time feedback on control execution status, generating control execution feedback data; dynamically optimize and adjust the intent reasoning engine and control strategy based on the control execution feedback data, generating optimization adjustment data.
[0007] The smart sofa based on a mobile terminal proposed in this invention includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method as described in any one of the above statements.
[0008] The beneficial effects of this invention are as follows: By integrating multi-dimensional data from physiology, environment, and devices to construct a graph neural network intent reasoning engine, it can accurately perceive the user's current state and infer potential needs. This reduces instances where users' immersive experience is interrupted by explicit sofa operations, avoids the inconvenience caused by traditional sofa controls relying on user-initiated actions, and improves the comfort and convenience of using the sofa. Furthermore, it formulates sofa control strategies in real-time based on user intent data and personalized preference information, and coordinates with other home appliances for integrated control. This quickly creates an environment that meets user needs and provides overall comfort. It also offers personalized services based on individual user differences, reducing the probability of low user satisfaction due to fragmented scenarios and lack of personalization, and minimizing user frustration caused by unintelligent sofa controls. This system addresses the challenges of controlling sofas and enhances the intelligent experience of home life. Real-time feedback on control execution via mobile terminals allows for dynamic optimization and adjustment of the intent reasoning engine and control strategies. It provides a clear understanding of sofa control effectiveness and analyzes the reasons for inaccurate control, avoiding resource waste and user experience degradation caused by blind control. This reduces deviations or errors during control, improving the accuracy and effectiveness of sofa control. Continuous improvement of sofa control services based on optimization data reduces the problem of fixed control logic failing to adapt to different scenarios and changing user needs, preventing long-term inefficiency or irrationality in sofa control. It elevates sofa control from explicit operation to a seamless service, enhancing the adaptability and sophistication of sofa control methods. Attached Figure Description
[0009] Figure 1 This is a diagram of the method described in this invention; Figure 2 This is a structural diagram of the intent reasoning engine described in this invention; Figure 3 This is a schematic diagram of the device collaborative control and feedback optimization described in this invention. Detailed Implementation
[0010] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0011] One embodiment of the present invention, such as Figure 1 As shown, a sofa control method based on a mobile terminal includes: S1. Collect multi-dimensional data on user physiology, environment, and device status to generate user physiological data (such as heart rate, body temperature, and body movement), environmental data (such as temperature, humidity, and light intensity), and device data (such as air conditioner operating status, light brightness, and audio playback content); construct a data fusion model based on the multi-dimensional data, which is used to provide basic data support for subsequent intent reasoning. S2. Based on the data fusion model, an intent reasoning engine is constructed using graph neural networks; user physiological data, environmental data, and device data are input into the intent reasoning engine, such as... Figure 2 As shown, deep analysis and learning are performed to infer the user's current needs and intentions, and generate user needs and intention data; for example, when it is detected that the user's heart rate is increasing, body temperature is rising and the ambient temperature is high, it can be inferred that the user may have a need to cool down. S3. Based on user demand intent data and combined with user personalized preference information (such as preset personalized parameters such as the elderly preferring hard seats and young people preferring massage), formulate sofa control strategies and generate personalized sofa control strategy data; this strategy data not only considers the user's current needs, but also takes into account the individual differences of the user. S4. Based on personalized sofa control strategy data, perform non-intrusive and predictive control operations on the sofa to generate sofa control command data; at the same time, coordinate and control the sofa control command data with other home appliances (such as air conditioners, lights, and audio equipment) to generate home appliance collaborative control command data; for example, when it is inferred that the user has a cooling need and prefers a soft sitting feel, not only adjust the sofa backrest angle and firmness to provide a more comfortable sitting feel, but also coordinate and control the air conditioner to lower the temperature and dim the lights to create a comfortable atmosphere; S5. Receive sofa control command data and home appliance collaborative control command data through the mobile terminal, and provide real-time feedback on control execution status, generating control execution feedback data; dynamically optimize and adjust the intent reasoning engine and control strategy based on the control execution feedback data, generating optimization adjustment data to continuously improve the accuracy of sofa control and user experience, realizing a paradigm shift in sofa control from explicit operation to seamless service, and generating the final optimized sofa control service data.
[0012] The working principle and effects of the above technical solution are as follows: By collecting multi-dimensional data on user physiology, environment, and devices, and combining it with a data fusion model and a graph neural network inference engine, the probability of misjudging user needs is reduced, inference bias caused by a single data dimension is avoided, the accuracy of demand identification is improved, and the scientific nature of control decisions is enhanced. By combining users' personalized preferences to formulate sofa control strategies, and linking home appliances such as air conditioners and lights to achieve scenario-based collaborative control, the differences in adaptation between individual needs and control effects are reduced, the sense of fragmentation in scenario control by a single device is avoided, the overall comfort of the home environment is improved, and the coordination of the user experience is enhanced. By reducing explicit operation steps through seamless predictive control and dynamically optimizing inference engine parameters and control strategies based on control execution feedback, user operation costs and the frequency of manual intervention are reduced. This avoids the problem of decreased adaptability after long-term use, reduces the waste of resources from ineffective control, improves the timeliness of service adjustments, and allows sofa control to shift from passive response to proactive adaptation, thus enhancing the intelligent closed loop of full-scenario services. By dynamically optimizing and iterating the control logic, and combining execution feedback to verify the control effect, we avoid the poor experience caused by blind control, reduce the limitations caused by the solidification of control strategies, improve the adaptability for long-term use, make the sofa control service more in line with the dynamic needs of users, and enhance the reliability of the seamless service.
[0013] One embodiment of the present invention, such as Figure 3 The S1, as described above, includes: S11. Determine the core data collection dimensions of user physiology, environment and device status, select suitable sensing devices and data collection interfaces, and generate a data collection configuration scheme. S12. Through the sensing devices associated with the mobile terminal, real-time collection of user physiological data, environmental data and device data are collected. The physiological data includes heart rate and body temperature, the environmental data includes ambient temperature and humidity, and the device data includes air conditioner operating status and light brightness, generating an original multidimensional dataset. S13. Perform outlier removal and data standardization on the original multidimensional dataset, and integrate it to generate regular user physiological data, environmental data and device data; S14. Based on the normalized three types of data, a multi-source data fusion model is constructed. The multi-source data fusion model integrates the three types of data in a structured manner through a data association algorithm, providing standardized basic data support for subsequent intent reasoning.
[0014] The working principle and effects of the above technical solution are as follows: By clearly defining the core data collection dimensions of user physiology, environment, and equipment, and selecting suitable sensing devices and data collection interfaces, resource redundancy caused by irrelevant data collection is reduced, data collection failures caused by device or interface incompatibility are avoided, the relevance and stability of data collection are improved, and the controllability of the collection process is enhanced. Outlier removal and standardization of the original multidimensional dataset reduced interference from data noise and outliers, avoided the potential impact of inconsistent data units across different dimensions on subsequent analysis, improved the regularity and reliability of the data, and provided a high-quality data foundation for subsequent processing. Based on the three types of data after normalization, a multi-source data fusion model is constructed. The structured integration is achieved through data association algorithms, which reduces the sense of fragmentation of physiological, environmental and equipment data, avoids inference bias caused by insufficient support from a single data dimension, improves the comprehensive support of basic data, and enhances the scientificity and accuracy of subsequent intention inference. The entire data collection and fusion process forms a closed loop, reducing data loss from collection to preprocessing, avoiding subsequent service deviations caused by data quality issues, improving data utilization efficiency, making basic data support more aligned with actual application needs, and enhancing the underlying reliability of the entire process service.
[0015] In one embodiment of the present invention, S13 includes: S131. Perform integrity verification on the original multidimensional dataset, check for issues such as missing data and incorrect formatting, and generate a data quality verification report. S132. Based on the data quality verification report, use a statistical anomaly detection algorithm to identify and remove outliers in physiological, environmental, and equipment data to generate an anomaly-free dataset. S133. For the outlier removal dataset, the Min-Max standardization method is used to unify the units and value ranges of each data dimension to generate a standardized processing dataset. S134. Based on the classification dimensions of physiological data, environmental data, and equipment data, the standardized processing dataset is structurally integrated to generate three types of target data.
[0016] The working principle and effects of the above technical solution are as follows: By performing integrity checks on the original multidimensional dataset, issues such as missing data and incorrect formats are identified and quality reports are generated. This reduces the inflow of invalid data into subsequent stages, avoids processing anomalies caused by inconsistent data formats, improves the pre-processing control of data quality, and enhances the standardization of the preprocessing stage. The statistical anomaly detection algorithm based on the verification report is used to remove outliers, which reduces the interference of extreme data and noisy data on the analysis results, avoids the hidden danger of outliers misleading subsequent reasoning, improves the purity and reliability of the dataset, and lays a solid foundation for standardized processing. By standardizing the data dimensions and ranges using the Min-Max standardization method, the analytical bias caused by differences in measurement standards for different types of data is reduced, the problem of insufficient data comparability due to inconsistent dimensions is avoided, the data adaptability is improved, and the accuracy of subsequent data fusion and inference is enhanced. By structurally integrating standardized datasets according to classification dimensions, the inefficiency caused by messy and disordered data is reduced, the risk of errors in using the three types of data interchangeably is avoided, the regularity of data and retrieval efficiency are improved, and subsequent model building and intent inference can call target data more efficiently, thus enhancing the practicality of data preprocessing.
[0017] In one embodiment of the present invention, S134 includes: Define the specific data item boundaries for physiological data, environmental data, and equipment data (e.g., physiological data is limited to heart rate, body temperature, etc., and environmental data is limited to temperature, humidity, etc.), and generate detailed classification dimension rules; Based on the classification dimension details, the standardized dataset is traversed and each data item is accurately matched to the corresponding classification dimension to generate the classified dataset. For each category of data in the categorized dataset, arrange the data in order according to the collection timestamp, complete the data association identifiers, and generate a regularized dataset within the category; By integrating three types of intraclass regularized datasets, unifying data storage formats and indexing rules, regularized user physiological data, environmental data, and device data are generated.
[0018] The working principle and effects of the above technical solution are as follows: By clearly defining the specific boundaries of physiological, environmental, and equipment data items and generating detailed classification dimension rules, the risk of confusion between different types of data items is reduced, the problem of unclear definition caused by vague classification standards is avoided, the clarity of data classification is improved, and a solid foundation is laid for subsequent accurate matching. By traversing the standardized dataset according to the classification rules, the data items are accurately matched, reducing misclassification and omission of data, avoiding the impact of classification bias on subsequent use, improving the accuracy of classification results, and enhancing the regularity of the dataset. After classifying the data, sorting it by collection timestamp and supplementing the association identifiers reduces the problem of disordered data within the category, avoids the analysis obstacles caused by broken data associations, improves the logic of data within the category, and enhances the traceability and relevance of the data. By integrating three types of standardized datasets and unifying storage formats and indexing rules, the cost of format adaptation during data retrieval is reduced, the inefficiency of retrieval caused by inconsistent indexes is avoided, and the reusability and retrieval efficiency of data are improved. This allows subsequent data fusion and intent reasoning to quickly and accurately retrieve target data, enhancing the practicality and support of the preprocessing stage.
[0019] In one embodiment of the present invention, S2 includes: S21. Based on the output data format and characteristics of the data fusion model, determine the hierarchical structure and training parameters of the graph neural network, and determine the data input-output mapping rules; S22. Build an intent reasoning engine according to the design scheme, and initialize and debug the data analysis process and reasoning logic of the intent reasoning engine to ensure that it is compatible with multi-source fusion data. S23. Input user physiological data, environmental data and device data into the debugged intent reasoning engine, and through deep feature extraction and association learning, mine the user behavior logic behind the data. S24. Based on the learning results, infer the user's current core needs, generate accurate user need intent data, and determine the user's sofa usage and related home furnishing needs in the current scenario.
[0020] The working principle and effects of the above technical solution are as follows: By combining the output format and features of the data fusion model, the hierarchical structure, training parameters and data mapping rules of the graph neural network are determined, which reduces the problem of mismatch between model structure and data features, avoids inefficient reasoning caused by blind parameter setting, improves the adaptability of the intent reasoning engine, and lays a solid foundation for subsequent accurate reasoning. The engine was built and the process and reasoning logic were debugged and analyzed according to the design scheme to ensure that it was compatible with multi-source fused data. This reduced problems such as process lag and logical conflicts after data input, avoided reasoning interruption caused by engine incompatibility with data, improved the stability of engine operation, and enhanced the controllability of the reasoning process. By inputting the three types of data into the debugged engine, user behavior logic is mined through deep feature extraction and association learning. This reduces the limitations of single-data-dimensional analysis, avoids the hidden danger that surface data cannot reflect real needs, improves the depth and comprehensiveness of behavior logic mining, and provides solid support for demand reasoning. Based on the learning results, the system accurately infers the core needs of users and generates demand intent data, which reduces the probability of misjudgment and omission of demands, avoids the problem of the inference results being out of touch with the actual scenario, improves the accuracy of demand identification, and makes the subsequent sofa control strategy more in line with the actual needs of users, thereby enhancing the practicality and guidance of the inference results.
[0021] In one embodiment of the present invention, step S21 includes: Extract standardized data from the data fusion model output, analyze its data format (such as dimension, data type, and structural hierarchy) and core features (such as feature distribution, correlation strength, and dynamic change patterns), and generate a data feature analysis report; Based on the data feature analysis report, the input layer dimension, the number of hidden layer nodes and the connection method of the graph neural network are determined, the complexity of the data features is matched, and a hierarchical structure design scheme is generated. By combining the dynamic nature of data features with learning requirements, the training parameters of the graph neural network (such as learning rate, number of iterations, and regularization coefficient) are determined, and a parameter configuration table is generated. Based on the hierarchical structure design scheme and parameter configuration table, determine the mapping relationship between data input to each layer of the graph neural network and the corresponding logic between the output results and the required intent, and generate a data input and output mapping rule description.
[0022] The working principle and effects of the above technical solution are as follows: By extracting standardized data and analyzing its format and core features, a data feature analysis report is generated, which reduces the problem of incomplete data information mining, avoids the blindness of subsequent model design that is divorced from the actual data, improves the control over the essence of the data, and provides a precise basis for model design. Based on the analysis report, the hierarchical structure of the graph neural network was determined, and the complexity of the data features was matched. This reduced the adaptation problem of incompatibility between the model structure and the data, avoided inefficient reasoning caused by overly simple or overly complex hierarchical design, improved the rationality of the model structure, and enhanced the adaptability to complex data features. By combining data dynamism and learning needs to determine training parameters and generate configuration tables, the subjectivity and arbitrariness of parameter settings are reduced, and unreasonable parameters such as learning rate and number of iterations are avoided, which can lead to model training oscillations or slow convergence. This improves the scientific nature of parameter configuration and speeds up model training efficiency. By clarifying the data input and output mapping rules based on the structural scheme and parameter table, logical confusion in the process of data flow and intent correspondence is reduced, reasoning deviations caused by ambiguous mapping relationships are avoided, and the accuracy of the correspondence between data and demand intent is improved, laying a solid logical foundation for the stable operation of the subsequent intent reasoning engine.
[0023] In one embodiment of the present invention, step S23 includes: Perform format validation and redundancy removal on user physiological data, environmental data, and device data to ensure that the data conforms to the input specifications of the debugged intent reasoning engine and generate an adaptive input dataset; Import the adaptive input dataset into the intent reasoning engine, start the deep feature extraction module, extract key features of each data dimension (such as the fluctuation features of physiological data and the gradient change features of environmental data), and generate a multi-dimensional feature set. Based on a multi-dimensional feature set, the engine's association learning algorithm analyzes the dynamic correlation between physiological, environmental, and equipment data (such as the response relationship between body temperature changes and environmental temperature) and generates a feature association rule set. By combining feature association rule sets, we can analyze the causal logic and user behavior patterns between data (such as the change pattern of user body movement frequency when ambient light is enhanced), mine the user behavior logic behind the data, and generate a user behavior logic map.
[0024] The working principle and effects of the above technical solution are as follows: By performing format validation and redundancy removal on the three types of data, we ensured that the data met the engine input specifications and generated an adaptive dataset. This reduced the problems of data format disorder and redundant information interference, avoided engine malfunctions caused by data incompatibility, improved the adaptability of data input, and laid a solid foundation for subsequent feature extraction. The deep feature extraction module is activated to extract key features from various dimensions and generate a multi-dimensional feature set. This reduces the limitations of relying solely on surface data, avoids the omission of core features, improves the accuracy of feature extraction, and enhances the data's ability to represent user needs. By analyzing the dynamic relationships between three types of data through association learning algorithms, a set of feature association rules is generated, which reduces the one-sidedness of analysis based on a single data dimension, avoids cognitive bias caused by the break in the logical relationship between data, improves the control over the relationship between scenarios, and makes the analysis more in line with actual use scenarios. By combining association rule sets to analyze causal logic and behavioral patterns, a user behavior logic graph is generated, which reduces the blindness of demand reasoning, avoids misjudgment of demand due to unclear behavioral logic, improves the support for subsequent intent reasoning, enhances the logic and reliability of the reasoning process, and makes demand identification more in line with the user's real behavioral logic.
[0025] In one embodiment of the present invention, S3 includes: S31. Obtain preset personalized preference information through the user profile database of the mobile terminal. The personalized preference information includes sitting posture preference and firmness requirement, and perform validity verification. S32. Establish a mapping mechanism between user demand intent data and personalized preference information, and determine preference adaptation rules under different demand scenarios; S33. Combine the scenario adaptation rules to construct a control strategy formulation framework, which covers core logic such as demand priority ranking and preference weight allocation. S34. Based on the control strategy framework, generate personalized sofa control strategy data to accurately match current needs with individual differences.
[0026] The working principle and effects of the above technical solution are as follows: By acquiring personalized information such as user sitting posture preferences and firmness requirements and conducting validity verification, the interference of invalid or outdated preferences on strategy formulation is reduced, the problem of generating control schemes based on failure information is avoided, the reliability of preference data is improved, and a solid foundation is laid for subsequent accurate adaptation. Establish a mapping mechanism between user needs and intentions and personalized preferences, clarify the adaptation rules under different scenarios, reduce the adaptation deviation of the disconnect between needs and preferences, avoid the blindness of strategy formulation, improve the matching accuracy between scenarios and individual preferences, and enhance the pertinence of control strategies. The framework for formulating control strategies, which includes prioritizing requirements and allocating preference weights, reduces decision-making confusion when multiple requirements conflict, avoids inefficiency caused by disordered strategy logic, improves the orderliness of strategy formulation, and makes the control logic more in line with actual use cases. The framework generates personalized sofa control strategies, achieving a precise match between current needs and individual differences. This reduces the gap between generalized strategies and users' actual needs, avoids poor experience caused by ignoring individual differences, improves the personalization of sofa control, and enhances user comfort and satisfaction.
[0027] In one embodiment of the present invention, step S4 includes: S41. Analyze the personalized sofa control strategy data, extract the core control parameters, including the sofa backrest angle and firmness adjustment, and generate basic control commands. S42. Perform non-sensory execution verification on the basic control commands to ensure that the commands meet the operating threshold of the sofa hardware and generate the final sofa control command data. S41. Analyze the coordination logic between the sofa control and other home appliances (air conditioner, lighting, etc.), and determine the linkage trigger conditions and parameter matching standards; S41. Based on the collaborative logic, the sofa control command data is combined with the home appliance control requirements to generate home appliance collaborative control command data, thereby achieving scenario-based linkage.
[0028] The working principle and effects of the above technical solution are as follows: By analyzing personalized sofa control strategies and extracting core parameters such as backrest angle and firmness adjustment, basic control commands are generated. This reduces the deviation and omission of core control parameters, avoids command redundancy caused by interference from irrelevant parameters, improves the pertinence of control commands, and lays a solid foundation for precise control. Conducting non-intrusive execution verification of basic control commands ensures compliance with the sofa hardware operating thresholds, reduces the risk of hardware failure caused by commands exceeding the range, avoids damage to the equipment caused by forcibly executing improper commands, improves the safety and stability of control execution, and enhances the durability of the equipment. By analyzing the collaborative logic between sofas and home appliances such as air conditioners and lights, and clarifying the triggering conditions and parameter matching standards for linkage, we have reduced logical conflicts when multiple devices are linked, avoided the sense of scene fragmentation caused by improper linkage timing and parameter mismatch, improved the coordination of home appliance collaboration, and made the scene-based service more coherent. By combining sofa control commands with home appliance control needs, collaborative control commands are generated to achieve scenario-based linkage. This reduces the tedious process of users manually operating multiple devices, avoids the problem that single device control cannot meet the overall scenario needs, improves the convenience of home control, and enhances the integrity and immersion of the user experience.
[0029] In one embodiment of the present invention, S41 includes: Load personalized sofa control strategy data, extract the core control dimensions from the strategy using a structured parsing algorithm, and generate a list of control dimensions; Based on the control dimension list, core control parameters such as sofa backrest angle and firmness adjustment are accurately identified, the target value range of each parameter is determined (e.g., backrest angle is 75°), and parameter details are generated. Based on the instruction communication protocol of the sofa hardware, the parameter details are converted into a machine-recognizable instruction format, generating a basic control instruction prototype. Verify the completeness of parameters and the compliance of format in the prototype of the basic control instruction, fill in missing information, correct format deviations, and finally generate the basic control instruction.
[0030] The working principle and effects of the above technical solution are as follows: By extracting the core control dimensions in the strategy and generating a list through structured parsing algorithms, the interference of irrelevant dimensions on instruction generation is reduced, redundant design caused by ambiguity in control direction is avoided, the focus of instruction generation is improved, and subsequent parameter extraction is more targeted. Based on the checklist, core parameters such as sofa backrest angle and firmness adjustment are accurately identified, and the target value range of each parameter is clearly defined. This reduces execution deviations caused by ambiguous parameter definitions, avoids the problem of control effects not meeting expectations, improves the accuracy of parameter settings, and makes control commands more in line with actual needs. According to the communication protocol of the sofa hardware, the parameter details are converted into a machine-recognizable format to generate a basic control instruction prototype. This reduces the risk of instruction failure caused by format incompatibility, avoids execution obstacles that the machine cannot parse, improves the compatibility between instructions and hardware, and ensures the smooth transmission of instructions. Verifying the completeness and format compliance of the prototype instructions and correcting deviations reduces instruction defects such as missing parameters and format errors, avoids control interruptions caused by instruction problems, improves the reliability of basic control instructions, and lays a solid foundation for subsequent seamless execution and device collaboration.
[0031] In one embodiment of the present invention, step S5 includes: S51. The mobile terminal receives sofa control command data and home appliance collaborative control command data, forwards them to the corresponding devices through the wireless communication module, and triggers execution. S52. Real-time acquisition of the control execution status of the sofa and related home appliances, recording the execution effect and implicit user feedback (such as changes in body movement), and generating control execution feedback data; S53. Based on control execution feedback data, locate the inference bias of the intent inference engine and the adaptation defects of the control strategy; S54. Optimize the algorithm parameters and control strategy adaptation rules of the intent reasoning engine in a targeted manner, and generate optimization and adjustment data; S55. By optimizing and adjusting the data update service logic, the final optimized sofa control service data is generated, realizing a paradigm shift from explicit operation to seamless service, and ensuring the continuous accuracy of sofa control.
[0032] The working principle and effects of the above technical solution are as follows: By receiving control commands via a mobile terminal and forwarding them to the corresponding device through a wireless communication module, the delay and loss risk of command transmission are reduced, execution interruption caused by the device not receiving commands is avoided, the reliability of command transmission is improved, and the timely triggering of control operations is guaranteed. Real-time collection of the execution status of sofas and related home appliances, recording execution effects and implicit feedback such as user movements, reduces the one-sidedness of feedback that relies solely on explicit results, avoids the problem of users' real experience not being captured, improves the comprehensiveness of feedback data, and provides a solid basis for subsequent optimization; By addressing the biases of the intent reasoning engine and the adaptation defects of the control strategy based on feedback data, we reduce the waste of resources from blind optimization, avoid ineffective optimization caused by not accurately identifying the root cause of the problem, improve the accuracy of problem localization, and make the optimization direction clearer and more definite. Targeted optimization of engine algorithm parameters and strategy adaptation rules reduces adaptation bias caused by generalization adjustments, avoids repeated effects after optimization, improves the effectiveness of optimization adjustments, and enhances the dynamic adaptation capability of service logic. By updating service logic to generate optimized service data, the transformation from explicit operation to seamless service is achieved, reducing the need for manual intervention, avoiding the problem of decreased accuracy after long-term use, improving the continuous stability of control, and enhancing the consistency and satisfaction of user experience.
[0033] In one embodiment of the present invention, S53 includes: Key information is extracted from the control execution feedback data, and core indicators such as the execution effect achievement status and the abnormal characteristics of implicit user feedback are screened to generate a localization analysis sample set; The analysis sample set is compared with the expected output of the intent reasoning engine and the preset target of the control strategy to identify data differences and generate a deviation comparison list. Based on the deviation comparison list, analyze the causes of the differences and distinguish whether they are feature extraction deviations or logical judgment deviations of the inference engine, or parameter adaptation defects or scene coverage defects of the control strategy. Summarize the types of deviations and categories of defects, clarify the specific manifestations of reasoning deviations (such as misjudgment of requirements and deviation in intensity estimation) and the core scenarios of strategy defects, and generate a problem location report.
[0034] The working principle and effects of the above technical solution are as follows: By extracting core indicators such as the performance of execution and the characteristics of hidden user feedback anomalies, a sample set for localization analysis is generated. This reduces the interference of redundant feedback information, avoids the analysis bias caused by key information being obscured by secondary data, improves the pertinence of problem localization, and lays a solid foundation for subsequent accurate analysis. By comparing the analysis sample set with the expected output of the inference engine and the preset targets of the control strategy, the data difference points are identified and a deviation list is generated. This reduces the subjectivity of judging problems based on experience, avoids the omission or misjudgment of differences, improves the accuracy of difference identification, and makes the problem clues more intuitive and clear. Based on the analysis of the differences in the list, the feature extraction of the inference engine, the logical judgment deviation and the parameter adaptation of the control strategy, and the scene coverage defects are distinguished. This reduces the positioning error caused by confusion of different types of problems, avoids the optimization direction deviation caused by general attribution, improves the identification of the root cause of the problem, and makes the positioning more in-depth. By summarizing deviation types and defect categories to generate a problem localization report, the specific manifestations of deviations and core scenarios of defects are clearly defined. This reduces communication costs associated with vague problem descriptions, avoids blind optimization without clear basis, improves the accuracy of problem communication, provides clear guidance for targeted optimization, and enhances the efficiency and effectiveness of the optimization process.
[0035] One embodiment of the present invention provides a smart sofa based on a mobile terminal, comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method as described in any one of the above statements.
[0036] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.
Claims
1. A sofa control method based on a mobile terminal, characterized in that, The method includes: S1. Collect multi-dimensional data on user physiology, environment, and device status to generate user physiological data, environmental data, and device data; construct a data fusion model based on the multi-dimensional data. S2. Based on the data fusion model, an intent reasoning engine is built using graph neural networks. User physiological data, environmental data, and device data are input into the intent reasoning engine for deep analysis and learning to infer the user's current needs and intentions, and generate user needs and intention data. S3. Based on user demand and intent data, combined with user personalized preference information, formulate sofa control strategies and generate personalized sofa control strategy data. S4. Based on the personalized sofa control strategy data, control the sofa to generate sofa control command data; at the same time, coordinate the sofa control command data with other home appliances to generate home appliance coordinated control command data. S5. Receive sofa control command data and home appliance collaborative control command data through the mobile terminal, and provide real-time feedback on control execution status, generating control execution feedback data; dynamically optimize and adjust the intent reasoning engine and control strategy based on the control execution feedback data, generating optimization adjustment data.
2. The sofa control method based on a mobile terminal according to claim 1, characterized in that, S1 includes: S11. Determine the core data collection dimensions of user physiology, environment and device status, select suitable sensing devices and data collection interfaces, and generate a data collection configuration scheme. S12. Through the sensing devices associated with the mobile terminal, collect user physiological data in real time, as well as environmental data and device data, to generate raw multidimensional datasets. S13. Perform outlier removal and data standardization on the original multidimensional dataset, and integrate it to generate regular user physiological data, environmental data and device data; S14. Based on the normalized three types of data, construct a multi-source data fusion model.
3. The sofa control method based on a mobile terminal according to claim 2, characterized in that, S13 includes: S131. Perform integrity verification on the original multidimensional dataset and generate a data quality verification report; S132. Based on the data quality verification report, use a statistical anomaly detection algorithm to identify and remove outliers in physiological, environmental, and equipment data to generate an anomaly-free dataset. S133. For the outlier removal dataset, the Min-Max standardization method is used to unify the units and value ranges of each data dimension to generate a standardized processing dataset. S134. Based on the classification dimensions of physiological data, environmental data, and equipment data, the standardized processing dataset is structurally integrated to generate three types of target data.
4. The sofa control method based on a mobile terminal according to claim 3, characterized in that, S134 includes: Determine the specific data item boundaries for physiological data, environmental data, and equipment data, and generate detailed classification dimension rules; Based on the classification dimension details, the standardized dataset is traversed, and each data item is matched to the corresponding classification dimension to generate the classified dataset. For each category of data in the categorized dataset, arrange the data in order according to the collection timestamp, complete the data association identifiers, and generate a regularized dataset within the category; By integrating three types of intraclass regularized datasets, unifying data storage formats and indexing rules, regularized user physiological data, environmental data, and device data are generated.
5. The sofa control method based on a mobile terminal according to claim 1, characterized in that, The S2 includes: S21. Based on the output data format and characteristics of the data fusion model, determine the hierarchical structure and training parameters of the graph neural network, and determine the data input-output mapping rules; S22. Build an intent reasoning engine according to the design scheme, and initialize and debug the data analysis process and reasoning logic of the intent reasoning engine. S23. Input user physiological data, environmental data and device data into the debugged intent reasoning engine, and through deep feature extraction and association learning, mine the user behavior logic behind the data. S24. Based on the learning results, infer the user's current core needs, generate user demand intent data, and determine the user's sofa usage and related home furnishing needs in the current scenario.
6. The sofa control method based on a mobile terminal according to claim 5, characterized in that, S21 includes: Extract standardized data from the data fusion model output, analyze its data format and core features, and generate a data feature analysis report; Based on the data feature analysis report, the input layer dimension, the number of hidden layer nodes and the connection method of the graph neural network are determined, the complexity of the data features is matched, and a hierarchical structure design scheme is generated. By combining the dynamic nature of data features with learning requirements, the training parameters of the graph neural network are determined, and a parameter configuration table is generated. Based on the hierarchical structure design scheme and parameter configuration table, determine the mapping relationship between data input to each layer of the graph neural network and the corresponding logic between the output results and the required intent, and generate a data input and output mapping rule description.
7. The sofa control method based on a mobile terminal according to claim 1, characterized in that, The S3 includes: S31. Obtain preset personalized preference information through the user profile database of the mobile terminal and perform validity verification; S32. Establish a mapping mechanism between user demand intent data and personalized preference information, and determine preference adaptation rules under different demand scenarios; S33. Combine scenario adaptation rules to construct a control strategy formulation framework; S34. Based on the control strategy framework, generate personalized sofa control strategy data and match current needs with individual differences.
8. The sofa control method based on a mobile terminal according to claim 1, characterized in that, The S4 includes: S41. Analyze the personalized sofa control strategy data, extract the core control parameters, and generate basic control commands; S42. Perform non-sensory execution verification on the basic control commands and generate the final sofa control command data; S41. Analyze the collaborative logic between the sofa control and other home appliances, and determine the linkage trigger conditions and parameter matching standards; S41. Based on the collaborative logic, combine the sofa control command data with the home appliance control requirements to generate home appliance collaborative control command data.
9. The sofa control method based on a mobile terminal according to claim 1, characterized in that, The S5 includes: S51. The mobile terminal receives sofa control command data and home appliance collaborative control command data, forwards them to the corresponding devices through the wireless communication module, and triggers execution. S52. Real-time acquisition of the control execution status of the sofa and related home appliances, recording the execution effect and implicit user feedback, and generating control execution feedback data; S53. Based on control execution feedback data, locate the inference bias of the intent inference engine and the adaptation defects of the control strategy; S54. Optimize the algorithm parameters and control strategy adaptation rules of the intent reasoning engine in a targeted manner, and generate optimization and adjustment data; S55. By optimizing and adjusting the data update service logic, the final optimized sofa control service data is generated.
10. Smart sofas based on mobile terminals, including: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.
