Smart home equipment state monitoring and abnormity early warning method and system

Abstract

The invention discloses a smart home equipment state monitoring and abnormity early warning method and system, and relates to the technical field of smart home equipment state monitoring and abnormity early warning, and the method comprises the steps: constructing a dynamic space-time topological graph between home equipment; monitoring state data of each node device in the topological graph in real time; identifying a source abnormal node in the topological graph based on the state data and a preset abnormal rule; by taking the source abnormal node as input, executing fault propagation simulation deduction in combination with the direction and the weight of the edge to obtain a risk equipment node set and a fault propagation path; and based on the set and the propagation path, generating diagnosis early warning information used for positioning a fault root cause and representing a propagation link. According to the invention, through dynamic topology modeling, equipment physical connection and a logic dependency relationship are uniformly represented, root causes can be quickly locked and potential affected equipment can be predicted when multiple equipment are abnormal at the same time, prospective early warning of hidden cascading failures is realized, and operation and maintenance efficiency and system reliability are improved through a visual report.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent home equipment state monitoring and abnormality early warning, in particular to an intelligent home equipment state monitoring and abnormality early warning method and system. BACKGROUND

[0002] With the rapid development of Internet of Things, cloud computing and edge computing technologies, the number and types of terminal devices connected to intelligent home systems are increasing, covering lighting, appliance control, security protection, environmental perception, energy management and other subsystems. Existing intelligent home platforms usually collect device operation state data through sensors and perform data aggregation and remote management with the help of gateways and cloud platforms. On this basis, the industry gradually introduces threshold alarms, rule engines, time series-based anomaly detection and some machine learning models to identify single-device operation anomalies and simple alarm push. At the same time, multi-device linkage control and scene automation configuration are increasingly popular, making the logical dependencies and functional coupling relationships between devices increasingly complex, and raising higher requirements for intelligent and systematic device state monitoring and abnormality early warning.

[0003] However, existing intelligent home equipment state monitoring and abnormality early warning technologies are mostly still based on a single-device-centric local perspective, and their anomaly recognition often relies on fixed thresholds or simple statistical features, lacking unified modeling of the physical connection relationships between multiple devices, energy and communication links, and functional call dependencies, making it difficult to build dynamic spatiotemporal topology maps that reflect the real structure of home environments. When facing multiple simultaneous or sequential anomalies, existing solutions can usually only perform joint alarms based on time sequence or simple correlation, failing to distinguish between source abnormal nodes and dependent abnormal nodes generated by their transmission, resulting in ambiguous root cause localization, redundant alarm information, and an inability to effectively depict the propagation links and impact range of faults in the system. In addition, existing technologies generally lack the ability to predictively model fault propagation based on the topology relationships and edge weights between devices, making it impossible to estimate and predict the risk of potentially affected downstream devices when only local minor anomalies occur, and unable to output structured diagnostic and early warning information at the system level, let alone intuitively present fault root causes, critical propagation paths and nodes of different risk levels in a unified topology view. SUMMARY

[0004] In view of the above problems, the present application is proposed.

[0005] Therefore, the present application solves the technical problems that the existing smart home device state monitoring and abnormal early warning method lacks unified space-time topology modeling of physical connection and logical dependency relationship between devices in a home environment, cannot accurately identify source abnormal nodes and distinguish root cause failures from chain propagation failures, cannot perform forward-looking fault propagation risk assessment and systematic diagnosis and early warning based on the topology structure between devices, and how to realize correlation analysis, propagation deduction and visual diagnosis and early warning of multi-device implicit failures on a dynamic space-time topology map.

[0006] To solve the above technical problems, the present application provides the following technical solutions. In a first aspect, an embodiment of the present application provides a smart home device state monitoring and abnormal early warning method, comprising: constructing a dynamic space-time topology map between devices in a home environment; monitoring devices corresponding to each node in the dynamic space-time topology map in real time to obtain running state data; identifying source abnormal nodes in the dynamic space-time topology map based on the running state data and a preset abnormality determination rule; inputting the source abnormal nodes, combining the directions and weights of edges in the dynamic space-time topology map, and executing a fault propagation simulation deduction algorithm to output a set of affected risk device nodes and a fault propagation path; generating systematic diagnosis and early warning information for locating fault root causes and representing propagation links based on the set of risk device nodes and the fault propagation path.

[0007] As a preferred scheme of the smart home device state monitoring and abnormal early warning method of the present application, the construction of the dynamic space-time topology map between devices in the home environment comprises: defining nodes in the dynamic space-time topology map based on the unique identifier, type and physical location information of the device; establishing edges connecting the nodes based on the physical connection relationship and logical dependency relationship between devices; wherein the physical connection relationship represents an entity link of energy and communication, and the logical dependency relationship represents a constraint condition of function call and control; allocating weights to the edges; the weights are determined based on device attributes, connection relationships and historical data; dynamically adjusting the nodes, edges and weights of the dynamic space-time topology map in response to device network entry, network exit, state change and dependency relationship update.

[0008] As a preferred scheme of the smart home device state monitoring and abnormal early warning method of the present application, the obtaining of the running state data comprises: Obtaining running state data corresponding to each device node in the dynamic spatiotemporal topology graph, the running state data including at least one of power consumption, temperature, communication signal strength, and device response time; According to the type of the device node, the connection relationship in the topology graph, and the historical failure rate of the device, the monitoring frequency and priority of the running state data are dynamically adjusted. The running state data is associated with and stored in the corresponding device node in the topology graph.

[0009] As a preferred scheme of the intelligent home device state monitoring and abnormality early warning method, the source abnormal node in the topology graph is identified, including: Based on the preset abnormality determination rule, the running state data is analyzed, and if the running state data exceeds the dynamic threshold range determined based on the device type or the historical running state, the corresponding device node is determined as a candidate abnormal node; Based on the dynamic spatiotemporal topology graph, the running state data of the adjacent node directly connected to the candidate abnormal node is obtained for collaborative verification; If the collaborative verification confirms that the abnormality of the running state data originates from the candidate abnormal node itself rather than being caused by abnormality transmission of the adjacent node, the candidate abnormal node is identified as the source abnormal node.

[0010] As a preferred scheme of the intelligent home device state monitoring and abnormality early warning method, the failure propagation simulation deduction algorithm is a graph calculation model based on the dynamic spatiotemporal topology graph, and the graph calculation model is configured to: Model the device nodes, the directions, and the weights of the edges in the dynamic spatiotemporal topology graph as a directed weighted graph structure; Receive a failure propagation analysis request starting from the source abnormal node, and perform risk assessment calculation on the reachable device nodes in the directed weighted graph to identify potential risk device nodes and corresponding propagation paths.

[0011] As a preferred scheme of the intelligent home device state monitoring and abnormality early warning method, the output of the set of affected risk device nodes and the failure propagation path includes: Starting from the source abnormal node, traverse along the directed edges in the dynamic spatiotemporal topology graph to determine all device nodes reachable from the source abnormal node; For each reachable device node, according to the weights of the edges on the traversal path from the source abnormal node to the device node, calculate the risk assessment value of failure propagation to the device node; comparing the risk assessment value with a preset propagation risk threshold, and including in the risk device node set a reachable device node whose risk assessment value exceeds the threshold; Based on the traversal path and the risk device node set, a fault propagation path from the source abnormal node to each device node in the risk device node set is generated.

[0012] As a preferred scheme of the intelligent home device state monitoring and abnormality early warning method, the generating of the systematic diagnostic early warning information for locating the fault root cause and representing the propagation link comprises: Integrating the source abnormal node, the risk device node set, the fault propagation path, and the risk assessment value corresponding to each node forms structured diagnostic data. According to the structured diagnostic data, the devices in the risk device node set are classified and sorted according to risk levels. Based on the structured diagnostic data, a structured diagnostic report containing fault root cause node identification, a risk device node list and its risk level, and a key fault propagation path is generated. The structured diagnostic report is presented in a visual alarm interface in a user terminal, and the visual alarm interface takes the dynamic space-time topology graph as a base map, and the fault propagation path and the device nodes of different risk levels are displayed differently.

[0013] In a second aspect, the embodiments of the present application provide an intelligent home device state monitoring and abnormality early warning system, comprising: A dynamic space-time topology graph construction module constructs a dynamic space-time topology graph among devices in a home environment. An operating state monitoring module monitors the devices corresponding to each node in the dynamic space-time topology graph in real time to obtain operating state data. A source abnormal node identification module identifies a source abnormal node in the topology graph based on the operating state data and a preset abnormality determination rule. A fault propagation deduction module takes the source abnormal node as input, combines the direction and weight of the edges in the topology graph, executes a fault propagation simulation deduction algorithm, and outputs a risk device node set affected and a fault propagation path. A systematic diagnostic early warning generation module generates systematic diagnostic early warning information for locating a fault root cause and representing a propagation link based on the risk device node set and the fault propagation path.

[0014] The application has the beneficial effects that the application realizes systematic modeling of physical connection and logical dependency relationship between devices by constructing a dynamic space-time topology atlas of home devices. Based on the atlas, not only can multi-source monitoring data be fused to accurately locate fault sources (source abnormal nodes), but also can the propagation path of the fault along the topology be simulated to realize implicit risk early warning of associated devices. Finally, the system generates a visual diagnostic report integrating root cause, influence range, risk level and propagation link, thereby upgrading the traditional fragmented alarm to systematic decision support that can guide disposal, and significantly improving the reliability, priori of early warning and operation and maintenance efficiency of the smart home system. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings. Figure 1 The application provides an intelligent home device state monitoring and abnormal early warning method. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned purposes, features and advantages of the application more apparent and easy to understand, the specific embodiments of the application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should be within the protection scope of the application.

[0017] Embodiment 1, refer to Figure 1 For an embodiment of the application, an intelligent home device state monitoring and abnormal early warning method is provided, comprising: S1: constructing a dynamic space-time topology atlas of devices in a home environment.

[0018] defining nodes in the dynamic space-time topology atlas based on unique identification, type and physical location information of the devices; establishing edges connecting the nodes based on physical connection relationship and logical dependency relationship between the devices; wherein the physical connection relationship represents entity links of energy and communication, and the logical dependency relationship represents constraint conditions of function call and control; allocating weights for the edges; the weights are determined based on device attributes, connection relationship and historical data; in response to device network entry, network exit, state change and dependency relationship update, dynamically adjusting the nodes, edges and weights of the dynamic space-time topology atlas.

[0019] It should be noted that the "unique identifier" can be the serial number of the device, the MAC address or the logical ID assigned by the gateway, which is used to ensure that each node in the topology map corresponds to a physical device one by one, avoiding device confusion in subsequent association analysis and propagation calculation; the "type" is used to distinguish lighting, motor, security monitoring, sensor and other functional devices, so that different abnormal threshold and propagation weight models can be used according to the type in the subsequent; the "physical location information" can be in the form of room number, spatial coordinates or floor / area marker, which is used to introduce the spatial dimension in the topology graph, so as to consider the physical proximity in the subsequent fault deduction, for example, the linkage risk of devices in the same power distribution circuit or the same room is usually significantly higher than that of remote devices.

[0020] In the construction of the edge, the "physical connection relationship" is limited to the power supply line, the communication link (such as bus connection, routing access relationship) and other entity connections, which can directly reflect the objective path of the fault conduction on the energy link or the communication link; the "logical dependency relationship" is extracted through scene linkage rules, automation scripts or control strategies, such as "magnetic door trigger - light lighting", "smoke alarm - air conditioning off" and other logical call links, which are used to depict the situation that even if there is no direct physical connection, the control logic can also cause the indirect propagation of state abnormalities between devices. By uniformly representing the physical connection and the logical dependency in the same topology, the subsequent propagation analysis can cover both "hard-wired chain faults" and "control logic caused linkage abnormalities", solving the problem of incomplete topology caused by traditional solutions focusing on only one of them.

[0021] The weight of the edge can be defined as a real value between 0 and 1, which is obtained by considering the device attributes (such as rated power, criticality), connection relationship (such as whether it is the same circuit, whether it is a single-point dependency) and historical data (such as historical fault co-occurrence frequency, correlation index, etc.). For example, based on the probability of the simultaneous or sequential occurrence of abnormalities of two devices in the historical operation log, combined with whether they share the power distribution branch and whether there is a mandatory control logic, the factors are normalized and summed or output by a machine learning model to obtain the edge weight. The weight obtained in this way not only has interpretability, but also provides a quantitative basis for subsequent graph calculation: the higher the weight, the greater the possibility and impact of fault propagation along the edge, so that risk attenuation and accumulation can be realized in the propagation deduction, rather than a simple "adjacent" or not coarse-grained judgment.

[0022] In the aspect of dynamic adjustment, by listening to device access protocols (such as network access / exit events), device self-reported capabilities (such as role changes, working mode switching), and user operations of adding, deleting, and modifying scene logic, the nodes, edges, and their weights in the topology are updated in real time, ensuring that the topology structure is consistent with the actual configuration of the home environment. Unlike existing static configurations or manually maintained topologies, this dynamic spatiotemporal topology map can reflect the evolution of the system structure over time, providing a reliable "spatiotemporal foundation" for subsequent time window-based anomaly detection and fault propagation deduction. Thus, the S1 step not only provides an implementable data structure and maintenance mechanism, but also fundamentally solves the problem of the lack of formal and computable representation of multi-device relationships in existing technologies, which makes it difficult to carry out system-level fault analysis. This provides a necessary prior structure and computable basis for subsequent source abnormal node identification and propagation simulation.

[0023] S2: Real-time monitoring of each node in the dynamic spatiotemporal topology map to obtain running state data.

[0024] Obtain running state data corresponding to each device node in the dynamic spatiotemporal topology map, the running state data including at least one of power consumption, temperature, communication signal strength, and device response time; According to the type of the device node, the connection relationship in the topology map, and the historical failure rate of the device, dynamically adjust the monitoring frequency and priority of the running state data; Associate and store the running state data with the corresponding device node in the topology map.

[0025] It should be noted that the running state data is uniformly collected by a monitoring agent or a central gateway deployed in the home network, and is obtained using a standardized Internet of Things communication method of polling or event-driven reporting. Each piece of data is attached with an accurate time stamp and a unique identifier of the source device node, forming a "device-time-state" triple basic data unit.

[0026] The selection of operational status data is based on the consideration of the coverage of the main failure modes of the device. The power consumption data can directly reflect the energy conversion state of the device, and its abnormal increase may indicate motor stall, circuit short circuit or standby leakage, while its abnormal decrease may mean that the device is unexpectedly powered off or enters an unexpected low-power mode. The temperature parameter represents the thermodynamic state of the device, and the continuous or instantaneous abnormal temperature rise is a key physical symptom of electronic component overload, heat dissipation channel blockage or environmental cooling failure. The communication signal strength (such as received signal strength indication RSSI, link quality indication LQI) quantifies the wireless link quality between the device and the network, and its attenuation trend can be used to diagnose the implicit degradation of the communication link caused by physical obstruction, distance increase or co-frequency interference. The device response time is obtained by measuring the time delay from the issuance of the control command to the receipt of the device state confirmation message, and the abnormally prolonged response time usually reveals that the device microprocessor is loaded saturated, the software process is blocked or the communication protocol stack processing is abnormal. Through continuous monitoring of at least one of the above parameters, the system can construct the health status portrait of the device from the multiple orthogonal dimensions of energy flow, thermodynamics, information transmission and control performance, and provide a multi-dimensional data basis with physical explainability for subsequent anomaly judgment.

[0027] The strategy of dynamically adjusting the monitoring frequency and priority is to allocate limited edge computing and communication resources to the device nodes with the highest risk of failure or the widest impact. This strategy is based on a multi-dimensional decision model: first, according to the type of the device, a basic weight is given, for example, safety devices such as smoke alarms and gas valves, and infrastructure devices such as home gateways and main routers are classified as high priority. Secondly, analyze the connection relationship of the node in the topology map, calculate its "structural criticality", for example, a smart socket that supplies power to multiple downstream devices, its degree is high and it is upstream in the energy path, its structural criticality is also relatively high. Finally, the historical failure rate based on time window statistics is integrated, and the nodes with recent frequent anomalies are monitored intensively. In a specific embodiment, the system can set a basic monitoring interval of 300 seconds for ordinary nodes, shorten it to 60 seconds for nodes with high priority or structural criticality higher than the threshold, and further shorten the monitoring interval to 30 seconds for nodes that meet both high priority and recent failure rate exceeding the standard, and place their data stream at the front end of the processing queue. This differentiated resource scheduling mechanism significantly improves the monitoring density of key risk points and the timeliness of anomaly capture under the premise that the total load of the system remains basically unchanged.

[0028] The association and storage of the running state data and the topology atlas are realized by establishing a bidirectional index of time series data and graph structure data. Specifically, all the time-stamped running state data is stored in a time series database with a device unique identifier as the key, forming an independent state history sequence for each device. At the same time, the dynamic space-time topology atlas maintained in the graph database includes a reference to the corresponding data sequence of the device in the time series database. The technical effect of this storage architecture is that it realizes the fusion of data dimensions: when individual historical behavior of the device is needed for trend analysis, time series data can be efficiently queried; when system-level analysis based on the correlation of devices is needed (such as the cooperative verification of step S4), the relevant node set can be quickly retrieved through the graph database, and the state sequence of these nodes in the same period can be obtained through indexing and concurrent access. This provides a unified data foundation for subsequent steps that can efficiently perform cross-device and cross-time correlation queries, and is a key engineering foundation for supporting the invention to realize the diagnosis from "point" to "surface".

[0029] S3: identifying a source abnormal node in the dynamic space-time topology atlas based on the running state data and a preset abnormality determination rule.

[0030] Based on the preset abnormality determination rule, the running state data is analyzed, and if the running state data exceeds the dynamic threshold range determined based on the device type or the historical running state, the corresponding device node is determined as a candidate abnormal node; Based on the dynamic space-time topology atlas, the running state data of the adjacent nodes directly connected with the candidate abnormal node is obtained for cooperative verification; If the cooperative verification confirms that the abnormality of the running state data originates from the candidate abnormal node itself, rather than being caused by the abnormality of the adjacent nodes, the candidate abnormal node is identified as the source abnormal node.

[0031] It should be noted that the "preset anomaly determination rule" includes a dynamic threshold model and an anomaly score model established for different device types and different operating state parameters. Specifically, the system, based on the historical operating state data stored in association in step S2, calculates the parameter distribution characteristics of each device node, such as mean, variance, and quantile range, within a sliding time window (such as 24 hours or one week), and identifies the seasonal periodic pattern. In combination with the safety specifications and design indicators of the device type, a dynamic threshold interval that is adaptive to the context (such as the working mode and the environmental cycle) is generated. For example, for the power consumption of the intelligent air conditioner compressor node, a threshold model can be established according to the historical data in different modes such as refrigeration, heating, and standby; for the temperature of the router node, a joint threshold surface can be constructed in combination with the real-time CPU load. When the newly collected data points or short sequences continuously exceed the corresponding dynamic threshold range, the system calculates the anomaly deviation degree as the anomaly score, and if the score exceeds the preset trigger condition, the device node is marked as a candidate abnormal node.

[0032] After becoming a candidate abnormal node, the system immediately starts the topology-based collaborative verification. This process first uses the dynamic spatiotemporal topology graph to obtain a set of adjacent nodes directly connected to the candidate node through physical connection edges or logical dependency edges. The system retrieves the operating state data of these adjacent nodes within a time window aligned with or slightly earlier than the abnormal period of the candidate node, and performs multidimensional comparison analysis. The core logic of the analysis is based on the causality and time characteristics of fault propagation: if the abnormality of the candidate node A is caused by its upstream node B, the abnormal state of B should generally occur earlier than or simultaneously with the abnormality of A, and the abnormal patterns of the two have reasonability on the topology propagation path.

[0033] For example, when detecting that the power consumption of a certain lighting device node abnormally drops, the system checks the upstream intelligent socket node and other devices on the same power supply branch. If it is found that the upstream socket node also has power consumption zero or is offline at the same time, and the abnormal occurrence time of the socket node is not later than that of the lamp node, it can be inferred that the abnormality is more likely to be caused by the common cause of upstream power interruption rather than the lamp itself. On the contrary, if only the lamp node shows a specific abnormal mode (such as a power consumption pulse corresponding to a specific flicker), and the states of its adjacent nodes (including the socket) are normal, it indicates that the abnormality has local independence. The collaborative verification process can further calculate the time correlation and conditional probability between abnormal events to quantify whether the abnormality is "spontaneous" or "conducted".

[0034] Through the above analysis, the system performs the traceability judgment: only when the abnormality of the candidate abnormal node cannot be explained by the states and topology relationships of its adjacent nodes, i.e., the abnormality is ahead of time or independent of its topological context in logic, is it finally identified as a source abnormal node.

[0035] The beneficial effect of this step is that it uses the system structure prior knowledge provided by the topology map to upgrade the traditional isolated point anomaly detection to context-based causal inference. This can effectively avoid misreporting a large number of affected downstream devices as independent fault sources in the scenarios of power grid flashover, network broadcast storm and other systematic disturbances, thereby effectively reducing false positives and focusing on the true fault starting point. The accurately identified source abnormal node provides a unique and reliable starting point for subsequent fault propagation simulation, ensuring the accuracy and operability of systematic early warning.

[0036] S4: Taking the source abnormal node as input, combining the direction and weight of the edge in the dynamic space-time topology map, executing a fault propagation simulation deduction algorithm, and outputting a set of affected risk device nodes and a fault propagation path.

[0037] Taking the source abnormal node as a starting point, traversing along the directed edges in the dynamic space-time topology map to determine all device nodes reachable from the source abnormal node; For each reachable device node, according to the weight of each edge on the traversal path from the source abnormal node to the device node, calculating a risk assessment value of fault propagation to the device node; Comparing the risk assessment value with a preset propagation risk threshold, and including the reachable device node whose risk assessment value exceeds the threshold in the set of risk device nodes; Based on the traversal path and the set of risk device nodes, generating a fault propagation path from the source abnormal node to each device node in the set of risk device nodes.

[0038] It should be noted that the fault propagation simulation deduction algorithm is a graph computing model based on the dynamic space-time topology map, and the graph computing model is configured to: Model the direction and weight of the device nodes and edges in the dynamic space-time topology map as a directed weighted graph structure; Receive a fault propagation analysis request starting from a source abnormal node, and perform risk assessment calculation on the reachable device nodes in the directed weighted graph to identify potential risk device nodes and corresponding propagation paths.

[0039] Further, the core of the fault propagation simulation algorithm lies in the transformation of the dynamic spatiotemporal topology graph constructed in step S1 into a directed weighted graph model that can be used for risk quantification. In this model, the direction of an edge defines the potential direction of fault influence (e.g., from a power supply device to a powered device, from a master controller to a controlled device); the weight of an edge quantifies the strength or likelihood of fault propagation along the edge, which is based on the device attributes, connection relationships, and historical data determined in step S1. The algorithm receives the source abnormal node output by step S3 as the only starting point for the simulation, simulating the process of fault spreading from the "epicenter" to the outside of the topology network.

[0040] The simulation starts from the source abnormal node and performs graph traversal along the directed edges (e.g., using a breadth-first search algorithm) to determine the set of reachable device nodes that may be affected in the topology structure. The traversal follows the direction of the edges to ensure that the simulation conforms to the physical and logical causal constraints. To control the computational complexity and focus on high-correlation risks, the traversal process can set a maximum depth or prune when the cumulative propagation strength falls below a certain threshold, avoiding unnecessary exploration of too far or extremely weak transmission paths.

[0041] For each reachable device node, the algorithm needs to calculate its risk assessment value. This calculation is essentially a process of attenuating and superimposing the "risk" of the source abnormal node along the propagation path. A typical implementation is to multiply the weights of all edges on each reachable path from the source abnormal node to the target node (or calculate the joint probability based on a probabilistic model), and the result represents the "path transmission strength" of the complete fault conduction along that specific path to the target node. A path length attenuation factor can be introduced in the calculation to reflect the reality that the propagation uncertainty increases with the number of hops. If there are multiple paths, take the maximum value or weighted sum of the path transmission strengths as the final risk assessment value of the target node. The greater the weight of the edge (close to 1), the smaller the conduction resistance and the higher the risk assessment value; conversely, the smaller the weight (close to 0), the greater the conduction attenuation.

[0042] The calculated risk assessment value will be compared with a preset propagation risk threshold. This threshold can be set and adjusted according to the overall risk tolerance of the family to device failures. If the risk assessment value of a node exceeds this threshold, it indicates that the node has a higher risk of being affected by the failure of the source abnormal node, and thus it is included in the set of risk device nodes.

[0043] Finally, the system generates one or more explicit fault propagation paths from the source abnormal node to each node in the set of risk device nodes based on the graph paths recorded during the traversal process. These paths not only list the affected devices, but also reveal the specific causal links on which the risk propagation depends (e.g., "abnormal smart socket → wireless router → smart TV"). This step completes the deduction from a single-point anomaly to a systematic risk network, and the output of the set of risk device nodes and fault propagation paths accurately depicts the scope of potential failure and the diffusion context.

[0044] The beneficial effect of this step is that it realizes the leap from "identifying an abnormality that has occurred" to "predicting a risk that has not occurred". By using the constructed and quantified topology model for simulation deduction, the system can warn of the risks that the associated devices may face in advance before they show any data anomalies, thereby realizing truly preventive maintenance. This fundamentally solves the "implicit chain failure warning problem" in the background art and upgrades the status monitoring of smart homes from passive alarm to active prediction to a new stage.

[0045] S5: Based on the set of risk device nodes and the fault propagation paths, generate systematic diagnostic warning information for locating fault causes and characterizing propagation links.

[0046] Integrate the source abnormal node, the set of risk device nodes, the fault propagation paths, and the risk assessment values corresponding to each node to form structured diagnostic data; According to the structured diagnostic data, divide and sort the devices in the set of risk device nodes by risk level; Based on the structured diagnostic data, generate a structured diagnostic report containing fault cause node identification, a list of risk device nodes and their risk levels, and key fault propagation paths; Present the structured diagnostic report in a visual alarm interface in the user terminal, with the dynamic spatiotemporal topology graph as the base map and differentiating display of the fault propagation paths and device nodes of different risk levels.

[0047] It should be noted that this step is the final integration and expression of all the aforementioned analysis and deduction results, and its core is to convert the system risks generated by the algorithm and implied in the data into systematic diagnostic warning information that can be directly understood and acted upon by users or maintenance personnel.

[0048] Firstly, the system integrates the outputs of steps S3 and S4 to form structured diagnostic data. The data unifies and correlates the root cause of the fault (source abnormal node), potential impact range (risk equipment node set), causal chain of risk propagation (fault propagation path), and quantitative degree of risk (risk assessment value corresponding to each node). In a specific embodiment, the data structure can be organized in JSON or protocol buffer format, including diagnostic ID, timestamp, source abnormal node details (ID, type, location), risk node list (each node contains ID, type, risk assessment value, and path ID), and detailed path sequence (sequentially recorded node ID and edge weight). This organization not only facilitates storage and retrieval, but also provides a basis for subsequent statistical analysis (such as high-frequency risk path mining).

[0049] Based on the above structured data, the system classifies and sorts the risk equipment nodes in the risk equipment node set. The classification logic can be based on the quantitative risk assessment value. In one embodiment, the system normalizes the risk assessment value to the interval [0, 1] and presets multiple level thresholds (e.g., 0.3, 0.6, 0.8) to divide the risk nodes into discrete levels such as "low risk", "medium risk", "high risk", and "extremely high risk". At the same time, for critical equipment types related to safety (such as gas and door locks) or infrastructure (such as gateways and main routers), their assessment values can be adjusted to reflect their business importance. When sorting, the system prioritizes the risk levels from high to low, and within the same level, it can further sort based on the risk assessment value or the structural importance of the node in the topology, thereby generating a clear processing priority list to guide users to focus their limited attention on the most critical risks.

[0050] Subsequently, the system generates a structured diagnostic report. This report is not a simple log list, but a diagnostic summary with standard fields that can be directly used for decision-making. Its core fields must include: fault root cause node identification (clear problem source), risk equipment node list grouped by risk level (clear impact range and severity), and key fault propagation path (revealing the main chain of fault diffusion). The report can also extend to include abnormal timeline, risk trend prediction based on historical data, and preliminary handling suggestions (such as "check the power supply of socket X" and "restart router Y"). This report is the crystallization of the system's intelligent analysis, which connects scattered abnormal points into a clear "fault impact network", fundamentally solving the problems of "alarm storm" and information fragmentation in traditional solutions.

[0051] Ultimately, to achieve optimal human-computer interaction and situational awareness, the system transforms the aforementioned structured diagnostic report into a visual alarm interface, presented on user terminals (such as mobile apps or home central touchscreens). This interface uses a dynamic spatiotemporal topology map as its visual base, intuitively recreating the layout and connection relationships of home devices. Based on this, the interface overlays information through differentiated visual encoding: prominent markers (such as flashing red icons) highlight source anomaly nodes; icons of different colors or sizes distinguish risk device nodes with different risk levels; and arrows or highlighted lines clearly depict key fault propagation paths. Through this interface, users can immediately grasp "where the fault started," "which devices might be affected," and "how the risk spreads," significantly shortening the time from receiving alarms to understanding the overall situation and making decisions.

[0052] The beneficial effect of this step lies in resolving the core contradictions of "information overload" and "insufficient insight" that have long existed in the smart home field. This invention, through integration, analysis, deduction, and visualization, ultimately outputs a unified, causally clear, and actionable "system health diagnosis report." This not only significantly improves the understandability and operability of warning information, transforming users from passive alarm receivers to proactive system managers, but also embodies the complete closed-loop value of the entire invention from perception and analysis to decision support, achieving a fundamental leap from "discrete device status monitoring" to "holistic home system health management."

[0053] Example 2 is the second embodiment of the present invention, which differs from the previous embodiment in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art or the current technical solution, can be embodied in the form of a software product. This current computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0054] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of executable instructions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical), and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be later executed by a computer. In this context, a "computer-readable medium" can be any means that can store the program for use by or in connection with the instruction execution system, apparatus, or device.

[0055] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical), and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be later executed by a computer.

[0056] Embodiment 3, as an embodiment of the present application, provides a smart home device state monitoring and abnormality early warning system, comprising a dynamic space-time topology graph construction module, a running state monitoring module, a source abnormal node identification module, a fault propagation deduction module and a systematic diagnosis early warning generation module.

[0057] The dynamic space-time topology graph construction module: constructs a dynamic space-time topology graph among devices in a home environment; The running state monitoring module: monitors in real time devices corresponding to each node in the dynamic space-time topology graph to obtain running state data; The source abnormal node identification module: identifies a source abnormal node in the topology graph based on the running state data and a preset abnormality determination rule; The fault propagation deduction module: takes the source abnormal node as input, combines the direction and weight of edges in the topology graph, executes a fault propagation simulation deduction algorithm, and outputs a set of affected risk device nodes and a fault propagation path; The systematic diagnosis early warning generation module: generates systematic diagnosis early warning information for locating a fault root cause and representing a propagation link based on the set of risk device nodes and the fault propagation path.

[0058] Embodiment 4, as an embodiment of the present application, provides a smart home device state monitoring and abnormality early warning method. In order to verify the beneficial effects of the present application, economic benefit calculation and simulation / contrast experiments are carried out for scientific demonstration.

[0059] In a simulated typical smart home environment, the present embodiment implements the method described in the present application by constructing a reproducible composite fault scenario, and systematically compares the technical effects of the method with those of the traditional single-point threshold early warning scheme, so as to comprehensively verify the innovation and superiority of the present application.

[0060] The test environment is a standard three-bedroom one-living room house, which is deployed with 22 smart devices covering four categories of lighting, environmental regulation, security and network infrastructure. To implement the present application, the system first constructs a dynamic spatiotemporal topology graph: all devices are automatically discovered through the home gateway, recording their MAC addresses, types (such as variable frequency air conditioners, smart sockets, main routers) and installation locations; the physical connections and logical dependencies between devices are automatically inferred through analysis of the network ARP table, smart socket energy consumption and scene rules of the home automation platform, forming an initial graph. In the graph, the direction of the edge represents the direction of the energy or control flow, and the weight is pre-assigned based on the connection type and historical data analysis (for example, the power supply edge weight from the living room smart socket to the living room air conditioner is 0.92, and to the living room main light is 0.88, indicating a strong conduction relationship; the logical edge weight from the door magnetic sensor to the living room main light based on the scene rule is 0.45, indicating a weak conduction relationship). The graph is set to dynamic mode and can respond to device online / offline events.

[0061] In the real-time monitoring stage, the system implements differentiated data collection strategies according to the types and topological importance of the nodes in the graph. For nodes on the critical power supply path (such as the living room smart socket and the main router) and security devices (such as the smoke alarm), a high-frequency sampling period of 30 seconds is set; for ordinary load devices (such as the bedroom light), a low-frequency period of 300 seconds is used. The collected operating state data includes power consumption, temperature, communication signal strength and control response time, all of which are associated with the graph node ID and stored with millisecond-level timestamps.

[0062] The test simulates a core fault scenario: the living room main smart socket circuit produces intermittent voltage fluctuations and microsecond-level transient power failure due to internal contact failure. This fault is a typical "upstream hidden fault". During the simulated evening high-load period (air conditioner and washing machine running simultaneously), the fault triggers a series of downstream device abnormalities: the living room air conditioner compressor records three abnormal restarts due to voltage drop (power consumption drops from 985W to 0W and then recovers); the home main router appears two short offline (packet loss rate rises from 0.5% to 100%) due to being connected to the same circuit; the living room main light observes a short flicker (current abnormal fluctuation).

[0063] The source abnormal node identification process is as follows: the monitoring system first marks the living room air conditioner and the main router as candidate abnormal nodes based on the dynamic threshold model of each device itself (for example, the power consumption dynamic threshold of the living room air conditioner is set to [850W, 1150W] based on its last week of the same mode operation data). Then, the system starts the collaborative verification: query the topology map to get the direct upstream nodes of these candidate nodes - the living room smart socket. Backtracking analysis found that about 2.1 seconds before the first abnormal restart of the living room air conditioner, the living room smart socket had continuously appeared multiple voltage drop events exceeding the normal range (fluctuation rate > 10%), and its time sequence completely covered the occurrence time of all downstream abnormalities. According to the causal logic of "upstream power supply abnormality preceding and possibly causing downstream multiple device abnormalities", the system accurately determines the living room smart socket as the source abnormal node, rather than misjudging the air conditioner or router as independent faults.

[0064] Then, the system takes the living room smart socket as the starting point to perform fault propagation simulation deduction. The algorithm traverses along the directed power supply edges in the graph, and integrates the edge weights and path length attenuation factors to calculate the risk assessment values of downstream nodes. The deduction result shows that the risk assessment values of the living room air conditioner and the home main router are 0.89 and 0.76 respectively, far exceeding the preset risk threshold of 0.65, and are marked as "high risk"; the risk assessment value of the living room main lamp is 0.58, which is marked as "medium risk"; and the risk values of other devices not on this power supply loop (such as the bedroom air conditioner and the kitchen refrigerator) are all below 0.15, which are determined as "low risk" or "no risk". At the same time, the system outputs the specific propagation path from the source socket to each risk node.

[0065] Finally, the system generates systematic diagnosis and warning information. Integrating all the above information, a structured diagnosis report is generated, the core content of which includes: "Fault root cause: poor contact or overload of the living room south wall smart socket circuit"; "High-risk devices: living room air conditioner (risk assessment 0.89), home main router (risk assessment 0.76)"; "Medium-risk device: living room main lamp (risk assessment 0.58)"; and "Key propagation path: smart socket → living room air conditioner; smart socket → home main router". The report is presented through a visual interface, with a family floor plan as the base map, clearly highlighting the source abnormal socket, and showing the risk propagation links with different colors and thicknesses, while providing "suggested immediate inspection or replacement of the smart socket" disposal guidance.

[0066] To quantitatively evaluate the effect, a traditional "single device independent threshold alarm" scheme is run in parallel as a control under the same environment and fault scenario as above. In the traditional scheme, the alarm threshold of each device is set based on the fixed recommended value in the device manual. In the repeated test of 7 consecutive days, the key data is recorded as shown in Tables 1 and 2: Table 1: Comparison of key fault event data

[0067] Table 2: Comparison of systematic performance indicators (7-day test period average)

[0068] The comparison test data fully reveals the substantial progress and creative effect of the present application compared with the prior art.

[0069] First of all, a fundamental breakthrough has been achieved in diagnostic accuracy and efficiency. Table 1 clearly shows a key fact: the traditional scheme completely failed to identify the true source of the fault (the living room smart socket), with a voltage fluctuation rate as high as 15.7% without triggering an alarm. Because the fixed threshold usually only focuses on overvoltage or undervoltage, it ignores the fluctuation rate, a key indicator of poor contact. On the contrary, it triggered alarms for the three downstream devices (air conditioner, router, main light) and incorrectly identified them as three independent root causes ("self-fault", "network fault", "drive fault"), as shown in Table 2, which resulted in up to 8 root cause positioning errors and an average positioning time of 187 seconds. The present application successfully traces the source of these concurrent anomalies to the only common upstream node through dynamic spatiotemporal topological mapping and collaborative verification mechanism, reducing root cause positioning errors to 0 times and shortening the time to 58 seconds. This directly proves that the present application effectively solves the core defect of the prior art that "root cause positioning is difficult and inaccurate".

[0070] Secondly, a leap from nothing to something has been achieved in the predictability of risk warning. The comparison of the "implicit risk identification rate" indicator (Table 2) is most significant. The 36.2% identification rate of the traditional scheme means that it can only alarm after the device parameters have exceeded the boundary, which is a post-response. The 87.4% identification rate of the present application is due to the fault propagation simulation algorithm. After the living room smart socket is identified as the source abnormal node, the algorithm immediately simulates and calculates the high-risk state of nodes such as the living room air conditioner and main router according to the topology and weight (risk assessment values 0.89, 0.76 in Table 1), and issues a warning before the device's own sensor triggers the traditional alarm. This is the main reason why the number of "chain failure unprewarning occurrences" in Table 2 is reduced from 11 to 2. The present application achieves active prediction of "implicit chain failure", changing the maintenance mode from "after-the-fact remedy" to "before-the-fact prevention".

[0071] Finally, the system optimization is realized in information integration and operation efficiency. Due to the lack of device correlation model, the traditional scheme will cause multiple downstream devices to issue independent alarms in the face of a single upstream fault, resulting in "alarm storm", as shown in Table 2, 23 invalid / repeated alarms per day. This not only wastes operation resources, but also seriously interferes with fault judgment. The present application generates a structured diagnostic report containing root cause, risk level and propagation path, and compresses multiple correlated alarms into one system diagnostic information, reducing invalid alarms by 74%. At the same time, clear visualization and accurate root cause positioning enable operation personnel to avoid blind troubleshooting of multiple devices, reducing the number of daily manual interventions from 9 to 3, and shortening the average fault downtime by 58%. This proves that the present application effectively solves the problem of "low operation efficiency caused by fragmented early warning information" in the prior art by providing systematic and operable diagnostic insights.

[0072] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit it. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, which should be covered by the scope of the claims of the present application.

Claims

1. A method for monitoring the status and providing early warning of anomalies in smart home devices, characterized in that, include: Constructing a dynamic spatiotemporal topology map of devices in a home environment; Real-time monitoring of devices corresponding to each node in the dynamic spatiotemporal topology map to obtain operational status data; Based on the aforementioned operational status data and preset anomaly determination rules, source anomaly nodes in the dynamic spatiotemporal topology map are identified. Using the source abnormal node as input, and combining the direction and weight of the edges in the dynamic spatiotemporal topology graph, a fault propagation simulation and deduction algorithm is executed to output the set of affected risky equipment nodes and the fault propagation path; Based on the set of risky device nodes and the fault propagation path, systematic diagnostic and early warning information is generated to locate the root cause of the fault and characterize the propagation link.

2. The method for monitoring and alerting to anomalies in smart home devices as described in claim 1, characterized in that, The construction of the dynamic spatiotemporal topology map between devices in the home environment includes: Nodes in the dynamic spatiotemporal topology map are defined based on the device's unique identifier, type, and physical location information. Based on the physical connection relationships and logical dependencies between devices, edges connecting the nodes are established; wherein, the physical connection relationship represents the physical link of energy and communication, and the logical dependency relationship represents the constraint conditions of function call and control. Assign weights to the edges; the weights are determined based on device attributes, connectivity, and historical data. In response to device network access, network decommissioning, status changes, and dependency updates, the nodes, edges, and weights of the dynamic spatiotemporal topology graph are dynamically adjusted.

3. The method for monitoring and warning of abnormalities in smart home devices as described in claim 2, characterized in that, The acquisition of operational status data includes: Obtain the operating status data corresponding to each device node in the dynamic spatiotemporal topology map. The operating status data includes at least one of power consumption, temperature, communication signal strength, and device response time. Based on the type of device nodes, the connection relationships in the topology map, and the historical failure rate of the devices, the monitoring frequency and priority of the operating status data are dynamically adjusted. The operational status data is associated with and stored with the corresponding device nodes in the topology map.

4. The method for monitoring and alerting to anomalies in smart home devices as described in claim 3, characterized in that, The identification of source anomaly nodes in the topology map includes: Based on preset anomaly determination rules, the operating status data is analyzed. If the operating status data exceeds the dynamic threshold range determined based on device type or historical operating status, the corresponding device node is determined as a candidate anomaly node. Based on the dynamic spatiotemporal topology map, the running status data of adjacent nodes that are directly connected to the candidate abnormal node are obtained for collaborative verification. If the collaborative verification confirms that the anomaly in the running status data originates from the candidate anomaly node itself, rather than from the anomaly propagation of adjacent nodes, then the candidate anomaly node is identified as the source anomaly node.

5. The method for monitoring and alerting to abnormalities in smart home devices as described in claim 4, characterized in that, The fault propagation simulation and deduction algorithm is a graph computation model based on a dynamic spatiotemporal topological graph, and the graph computation model is configured as follows: The orientation and weight of device nodes and edges in the dynamic spatiotemporal topology graph are modeled as a directed weighted graph structure; The system receives a fault propagation analysis request starting from the source abnormal node and performs risk assessment calculations on the reachable device nodes in the directed weighted graph to identify potential risky device nodes and their corresponding propagation paths.

6. The method for monitoring and alerting to anomalies in smart home devices as described in claim 5, characterized in that, The set of risky device nodes affected by the output and the fault propagation path include: Starting from the source anomalous node, traverse along the directed edges in the dynamic spatiotemporal topology graph to determine all device nodes reachable from the source anomalous node. For each reachable device node, the risk assessment value of fault propagation to the device node is calculated based on the weight of each edge on the traversal path from the source fault node to the device node. The risk assessment value is compared with a preset propagation risk threshold, and reachable device nodes whose risk assessment values ​​exceed the threshold are included in the risk device node set. Based on the traversal path and the set of risky device nodes, a fault propagation path is generated from the source abnormal node to each device node in the set of risky device nodes.

7. The method for monitoring and warning of abnormalities in smart home devices as described in claim 6, characterized in that, The systemic diagnostic and early warning information generated for locating the root cause of the fault and characterizing the propagation path includes: The source anomaly node, the set of risky device nodes, the fault propagation path, and the risk assessment value corresponding to each node are integrated to form structured diagnostic data; Based on the structured diagnostic data, the devices in the risk device node set are classified and sorted according to their risk levels. Based on the structured diagnostic data, a structured diagnostic report is generated, which includes fault root cause node identifiers, a list of risky device nodes and their risk levels, and key fault propagation paths. The structured diagnostic report is presented in the user terminal as a visual alarm interface. The visual alarm interface uses the dynamic spatiotemporal topology map as the base map and distinguishes and displays the fault propagation path and equipment nodes with different risk levels.

8. A smart home device status monitoring and anomaly early warning system, used to implement the smart home device status monitoring and anomaly early warning method as described in any one of claims 1 to 7, characterized in that, include: Dynamic Spatiotemporal Topology Graph Construction Module: Constructs a dynamic spatiotemporal topology graph between devices in a home environment; Operational status monitoring module: Real-time monitoring of devices corresponding to each node in the dynamic spatiotemporal topology map to obtain operational status data; Source anomaly node identification module: Based on the running status data and preset anomaly judgment rules, identify source anomaly nodes in the topology map; Fault propagation simulation module: Taking the source abnormal node as input, and combining the direction and weight of the edge in the topology graph, executes the fault propagation simulation algorithm to output the set of affected risky equipment nodes and the fault propagation path; Systematic Diagnostic Early Warning Generation Module: Based on the set of risky device nodes and the fault propagation path, it generates systematic diagnostic early warning information for locating the root cause of the fault and characterizing the propagation link.

Images