An intelligent home joint control method and system based on data analysis

Abstract

This invention discloses a system and method belonging to the field of smart home technology. The method includes the following steps: constructing a dynamic influence relationship hypergraph; triggering model verification logic; generating a virtual probe task based on the triggering result of the model verification logic and obtaining the expected influence propagation path; executing the virtual probe task while simultaneously collecting actual response data; comparing and analyzing the collected actual response data with the expected state in the dynamic influence relationship hypergraph, and updating and correcting accordingly; generating the final joint control command and sending it to the smart home execution device. This invention, by constructing a device relationship cognitive model that can dynamically perceive and self-correct, achieves the effect of ensuring the long-term accuracy and reliability of smart home automation joint control, solving the problem of control failure caused by the inability of static knowledge models in existing technologies to adapt to the dynamic evolution of the real home environment.

Description

Technical Field

[0001] This invention relates to the field of smart home technology, and in particular to a smart home joint control method and system based on data analysis. Background Technology

[0002] With the convergence of IoT and AI technologies, smart home systems are evolving from independent control of single devices to cross-device collaborative operation based on environmental perception and user habits. This joint control aims to provide users with a more comfortable, energy-efficient, and convenient overall living experience.

[0003] Existing smart home integrated control systems typically achieve automatic collaborative control of devices by pre-setting linkage rules or using machine learning algorithms to analyze historical data to build a user behavior and environmental adjustment model.

[0004] In existing technologies, the aforementioned methods heavily rely on static knowledge models formed through initial presets or historical learning, treating parameters such as device performance and home topology as fixed premises. However, in real-world home scenarios, device performance degrades, device locations may change, and network connectivity fluctuates, leading to a gradual mismatch between the static model and the dynamic real-world environment. Therefore, existing technologies suffer from static control models that struggle to adapt to the dynamic changes in the real-world home environment. They often issue invalid or even contradictory joint control commands due to outdated or erroneous models, ultimately resulting in automation failure, increased energy consumption, and a degraded user experience. Summary of the Invention

[0005] This application provides a data analysis-based smart home joint control method and system, which solves the problem in the prior art where static control models fail to adapt to dynamic changes in the home environment, such as equipment aging and relocation, leading to joint control failure. By constructing a dynamic self-verifying cognitive model, the smart home system can continuously maintain an accurate understanding of the real environment, thereby generating reliable and effective automated joint control commands.

[0006] This application provides a data analysis-based smart home joint control method, including: constructing a dynamic influence relationship hypergraph containing smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes;

[0007] Real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence relationship hypergraph and received external operation commands for smart home devices triggers model verification logic;

[0008] Based on the model validation logic triggering results, a virtual probe task is generated, and the expected impact propagation path of the virtual probe task in the smart home network is obtained.

[0009] Execute virtual probe tasks while simultaneously collecting actual response data;

[0010] The actual response data collected is compared and analyzed with the expected state in the dynamic impact relationship diagram, and then updated and corrected.

[0011] Based on the updated and corrected dynamic influence relationship hypergraph, the final joint control command is generated and sent to the smart home execution device.

[0012] Furthermore, the steps for constructing a dynamic influence hypergraph containing smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes include:

[0013] Iterate through all online hardware entities in the smart home network, create a corresponding data node for each smart home device, and initialize the basic attributes of each node;

[0014] Establish a four-dimensional connection edge architecture to describe the relationships between smart home devices: the first dimension is the physical influence connection edge, the second dimension is the logical collaboration connection edge, the third dimension is the resource competition connection edge, and the fourth dimension is the state mutual exclusion connection edge.

[0015] Assign a confidence attribute value to each established connection edge;

[0016] The smart home device nodes, four-dimensional connection edges, and their real-time confidence attributes are combined and stored to form a dynamic influence relationship hypergraph.

[0017] Furthermore, the steps for triggering model validation logic, including real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence hypergraph and received external operation commands for smart home devices, include:

[0018] Run a passive trigger monitoring mechanism. When it receives an external operation command initiated by a user for a smart home device, it parses the target smart home device chain driven by the command's intent.

[0019] Retrieve all physical influence connection edges involved in the target smart home device chain in the dynamic influence relationship hypergraph, and read the current confidence attribute value of these connection edges;

[0020] Determine whether there is any connection edge whose confidence attribute value is lower than the preset security trust threshold;

[0021] If it exists, the current smart home control request is determined to be in a low cognitive reliability state, the direct execution logic is blocked, and instead an accompanying verification request is triggered for the weak connection edge.

[0022] Run an active trigger monitoring mechanism to periodically scan the confidence attribute values ​​of all connected edges in the dynamic influence relationship hypergraph in the background;

[0023] Identify all connection edges whose confidence level falls below a preset aging threshold due to time decay, mark these failure edges as objects to be verified, and classify them according to the type of connection edge;

[0024] When the number of relationships between smart home devices of a certain type or within a certain area exceeds a preset aggregation threshold, a proactive batch verification request is generated.

[0025] Furthermore, the steps of generating virtual probe tasks and obtaining the expected propagation path of the virtual probe tasks' impact in the smart home network include:

[0026] The specific source of the trigger model verification logic is analyzed. If it is a passive trigger based on external operation instructions, an embedded micro probe task is constructed.

[0027] If it is an active trigger based on active scanning, then construct an aggregated probe task;

[0028] After generating the probe task, a breadth-first search is performed in the dynamic influence relationship hypergraph, starting from the source smart home device node that executes the probe task.

[0029] Based on the type and weight of the connection edge, predict the flow direction of the physical or logical signals generated by the probe action in the smart home network, and mark all downstream smart home device nodes that will be affected by the physical or logical signals as potential response nodes.

[0030] The set of connected paths from the source node to all potential response nodes is defined as the expected impact propagation path.

[0031] Furthermore, the steps for executing virtual probe tasks while simultaneously collecting actual response data include:

[0032] Send control commands to the smart home execution device specified by the probe task to drive the execution device to perform actions according to the preset timing, intensity and duration;

[0033] A wake-up command is sent to all smart home device nodes with sensing capabilities located along the expected propagation path, temporarily switching the operating mode of these devices to high-frequency sampling mode;

[0034] During the time window of the probe task execution, all activated smart home sensor components synchronously record the sequence of changes in home environment parameters;

[0035] The time axis alignment process is performed on the collected sequence of changes in home environment parameters;

[0036] This generates a panoramic data package containing the stimulus, main response, and secondary response.

[0037] Furthermore, the steps of comparing and analyzing the collected actual response data with the expected state in the dynamic influence relationship supergraph, and updating and correcting it, include:

[0038] Extract the smart home sensor response curves from the panoramic data package and perform cross-correlation analysis between them and the model prediction curves pre-stored in the dynamic influence relationship hypermap;

[0039] For each verified connection edge, calculate the goodness of fit between the actual response and the expected response;

[0040] If the goodness of fit is higher than the preset confirmation threshold, the confidence attribute value of the connection edge is restored to the maximum value, and the weight coefficient of the connection edge is finely adjusted according to the strength of the actual response.

[0041] If the goodness of fit is lower than the rejection threshold, the smart home device relationship described by the connection edge is determined to be invalid.

[0042] Analyze the unexpected related data in the panoramic data packet. If a smart home device node outside the expected propagation path shows a response feature that is strongly correlated with the probe action, it is determined that a potential new connection edge has been discovered.

[0043] In the dynamic influence hypergraph, a new hypothesis layer connection edge is established for this pair of smart home device nodes, and an initial confidence level is assigned.

[0044] Furthermore, based on the updated and corrected dynamic influence relationship hypergraph, the steps for generating the final joint control command and sending it to the smart home execution device include:

[0045] After obtaining the updated and corrected dynamic influence relationship hypergraph, the original home environment adjustment needs of the current users are reassessed.

[0046] An optimal path planning algorithm based on graph search is used to find the optimal solution to achieve the goal in the updated hypergraph;

[0047] The planned optimal operation sequence is converted into specific machine code and sent as the final joint control command to each smart home actuator.

[0048] Furthermore, the steps for calculating the goodness of fit between the actual response and the expected response include:

[0049] The goodness of fit between the actual response and the expected response is calculated using the formula for calculating the goodness of fit of the expected response.

[0050] The formula for calculating the goodness of fit of the expected response is:

[0051] ;

[0052] In the formula, For goodness of fit, For the sum of squared residuals, This is the total sum of squares.

[0053] This application provides a data analysis-based smart home joint control system, used to implement a data analysis-based smart home joint control method, including:

[0054] The module includes a relationship hypergraph construction module, a logic triggering module, a task generation module, a task execution module, an update and correction module, and an instruction sending module.

[0055] The relationship hypergraph construction module is used to construct a dynamic influence relationship hypergraph that includes smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes.

[0056] The logic triggering module is used to monitor in real time the changes in the confidence attribute of the connecting edges in the dynamic influence relationship hypergraph and the received external operation commands for smart home devices, and trigger the model verification logic.

[0057] The task generation module is used to generate virtual probe tasks based on the model verification logic triggering results, and to obtain the expected impact propagation path of the virtual probe tasks in the smart home network.

[0058] The task execution module is used to execute virtual probe tasks and simultaneously collect actual response data.

[0059] The update and correction module is used to compare and analyze the collected actual response data with the expected state in the dynamic influence relationship super graph, and then update and correct it.

[0060] The instruction sending module is used to generate the final joint control instruction based on the updated and corrected dynamic influence relationship hypergraph and send it to the smart home execution device.

[0061] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0062] By constructing a dynamic influence relationship hypergraph, the multi-dimensional relationships such as physical influence and logical coordination among home devices are modeled in the form of a credibility score. This enables control decisions to continuously perceive changes such as device performance degradation and location movement, and update internal cognition in a timely manner, effectively solving the problem of static models being out of touch with reality.

[0063] Furthermore, by introducing automated verification logic, when the confidence level of a relationship is detected to be below the threshold, whether it is due to user operation or periodic scanning, a targeted virtual probe task will be triggered to perform micro-tests with minimal interference, thereby discovering and correcting model deviations in advance and avoiding issuing invalid instructions based on erroneous models.

[0064] Furthermore, when performing probe tasks, the influence propagation path is predicted based on the supergraph, and all sensors along the path are activated to collect high-frequency data. By comparing the goodness of fit between the actual response and the expected state, not only can the target relationship be verified, but also unexpected device associations can be discovered, thereby enhancing the comprehensiveness and adaptability of the model. Attached Figure Description

[0065] Figure 1 A flowchart illustrating a data analysis-based smart home joint control method provided in this application embodiment;

[0066] Figure 2 This is a schematic diagram of a data analysis-based smart home integrated control system provided in an embodiment of this application. Detailed Implementation

[0067] This application provides a data analysis-based smart home joint control method and system, which solves the problem of inaccurate automation decisions caused by static knowledge models in the prior art due to their inability to continuously adapt to real-world changes such as device aging, location relocation, and network fluctuations. By constructing and dynamically maintaining a cognitive graph reflecting the multidimensional relationships between devices, and automatically executing micro-tests under the guidance of this graph to verify and correct the model, the understanding of the real situation of the home can be automatically kept accurate, thereby always outputting effective and reliable joint control commands and ensuring the long-term stability of automation effects.

[0068] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0069] like Figure 1 The diagram shown is a flowchart of a data analysis-based smart home joint control method provided in an embodiment of this application. The method is applied to a data analysis-based smart home joint control system and includes the following steps: constructing a dynamic influence relationship hypergraph containing smart home device nodes and multi-dimensional logical connection edges and their confidence attributes, as a basic cognitive model for smart home system decision-making;

[0070] Real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence relationship hypergraph and received external operation commands for smart home devices. When the confidence of a connected edge is lower than a preset threshold or the external operation command involves a low-confidence connected edge, the model verification logic is triggered.

[0071] Based on the model validation logic triggering results, a virtual probe task for a specific smart home device is generated in the dynamic influence relationship hypergraph, and the expected influence propagation path of the virtual probe task in the smart home network is obtained according to the connection structure of the dynamic influence relationship hypergraph.

[0072] Control relevant smart home devices to perform virtual probe tasks, while activating smart home sensor components located on the expected impact propagation path to collaboratively collect actual response data;

[0073] The actual response data collected is compared and analyzed with the expected state in the dynamic influence relationship hypergraph. Based on the degree of difference in comparison, the weight of the connection edge, topology and confidence attributes between smart home devices in the dynamic influence relationship hypergraph are updated and corrected.

[0074] Based on the updated and corrected dynamic influence relationship hypergraph, the smart home device scheduling strategy for the current home environment target is re-acquired, and the final joint control command is generated and sent to the smart home execution device.

[0075] Furthermore, the steps for constructing a dynamic influence hypergraph containing smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes include:

[0076] Iterate through all online hardware entities in the smart home network, create a corresponding data node for each smart home device, and initialize the basic attributes of each node;

[0077] A four-dimensional connection edge architecture is established to describe the relationships between smart home devices: The first dimension is the physical influence connection edge, which is used to map the causal path of changes in sensor readings of home environmental parameters caused by the actions of smart home devices; the second dimension is the logical collaboration connection edge, which is used to map the temporal linkage or combination relationship between smart home devices in specific home scene modes; the third dimension is the resource competition connection edge, which is used to map the mutual exclusion or constraint relationship between smart home devices that share home power load or home network bandwidth resources; the fourth dimension is the state mutual exclusion connection edge, which is used to map the physical or logical conflict relationship between the functional states of smart home devices.

[0078] Each established connection edge is assigned a dynamically changing confidence attribute value. This confidence attribute value is configured with time decay logic, that is, as the last verification timestamp moves forward, the confidence attribute value will automatically decrease according to a preset function curve, so as to characterize the system's confidence in the validity of the relationship between the smart home devices decreases over time.

[0079] The formula for calculating the confidence attribute value is as follows:

[0080] ;

[0081] In the formula, CS represents the confidence level of the connected edges at the current time. It is dimensionless and its value range is usually [0,1]. The initial confidence level for establishing the connection edge or after the last successful verification, usually set to 1, is a dimensionless scalar. This is a time decay function used to simulate "memory fuzziness," and its formula is: ,in This is the decay coefficient, measured in units of "1 / time" (e.g., 1 / day). This parameter is obtained through historical data analysis, such as determining the average time it takes for various relationships in a statistical system to begin showing signs of failure, and then taking its reciprocal as the decay coefficient. Reference value, This is the difference between the current time and the timestamp of the last successful verification of the connected edge, in time (e.g., days). The exponential decay model is a commonly used mathematical model to describe the decay of information value or memory reliability over time (such as radioactive decay and the forgetting curve). It ensures that the confidence level decreases smoothly over time and never returns to zero, but can infinitely approach 0. To verify the volatility factor, which is dimensionless, its formula is as follows: , This is an adjustment coefficient (0 ≤ α ≤ 1), used to control the impact of a single validation result on long-term confidence. It is based on an empirical preset, for example, set to 0.3. This represents the accuracy of the most recent validation, with a value ranging from [0,1]. For example, in validation, if the actual response perfectly matches the expected response, then... If there is no match at all, the value is 0. The value is obtained by comparing the actual sensor data curve with the model prediction curve and calculating the complement of the Pearson correlation coefficient or the normalized root mean square error.

[0082] The preset function curve is In the coordinate system, the horizontal axis represents the time difference. The vertical axis represents the attenuation factor. The curve starts from 1 ( (Time) begins, with As the curve increases, it first rapidly and then slowly approaches 0.

[0083] The smart home device nodes, four-dimensional connection edges, and their real-time confidence attributes are combined and stored to form a dynamic influence relationship hypergraph that can dynamically reflect the complex interactions between smart home devices, serving as the data source for all probe generation and control decisions in subsequent steps.

[0084] Furthermore, the steps for triggering model validation logic, including real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence hypergraph and received external operation commands for smart home devices, include:

[0085] Run a passive trigger monitoring mechanism. When it receives an external operation command initiated by a user for a smart home device, it parses the target smart home device chain driven by the command's intent.

[0086] Retrieve all physical influence connection edges involved in the target smart home device chain in the dynamic influence relationship hypergraph, and read the current confidence attribute value of these connection edges;

[0087] Determine whether there is any connection edge whose confidence attribute value is lower than the preset security trust threshold;

[0088] Security trust thresholds are usually determined empirically.

[0089] If it exists, the current smart home control request is determined to be in a low cognitive reliability state, the direct execution logic is blocked, and instead an accompanying verification request is triggered for the weak connection edge.

[0090] In this context, a low cognitive reliability state refers to a system's level of confidence in the accuracy of its internal knowledge model (specifically manifested as a single edge or set of connecting edges in a dynamic influence hypergraph) falling below the minimum level required for safe and reliable automated decision-making. In this state, directly implementing control based on this low-confidence model may lead to ineffective actions, wasted resources, or even conflicts.

[0091] For accompanying verification requests on weak connections: When a user operation necessitates the use of a low-confidence connection, the user request is not blocked. Instead, a small, diagnostic test task is synchronously embedded within the main task flow of executing the user request. For example, the user commands "turn on the air conditioner," but the "air conditioner-temperature sensor" edge has low confidence. While executing the regular cooling command, the system adds a small, specific power fluctuation or frequency change as a test signal and monitors the temperature sensor's response at high frequency. This test signal has a negligible impact on the main cooling function (non-intrusive), but is sufficient to verify whether the physical causal relationship still holds. This verification request occurs alongside the main task, efficiently acquiring verification data using real-world operational scenarios.

[0092] Run an active trigger monitoring mechanism to periodically scan the confidence attribute values ​​of all connected edges in the dynamic influence relationship hypergraph in the background;

[0093] Identify all connection edges whose confidence level falls below a preset aging threshold due to time decay, mark these failed edges as objects to be verified, and classify them according to the type of connection edge (physical, logical, resource, or mutual exclusion).

[0094] The purpose of marking failed edges as objects to be verified is to manage and optimize the model aging problem in a task-oriented manner. It collects these "cognitive vulnerabilities" that need to be verified so that the system can plan and execute efficient and non-intrusive verification tasks in a unified manner when resources are idle (such as at night or when no one is home), rather than triggering them temporarily and sporadically when users use the system, thereby maximizing verification efficiency.

[0095] When the number of relationships between smart home devices of a certain type or within a certain area exceeds a preset aggregation threshold, an active batch verification request is generated, thereby triggering the model verification logic during system idle periods to prevent the model from silently deviating from the real home environment.

[0096] Proactive batch verification requests involve the system packaging multiple low-confidence connection edges that are geographically adjacent or logically related into a single composite verification task. For example, a set of lights in a living room can be made to flash in a specific sequence at once, simultaneously verifying the responses of multiple light sensors, cameras, and power monitoring devices. This allows multiple relationships to be verified with a single probe action, improving verification efficiency.

[0097] Furthermore, the steps of generating virtual probe tasks and obtaining the expected propagation path of the virtual probe tasks' impact in the smart home network include:

[0098] The specific source of the trigger model verification logic is analyzed. If it is a passive trigger based on external operation commands, an embedded micro probe task is constructed. This task is designed to add a small, non-intrusive specific action feature to the action sequence of smart home devices that execute user commands. For example, inserting a wind speed fluctuation of a specific frequency as a signal marker during the startup process of a smart air conditioner.

[0099] Embedded micro probe task construction steps:

[0100] Intent parsing: Parses the core device chain that needs to be called for the current user command (such as "set the air conditioner to 26℃").

[0101] Weak point location: Locate the specific connection edge (such as "air conditioner compressor power - room temperature drop rate") on the device chain in the dynamic influence relationship hypergraph with a confidence level lower than a preset threshold.

[0102] Micro-motion design: Design a small-amplitude, short-duration characteristic motion superimposed on the normal control signal. For example, when the air conditioner compressor starts, apply an additional sinusoidal power modulation lasting 5 seconds with an amplitude of 5% of the rated power.

[0103] Signal injection: The timing and waveform parameters of the micro-actions are encoded and embedded as probe payloads into the normal control command stream sent to the air conditioner.

[0104] If it is an active trigger based on active scanning, then an aggregated probe task is constructed. This task selects a core smart home device action that can simultaneously stimulate multiple connection edges to be verified, such as controlling the smart main light in the living room to flash in a specific encoding pattern, so as to simultaneously test the light sensor response, camera light sensitivity and circuit load detection response.

[0105] Steps for building a clustered probe task:

[0106] Cluster analysis: Perform cluster analysis on the scanned low-confidence connection edges to find the set of edges that can be excited by the same device action (such as multiple edges that are related to the light output of the "living room main light").

[0107] Excitation source selection: Select the core device in this set (i.e., the "living room main light") as the probe actuator.

[0108] Encoding Pattern Design: Design an action code that can carry multi-dimensional information. For example, design a "bright-dark-bright" flashing pattern, where the duration of the bright and dark phases, the slope of the intensity change, and other parameters correspond to different relational dimensions that need to be verified (such as verifying the response delay of the light sensor or the overexposure recovery capability of the camera).

[0109] Task encapsulation: The designed coding pattern is encapsulated into an independent control task and scheduled to be executed during periods of low system load.

[0110] After generating probe tasks (including embedded micro-probe tasks and aggregated probe tasks), the breadth-first search algorithm starts from the source smart home device node that executes the probe task and performs a breadth-first search in the dynamic influence relationship hypergraph.

[0111] The search steps are as follows:

[0112] 1. Add the source device node that performed the probe task to the search queue and mark it as visited.

[0113] 2. Take a node from the queue and traverse all its unvisited neighbor nodes (nodes directly reachable by connecting edges) in the dynamic influence hypergraph.

[0114] 3. Add these neighboring nodes to the tail of the queue and mark them as visited.

[0115] 4. Repeat steps 2 and 3 until the queue is empty. All visited nodes and the edges they form together form a connected subgraph originating from the source node. This subgraph is the basis for the expected influence propagation path.

[0116] Based on the type and weight of the connection edge, predict the flow direction of the physical or logical signals generated by the probe action in the smart home network, and mark all downstream smart home device nodes that will be affected by the physical or logical signals as potential response nodes.

[0117] The steps for predicting the flow direction include:

[0118] For physical influence edges (such as light, heat, and sound), a simplified propagation model is established based on physical principles (such as inverse square attenuation and the heat conduction equation). By combining the weights of the connecting edges (which can be regarded as attenuation coefficients), the signal strength attenuation with "distance" (hop count or physical distance) is predicted, thereby determining the main direction of influence.

[0119] The steps to determine if a node is affected include: setting a detection threshold for each type of signal (physical or logical); acquiring the predicted intensity of the signal when it reaches each potential node while predicting the propagation path; and identifying a node as a potential responder if the predicted intensity is higher than its detection threshold. For example, if the predicted sound pressure level of a sound probe at a distance of 3 meters is 40 dB, and the sensitivity threshold of a microphone sensor is 35 dB, then the microphone is considered to be affected.

[0120] For logically cooperative edges or state-exclusive edges, symbolic reasoning is performed based on the logical rules represented by the edges (such as IF-THEN) to predict the propagation path of the logical signals.

[0121] The set of connected paths from the source node to all potential response nodes is defined as the expected impact propagation path. This path includes not only the direct target verification object, but also indirect smart home devices that may generate bypass responses, thus providing a path plan for subsequent comprehensive data harvesting.

[0122] Furthermore, the steps for executing virtual probe tasks while simultaneously collecting actual response data include:

[0123] The central control unit (the core decision-making and scheduling server or high-performance gateway device in the smart home joint control system) sends control commands to the smart home execution device specified by the probe task, driving the execution device to perform actions according to the preset timing, intensity and duration;

[0124] A wake-up command is sent to all smart home device nodes with sensing capabilities located along the expected propagation path, temporarily switching the operating mode of these devices to high-frequency sampling mode;

[0125] The steps to determine if a smart home device node possesses sensing capabilities include: During system initialization or device network access, each device reports its device type and a list of built-in sensors (following an IoT device description framework, such as the OMA / WM2M object model or a manufacturer-defined model) to the central control unit based on its hardware capabilities. The central control unit then establishes a device capability registry, clearly identifying which device nodes have sensors for temperature, light intensity, power meters, etc.

[0126] The working modes include: regular event-triggered or low-frequency periodic reporting mode, and temporary switching to a fixed high-frequency (e.g., 10 times per second) data acquisition and reporting mode.

[0127] During the probe task execution time window, all activated smart home sensor components (i.e., smart home devices switching working modes) synchronously record the sequence of changes in home environment parameters, including but not limited to temperature, humidity, light, power, sound decibels, and image change data.

[0128] The time-axis alignment of the collected home environment parameter change sequence is performed to ensure that the data streams of all smart home sensors are strictly synchronized with the execution time of the probe task, forming a panoramic data package containing "excitation-main response-secondary response".

[0129] Furthermore, the steps of comparing and analyzing the collected actual response data with the expected state in the dynamic influence relationship supergraph, and updating and correcting it, include:

[0130] Extract the smart home sensor response curves from the panoramic data package and perform cross-correlation analysis between them and the model prediction curves pre-stored in the dynamic influence relationship hypermap;

[0131] The steps for performing cross-correlation analysis include:

[0132] 1. Data preprocessing: processing the collected actual response data curves. And the model prediction curves pre-stored in the hypergraph Zero-mean processing (subtracting their respective average values) is performed to eliminate the influence of the DC component.

[0133] 2. Calculate the cross-correlation sequence: For a series of lag times ,calculate and The integral of the product (or summation for discrete data): Generate a statement about function .

[0134] In the above formula, In order to delay time The numerical value of the cross-relationship at the point is the product of the actual response and the predicted response (e.g., when the temperature unit is ℃, the product unit is (℃)²). This is the actual response time series after zero-mean normalization. After zero mean and lag Time-based models predict time series.

[0135] 3. Finding the peak: In Find the maximum value in a sequence The lag time corresponding to this peak. This reflects the delay in the actual response relative to the prediction.

[0136] 4. Normalization and Evaluation: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Dividing by the product of the standard deviations of the two sequences yields the normalized cross-correlation coefficient, which ranges from -1 to 1. The closer this coefficient is to 1, the more similar the shapes of the two curves are after accounting for time delays. This coefficient can be used as a measure of the accuracy of the most recent validation. One of the calculation bases.

[0137] For each verified connection edge, calculate the goodness of fit between the actual response and the expected response;

[0138] Verified connections refer to low-confidence connections that are used as verification targets when generating virtual probe tasks. The probe task is designed to stimulate the relationships described by these edges.

[0139] If the goodness of fit is higher than the preset confirmation threshold, the confidence attribute value of the connection edge is restored to the maximum value, and the weight coefficient of the connection edge is finely adjusted according to the strength of the actual response to make it more consistent with the physical performance of the current smart home device.

[0140] If the goodness of fit is lower than the rejection threshold, the smart home device relationship described by the connection edge is determined to be invalid. Its confidence is then significantly reduced, and a secondary diagnostic process is triggered or the edge is marked as blocked in the hypergraph.

[0141] The confirmation threshold and rejection threshold are preset based on the principle of hypothesis testing.

[0142] By using anomaly detection algorithms to analyze unexpected related data in panoramic data packets, if a smart home device node outside the expected propagation path is found to have response characteristics strongly correlated with probe actions, it is determined that a potential new connection edge has been discovered.

[0143] The existing technique of combining mutation detection and correlation analysis is used to determine the characteristics of a strong correlation response. The specific steps are as follows:

[0144] 1. Background noise estimation: Before the probe mission is executed, the baseline noise level of the target edge device sensor readings is analyzed, and its mean and standard deviation are calculated. ).

[0145] 2. Sudden Change Detection: Within the probe's action time window, detect whether the sensor reading significantly exceeds the baseline noise range (e.g., change amplitude > 3). The abrupt change point (step, impulse, or trend reversal) of ).

[0146] 3. Temporal alignment and correlation calculation: If a mutation can be detected, the mutation signal is aligned with the timing of the probe action. The Pearson correlation coefficient between the two is obtained.

[0147] 4. Threshold Judgment: If the correlation coefficient exceeds a high preset threshold (e.g., 0.7 or 0.8), and the delay between the mutation point and the probe action is within a reasonable physical range (e.g., sound propagation delay), then it is determined that a "response feature strongly correlated with the probe action" has occurred. This indicates that there is a physical or logical connection between the edge device and the probe source device that has not been recorded in the hypergraph.

[0148] In the dynamic influence hypergraph, a new hypothesis layer connection edge is established for this pair of smart home device nodes, and an initial confidence level is assigned.

[0149] This mechanism not only repairs known relationship parameters but also enables the automatic discovery and topology growth of implicit relationships between smart home devices (such as new physical interference or unrecorded linkage effects).

[0150] Furthermore, based on the updated and corrected dynamic influence relationship hypergraph, the steps for generating the final joint control command and sending it to the smart home execution device include:

[0151] After obtaining the updated and corrected dynamic influence relationship hypergraph, reassess the current user's original home environment adjustment needs (such as "adjust the room temperature to 26 degrees").

[0152] The scheduling algorithm no longer uses preset static rules, but adopts a graph search-based optimal path planning algorithm to find the optimal solution to achieve the goal in the updated hypergraph;

[0153] During the path search process, the algorithm automatically avoids connection edges with low confidence attribute values ​​or marked as "blocked", thereby avoiding the use of smart home device paths with severe performance degradation or damage.

[0154] The algorithm prioritizes strong connections with high weighting coefficients and recent probe verification to ensure that control commands can be transmitted to home environment terminals through the most reliable physical or logical links.

[0155] If it is found that the home environment goal cannot be achieved through the existing high-confidence connection edge, an anomaly report is generated to prompt the user to connect, instead of forcibly executing invalid automation logic;

[0156] The planned optimal operation sequence is converted into specific machine code, which is then sent to each smart home actuator as the final joint control command, thereby achieving precise control.

[0157] Planning the optimal sequence of operations involves solving a constrained optimization problem on the updated dynamic influence hypergraph using a graph search algorithm. The specific steps include:

[0158] Problem modeling: Transform user requirements (e.g., "room temperature 26℃") into target states. Treat device nodes in the hypergraph as state variables, and connect edges, their weights, and confidence levels as the "actions" and "costs / reliability" of state transitions.

[0159] Path search: A shortest path algorithm using confidence weighting. The algorithm starts from the current environmental state and explores all possible paths to the target state through a series of device actions (such as "turn on the air conditioner" or "close the window").

[0160] Cost calculation: The "cost" of each path is the sum of the cost functions of all connected edges traversed. The cost function is negatively correlated with the confidence level (the lower the confidence level, the higher the cost).

[0161] Sequence generation: The algorithm ultimately finds a path from the initial state to the target state with the minimum total cost. The list of device actions executed sequentially along this path is the planned optimal operation sequence. This sequence is then compiled into a specific sequence of device control commands and issued.

[0162] Furthermore, the steps for calculating the goodness of fit between the actual response and the expected response include:

[0163] The goodness of fit between the actual response and the expected response is calculated using the formula for calculating the goodness of fit of the expected response.

[0164] The formula for calculating the goodness of fit of the expected response is:

[0165] ;

[0166] In the formula, For goodness of fit, dimensionless, ranging from [0,1]. The sum of squared residuals is calculated using the following formula: .in These are the actual response data points, derived from the actual sensor data stream collected after the probe is executed. The expected response data point for the model comes from the dynamic influence hypergraph. It represents the predicted data stream calculated by the mathematical model (e.g., a differential equation describing "air conditioner power-temperature change") associated with the connected edge under probe input excitation, expressed in units of the square of the sensor reading. The total sum of squares is calculated using the following formula: .in It is the average value of the actual response data. The unit is also the square of the sensor reading.

[0167] like Figure 2The diagram shown is a structural schematic of a data analysis-based smart home joint control system provided in this application embodiment. The data analysis-based smart home joint control system provided in this application embodiment includes: a relationship hypergraph construction module, a logic triggering module, a task generation module, a task execution module, an update and correction module, and an instruction sending module.

[0168] The relationship hypergraph construction module is used to construct a dynamic influence relationship hypergraph that includes smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes.

[0169] The logic triggering module is used to monitor in real time the changes in the confidence attribute of the connecting edges in the dynamic influence relationship hypergraph and the received external operation commands for smart home devices, and trigger the model verification logic.

[0170] The task generation module is used to generate virtual probe tasks based on the model verification logic triggering results, and to obtain the expected impact propagation path of the virtual probe tasks in the smart home network.

[0171] The task execution module is used to execute virtual probe tasks and simultaneously collect actual response data.

[0172] The update and correction module is used to compare and analyze the collected actual response data with the expected state in the dynamic influence relationship super graph, and then update and correct it.

[0173] The instruction sending module is used to generate the final joint control instruction based on the updated and corrected dynamic influence relationship hypergraph and send it to the smart home execution device.

[0174] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0175] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0176] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0177] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0178] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0179] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart home joint control method based on data analysis, characterized in that, Includes the following steps: Construct a dynamic influence relationship hypergraph that includes smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes; Real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence relationship hypergraph and received external operation commands for smart home devices triggers model verification logic; Based on the model validation logic triggering results, a virtual probe task is generated, and the expected impact propagation path of the virtual probe task in the smart home network is obtained. Execute virtual probe tasks while simultaneously collecting actual response data; The actual response data collected is compared and analyzed with the expected state in the dynamic impact relationship diagram, and then updated and corrected. Based on the updated and corrected dynamic influence relationship hypergraph, the final joint control command is generated and sent to the smart home execution device.

2. The smart home joint control method based on data analysis as described in claim 1, characterized in that, The steps to construct a dynamic influence hypergraph containing smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes include: Iterate through all online hardware entities in the smart home network, create a corresponding data node for each smart home device, and initialize the basic attributes of each node; Establish a four-dimensional connection edge architecture to describe the relationships between smart home devices: the first dimension is the physical influence connection edge, the second dimension is the logical collaboration connection edge, the third dimension is the resource competition connection edge, and the fourth dimension is the state mutual exclusion connection edge. Assign a confidence attribute value to each established connection edge; The smart home device nodes, four-dimensional connection edges, and their real-time confidence attributes are combined and stored to form a dynamic influence relationship hypergraph.

3. The smart home joint control method based on data analysis as described in claim 1, characterized in that, The steps for triggering model validation logic, which involve real-time monitoring of changes in the confidence attribute of connected edges in the dynamic influence hypergraph and receiving external operation commands for smart home devices, include: Run a passive trigger monitoring mechanism. When it receives an external operation command initiated by a user for a smart home device, it parses the target smart home device chain driven by the command's intent. Retrieve all physical influence connection edges involved in the target smart home device chain in the dynamic influence relationship hypergraph, and read the current confidence attribute value of these connection edges; Determine whether there is any connection edge whose confidence attribute value is lower than the preset security trust threshold; If it exists, the current smart home control request is determined to be in a low cognitive reliability state, the direct execution logic is blocked, and instead an accompanying verification request is triggered for the weak connection edge. Run an active trigger monitoring mechanism to periodically scan the confidence attribute values ​​of all connected edges in the dynamic influence relationship hypergraph in the background; Identify all connection edges whose confidence level falls below a preset aging threshold due to time decay, mark these failure edges as objects to be verified, and classify them according to the type of connection edge; When the number of relationships between smart home devices of a certain type or within a certain area exceeds a preset aggregation threshold, a proactive batch verification request is generated.

4. The smart home joint control method based on data analysis as described in claim 1, characterized in that, The steps for generating a virtual probe task and obtaining the expected propagation path of the virtual probe task's impact in the smart home network include: The specific source of the trigger model verification logic is analyzed. If it is a passive trigger based on external operation instructions, an embedded micro probe task is constructed. If it is an active trigger based on active scanning, then construct an aggregated probe task; After generating the probe task, a breadth-first search is performed in the dynamic influence relationship hypergraph, starting from the source smart home device node that executes the probe task. Based on the type and weight of the connection edge, predict the flow direction of the physical or logical signals generated by the probe action in the smart home network, and mark all downstream smart home device nodes that will be affected by the physical or logical signals as potential response nodes. The set of connected paths from the source node to all potential response nodes is defined as the expected impact propagation path.

5. The smart home joint control method based on data analysis as described in claim 1, characterized in that, The steps for executing a virtual probe task while simultaneously collecting actual response data include: Send control commands to the smart home execution device specified by the probe task to drive the execution device to perform actions according to the preset timing, intensity and duration; A wake-up command is sent to all smart home device nodes with sensing capabilities located along the expected propagation path, temporarily switching the operating mode of these devices to high-frequency sampling mode; During the time window of the probe task execution, all activated smart home sensor components synchronously record the sequence of changes in home environment parameters; The time axis alignment process is performed on the collected sequence of changes in home environment parameters; This generates a panoramic data package containing the stimulus, main response, and secondary response.

6. The smart home joint control method based on data analysis as described in claim 1, characterized in that, The steps for comparing and analyzing the collected actual response data with the expected state in the dynamic impact relationship diagram, and then updating and correcting it, include: Extract the smart home sensor response curves from the panoramic data package and perform cross-correlation analysis between them and the model prediction curves pre-stored in the dynamic influence relationship hypermap; For each verified connection edge, calculate the goodness of fit between the actual response and the expected response; If the goodness of fit is higher than the preset confirmation threshold, the confidence attribute value of the connection edge is restored to the maximum value, and the weight coefficient of the connection edge is finely adjusted according to the strength of the actual response. If the goodness of fit is lower than the rejection threshold, the smart home device relationship described by the connection edge is determined to be invalid. Analyze the unexpected related data in the panoramic data packet. If a smart home device node outside the expected propagation path shows a response feature that is strongly correlated with the probe action, it is determined that a potential new connection edge has been discovered. In the dynamic influence hypergraph, a new hypothesis layer connection edge is established for this pair of smart home device nodes, and an initial confidence level is assigned.

7. The smart home joint control method based on data analysis as described in claim 1, characterized in that, Based on the updated and corrected dynamic influence hypergraph, the steps for generating the final joint control command and sending it to the smart home execution device include: After obtaining the updated and corrected dynamic influence relationship hypergraph, the original home environment adjustment needs of the current users are reassessed. An optimal path planning algorithm based on graph search is used to find the optimal solution to achieve the goal in the updated hypergraph; The planned optimal operation sequence is converted into specific machine code and sent as the final joint control command to each smart home actuator.

8. The smart home joint control method based on data analysis as described in claim 6, characterized in that, The steps for calculating the goodness of fit between the actual response and the expected response include: The goodness of fit between the actual response and the expected response is calculated using the formula for calculating the goodness of fit of the expected response. The formula for calculating the goodness of fit of the expected response is: ； In the formula, For goodness of fit, For the sum of squared residuals, This is the total sum of squares.

9. A data analysis-based smart home joint control system, used to implement the data analysis-based smart home joint control method according to any one of claims 1-8, characterized in that, include: The module includes a relationship hypergraph construction module, a logic triggering module, a task generation module, a task execution module, an update and correction module, and an instruction sending module. The relationship hypergraph construction module is used to construct a dynamic influence relationship hypergraph that includes smart home device nodes, multi-dimensional logical connection edges, and their confidence attributes. The logic triggering module is used to monitor in real time the changes in the confidence attribute of the connecting edges in the dynamic influence relationship hypergraph and the received external operation commands for smart home devices, and trigger the model verification logic. The task generation module is used to generate virtual probe tasks based on the model verification logic triggering results, and to obtain the expected impact propagation path of the virtual probe tasks in the smart home network. The task execution module is used to execute virtual probe tasks and simultaneously collect actual response data. The update and correction module is used to compare and analyze the collected actual response data with the expected state in the dynamic influence relationship super graph, and then update and correct it. The instruction sending module is used to generate the final joint control instruction based on the updated and corrected dynamic influence relationship hypergraph and send it to the smart home execution device.

Images