Temperature sensing data processing method and system combined with edge computing

Abstract

The application provides a temperature sensing data processing method and system combined with edge computing, based on an edge computing node distributed deployment architecture, receiving temperature sensing data transmitted by each edge node, carrying node identification and collection time identification. Each edge node locally performs distributed feature linkage processing, generates a local correlation feature set and transmits it to an edge collaborative node, and the edge collaborative node performs global feature integration to generate a global temperature state representation. A pre-trained temperature state analysis model is called to perform correlation judgment on the global temperature state representation, generating a temperature state correlation judgment result. Based on the judgment result, temperature data processing instructions adapted to each edge node are generated and distributed, and each edge node performs local temperature sensing data processing operations according to the instructions. The application makes full use of edge computing resources, improves data processing efficiency and accuracy, and enhances system reliability and real-time performance.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence, and more specifically, to a method and system for processing temperature sensing data using edge computing. Background Technology

[0002] In the field of temperature sensor data processing, traditional centralized data processing methods face numerous challenges. Centralized processing requires transmitting large amounts of temperature sensor data from various collection points to a central server. This not only places extremely high demands on network bandwidth but also makes data incomplete or untimely due to network latency and packet loss during transmission. Simultaneously, the central server needs to process massive amounts of data, resulting in immense computational pressure and making it difficult to meet the real-time requirements of applications. Furthermore, centralized processing struggles to fully utilize the local computing resources of each collection point, leading to resource waste. While some existing distributed processing methods alleviate some of the problems of centralized processing to a certain extent, most lack in-depth analysis of the temporal variation patterns of temperature sensor data and the spatial correlations between nodes. During processing, each node processes data independently, making it difficult to form a global temperature state representation. This results in inaccurate temperature state judgments, failing to provide effective guidance for subsequent data processing and ultimately affecting the performance and reliability of the entire temperature sensor data processing system. 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 temperature sensing data processing method incorporating edge computing, the method comprising: Based on a distributed deployment architecture of edge computing nodes, the system receives temperature sensing data transmitted from each edge node. The temperature sensing data is the ambient temperature sensing data collected in real time by each edge node, and carries node identification information and collection time identifier. Each edge node performs distributed feature linkage processing of temperature sensing data locally, and establishes data association relationships by combining node identification information and acquisition time identifier to generate a corresponding local association feature set. The local associated feature sets of each edge node are transmitted to the edge collaboration node, and the global features are integrated through the cross-node feature fusion mechanism of the edge collaboration node to generate a global temperature state representation with inter-node correlation. The pre-trained temperature state analysis model is invoked to make correlation judgments on the global temperature state representation. The temperature state correlation judgment results are generated by combining the temporal variation pattern of temperature sensing data and the spatial correlation pattern between nodes. Based on the temperature status correlation judgment results, temperature data processing instructions adapted to each edge node are generated and distributed to the corresponding edge nodes. Each edge node then executes the local temperature sensing data processing operation according to the instructions.

[0004] In another aspect, embodiments of the present invention also provide a temperature sensing data processing system combined with edge computing, 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 embodiment of the invention achieves efficient processing of temperature sensing data through a distributed deployment architecture based on edge computing nodes. Each edge node performs distributed feature linkage processing locally, establishing data associations by combining node identifiers and acquisition time identifiers to generate a local associated feature set. This fully utilizes the local computing resources of the edge nodes, reducing data transmission volume and lowering network bandwidth requirements. The local associated feature set is transmitted to edge collaborative nodes for global feature integration, generating a global temperature state representation with inter-node correlations, which can comprehensively and accurately reflect the temperature state of the entire monitoring area. A pre-trained temperature state analysis model is invoked to perform association judgments on the global temperature state representation, combining temporal change patterns and spatial correlation patterns to generate accurate temperature state association judgment results. Based on the judgment results, temperature data processing instructions adapted to each edge node are generated, enabling each edge node to perform targeted processing of local data according to actual needs, improving the accuracy and efficiency of data processing, and enhancing the reliability and real-time performance of the entire system. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the execution flow of the temperature sensing data processing method combined with edge computing provided in an embodiment of the present invention.

[0007] Figure 2 This is a schematic diagram of exemplary hardware and software components of a temperature sensing data processing system combining edge computing, 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 schematic flowchart of a temperature sensing data processing method combining edge computing according to an embodiment of the present invention. The following is a detailed description of the temperature sensing data processing method combining edge computing.

[0009] Step S110: Based on the distributed deployment architecture of edge computing nodes, receive temperature sensing data transmitted by each edge node. The temperature sensing data is the ambient temperature sensing data collected in real time by each edge node, and carries node identification information and collection time identifier.

[0010] In this embodiment, the distributed deployment architecture of the edge computing nodes is specifically a self-organizing network structure composed of twenty edge nodes dispersed in a grid-like layout within a closed campus. Each edge node is equipped with a high-precision digital temperature sensor, a low-power communication module, and a local storage unit. The sampling frequency of the temperature sensor is set to once per second, and the accuracy of the sampled data is retained to two decimal places. After completing a single temperature sampling, each edge node automatically calls the built-in identifier generation module to add a node identifier information consisting of sixteen characters to the temperature sensing data. The first eight characters are the hardware code of the edge node, and the last eight characters are the logical address code of the node in the deployment architecture. At the same time, a collection time identifier consisting of twelve characters is added. The identifier format is a combination of year, month, day, hour, minute, and second. For example, the collection time identifier of a certain data is 20240520143005, which means that the data was collected at 2:30:05 PM on May 20, 2024. The edge collaboration node is deployed in the center of the park and is equipped with a high-performance processor and a large-capacity storage device. It receives temperature sensing data transmitted by each edge node through a network module that supports concurrent communication of multiple nodes. During the reception process, it checks the header format, data length and checksum of each data packet in real time. The header format must strictly conform to the preset frame structure specification, the data length must be consistent with the preset length of a single temperature sensing data packet, and the checksum must be the same as the result calculated by the preset algorithm. Only data that meets all three conditions will be judged to conform to the transmission specification and be received and stored.

[0011] Step S120: Each edge node performs distributed feature linkage processing of temperature sensing data locally, establishes data association relationship by combining node identification information and acquisition time identifier, and generates corresponding local association feature set.

[0012] Step S121: Each edge node locally extracts the data association dimension from the received temperature sensing data, and obtains the node location association attribute corresponding to the node identification information and the time series association attribute corresponding to the acquisition time identifier.

[0013] In this embodiment, upon receiving temperature sensing data, each edge node immediately activates its local data preprocessing module. This module first calls the string parsing submodule to parse the header of the temperature sensing data, extracting a 16-bit node identifier and a 12-bit acquisition time identifier. Subsequently, the data preprocessing module queries the locally stored node configuration database based on the extracted node identifier. This database stores the mapping relationship between the node identifier information and node location association attributes of all edge nodes. The node location association attributes specifically include the X-axis coordinates, Y-axis coordinates, and altitude of the edge node's installation location in the park coordinate system. Simultaneously, the data preprocessing module converts the extracted acquisition time identifier into a timestamp format. The timestamp format is the number of seconds calculated from a fixed start time. This timestamp determines the specific location of the temperature sensing data in the time series, thus forming a time series association attribute. The time series association attribute also includes the time interval between the data and the previous data, the time interval category to which the data belongs, and other information. The time interval category is divided into multiple fixed intervals based on different time periods of the day.

[0014] Step S122: Determine the spatial range of adjacent nodes based on the node location association attribute, and filter the temperature sensing data transmitted by the spatially adjacent nodes as the adjacent node temperature data. The adjacent node temperature data carries the corresponding node identification information and acquisition time identifier.

[0015] In this embodiment, after obtaining its own node location association attributes, the edge node calls the spatial calculation submodule to calculate the spatial distance between itself and all other edge nodes. The spatial distance is calculated based on the X-axis coordinates, Y-axis coordinates, and altitude in the park coordinate system, using a three-dimensional spatial distance calculation formula. Subsequently, the spatial calculation submodule compares the calculated spatial distance with a preset spatial distance threshold, identifying edge nodes with spatial distances less than or equal to the threshold as spatially adjacent nodes. The spatial distance threshold is set to fifty meters based on the actual layout of the park and the requirements of temperature sensing. After determining the range of spatially adjacent nodes, the edge node calls the data filtering submodule to filter the temperature sensing data transmitted by these spatially adjacent nodes from the locally received and stored temperature sensing data as adjacent node temperature data. During the filtering process, precise matching is performed based on node identification information. Only temperature sensing data whose node identification information falls within the range of spatially adjacent nodes is selected. The selected adjacent node temperature data is stored separately in a temporary data buffer, retaining its original node identification information and acquisition time identifier.

[0016] Step S123: Align the temperature sensing data of this edge node with the temperature data of adjacent nodes according to the acquisition time identifier to form a time-synchronized data group.

[0017] In this embodiment, the edge node invokes the time alignment submodule to process the temperature sensing data of its own edge node stored locally and the temperature data of adjacent nodes in the temporary data buffer. The time alignment submodule first extracts the acquisition time identifier of all data and converts it into a timestamp format, then sorts all data according to the order of the timestamps. After sorting, the time alignment submodule uses the timestamp of the edge node's temperature sensing data as a reference and searches for data in the adjacent node temperature data whose timestamp differs from the reference timestamp by a preset time difference threshold, which is set to one second. For each reference timestamp, if a matching adjacent node temperature data is found, the temperature sensor data of this edge node is combined with the corresponding adjacent node temperature data to form a time-synchronized data group. If no matching adjacent node temperature data is found, a virtual adjacent node temperature data is generated using linear interpolation. The temperature value of the generated virtual data is the linear interpolation result of the temperature values ​​of the two valid data before and after the adjacent node. The acquisition time identifier is the time identifier corresponding to the reference timestamp, and the node identifier information is the identifier information of the adjacent node. Then, the temperature sensor data of this edge node is combined with the generated virtual data to form a time-synchronized data group.

[0018] Step S124: Perform feature linkage modeling on the time-synchronized data group, construct spatial correlation constraints by combining node location correlation attributes, and mine the correlation features between the temperature data of this edge node and adjacent nodes based on the constraints.

[0019] Step S1241: Extract features from the temperature data of the local edge node and adjacent nodes in the time-synchronized data group to obtain basic temperature features, which reflect the core attributes of the temperature data.

[0020] In this embodiment, after the time alignment submodule completes the construction of the time-synchronized data group, it transmits the data group to the feature extraction module. The feature extraction module first calls the basic feature extraction submodule to process the temperature sensing data of the current edge node and the temperature data of adjacent nodes in the time-synchronized data group. The basic feature extraction submodule analyzes the temperature value portion of each temperature sensing data and extracts basic temperature features. These basic temperature features specifically include the instantaneous temperature value, the five-minute moving average of the temperature, the ten-minute moving maximum value of the temperature, the ten-minute moving minimum value of the temperature, and the rate of temperature change. The rate of temperature change is calculated by dividing the difference between the current temperature value and the temperature value of the previous data by the time interval. These basic temperature features can reflect the core attributes of the temperature data from different perspectives, providing a foundation for subsequent correlation feature mining.

[0021] Step S1242: Determine the spatial distance and orientation relationships between this edge node and each adjacent node based on the node location association attributes, and quantify the spatial distance and orientation relationships between this edge node and each adjacent node into spatial association parameters.

[0022] In this embodiment, while extracting basic temperature features, the feature extraction module calls the spatial correlation quantization submodule to process the spatial relationship between the current edge node and its neighboring nodes. The spatial correlation quantization submodule first calculates the spatial distance between the current edge node and its neighboring nodes based on their node position association attributes, using a three-dimensional spatial distance calculation formula. Then, it calculates the azimuth angle of each neighboring node relative to the current edge node based on their X-axis and Y-axis coordinates, using a planar coordinate system angle calculation method. Next, the submodule inputs the calculated spatial distance and azimuth angle into a preset quantization model. This model maps the spatial distance and azimuth angle to corresponding quantized values ​​based on their ranges. The quantized value of the spatial distance is inversely proportional to the spatial distance, and the quantized value of the azimuth angle is divided according to a preset azimuth interval. Finally, the quantized values ​​of the spatial distance and azimuth angle are weighted and combined to obtain spatial correlation parameters. The weights of the weighted combination are set according to the actual layout of the park and the temperature propagation characteristics.

[0023] Step S1243: Construct spatial association constraints by combining spatial association parameters. The spatial association constraints clarify the spatial association logic that should be satisfied between the temperature data of this edge node and adjacent nodes.

[0024] In this embodiment, after obtaining the spatial association parameters, the feature extraction module calls the constraint construction submodule to construct spatial association constraints. The constraint construction submodule first divides the spatial association parameters into multiple intervals based on their value range. Each interval corresponds to a spatial association strength level, which is divided into three levels: strong association, medium association, and weak association. Subsequently, the constraint construction submodule formulates corresponding spatial association logic rules for each spatial association strength level. For example, for adjacent nodes of the strong association level, the temperature difference between the current edge node and its adjacent node should be less than or equal to a preset difference threshold, and the difference in temperature change rate should be less than or equal to a preset change rate difference threshold; for adjacent nodes of the medium association level, the temperature difference between the current edge node and its adjacent node should be less than or equal to a preset medium difference threshold, and the difference in temperature change rate should be less than or equal to a preset medium change rate difference threshold; for adjacent nodes of the weak association level, the temperature difference between the current edge node and its adjacent node should be less than or equal to a preset weak difference threshold, and the difference in temperature change rate should be less than or equal to a preset weak change rate difference threshold. Finally, the constraint construction submodule combines these spatial association logic rules to form spatial association constraints.

[0025] Step S1244: Substitute the basic temperature features of this edge node and the basic temperature features of adjacent nodes into the spatial association constraints, perform association feature mining, and identify feature association patterns that conform to the spatial association logic.

[0026] In this embodiment, the feature extraction module calls the association feature mining submodule to process the base temperature features of the current edge node and the base temperature features of adjacent nodes. The association feature mining submodule first combines the base temperature features of the current edge node and the base temperature features of adjacent nodes to form a feature combination vector. Then, the association feature mining submodule substitutes the feature combination vector into the spatial association constraints, checking one by one whether the feature combination vector satisfies each rule in the spatial association constraints. For feature combination vectors that satisfy the rules, the association feature mining submodule records the combination relationship between the current edge node and adjacent nodes, the spatial association strength level, and the specific value of the feature combination vector. Next, the association feature mining submodule performs statistical analysis on the recorded information to identify frequently occurring feature combination patterns; these patterns are the feature association patterns that conform to the spatial association logic.

[0027] Step S1245: Perform feature quantization on the identified feature association patterns and convert them into representative association feature vectors. The dimension of the association feature vectors is consistent with the dimension of the basic temperature features, forming the association features between the temperature data of this edge node and adjacent nodes.

[0028] In this embodiment, the feature extraction module calls the feature quantization submodule to process the identified feature association patterns. The feature quantization submodule first analyzes the feature combination vector in each feature association pattern to determine the value range and distribution of each feature dimension within that pattern. Then, it uses a preset quantization method to map the value of each feature dimension to its corresponding quantized value. Quantization methods include normalization and discretization. Next, the feature quantization submodule combines the quantized values ​​of each feature dimension to form an associated feature vector. The dimensions of the associated feature vector are consistent with the dimensions of the base temperature feature, with each dimension corresponding to a quantized value of the base temperature feature. Finally, the feature quantization submodule verifies the associated feature vector to ensure it accurately reflects the information of the feature association pattern. The verified associated feature vector is the association feature between the temperature data of this edge node and its neighboring nodes.

[0029] Step S125: Integrate the inherent features of the temperature sensing data of this edge node with the associated features obtained by mining to form a local associated feature set containing spatial association information and temporal synchronization information. The feature dimensions of the local associated feature set are adapted to the local processing capabilities of each edge node.

[0030] In this embodiment, after obtaining the intrinsic and correlated features of the temperature sensing data of the current edge node, the feature extraction module calls the feature integration submodule to integrate these features. The feature integration submodule first performs feature selection on the intrinsic and correlated features using a preset feature importance evaluation method, including variance analysis and mutual information calculation. Based on the evaluation results, features with higher importance are retained, while redundant features with lower importance are removed. Subsequently, the feature integration submodule performs feature encoding on the retained features, using one-hot encoding, label encoding, and other methods to convert non-numerical features into numerical features. Next, the feature integration submodule adjusts the feature dimensions according to the local processing capabilities of each edge node. Local processing capabilities include the edge node's CPU processing power, memory capacity, and storage capacity. The feature dimension adjustment uses feature dimensionality reduction methods, including principal component analysis and linear discriminant analysis, to ensure that the adjusted feature dimensions can be successfully processed and stored locally at the edge node. Finally, the feature integration submodule combines the adjusted features to form a local correlated feature set containing spatial correlation information and temporal synchronization information.

[0031] Step S130: Transmit the local associated feature set of each edge node to the edge collaboration node, and integrate the global features through the cross-node feature fusion mechanism of the edge collaboration node to generate a global temperature state representation with inter-node correlation.

[0032] Step S131: Each edge node transmits its local associated feature set to the edge collaboration node according to the preset transmission protocol. The edge collaboration node receives all local associated feature sets and performs transmission integrity verification.

[0033] In this embodiment, after generating a local associated feature set, each edge node calls the data transmission submodule to process the local associated feature set. The data transmission submodule first packages the local associated feature set according to a preset transmission protocol. During packaging, a frame header, frame trailer, data length, and checksum are added. The frame header contains a synchronization word and node identification information, the frame trailer contains an end word, the data length is the number of bytes in the local associated feature set, and the checksum is calculated using a preset cyclic redundancy check algorithm. Subsequently, the data transmission submodule transmits the packaged local associated feature set to the edge collaboration node via network communication. During the receiving process, the edge collaboration node calls the data receiving submodule to receive and store the local associated feature set transmitted by each edge node. The data receiving submodule checks the frame header, frame trailer, data length, and checksum of the received data in real time. The synchronization word in the frame header must match the preset synchronization word, the end word in the frame trailer must match the preset end word, the data length must match the data length recorded in the frame header, and the checksum must match the result calculated by the preset algorithm. Only data that simultaneously meets all four conditions is considered to have been transmitted completely and is received and stored.

[0034] Step S132: Unify the feature dimensions of the verified local association feature sets, and construct global feature dimension mapping rules based on the feature description information of the local association feature sets of each edge node.

[0035] In this embodiment, after completing the transmission integrity verification, the edge collaborative node calls the feature dimension unification module to process the verified local associated feature sets. The feature dimension unification module first calls the feature description information extraction submodule to parse the metadata of each local associated feature set, extracting feature description information, including feature name, feature type, feature dimension, and feature value range. Subsequently, the feature dimension unification module performs statistical analysis on the extracted feature description information to determine the common feature names and feature types of all local associated feature sets, as well as the differences in feature dimension and feature value range corresponding to each feature name. Next, the feature dimension unification module constructs global feature dimension mapping rules based on the statistical analysis results. The global feature dimension mapping rules specifically include a unified mapping relationship for feature names, a unified conversion rule for feature types, a unified adjustment method for feature dimensions, and a unified standardization rule for feature value ranges. The unified mapping relationship for feature names is used to map the local feature names of different edge nodes to a unified global feature name. The unified conversion rule for feature types is used to convert the local feature types of different edge nodes to a unified global feature type. The unified adjustment method for feature dimensions is used to adjust the local feature dimensions of different edge nodes to a unified global feature dimension. The unified standardization rule for feature value ranges is used to standardize the local feature value ranges of different edge nodes to a unified global feature value range.

[0036] Step S133: Perform dimensional transformation on the local associated feature set of each edge node according to the global feature dimension mapping rule to obtain a standardized feature set with a unified dimension.

[0037] In this embodiment, the feature dimension unification module for edge collaborative nodes calls the dimension transformation submodule to process the local associated feature set of each edge node. First, the dimension transformation submodule replaces each feature name in the local associated feature set with its corresponding global feature name according to the unified feature name mapping relationship in the global feature dimension mapping rules. Then, according to the unified feature type conversion rules, the dimension transformation submodule converts each feature type in the local associated feature set to its corresponding global feature type, such as converting integer features to floating-point features and string features to numeric features. Next, the dimension transformation submodule adjusts each feature dimension in the local associated feature set according to the unified feature dimension adjustment method. The adjustment method includes feature splitting and feature merging, such as splitting a high-dimensional feature into multiple low-dimensional features and merging multiple low-dimensional features into a single high-dimensional feature, ensuring that the adjusted feature dimensions are consistent with the global feature dimensions. Finally, the dimension transformation submodule will standardize each feature value in the local associated feature set according to the unified standardization rules of the feature value range. The standardization process adopts methods such as min-max standardization and Z-score standardization to standardize the feature values ​​to a unified value range, thereby obtaining a standardized feature set with a unified dimension.

[0038] Step S134: Based on the node identification information and node location association attributes of each edge node, construct the node spatial topology structure and associate and bind the standardized feature set with the node spatial topology structure.

[0039] In this embodiment, the edge collaboration node calls the node topology construction module to process the information of each edge node. The node topology construction module first queries the node configuration database based on the node identifier information of each edge node to obtain the corresponding node location association attributes. Then, the node topology construction module calls the topology structure generation submodule to construct a node spatial topology structure based on the obtained node location association attributes. The topology structure generation submodule uses a graph structure to represent the node spatial topology structure. Each node in the graph structure represents an edge node, and the node's attributes include node identifier information and node location association attributes. Each edge in the graph structure represents the spatial adjacency relationship between two edge nodes, and the edge weight represents the spatial association strength between the two edge nodes, which is calculated based on the spatial distance between the two edge nodes. Next, the node topology construction module calls the association binding submodule to associate and bind the standardized feature set with the node spatial topology structure. The association binding method is to add the standardized feature set of each edge node as an attribute of the corresponding graph node to the graph structure, ensuring that each node in the graph structure contains the corresponding standardized feature set information.

[0040] Step S135: Through the cross-node feature fusion mechanism, global information aggregation is performed on the standardized feature set after association and binding, the hidden association relationships between different edge nodes are mined, the spatial topological association information and the feature information of the standardized feature set are integrated, and a global temperature state representation is generated.

[0041] Step S1351: Determine the global association weight of each edge node based on the node spatial topology. The global association weight is related to the spatial importance of the edge node and the reliability of data transmission.

[0042] In this embodiment, the edge collaborative nodes invoke the global weight calculation module to process the node spatial topology. The global weight calculation module first invokes the spatial location importance assessment submodule to evaluate the spatial location importance of each edge node based on its node location association attributes. This evaluation uses preset assessment indicators, including the distance between the edge node and the park center, the node density around the edge node, and the importance of the area where the edge node is located. A spatial location importance score is calculated based on the values ​​of these indicators. Next, the global weight calculation module invokes the data transmission reliability assessment submodule to evaluate the data transmission reliability of each edge node based on its historical data transmission records. This evaluation uses preset assessment indicators, including data transmission success rate, data transmission latency, and data transmission packet loss rate. A data transmission reliability score is calculated based on the values ​​of these indicators. Finally, the global weight calculation module weights the spatial location importance score and the data transmission reliability score to obtain the global association weight for each edge node. The weights for this weighted combination are set according to the actual needs of the park and the application scenario.

[0043] Step S1352: Weight the standardized feature set of each edge node according to the global association weight to obtain the weighted standardized feature set, keeping the feature dimension unchanged during the weighting process.

[0044] In this embodiment, the edge collaboration node invokes the feature weighting module to process the standardized feature sets of each edge node. The feature weighting module first obtains the global association weights of each edge node, and then weights the value of each feature dimension in the standardized feature set of each edge node. The weighting method involves multiplying the feature dimension value by the corresponding global association weight to obtain the weighted feature value. During the weighting process, the feature dimensions remain unchanged to ensure that the weighted standardized feature set has the same feature dimensions as the original standardized feature set.

[0045] Step S1353: Perform neighborhood association mining on the weighted standardized feature set. Taking the weighted standardized feature set corresponding to each edge node as the center, mine the association pattern of the weighted standardized feature set corresponding to each edge node within the preset range of the neighboring nodes in the node space topology.

[0046] In this embodiment, the edge collaborative node invokes the neighborhood association mining module to process the weighted standardized feature set. The neighborhood association mining module first determines the neighborhood node range of each edge node based on the node spatial topology. The neighborhood node range is determined using a preset neighborhood radius, which is set according to the actual layout of the park and the temperature sensing requirements. Then, the neighborhood association mining module uses the weighted standardized feature set corresponding to each edge node as the center and searches for neighboring nodes within the neighborhood radius in the node spatial topology. Next, the neighborhood association mining module performs feature comparison and analysis on the weighted standardized feature sets of the central edge node and the weighted standardized feature sets of the neighboring nodes. Feature comparison and analysis uses a preset similarity measurement method, including Euclidean distance calculation and cosine similarity calculation. Based on the similarity measurement results, the module identifies the association patterns between features, including feature similarity patterns, feature complementarity patterns, and feature correlation patterns.

[0047] Step S1354: Construct a cross-node association feature matrix based on the association patterns obtained from mining. The row dimension of the cross-node association feature matrix corresponds to the number of edge nodes, and the column dimension corresponds to the feature dimension of the standardized feature set.

[0048] In this embodiment, the edge collaboration nodes invoke the feature matrix construction module to process the mined association patterns. The feature matrix construction module first determines the row and column dimensions of the cross-node association feature matrix. The row dimension corresponds to the number of edge nodes, with each row representing one edge node. The column dimension corresponds to the feature dimensions of the standardized feature set, with each column representing one feature dimension. Subsequently, the feature matrix construction module analyzes the association pattern of each edge node to determine the value of each feature dimension within the association pattern. Next, the feature matrix construction module fills the value of each feature dimension of each edge node into the corresponding position in the cross-node association feature matrix, ensuring that each element in the matrix accurately reflects the association pattern information of the corresponding edge node and feature dimension.

[0049] Step S1355: Perform feature compression on the cross-node association feature matrix, retain the core association feature information, combine the global topological features of the node spatial topology, and fuse the compressed matrix with the global topological features to generate a global temperature state representation.

[0050] In this embodiment, the edge collaborative node invokes the feature compression and fusion module to process the cross-node association feature matrix. First, the feature compression and fusion module calls the feature compression submodule to compress the cross-node association feature matrix using a preset feature compression algorithm, including principal component analysis and singular value decomposition. This compression removes redundant information from the matrix while retaining core association features. Next, the feature compression and fusion module calls the global topology feature extraction submodule to extract global topology features from the node space topology. These global topology features include node distribution density, connection strength between nodes, and node clustering. Then, the feature compression and fusion module fuses the compressed cross-node association feature matrix with the global topology features using methods such as feature concatenation and weighted summation to combine the compressed matrix and global topology features into a new feature set. Finally, the feature compression and fusion module performs feature encoding and standardization on the fused feature set to generate a global temperature state representation with inter-node associations.

[0051] Step S140: Call the pre-trained temperature state analysis model to make a correlation judgment on the global temperature state representation, and generate the temperature state correlation judgment result by combining the temporal change pattern of temperature sensing data and the spatial correlation pattern between nodes.

[0052] Step S141: Adapt the feature format of the global temperature state representation into the feature input layer of the temperature state analysis model so that the global temperature state representation meets the model input requirements.

[0053] In this embodiment, the edge collaboration node invokes the model input adaptation module to process the global temperature state representation. The model input adaptation module first obtains the input requirements of the pre-trained temperature state analysis model, including the shape, data type, and value range of the input data. Then, the model input adaptation module adjusts the shape of the global temperature state representation using methods such as dimensional expansion and dimensional compression to ensure the adjusted shape matches the model input requirements. Next, the model input adaptation module converts the data type of the global temperature state representation using methods such as type casting and data format conversion to ensure the converted data type matches the model input requirements. Finally, the model input adaptation module standardizes the value range of the global temperature state representation using a preset standardization method to ensure the standardized value range matches the model input requirements.

[0054] Step S142: The spatial correlation modeling layer of the temperature state analysis model is used to strengthen the spatial correlation of the adapted global temperature state representation. A spatial correlation constraint model is constructed by combining the spatial correlation rules between nodes. The spatial correlation features in the global temperature state representation are strengthened based on the constraint model.

[0055] Step S1421: Extract initial spatial correlation features from the adapted global temperature state representation. The initial spatial correlation features directly reflect the spatial correlation relationship of temperature data of each edge node.

[0056] In this embodiment, the spatial correlation modeling layer of the temperature state analysis model first calls the feature extraction submodule to process the adapted global temperature state representation. The feature extraction submodule uses a preset feature extraction algorithm, including convolutional neural networks, recurrent neural networks, etc., to extract initial spatial correlation features from the adapted global temperature state representation that can directly reflect the spatial correlation of temperature data of each edge node. The initial spatial correlation features include the spatial distribution features, spatial clustering features, and spatial correlation features of temperature data of each edge node.

[0057] Step S1422: Determine the spatial association influencing factor based on the spatial association pattern between nodes. The spatial association influencing factor is used to quantify the strength of spatial association between different edge nodes.

[0058] In this embodiment, the spatial correlation modeling layer of the temperature state analysis model calls the influence factor calculation submodule to process the spatial correlation patterns between nodes. The influence factor calculation submodule first obtains the spatial correlation patterns between nodes, including the relationship between spatial distance and correlation strength, the relationship between spatial orientation and correlation strength, and the relationship between spatial clustering and correlation strength. Subsequently, the influence factor calculation submodule formulates a method for calculating the spatial correlation influence factor based on the spatial correlation patterns between nodes. This method includes influence factor calculation based on spatial distance, influence factor calculation based on spatial orientation, and influence factor calculation based on spatial clustering. Next, the influence factor calculation submodule analyzes the spatial relationship between each edge node and other edge nodes, and calculates the corresponding spatial correlation influence factor according to the calculation method. The value of the spatial correlation influence factor ranges from 0 to 1, with a larger value indicating a higher spatial correlation strength.

[0059] Step S1423: Construct a spatial correlation constraint model based on spatial correlation influencing factors. The spatial correlation constraint model contains multiple spatial correlation constraints, and each condition corresponds to a set of spatial correlation rules for edge nodes.

[0060] In this embodiment, the spatial correlation modeling layer of the temperature state analysis model calls the constraint model construction submodule to process the spatial correlation influencing factors. The constraint model construction submodule first divides the spatial correlation influencing factors into multiple intervals based on their value ranges, with each interval corresponding to a spatial correlation strength level. Then, the constraint model construction submodule formulates corresponding spatial correlation constraints for each spatial correlation strength level, including constraints on temperature value differences, temperature change rate differences, and temperature distribution consistency. Next, the constraint model construction submodule combines these spatial correlation constraints to form a spatial correlation constraint model, ensuring that each constraint in the model accurately reflects the corresponding spatial correlation rule.

[0061] Step S1424: Input the initial spatial association features into the spatial association constraint model, and filter and strengthen the initial spatial association features through the constraint conditions, retain the effective spatial association features that conform to the spatial association rules, and enhance the feature response strength of the effective spatial association features.

[0062] In this embodiment, the spatial correlation modeling layer of the temperature state analysis model calls the feature filtering and enhancement submodule to process the initial spatial correlation features. The feature filtering and enhancement submodule first inputs the initial spatial correlation features into the spatial correlation constraint model and checks whether each initial spatial correlation feature satisfies the constraints in the model. For initial spatial correlation features that satisfy the constraints, the feature filtering and enhancement submodule retains them as valid spatial correlation features; for initial spatial correlation features that do not satisfy the constraints, the feature filtering and enhancement submodule removes them. Subsequently, the feature filtering and enhancement submodule performs feature response intensity enhancement processing on the retained valid spatial correlation features. Enhancement methods include feature amplification and feature weighting, thereby increasing the importance of valid spatial correlation features in subsequent processing.

[0063] Step S1425: Normalize the enhanced effective spatial correlation features to eliminate the dimensional differences between different spatial correlation features and generate standardized spatial correlation features. The standardized spatial correlation features are consistent with the feature dimensions of the global temperature state representation.

[0064] In this embodiment, the spatial correlation modeling layer of the temperature state analysis model calls the feature normalization submodule to process the enhanced effective spatial correlation features. The feature normalization submodule first determines the value range of the effective spatial correlation features, and then normalizes them using a preset normalization method, including min-max normalization and Z-fraction normalization. This normalization process eliminates dimensional differences between different spatial correlation features, making the features comparable. During normalization, the feature dimensions remain unchanged, ensuring that the standardized spatial correlation features are consistent with the feature dimensions of the global temperature state representation.

[0065] Step S143: The global temperature state representation after strengthening spatial correlation features is subjected to temporal correlation mining through the temporal correlation modeling layer of the temperature state analysis model. The temporal correlation features are extracted by combining the temporal change pattern of temperature sensing data. The temporal correlation features reflect the change correlation of temperature data under different acquisition time markers.

[0066] Step S1431: Perform time-series segmentation on the global temperature state representation after enhancing spatial correlation features, dividing it into multiple continuous time-series segment features according to the order of acquisition time identifiers.

[0067] In this embodiment, the temporal correlation modeling layer of the temperature state analysis model calls the temporal segmentation submodule to process the global temperature state representation after enhancing spatial correlation features. The temporal segmentation submodule first extracts the acquisition time identifiers from the global temperature state representation and sorts them according to their chronological order. Then, it divides the sorted acquisition time identifiers into multiple consecutive time intervals according to a preset time interval, with each time interval corresponding to a temporal segment. Next, the temporal segmentation submodule combines the feature information of the global temperature state representation within each time interval to form corresponding temporal segment features, ensuring that each temporal segment feature accurately reflects the temperature data and spatial correlation feature information within the corresponding time interval.

[0068] Step S1432: Calculate the feature change between adjacent time segment features. The feature change reflects the numerical change of the corresponding feature dimension in the features of two consecutive time segments.

[0069] In this embodiment, the time-series correlation modeling layer of the temperature state analysis model calls the feature change calculation submodule to process the time-series segment features. The feature change calculation submodule first aligns the features of two adjacent time-series segments to ensure that the feature dimensions of the two time-series segment features are consistent. Subsequently, the feature change calculation submodule calculates the difference between the values ​​of each feature dimension. The difference calculation method is to subtract the feature dimension value of the previous time-series segment feature from the feature dimension value of the later time-series segment feature to obtain the corresponding feature change. The feature change accurately reflects the numerical change of the corresponding feature dimension in two consecutive time-series segment features.

[0070] Step S1433: Determine the temporal correlation threshold based on the temporal variation pattern of temperature sensing data. The temporal correlation threshold is used to determine whether there is an effective temporal correlation between features of adjacent temporal segments.

[0071] In this embodiment, the temporal correlation modeling layer of the temperature state analysis model calls the threshold determination submodule to process the temporal variation patterns of the temperature sensing data. The threshold determination submodule first obtains the temporal variation patterns of the temperature sensing data, including periodic, trend-based, and random temperature variations. Then, it formulates a calculation method for the temporal correlation threshold based on these patterns, including statistical analysis and machine learning methods. Next, the threshold determination submodule determines the temporal correlation threshold based on the calculation method. The value of the temporal correlation threshold is used to determine whether there is a valid temporal correlation between features of adjacent time segments. When the feature change is less than or equal to the temporal correlation threshold, a valid temporal correlation is considered to exist.

[0072] Step S1434: Compare the feature change with the time series correlation threshold, and select the valid time series segment feature pairs whose feature change meets the threshold requirements.

[0073] In this embodiment, the temporal correlation modeling layer of the temperature state analysis model calls the feature comparison and filtering submodule to process the feature changes and temporal correlation thresholds. The feature comparison and filtering submodule first compares the calculated feature changes with the temporal correlation thresholds one by one. Then, it filters out temporal segment feature pairs whose feature changes are less than or equal to the temporal correlation thresholds. These feature pairs are considered valid temporal segment feature pairs, accurately reflecting the continuous changes and correlations of temperature data over time.

[0074] Step S1435: Extract the correlation patterns between valid time-series feature pairs, integrate the correlation patterns of all valid time-series feature pairs, and generate time-series correlation features that reflect the correlation of time-series changes in global temperature state.

[0075] In this embodiment, the temporal correlation modeling layer of the temperature state analysis model calls the correlation pattern extraction and integration submodule to process valid temporal segment feature pairs. The correlation pattern extraction and integration submodule first extracts correlation patterns for each valid temporal segment feature pair using a preset pattern recognition algorithm, including clustering analysis and association rule mining. Subsequently, the correlation pattern extraction and integration submodule integrates the extracted correlation patterns using methods such as pattern merging and pattern filtering. Next, the correlation pattern extraction and integration submodule converts the integrated correlation patterns into temporal correlation features, which accurately reflect the temporal change correlation information of the global temperature state characterization.

[0076] Step S144: Input the spatial correlation features and temporal correlation features into the feature fusion layer of the temperature state analysis model to perform cross-dimensional feature fusion and generate comprehensive correlation features that have both spatial and temporal correlation.

[0077] In this embodiment, the feature fusion layer of the temperature state analysis model calls the cross-dimensional feature fusion submodule to process spatial and temporal correlation features. The cross-dimensional feature fusion submodule first aligns the spatial and temporal correlation features to ensure that their feature dimensions are consistent. Then, it uses a preset feature fusion method to fuse the spatial and temporal correlation features, including feature concatenation, weighted summation, and concatenation. Next, it performs feature encoding and standardization on the fused features to generate comprehensive correlation features that possess both spatial and temporal correlation.

[0078] Step S145: Call the judgment output layer of the temperature state analysis model to classify the correlation state of the comprehensive correlation features and generate a temperature state correlation judgment result containing the correlation state category and correlation confidence.

[0079] In this embodiment, the output layer of the temperature state analysis model calls the correlation state classification submodule to process the comprehensive correlation features. The correlation state classification submodule first uses a preset classification algorithm to classify the correlation state of the comprehensive correlation features. Classification algorithms include support vector machines, decision trees, and neural networks. Then, the correlation state classification submodule determines the correlation state category based on the classification results. Correlation state categories include strongly correlated, moderately correlated, weakly correlated, and no correlated states. Next, the correlation state classification submodule calculates the correlation confidence score for each correlation state category. The correlation confidence score is calculated using a preset confidence score evaluation method, including probability calculation and likelihood estimation. Finally, the correlation state classification submodule combines the correlation state category and the correlation confidence score to generate the temperature state correlation judgment result.

[0080] Step S150: Generate temperature data processing instructions adapted to each edge node based on the temperature state association judgment result, and distribute them to the corresponding edge nodes. Each edge node then performs local temperature sensing data processing operations according to the instructions.

[0081] Step S151: Analyze the association state category and association confidence in the temperature state association judgment result, and determine the temperature data processing requirements corresponding to each edge node. The processing requirements are directly related to the association state category.

[0082] In this embodiment, the edge collaboration node invokes the processing requirement determination module to process the temperature state correlation judgment results. The processing requirement determination module first parses the temperature state correlation judgment results, extracting the correlation state category and correlation confidence information. Subsequently, the processing requirement determination module determines the temperature data processing requirements for each edge node based on the correlation state category. Different correlation state categories correspond to different processing requirements. For example, processing requirements for a strong correlation state include detailed data analysis, real-time data monitoring, and data alerts; processing requirements for a medium correlation state include routine data analysis and periodic data monitoring; processing requirements for a weak correlation state include simple data analysis and occasional data monitoring; and processing requirements for a no-correlation state include data storage and data backup. The determination of processing requirements also considers the correlation confidence information; processing requirements with higher correlation confidence are given higher priority.

[0083] Step S152: Based on the local processing capabilities of each edge node and the temperature data processing requirements, construct processing instruction generation rules. The processing instruction generation rules include instruction operation types and operation parameters corresponding to different processing requirements.

[0084] In this embodiment, the edge collaboration nodes invoke the instruction rule construction module to process the local processing capabilities and temperature data processing requirements of each edge node. The instruction rule construction module first obtains the local processing capability information of each edge node, including CPU processing power, memory capacity, storage capacity, and network bandwidth. Subsequently, the instruction rule construction module formulates processing instruction generation rules based on the temperature data processing requirements and local processing capability information. These rules include instruction operation types and parameters corresponding to different processing requirements. Instruction operation types include data filtering, data aggregation, data storage, data transmission, and data alerts. Operation parameters include filtering conditions, aggregation methods, storage paths, transmission protocols, and alert thresholds. The formulation of processing instruction generation rules must ensure that the instruction operation types and parameters are compatible with the local processing capabilities of each edge node to avoid situations where edge nodes are unable to execute instructions.

[0085] Step S153: Generate a unique temperature data processing instruction for each edge node according to the processing instruction generation rules. The instruction includes node identification information, processing operation type and corresponding operation parameters, so that the instruction is accurately matched with the edge node.

[0086] In this embodiment, the edge collaborative node invokes the instruction generation module to process the instruction generation rules. The instruction generation module first determines the instruction operation type and operation parameters corresponding to each edge node based on the instruction generation rules. Then, the instruction generation module generates a unique temperature data processing instruction for each edge node. The instruction format adopts a preset instruction format specification, which includes an instruction header, instruction body, and instruction tail. The instruction header contains node identification information and instruction type; the instruction body contains the processing operation type and corresponding operation parameters; and the instruction tail contains a checksum and an end character. The instruction generation module ensures that the node identification information in the instruction matches the identification information of the corresponding edge node, and that the processing operation type and operation parameters match the specifications in the instruction generation rules, enabling precise matching between the instruction and the edge node.

[0087] Step S154: The temperature data processing command is distributed to the corresponding edge node according to the node identification information through the command distribution module of the edge collaborative node. The command is encrypted during the distribution process to ensure transmission security.

[0088] In this embodiment, the edge collaborative node invokes the instruction distribution module to process the temperature data processing instruction. The instruction distribution module first encrypts the generated temperature data processing instruction using a preset encryption algorithm, including symmetric and asymmetric encryption algorithms. An encryption key is generated during the encryption process and must be securely stored for decryption by the edge node. Subsequently, the instruction distribution module distributes the encrypted temperature data processing instruction to the corresponding edge node based on the node identification information in the instruction. The distribution process uses a preset transmission protocol that supports the transmission of encrypted data to ensure the security of the instruction during transmission.

[0089] Step S155: After receiving the temperature data processing instruction, each edge node parses it, extracts the processing operation type and operation parameters, and performs the local temperature sensing data processing operation according to the processing operation type and operation parameters.

[0090] For example, in step S1551: the instruction receiving module of each edge node receives the temperature data processing instruction, decrypts the encrypted instruction through the built-in decryption algorithm, and obtains the temperature data processing instruction in plaintext form.

[0091] In this embodiment, the instruction receiving module of each edge node listens to the network port in real time to receive temperature data processing instructions distributed by the edge collaborative nodes. The instruction receiving module has a built-in decryption algorithm that corresponds to the encryption algorithm used by the edge collaborative nodes. The instruction receiving module uses a preset decryption key to decrypt the encrypted instructions, obtaining the plaintext temperature data processing instructions. During the decryption process, an integrity check is performed on the decrypted instructions to ensure that they have not been tampered with during transmission.

[0092] Step S1552: Call the instruction parsing module of the edge node to perform syntax parsing and semantic analysis on the plaintext instruction, and extract the node identification information, processing operation type and corresponding operation parameters in the instruction.

[0093] In this embodiment, each edge node invokes the instruction parsing module to process plaintext temperature data processing instructions. The instruction parsing module first performs syntactic parsing on the plaintext instructions, using preset syntactic rules. These rules include the instruction's structure, keywords, and data type, and the module checks the syntactic correctness of the instructions according to these rules. Subsequently, the instruction parsing module performs semantic analysis on syntactically correct instructions, using preset semantic rules. These rules include the meaning of the instruction's operation type and the range of operation parameter values, and the module extracts node identification information, processing operation type, and corresponding operation parameters from the instructions based on these semantic rules.

[0094] Step S1553: Compare and verify the extracted node identification information with the preset identification information of this edge node to make the temperature data processing instruction a dedicated instruction adapted to this edge node.

[0095] In this embodiment, each edge node calls the identifier verification module to process the extracted node identifier information. The identifier verification module first obtains the preset identifier information of its own edge node, which is stored in the edge node's local configuration file. Then, the identifier verification module compares the extracted node identifier information with the preset identifier information using a string matching method. Only when the two are completely identical is the temperature data processing instruction determined to be a unique instruction adapted to this edge node; otherwise, the instruction is discarded.

[0096] Step S1554: Call the local data processing module corresponding to the edge node according to the processing operation type, and configure the operation parameters to the module. The local data processing module corresponds one-to-one with the processing operation type.

[0097] In this embodiment, each edge node invokes the module invocation and parameter configuration module to process the operation type and operation parameters. The module invocation and parameter configuration module first invokes the corresponding local data processing module based on the operation type. Each local data processing module corresponds one-to-one with the operation type, and each module is specifically designed to execute a particular type of operation. Subsequently, the module invocation and parameter configuration module configures the extracted operation parameters into the invoked local data processing module. Parameter configuration uses a preset parameter configuration method to ensure that the parameters can be correctly recognized and used by the local data processing module.

[0098] Step S1555: The local data processing module performs processing operations on the temperature sensing data stored in this edge node based on the configured operation parameters, and references relevant feature information in the local associated feature set to assist in the processing.

[0099] In this embodiment, after configuring the operation parameters, the local data processing module immediately performs processing operations on the temperature sensing data stored in this edge node. The processing operations are executed strictly according to the requirements of the operation parameters. For example, data filtering filters out temperature sensing data that meets the requirements based on filtering conditions; data aggregation performs aggregation calculations on the temperature sensing data according to the aggregation method; data storage stores the processed data to a specified location according to the storage path; data transmission transmits the processed data to a specified target node according to the transmission protocol; and data early warning monitors the temperature sensing data based on an early warning threshold, triggering an early warning mechanism when the data exceeds the threshold. During the execution of the processing operations, the local data processing module references relevant feature information from the local associated feature set to assist in processing. This relevant feature information includes spatial association features and temporal association features, etc. Referencing this feature information improves the accuracy and efficiency of the processing operations.

[0100] Step S1556: After the processing operation is completed, the local data processing module generates a processing result report, which includes the processing operation type, operation parameters, processing data volume, and processing completion status.

[0101] In this embodiment, the local data processing module generates a processing result report immediately after completing the processing operation. The processing result report follows a preset report format specification, which includes a report header, report body, and report footer. The report header contains edge node identification information and the report generation time. The report body includes the processing operation type, operation parameters, processed data volume, and processing completion status, which includes both success and failure. If the processing operation fails, the report body will also include the reason for the failure. The report footer includes a checksum and an end marker. The processing result report can record the execution status of the processing operation in detail, providing a basis for subsequent result analysis and feedback.

[0102] Step S1557: Transmit the processing result report to the edge collaboration node to complete the processing operation of the local temperature sensing data.

[0103] In this embodiment, after generating the processing result report, the local data processing module calls the data transmission module to transmit it to the edge collaboration node. The data transmission module uses a preset transmission protocol to package and transmit the processing result report. The transmission protocol must support reliable transmission of the report data to ensure that the report can be transmitted completely and accurately to the edge collaboration node. After receiving the processing result report, the edge collaboration node stores and analyzes the report, completing the entire local temperature sensing data processing workflow.

[0104] Figure 2 The illustration shows exemplary hardware and software components of a temperature sensing data processing system 100 incorporating edge computing, which can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 may be used in the temperature sensing data processing system 100 incorporating edge computing and to perform the functions described in this application.

[0105] The temperature sensing data processing system 100 incorporating edge computing can be a general-purpose server or a special-purpose server; both can be used to implement the temperature sensing data processing method incorporating edge computing of 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 processing load.

[0106] For example, the edge computing-integrated temperature sensing data processing system 100 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 edge computing-integrated temperature sensing data processing system 100 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 edge computing-integrated temperature sensing data processing system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0107] For ease of explanation, only one processor is described in the temperature sensing data processing system 100 incorporating edge computing. However, it should be noted that the temperature sensing data processing system 100 incorporating edge computing may also include multiple processors, and 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 temperature sensing data processing system 100 incorporating edge computing performs 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 performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0108] 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 temperature sensing data processing method combined with edge computing as described above is implemented.

[0109] 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 processing temperature sensing data combined with edge computing, characterized in that, The method includes: Based on a distributed deployment architecture of edge computing nodes, the system receives temperature sensing data transmitted from each edge node. The temperature sensing data is the ambient temperature sensing data collected in real time by each edge node, and carries node identification information and collection time identifier. Each edge node performs distributed feature linkage processing of temperature sensing data locally, and establishes data association relationships by combining node identification information and acquisition time identifier to generate a corresponding local association feature set. The local associated feature sets of each edge node are transmitted to the edge collaboration node, and the global features are integrated through the cross-node feature fusion mechanism of the edge collaboration node to generate a global temperature state representation with inter-node correlation. The pre-trained temperature state analysis model is invoked to make correlation judgments on the global temperature state representation. The temperature state correlation judgment results are generated by combining the temporal variation pattern of temperature sensing data and the spatial correlation pattern between nodes. Based on the temperature status correlation judgment results, temperature data processing instructions adapted to each edge node are generated and distributed to the corresponding edge nodes. Each edge node then executes the local temperature sensing data processing operation according to the instructions.

2. The method according to claim 1, characterized in that, Each edge node performs distributed feature linkage processing on its local temperature sensing data, establishing data association relationships by combining node identification information and acquisition time identifiers, and generating a corresponding local association feature set, including: Each edge node locally extracts the data association dimension from the received temperature sensing data, and obtains the node location association attribute corresponding to the node identification information and the time series association attribute corresponding to the acquisition time identifier. Based on the node location association attribute, the spatial adjacent node range is determined, and the temperature sensing data transmitted by the spatial adjacent nodes is selected as the adjacent node temperature data. The adjacent node temperature data carries the corresponding node identification information and the acquisition time identifier. The temperature sensing data of this edge node is aligned with the temperature data of adjacent nodes according to the time sequence of the acquisition time identifier to form a time-synchronized data group; Feature linkage modeling is performed on time-synchronized data groups, spatial correlation constraints are constructed by combining node location correlation attributes, and correlation features between temperature data of this edge node and adjacent nodes are mined based on the constraints. By integrating the inherent characteristics of the temperature sensing data of this edge node with the associated characteristics mined out, a local associated feature set containing spatial association information and temporal synchronization information is formed. The feature dimensions of the local associated feature set are adapted to the local processing capabilities of each edge node.

3. The method according to claim 1, characterized in that, The process of transmitting the local association feature sets of each edge node to the edge collaboration node, and integrating global features through the cross-node feature fusion mechanism of the edge collaboration node to generate a global temperature state representation with inter-node correlation includes: Each edge node transmits its local associated feature set to the edge collaboration node according to a preset transmission protocol. The edge collaboration node receives all local associated feature sets and performs transmission integrity verification. The feature dimensions of the verified local associated feature sets are unified, and a global feature dimension mapping rule is constructed based on the feature description information of the local associated feature sets of each edge node. The local associated feature set of each edge node is transformed according to the global feature dimension mapping rule to obtain a standardized feature set with a unified dimension; Based on the node identification information and node location association attributes of each edge node, a node spatial topology is constructed, and a standardized feature set is associated and bound with the node spatial topology. By using a cross-node feature fusion mechanism, global information is aggregated from the standardized feature set after association and binding, hidden relationships between different edge nodes are discovered, spatial topological association information and feature information of the standardized feature set are integrated, and a global temperature state representation is generated.

4. The method according to claim 3, characterized in that, The process involves global information aggregation of the standardized feature set after association and binding through a cross-node feature fusion mechanism, mining hidden relationships between different edge nodes, integrating spatial topological association information and feature information of the standardized feature set, and generating a global temperature state representation, including: The global association weight of each edge node is determined based on the node spatial topology. The global association weight is related to the spatial importance of the edge node and the reliability of data transmission. The standardized feature sets of each edge node are weighted according to the global association weight to obtain a weighted standardized feature set, while keeping the feature dimension unchanged during the weighting process; Neighborhood association mining is performed on the weighted standardized feature set. Taking the weighted standardized feature set corresponding to each edge node as the center, the association pattern of the weighted standardized feature set corresponding to each edge node in the neighboring nodes within a preset range in the node spatial topology is mined. A cross-node association feature matrix is ​​constructed based on the association patterns obtained from mining. The row dimension of the cross-node association feature matrix corresponds to the number of edge nodes, and the column dimension corresponds to the feature dimension of the standardized feature set. The cross-node correlation feature matrix is ​​compressed to retain the core correlation feature information. Combined with the global topological features of the node spatial topology, the compressed matrix is ​​fused with the global topological features to generate a global temperature state representation.

5. The method according to claim 1, characterized in that, The pre-trained temperature state analysis model is invoked to perform correlation judgment on the global temperature state representation. Combining the temporal variation patterns and spatial correlation patterns between nodes in the temperature sensing data, a temperature state correlation judgment result is generated, including: The feature format of the global temperature state representation is adapted to the feature input layer of the temperature state analysis model so that the global temperature state representation meets the model input requirements. The spatial correlation modeling layer of the temperature state analysis model is used to strengthen the spatial correlation of the adapted global temperature state representation. A spatial correlation constraint model is constructed by combining the spatial correlation rules between nodes, and the spatial correlation features in the global temperature state representation are strengthened based on the constraint model. The temporal correlation modeling layer of the temperature state analysis model is used to mine the temporal correlation of the global temperature state representation after strengthening the spatial correlation features. The temporal correlation features are extracted by combining the temporal change pattern of temperature sensing data. The temporal correlation features reflect the change correlation of temperature data under different acquisition time markers. Spatial and temporal correlation features are input into the feature fusion layer of the temperature state analysis model for cross-dimensional feature fusion, generating comprehensive correlation features that combine spatial and temporal correlation. The judgment output layer of the temperature state analysis model is invoked to classify the correlation state of the comprehensive correlation features, and generate temperature state correlation judgment results that include correlation state category and correlation confidence.

6. The method according to claim 5, characterized in that, The spatial correlation modeling layer of the temperature state analysis model enhances the spatial correlation of the adapted global temperature state representation, constructs a spatial correlation constraint model based on the spatial correlation rules between nodes, and strengthens the spatial correlation features in the global temperature state representation based on the constraint model, including: Initial spatial correlation features are extracted from the adapted global temperature state representation. These initial spatial correlation features directly reflect the spatial correlation relationship of temperature data at each edge node. Based on the spatial association patterns between nodes, spatial association influencing factors are determined, which are used to quantify the strength of spatial association between different edge nodes. A spatial correlation constraint model is constructed based on spatial correlation influencing factors. The spatial correlation constraint model contains multiple spatial correlation constraints, and each condition corresponds to a set of spatial correlation rules for edge nodes. The initial spatial association features are input into the spatial association constraint model. The initial spatial association features are filtered and strengthened through the constraint conditions, retaining the effective spatial association features that conform to the spatial association rules and enhancing the feature response strength of the effective spatial association features. The enhanced effective spatial correlation features are normalized to eliminate the dimensional differences between different spatial correlation features and generate standardized spatial correlation features. The standardized spatial correlation features are consistent with the feature dimensions of the global temperature state representation.

7. The method according to claim 5, characterized in that, The process involves using the temporal correlation modeling layer of the temperature state analysis model to perform temporal correlation mining on the global temperature state representation after enhancing spatial correlation features. This is combined with extracting temporal correlation features based on the temporal variation patterns of temperature sensing data, including: The global temperature state representation after enhancing spatial correlation features is segmented temporally and divided into multiple continuous temporal segment features according to the order of acquisition time. Calculate the feature change between adjacent time segments. The feature change reflects the numerical change of the corresponding feature dimension in the features of two consecutive time segments. The temporal correlation threshold is determined based on the temporal variation pattern of temperature sensing data. The temporal correlation threshold is used to determine whether there is an effective temporal correlation between features of adjacent time segments. By comparing the feature change with the temporal correlation threshold, valid temporal segment feature pairs whose feature change meets the threshold requirement are selected. The association patterns between valid time-series feature pairs are extracted, and the association patterns of all valid time-series feature pairs are integrated to generate time-series association features that reflect the time-series changes in global temperature state.

8. The method according to claim 1, characterized in that, The process involves generating temperature data processing instructions adapted to each edge node based on the temperature state correlation judgment result, distributing these instructions to the corresponding edge nodes, and then having each edge node execute local temperature sensing data processing operations according to the instructions, including: Analyze the association status category and association confidence in the temperature status association judgment results to determine the temperature data processing requirements for each edge node. The processing requirements are directly related to the association status category. Based on the local processing capabilities and temperature data processing requirements of each edge node, processing instruction generation rules are constructed. The processing instruction generation rules include instruction operation types and operation parameters corresponding to different processing requirements. According to the processing instruction generation rules, a unique temperature data processing instruction is generated for each edge node. The instruction includes node identification information, processing operation type and corresponding operation parameters, so that the instruction is accurately matched with the edge node. The temperature data processing instructions are distributed to the corresponding edge nodes according to the node identification information through the instruction distribution module of the edge collaboration node. The instructions are encrypted during the distribution process to ensure secure transmission. After receiving the temperature data processing instructions, each edge node parses them, extracts the processing operation type and operation parameters, executes the local temperature sensing data processing operation according to the processing operation type and operation parameters, and sends back processing completion confirmation information to the edge collaboration node after completion.

9. The method according to claim 2, characterized in that, The process involves performing feature-linked modeling on time-synchronized data sets, constructing spatial correlation constraints by combining node location association attributes, and mining the correlation features between the temperature data of the current edge node and its neighboring nodes based on these constraints. This includes: Feature extraction is performed on the temperature data of the local edge node and adjacent nodes in the time-synchronized data group to obtain basic temperature features, which reflect the core attributes of the temperature data. Based on the node location association attributes, determine the spatial distance and orientation relationships between this edge node and each adjacent node, and quantify the spatial distance and orientation relationships between this edge node and each adjacent node into spatial association parameters; Spatial correlation constraints are constructed by combining spatial correlation parameters. These constraints clarify the spatial correlation logic that should be satisfied between the temperature data of this edge node and its neighboring nodes. Substitute the basic temperature features of this edge node and the basic temperature features of adjacent nodes into the spatial association constraints to perform association feature mining and identify feature association patterns that conform to spatial association logic. The identified feature association patterns are quantized and converted into a representative association feature vector. The dimension of the association feature vector is consistent with the dimension of the basic temperature feature, forming the association feature between the temperature data of this edge node and the adjacent nodes.

10. A temperature sensing data processing system incorporating edge computing, characterized in that, The temperature sensing data processing system combining edge computing 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 temperature sensing data processing method combining edge computing as described in any one of claims 1-9.

Images