Heating device remote monitoring method and system combined with internet of things

Abstract

The application provides a heating device remote monitoring method and system combined with the Internet of Things, which first acquires real-time running state data and environment-related data of the heating device connected with the Internet of Things, performs cross-dimension correlation modeling based on a pre-trained analysis model, and generates a state correlation graph containing correlation rules. Then, the state correlation graph is reversely traced to analyze and locate the abnormal source potentially affecting the running stability of the heating device. Then, the pre-set control parameter adaptation library is queried according to the potential abnormal source to determine the matched target control parameter combination. Finally, the target control parameter combination is packaged as a standardized remote control instruction and sent to the Internet of Things control module of the heating device through the Internet of Things communication link to adjust the running parameters. The application realizes comprehensive perception, accurate analysis and effective control of the running state of the heating device, and improves the equipment running efficiency and stability.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and more specifically, to a method and system for remote monitoring of heating devices that incorporates the Internet of Things. Background Technology

[0002] Heating devices are widely used in industrial production and civilian equipment, such as industrial boilers and household heaters. Traditional monitoring of these devices relies primarily on manual on-site inspections. Staff must periodically visit the equipment to observe instrument readings and inspect its appearance to obtain operational status information. This method is not only labor-intensive and time-consuming but also fails to provide real-time monitoring of equipment operation and timely detection of potential problems. With the development of IoT technology, some heating devices have achieved simple remote data acquisition functions, obtaining some basic operating parameters. However, existing technologies can only monitor single operational status data of the heating device, failing to comprehensively consider the impact of environmental factors such as ambient temperature and humidity on equipment operation. Furthermore, the lack of in-depth analysis and correlation modeling of the collected data makes it difficult to uncover the intrinsic connections between operational status data and environmental data, and to accurately locate potential sources of anomalies affecting the stability of the heating device's operation. Consequently, effective remote control is difficult to achieve, limiting the operating efficiency and stability of the heating device. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for remote monitoring of a heating device incorporating the Internet of Things, the method comprising: Acquire real-time operating status data of IoT-connected heating devices and environmental data of the environment in which the heating devices are located. The real-time operating status data reflects the operating conditions of the heating devices, and the environmental data reflects the environmental conditions around the heating devices. Based on the pre-trained analysis model, cross-dimensional correlation modeling is performed on real-time operation status data and environmental correlation data to mine the inherent correlation between real-time operation status data and environmental correlation data and generate a state correlation graph containing correlation rules. By performing reverse source analysis on the operating status of the heating device through state association graph, potential anomalies that may affect the operating stability of the heating device can be located. The potential anomalies correspond to the association rules in the state association graph. Based on the query of the preset control parameter adaptation library for potential anomalies, the target control parameter combination that matches the potential anomaly is determined. The target control parameter combination can adapt to the current operating status and environmental conditions of the heating device. The target control parameters are packaged into standardized remote control commands, which are sent to the IoT control module of the heating device via an IoT communication link. The IoT control module then executes the remote control commands to adjust the operating parameters of the heating device.

[0004] In another aspect, embodiments of the present invention also provide a remote monitoring system for heating devices that incorporates the Internet of Things, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0005] Based on the above, this invention achieves comprehensive perception of the heating device's operating conditions and surrounding environment by acquiring real-time operating status data of the IoT-connected heating device and environmental correlation data of its surrounding environment. Cross-dimensional correlation modeling is performed based on a pre-trained analysis model to deeply explore the inherent correlation between real-time operating status data and environmental correlation data, generating a state correlation graph containing correlation rules, providing strong evidence for accurate analysis of device operation. Reverse tracing analysis of the heating device's operating status through the state correlation graph can quickly locate potential anomalies that may affect operational stability, improving the efficiency and accuracy of fault diagnosis. Based on the potential anomaly sources, a preset control parameter adaptation library is queried to determine the appropriate target control parameter combination, which is then encapsulated into standardized remote control commands and sent to the heating device, achieving precise adjustment of the heating device's operating parameters. This method forms a complete closed loop from data acquisition, correlation analysis, anomaly location to remote control, effectively improving the operating efficiency and stability of the heating device, reducing maintenance costs, and demonstrating high innovation and practicality. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the execution flow of the remote monitoring method for heating devices combined with the Internet of Things provided in the embodiments of the present invention.

[0007] Figure 2 This is a schematic diagram of exemplary hardware and software components of a remote monitoring system for heating devices that integrates the Internet of Things, provided in an embodiment of the present invention. Detailed Implementation

[0008] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a remote monitoring method for a heating device integrated with the Internet of Things (IoT) according to an embodiment of the present invention. The following is a detailed description of this remote monitoring method for a heating device integrated with the IoT.

[0009] Step S110: Obtain real-time operating status data of the IoT-connected heating device and environmental correlation data of the environment in which the heating device is located. The real-time operating status data reflects the operating condition of the heating device, and the environmental correlation data reflects the environmental conditions around the heating device.

[0010] In this embodiment, an industrial pipeline electric heating device is selected as the application scenario. This heating device is used to maintain a preset temperature of the fluid inside the pipeline. Its IoT control module integrates multiple sensors to collect data in real time. Real-time operating status data includes attributes such as the heating device's output power, operating current, operating voltage, pipeline inlet fluid temperature, pipeline outlet fluid temperature, heating element surface temperature, cumulative operating time, number of start-stop cycles, and fault codes. Each attribute corresponds to continuous or discrete time-series data, and the data collection frequency is a fixed period. Environmental correlation data includes attributes such as ambient temperature, ambient humidity, ambient air pressure, dust concentration in the air, external wind speed of the pipeline, external vibration amplitude of the pipeline, voltage fluctuation amplitude of the power supply line, and frequency fluctuation amplitude of the power supply line. The data collection frequency is consistent with the real-time operating status data. During the data collection process, for privacy-sensitive data, such as fluctuation data of power supply lines (which may be related to the company's power consumption layout), differential privacy technology is used for processing. Specifically, noise that conforms to the Laplace distribution is added to the original data. The magnitude of the noise addition is determined according to the preset privacy budget to ensure that the data after adding noise can still retain the trend characteristics of the original data, while the original data cannot be restored by reverse calculation, thus preventing data leakage.

[0011] Step S120: Based on the pre-trained analysis model, perform cross-dimensional correlation modeling processing on real-time operating status data and environmental correlation data, mine the inherent correlation between real-time operating status data and environmental correlation data, and generate a state correlation graph containing correlation rules.

[0012] In this embodiment, the pre-trained analysis model is a hybrid model based on graph neural networks and attention mechanisms. The model input is a time series feature matrix of real-time running state data and environmental correlation data, and the output is a structured state correlation graph.

[0013] Step S121: Input the real-time operating status data and environmental correlation data into the feature input layer of the analysis model, perform feature channel alignment processing on the real-time operating status data and environmental correlation data to keep the feature dimensions of the real-time operating status data and environmental correlation data consistent, and form a dual-channel input feature stream with a unified dimension.

[0014] In this embodiment, both the real-time operational status data and the environmental correlation data have multiple original feature dimensions. The feature input layer first standardizes the data for each attribute by mapping the value of each attribute to the same numerical range, eliminating the influence of different units. Next, the feature input layer expands the feature dimensions of the standardized real-time operational status data by extracting statistical features from the time-series data of each attribute. These statistical features include mean, variance, maximum, minimum, median, and slope, with each attribute corresponding to multiple statistical features, thus expanding the feature dimensions of the real-time operational status data to match those of the environmental correlation data. Simultaneously, the feature input layer adjusts the feature dimensions of the standardized environmental correlation data by merging the statistical features of some attribute items. The merging rule involves weighted summation of highly correlated statistical features, with the weighting coefficients determined based on the physical meaning of the attribute items, ensuring that the feature dimensions of the environmental correlation data are consistent with those of the expanded real-time operational status data. Finally, a unified dual-input feature stream is formed, with one stream being the real-time operational status feature stream and the other being the environmental correlation feature stream.

[0015] Step S122: Perform collaborative correlation analysis on the dual-input feature streams through the correlation mining layer of the analysis model, extract the common change features of real-time running status data and environmental correlation data in the time dimension and attribute dimension, and generate cross-dimensional correlation feature vectors.

[0016] In this embodiment, the association mining layer includes a time series processing module, a collaborative attention module, a feature interaction module, and a feature refinement module.

[0017] Step S1221: In the association mining layer, time series trend extraction processing is performed on the dual input feature streams respectively, retaining the core change features of the data, filtering out instantaneous fluctuation interference, and obtaining a stable real-time running state feature stream and a stable environmental association feature stream.

[0018] In this embodiment, the time series trend extraction process employs a sliding window filtering method. The length of the sliding window is determined based on the data acquisition frequency and fluctuation characteristics. For each attribute item's time series data in the real-time running status feature stream, the sliding window sequentially covers each data point of the time series. The average value of the data within the window is calculated as the filtered value for that data point, replacing the original data point's value, thus completing the stabilization processing of the real-time running status feature stream. For each attribute item's time series data in the environmental association feature stream, the same sliding window filtering method is used for stabilization processing, resulting in a stabilized environmental association feature stream.

[0019] Step S1222: Input the stable real-time running state feature stream and the stable environment association feature stream into the collaborative attention module, calculate the mutual information value between the stable real-time running state feature stream and the stable environment association feature stream, and generate bidirectional attention weights based on the mutual information value. The bidirectional attention weights are used to characterize the degree of influence of one type of data on another type of data.

[0020] In this embodiment, the collaborative attention module includes a mutual information calculation submodule and an attention weight generation submodule. The mutual information calculation submodule first converts the stationary real-time running state feature stream and the stationary environment-related feature stream into probability distributions. The conversion method involves normalizing the values ​​of each feature stream so that the sum of the values ​​of each feature stream is one. Next, it calculates the mutual information value between the two probability distributions. This is done by weighted summing the logarithms of the products of the joint probability distribution and the marginal probability distributions, according to the definition of mutual information. The weight is the value of the joint probability distribution. The attention weight generation submodule determines the bidirectional attention weights based on the mutual information values. For each feature component in the real-time running state feature stream, it determines the attention weight of that feature component to the environment-related feature stream based on its mutual information value with the corresponding feature component in the environment-related feature stream. Simultaneously, for each feature component in the environment-related feature stream, it determines the attention weight of that feature component to the real-time running state feature stream based on its mutual information value with the corresponding feature component in the real-time running state feature stream. The numerical range of the attention weights is consistent with the range of the mutual information values; the larger the mutual information value, the larger the corresponding attention weight.

[0021] Step S1223: Based on bidirectional attention weights, perform weighted enhancement processing on the stable real-time running state feature stream and the stable environment correlation feature stream, highlighting feature components whose mutual information values ​​meet preset conditions and weakening feature components whose mutual information values ​​do not meet preset conditions, to obtain the enhanced correlation feature stream.

[0022] In this embodiment, the weighted enhancement processing method involves multiplying each feature component in the stable real-time running state feature stream by its corresponding attention weight to obtain a weighted real-time running state feature component. Simultaneously, each feature component in the stable environment association feature stream is multiplied by its corresponding attention weight to obtain a weighted environment association feature component. Next, the weighted real-time running state feature component and the weighted environment association feature component are concatenated, arranged sequentially according to the order of the feature components, to form an enhanced association feature stream. A preset condition is that the mutual information value is greater than a preset threshold. The preset threshold is determined based on the statistical characteristics of the data. For feature components with a mutual information value greater than the preset threshold, their corresponding attention weight is greater than one, achieving highlighting processing; for feature components with a mutual information value less than or equal to the preset threshold, their corresponding attention weight is less than one, achieving weakening processing.

[0023] Step S1224: Perform cross-dimensional feature interaction processing on the enhanced correlation feature stream. Through the corresponding correlation operation of feature elements, fuse the common change information of the time dimension and attribute dimension to generate a preliminary correlation feature vector.

[0024] In this embodiment, cross-dimensional feature interaction processing employs pairwise multiplication of feature components. For each time step feature vector in the enhanced associated feature flow, each feature component in the feature vector is multiplied by all other feature components to obtain multiple interactive feature components. Next, the interactive feature components are concatenated with the original feature components, arranged sequentially according to the order of the feature components, to form a preliminary associated feature vector. The corresponding association operation of feature elements is based on the correspondence between the time dimension and the attribute dimension. The correspondence in the time dimension involves operation on feature components at the same time step, while the correspondence in the attribute dimension involves operation on feature components of different attribute items. This method integrates the common change information of the time dimension and the attribute dimension.

[0025] Step S1225: Input the preliminary associated feature vector into the feature refinement module, extract the global association information in the preliminary associated feature vector through non-local feature aggregation operation, eliminate the interference of local isolated features, and generate a cross-dimensional associated feature vector with global consistency.

[0026] In this embodiment, the feature refinement module adopts a nonlocal neural network structure, which includes an embedding layer, an attention calculation layer, and an aggregation layer. The embedding layer maps the initially associated feature vectors to a high-dimensional feature space using a linear transformation through a fully connected layer. The weight parameters of the fully connected layer are determined through model training. The attention calculation layer calculates the attention weight of each feature point in the high-dimensional feature space with all other feature points. The attention weight is calculated based on the similarity between feature points; the higher the similarity, the larger the corresponding attention weight. The aggregation layer performs a weighted summation of the feature points in the high-dimensional feature space according to the attention weights, obtaining an aggregated feature vector that contains global association information. Finally, the aggregated feature vector is mapped back to the original feature space using a linear transformation through a fully connected layer, resulting in a cross-dimensional associated feature vector with global consistency.

[0027] Step S123: Use the rule generation module of the analysis model to perform association pattern recognition on the cross-dimensional association feature vectors, filter out association patterns with stability and repeatability, and transform the association patterns into structured association rules.

[0028] In this embodiment, the rule generation module includes a pattern recognition submodule and a rule conversion submodule.

[0029] Step S1231: Perform feature clustering on the cross-dimensional associated feature vectors, grouping feature points that are close to each other in the feature space into the same cluster, with each cluster corresponding to a potential association pattern.

[0030] In this embodiment, the feature clustering process employs a hierarchical clustering algorithm. This algorithm first treats each feature point as a separate cluster, then calculates the distance between clusters, merges the two closest clusters into a new cluster, and repeats this process until the number of clusters reaches a preset number. The distance between clusters is calculated using the average distance, that is, the average distance between all feature points in two clusters is taken as the distance between the clusters. The preset number is determined based on the dimension and data volume of the cross-dimensional associated feature vectors, ensuring that each cluster contains a sufficient number of feature points to reflect a stable association pattern.

[0031] Step S1232: Calculate the internal compactness and inter-cluster separation of each cluster. Internal compactness reflects the concentration of feature points within a cluster, while inter-cluster separation reflects the degree of difference between different clusters.

[0032] In this embodiment, the internal compactness is calculated by averaging the distances from all feature points within a cluster to the cluster center using Euclidean distance. A smaller average distance indicates higher internal compactness. The inter-cluster separation is calculated by averaging the distance between the centers of two clusters using Euclidean distance. A larger distance indicates higher inter-cluster separation. For each cluster, its inter-cluster separation is calculated compared to all other clusters, and the minimum value is taken as the inter-cluster separation of that cluster.

[0033] Step S1233: Set screening conditions based on internal compactness and inter-cluster separation, retain clusters that meet both the internal compactness and inter-cluster separation conditions, and remove clusters that do not meet the conditions to obtain a set of effective clusters.

[0034] In this embodiment, the selection criteria are that the internal compactness is less than a preset compactness threshold and the inter-cluster separation is greater than a preset separation threshold. The preset compactness threshold and preset separation threshold are determined based on the statistical characteristics of the clustering results. For each cluster, it is determined whether its internal compactness is less than the preset compactness threshold and whether its inter-cluster separation is greater than the preset separation threshold. If both conditions are met, the cluster is retained; otherwise, it is discarded. Finally, a set of valid clusters is obtained, and each valid cluster corresponds to a stable association pattern.

[0035] Step S1234: Extract pattern features for each effective cluster, extract the central feature vector and feature distribution range of the effective cluster, and use the central feature vector and feature distribution range as the core features of the association pattern corresponding to the effective cluster.

[0036] In this embodiment, the central feature vector is extracted by calculating the average value of all feature points within the effective cluster, thus obtaining the feature vector of the cluster center. The feature distribution range is extracted by calculating the maximum and minimum values ​​of the distances from all feature points within the effective cluster to the central feature vector; the interval between the maximum and minimum values ​​is the feature distribution range. The central feature vector reflects the typical characteristics of the association pattern, while the feature distribution range reflects the fluctuation range of the association pattern.

[0037] Step S1235: Match the core features of each association pattern with the preset pattern description template, and transform the core features into structured association rules.

[0038] In this embodiment, the preset pattern description template is "When the attribute items of the environmental associated data are within the feature distribution range, the central feature vector of the attribute items of the real-time operating status data is the corresponding feature vector," where the attribute items are the attribute items of the environmental associated data, the attribute items are the attribute items of the real-time operating status data, and the corresponding feature vector is the central feature vector. The matching method is to substitute the attribute items of the environmental associated data and the attribute items of the real-time operating status data corresponding to the effective clusters into the attribute item positions in the template, substitute the central feature vector into the corresponding feature vector positions in the template, and substitute the feature distribution range into the feature distribution range part of the template to form a structured association rule. The structured association rule adopts the form of conditional result, such as "When the ambient temperature is within the feature distribution range, the central feature vector of the output power is the corresponding feature vector," where the corresponding feature vector is a specific feature vector value.

[0039] Step S124: Construct an initial association graph based on association rules. The nodes of the initial association graph correspond to the attribute items of real-time running status data and environmental association data, and the edges correspond to the association strength of the association rules.

[0040] In this embodiment, the initial association graph is constructed by first determining a node set, which includes all attribute items of the real-time running status data and all attribute items of the environmental association data, with each attribute item corresponding to one node. Next, an edge set is determined. For each association rule, if the rule contains attribute items of both environmental association data and real-time running status data, an edge is added between the corresponding nodes, with the edge pointing from the environmental association data attribute item to the real-time running status data attribute item. The association strength of the edge is determined based on the core features of the association rule, calculated as the magnitude of the central feature vector; the larger the magnitude, the higher the association strength. The number of nodes and edges in the initial association graph corresponds to the number of association rules, and each node may be connected to multiple other nodes.

[0041] Step S125: The graph optimization module of the analysis model is used to remove redundant edges and merge nodes in the initial association graph, strengthen the edge weights corresponding to key association rules, weaken the edge weights corresponding to meaningless associations, and generate a concise and accurate state association graph.

[0042] In this embodiment, the graph optimization module includes a redundant edge removal submodule, a node merging submodule, and a weight adjustment submodule.

[0043] For example, in step S1251: the redundant edge removal submodule performs redundant edge removal processing on the initial association graph. The removal method is to calculate the association strength of each edge. If the association strength is less than a preset redundancy threshold, the edge is removed. The preset redundancy threshold is determined based on the statistical characteristics of the association rules. The association rules corresponding to edges with association strength less than the redundancy threshold are meaningless associations. Removing these edges can reduce the redundant information in the graph.

[0044] Step S1252: The node merging submodule performs node merging processing on the initial association graph. The merging method is to calculate the similarity between nodes. If the similarity between two nodes is greater than a preset merging threshold, the two nodes are merged into one node. The similarity between nodes is determined based on the physical meaning and data distribution of the attribute items corresponding to the nodes. For example, if two attribute items are both temperature-related and their data distributions have a high degree of overlap, then the similarity is high. The merging threshold is determined based on the classification characteristics of the attribute items. Nodes with a similarity greater than the merging threshold are merged. The name of the merged node is a combination of the names of the two original nodes, and the attribute items corresponding to the merged node are the set of attribute items corresponding to the two original nodes.

[0045] Step S1253: The weight adjustment submodule performs edge weight adjustment on the initial association graph, strengthening the edge weights corresponding to key association rules and weakening the edge weights corresponding to meaningless associations.

[0046] In this embodiment, the edge weights are adjusted by first determining key association rules, which are association rules with a strength greater than a preset key threshold. The preset key threshold is determined based on the statistical characteristics of the association rules. For edges corresponding to key association rules, their weights are multiplied by a coefficient greater than one to strengthen the weights; for edges corresponding to meaningless associations, their weights are multiplied by a coefficient less than one to weaken the weights. The values ​​of the coefficients are determined based on the model training results to ensure that the adjusted edge weights accurately reflect the importance of the association rules.

[0047] Step S1254: Generate a simplified and accurate state association graph. The nodes in the state association graph are the merged set of nodes, the edges are the adjusted set of edges, and the edge weights are the adjusted weight values. The number of nodes and edges in the state association graph is reduced compared to the initial association graph. The edge weights corresponding to key association rules are larger, while the edge weights corresponding to meaningless associations are smaller, thus achieving simplification and accuracy of the graph.

[0048] Step S130: Perform reverse source analysis on the operating status of the heating device through the state association map to locate potential anomalies that may affect the operating stability of the heating device. The potential anomalies correspond to the association rules in the state association map.

[0049] In this embodiment, the reverse source tracing analysis is based on the structure and edge weights of the state association graph.

[0050] Step S131: Obtain abnormal feature items from the real-time operating status data of the heating device. Abnormal feature items are attribute items that deviate from the preset normal operating range.

[0051] In this embodiment, the preset normal operating range is the statistical range of historical normal data for each real-time operating status data attribute item. For example, the normal operating range of output power is an interval determined based on historical data. Abnormal features are obtained by real-time monitoring of the value of each real-time operating status data attribute item. If the value exceeds the preset normal operating range, the real-time operating status data attribute item is identified as an abnormal feature item. For example, if the value of output power exceeds the normal operating range, then output power is an abnormal feature item.

[0052] Step S132: Starting from the abnormal feature item, traverse the associated edges in the state association graph, track all associated nodes that are directly or indirectly associated with the abnormal feature item, and form an associated node chain.

[0053] In this embodiment, the traversal method is as follows: First, determine the node corresponding to the abnormal feature item and use this node as the starting point. Next, traverse all incoming edges from the starting point. The source node corresponding to the incoming edge is a directly related node, and this directly related node is added to the related node chain. Then, using the directly related node as the new starting point, traverse all its incoming edges. The source node corresponding to the incoming edge is an indirectly related node, and this indirectly related node is added to the related node chain. Repeat the above process until all related nodes have been traversed, forming a complete related node chain. The related node chain adopts a linked list structure, and each node contains a node name and its connection relationship with the previous node.

[0054] Step S133: Perform status verification on the attribute data corresponding to each associated node in the associated node chain, determine whether the attribute data corresponding to the associated node is within the preset normal data range, and filter out abnormal associated nodes that are within the abnormal data range.

[0055] In this embodiment, the status verification method involves first obtaining the real-time values ​​of the attribute data corresponding to the associated node, and then comparing these real-time values ​​with a preset normal data range. The preset normal data range is the statistical range of historical normal data for each environmental associated data attribute item; for example, the normal data range for ambient temperature is an interval determined based on historical data. If the real-time value of the attribute data corresponding to the associated node exceeds the preset normal data range, then the associated node is an abnormal associated node; otherwise, it is a normal associated node. For example, if the real-time value of the ambient temperature exceeds the normal data range, then the associated node corresponding to the ambient temperature is an abnormal associated node.

[0056] Step S134: Analyze the association path length and association edge weight between abnormal association nodes and abnormal feature items. The smaller the value of the association path length and the larger the value of the association edge weight, the greater the influence of abnormal association nodes on abnormal feature items.

[0057] In this embodiment, the analysis method is to calculate the length of the associated path and the weight of the associated edge between the abnormal associated node and the node corresponding to the abnormal feature item.

[0058] Step S1341: Calculate the number of associated edges in the shortest path between each abnormal associated node and the node corresponding to the abnormal feature item, and use the number of associated edges as the length of the associated path.

[0059] In this embodiment, the shortest path is calculated using a breadth-first search algorithm. Starting from the node corresponding to the abnormal associated node, all possible paths to the node corresponding to the abnormal feature item are searched. The number of associated edges contained in each path is calculated, and the path with the fewest associated edges is selected as the shortest path. The number of associated edges is the length of the associated path.

[0060] Step S1342: Extract the weight value of each associated edge on the shortest path, calculate the product of the weight values ​​of all associated edges as the total path weight, and the total path weight reflects the association strength of the path.

[0061] In this embodiment, the weight values ​​are extracted by obtaining the weight value of each associated edge on the shortest path from the state association graph, and then multiplying these weight values ​​to obtain the total path weight.

[0062] Step S1343: Set the path length weight coefficient and the total path weight coefficient. Construct an impact assessment method based on the path length and the total path weight. The impact assessment method is negatively correlated with the path length and positively correlated with the total path weight.

[0063] In this embodiment, the path length weight coefficient and the total path weight coefficient are preset constants determined through model training. The impact degree is evaluated by multiplying the total path weight coefficient by the total path weight and dividing by multiplying the path length weight coefficient by the path length. This method achieves a negative correlation with path length and a positive correlation with the total path weight.

[0064] Step S1344: Substitute the sum of the associated path length and path weight of each abnormal associated node into the influence degree assessment method to calculate the influence degree score corresponding to each abnormal associated node.

[0065] In this embodiment, the substitution method involves substituting the sum of the associated path length and path weight into the influence assessment method to calculate the influence score.

[0066] Step S1345: Perform uniform range processing on the influence degree scores of all abnormal associated nodes to make the score range a fixed interval, and determine the influence degree of each abnormal associated node on the abnormal feature item based on the influence degree scores after uniform range processing.

[0067] In this embodiment, the unified range processing method is to first determine the maximum and minimum values ​​of all influence scores, then subtract the minimum value from each influence score, and finally divide by the difference between the maximum and minimum values ​​to obtain the normalized score. The higher the normalized score, the greater the influence of the abnormal association node on the abnormal feature item.

[0068] Step S135: Sort the abnormal associated nodes according to the degree of influence, and select the abnormal associated nodes with significant influence as potential sources of abnormality that may affect the operational stability of the heating device.

[0069] In this embodiment, the sorting method involves ranking the normalized impact scores from highest to lowest, and selecting the anomaly-related nodes with the highest scores as potential anomaly sources. The criterion for determining significant impact is that the normalized impact score is greater than a preset impact threshold, which is determined based on the model training results.

[0070] Step S140: Based on the potential anomaly source, query the preset control parameter adaptation library to determine the target control parameter combination that matches the potential anomaly source. The target control parameter combination can adapt to the current operating status and environmental conditions of the heating device.

[0071] In this embodiment, the control parameter adaptation library is a pre-built database that contains the correspondence between potential anomaly sources and control parameter combinations.

[0072] Step S141: Perform feature encoding processing on potential anomaly sources, converting the attribute information of each potential anomaly source into a standardized feature encoding vector. The dimension of the feature encoding vector is consistent with the index encoding dimension in the control parameter adaptation library.

[0073] In this embodiment, the feature encoding process first determines the attribute information of potential anomaly sources. This attribute information includes the attribute item name corresponding to the anomaly-related node, the range of anomaly data, and the impact score. Next, the attribute information is converted into numerical features. This conversion involves one-hot encoding of the attribute item name, interval encoding of the anomaly data range, and normalization of the impact score. Then, the numerical features are arranged in a preset order to form a feature encoding vector. The dimension of the feature encoding vector is consistent with the index encoding dimension in the control parameter adaptation library.

[0074] Step S142: Based on the feature encoding vector, traverse the adjustment parameter adaptation library, calculate the similarity between the feature encoding vector and each preset anomaly source encoding vector in the adaptation library, and select candidate encoding vectors whose similarity meets the preset requirements.

[0075] In this embodiment, the traversal method involves sequentially reading each preset anomaly source encoding vector in the control parameter adaptation library and calculating its similarity to the feature encoding vector. The similarity is calculated using cosine similarity, which is calculated by dividing the dot product of the feature encoding vector and the preset anomaly source encoding vector by the product of the magnitudes of the feature encoding vector and the preset anomaly source encoding vector. A preset requirement is that the similarity must be greater than a preset similarity threshold, which is determined based on the coverage of the control parameter adaptation library.

[0076] Step S143: Extract the candidate control parameter combinations corresponding to the candidate encoding vectors. Each candidate control parameter combination contains a set of operation adjustment parameters adapted to the corresponding anomaly source.

[0077] In this embodiment, the extraction method involves reading the control parameter combination corresponding to the index position of the candidate encoding vector in the control parameter adaptation library, and using this combination as the candidate control parameter combination. The candidate control parameter combination includes operating adjustment parameters such as output power adjustment value, operating current adjustment value, and operating voltage adjustment value.

[0078] Step S144: Perform adaptability verification on each candidate control parameter combination. Combine the current real-time operating status data of the heating device with the environmental correlation data to determine whether the candidate control parameter combination can adapt to the current operating conditions.

[0079] In this embodiment, the adaptability verification method is as follows: First, a feature vector of the current operating condition is constructed. The feature vector contains the real-time values ​​of all attribute items of the real-time operating status data and the real-time values ​​of all attribute items of the environmental correlation data. Next, the candidate control parameter combination is applied to the current operating condition to simulate the changes in the operating status of the heating device, obtaining simulated operating status data. Then, the simulated operating status data is compared with the preset expected operating status data, and the similarity between the two is calculated. If the similarity is greater than the preset adaptability threshold, the candidate control parameter combination can adapt to the current operating condition; otherwise, it cannot adapt to the current operating condition. The preset expected operating status data is the operating status data determined based on historical normal operating conditions, and the adaptability threshold is determined based on the model training results.

[0080] Step S1441: Input each candidate control parameter combination into the operating condition simulation module, and construct a simulated operating scenario by combining the current real-time operating status data and environmental correlation data.

[0081] In this embodiment, the operating condition simulation module is a pre-built simulation system for the operation of a heating device. This simulation system has a built-in physical operating model of the heating device and can simulate changes in the device's operating state based on the input parameters. When constructing the simulated operating scenario, the attribute values ​​in the current real-time operating status data (such as output power, operating current, and pipe inlet fluid temperature) are first used as the initial operating parameters of the simulation system. Then, the attribute values ​​in the current environmental correlation data (such as ambient temperature, ambient humidity, and external wind speed of the pipe) are used as the environmental input parameters of the simulation system. Finally, the adjustment parameters in the candidate control parameter combination (such as output power adjustment value and operating current adjustment value) are used as the control input parameters of the simulation system. By combining the above three types of parameters, a simulated operating scenario consistent with the current actual operating conditions is constructed.

[0082] Step S1442: In the simulated operation scenario, simulate the adjustment operation corresponding to the candidate control parameter combination, and record the change data of the heating device's operating status and the response change data of the environmental correlation data during the simulation.

[0083] In this embodiment, the simulation process of performing the adjustment operation is as follows: the simulation system adjusts the initial operating parameters according to the adjustment parameters in the candidate control parameter combination. For example, if the candidate control parameter combination includes an output power adjustment value, the simulation system adjusts the initial output power parameter to the superposition result of the initial value and the adjustment value. Subsequently, the simulation system runs according to a preset time step. After each time step, it records the operating status change data of the heating device (including the adjusted output power, operating current, operating voltage, pipe outlet fluid temperature, heating element surface temperature, etc.) and records the response change data of environmental related data (including the fluctuation value of ambient temperature, the fluctuation value of ambient humidity, the fluctuation value of external wind speed of the pipe, etc.). The simulation execution time is consistent with the real-time data acquisition cycle to ensure that the recorded data can reflect the complete state change process.

[0084] Step S1443: Analyze whether the changes in operating status conform to the preset normal operating trend, and determine whether the changes in response data are within a reasonable fluctuation range.

[0085] In this embodiment, the preset normal operation trend is a pattern of state changes summarized from the design parameters of the heating device and historical normal operation data. For example, the outlet fluid temperature should rise steadily with the increase of output power, and the rate of increase should remain within a fixed range; the operating current should change linearly with the adjustment of output power, and the change amplitude should not exceed the preset safety threshold. When analyzing the operating state change data, the simulated and recorded operating state change data is compared with the preset normal operation trend step by step to determine whether the operating state data at each time step meets the trend requirements. When judging the response change data, a reasonable fluctuation range is first set based on the historical fluctuation of environmental data. For example, the reasonable fluctuation range of ambient temperature is a fixed range above and below the current ambient temperature value, and the reasonable fluctuation range of ambient humidity is a fixed range above and below the current ambient humidity value. Then, the simulated and recorded response change data is compared with the corresponding reasonable fluctuation range to determine whether the data is within the range.

[0086] Step S1444: If the change data of the operating status conforms to the preset normal operating trend and the change data of the response is within a reasonable fluctuation range, then the adaptability verification of the candidate control parameter combination is determined to be successful.

[0087] In this embodiment, the criteria for passing the adaptability verification are that two sub-conditions are met simultaneously: First, the comparison results of the operating status change data at all time steps conform to the preset normal operation trend, that is, the deviation value of the operating status data at each time step from the trend requirement is less than the preset deviation threshold; second, the values ​​of the response change data at all time steps are within the corresponding reasonable fluctuation range, that is, the difference between the response change data at each time step and the current environmental correlation data is less than the preset fluctuation threshold. When both sub-conditions are met simultaneously, it indicates that the candidate control parameter combination can enable the heating device to maintain normal operation under the current operating conditions and will not have an unexpected impact on the environmental correlation data. Therefore, the adaptability verification is deemed to have passed.

[0088] Step S1445: If the change data of the operating status does not conform to the preset normal operating trend or the change data of the response exceeds the reasonable fluctuation range, it is determined that the adaptability verification of the candidate control parameter combination has failed, and the candidate control parameter combination is removed from the candidate set.

[0089] In this embodiment, the condition for failing the adaptability verification is that either of the following two sub-conditions is not met: First, the comparison result of the operating status change data at least at one time step does not conform to the preset normal operating trend, that is, the deviation value of the operating status data at that time step from the trend requirement is greater than or equal to the preset deviation threshold; second, the value of the response change data at least at one time step exceeds the corresponding reasonable fluctuation range, that is, the difference between the response change data at that time step and the current environmental correlation data is greater than or equal to the preset fluctuation threshold. When either sub-condition is met, it indicates that the candidate control parameter combination cannot enable the heating device to maintain normal operation under the current operating conditions, or will have an unexpected impact on the environmental correlation data. Therefore, the adaptability verification is determined to have failed, and the candidate control parameter combination is removed from the candidate set and will no longer participate in the subsequent priority ranking.

[0090] Step S145: Retain the candidate control parameter combinations that have passed the adaptability verification, sort the candidate control parameter combinations that have passed the adaptability verification by priority, and select the candidate control parameter combination with the highest priority as the target control parameter combination to match the potential anomaly source.

[0091] In this embodiment, the priority ranking is based on the similarity of the candidate control parameter combinations in the suitability verification, from high to low. The higher the similarity, the higher the priority. The candidate control parameter combination with the highest priority is selected as the target control parameter combination.

[0092] Step S150: The target control parameters are packaged into standardized remote control commands and sent to the IoT control module of the heating device through the IoT communication link. The IoT control module then executes the remote control commands to adjust the operating parameters of the heating device.

[0093] In this embodiment, the encapsulation and transmission of remote control commands follow the Internet of Things (IoT) communication protocol.

[0094] Step S151: Extract each adjustment parameter from the target control parameter combination, divide and arrange each adjustment parameter into fields according to the preset instruction format specification to form the instruction raw data.

[0095] In this embodiment, the preset instruction format is a fixed field combination, which includes the instruction type and various adjustment parameters. The fields are divided by treating each adjustment parameter in the target control parameter combination as a single field, and arranged in descending order of importance. The resulting raw instruction data is in string format.

[0096] Step S152: Perform integrity verification on the original instruction data to check for missing parameters or abnormal parameter formats. If any are found, return to the step of re-determining the target control parameter combination.

[0097] In this embodiment, the integrity verification method is as follows: First, check whether the number of fields in the original instruction data is consistent with the preset number of fields. If they are inconsistent, there is a missing parameter. Then, check whether the format of each field conforms to the preset format requirements. For example, the format of the output power adjustment value is numerical. If the field format is non-numerical, there is an abnormal parameter format. If there are missing parameters or abnormal parameter formats, return to step S145 to redetermine the target control parameter combination; otherwise, proceed to the next step.

[0098] Step S153: Encrypt the verified original instruction data by using a preset encryption algorithm to protect the original instruction data and generate encrypted instruction data.

[0099] In this embodiment, the preset encryption algorithm is a symmetric encryption algorithm. The encryption process involves first obtaining a preset encryption key, and then using the encryption key to perform encryption operations on the original instruction data to obtain the encrypted instruction data. The encryption key is a key pre-stored in the IoT control module and monitoring system to ensure consistency between encryption and decryption.

[0100] Step S154: Encapsulate the encrypted instruction data into a remote control instruction conforming to the Internet of Things communication protocol, and add instruction identifier, sending timestamp and verification code fields.

[0101] In this embodiment, the IoT communication protocol is a fixed protocol. The encapsulation method involves using encrypted instruction data as the message content, adding an instruction identifier as the message identifier, adding a sending timestamp as the message time, and adding a checksum as the message verification. The instruction identifier is a unique string used to identify the type and purpose of the remote control instruction. The sending timestamp is the current system time, in a fixed time string format. The checksum is a value obtained by hashing the encrypted instruction data, used to verify the integrity of the instruction data.

[0102] Step S155: Send remote control commands to the IoT control module of the heating device via the IoT communication link, and record the sending log. The sending log is used for querying and tracing the command transmission process.

[0103] In this embodiment, the IoT communication link is a wireless communication link. The transmission method involves calling the transmission interface of the IoT communication module to send the remote control command to the designated address of the IoT control module. The transmission log is recorded by storing the command identifier, transmission time, transmission link information, reception confirmation time, and command content digest in a log file. The command content digest is a value obtained by hashing the encrypted command data, used for quickly identifying the command content.

[0104] For example, in step S1551: before sending the remote control command, the connection status and transmission quality of the IoT communication link are detected to determine whether the IoT communication link meets the command transmission requirements.

[0105] In this embodiment, the connection status is detected by sending a test data packet to the IoT control module. If a response is received, the connection status is normal; otherwise, the connection status is abnormal. The transmission quality is detected by calculating the transmission delay and packet loss rate of the test data packet. If the transmission delay is less than a preset delay threshold and the packet loss rate is less than a preset packet loss rate threshold, the transmission quality meets the requirements; otherwise, the transmission quality does not meet the requirements. If the IoT communication link's connection status is normal and the transmission quality meets the requirements, the next step is performed; otherwise, link switching is initiated.

[0106] Step S1552: If the IoT communication link meets the transmission requirements, then send a remote control command through the IoT communication link; if the IoT communication link does not meet the transmission requirements, then switch to the backup IoT communication link, check the transmission status of the backup IoT communication link, until a communication link that meets the requirements is found.

[0107] In this embodiment, the backup IoT communication link is another wireless communication link. The switching method is to shut down the current communication link, open the backup communication link, and repeat the detection process in step S1551 until a suitable communication link is found. After finding a suitable communication link, remote control commands are sent through that communication link.

[0108] Step S1553: During the instruction sending process, monitor the transmission progress and transmission status in real time. If a transmission interruption or transmission error occurs, trigger the retransmission mechanism to resend the remote control instruction.

[0109] In this embodiment, the transmission progress is monitored by obtaining the number of bytes transmitted returned by the communication module and comparing it with the total number of bytes of the remote control command to determine the transmission progress. The transmission status is monitored by obtaining the status code returned by the communication module. If the status code is successful, the transmission status is normal; if the status code is failed, the transmission status is abnormal. If a transmission interruption or transmission error occurs, i.e., the transmission status is abnormal, a retransmission mechanism is triggered. The retransmission mechanism waits for a preset time and then retransmits the remote control command, repeating the retransmission a preset number of times. If it still fails, an error log is recorded and the administrator is notified.

[0110] Step S1554: After confirming that the IoT control module of the heating device has successfully received and returned the confirmation information, record the sending log. The sending log includes the instruction identifier, sending time, transmission link information, receiving confirmation time, and instruction content summary.

[0111] In this embodiment, the confirmation message is received by the IoT control module sending a confirmation message to the monitoring system after successfully receiving the remote control command. The confirmation message has a fixed string format, containing a command identifier and a success status. The sending log is recorded by writing the command identifier, sending time, transmission link information, confirmation time, and command content summary to a log file, where the confirmation time is the system time at which the confirmation message was received.

[0112] Step S1555: Store the sending log in the preset log database and create a log index. The log index is used for subsequent querying and tracing of the instruction transmission process.

[0113] In this embodiment, the log database is a relational database. The storage method involves inserting each field of the transmitted log into a database table. The database table fields include the instruction identifier, transmission time, transmission link information, reception acknowledgment time, and instruction content summary. The log index is created by using the instruction identifier as the index key to establish an index in the database table, facilitating quick retrieval of transmitted logs based on the instruction identifier.

[0114] Figure 2The illustration shows exemplary hardware and software components of an IoT-integrated remote monitoring system 100 for heating devices, which can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the IoT-integrated remote monitoring system 100 for heating devices and to perform the functions described in this application.

[0115] The IoT-integrated remote monitoring system 100 for heating devices can be a general-purpose server or a special-purpose server; both can be used to implement the IoT-integrated remote monitoring method for heating devices described in this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0116] For example, the IoT-integrated remote monitoring system 100 for heating devices may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the IoT-integrated remote monitoring system 100 for heating devices may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The IoT-integrated remote monitoring system 100 for heating devices also includes an I / O interface 150 between the computer and other input / output devices.

[0117] For ease of explanation, only one processor is described in the IoT-integrated remote monitoring system 100 for heating devices. However, it should be noted that the IoT-integrated remote monitoring system 100 for heating devices in this application may also include multiple processors. Therefore, the steps performed by one processor as described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the IoT-integrated remote monitoring system 100 for heating devices executes steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0118] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned method for remote monitoring of heating devices combined with the Internet of Things is implemented.

[0119] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for remote monitoring of a heating device combined with the Internet of Things, characterized in that, The method includes: Acquire real-time operating status data of IoT-connected heating devices and environmental data of the environment in which the heating devices are located. The real-time operating status data reflects the operating conditions of the heating devices, and the environmental data reflects the environmental conditions around the heating devices. Based on the pre-trained analysis model, cross-dimensional correlation modeling is performed on real-time operation status data and environmental correlation data to mine the inherent correlation between real-time operation status data and environmental correlation data and generate a state correlation graph containing correlation rules. By performing reverse source analysis on the operating status of the heating device through state association graph, potential anomalies that may affect the operating stability of the heating device can be located. The potential anomalies correspond to the association rules in the state association graph. Based on the query of the preset control parameter adaptation library for potential anomalies, the target control parameter combination that matches the potential anomaly is determined. The target control parameter combination can adapt to the current operating status and environmental conditions of the heating device. The target control parameters are packaged into standardized remote control commands, which are sent to the IoT control module of the heating device via an IoT communication link. The IoT control module then executes the remote control commands to adjust the operating parameters of the heating device.

2. The remote monitoring method for heating devices combined with the Internet of Things according to claim 1, characterized in that, The pre-trained analysis model performs cross-dimensional correlation modeling on real-time operational status data and environmental correlation data, mines the inherent correlation between real-time operational status data and environmental correlation data, and generates a state correlation graph containing correlation rules, including: Real-time operating status data and environmental correlation data are input into the feature input layer of the analysis model. Feature channel alignment processing is performed on the real-time operating status data and environmental correlation data to ensure that the feature dimensions of the real-time operating status data and environmental correlation data are consistent, forming a dual-channel input feature stream with a unified dimension. By analyzing the correlation mining layer of the analysis model, collaborative correlation analysis is performed on the dual-input feature streams to extract the common change features of real-time operating status data and environmental correlation data in the time and attribute dimensions, and to generate cross-dimensional correlation feature vectors. The rule generation module of the analysis model is used to identify association patterns in cross-dimensional association feature vectors, filter out association patterns with stability and repeatability, and transform the association patterns into structured association rules. An initial association graph is constructed based on association rules. The nodes of the initial association graph correspond to the attribute items of real-time running status data and environmental association data, and the edges correspond to the association strength of the association rules. The graph optimization module of the analysis model performs redundant edge removal and node merging on the initial association graph, strengthens the edge weights corresponding to key association rules, weakens the edge weights corresponding to meaningless associations, and generates a concise and accurate state association graph.

3. The remote monitoring method for heating devices combined with the Internet of Things according to claim 2, characterized in that, The method involves performing collaborative correlation analysis on the dual-input feature streams through the correlation mining layer of the analysis model, extracting common change features of real-time operating status data and environmental correlation data in the time and attribute dimensions, and generating cross-dimensional correlation feature vectors, including: In the correlation mining layer, time series trend extraction processing is performed on the dual input feature streams to retain the core change features of the data and filter out instantaneous fluctuation interference, so as to obtain a stable real-time running status feature stream and a stable environmental correlation feature stream. The stable real-time running state feature stream and the stable environment association feature stream are input into the collaborative attention module. The mutual information value between the stable real-time running state feature stream and the stable environment association feature stream is calculated. Based on the mutual information value, bidirectional attention weights are generated. The bidirectional attention weights are used to characterize the degree of influence of one type of data on another type of data. Based on bidirectional attention weights, the stable real-time running state feature stream and the stable environment correlation feature stream are weighted and enhanced, highlighting the feature components whose mutual information values ​​meet the preset conditions and weakening the feature components whose mutual information values ​​do not meet the preset conditions, thus obtaining the enhanced correlation feature stream. Cross-dimensional feature interaction processing is performed on the enhanced correlation feature stream. By performing corresponding correlation operations on feature elements, the common change information of the time dimension and attribute dimension is fused to generate a preliminary correlation feature vector. The initial correlation feature vector is input into the feature refinement module. The global correlation information in the initial correlation feature vector is extracted through non-local feature aggregation operation, eliminating the interference of local isolated features and generating a cross-dimensional correlation feature vector with global consistency.

4. The remote monitoring method for heating devices combined with the Internet of Things according to claim 2, characterized in that, The rule generation module of the analytical model performs association pattern recognition on cross-dimensional association feature vectors, filters out association patterns with stability and repeatability, and transforms the association patterns into structured association rules, including: Feature clustering is performed on cross-dimensional correlation feature vectors, grouping feature points that are close to each other in the feature space into the same cluster, with each cluster corresponding to a potential correlation pattern; Calculate the internal compactness and inter-cluster separation of each cluster. Internal compactness reflects the concentration of feature points within a cluster, while inter-cluster separation reflects the degree of difference between different clusters. The selection criteria are set based on internal compactness and inter-cluster separation. Clusters that meet both the criteria are retained, while clusters that do not meet the criteria are removed, resulting in a set of effective clusters. For each effective cluster, pattern features are extracted, including the central feature vector and feature distribution range of the effective cluster. The central feature vector and feature distribution range are used as the core features of the association patterns corresponding to the effective clusters. The core features of each association pattern are matched with a preset pattern description template, and the core features are transformed into association rules described in natural language. The association rules clarify the correspondence between the attribute items of the real-time running status data and the attribute items of the environmental association data.

5. The remote monitoring method for a heating device combined with the Internet of Things according to claim 1, characterized in that, The reverse source analysis of the heating device's operating status using state correlation maps to locate potential anomalies that may affect the stability of the heating device's operation includes: Obtain abnormal feature items from the real-time operating status data of the heating device. Abnormal feature items are attribute items that deviate from the preset normal operating range. Starting with the abnormal feature item, traverse the association edges in the state association graph, track all association nodes that have direct and indirect association with the abnormal feature item, and form a chain of association nodes. Perform status verification on the attribute data corresponding to each associated node in the associated node chain, determine whether the attribute data corresponding to the associated node is within the preset normal data range, and filter out abnormal associated nodes that are within the abnormal data range; Analyze the association path length and association edge weight between abnormal association nodes and abnormal feature items. The smaller the value of the association path length and the larger the value of the association edge weight, the greater the influence of the abnormal association node on the abnormal feature item. The abnormal associated nodes are sorted according to the degree of their impact, and the abnormal associated nodes with significant impact are selected as potential sources of anomalies that may affect the operational stability of the heating device.

6. The remote monitoring method for a heating device combined with the Internet of Things according to claim 5, characterized in that, The analysis examines the path length and edge weights between abnormally associated nodes and abnormal feature items. A smaller path length and a larger edge weight indicate a greater influence of the abnormally associated node on the abnormal feature item. This includes: Calculate the number of associated edges in the shortest path between each associated node and the node corresponding to the abnormal feature item, and use the number of associated edges as the length of the associated path; Extract the weight value of each associated edge on the shortest path, calculate the product of the weight values ​​of all associated edges as the total path weight, and the total path weight reflects the association strength of the path; Set a path length weight coefficient and a path weight sum coefficient, and construct an impact assessment method based on the path length and path weight sum. The impact assessment method is negatively correlated with the path length and positively correlated with the path weight sum. Substitute the total path length and path weight of each abnormal associated node into the impact assessment method to calculate the impact score corresponding to each abnormal associated node. The impact scores of all abnormal associated nodes are processed to a uniform range, so that the score range is within a consistent interval. The impact score of each abnormal associated node on the abnormal feature item is determined based on the impact score after the uniform range processing.

7. The remote monitoring method for a heating device combined with the Internet of Things according to claim 1, characterized in that, The step of determining the target control parameter combination that matches the potential anomaly source by querying a preset control parameter adaptation library includes: The potential anomaly sources are processed by feature encoding, and the attribute information of each potential anomaly source is transformed into a standardized feature encoding vector. The dimension of the feature encoding vector is consistent with the index encoding dimension in the control parameter adaptation library. Based on the feature encoding vector traversal of the control parameter adaptation library, the similarity between the feature encoding vector and each preset anomaly source encoding vector in the adaptation library is calculated, and candidate encoding vectors with similarity that meet the preset requirements are selected. Extract the candidate control parameter combinations corresponding to the candidate encoding vectors. Each candidate control parameter combination contains a set of operation adjustment parameters adapted to the corresponding anomaly source. For each candidate control parameter combination, the adaptability is verified. Based on the current real-time operating status data of the heating device and the environmental correlation data, it is determined whether the candidate control parameter combination can adapt to the current operating conditions. Candidate control parameter combinations that have passed the adaptability verification are retained. These candidate control parameter combinations are then prioritized, and the candidate control parameter combination with the highest priority is selected as the target control parameter combination to match the potential anomaly source.

8. The remote monitoring method for a heating device combined with the Internet of Things according to claim 7, characterized in that, The adaptation verification of each candidate control parameter combination, combined with the current real-time operating status data of the heating device and environmental correlation data, determines whether the candidate control parameter combination can adapt to the current operating conditions, including: Each candidate control parameter combination is input into the operating condition simulation module, and a simulated operating scenario is constructed by combining the current real-time operating status data and environmental correlation data. In the simulated operation scenario, the adjustment operation corresponding to the candidate control parameter combination is simulated and executed, and the operating status change data of the heating device and the response change data of the environmental related data are recorded during the simulation. Analyze whether the changes in operating status data conform to the preset normal operating trend, and determine whether the changes in response data are within a reasonable fluctuation range; If the data on changes in operating status conform to the preset normal operating trend and the data on changes in response are within a reasonable fluctuation range, then the adaptability verification of the candidate control parameter combination is deemed to have passed. If the data on changes in operating status do not conform to the preset normal operating trend or the data on changes in response exceed the reasonable fluctuation range, the suitability verification of the candidate control parameter combination is deemed to have failed, and the candidate control parameter combination is removed from the candidate set.

9. The remote monitoring method for a heating device combined with the Internet of Things according to claim 1, characterized in that, The step of encapsulating the target control parameters into standardized remote control commands and sending them to the IoT control module of the heating device via an IoT communication link includes: Extract each adjustment parameter from the target control parameter combination, divide and arrange each adjustment parameter into fields according to the preset instruction format specification to form the instruction raw data; Perform integrity verification on the original instruction data to check for missing parameters or abnormal parameter formats. If any are found, return to the step of re-determining the target control parameter combination. The original instruction data that has passed verification is encrypted. A preset encryption algorithm is used to protect the original instruction data and generate encrypted instruction data. The encrypted command data is encapsulated into remote control commands that conform to the Internet of Things communication protocol, and command identifier, sending timestamp and verification code fields are added; Remote control commands are sent to the IoT control module of the heating device via an IoT communication link, and a transmission log is recorded. The transmission log is used for querying and tracing the command transmission process.

10. A remote monitoring system for heating devices integrated with the Internet of Things, characterized in that, The IoT-integrated remote monitoring system for heating devices includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the memory to implement the IoT-integrated remote monitoring method for heating devices as described in any one of claims 1-9.

Images