Method and device for maintaining emergency equipment, electronic device and storage medium

Abstract

The application discloses a kind of maintenance optimization method, device, electronic equipment and storage medium of emergency equipment, for solving how to meet the device state retention while, the allocation of emergency equipment maintenance resource is optimized, the problem of improving maintenance efficiency.Acquire the emergency equipment state data of to be maintained area and the discriminant text information of user input;According to the discriminant text information from the abnormal feature matrix of emergency equipment state data;Based on abnormal feature matrix and the maintenance example data of preestablished constructs multiple associated hypergraphs;According to multiple associated hypergraphs, generate regional maintenance information.Thereby, through the abnormal feature matrix of extraction associated hypergraph, the complex causal relationship between multiple emergency equipment and maintenance factor can be accurately described;Further, according to the regional maintenance information of associated hypergraph, thereby optimize maintenance strategy, realize reasonable scheduling maintenance resource, reduce maintenance cost, improve the maintenance response efficiency and efficiency of emergency equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of device management, and in particular to a maintenance optimization method and device for emergency equipment, an electronic device and a storage medium. BACKGROUND

[0002] Emergency equipment refers to various special equipment and technical devices used to protect people's lives and safety, reduce property losses, maintain social order and assist rescue operations in emergency situations such as sudden natural disasters, accident disasters, public health incidents or social security incidents. These devices usually have the characteristics of fast response, strong adaptability, high reliability and can operate normally in extreme environments and complex conditions. Modern emergency equipment covers communication equipment, lighting equipment, medical equipment, search and rescue equipment, protective equipment and other fields. Its technological development mainly depends on the progress of modern science and technology such as Internet of Things, artificial intelligence, big data and satellite communication.

[0003] With the widespread use of emergency equipment in various emergency response, the demand for state perception, maintenance and resource scheduling of emergency equipment is increasing. However, in the management and maintenance process of a large number of emergency equipment, scheduling difficulties often arise. The maintenance of emergency equipment not only involves the wear and tear of the equipment itself, but also includes the scheduling of maintenance personnel, the allocation of maintenance supplies and the commonality of maintenance equipment.

[0004] Due to limited maintenance resources, how to optimize the allocation of maintenance resources while meeting the equipment state retention and improve the maintenance efficiency is a difficult point in current emergency equipment management. SUMMARY

[0005] The present application provides a maintenance optimization method and device for emergency equipment, an electronic device and a storage medium, which are used to solve or partially solve the technical problem of how to optimize the allocation of emergency equipment maintenance resources while meeting the equipment state retention and improve the maintenance efficiency.

[0006] The present application provides a maintenance optimization method for emergency equipment, which comprises:

[0007] Obtaining emergency equipment state data of a to-be-maintained area and user input discriminant text information;

[0008] Extracting an abnormal feature matrix from the emergency equipment state data according to the discriminant text information;

[0009] Constructing a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data;

[0010] Generating area maintenance information according to the plurality of associated hypergraphs, the area maintenance information being used for maintenance and resource allocation of emergency equipment in the to-be-maintained area.

[0011] Optionally, the extracting an abnormal feature matrix from the emergency equipment state data according to the discriminative text information comprises:

[0012] performing semantic analysis on the discriminative text information to generate a discriminative feature vector;

[0013] performing context feature extraction on the emergency equipment state data based on the discriminative feature vector to obtain an initial abnormal feature set;

[0014] extracting nodes with feature activation values exceeding a preset threshold from the initial abnormal feature set to construct a first abnormal feature tensor;

[0015] performing clustering analysis on the first abnormal feature tensor to obtain a same-class data set;

[0016] performing time-frequency domain joint feature extraction on the same-class data set to generate a second abnormal feature tensor;

[0017] performing high-order feature fusion on the first abnormal feature tensor and the second abnormal feature tensor to generate an abnormal feature matrix.

[0018] Optionally, the performing clustering analysis on the first abnormal feature tensor to obtain a same-class data set comprises:

[0019] calculating a cosine similarity of the first abnormal feature tensor, and constructing a feature similarity matrix according to the cosine similarity;

[0020] calculating a dynamic neighborhood radius parameter according to the feature similarity matrix, and screening the feature similarity matrix according to the dynamic neighborhood radius parameter to obtain an adjacency matrix;

[0021] extracting a core point set of the adjacency matrix, and performing spectral clustering segmentation on the core point set to generate a same-class data set.

[0022] Optionally, the performing time-frequency domain joint feature extraction on the same-class data set to generate a second abnormal feature tensor comprises:

[0023] allocating dynamic weights to the same-class data set through a multi-head attention mechanism to generate an attention feature matrix;

[0024] adopting discrete wavelet transform to decompose the attention feature matrix into a low-frequency component and a high-frequency component;

[0025] mapping the low-frequency component into a first sub-tensor and mapping the high-frequency component into a second sub-tensor;

[0026] performing orthogonal projection on the first sub-tensor and the second sub-tensor to generate a second abnormal feature tensor.

[0027] Optionally, the constructing a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data comprises:

[0028] performing vectorization processing on the preset maintenance example data to obtain a first example feature matrix;

[0029] matching the abnormal feature matrix and the first example feature matrix to obtain a second example feature matrix;

[0030] generating maintenance instruction information according to the abnormal feature matrix and the second example feature matrix;

[0031] obtaining map information associated with the abnormal feature matrix;

[0032] performing hypergraph construction according to the map information, the maintenance instruction information and a preset hypergraph construction condition to obtain a plurality of associated hypergraphs.

[0033] Optionally, the performing hypergraph construction according to the map information, the maintenance instruction information and a preset hypergraph construction condition to obtain a plurality of associated hypergraphs comprises:

[0034] performing clustering according to the maintenance instruction information to obtain a maintenance type distribution index;

[0035] determining a target hypergraph construction condition from the preset hypergraph construction condition according to the maintenance type distribution index;

[0036] determining a hypergraph node from the map information according to the target hypergraph construction condition;

[0037] generating edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information and the map information to obtain a plurality of associated hypergraphs.

[0038] Optionally, the map information comprises emergency equipment positioning coordinate information; the maintenance instruction information comprises maintenance equipment, supplies and personnel configuration; and the generating edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information and the map information to obtain a plurality of associated hypergraphs comprises:

[0039] performing Gaussian kernel space encoding processing on the emergency equipment positioning coordinate information to generate a hypergraph node set;

[0040] calculating a resource sharing rate matrix of the hypergraph node set according to the maintenance equipment, supplies and personnel configuration;

[0041] performing geographical proximity analysis on the hypergraph node set based on the traffic distance threshold in the target hypergraph construction condition, for each node to be analyzed in the hypergraph node set, marking nodes having a Euclidean distance less than or equal to the traffic distance threshold from the node to be analyzed as adjacent nodes to generate a traffic adjacency matrix;

[0042] fusing the resource sharing rate matrix and the traffic adjacency matrix to generate a hybrid similarity matrix through weighted summation, and screening a hyperedge candidate set from the hybrid similarity matrix according to a dynamic threshold;

[0043] aggregating all adjacent nodes and indirectly associated nodes of each node in the hyperedge candidate set to generate a hyperedge set;

[0044] performing multi-level hypergraph convolution optimization on the hyperedge set to generate a plurality of associated hypergraphs.

[0045] The application further provides a maintenance optimization device for emergency equipment, comprising:

[0046] a data acquisition unit configured to acquire emergency equipment state data of a region to be maintained and user-input discrimination text information;

[0047] an abnormal feature matrix extraction unit configured to extract an abnormal feature matrix from the emergency equipment state data according to the discrimination text information;

[0048] an associated hypergraph construction unit configured to construct a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data;

[0049] a region maintenance information generation unit configured to generate region maintenance information according to the plurality of associated hypergraphs, the region maintenance information being used for maintenance and resource allocation of emergency equipment in the region to be maintained.

[0050] The application further provides an electronic device, which comprises a processor and a memory:

[0051] The memory is configured to store program code and transmit the program code to the processor;

[0052] The processor is configured to execute the maintenance optimization method for emergency equipment according to instructions in the program code.

[0053] The application further provides a computer-readable storage medium configured to store program code, the program code being used for executing the maintenance optimization method for emergency equipment.

[0054] As can be seen from the above technical solutions, the application has the following advantages:

[0055] A maintenance optimization method of emergency equipment is provided. First, emergency equipment state data of a region to be maintained and user input discriminant text information are acquired; then, an abnormal feature matrix is extracted from the emergency equipment state data according to the discriminant text information; then, a plurality of associated hypergraphs are constructed based on the abnormal feature matrix and preset maintenance example data; and region maintenance information is generated according to the plurality of associated hypergraphs. Thus, by extracting the abnormal feature matrix to construct the associated hypergraph, the complex causal relationship between the plurality of emergency equipment and the maintenance factors can be accurately described; and further, the region maintenance information for maintaining and resource allocation of the emergency equipment in the region to be maintained is generated according to the associated hypergraph, so as to optimize the maintenance strategy, realize reasonable scheduling of maintenance resources, reduce the maintenance cost, and improve the maintenance response efficiency and efficiency of the emergency equipment. BRIEF DESCRIPTION OF DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0057] Figure 1 A step flow chart of a maintenance optimization method of emergency equipment;

[0058] Figure 2 A structure schematic diagram of a maintenance optimization system of emergency equipment;

[0059] Figure 3 A structure block diagram of a maintenance optimization device of emergency equipment. DETAILED DESCRIPTION

[0060] The embodiments of the present application provide a maintenance optimization method, device, electronic equipment and storage medium of emergency equipment, which are used to solve or partially solve the technical problem of how to optimize the allocation of emergency equipment maintenance resources while meeting the device state maintenance, and improve the maintenance efficiency.

[0061] In order to make the inventive purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the following described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0062] As an example, with the wide application of emergency equipment in various emergency response, the demand for state awareness, maintenance and resource scheduling of the emergency equipment is increasing. However, in the management and maintenance process of a large number of emergency equipment, the scheduling difficulty is often encountered. The maintenance of the emergency equipment not only involves the wear and tear of the equipment itself, but also includes the scheduling of maintenance personnel, the allocation of maintenance materials and the commonality of the maintenance equipment.

[0063] Due to the limited maintenance resources, how to optimize the allocation of maintenance resources while meeting the equipment state retention and improve the maintenance efficiency is a difficulty in the current emergency equipment management.

[0064] Therefore, one of the core invention points of the embodiment of the present application is to provide a maintenance optimization method of emergency equipment. First, the abnormal feature matrix is automatically extracted based on the pre-set natural language model, and the unstructured data is converted into structured information; then the complex causal relationship between multiple emergency equipment and maintenance factors is accurately described by constructing an associated hypergraph; further, the regional maintenance information is generated according to the associated hypergraph, so as to optimize the maintenance strategy, realize the reasonable scheduling of maintenance resources, reduce the maintenance cost, and improve the maintenance response efficiency and efficiency of the emergency equipment.

[0065] In the embodiment of the present application, the emergency equipment mainly includes various sensors, device functions, programmable logic controllers (PLC), intelligent communication modules and device control screens.

[0066] Among them, various sensors needed for monitoring the emergency equipment will be carried on the emergency equipment. The sensors will collect environmental data according to the function, and store the collected data to the PLC through serial communication. The PLC is also connected with various relays, intelligent air switches through serial ports. The relay can control the power-on of the equipment, lamps and supports. The PLC communicates with the device control screen and the intelligent communication module in the form of TCP / IP (Transmission Control Protocol / Internet Protocol) through the network port. The emergency equipment reads the data on the PLC based on the control screen to perform front-end design and data layout, so as to realize the display of equipment data. The control screen of the emergency equipment also writes control commands for the PLC. The control screen sends control instructions to make the relays corresponding to the PLC address and the intelligent air switch respond, so as to control the equipment to perform various types of operations. The intelligent communication module is connected with the PLC through the network port in the form of TCP / IP. The intelligent communication module receives queries, write, control instructions from the server through the 4G network, and sends them to the PLC in the form of S7Comm (S7 Communication Protocol) communication protocol to perform related operations.

[0067] Referring to Figure 1 , a step flow chart of a method for maintaining an emergency device is shown, which can specifically include the following steps:

[0068] Step 101, obtaining state data of emergency devices in a region to be maintained and user-input discriminant text information;

[0069] In a specific implementation, first, state data of each emergency device in a region to be maintained, which is previously delimited, needs to be obtained.

[0070] Exemplarily, the emergency device can specifically include a device power supply, a generator set, a circuit breaker, a LED (Light Emitting Diode) lighting lamp group, an advertising lamp, a side lamp, a searchlight, a warning lamp, etc. The state data of the address points is collected through the PLC on the emergency device. The collected state data can specifically include the starting condition, online condition, battery capacity, inclination angle, obstacle distance, device input voltage and current, device power, support leg stress condition, device lamp current, device height, support leg elongation length, device lamp pole air pressure, device inclination temperature, device specific position, oil tank oil quantity, device input form, device running time, etc.

[0071] Meanwhile, according to the model of the emergency device, the factory information, function information, device attribution information, etc. of the emergency device can also be collected in the database as a supplement of the state data, so as to cope with the situation that the device cannot find the responsible person and understand the specific configuration information of the device.

[0072] It should be noted that the collection of the above information needs to pass through the permission system of the platform to ensure the legality and compliance of the information collection. Specifically, the system needs to grant the platform function permission to specific roles and users, and will record the operation records of the users to provide protection for the network security of the device.

[0073] The user-input discriminant text information refers to the discriminant text information input by the user in advance. For example, "abnormal pressure fluctuation of the hydraulic system". For another example, "battery module temperature overrun", etc.

[0074] Step 102, extracting an abnormal feature matrix from the state data of the emergency device according to the discriminant text information;

[0075] This step mainly realizes extracting an abnormal feature matrix from the state data of the emergency device according to the pre-input discriminant text information based on a pre-set natural language model.

[0076] In some embodiments, according to the process of extracting an abnormal feature matrix from the emergency equipment state data according to the discriminative text information, the following sub-steps S01 to S06 can be performed:

[0077] Step S01: performing semantic analysis on the discriminative text information to generate a discriminative feature vector;

[0078] Step S02: performing context feature extraction on the emergency equipment state data based on the discriminative feature vector to obtain an initial abnormal feature set;

[0079] Step S03: extracting nodes with feature activation values exceeding a preset threshold in the initial abnormal feature set to construct a first abnormal feature tensor;

[0080] Step S04: performing clustering analysis on the first abnormal feature tensor to obtain a same-class data set;

[0081] Further, the implementation process of step S04 can mainly be as follows: calculating the cosine similarity of the first abnormal feature tensor, and constructing a feature similarity matrix according to the cosine similarity; calculating a dynamic neighborhood radius parameter according to the feature similarity matrix, and filtering the feature similarity matrix according to the dynamic neighborhood radius parameter to obtain an adjacency matrix; extracting a core point set of the adjacency matrix, and performing spectral clustering segmentation on the core point set to generate the same-class data set.

[0082] Step S05: performing time-frequency domain joint feature extraction on the same-class data set to generate a second abnormal feature tensor;

[0083] Further, the implementation process of step S05 can mainly be as follows: assigning a dynamic weight to the same-class data set through a multi-head attention mechanism to generate an attention feature matrix; decomposing the attention feature matrix into a low-frequency component and a high-frequency component using discrete wavelet transform; mapping the low-frequency component into a first sub-tensor and the high-frequency component into a second sub-tensor; performing orthogonal projection on the first sub-tensor and the second sub-tensor to generate the second abnormal feature tensor.

[0084] Step S06: performing high-order feature fusion on the first abnormal feature tensor and the second abnormal feature tensor to generate an abnormal feature matrix.

[0085] Exemplarily, a data analysis module is arranged in the server. In the data analysis module, a preset natural language model is installed. The preset natural language model mainly includes an embedding layer, a bidirectional long short-term memory (BiLSTM), a gated linear unit (GLU) and a tensor fusion layer. The preset natural language model can perform correlation analysis on the pre-input discriminant text information and the real-time collected emergency equipment state data through a multi-modal data fusion and deep feature extraction technology, and output a structured abnormal feature matrix.

[0086] The server can be an independent server. It can also be a server cluster. It can also be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing services such as big data and artificial intelligence platforms.

[0087] First, based on the preset natural language model, abnormal data is extracted from the emergency equipment state data. At the same time, the abnormal data also needs to be output as structured data.

[0088] The user pre-input discriminant text information (such as "hydraulic system pressure fluctuation anomaly" and "battery module temperature overrun") is subjected to semantic analysis. The key words in the text are mapped to high-dimensional vectors through the embedding layer, and the core semantic features in the text are extracted using the self-attention mechanism (Self-Attention) to generate a discriminant feature vector. For example, "pressure fluctuation anomaly" will be analyzed as a discriminant feature vector containing device state, fault type, and severity.

[0089] The generated discriminant feature vector is input into the bidirectional long short-term memory network to model the context of the time series of the structured abnormal data corresponding to the state data. The bidirectional long short-term memory network captures the time sequence dependence of the sensor data through forward and reverse double channels to identify potential abnormal patterns (such as periodic fluctuations, sudden increases and decreases). For example, for the pressure data of the hydraulic system, the model can extract context features such as mean, variance, and gradient change in the continuous time window to form an initial abnormal feature set.

[0090] The initial abnormal feature set is nonlinearly filtered based on a gated linear unit. The gated linear unit dynamically activates feature channels through a gating mechanism, and only retains feature nodes with activation values exceeding a preset threshold (such as the abnormal threshold can be set to 3σ range of historical data distribution). The filtered feature nodes are reorganized according to the space-time dimension to construct a first abnormal feature tensor. The dimensions include device number, sensor type, timestamp, and multiple indexes.

[0091] For feature processing of the tensor fusion layer, a dynamic density clustering algorithm is used to group the first abnormal feature tensor. According to similarity measurement (such as cosine similarity) and adaptive neighborhood radius adjustment, the associated emergency equipment data is classified into the same set to generate a second abnormal feature tensor. The root cause is determined from the data features of the first abnormal feature tensor and the second abnormal feature tensor, high-order feature fusion is performed, and an abnormal feature matrix is generated.

[0092] Step 103, based on the abnormal feature matrix and the preset maintenance example data, a plurality of associated hypergraphs are constructed;

[0093] There is an associated influence between multiple interacting devices. For example, it is assumed that the normal fluctuation is determined in the A device, which causes the B device to appear abnormal. In this case, if A device and B device are analyzed separately, it is difficult to find such an influence, thereby causing abnormal analysis error.

[0094] Therefore, a plurality of associated hypergraphs can be constructed based on the previously constructed abnormal feature matrix and the preset maintenance example data to avoid such abnormal analysis error.

[0095] In some embodiments, the implementation process of constructing a plurality of associated hypergraphs based on the abnormal feature matrix and the preset maintenance example data can specifically include the following sub-steps S11 to S15:

[0096] Step S11: The preset maintenance example data is vectorized to obtain a first example feature matrix;

[0097] Step S12: The abnormal feature matrix and the first example feature matrix are matched to obtain a second example feature matrix;

[0098] Step S13: Maintenance instruction information is generated according to the abnormal feature matrix and the second example feature matrix;

[0099] Step S14: Map information associated with the abnormal feature matrix is obtained;

[0100] Step S15: Hypergraph construction is performed according to the map information, the maintenance instruction information, and the preset hypergraph construction condition to obtain a plurality of associated hypergraphs.

[0101] Further, the supergraph is constructed according to the map information, the maintenance instruction information, and a preset supergraph construction condition, and a plurality of associated supergraphs is obtained. Specifically, the supergraph can be constructed as follows:

[0102] The clustering is performed according to the maintenance instruction information, and a maintenance type distribution index is obtained.

[0103] The target supergraph construction condition is determined from the preset supergraph construction condition according to the maintenance type distribution index.

[0104] The supergraph node is determined from the map information according to the target supergraph construction condition.

[0105] The edge is generated between the supergraph nodes according to the target supergraph construction condition, the maintenance instruction information, and the map information, and a plurality of associated supergraphs is obtained.

[0106] Further, the map information can include emergency equipment positioning coordinate information, and the maintenance instruction information can include maintenance equipment, material, and personnel configuration. The edge is generated between the supergraph nodes according to the target supergraph construction condition, the maintenance instruction information, and the map information, and a plurality of associated supergraphs is obtained. Specifically, the supergraph can be constructed as follows:

[0107] The emergency equipment positioning coordinate information is subjected to Gaussian kernel space coding processing, and a supergraph node set is generated.

[0108] The resource sharing rate matrix of the supergraph node set is calculated according to the maintenance equipment, material, and personnel configuration.

[0109] The geographical proximity of the supergraph node set is analyzed based on the traffic distance threshold in the target supergraph construction condition. For each node to be analyzed in the supergraph node set, the node with a Euclidean distance less than or equal to the traffic distance threshold from the node to be analyzed is marked as an adjacent node, and a traffic adjacency matrix is generated.

[0110] The resource sharing rate matrix and the traffic adjacency matrix are fused, a hybrid similarity matrix is generated by weighted summation, and a superedge candidate set is screened from the hybrid similarity matrix according to a dynamic threshold.

[0111] All adjacent nodes and indirectly associated nodes of each node in the superedge candidate set are aggregated to generate a superedge set.

[0112] The superedge set is subjected to multi-level supergraph convolution optimization, and a plurality of associated supergraphs is generated.

[0113] Exemplarily, a result generation module can also be arranged in the server. In the result generation module, a multi-dimensional associated supergraph is constructed by fusing the abnormal feature matrix and the preset maintenance example data, and a regionalized maintenance scheme is generated based on the supergraph topology.

[0114] First, the large-scale emergency equipment to be maintained in the area to be maintained needs to be divided into multiple maintenance areas or maintenance routes, and maintenance personnel and maintenance resources are arranged. Before the maintenance personnel start, maintenance planning and maintenance prompts need to be prepared. Therefore, the embodiment of the present application solves the problem of difficult scheduling by automatically performing maintenance configuration, thereby improving the maintenance efficiency.

[0115] The preset maintenance example data (such as "replace hydraulic oil" and "adjust battery heat dissipation module") is semantically coded (vectorized), and a natural language model is used to convert it into a first example feature matrix. Each row of the matrix corresponds to a feature vector of a maintenance operation. It specifically contains dimension information such as required equipment type, material list, personnel skill, etc.

[0116] The abnormal feature matrix and the first example feature matrix are matched in similarity to filter out the target maintenance example most related to the current abnormal mode as the second example feature matrix. The matching process uses a bidirectional attention mechanism to calculate the correlation weight of the abnormal feature and the maintenance feature. For example, for "hydraulic pressure anomaly", the model will preferentially match "hydraulic system overhaul", "seal replacement" and other high-correlation maintenance items to generate the second example feature matrix.

[0117] The abnormal feature matrix and the second example feature matrix are input into the preset natural language model to generate structured maintenance instruction information through the decoder. The information includes maintenance priority, required resource list, operation steps, etc. For example: "Priority 1: Replace hydraulic pump (need hydraulic oil 20L, 2 technical personnel); Priority 2: Calibrate pressure sensor (need calibration instrument 1)".

[0118] Obtain the map information (such as device latitude and longitude coordinates, traffic road network) associated with the abnormal feature matrix. The positioning coordinates of each emergency device are encoded into a hypergraph node through a Gaussian kernel function. The node features include geographic location, abnormal level, resource demand, etc. For example, a device node can be represented as {longitude X, latitude Y, abnormal level: 3, need hydraulic oil 30L}.

[0119] Consider the resource collaboration between hypergraph nodes. Specifically, based on the analysis of maintenance equipment, materials and personnel configuration, if the commonality of the required maintenance materials of two nodes exceeds a threshold (such as 60%), the connection weight between them is enhanced, and a resource commonality matrix of the hypergraph node set is constructed. The resource commonality matrix reflects the resource commonality between every two nodes in the hypergraph node set.

[0120] According to the preset hypergraph construction conditions (traffic distance, resource commonality, personnel configuration), a hyperedge is dynamically generated. The geographical proximity analysis specifically includes: if the Euclidean distance between two device hypergraph nodes is less than a traffic threshold (such as 5 kilometers), it is marked as a potential association.

[0121] In addition, personnel scheduling constraints can also be considered. Specifically, according to the skills and positions of technical personnel, the hyperedge connection is optimized to minimize scheduling costs.

[0122] The multi-level convolution optimization is performed on the hyperedge set, the feature representation is updated through node-hyperedge bidirectional message passing, and multiple associated hypergraphs are generated.

[0123] In step 104, regional maintenance information is generated according to the multiple associated hypergraphs, and the regional maintenance information is used for maintenance and resource allocation of emergency equipment in the region to be maintained.

[0124] Based on the constructed multiple associated hypergraphs (which can be understood as a hypergraph topology composed of multiple associated hypergraphs), a spectral clustering algorithm is used to divide the hypergraph nodes into multiple maintenance groups. Each group satisfies the following conditions: the geographic distribution of the equipment in the group is concentrated to ensure traffic accessibility; the resource demand in the group is highly overlapped to improve the efficiency of material sharing; the personnel configuration is matched with the skills to avoid cross-group scheduling. The generated regional maintenance information includes the geographic range of each maintenance group, the resource allocation table, the personnel scheduling plan, etc. Through the regional maintenance information, each emergency equipment in the region to be maintained is maintained and resources are allocated. Thus, the command center can quickly maintain and allocate resources to the emergency equipment in the region and related personnel scheduling.

[0125] In the embodiments of the present application, a maintenance optimization method for emergency equipment is provided. Based on the pre-set natural language model, an abnormal feature matrix is automatically extracted, and unstructured data is converted into structured information; by constructing an associated hypergraph, the complex causal relationship between multiple emergency equipment and maintenance factors can be accurately described; further, according to the associated hypergraph, regional maintenance information is generated, thereby optimizing the maintenance strategy, realizing reasonable scheduling of maintenance resources, reducing the maintenance cost, and improving the maintenance response efficiency and efficiency of the emergency equipment.

[0126] In some embodiments, based on the pre-set natural language model, an abnormal feature matrix is extracted from the emergency equipment state data according to the pre-input discriminative text information. The pre-set natural language model includes an embedding layer, a bidirectional long short-term memory network, a gated linear unit, and a tensor fusion layer. In a specific implementation, based on the pre-set natural language model, an abnormal feature matrix is extracted from the emergency equipment state data according to the pre-input discriminative text information, including the following sub-steps S1021 to S1026:

[0127] S1021: The pre-input discriminative text information is processed in the embedding layer to generate a discriminative feature vector.

[0128] Exemplarily, in the embedding layer processing stage of the natural language model, the system first performs semantic analysis and vector mapping on the user pre-input discriminant text information (such as "hydraulic system pressure anomaly" "battery module temperature overrun").

[0129] The embedding layer uses a pre-trained word vector model (such as Word2Vec or BERT) to convert key terms in the text (such as "pressure", "temperature", "abnormal") into high-dimensional dense vectors, while preserving the semantic relevance between words. For example, "pressure" and "hydraulic" have a high cosine similarity in the vector space, while "temperature" and "heat dissipation" are mapped to similar semantic clusters.

[0130] After completing the word-level embedding, the text sequence is globally semantically modeled through a self-attention mechanism. Specifically, the self-attention layer calculates the correlation weight of each word with the context words, dynamically focusing on the core words describing the abnormal state (such as "fluctuation", "overrun"), and suppressing the interference of redundant modifier words (such as "slight", "possible"). By weightedly aggregating the attention vectors of all words, a discriminant feature vector is generated. This vector not only contains the semantic label of the abnormal type (such as "hydraulic system failure"), but also encodes the multi-dimensional information of the abnormal severity (such as "emergency", "general") and the potential impact range (such as "local", "global"), providing strong semantic guidance for subsequent time series feature extraction.

[0131] S1022: input the discriminant feature vector and the emergency equipment state data into a bidirectional long short-term memory network, and perform context feature extraction on the emergency equipment state data based on the discriminant feature vector to obtain an initial abnormal feature set.

[0132] Exemplarily, the bidirectional long short-term memory network provides semantic prior based on the discriminant feature vector, and performs context-aware feature extraction on the time series of the emergency equipment state data. The network uses a forward and reverse dual-channel structure. Specifically, the forward LSTM captures the evolution trend of the sensor data (such as the gradual rise of the pressure value) in time order (t=1→T), and the reverse LSTM identifies the lagging influence of the abnormal event (such as the sustained low value after the temperature drops).

[0133] The dual-channel output is spliced at each time step to form a hidden state vector containing complete context information. For example, for pressure sensor data of a hydraulic system, the bidirectional long short-term memory network can simultaneously capture the starting time of a pressure surge (forward feature) and its continued impact on subsequent time windows (backward feature). In addition, the network introduces a residual connection (RC) to alleviate the gradient vanishing problem, ensuring effective modeling of long sequence data. After multiple time step iterations, the network outputs an initial abnormal feature set. Its dimension covers device ID, sensor type, timestamp, statistical quantity (mean, variance, gradient), etc., providing a fine-grained data basis for subsequent feature selection and pattern mining.

[0134] S1023: Based on the gated linear unit, nodes with feature activation values exceeding a preset threshold in the initial abnormal feature set are extracted to construct a first abnormal feature tensor.

[0135] By way of example, the gated linear unit dynamically filters the initial abnormal feature set through a gating mechanism, separating abnormal signals from normal noise. Specifically, the gated linear unit divides the input features into two groups of channels. One group generates feature representations through linear transformation, and the other group generates gating weights (range 0~1) through a Sigmoid activation function. The two groups of channels are multiplied element by element, and only significant abnormal features with weights close to 1 (such as a pressure surge exceeding 3 times the standard deviation of historical mean) are retained, while irrelevant fluctuations with weights close to 0 (such as minor disturbances caused by environmental noise) are suppressed.

[0136] The gating threshold is adaptively set according to the statistical distribution of historical data. For example, for temperature sensor data, if the temperature value continuously exceeds the normal range (such as ±5℃) within a certain time window and the duration exceeds a preset threshold (such as 10 seconds), the activation value of the corresponding feature node is determined to be an effective anomaly.

[0137] The filtered feature nodes are reorganized according to the three dimensions of device, sensor, and time to construct a first abnormal feature tensor. The high-order structure of this tensor can represent cross-device collaborative anomalies (such as multiple hydraulic machines simultaneously experiencing pressure anomalies) or cross-sensor correlated faults (such as temperature rise accompanied by current fluctuations), providing a structured input for subsequent clustering analysis.

[0138] S1024: For feature processing of the tensor fusion layer, clustering analysis is performed on the first abnormal feature tensor to obtain a same or similar abnormal mode data set through dynamic radius parameter adjustment.

[0139] Exemplarily, based on the spatial distribution characteristics of the first abnormal feature tensor, a density clustering algorithm (such as OPTICS (Ordering Points To Identify the Clustering Structure) or DBSCAN (Density-Based Spatial Clustering of Applications with Noise)) is used to identify clusters of devices or sensors with similar abnormal patterns.

[0140] Specifically, the algorithm first calculates the cosine similarity between feature vectors to measure the semantic consistency of abnormal patterns (for example, “pressure fluctuation” and “flow anomaly” may belong to the same hydraulic system failure). Unlike traditional clustering methods with fixed neighborhood radius, a dynamic radius adjustment mechanism is introduced in this step. That is, according to the time series rate of change of feature similarity (such as the similarity difference between adjacent time windows), the search range is adaptively scaled. For example, when the similarity fluctuates sharply (such as a sudden multi-device failure), the neighborhood radius is expanded to capture a wider range of abnormal associations. When the similarity is stable (such as a periodic slight anomaly), the radius is reduced to improve clustering accuracy.

[0141] Through iterative optimization, the algorithm divides the feature tensor into multiple homogeneous data sets. Each set represents a specific abnormal pattern (such as “hydraulic system periodic leakage” and “battery pack heat dissipation efficiency decline”). At the same time, the algorithm automatically removes isolated noise points (such as single false alarm data), ensuring that the clustering results only contain statistically significant abnormal patterns, providing high-quality input for subsequent time-frequency domain joint analysis and maintenance strategy generation.

[0142] In some embodiments, the execution flow of the clustering analysis of the first abnormal feature tensor in step S1024, through dynamic radius parameter adjustment, to obtain homogeneous data sets with the same or similar abnormal patterns, can further include the following sub-steps S241 to S244:

[0143] S241: Calculate the cosine similarity of the first abnormal feature tensor, and construct a feature similarity matrix according to the cosine similarity.

[0144] Exemplarily, the cosine similarity between different first abnormal feature tensors is calculated by the following formula :

[0145]

[0146] wherein, and are different first abnormal feature tensors.

[0147] Then, according to the cosine similarity Construct a feature similarity matrix S. Where S∈R N×N , where S is an N-row N-column real number matrix, R represents the set of real numbers, and N is a preset feature dimension.

[0148] S242: Calculate the dynamic neighborhood radius parameter based on the feature similarity matrix.

[0149] Among them, the dynamic neighborhood radius parameter The specific calculation formula is as follows:

[0150]

[0151] In the formula, t represents time; λ is the decay coefficient; δ is the smoothing factor; ||·|| F This represents the Frobenius norm of the matrix. The rate of adjustment is controlled by the exponential decay coefficient λ to prevent cluster instability caused by abrupt changes in similarity.

[0152] S243: Filter the feature similarity matrix based on the dynamic neighborhood radius parameter to obtain the adjacency matrix.

[0153] For example, for satisfying Construct an adjacency matrix A∈{0,1} from the eigenvectors. n×n When processing data, the size of the neighborhood radius of the adjacency matrix is ​​dynamically adjusted based on the distribution and characteristics of the dynamic neighborhood radius parameter to adapt to the local characteristics of the data. This dynamic adjustment can improve the adaptability and accuracy of the algorithm.

[0154] S244: Extract the core point set of the adjacency matrix, and perform spectral clustering segmentation on the core point set to generate a similar data set.

[0155] Extract the core point set C={c through graph connectivity analysis. k |deg(c k )≥θ}. Where, deg(c k ) represents the node degree, and θ is the core point threshold.

[0156] Perform spectral clustering segmentation on the core point set, and divide the feature space into K sub-clusters through Laplacian matrix eigenvalue decomposition to generate clusters that satisfy... and Similar data sets {D1, D2, ..., D} K}, and remove the noise point set N=C c .

[0157] It should be noted that spectral clustering segmentation includes the following steps:

[0158] Construct the normalized Laplace matrix Where D is the degree matrix;

[0159] computing the eigenvectors U corresponding to the K smallest eigenvalues of L n×K ; wherein U is an n by K real matrix, and R denotes the set of real numbers;

[0160] performing K-means clustering on the row vectors of U to complete data partitioning.

[0161] S1025: performing time-frequency domain joint feature extraction on the same type of data set to generate a second abnormal feature tensor.

[0162] In some embodiments, the implementation process of performing time-frequency domain joint feature extraction on the same type of data set to generate a second abnormal feature tensor in step S1025 can further include the following sub-steps S251 to S253:

[0163] S251: Assigning dynamic weights to the same type of data set through a multi-head attention mechanism to generate an attention feature matrix.

[0164] For example, the multi-head attention mechanism is used to assign dynamic weights to the same type of data set (i.e., time series data of multi-source sensor channels) to generate an attention feature matrix with channel correlation. The data of multi-source sensor channels usually has strong time series dependence and inter-channel correlation features, which are difficult to fully exploit by traditional methods. By introducing the multi-head attention mechanism, the model can capture multi-angle feature information from different subspaces, avoiding information omission caused by single weight assignment.

[0165] Specifically, when processing multi-source sensor channel data, the data of each sensor channel is treated as an independent feature stream. The model first embeds the time series data of each channel to extract basic features. Then, the multi-head attention mechanism adaptively assigns dynamic weights to each channel according to the importance of the channel features. The dynamic weights can be adjusted according to the fluctuation characteristics, abnormal change trend or periodicity of the channel signals, thereby improving the adaptability of the model to complex signals. Then, the weighted aggregation operation is used to integrate the channel features to generate an attention feature matrix with channel correlation, so as to ensure that the model can fully capture the correlation and difference between sensor channels, laying a foundation for subsequent feature extraction and anomaly detection.

[0166] S252: Decomposing the attention feature matrix into low-frequency approximation components and high-frequency detail components using discrete wavelet transform, and mapping the low-frequency approximation components to a first sub-tensor and the high-frequency detail components to a second sub-tensor.

[0167] Exemplarily, the attention feature matrix is decomposed by using a discrete wavelet transform (DWT) to convert it into a low-frequency approximation component and a high-frequency detail component, so as to construct a first sub-tensor and a second sub-tensor. Since the attention feature matrix has integrated the time-series data and channel correlation features of the multi-source sensor, the internal signals often contain both smooth change components and sudden abnormal signals. As a time-frequency analysis method, the discrete wavelet transform can capture the local features of the signals at multiple scales and has strong separation ability for sudden signals and noise.

[0168] In a specific implementation, multi-layer decomposition is performed on the attention feature matrix to divide the signal into a low-frequency approximation component and a high-frequency detail component. The low-frequency approximation component retains the overall trend and long-term change features of the original signal and can represent the smooth information in the sensor channel. The high-frequency detail component mainly contains sudden abnormal signals, noise or other short-term dramatic change features and can highlight the short-time dynamic changes in the sensor signal. Further, the low-frequency approximation component is mapped to the first sub-tensor and the high-frequency detail component is mapped to the second sub-tensor, realizing multi-scale decomposition processing of the features and providing more discriminative data representation for subsequent feature orthogonalization and abnormality detection.

[0169] S253: Orthogonal projection is performed on the first sub-tensor and the second sub-tensor to generate a second abnormal feature tensor.

[0170] Exemplarily, the first sub-tensor and the second sub-tensor are subjected to orthogonal projection based on the Hilbert-Schmidt orthogonalization criterion to eliminate the collinearity interference between the features and generate the second abnormal feature tensor. Since the first sub-tensor and the second sub-tensor extracted in the previous steps may have a certain degree of feature redundancy, especially in multi-source sensor data, the features of each signal channel have a certain similarity and overlap, which is likely to cause collinearity problems in model training.

[0171] Collinearity interference can reduce the generalization ability of the model and affect the accuracy of abnormality detection. Therefore, the embodiment of the application introduces the Hilbert-Schmidt orthogonalization criterion to perform orthogonal projection on the first sub-tensor and the second sub-tensor. Specifically, first, the two types of feature sub-tensors are subjected to normalization processing to eliminate the influence of amplitude difference on the projection result. Then, based on the Hilbert-Schmidt criterion, the optimal projection matrix between the first sub-tensor and the second sub-tensor is calculated to make the projection results of the two as orthogonal as possible in the feature space, so as to reduce the feature redundancy to the maximum extent. Through the orthogonalized feature fusion operation, the second abnormal feature tensor is generated, so that the feature tensor has stronger discriminability and robustness and provides more accurate feature description for abnormal pattern recognition and device state perception.

[0172] In one embodiment, the first sub-tensor T∈Rn×m and a second sub-tensor F e R n×m , a Gram matrix G = T T F is constructed. Wherein, T and F are both real matrices of n rows and m columns, and R represents a real set. Then, by QR decomposition on G, an orthogonal basis matrix Q and an upper triangular matrix R can be obtained. After QR decomposition operation, the orthogonalized characteristic tensor Q satisfying Q T Q = I is generated, which is the orthogonal basis matrix Q, as the second abnormal characteristic tensor.

[0173] S1026: High-order feature fusion is performed on the first abnormal characteristic tensor and the second abnormal characteristic tensor to generate an abnormal feature matrix.

[0174] In some embodiments, the implementation process of step S1026 of high-order feature fusion of the first abnormal characteristic tensor and the second abnormal characteristic tensor to generate an abnormal feature matrix can further include the following sub-steps S261 to S263:

[0175] S261: The tensor product of the first abnormal characteristic tensor and the second abnormal characteristic tensor is calculated to generate a fusion tensor.

[0176] Exemplarily, the first abnormal characteristic tensor and the second abnormal characteristic tensor come from different feature extraction processes, respectively representing abnormal features capturing different aspects of the signal in the time domain and the frequency domain. In order to more comprehensively represent these abnormal features, high-order feature fusion is performed on the two. By calculating the tensor product of the two tensors, the features in the time domain and the frequency domain can be combined in a high-order space.

[0177] The tensor product is an operation that can fuse multi-dimensional data. It not only considers the product relationship between the elements of the two tensors, but also retains more multi-dimensional feature information. Thus, it helps to more deeply reveal the potential abnormal patterns in time series data. The fusion tensor obtained by calculation is multi-dimensional. It contains information from different feature spaces, ensuring the organic combination of time domain and frequency domain features, laying a foundation for subsequent feature dimension reduction and anomaly detection. Through high-order feature fusion, a fusion tensor with rich information can be obtained, thereby more accurately describing abnormal phenomena in multi-source sensor data.

[0178] S262: Singular value decomposition is performed on the fusion tensor layer by layer, the first r principal component feature vectors are retained, and dimension reduction mapping is performed through a projection matrix to generate a dimension-reduced feature matrix.

[0179] Exemplarily, although the fusion tensor obtained through tensor product contains rich feature information, in actual application, too high a data dimension may lead to too high a calculation complexity, and contains much redundant information. Therefore, dimension reduction processing is needed.

[0180] Singular Value Decomposition (SVD) is a common dimensionality reduction method. By performing singular value decomposition on the fused tensor, high-dimensional data can be mapped to a lower-dimensional space. Specifically, the fused tensor is decomposed into singular values and corresponding eigenvectors using singular value decomposition. By selecting the top r largest singular values and their corresponding eigenvectors, the main information in the data is preserved. Then, these principal component eigenvectors are mapped to a new low-dimensional space by a projection matrix to generate a reduced dimension feature matrix.

[0181] The reduced dimension matrix not only removes redundant information, but also preserves the core features of the data, making subsequent calculations more efficient. This step helps to extract the most representative abnormal features through effective dimension reduction, providing a refined and efficient data representation for anomaly detection.

[0182] S263: Apply spectral norm constraint to the reduced dimension feature matrix to generate an abnormal feature matrix.

[0183] Exemplarily, the reduced dimension feature matrix has significantly reduced redundant information and preserved the core features of the data. However, in practical applications, there may still be some noise or unstable components that affect the subsequent anomaly recognition performance. To solve this problem, the embodiment of the present application introduces spectral norm constraint to further optimize the feature matrix.

[0184] where the spectral norm constraint is a regularization method. By imposing a limit on the spectral norm of the matrix (i.e., the largest singular value of the matrix), it can prevent overfitting of the model during training and improve the generalization ability of the model.

[0185] In this step, by imposing spectral norm constraint on the reduced dimension feature matrix, the noise components in the matrix can be further suppressed, ensuring that the model pays more attention to the main abnormal features in the data and reduces false positives caused by data noise. In addition, the spectral norm constraint can also make the generated abnormal feature matrix have good numerical stability and robustness, making it more suitable for complex dynamic environments. The abnormal feature matrix optimized by spectral norm constraint has higher accuracy and can be more effectively used for anomaly detection and device state perception.

[0186] In some embodiments, the implementation process of constructing multiple associated hypergraphs based on the abnormal feature matrix and the preset maintenance example data can include the following sub-steps S1031 to S1035:

[0187] S1031: Based on the pre-set natural language model, perform vectorization processing on the preset maintenance example data to obtain a first example feature matrix.

[0188] The maintenance example data is vectorized by the preset natural language model, and each key element in the maintenance information is converted into a vector representation. For example, assume that the preset maintenance example data includes "replace hydraulic oil filter", "check battery voltage", and "calibrate temperature control device", and these text information is converted into a high-dimensional dense vector using a pre-trained word vector model (such as BERT or Word2Vec).

[0189] In this process, similar words can be mapped to similar vector space positions according to their semantic relationships. For example, "hydraulic" and "oil filter" may have a high degree of similarity in the vector space, while "battery" and "voltage" form relatively independent vector representations.

[0190] By processing each maintenance example data in this way, the semantic features of the maintenance task are captured, and a first example feature matrix is generated. It includes the type of maintenance task, the components involved, the urgency of the operation, and other information. Thus providing a basis for subsequent feature matching and hypergraph construction.

[0191] S1032: Match the abnormal feature matrix and the first example feature matrix to obtain a second example feature matrix.

[0192] Exemplarily, the abnormal feature matrix extracted from the device state perception is matched with the first example feature matrix. Assume that the state data of the emergency device shows "abnormal hydraulic pressure" at a certain time, first convert this abnormal information into an abnormal feature matrix, and extract the key features related to it (such as "pressure" exceeding the limit, "hydraulic" failure, etc.). Match these features with the first example feature matrix. Assume that there is a task "replace hydraulic oil filter" in the maintenance example, which may have a high degree of matching with the abnormal features of "abnormal hydraulic pressure" (because both are related to hydraulic pressure).

[0193] By calculating the similarity between the abnormal feature matrix and the first example feature matrix, the most relevant maintenance example task is selected and extracted to form a second example feature matrix. The second example feature matrix contains the maintenance task that best matches the current device abnormal features, and provides guidance for subsequent maintenance decisions.

[0194] S1033: Input the abnormal feature matrix and the second example feature matrix into the preset natural language model to generate maintenance instruction information.

[0195] Exemplarily, the abnormal features are combined with the maintenance example data using natural language processing techniques to generate specific maintenance recommendations for the current equipment condition. For example, suppose the abnormal feature matrix shows "hydraulic pressure out of limits", and the second example feature matrix contains tasks related to "hydraulic maintenance", such as "replace hydraulic oil" or "check pressure regulating valve". These feature information is input into a natural language model (such as BERT). The model generates specific maintenance instruction information according to these inputs. For example, "recommend replacing hydraulic oil and checking the regulating valve".

[0196] In addition, the model will also generate priority labels such as "urgent treatment" or "ordinary treatment" according to the urgency of the maintenance task and the operation of the equipment. Thus, a detailed maintenance instruction can be intelligently generated according to the actual situation of the equipment, ensuring that the equipment can be timely and properly maintained.

[0197] S1034: Obtain map information associated with the abnormal feature matrix.

[0198] By obtaining the map information associated with the abnormal feature matrix, the maintenance decision-making process of the equipment is further improved. For example, suppose the abnormal feature matrix shows that the hydraulic pressure is out of limits, and the relevant equipment location and geographic information need to be identified according to this abnormal state in order to optimize the maintenance scheduling. In actual application, equipment may be distributed in different geographic locations. Through the map information, technicians can understand the specific location of the equipment, the state of adjacent equipment, and factors such as traffic routes.

[0199] S1035: Construct a hypergraph according to the map information, the maintenance instruction information, and the preset hypergraph construction condition to obtain a plurality of associated hypergraphs.

[0200] In some embodiments, the implementation process of step S1035 of constructing a hypergraph according to the map information, the maintenance instruction information, and the preset hypergraph construction condition to obtain a plurality of associated hypergraphs can further include the following sub-steps S351-S354:

[0201] S351: Cluster according to the maintenance instruction information to obtain a maintenance type distribution index.

[0202] Through this step, the classification of maintenance types can be realized, so as to determine the maintenance resources corresponding to the maintenance task and determine the size of the associated hypergraph in advance.

[0203] S352: Determine the target hypergraph construction condition from the preset hypergraph construction condition according to the maintenance type distribution index.

[0204] Exemplarily, the size of the associated hypergraph is determined, i.e., the target hypergraph construction condition can be selected from the hypergraph construction conditions. For example, the total amount of traffic distance, the total amount of maintenance resources required, and the maintenance personnel configuration, etc. Thus, the construction of the associated hypergraph corresponds to the current maintenance task, and the maintenance efficiency is improved.

[0205] S353: determining the hypergraph nodes from the map information according to the target hypergraph construction condition.

[0206] S354: generating edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information, and the map information, to obtain a plurality of associated hypergraphs.

[0207] In one embodiment, assuming that the maintenance task of a certain device is "replace hydraulic oil", the device node and the task node related thereto are connected by a hyperedge. These nodes and their relationships are constructed into a hypergraph, and adjusted according to the preset hypergraph construction condition (such as common maintenance resources of the device, priority of the task, etc.) to ensure the effectiveness and executability of the hypergraph. The generated associated hypergraph can not only be used for scheduling of the device maintenance task, but also serve as a basis for global decision support, helping to dynamically adjust the maintenance strategy and resource allocation, so as to ensure the efficiency and reliability of the device in the running process.

[0208] In some embodiments, the implementation process of generating edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information, and the map information in step S354 to obtain a plurality of associated hypergraphs can further include the following sub-steps S3541 to S3546:

[0209] S3541: performing Gaussian kernel space encoding processing on the emergency device positioning coordinates in the map information to generate a hypergraph node set.

[0210] Exemplarily, the positioning coordinates of the emergency devices are extracted from the map information, and the coordinates are subjected to Gaussian kernel space encoding processing. The map information usually includes the accurate positions of various emergency devices, such as GPS (Global Positioning System) coordinates of the devices, etc. Gaussian kernel space encoding is a technology for mapping position coordinates to high-dimensional space through Gaussian function. It can capture the relative distance and spatial distribution characteristics between positions.

[0211] Specifically, the position coordinates of each emergency device will generate a new spatial vector under the action of the Gaussian kernel function. This vector not only maintains the original geographical space information, but also effectively maps the similarity between positions to a more compact high-dimensional space. In this way, a high-dimensional hypergraph node representation can be generated for each emergency device. These nodes represent the positions of the emergency devices in space, thereby providing accurate geographical information basis for subsequent resource allocation, scheduling, and maintenance tasks.

[0212] S3542: Calculate the resource sharing rate matrix of the hypergraph node set according to the maintenance equipment, materials, and personnel configuration in the maintenance instruction information.

[0213] Illustratively, according to the maintenance equipment, materials, and personnel configuration in the maintenance instruction information, the resource sharing rate between each two nodes is calculated to further construct the resource sharing rate matrix of the hypergraph node set. The resource sharing rate matrix reflects the resource sharing rate between each two nodes in the hypergraph node set.

[0214] The purpose of this step is to evaluate the resource sharing between different emergency equipment nodes to help understand the resources that each device may share during maintenance. Assuming that two nodes represent emergency equipment at different locations, the maintenance tasks of both require the same maintenance equipment or staff. At this time, the degree of resource overlap between the two is measured by calculating the resource sharing rate between the two nodes. For example, if the maintenance tasks of both nodes require "hydraulic oil filter element" replacement, and the filter element can only be operated by the same group of personnel, then the resource sharing rate of the two nodes is high. By analyzing the maintenance equipment, materials, and personnel configuration, a resource sharing rate matrix is generated. This matrix can represent the resource allocation relationship between nodes and provide a basis for the connection between nodes and the generation of hyperedges in the subsequent steps.

[0215] S3543: Perform geographic proximity analysis on the hypergraph node set based on the traffic distance threshold in the target hypergraph construction conditions. For each node to be analyzed in the hypergraph node set, mark the nodes with a Euclidean distance less than or equal to the traffic distance threshold as adjacent nodes, and generate a traffic adjacency matrix.

[0216] Illustratively, geographic proximity analysis is performed on the hypergraph node set based on the traffic distance threshold in the target hypergraph construction conditions. Specifically, the Euclidean distance between each pair of nodes is calculated and compared with the preset traffic distance threshold. If the Euclidean distance between two nodes is less than or equal to the preset threshold, the two nodes are considered "adjacent nodes". This means that the transportation connection between them can be made within a reasonable distance.

[0217] Through this geographic proximity analysis, it can be determined which emergency equipment can be connected to each other through short-distance transportation or walking, and which requires remote scheduling or additional transportation arrangements. The generated traffic adjacency matrix reflects the relative geographical relationship between nodes and provides an important spatial reference for the subsequent generation of hyperedges, making the construction of the hypergraph more consistent with the actual geographical distribution and resource allocation.

[0218] S3544: Fuse the resource sharing rate matrix with the traffic adjacency matrix, generate a hybrid similarity matrix through weighted summation, and screen out a super-edge candidate set from the hybrid similarity matrix according to a dynamic threshold.

[0219] Exemplarily, the resource sharing rate matrix is fused with the traffic adjacency matrix to generate a hybrid similarity matrix through weighted summation. The process of weighted summation combines the information of resource sharing rate and traffic adjacency to comprehensively evaluate the correlation between nodes. For example, two nodes may be considered to have strong correlation even if they are geographically far apart, as long as they share resources highly. The hybrid similarity matrix is a comprehensive evaluation of the two kinds of information and can reflect the overall similarity between nodes under the conditions of geography, resource sharing, etc.

[0220] Subsequently, the hybrid similarity matrix is screened according to a dynamic threshold, and node pairs with similarity higher than the threshold are marked as candidate super-edges. These candidate super-edges represent possible resource sharing and scheduling relationships. The screened super-edge candidate set provides a basis for subsequent super-edge construction and helps determine which devices or tasks need to be scheduled and allocated resources in priority.

[0221] S3545: Aggregate all adjacent nodes and indirectly associated nodes of each node in the super-edge candidate set to generate a super-edge set.

[0222] Exemplarily, by analyzing the direct adjacent nodes of each node in the super-edge candidate set and the nodes indirectly associated through other nodes, these nodes are combined into a super-edge set. Specifically, in a hypergraph, a super-edge can connect not only two nodes but also multiple nodes, indicating that there is a certain resource sharing or scheduling association between these nodes. For example, a node shares the same group of maintenance personnel with multiple devices, or is in the same geographic area with multiple devices. These nodes are combined into a super-edge so as to be considered together in subsequent scheduling and maintenance tasks.

[0223] This step, through the aggregation operation of multiple nodes, more finely reflects the complex relationships and mutual dependencies between devices in the entire hypergraph structure, ensuring that the hypergraph can accurately represent the dynamic connections between all devices and resources.

[0224] S3546: Perform multi-level hypergraph convolution optimization on the super-edge set to generate multiple associated hypergraphs.

[0225] Exemplarily, the hypergraph convolution is a way of processing the connection relationship between hyperedges in a hierarchical manner. It can effectively learn the high-order dependency relationship between nodes. Through multi-layer convolution operation, the representation of the hyperedge set is gradually optimized, so as to extract more detailed association information between nodes. Each layer of convolution operation is aimed at aggregating the features of nodes and updating the connection strength between nodes according to the structure of hyperedge, so that the hyperedge can more accurately reflect the complex dependency relationship between nodes.

[0226] In the convolution process, the topological structure of multiple associated hypergraphs is also optimized layer by layer to reduce redundant relationships and enhance the influence of key nodes and hyperedges, thereby generating optimized multiple associated hypergraphs. The hypergraph topological structure can not only effectively represent the relationship between emergency equipment, maintenance tasks and resources, but also provide accurate support for subsequent decision and scheduling, ensuring that resources are reasonably configured and the maintenance and scheduling process is optimized.

[0227] In some embodiments, the present application also provides an emergency equipment maintenance optimization system as shown in Figure 2 The emergency equipment maintenance optimization system can be used to perform the emergency equipment maintenance optimization method of any one of the embodiments of the present application.

[0228] As shown in Figure 2 The emergency equipment maintenance optimization system 200 can mainly include a data acquisition module 201, a data analysis module 202 and a result generation module 203.

[0229] It should be noted that the data acquisition module 201, the data analysis module 202 and the result generation module 203 can be communicatively connected to each other to transmit data or instructions to each other.

[0230] The data acquisition module 201 is configured to acquire state data of the emergency equipment.

[0231] The data analysis module 202 is configured to extract an abnormal feature matrix from the state data based on a preset natural language model.

[0232] Exemplarily, the database of the data analysis module 202 has a maintenance record function. When a user finds an abnormality, the abnormality of the equipment can be filled in through a pre-designed equipment management applet. The equipment maintenance personnel can prepare maintenance related tools according to the abnormality, and record the maintenance after the maintenance is completed. Through this function, the maintenance efficiency of the product can be greatly improved, and the maintenance of the equipment can be traced.

[0233] The database of the data analysis module 202 also has a device maintenance record and maintenance configuration function. When the device has abnormal parameters or a certain device reaches the maintenance time, the user can be notified in the form of a short message or device end notification, so that the user pays attention to the need to maintain the device. The user can also choose to maintain it himself or contact the manufacturer to maintain the device according to the actual situation and maintenance category. On the one hand, the maintenance record can be filled in to update the maintenance list of the device. On the other hand, through the self-defined maintenance configuration, maintenance notification, maintenance method and maintenance record, the device maintenance can be more rationalized, and the risk of device abnormality or damage can be reduced.

[0234] The result generation module 203 is configured to construct an abnormal hypergraph according to the abnormal feature matrix and the preset maintenance example data, and generate regional maintenance information.

[0235] In some embodiments, the emergency device maintenance optimization system 200 further comprises an emergency device. The emergency device supports one-key self-checking, can execute a self-checking program, actively detects abnormal states, and is mainly used for the user to understand the device condition when the device is not used for a long time. It can also be realized that non-professionals can check the device in a simple form.

[0236] The emergency device supports device authorization, and the user can temporarily borrow the devices of other customers by initiating a device authorization process. In the case of device shortage and emergency, the device can be quickly deployed.

[0237] In some embodiments, the emergency device maintenance optimization system 200 can further comprise a remote control terminal. Through remote control of the emergency device by the remote control terminal, control signals are sent to the PLC address points, so as to control the opening and closing of the power supply of the device, the opening and closing of the generator, the opening and closing of the circuit breaker, the opening and closing of the LED lamp group, the opening and closing of the advertising lamp, the opening and closing of the side lamp, the opening and closing of the searchlight, the opening and closing of the warning light, the lifting of the supporting leg, the lifting of the lighthouse lamp pole, the pitching of the lighthouse, and the like, to realize remote scheduling of the device. In addition, the remote control terminal can configure a batch intelligent control task, connect multiple devices, realize one-key control of multiple devices, reduce the consumption of manpower, and increase the use efficiency of the device.

[0238] Exemplarily, the remote control terminal is equipped with a device notification configuration function. The user can customize the judgment method of the abnormal field, and can independently select the abnormal notification method (short message notification / client notification). It can be specifically used for voltage abnormality notification, current abnormality notification, low power warning notification, low oil quantity warning notification, device obstacle blocking notification, device height warning, and the like.

[0239] In addition, the data exhibition stand of the remote control terminal can be filtered according to regions and time, and device quantity statistics, device usage statistics, device abnormal data statistics and device maintenance data statistics are displayed. Thus, the overall device condition and maintenance information can be more intuitively understood in the form of a statistical chart.

[0240] With reference to Figure 3 The structure block diagram of the maintenance optimization device for emergency equipment is shown, and the device can specifically include:

[0241] The data acquisition unit 301 is configured to acquire emergency equipment state data of a region to be maintained and user input discriminant text information.

[0242] The abnormal feature matrix extraction unit 302 is configured to extract an abnormal feature matrix from the emergency equipment state data according to the discriminant text information.

[0243] The associated hypergraph construction unit 303 is configured to construct a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data.

[0244] The region maintenance information generation unit 304 is configured to generate region maintenance information according to the plurality of associated hypergraphs, and the region maintenance information is used for maintenance and resource allocation of emergency equipment in the region to be maintained.

[0245] In an optional embodiment, the abnormal feature matrix extraction unit 302 includes:

[0246] The semantic analysis unit is configured to perform semantic analysis on the discriminant text information to generate a discriminant feature vector.

[0247] The context feature extraction unit is configured to perform context feature extraction on the emergency equipment state data based on the discriminant feature vector to obtain an initial abnormal feature set.

[0248] The first abnormal feature tensor construction unit is configured to extract nodes with feature activation values exceeding a preset threshold value in the initial abnormal feature set to construct a first abnormal feature tensor.

[0249] The clustering analysis unit is configured to perform clustering analysis on the first abnormal feature tensor to obtain a same-class data set.

[0250] The time-frequency domain joint feature extraction unit is configured to perform time-frequency domain joint feature extraction on the same-class data set to generate a second abnormal feature tensor.

[0251] The high-order feature fusion unit is configured to perform high-order feature fusion on the first abnormal feature tensor and the second abnormal feature tensor to generate an abnormal feature matrix.

[0252] In an optional embodiment, the clustering analysis unit comprises:

[0253] a feature similarity matrix construction unit configured to calculate cosine similarity of the first abnormal feature tensor and construct a feature similarity matrix according to the cosine similarity;

[0254] a neighborhood matrix construction unit configured to calculate a dynamic neighborhood radius parameter according to the feature similarity matrix and screen the feature similarity matrix according to the dynamic neighborhood radius parameter to obtain a neighborhood matrix;

[0255] a spectral clustering segmentation unit configured to extract a core point set of the neighborhood matrix and perform spectral clustering segmentation on the core point set to generate a same-class data set.

[0256] In an optional embodiment, the time-frequency domain joint feature extraction unit comprises:

[0257] an attention feature matrix generation unit configured to assign a dynamic weight to the same-class data set by a multi-head attention mechanism to generate an attention feature matrix;

[0258] a discrete wavelet transform unit configured to decompose the attention feature matrix into a low-frequency component and a high-frequency component by using a discrete wavelet transform;

[0259] a mapping unit configured to map the low-frequency component into a first sub-tensor and map the high-frequency component into a second sub-tensor;

[0260] an orthogonal projection unit configured to perform orthogonal projection on the first sub-tensor and the second sub-tensor to generate a second abnormal feature tensor.

[0261] In an optional embodiment, the correlation hypergraph construction unit 303 comprises:

[0262] a vectorization processing unit configured to perform vectorization processing on a preset maintenance example data to obtain a first example feature matrix;

[0263] a feature matrix matching unit configured to match the abnormal feature matrix and the first example feature matrix to obtain a second example feature matrix;

[0264] a maintenance instruction information generation unit configured to generate maintenance instruction information according to the abnormal feature matrix and the second example feature matrix;

[0265] a map information acquisition unit configured to acquire map information associated with the abnormal feature matrix;

[0266] a hypergraph construction unit configured to perform hypergraph construction according to the map information, the maintenance instruction information and a preset hypergraph construction condition to obtain a plurality of correlation hypergraphs.

[0267] In an optional embodiment, the supergraph construction unit comprises:

[0268] a maintenance instruction information clustering unit configured to cluster the maintenance instruction information to obtain a maintenance type distribution index;

[0269] a target supergraph construction condition determination unit configured to determine a target supergraph construction condition from preset supergraph construction conditions according to the maintenance type distribution index;

[0270] a supergraph node determination unit configured to determine supergraph nodes from the map information according to the target supergraph construction condition;

[0271] an associated supergraph construction subunit configured to generate edges between the supergraph nodes according to the target supergraph construction condition, the maintenance instruction information and the map information to obtain a plurality of associated supergraphs.

[0272] In an optional embodiment, the map information comprises emergency equipment positioning coordinate information; the maintenance instruction information comprises maintenance equipment, material and personnel configuration; and the associated supergraph construction subunit comprises:

[0273] a Gaussian kernel space encoding processing unit configured to perform Gaussian kernel space encoding processing on the emergency equipment positioning coordinate information to generate a supergraph node set;

[0274] a resource sharing rate matrix calculation unit configured to calculate a resource sharing rate matrix of the supergraph node set according to the maintenance equipment, material and personnel configuration;

[0275] a traffic adjacency matrix generation unit configured to perform geographical proximity analysis on the supergraph node set based on a traffic distance threshold in the target supergraph construction condition, and mark nodes having a Euclidean distance less than or equal to the traffic distance threshold from each to-be-analyzed node in the supergraph node set as adjacent nodes to generate a traffic adjacency matrix;

[0276] a superedge candidate set screening unit configured to fuse the resource sharing rate matrix and the traffic adjacency matrix, generate a hybrid similarity matrix by weighted summation, and screen a superedge candidate set from the hybrid similarity matrix according to a dynamic threshold;

[0277] a superedge set generation unit configured to aggregate all adjacent nodes and indirectly associated nodes of each node in the superedge candidate set to generate a superedge set;

[0278] a multi-level supergraph convolution optimization unit configured to perform multi-level supergraph convolution optimization on the superedge set to generate a plurality of associated supergraphs.

[0279] For the device embodiment, since it is basically similar to the method embodiment, it is described more simply, and the relevant part refers to the part of the foregoing method embodiment.

[0280] It should be noted that, in order to better distinguish the data of the same type but actually pointing to different meanings for those skilled in the art, part of the technical features in the embodiments of the application are distinguished as first and second, and the first and second are only used for data distinction and have no other special meanings. It can be understood that the application does not limit this.

[0281] The embodiment of the application further provides an electronic device, the device comprising a processor and a memory:

[0282] The memory is used for storing program code and transmitting the program code to the processor.

[0283] The processor is used for executing the maintenance optimization method of the emergency device according to the instructions in the program code.

[0284] The embodiment of the application further provides a computer readable storage medium, the computer readable storage medium is used for storing program code, and the program code is used for executing the maintenance optimization method of the emergency device according to any one of the embodiments of the application.

[0285] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here.

[0286] In the several embodiments of the application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There can be another division manner for actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0287] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment scheme.

[0288] In addition, each function unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software function unit.

[0289] When the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0290] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features. These modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A method for optimizing maintenance of emergency equipment, characterized in that The method comprises the following steps: acquiring emergency equipment state data of a region to be maintained and user input discriminant text information; extracting an abnormal feature matrix from the emergency equipment state data according to the discriminant text information; constructing a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data; generating region maintenance information according to the plurality of associated hypergraphs, the region maintenance information being used for maintenance and resource allocation of emergency equipment in the region to be maintained; wherein the step of constructing a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data comprises: vectorizing the preset maintenance example data to obtain a first example feature matrix; matching the abnormal feature matrix and the first example feature matrix to obtain a second example feature matrix; generating maintenance indication information according to the abnormal feature matrix and the second example feature matrix; acquiring map information associated with the abnormal feature matrix; performing hypergraph construction according to the map information, the maintenance indication information and preset hypergraph construction conditions to obtain a plurality of associated hypergraphs; the preset hypergraph construction conditions include traffic distance, resource sharing rate and personnel configuration.

2. The method of claim 1, wherein The step of extracting an abnormal feature matrix from the emergency equipment state data according to the discriminant text information comprises: performing semantic analysis on the discriminant text information to generate a discriminant feature vector; extracting a context feature from the emergency equipment state data based on the discriminant feature vector to obtain an initial abnormal feature set; extracting feature nodes with feature activation values exceeding a preset threshold from the initial abnormal feature set to construct a first abnormal feature tensor; the feature activation value is an activation value obtained by dynamically activating feature nodes in a feature channel through a gating mechanism; performing clustering analysis on the first abnormal feature tensor to obtain a same-type data set; performing time-frequency domain joint feature extraction on the same-type data set to generate a second abnormal feature tensor; performing high-order feature fusion on the first abnormal feature tensor and the second abnormal feature tensor to generate an abnormal feature matrix.

3. The method of claim 2, wherein, The step of performing clustering analysis on the first abnormal feature tensor to obtain a same-type data set comprises: calculating the cosine similarity of the first abnormal feature tensor, and constructing a feature similarity matrix according to the cosine similarity; calculating a dynamic neighborhood radius parameter according to the feature similarity matrix, and screening the feature similarity matrix according to the dynamic neighborhood radius parameter to obtain an adjacency matrix; extracting a core point set of the adjacency matrix, and performing spectral clustering segmentation on the core point set to generate a same-type data set.

4. The method of claim 2, wherein, The step of performing time-frequency domain joint feature extraction on the same-type data set to generate a second abnormal feature tensor comprises: allocating a dynamic weight to the same-type data set through a multi-head attention mechanism to generate an attention feature matrix; decomposing the attention feature matrix into a low-frequency component and a high-frequency component using discrete wavelet transform; mapping the low-frequency component into a first sub-tensor and the high-frequency component into a second sub-tensor; performing orthogonal projection on the first sub-tensor and the second sub-tensor to generate a second abnormal feature tensor.

5. The method of claim 1, wherein, The hypergraph construction according to the map information, the maintenance instruction information, and a preset hypergraph construction condition comprises: performing clustering according to the maintenance instruction information to obtain a maintenance type distribution index; determining a target hypergraph construction condition from the preset hypergraph construction condition according to the maintenance type distribution index; determining a hypergraph node from the map information according to the target hypergraph construction condition; generating edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information, and the map information to obtain a plurality of associated hypergraphs.

6. The method of claim 5, wherein, The map information comprises emergency equipment positioning coordinate information; the maintenance instruction information comprises maintenance equipment, supplies, and personnel allocation; and the generation of edges between the hypergraph nodes according to the target hypergraph construction condition, the maintenance instruction information, and the map information to obtain a plurality of associated hypergraphs comprises: performing Gaussian kernel space coding processing on the emergency equipment positioning coordinate information to generate a hypergraph node set; calculating a resource sharing rate matrix of the hypergraph node set according to the maintenance equipment, supplies, and personnel allocation; performing geographical proximity analysis on the hypergraph node set based on a traffic distance threshold value in the target hypergraph construction condition, and marking, for each to-be-analyzed node in the hypergraph node set, a node having a Euclidean distance from the to-be-analyzed node less than or equal to the traffic distance threshold value as an adjacent node to generate a traffic adjacency matrix; fusing the resource sharing rate matrix and the traffic adjacency matrix to generate a hybrid similarity matrix through weighted summation, and screening a hyperedge candidate set from the hybrid similarity matrix according to a dynamic threshold value; aggregating all adjacent nodes and indirectly associated nodes of each node in the hyperedge candidate set to generate a hyperedge set; performing multi-level hypergraph convolution optimization on the hyperedge set to generate a plurality of associated hypergraphs.

7. An apparatus for optimizing maintenance of emergency equipment, characterized in that The method comprises: a data acquisition unit configured to acquire emergency equipment state data of a region to be maintained and user-input discrimination text information; an abnormal feature matrix extraction unit configured to extract an abnormal feature matrix from the emergency equipment state data according to the discrimination text information; an associated hypergraph construction unit configured to construct a plurality of associated hypergraphs based on the abnormal feature matrix and preset maintenance example data; a region maintenance information generation unit configured to generate region maintenance information according to the plurality of associated hypergraphs, the region maintenance information being used for maintenance and resource allocation of emergency equipment in the region to be maintained; wherein the associated hypergraph construction unit comprises: a vectorization processing unit configured to perform vectorization processing on the preset maintenance example data to obtain a first example feature matrix; a feature matrix matching unit configured to match the abnormal feature matrix and the first example feature matrix to obtain a second example feature matrix; a maintenance instruction information generation unit configured to generate maintenance instruction information according to the abnormal feature matrix and the second example feature matrix; a map information acquisition unit configured to acquire map information associated with the abnormal feature matrix; and The supergraph construction unit is configured to construct supergraphs according to the map information, the maintenance instruction information, and preset supergraph construction conditions, and obtain a plurality of associated supergraphs; the preset supergraph construction conditions include a traffic distance, a resource sharing rate, and personnel allocation.

8. An electronic device, comprising: The device comprises a processor and a memory: The memory is configured to store program code and transmit the program code to the processor; The processor is configured to execute the maintenance optimization method of the emergency device according to instructions in the program code.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium is configured to store program code, and the program code is configured to execute the maintenance optimization method of the emergency device.

Images