Intelligent operation and maintenance and emergency response linked digital analysis, prediction and disposal system

Abstract

The invention discloses an intelligent operation and maintenance and emergency response linked digital analysis, prediction and disposal system, and relates to the technical field of intelligent operation and maintenance, and the system comprises a fault thermodynamic analysis module which constructs a fault conduction thermodynamic diagram based on equipment topology, resource states and fault data, and quantifies the fault diffusion intensity; the blocking point identification module calculates and positions a key blocking node through the node blocking efficiency value, and optimizes the resource input efficiency; the feedback channel construction module plans an optimal resource transmission path and minimizes transmission delay and loss; the dynamic processing execution module generates a precise blocking instruction and a resource circulation scheme to realize fault closed-loop processing; the data interaction module guarantees standardized interaction of multi-source data, the system adopts an edge + cloud hybrid architecture, and is combined with a machine learning self-optimization model, so that the accuracy, the real-time performance and the resource utilization rate of fault handling are improved, and the system is suitable for intelligent operation and maintenance of complex industrial scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent operation and maintenance, more particularly to a digital analysis, prediction and disposal system for linkage between intelligent operation and emergency response. BACKGROUND

[0002] Under the background of deepening industrial digital transformation, the operation and emergency response of complex equipment clusters face multi-dimensional challenges. Modern industrial systems exhibit deep coupling characteristics of physical equipment, network architecture and business processes. The device topology evolves from linear connection to heterogeneous mesh structure. The resource pool covers multiple types of elements such as computing, storage, spare parts and bandwidth. The fault conduction path presents dynamic and nonlinear characteristics. The traditional operation and maintenance mode relies on preset rules and manual experience, which is difficult to meet the fault disposal needs of such complex systems. On the one hand, when a fault occurs, a single-point alarm often triggers a chain reaction. The existing system lacks quantitative characterization of the diffusion intensity of the fault in the topology network and can only locate the initial fault point, but cannot predict the heat conduction trend, resulting in a lag in blocking measures compared to the speed of fault spread. On the other hand, there is a time and space mismatch between resource scheduling and fault disposal. The dynamic matching of resource pool and fault area relies on static path planning, without considering the real-time changes of transmission delay and loss. Moreover, the resource feedback lacks a collaborative mechanism with blocking operations, resulting in resource redundancy or insufficient supply at critical nodes. The allocation of computing power at the edge and cloud and the data interaction delay further exacerbate the efficiency loss of fault response. When the network fluctuates or is interrupted, systems that rely on centralized decision-making in the cloud are prone to disposal stagnation, while pure edge deployment is limited by local computing power and cannot complete global path optimization. In addition, the value of historical data has not been fully tapped, and the selection of key blocking points and resource channel planning lack self-learning ability, making it difficult to adapt to changes in system characteristics caused by equipment aging and business iteration. These problems collectively make it difficult to balance the accuracy, real-time performance and resource utilization of fault disposal in complex industrial scenarios, and there is an urgent need to build an intelligent system that integrates digital analysis, prediction and linkage disposal.

[0003] The prior art has problems such as lack of quantitative characterization of fault conduction, lack of coordination between resource scheduling and blocking, and insufficient edge cloud collaboration. SUMMARY

[0004] To overcome the problems of lack of quantitative characterization of fault conduction, lack of coordination between resource scheduling and blocking, and insufficient edge cloud collaboration in the prior art, the present application discloses a digital analysis, prediction and disposal system for linkage between intelligent operation and emergency response, which can effectively solve the above technical problems.

[0005] To solve the above technical problems, the technical solution of the present application is as follows:

[0006] The digital analysis, prediction and disposal system for linkage between intelligent operation and emergency response comprises:

[0007] A fault heat analysis module is configured to collect device topology knowledge base, resource pool state monitoring data and real-time fault prediction data, construct a fault heat conduction map based on a heat conduction map generation algorithm, and represent the heat conduction distribution of the fault in the device topology.

[0008] A blocking point identification module is configured to locate multiple key blocking nodes by calculating the full path score of the device topology in real time according to the key blocking point positioning algorithm, and taking the node that can block the maximum heat value with the unit resource input as the standard.

[0009] A counter-feeding channel construction module is configured to construct resource counter-feeding channels according to the device topology relationship, resource pool state and fault heat conduction map, and to schedule resource pool resources to counter-feed the fault disposal.

[0010] A dynamic disposal execution module is configured to generate precise blocking instructions and resource circulation schemes according to the multiple key blocking nodes output by the key blocking point identification module and the resource counter-feeding channels constructed by the counter-feeding channel construction module, execute fault disposal, and feed back the efficiency feedback data to the fault heat analysis module.

[0011] A data interaction module is configured to interact with the fault prediction module, device topology knowledge base and resource pool state monitoring module to obtain real-time data, knowledge base information and state monitoring data, and output precise blocking instructions and resource circulation schemes to the outside.

[0012] Preferably, the fault heat analysis module includes the following when generating the fault heat conduction map:

[0013] Extracting the device connection relationship and physical location information in the device topology knowledge base as the basis for the fault conduction path;

[0014] Collecting resource pool state monitoring data, including resource load rate, resource type distribution and resource available amount, and marking the influence weight of the resource on the fault conduction;

[0015] Receiving real-time fault data output by the fault prediction module, including fault type, fault initial node and fault development trend, combining the device topology and resource state, and calculating the heat value of each node in the device topology through a heat conduction simulation algorithm to generate a fault heat conduction map, the formula being:

[0016]

[0017] wherein, is the fault heat value of node i, is the fault severity factor, is the resource influence factor, is the conduction coefficient, + + = 1.

[0018] Preferably, in the key blocking point positioning algorithm of the blocking point identification module, the node selection criterion of the maximum heat value blocked by unit resource investment is quantified by the following method:

[0019] Define the node blocking efficiency value , wherein is the amount of total heat value in the fault conduction heat map after blocking node j, is the resource cost required to block node j;

[0020] Iterate through all nodes in the device topology, calculate the of each node, sort them in descending order according to , and select multiple key blocking nodes;

[0021] When calculating the full path score, the path score , wherein P is the conduction path in the device topology, and by iterating through all possible paths, the path and associated nodes with the best fault conduction suppression effect are selected.

[0022] Preferably, when the resource backfeeding channel construction module constructs the resource backfeeding channel, it includes:

[0023] Analyze the resource demand type and demand priority of the fault impact area in the fault conduction heat map;

[0024] Query the resource pool state monitoring data, match the location, type and quantity of available resources;

[0025] Based on the device topology relationship, use the shortest path algorithm and resource scheduling algorithm to plan the resource transportation path from the resource pool to the fault impact area, construct the resource backfeeding channel, and the channel construction satisfies the formula:

[0026]

[0027] , wherein is the resource transmission delay of path node k, is the resource transmission loss coefficient of path node k, and the optimal resource backfeeding channel is determined by minimizing the weighted sum of transmission delay and loss.

[0028] Preferably, when the dynamic disposal execution module generates accurate blocking instructions, it includes for each key blocking node:

[0029] Analyze the device type, running state and carried service of the key blocking node;

[0030] Combined with the fault type, formulate accurate blocking instructions including resource injection strategy, service migration strategy and hardware restart / reset strategy;

[0031] The instruction generation follows the principle of minimum business impact, and the business impact degree evaluation model is used Ensure that the business impact degree I after instruction execution is lower than the preset threshold, wherein The business importance is The business interruption duration is The business recovery cost is

[0032] Preferably, the dynamic disposal execution module includes the following when generating the resource circulation scheme:

[0033] Identify the type of disposal resource to be released;

[0034] If it is a computing resource, perform an injection adjacent overload node operation, update the resource pool computing resource distribution, and trigger real-time heat map updating;

[0035] If it is a spare part resource, perform a return dynamic resource pool operation to supplement the resource pool spare part reserve;

[0036] If it is network bandwidth, perform a rerouting to the fault area operation to adjust the network topology bandwidth allocation;

[0037] After all resource circulation operations are completed, call the fault heat analysis module interface to perform real-time heat map updating, and correct the heat value calculation model of the fault conduction heat map according to the heat value changes before and after resource circulation and the resource input-output ratio.

[0038] Preferably, when the data interaction module interacts with external modules, a standardized data interface is used, specifically including:

[0039] Interact with the fault prediction module to receive fault prediction data encapsulated in JSON format, including fault probability, fault type, and predicted occurrence timestamp, and the data transmission follows the MQTT protocol;

[0040] Interact with the device topology knowledge base to read device three-dimensional model data and connection relationship graph, and use RESTful API for data query and update;

[0041] Interact with the resource pool state monitoring module to obtain real-time state data of resources, which are stored in a time series database format and synchronized through a Telegraf collection agent;

[0042] When outputting precise blocking instructions and resource circulation schemes to the outside, the instructions and schemes are encapsulated in XML format and transmitted to the operation and maintenance execution terminal or the upper management system through the HTTPS protocol.

[0043] Preferably, in the key blocking point positioning algorithm of the blocking point identification module, the full path score calculation is specifically as follows:

[0044] For each conduction path P in the device topology, define a path score wherein is the failure heat value of path node n, is the resource cost of blocking node n;

[0045] Traverse all possible conduction paths, calculate the of each path, and select the path with the highest score and the associated nodes as the key blocking point candidate;

[0046] Perform blocking synergy verification on the candidate nodes. If multiple candidate nodes belong to the same short path, merge and evaluate the joint blocking effectiveness to avoid repeated resource investment. The joint blocking effectiveness is wherein is the resource investment overlap degree between nodes, is the overlap penalty coefficient.

[0047] Preferably, the system further comprises a machine learning training module for:

[0048] Collect historical fault handling data, including fault type, key blocking point selection, resource feedback channel planning, and handling effectiveness result;

[0049] Construct a training data set, with fault recovery time, business interruption loss, and resource investment cost as the optimization objectives, and use a reinforcement learning algorithm to train a key blocking point positioning model and a resource feedback channel planning model;

[0050] Periodically update the model parameters optimized by training to the blocking point identification module and the feedback channel construction module to realize algorithm self-optimization. The model training objective function is:

[0051]

[0052] wherein is the fault recovery time, is the business interruption loss, is the resource investment cost, + + = 1.

[0053] Preferably, when the system is deployed, a hybrid architecture of edge nodes + cloud platform is used:

[0054] The edge node is deployed locally in the device room and integrates a lightweight version of the failure heat analysis module and a basic instruction execution unit of the dynamic handling execution module, which is used for real-time collection of device data and execution of rapid blocking instructions;

[0055] The cloud platform deployment system deploys complete function modules, carries out fault conduction thermal map calculation of large data, global path score calculation of key blocking points and resource counter-acting channel cross-region planning;

[0056] The edge node and the cloud platform carry out data interaction through 5G / industrial Ethernet, adopt an edge computing unloading strategy, unload complex computing tasks to the cloud when network bandwidth is sufficient, enable a local cache model to independently execute basic fault disposal when network interruption occurs, guarantee system resilience, and data interaction follows an OPC UA protocol, supports data encryption and breakpoint resume.

[0057] Compared with the prior art, the beneficial effects of the present application are that the digital analysis, prediction and disposal system of intelligent operation and emergency response linkage effectively solves the problems of non-quantitative representation of fault conduction, lack of resource scheduling and blocking cooperation and insufficient edge cloud cooperation in the prior art through innovative modules and algorithms, and improves operation efficiency and system reliability.

[0058] Firstly, the system realizes the quantitative characterization of fault conduction through the fault thermal analysis module. The module collects device topology knowledge base, resource pool state monitoring data and fault prediction real-time data, and constructs the fault conduction thermal map based on the conduction thermal map generation algorithm. This process not only represents the conduction thermal distribution of the fault in the device topology, but also calculates the thermal value of each node through the heat conduction simulation algorithm, providing basis for fault positioning and prediction. This quantitative characterization enables operation and maintenance personnel to intuitively understand the propagation path and impact range of the fault, so as to take measures in advance to avoid further spread of the fault; Secondly, the system realizes the collaborative optimization of resource scheduling and blocking through the blocking point identification module and the counter-feeding channel construction module. The blocking point identification module uses the key blocking point positioning algorithm to calculate the full path score of the device topology in real time, and locates multiple key blocking nodes. The counter-feeding channel construction module constructs resource counter-feeding channels based on device topology relationship, resource pool state and fault conduction thermal map, which are used to schedule resource pool resources for counter-feeding fault disposal. This collaborative mechanism ensures that resources can be quickly and accurately scheduled to key locations when a fault occurs, minimizing the impact of the fault on business. At the same time, by minimizing the weighted sum of resource transmission delay and loss, the system further optimizes the resource scheduling efficiency and avoids resource waste; Finally, the system solves the problem of insufficient edge cloud cooperation by using a hybrid architecture of edge nodes and cloud platforms. The edge nodes are deployed locally in the device room, integrating a lightweight version of the fault thermal analysis module and a dynamic disposal execution module basic instruction execution unit, which are used to collect device data and execute rapid blocking instructions in real time. The cloud platform deploys complete system function modules for large data fault conduction thermal map calculation, key blocking point global path score calculation and resource counter-feeding channel cross-regional planning. The edge nodes and the cloud platform interact through 5G / industrial Ethernet, using edge computing offloading strategy. When the network bandwidth is sufficient, complex computing tasks are offloaded to the cloud. When the network is interrupted, the edge node enables the local cache model to execute basic fault disposal independently. This hybrid architecture not only improves the response speed and resilience of the system, but also ensures the security and reliability of data transmission through data encryption and breakpoint resume technology. In summary, the system improves operation and maintenance efficiency and system reliability through quantitative characterization of fault conduction, collaborative optimization of resource scheduling and blocking, and efficient cooperation of edge cloud, providing support for efficient operation and maintenance and rapid emergency response of complex industrial systems. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be obtained from the provided drawings without creative labor.

[0060] Figure 1 The system structure diagram of the present application. DETAILED DESCRIPTION

[0061] The drawings are only used for illustrative description, and cannot be understood as a limitation on the patent;

[0062] In order to better illustrate the present embodiment, some components in the drawings may be omitted, enlarged or reduced, and do not represent the actual product size;

[0063] For those skilled in the art, it is understandable that some known structures in the drawings and their descriptions may be omitted.

[0064] The technical solutions of the present application will be further described below in combination with the drawings and embodiments.

[0065] EMBODIMENT

[0066] The present embodiment takes a certain chemical production park as the application scene, which contains multiple sets of reaction devices, conveying pump groups, control systems and utility devices, forms a complex interconnection topology through industrial network, the device cluster covers production devices of different years and different types, the resource pool contains computing nodes required for process control, various device spare parts and network transmission resources, and undertakes the stable operation guarantee task of continuous production process.

[0067] Please refer to Figure 1 , the digital analysis prediction and disposal system of intelligent operation and emergency response linkage, the system comprises:

[0068] A fault heat analysis module is used to collect device topology knowledge base, resource pool state monitoring data and fault prediction real-time data, construct a fault heat conduction map based on a heat conduction map generation algorithm, and represent the heat conduction distribution of the fault in the device topology;

[0069] A block point identification module adopts a key block point positioning algorithm, takes the node that can block the maximum heat value with unit resource input as the standard, calculates the full path score of the device topology in real time, and locates multiple key block nodes;

[0070] A counter-benefiting channel construction module constructs resource counter-benefiting channels according to the device topology relationship, resource pool state and fault heat conduction map, and is used for scheduling resource pool resources to counter-benefit the fault disposal;

[0071] A dynamic treatment execution module generates precise blocking instructions and resource recycling schemes based on the multiple key blocking nodes output by the key blocking point identification module and the resource recycling channels constructed by the resource recycling channel construction module, executes fault treatment, and feeds back performance feedback data to the fault heat analysis module;

[0072] A data interaction module is configured to interact with the fault prediction module, the device topology knowledge base, and the resource pool state monitoring module to obtain real-time data, knowledge base information, and state monitoring data, and output precise blocking instructions and resource recycling schemes to the outside.

[0073] The fault heat analysis module includes the following when generating the fault conduction heat map:

[0074] The device connection relationship and physical location information in the device topology knowledge base are extracted as the basis for the fault conduction path;

[0075] Resource pool state monitoring data, including resource load rate, resource type distribution, and resource availability, are collected to mark the influence weight of resources on fault conduction;

[0076] Real-time fault data output by the fault prediction module, including fault type, fault initial node, and fault development trend, are received, and the device topology and resource state are combined to calculate the heat value of each node in the device topology through heat conduction simulation algorithm to generate the fault conduction heat map, with the formula being:

[0077]

[0078] wherein, is the fault heat value of node i, is the fault severity factor, is the resource influence factor, is the conduction coefficient, + + = 1.

[0079] In the key blocking point positioning algorithm of the blocking point identification module, the selection criteria for the unit resource investment to block the node with the maximum heat value are quantified in the following way:

[0080] The node blocking performance value is defined as wherein, is the amount of total heat value decrease in the fault conduction heat map after blocking node j, is the resource cost required for blocking node j;

[0081] All nodes in the device topology are traversed to calculate the of each node, and the is sorted from large to small, and multiple key blocking nodes are selected;

[0082] The path score is calculated when the full path score is calculated Where P is the conduction path in the device topology, by traversing all possible paths, the path and associated nodes with the best fault conduction suppression effect are screened out.

[0083] The anti-nourishing channel construction module includes the following when constructing the resource anti-nourishing channel:

[0084] Analyze the resource demand type and demand priority of the fault impact area in the fault conduction heat map;

[0085] Query resource pool state monitoring data, match the location, type and quantity of available resources;

[0086] Based on the device topology relationship, use the shortest path algorithm and resource scheduling algorithm to plan the resource delivery path from the resource pool to the fault impact area, construct the resource anti-nourishing channel, and the channel construction satisfies the formula:

[0087]

[0088] Wherein is the resource transmission delay of path node k, is the resource transmission loss coefficient of path node k, by minimizing the weighted sum of transmission delay and loss, the optimal resource anti-nourishing channel is determined.

[0089] The dynamic disposal execution module includes the following when generating precise blocking instructions for each key blocking node:

[0090] Analyze the device type, running state and carried service of the key blocking node;

[0091] Combined with the fault type, formulate precise blocking instructions containing resource injection strategy, service migration strategy and hardware restart / reset strategy;

[0092] The instruction generation follows the principle of minimum service impact, through the service impact degree evaluation model Ensure that the service impact degree I after executing the instructions is lower than the preset threshold, wherein is the service importance, is the service interruption time, is the service recovery cost.

[0093] The dynamic disposal execution module includes the following when generating resource circulation scheme:

[0094] Identify the type of released disposal resources;

[0095] If it is a computing resource, execute the injection adjacent overload node operation, update the resource pool computing resource distribution, and trigger real-time heat map update;

[0096] If it is spare resource, execute the return dynamic resource pool operation to supplement the resource pool spare reserve;

[0097] If it is network bandwidth, execute the re-routing to the failure area operation to adjust the network topology bandwidth allocation;

[0098] After all resource circulation operations are completed, call the fault heat analysis module interface to execute real-time heat map updating, and correct the heat value calculation model of the fault conduction heat map according to the heat value changes before and after resource circulation and the resource input-output ratio.

[0099] When the data interaction module interacts with external modules, a standardized data interface is adopted, specifically including:

[0100] Interact with the fault prediction module to receive fault prediction data encapsulated in JSON format, including fault probability, fault type, and predicted occurrence timestamp, and the data transmission complies with the MQTT protocol;

[0101] Interact with the device topology knowledge base to read device three-dimensional model data and connection relationship graph, and use RESTful API for data query and update;

[0102] Interact with the resource pool state monitoring module to obtain real-time state data of resources, and the data is stored in a time series database format and synchronized through a Telegraf collection agent;

[0103] When outputting precise blocking instructions and resource circulation schemes to the outside, the instructions and schemes are encapsulated in XML format and transmitted to the operation and maintenance execution terminal or the upper management system through the HTTPS protocol.

[0104] In the key blocking point positioning algorithm of the blocking point identification module, the full path score calculation is specifically:

[0105] For each conduction path P in the device topology, define the path score , wherein is the fault heat value of the path node n, is the resource cost of the blocking node n;

[0106] Iterate through all possible conduction paths to calculate the of each path, and select the path with the highest score and the associated node as the key blocking point candidate;

[0107] Verify the blocking synergy of the candidate node, if multiple candidate nodes belong to the same short path, then combine to evaluate the joint blocking effectiveness to avoid repeated resource investment, and the joint blocking effectiveness , wherein is the resource investment overlap degree between nodes, is the overlap penalty coefficient.

[0108] The system also includes a machine learning training module for:

[0109] Collecting historical fault handling data, including fault type, key blocking point selection, resource counter-attack channel planning, and handling performance results;

[0110] Constructing a training data set, with fault recovery time after fault handling, business interruption loss, and resource investment cost as optimization objectives, using reinforcement learning algorithm to train key blocking point positioning model and resource counter-attack channel planning model;

[0111] Periodically update the model parameters optimized by training to the blocking point identification module and counter-attack channel construction module to realize algorithm self-optimization, and the model training objective function is:

[0112]

[0113] Wherein is the fault recovery time, is the business interruption loss, is the resource investment cost, + + = 1.

[0114] When the system is deployed, a hybrid architecture of edge node + cloud platform is used:

[0115] The edge node is deployed locally in the equipment room, integrating a lightweight version of the fault thermal analysis module and a basic instruction execution unit of the dynamic handling execution module, for real-time collection of equipment data and execution of rapid blocking instructions;

[0116] The cloud platform deploys complete function modules of the system, performs large data fault conduction thermal map calculation, key blocking point global path score calculation, and resource counter-attack channel cross-regional planning;

[0117] The edge node and the cloud platform interact through 5G / industrial Ethernet, using edge computing offloading strategy, offloading complex computing tasks to the cloud when network bandwidth is sufficient, and enabling the edge node to execute basic fault handling independently using local cache model when network is interrupted, to ensure system resilience. Data interaction complies with OPC UA protocol, supports data encryption and breakpoint resume.

[0118] The system uses a hybrid architecture of edge node + cloud platform for deployment:

[0119] Edge node: Deployed in the control rooms and near the equipment of each production device in the park, the edge node integrates a lightweight version of the fault thermal analysis module and the basic instruction unit of the dynamic handling execution module. It directly accesses the sensor network of the field equipment, collects real-time operation parameters such as temperature, pressure, and flow, and can quickly perform simple fault blocking operations such as emergency valve shutdown and temporary suspension of local equipment operation, ensuring preliminary control at the initial stage of the fault.

[0120] Cloud platform: Deployed in the central control center of the park, it contains the complete functional modules of the system and has data analysis and processing capabilities. The cloud platform is responsible for processing large-scale device topology data, resource state data, and fault history data, completing tasks such as fault conduction thermal map drawing, full-path score calculation, and cross-regional resource feedback channel planning.

[0121] Data interaction mechanism: Edge nodes and cloud platforms interact through industrial Ethernet and wireless networks, using communication protocols that meet industrial standards, supporting data encryption and breakpoint resume. When the network state is good, edge nodes upload detailed data to the cloud for deep analysis and global optimization. When the network is abnormal or interrupted, edge nodes can rely on locally cached models and historical data to independently execute basic fault handling processes, ensuring the basic stability of the production process.

[0122] In specific implementation, device topology data: Extract the connection relationships and physical location information of all devices in the park from the device topology knowledge base, such as that reaction kettle A is connected to delivery pump B through a pipeline, and that the temperature sensor of reaction kettle A is connected to control system C through a cable. These connection relationships and relative positions provide a basic framework for analyzing possible fault conduction paths.

[0123] Resource pool state data: Collect real-time state of resource pool, including load of each computing node, inventory quantity and distribution location of different types of spare parts, bandwidth occupation of network link, etc. Mark the influence weight of different resources on fault conduction according to these data, such as that the support ability of a computing node with high load for fault analysis and processing decreases, and the inhibition effect on fault conduction weakens; while the area with sufficient spare parts inventory can obtain resource support faster during fault handling, reducing the risk of fault diffusion.

[0124] Fault prediction data: Receive real-time fault information sent by the fault prediction module, including fault type such as temperature anomaly and pressure fluctuation, initial device node, and fault trend prediction, etc. These information provide input for thermal value calculation and fault conduction analysis.

[0125] The fault heat analysis module uses a heat conduction simulation algorithm, combined with the above collected device topology, resource state and fault data, to calculate the fault heat value of each device node, and its calculation formula is:

[0126] wherein, is the fault heat value of node i, is the fault severity factor, is the resource influence factor, is the conduction coefficient, + + = 1, specifically the fault severity factor is determined according to the fault type and the degree of influence on production, for example, the severity of the abnormal temperature of the reaction kettle is higher than that of the slight leakage of the ordinary valve; the resource influence factor is related to the resource state around the node, the more abundant the resources, the lower the factor value, indicating that the inhibition of the resources on the fault conduction is stronger; the conduction coefficient reflects the coupling degree between the node and the fault source, the closer the connection and the more direct the influence, the larger the conduction coefficient value.

[0127] According to the calculated heat value of each node, a visual fault conduction heat map is generated, in which different colors represent different heat values, and the deeper the color, the more serious the degree of influence of the node by the fault and the higher the probability of fault conduction. Through the heat map, the operation and maintenance personnel can intuitively master the distribution and spread of the fault in the device topology, and provide a clear visual reference for subsequent blocking decisions.

[0128] The blocking point identification module evaluates the blocking value of each node by calculating the node blocking efficiency value, which is defined as: wherein, is the amount of total heat value in the fault conduction heat map after blocking node j, is the resource cost required to block the node, the module traverses all nodes in the device topology, and calculates the blocking efficiency value of each node one by one, for example, blocking a certain key valve node may greatly reduce the total heat value, but the resource cost required is also higher; while blocking a auxiliary monitoring node, the total heat value may decrease limitedly, but the resource cost is lower. Through calculation and comparison, each node is sorted from high to low according to the blocking efficiency value, and the nodes with high blocking value are preliminarily screened out as candidate key blocking nodes.

[0129] In order to more accurately locate the key blocking node, the module analyzes all possible fault conduction paths in the device topology, and calculates the score of each path, and the calculation formula of the path score is:

[0130] P is a certain conduction path in the device topology, the failure thermal value of the path node n, the resource cost of the blocking node n, traverse all possible conduction paths, calculate the score of each path, select the path with the highest score and the associated node as the candidate of the key blocking point, then perform blocking synergy verification on these candidate nodes, if multiple candidate nodes belong to a shorter path, combine them to evaluate the joint blocking effectiveness, the calculation formula is:

[0131] wherein, the resource input overlap degree between nodes, the overlap penalty coefficient, through this synergy verification, avoid duplication in resource input, ensure that resources can be used most reasonably, and finally determine multiple key blocking nodes.

[0132] The counter-feeding channel construction module analyzes the demand type and demand priority of various resources in the failure influence area according to the failure conduction thermal map, for example, the temperature abnormality of the reaction kettle in the failure influence area may require temperature control related spare parts and additional computing resources to support real-time temperature simulation and regulation; and the pump failure may require pump body spare parts and maintenance personnel resources.

[0133] Query the state data of the resource pool, find the available resources that match the demand of the failure influence area, and determine the location, type and quantity of these resources, based on the device topology relationship, use the shortest path algorithm and resource scheduling algorithm to plan the resource delivery path from the resource pool to the failure influence area, and the channel construction satisfies the following formula:

[0134] wherein the resource transmission delay of path node k, the resource transmission loss coefficient of path node k, by minimizing the weighted sum of transmission delay and loss, determine the optimal resource counter-feeding channel, ensure that resources can be delivered to where they are needed with the highest efficiency and the smallest loss, and provide timely resource support for fault handling.

[0135] The dynamic handling execution module analyzes the device type, current running state and carried business of each key blocking node, combined with the specific type of failure, formulates precise blocking instructions including resource injection strategy, business migration strategy and hardware restart / reset strategy, the instruction generation follows the principle of minimum business impact, the possible impact of the instruction is evaluated through the business impact degree evaluation model, the calculation formula is:

[0136] wherein, the business importance, For the duration of business interruption, To minimize the cost of business recovery, ensure that the business impact I after the instruction is executed is below a preset threshold, effectively blocking the transmission of faults while minimizing interference with normal production operations.

[0137] The module identifies the types of resources released during fault handling and takes corresponding recycling measures based on the resource type: if it is a computing resource, it injects it into a nearby overloaded node, updates the computing resource distribution in the resource pool, and triggers real-time updates to the fault heatmap; if it is a spare part resource, it returns it to the dynamic resource pool to replenish spare parts reserves and improve resource reuse rate; if it is a network bandwidth resource, it reroutes it to the fault area, adjusts the bandwidth allocation of the network topology, and optimizes the utilization of network resources. After all resource recycling operations are completed, the module calls the interface of the fault heatmap analysis module to update the fault propagation heatmap. Based on the changes in heatmap values ​​before and after resource recycling and the resource input-output ratio, the module corrects the heatmap value calculation model to continuously improve the accuracy and adaptability of the model.

[0138] The machine learning training module collects relevant data on historical fault handling within the park, including fault types, selection of key blocking points, planning of resource feedback channels, and final handling effects, such as fault recovery time and business losses. This data is then organized, cleaned, and labeled to construct a training dataset, providing high-quality samples for model training.

[0139] With recovery time after fault handling, business interruption losses, and resource investment costs as optimization objectives, a reinforcement learning algorithm is used to train the critical blockage point location model and the resource feedback channel planning model. The objective function is:

[0140] in, For fault recovery time, For business interruption losses, Cost of resource investment + + = 1. Through continuous learning and iteration, the model parameters are optimized to improve the model's ability to plan key blocking points and resource feedback channels. The optimized model parameters are regularly updated to the blocking point identification module and the feedback channel construction module to achieve self-optimization of the algorithm. This enables the system to continuously adapt to changes in equipment status and business needs, and continuously improve the effectiveness of fault handling.

[0141] The data interaction module uses standardized data interfaces to communicate with external modules, ensuring the accuracy and compatibility of data transmission: When interacting with the fault prediction module, it receives fault prediction data encapsulated in a specific format, and data transmission follows the corresponding protocol to ensure real-time acquisition of accurate fault warning information; when interacting with the equipment topology knowledge base, it reads the equipment's 3D model data and connection relationship map through a specified interface, and performs data querying and updating to ensure the timeliness and accuracy of equipment topology information; when interacting with the resource pool status monitoring module, it acquires real-time resource status data, which is stored in a specific database format and synchronized through professional data acquisition tools to provide data support for resource scheduling and feedback channel planning. When outputting precise blocking instructions and resource recycling schemes to external systems, the instructions and schemes are encapsulated in a standard format and transmitted to the operation and maintenance execution terminal or upper-level management system through a secure protocol to ensure accurate and reliable information transmission and guide operation and maintenance personnel in actual fault handling operations.

[0142] The application of this system in chemical production parks has improved the response speed of fault handling. It can quickly locate key blocking nodes and plan resource feedback channels after a fault occurs, effectively curbing the spread of the fault. The efficiency of resource utilization has also been greatly improved, avoiding redundant investment and waste of resources. The system's self-optimization capability enables it to continuously adapt to changes in equipment and business within the park, continuously improving the accuracy and effectiveness of fault handling. At the same time, the collaborative working mode between edge nodes and the cloud platform ensures that the system can still maintain basic fault handling capabilities when network conditions are poor, providing a guarantee for the stable production of the park and reducing production interruption losses and resource consumption caused by faults.

[0143] The same or similar labels correspond to the same or similar parts;

[0144] The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this patent.

[0145] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all implementation methods here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A digital analysis, prediction and disposal system for intelligent operation and emergency response linkage, characterized in that, The system comprises: a fault thermal analysis module for collecting device topology knowledge base, resource pool state monitoring data and real-time fault prediction data, constructing a fault conduction thermal map based on a conduction thermal map generation algorithm, and representing the conduction thermal distribution of the fault in the device topology; a blocking point identification module for adopting a key blocking point positioning algorithm, taking the node with the maximum thermal value blocked by unit resource input as the standard, calculating the full path score of the device topology in real time, and positioning multiple key blocking nodes; a counter-feeding channel construction module for constructing resource counter-feeding channels based on the device topology relationship, resource pool state and fault conduction thermal map, and feeding back resource pool resources to fault disposal; a dynamic disposal execution module for generating precise blocking instructions and resource circulation schemes based on the multiple key blocking nodes output by the key blocking point identification module and the resource counter-feeding channels constructed by the counter-feeding channel construction module, executing fault disposal, and feeding back performance feedback data to the fault thermal analysis module; a data interaction module for data interaction with the fault prediction module, device topology knowledge base and resource pool state monitoring module, obtaining real-time data, knowledge base information and state monitoring data, and outputting precise blocking instructions and resource circulation schemes to the outside.

2. The system of claim 1, wherein, When the fault thermal analysis module generates the fault conduction thermal map, it includes: extracting the device connection relationship and physical location information in the device topology knowledge base as the basis for fault conduction path; collecting resource pool state monitoring data, including resource load rate, resource type distribution and resource available quantity, and marking the influence weight of resources on fault conduction; receiving real-time fault data output by the fault prediction module, including fault type, fault initial node and fault development trend, combining device topology and resource state, and calculating the thermal value of each node in the device topology through a heat conduction simulation algorithm to generate a fault conduction thermal map, with the formula being: ； wherein, is a failure thermal value for node i, is a failure severity factor, is a resource impact factor, is a conduction coefficient, + + = 1.

3. The system of claim 1, wherein, In the key blocking point positioning algorithm of the blocking point identification module, the selection standard of the node with the maximum thermal value blocked by unit resource input is quantified by the following method: Defining node outage performance values wherein is the decrease in total heat value in the failure propagation heat map after blocking node j, is the resource cost required to block node j; Traverse all nodes in the device topology and calculate the value of each node. ,according to Sort the nodes from largest to smallest and select multiple key blocking nodes; The path score is calculated when the full path score is calculated where P is a conduction path in the device topology, and the optimal path and associated nodes for failure conduction suppression effect are screened out by traversing all possible paths.

4. The system of claim 1, wherein, When the counter-feeding channel construction module constructs resource counter-feeding channels, it includes: analyzing the resource demand type and demand priority of the fault impact area in the fault conduction thermal map; querying resource pool state monitoring data to match the location, type and quantity of available resources; based on the device topology relationship, using the shortest path algorithm and resource scheduling algorithm to plan the resource delivery path from the resource pool to the fault impact area, and constructing the resource counter-feeding channel, which satisfies the formula: ； wherein is the resource transmission delay for path node k, is the resource transmission loss coefficient for path node k, the optimal resource feedback channel is determined by minimizing the weighted sum of transmission delay and loss.

5. The system of claim 1, wherein, When the dynamic disposal execution module generates precise blocking instructions, it includes for each key blocking node: analyzing the device type, running state and carried service of the key blocking node; combining the fault type to develop precise blocking instructions including resource injection strategy, service migration strategy and hardware restart / reset strategy; The instruction generation follows a principle of minimum business impact, through a business impact degree evaluation model Ensure that the business impact degree I after the instruction execution is lower than a preset threshold, wherein The business importance degree, The business interruption duration, The business recovery cost.

6. The system of claim 1, wherein, When the dynamic disposal execution module generates resource circulation schemes, it includes: identifying the type of released disposal resources; if it is a computing resource, performing an injection adjacent overload node operation, updating the resource pool computing resource distribution, and triggering real-time thermal map update; if it is a spare part resource, performing a return dynamic resource pool operation to supplement the resource pool spare part reserve; If it is network bandwidth, perform re-routing to the failure area operation to adjust the network topology bandwidth allocation; After all resource circulation operations are completed, a fault thermal analysis module interface is called to perform real-time thermal map updating, and a thermal value calculation model of the fault conduction thermal map is corrected based on thermal value changes before and after resource circulation and resource input-output ratios.

7. The system of claim 1, wherein, When the data interaction module interacts with external modules, a standardized data interface is adopted, specifically including: Interacting with the fault prediction module, receiving fault prediction data encapsulated in JSON format, including fault probability, fault type, and predicted occurrence timestamp, and data transmission complying with the MQTT protocol; Interacting with the device topology knowledge base to read device three-dimensional model data and connection relationship graphs, and using RESTful API for data query and update; Interacting with the resource pool state monitoring module to obtain real-time state data of resources, and storing the data in a time series database format and synchronizing the data through a Telegraf collection agent; When outputting precise blocking instructions and resource circulation schemes to the outside, the instructions and schemes are encapsulated in XML format and transmitted to operation and maintenance execution terminals or upper-level management systems through the HTTPS protocol.

8. The system of claim 1, wherein, In the key blocking point positioning algorithm of the blocking point identification module, the full-path score calculation is specifically: For each conducted path P in the device topology, define a path score wherein is a failure heat value for path node n, is a resource cost for blocking node n; Traverse all possible conduction paths, calculate the score of each path , and select the path with the highest score and the associated nodes as the key blocking point candidate; The blocking synergy check is performed on the candidate nodes, if multiple candidate nodes belong to one short path, the joint blocking performance is evaluated to avoid repeated investment of resources, and the joint blocking performance wherein is the resource investment overlap degree between nodes, is an overlap penalty coefficient.

9. The system of claim 1, wherein, The system further includes a machine learning training module for: Collecting historical fault handling data, including fault type, key blocking point selection, resource counter-feeding channel planning, and handling efficiency results; Building a training data set, taking fault recovery time after fault handling, business interruption loss, and resource input cost as optimization objectives, and using a reinforcement learning algorithm to train a key blocking point positioning model and a resource counter-feeding channel planning model; Periodically updating the model parameters optimized through training to the blocking point identification module and the counter-feeding channel construction module to realize algorithm self-optimization, and the model training objective function is: ； wherein is the failure recovery time, is the business interruption loss, is the resource investment cost, + + = 1.

10. The system of claim 1, wherein, When the system is deployed, a hybrid architecture of edge nodes and cloud platforms is adopted: The edge node is deployed locally in the device room, integrates a lightweight version of the fault thermal analysis module and a basic instruction execution unit of the dynamic handling execution module, and is used for real-time collection of device data and execution of rapid blocking instructions; The cloud platform deploys complete function modules of the system to perform large data fault conduction thermal map calculation, key blocking point global path score calculation, and resource counter-feeding channel cross-region planning; The edge node and the cloud platform interact through 5G / industrial Ethernet, adopt an edge computing offloading strategy, offload complex calculation tasks to the cloud when network bandwidth is sufficient, and enable the local cache model to independently execute basic fault handling when the network is interrupted, thereby ensuring system resilience, and data interaction complies with the OPC UA protocol and supports data encryption and breakpoint continuation.

Images