Safety inspection method, device and equipment for smart power plant, storage medium and program product

Abstract

The invention discloses a safety inspection method and device for a smart power plant, equipment, a storage medium and a program product, and relates to the technical field of power plant inspection, and the safety inspection method for the smart power plant comprises the steps: obtaining multi-modal data in the smart power plant, and carrying out the multi-modal fusion based on the multi-modal data, and obtaining an equipment health state tensor; dynamically updating a fault propagation network based on the equipment health state tensor, and predicting a fault propagation path according to the updated fault propagation network; performing routing inspection path self-optimization based on risk nodes in the fault propagation path to obtain an optimized routing inspection path; and performing safety inspection based on the inspection path, and determining potential faults of the smart power plant based on an inspection result of safety inspection. The multi-modal data in the smart power plant are obtained and fused to obtain the equipment health state tensor, so that the limitation of traditional single-modal analysis is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power plant inspection, and in particular to a safety inspection method, device, equipment, storage medium and program product for a smart power plant. BACKGROUND

[0002] In modern power plant operation management, safety inspection can ensure stable operation of the power plant. With the continuous expansion of the scale of the power plant and the increase in the complexity of equipment, the traditional safety inspection method is difficult to meet the efficient and accurate safety management needs of the power plant. The concept of a smart power plant has emerged, which uses Internet of Things technology, big data analysis and artificial intelligence to realize real-time monitoring, intelligent analysis and early warning of the operation state of the power plant.

[0003] Currently, when performing safety inspection of a smart power plant, the dynamic changes of equipment states are not taken into account in the influence on each device in the smart power plant. SUMMARY

[0004] The main purpose of the present application is to provide a safety inspection method, device, equipment, storage medium and program product for a smart power plant, which aims to solve the technical problem that the existing smart power plant uses static path planning and does not take into account the dynamic changes of equipment states.

[0005] To achieve the above-mentioned purpose, the present application provides a safety inspection method for a smart power plant, which comprises: acquiring multi-modal data in the smart power plant, and performing multi-modal fusion based on the multi-modal data to obtain a device health state tensor; dynamically updating a fault propagation network based on the device health state tensor, and predicting a fault propagation path according to the updated fault propagation network; performing self-optimization of an inspection path based on a risk node in the fault propagation path to obtain an optimized inspection path; performing safety inspection based on the inspection path, and determining potential faults of the smart power plant based on the inspection results of the safety inspection.

[0006] In an embodiment, the step of dynamically updating the fault propagation network based on the device health state tensor comprises: performing dimension reduction processing based on the device health state tensor to obtain a three-dimensional health vector; updating the node attributes of the device nodes in the fault propagation network based on the three-dimensional health vector to obtain updated device nodes; performing edge weight updating and node degree correction of the fault propagation network through the updated device nodes to obtain an updated fault propagation network.

[0007] In an embodiment, the step of predicting the fault propagation path according to the updated fault propagation network comprises: constructing a time-varying adjacency matrix of the updated fault propagation network according to the updated edge weight; performing graph neural network forward inference prediction based on the time-varying adjacency matrix and the three-dimensional health vector to obtain a fault probability of each device node; selecting a risk node from each device node based on the fault probability; taking the risk node as a risk propagation starting point and generating a fault propagation path based on the updated edge weight.

[0008] In an embodiment, the step of performing inspection path self-optimization based on the risk nodes in the fault propagation path to obtain an optimized inspection path comprises: merging the risk nodes on each fault propagation path to obtain a risk node pool; constructing a risk subgraph based on the edges composed of the risk nodes and adjacent risk nodes in the risk node pool; performing inspection on the risk subgraph based on a preset double-target reward function to obtain a path reward score; performing inspection path self-optimization based on the path reward score to obtain an optimized inspection path.

[0009] In an embodiment, the step of performing inspection path self-optimization based on the path reward score to obtain an optimized inspection path comprises: taking the current inspection position as a state input, taking the inspection path as an action input, and taking the path reward score as a reward parameter input to obtain an optimized inspection path.

[0010] In an embodiment, the step of performing safety inspection based on the inspection path and determining potential faults of the smart power plant based on the inspection result of the safety inspection comprises: moving based on the inspection path and obtaining risk device data of the risk nodes in the inspection path; taking the risk device data as an inspection result of safety inspection and determining potential faults of the smart power plant based on the inspection result.

[0011] In addition, to achieve the above-mentioned purpose, the present application also provides a safety inspection device for a smart power plant, which comprises: a data acquisition module configured to acquire multi-modal data in the smart power plant and perform multi-modal fusion based on the multi-modal data to obtain a device health state tensor; The network update module is used to dynamically update the fault propagation network based on the device health status tensor, and predict the fault propagation path based on the updated fault propagation network. The path optimization module is used to perform self-optimization of the inspection path based on the risk nodes in the fault propagation path, so as to obtain the optimized inspection path. The intelligent inspection module is used to perform safety inspections based on the inspection path and to determine potential faults in the smart power plant based on the inspection results.

[0012] In addition, to achieve the above objectives, this application also proposes a safety inspection device for a smart power plant, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the safety inspection method for a smart power plant as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the safety inspection method for smart power plants as described above.

[0014] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the safety inspection method for smart power plants as described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: This application acquires multimodal data from smart power plants and performs multimodal fusion to obtain an equipment health state tensor. It then dynamically updates the fault propagation network based on this tensor and predicts fault propagation paths using the updated network. Finally, it performs self-optimization of inspection paths based on risk nodes within these paths, resulting in optimized inspection paths. Safety inspections are then conducted based on these paths, and the inspection results are used to identify potential faults in the smart power plant. Because the equipment health state tensor is obtained by fusing multimodal data from the smart power plant, the limitations of traditional single-modal analysis are avoided. By dynamically updating the fault propagation network and predicting fault propagation paths using the equipment health state tensor, and by performing self-optimization of inspection paths based on risk nodes, static path planning is transformed into dynamic, real-time data-driven reasoning, thus improving the applicability of the smart power plant safety inspection solution. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the safety inspection method for smart power plants provided in this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the safety inspection method for smart power plants in this application. Figure 3 This is a flowchart illustrating Embodiment 3 of the safety inspection method for smart power plants in this application. Figure 4 This is a schematic diagram of the module structure of the safety inspection device for a smart power plant, as described in an embodiment of this application. Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the safety inspection method of the smart power plant in this application embodiment.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application embodiment is as follows: acquire multimodal data in the smart power plant, and perform multimodal fusion based on the multimodal data to obtain the equipment health state tensor; dynamically update the fault propagation network based on the equipment health state tensor, and predict the fault propagation path based on the updated fault propagation network; perform self-optimization of the inspection path based on the risk nodes in the fault propagation path to obtain the optimized inspection path; perform safety inspection based on the inspection path, and determine the potential faults of the smart power plant based on the inspection results of the safety inspection.

[0023] This application provides a solution that avoids the limitations of traditional single-modal analysis by fusing multimodal data from smart power plants to obtain the equipment health state tensor. By dynamically updating the fault propagation network and predicting fault propagation paths using the equipment health state tensor, and by performing self-optimization of inspection paths based on risk nodes, static path planning is transformed into dynamic, real-time data-based reasoning, thus improving the applicability of the smart power plant safety inspection solution.

[0024] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a computer, server, or security inspection robot, or an electronic device or virtual device capable of performing the above functions. The following description uses a smart power plant security inspection device (hereinafter referred to as the inspection device) as an example to illustrate this embodiment and the subsequent embodiments.

[0025] Based on this, the embodiments of this application provide a safety inspection method for smart power plants, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the safety inspection method for smart power plants in this application.

[0026] In this embodiment, the safety inspection method for the smart power plant includes steps S10 to S40: Step S10: Obtain multimodal data from the smart power plant, and perform multimodal fusion based on the multimodal data to obtain the equipment health status tensor.

[0027] It is understood that the aforementioned multimodal data can be at least one type of data collected from different sensors, systems, and sources, which may have different structures and properties. This multimodal data can collectively describe the equipment status in a smart power plant from different dimensions and perspectives. In the embodiments of this application, the aforementioned multimodal data may specifically include pulse reflection signals, acoustic emission signals, infrared thermal imaging signals, electromagnetic disturbance signals, etc., and the embodiments of this application do not impose any limitations on this.

[0028] It should be noted that the device health status tensor in this application embodiment can be a high-bit vector obtained by fusing multimodal data. Specifically, the fusion method of multimodal data in this application embodiment can be based on a Transformer encoder or other methods, and this application embodiment does not limit this.

[0029] In some embodiments of this application, the input multimodal data is fused by a Transformer encoder to obtain a 128-dimensional vector, where each dimension can represent the cross-modal fusion features and can be used to directly quantify the health status of each device node in a smart power plant.

[0030] In its specific implementation, the inspection equipment in this application embodiment can acquire multimodal data from a smart power plant. By fusing this multimodal data, an equipment health state tensor can be obtained to comprehensively characterize the equipment's health status. Since the equipment health state tensor is obtained by fusing multimodal data from a smart power plant, the limitations of traditional single-modal analysis are avoided.

[0031] Step S20: Dynamically update the fault propagation network based on the device health status tensor, and predict the fault propagation path based on the updated fault propagation network.

[0032] It is understandable that in a smart power plant environment, the operation of equipment is not isolated; they are typically closely interconnected through the interaction of energy flow, information flow, and control flow. The aforementioned fault propagation network is a network established to describe the interdependencies between these devices or critical components and how faults propagate within them. Specifically, this fault propagation network can be represented in the form of a directed weighted graph or other forms, which are not limited in this embodiment.

[0033] It should be noted that the fault propagation network in this application embodiment may include device nodes and edges composed of adjacent device nodes. These device nodes may represent various devices or key components (such as generators, boilers, transformers, etc.) in a smart power plant, and each device node may correspond to a device health status tensor, which is used to characterize the real-time health status information of the device.

[0034] It should be explained that the aforementioned adjacent device nodes are those with mutual or unilateral dependencies. The edge between adjacent device nodes (such as device node A and device node B) can represent the functional dependency between these two device nodes. When device node B depends on device node A, it means that if device node A fails, device node B may also fail or experience performance degradation.

[0035] It should be noted that each edge in a fault propagation network has a corresponding edge weight, which can be used to quantify the strength of the dependency between adjacent device nodes. The higher the edge weight, the greater the probability that the fault will propagate along that path.

[0036] It is understandable that a decrease in the indicators in the device health state tensor indicates an increase in the failure risk level of that device node. Therefore, in the fault propagation network, the probability of that device node failing and the probability of fault propagation will also be higher. Thus, in this embodiment, the node degree and edge weights between adjacent device nodes can be updated based on the device health state tensor of each device node, thereby achieving dynamic updates to the fault propagation network.

[0037] It should be understood that the aforementioned fault propagation path refers to one or more propagation paths in a fault propagation network where a fault in a potentially faulty device node sequentially affects other device nodes along the edges. This fault propagation path can be used to describe the possible trajectory of a fault propagating within the system. Since different edges have different weights, the strength of these fault propagation paths will also vary. In practical applications, one or more unique fault propagation paths can be determined as the predicted fault propagation paths based on the edge weights involved in the multiple fault propagation paths corresponding to a device node.

[0038] In a specific implementation, the inspection device of this application embodiment can dynamically update the fault propagation network based on the device health status tensor and predict the fault propagation path in it according to the updated fault propagation network. It can capture signs when the performance of device nodes begins to slowly degrade. By dynamically updating the network, it can extract and identify device nodes that degrade due to the degradation of dependent device nodes, and realize early warning of device faults.

[0039] Step S30: Based on the risk nodes in the fault propagation path, perform self-optimization of the inspection path to obtain the optimized inspection path; Step S40: Perform a safety inspection based on the inspection path, and determine the potential faults of the smart power plant based on the inspection results of the safety inspection.

[0040] It should be noted that, along the fault propagation path, the fault probability of a device node can be determined based on its device health status tensor. When the fault probability of a device node exceeds a preset risk probability threshold, the device node can be identified as a risk node. Based on the location of risk nodes in the fault propagation path, the inspection path can be self-optimized, thereby enabling dynamic and automatic adjustment of the inspection route in the fault propagation network to obtain the final optimized inspection path.

[0041] It is understandable that by conducting safety inspections based on this optimized inspection path, the corresponding inspection results can be obtained, thereby enabling the identification of potential faults in smart power plants and the acquisition of a new round of multimodal data.

[0042] In some embodiments of this application, the step of performing a safety inspection based on the inspection path and determining potential faults of the smart power plant based on the inspection results includes: moving based on the inspection path and acquiring risk equipment data of risk nodes in the inspection path; using the risk equipment data as the inspection results of the safety inspection and determining potential faults of the smart power plant based on the inspection results.

[0043] It should be noted that the method for determining the aforementioned potential faults can be selected based on the needs of actual applications, and can be implemented using machine learning, neural networks, etc. This application does not impose any limitations on this.

[0044] It should be understood that the aforementioned risk device data can be multi-dimensional data related to risk device nodes, and this application embodiment does not limit this.

[0045] In its specific implementation, the inspection equipment in this embodiment can perform self-optimization of the inspection path based on risk nodes in the fault propagation path, resulting in an optimized inspection path. This enables the timely detection and handling of potential faults. Safety inspections are then conducted based on this optimized inspection path to identify potential faults in the smart power plant. This transforms static path planning into dynamic, real-time data-based reasoning, improving the applicability of the smart power plant safety inspection solution.

[0046] This application's embodiments acquire multimodal data from a smart power plant and perform multimodal fusion based on this data to obtain an equipment health state tensor. The fault propagation network is dynamically updated based on the equipment health state tensor, and fault propagation paths are predicted according to the updated network. Inspection paths are self-optimized based on risk nodes in the fault propagation paths to obtain optimized paths. Safety inspections are conducted based on these paths, and potential faults in the smart power plant are determined based on the inspection results. Because the equipment health state tensor is obtained by fusing multimodal data from the smart power plant, the limitations of traditional single-modal analysis are avoided. By dynamically updating the fault propagation network and predicting fault propagation paths using the equipment health state tensor, and by performing self-optimization of inspection paths based on risk nodes, static path planning is transformed into dynamic, real-time data-based reasoning, improving the applicability of the smart power plant safety inspection solution.

[0047] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 , Figure 2 This is a flowchart illustrating the second embodiment of the safety inspection method for smart power plants in this application.

[0048] like Figure 2As shown in the embodiment of this application, the step of dynamically updating the fault propagation network based on the device health state tensor includes: Step S21: Perform dimensionality reduction processing based on the device health status tensor to obtain a three-dimensional health vector; Step S22: Update the node attributes of the device nodes in the fault propagation network based on the three-dimensional health vector to obtain the updated device nodes; Step S23: Update the edge weights and correct the node degrees of the fault propagation network using the updated device nodes to obtain the updated fault propagation network.

[0049] It should be noted that the above-mentioned three-dimensional health vector can be a feature vector composed of three features. By performing dimensionality reduction processing on the device health state tensor, the three-dimensional health vector of the device node can be obtained.

[0050] In some embodiments of this application, a feature network for dimensionality reduction can be designed. This feature network may include two fully connected layers to map the device health state tensor to a three-dimensional health vector containing health score, stress concentration index, and degradation rate. The specific structure of the feature network is not limited in this application; it can be configured based on the actual application.

[0051] It should be noted that the above node attributes are the three features in the three-dimensional health vector, which can be specifically represented as: ; in, Represents a three-dimensional health vector. Indicates health score, Indicates the stress concentration index. Indicates the rate of degradation.

[0052] It is understandable that upon obtaining the 3D health vector corresponding to a device, this vector can be used to synchronously update the device nodes in the fault propagation network. This involves updating the node attributes of the device nodes to obtain the updated device nodes. Based on these updated device nodes, the updated fault propagation network can then be obtained.

[0053] In some embodiments of this application, the edge weights can be updated as follows: ; Where i and j are adjacent device nodes. For the updated edge weights, To update the previous edge weights, and This is the stress amplification factor (which can be preset based on actual conditions). Let i and j represent the propagation frequency coefficients between device nodes (i.e., the ratio of the number of fault propagations to the total number of faults).

[0054] In some embodiments of this application, the node degree can be updated as follows: ; in, Used to represent the updated degree of device node i. Used to represent the updated node degree of device node j.

[0055] It should be noted that the node degree mentioned above can be used to represent the number of edges connected to device node i.

[0056] In some embodiments of this application, the step of predicting the fault propagation path based on the updated fault propagation network includes: constructing a time-varying adjacency matrix of the updated fault propagation network based on the updated edge weights; performing forward inference prediction using a graph neural network based on the time-varying adjacency matrix and the three-dimensional health vector to obtain the fault probability of each device node; selecting risk nodes among the device nodes based on the fault probabilities; using the risk nodes as the starting point for risk propagation, and generating a fault propagation path based on the updated edge weights.

[0057] It should be noted that the aforementioned time-varying adjacency matrix can be a matrix composed of the edge weights in a fault propagation network that changes over time, and can be used to reflect a quantitative value of the fault propagation capability between device nodes. Specifically, the matrix elements in the time-varying adjacency matrix can be determined as follows: .

[0058] In this embodiment, a node feature matrix can be constructed based on the three-dimensional feature health vectors of each device node. The edge weights and edge connection indices between each device node in the fault propagation network can be determined based on the time-varying adjacency matrix.

[0059] It should be noted that the aforementioned edge connection index is a type of data that records all adjacent device nodes with fault propagation relationships. By inputting the node feature matrix, edge weights, and edge connection index into a pre-trained graph neural network model, the model can perform forward inference prediction to obtain the fault probability of each device node.

[0060] Specifically, the graph neural network model in this application embodiment may include graph convolutional layers and graph attention layers. For the graph convolutional layer, starting from each device node (taking the initial device node as an example), all adjacent device nodes can be found based on their edge connection indices. These adjacent device nodes are the first-order neighbor nodes of the initial device node. In this application embodiment, the initial device node can perform a weighted summation of the information of these neighbor nodes based on the edge weights with each first-order neighbor node, thereby realizing the aggregation of the neighbor node feature information (3D health vector) of the initial device node. The aggregated neighbor node information can be fused with the 3D health vector of the initial device node itself, and a richer feature representation can be generated through a preset nonlinear activation function.

[0061] In some embodiments of this application, deeper aggregation can be performed on the graph attention layer. In this case, each initial device node can not only consider its first-order neighbors, but also indirectly receive feature information from its neighbors' neighbors (i.e., second-order neighbors), thereby capturing a wider range of system influences. In the graph attention layer, a dynamic attention weight can be assigned to each neighbor of the initial device node. This weight can determine which first-order neighbors are more important at the current moment, thereby achieving focus on key propagation paths.

[0062] It should be noted that by aggregating multiple layers of graph convolutional layers and graph attention layers, high-level features of each device node can be obtained. These high-level features can be identified by a classifier to obtain a single numerical value (fault parameter). After processing the fault parameter through an activation function to achieve normalization, the fault probability of the fault node can be obtained.

[0063] It should be noted that the above classifier can be a classifier trained based on machine learning, and this application embodiment does not limit this.

[0064] It is understood that, upon obtaining the failure probability of each device node, risk nodes with a failure probability greater than a preset failure probability can be selected from these device nodes. This preset failure probability can be set according to the needs of actual applications, such as 40% or 50%, and this embodiment does not impose any limitations on it.

[0065] It should be noted that for risk nodes with a failure probability greater than a preset failure probability, they can be used as the starting point for failure propagation (i.e., the risk propagation starting point). At the same time, a failure propagation path is generated based on the updated edge weights of each edge connected to the risk starting point. Specifically, the failure propagation path can be one or several paths with the largest sum of edge weights among all propagation paths, and this application embodiment does not impose any restrictions on this.

[0066] This application embodiment performs dimensionality reduction based on the device health state tensor to obtain a three-dimensional health vector. The node attributes of device nodes in the fault propagation network are then updated based on this three-dimensional health vector, resulting in updated device nodes. Finally, the edge weights and node degrees of the fault propagation network are updated using these updated device nodes, resulting in an updated fault propagation network. Because dimensionality reduction is based on the device health state tensor, data complexity is reduced. Updating device nodes in the fault propagation network using the three-dimensional health vector transforms static paths into dynamic paths, enabling the fault propagation network to accurately reflect the current system operating status, laying the foundation for accurate risk identification. Furthermore, updating edge weights and correcting node degrees using the updated device nodes achieves greater precision in fault propagation probability, improving the accuracy of path prediction.

[0067] Based on the first and / or second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to the first and / or second embodiments described above can be referred to the above description and will not be repeated hereafter. Based on this, please refer to... Figure 3 , Figure 3 This is a flowchart illustrating the third embodiment of the safety inspection method for smart power plants in this application.

[0068] In the embodiments of this application, such as Figure 3 As shown, the step of performing self-optimization of the inspection path based on the risk nodes in the fault propagation path to obtain the optimized inspection path includes: Step S31: Merge the risk nodes on each of the fault propagation paths to obtain a risk node pool; Step S32: Construct a risk subgraph based on the risk nodes in the risk node pool and the edges formed by adjacent risk nodes; Step S33: Based on the preset dual-objective reward function, perform inspection on the risk subgraph to obtain the path reward score; Step S34: Based on the path reward score, perform self-optimization of the inspection path to obtain the optimized inspection path.

[0069] It should be noted that by merging risk nodes, a shift from decentralized multi-path management to centralized management can be achieved, improving the efficiency of inspection path self-optimization. During risk node merging, duplicate risk nodes can also be deduplicated to obtain a deduplicated risk node pool.

[0070] It's important to explain that after obtaining the risk node pool, the risk nodes can be sorted in descending order based on the merged failure probabilities, thereby adjusting the priority of each risk node. When constructing the risk subgraph, all risk nodes in the risk node pool can be used as core nodes, and target neighbor nodes directly connected to these core nodes can be identified in the original failure propagation network. Target neighbor nodes whose failure probabilities exceed the secondary risk probability threshold are then selected as the neighbor child nodes of the core nodes. This reconstructs the edge relationships between core nodes and between core nodes and their neighbor child nodes, resulting in the risk subgraph.

[0071] It should be noted that when reconstructing edge relationships, the edge weights can be set to the corresponding edge weights in the original fault propagation network.

[0072] In this embodiment of the application, the above-mentioned dual-objective reward function can be as follows: ; in, This represents the sum of the failure probabilities of the nodes involved in the inspection path. This represents the weight value of the failure probability term; This represents the sum of the propagation speeds of the nodes involved in the inspection path. This represents the weight value of the propagation speed term; This represents the sum of the lengths of the nodes involved in the inspection path. This represents the weight value of the path length penalty term.

[0073] It should be noted that the propagation speed mentioned above is related to the value of the edge weight. The larger the edge weight value, the faster the propagation speed. The specific corresponding relationship is not limited in the embodiments of this application.

[0074] It is understandable that the reward score corresponding to each inspection path in the risk subgraph can be calculated using a bi-objective reward function. In practical applications, the weights mentioned above can be adjusted using ant colony optimization, multi-objective particle swarm optimization, etc., and this application embodiment does not impose any limitations on this.

[0075] It should be noted that when performing self-optimization of the inspection path, the action space and state space of the inspection equipment can be determined first. The state space can include the current position of the inspection equipment, the environmental state, and the resource state, etc.; the action space can include each inspection path, and the reward score corresponding to the inspection path can be used as a reward parameter input. A preset loss function is then used to achieve self-optimization of the inspection path, resulting in the final optimized inspection path. That is, the step of self-optimizing the inspection path based on the path reward score to obtain the optimized inspection path includes: using the current inspection position as the state input, the inspection path as the action input, and the path reward score as the reward parameter input to obtain the optimized inspection path.

[0076] Specifically, this application embodiment can achieve self-optimization of the inspection path using a multilayer perceptron. The input of the multilayer perceptron is set to the current inspection location, the inspection path, and the path reward score, thereby selecting the next risk node of the inspection equipment from the inspection path. At the same time, the model parameters of the multilayer perceptron can be self-optimized by a preset inspection function to achieve self-optimization of the inspection path planning, resulting in an optimized inspection path.

[0077] This application embodiment merges risk nodes on each fault propagation path to obtain a risk node pool; constructs a risk subgraph based on the risk nodes in the risk node pool and the edges formed by adjacent risk nodes; performs inspections on the risk subgraph based on a preset bi-objective reward function to obtain path reward scores; and performs self-optimization of the inspection path based on the path reward scores to obtain an optimized inspection path, thereby improving the daily maintenance efficiency of inspections. By generating inspection routes through the risk subgraph, it ensures that in complex power plant environments, limited inspection resources can be prioritized for the most important risk prevention and control, achieving a balance between safety and economy.

[0078] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the safety inspection method of smart power plants in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0079] This application also provides a safety inspection device for smart power plants; please refer to [reference needed]. Figure 4 , Figure 4 This is a schematic diagram of the module structure of a safety inspection device for a smart power plant according to an embodiment of this application. The safety inspection device for the smart power plant includes: The data acquisition module 10 is used to acquire multimodal data in the smart power plant and perform multimodal fusion based on the multimodal data to obtain the equipment health status tensor. The network update module 20 is used to dynamically update the fault propagation network based on the device health status tensor, and predict the fault propagation path based on the updated fault propagation network. The path optimization module 30 is used to perform self-optimization of the inspection path based on the risk nodes in the fault propagation path to obtain the optimized inspection path. The intelligent inspection module 40 is used to perform safety inspections based on the inspection path and to determine potential faults in the smart power plant based on the inspection results.

[0080] The safety inspection device for smart power plants provided in this application, employing the safety inspection method for smart power plants described in the above embodiments, can solve the technical problem that existing smart power plants use static path planning and do not consider the dynamic changes in equipment status. Compared with the prior art, the beneficial effects of the safety inspection device for smart power plants provided in this application are the same as those of the safety inspection method for smart power plants provided in the above embodiments, and other technical features in the safety inspection device for smart power plants are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0081] This application provides a safety inspection device for a smart power plant. The safety inspection device for a smart power plant includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the safety inspection method for a smart power plant in the above embodiment 1.

[0082] The following is for reference. Figure 5 The diagram illustrates a structural schematic of a safety inspection device suitable for implementing the embodiments of this application in a smart power plant. The safety inspection device for a smart power plant in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), vehicle terminals (e.g., vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The safety inspection equipment for the smart power plant shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0083] like Figure 5As shown, the safety inspection equipment of a smart power plant may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the safety inspection equipment of the smart power plant. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the safety inspection equipment of the smart power plant to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows safety inspection equipment for a smart power plant with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0084] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0085] The safety inspection equipment for smart power plants provided in this application, employing the safety inspection method for smart power plants described in the above embodiments, can solve the technical problem that existing smart power plants use static path planning and do not consider the dynamic changes in equipment status. Compared with the prior art, the beneficial effects of the safety inspection equipment for smart power plants provided in this application are the same as those of the safety inspection method for smart power plants provided in the above embodiments, and other technical features of this safety inspection equipment are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0086] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0087] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0088] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the safety inspection method for a smart power plant in the above embodiments.

[0089] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0090] The aforementioned computer-readable storage medium may be included in the safety inspection equipment of the smart power plant; or it may exist independently and not be installed in the safety inspection equipment of the smart power plant.

[0091] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the safety inspection equipment of the smart power plant, cause the safety inspection equipment of the smart power plant to: Acquire multimodal data from the smart power plant, and perform multimodal fusion based on the multimodal data to obtain the equipment health status tensor; The fault propagation network is dynamically updated based on the device health status tensor, and the fault propagation path is predicted based on the updated fault propagation network. Based on the risk nodes in the fault propagation path, the inspection path is self-optimized to obtain the optimized inspection path. Safety inspections are conducted based on the inspection path, and potential faults in the smart power plant are determined based on the inspection results.

[0092] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0093] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0094] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0095] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned safety inspection method for smart power plants. This addresses the technical problem that existing smart power plants employ static path planning, failing to consider dynamic changes in equipment status. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the safety inspection method for smart power plants provided in the above embodiments, and will not be elaborated upon here.

[0096] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the safety inspection method for a smart power plant as described above.

[0097] The computer program product provided in this application can solve the technical problem that existing smart power plants use static path planning, which does not take into account the dynamic changes in equipment status. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the safety inspection method for smart power plants provided in the above embodiments, and will not be repeated here.

[0098] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A safety inspection method for a smart power plant, characterized in that, The method includes: Acquire multimodal data from the smart power plant, and perform multimodal fusion based on the multimodal data to obtain the equipment health status tensor; The fault propagation network is dynamically updated based on the device health status tensor, and the fault propagation path is predicted based on the updated fault propagation network. Based on the risk nodes in the fault propagation path, the inspection path is self-optimized to obtain the optimized inspection path. Safety inspections are conducted based on the inspection path, and potential faults in the smart power plant are determined based on the inspection results.

2. The safety inspection method for a smart power plant as described in claim 1, characterized in that, The step of dynamically updating the fault propagation network based on the device health status tensor includes: Dimensionality reduction is performed on the device health status tensor to obtain a three-dimensional health vector; The node attributes of the device nodes in the fault propagation network are updated based on the three-dimensional health vector to obtain the updated device nodes. The updated fault propagation network is obtained by updating the edge weights and correcting the node degrees of the updated device nodes.

3. The safety inspection method for a smart power plant as described in claim 2, characterized in that, The step of predicting the fault propagation path based on the updated fault propagation network includes: Construct the updated time-varying adjacency matrix of the fault propagation network based on the updated edge weights; Based on the time-varying adjacency matrix and the three-dimensional health vector, a graph neural network is used for forward inference prediction to obtain the failure probability of each device node; Based on the failure probability, risk nodes are selected from each of the device nodes; The risk node is used as the starting point for risk propagation, and a fault propagation path is generated based on the updated edge weights.

4. The safety inspection method for a smart power plant as described in claim 1, characterized in that, The step of performing self-optimization of the inspection path based on the risk nodes in the fault propagation path to obtain the optimized inspection path includes: The risk nodes on each of the aforementioned fault propagation paths are merged to obtain a risk node pool; A risk subgraph is constructed based on the risk nodes in the risk node pool and the edges formed by adjacent risk nodes. Based on a preset dual-objective reward function, an inspection is performed on the risk subgraph to obtain a path reward score; The inspection path is self-optimized based on the path reward score to obtain the optimized inspection path.

5. The safety inspection method for a smart power plant as described in claim 4, characterized in that, The step of self-optimizing the inspection path based on the path reward score to obtain the optimized inspection path includes: The current inspection location is used as the status input, the inspection path is used as the action input, and the path reward score is used as the reward parameter input to obtain the optimized inspection path.

6. The safety inspection method for a smart power plant as described in claim 1, characterized in that, The steps of conducting safety inspections based on the inspection path and determining potential faults in the smart power plant based on the inspection results include: Based on the inspection path movement, risk equipment data of risk nodes in the inspection path are obtained; The data on the risky equipment is used as the inspection result of the safety inspection, and potential faults of the smart power plant are determined based on the inspection result.

7. A safety inspection device for a smart power plant, characterized in that, The safety inspection device for the smart power plant includes: The data acquisition module is used to acquire multimodal data in the smart power plant and perform multimodal fusion based on the multimodal data to obtain the equipment health status tensor. The network update module is used to dynamically update the fault propagation network based on the device health status tensor, and predict the fault propagation path based on the updated fault propagation network. The path optimization module is used to perform self-optimization of the inspection path based on the risk nodes in the fault propagation path, so as to obtain the optimized inspection path. The intelligent inspection module is used to perform safety inspections based on the inspection path and to determine potential faults in the smart power plant based on the inspection results.

8. A safety inspection device for a smart power plant, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the safety inspection method for a smart power plant as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the safety inspection method for a smart power plant as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the safety inspection method for a smart power plant as described in any one of claims 1 to 6.

Images