Nuclear emergency decontamination station partition management method and device based on artificial intelligence

Abstract

The invention relates to the technical field of nuclear emergency decontamination management, and discloses a nuclear emergency decontamination station partition management method and device based on artificial intelligence, and the method comprises the steps: carrying out the three-dimensional laser scanning and data collection of a pollution region, a decontamination region and a clean region of a nuclear emergency decontamination station, and constructing a decontamination station digital twinborn model; according to the digital twin model of the decontamination station, performing multi-dimensional analysis and weight calculation on the polluted area, the decontamination area and the clean area by adopting an analytic hierarchy process to obtain a partition evaluation parameter matrix; inputting the partition evaluation parameter matrix into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters; and performing decision analysis and threshold judgment on the radiation dose rate data, the pollution range data, the personnel quantity data and the equipment state data according to the partition management control parameters, and generating an emergency disposal control instruction, thereby realizing cooperative linkage of the pollution area, the decontamination area and the clean area, and improving the accuracy and timeliness of emergency disposal.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear emergency decontamination management, and in particular to a nuclear emergency decontamination station zoning management method and device based on artificial intelligence. Background Art

[0002] Nuclear emergency decontamination is a key link in ensuring emergency response to nuclear accidents, and plays an important role in reducing casualties and preventing the spread of pollution. Traditional nuclear emergency decontamination stations are mainly divided into two types: fixed and temporary. Although fixed decontamination stations have advantages such as perfect configuration and good effect, they have problems such as high construction and maintenance costs, long-term occupation of public resources, and low efficiency. Although temporary decontamination stations can solve the problem of resource occupation, they increase the emergency response time due to the need for temporary dispatch, and have defects such as insufficient thermal protection function, low decontamination efficiency, and weak pollution disposal capacity.

[0003] With the rapid development of artificial intelligence technology, the need to introduce intelligent management into nuclear emergency decontamination stations is becoming increasingly urgent. However, there are still many problems in the zoning management of current nuclear emergency decontamination stations: the lack of real-time dynamic monitoring and evaluation mechanisms makes it impossible to accurately grasp the pollution status and equipment operation status of each zone; personnel diversion and path planning rely on manual experience, making it difficult to make the best decision in an emergency; the adjustment of decontamination process parameters and waste treatment parameters lacks scientific basis, affecting the disposal effect. In addition, the multi-regional collaborative management of nuclear emergency decontamination stations also faces challenges: the information exchange between contaminated areas, decontamination areas and clean areas is not smooth, making it difficult to achieve rapid response; the equipment linkage of each functional area is poor, and it is impossible to form an efficient emergency disposal process; the lack of an intelligent decision support system makes it difficult to cope with complex and changeable emergency situations. Summary of the invention

[0004] The present invention provides a nuclear emergency decontamination station zoning management method and device based on artificial intelligence. The present invention realizes the coordinated linkage of contaminated areas, decontamination areas and clean areas, and improves the accuracy and timeliness of emergency disposal.

[0005] In a first aspect, the present invention provides a nuclear emergency decontamination station zoning management method based on artificial intelligence, the nuclear emergency decontamination station zoning management method based on artificial intelligence comprising:

[0006] Conduct 3D laser scanning and data collection on the contaminated area, decontamination area, and clean area of ​​the nuclear emergency decontamination station to build a digital twin model of the decontamination station;

[0007] According to the digital twin model of the decontamination station, a hierarchical analysis method is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area to obtain a partition evaluation parameter matrix;

[0008] Inputting the partition evaluation parameter matrix into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters;

[0009] According to the zoning management control parameters, decision analysis and threshold judgment are performed on the radiation dose rate data, contamination range data, personnel quantity data and equipment status data to generate emergency disposal control instructions.

[0010] In a second aspect, the present invention provides a nuclear emergency decontamination station partition management device based on artificial intelligence, the nuclear emergency decontamination station partition management device based on artificial intelligence comprising:

[0011] The acquisition module is used to perform 3D laser scanning and data acquisition on the contaminated area, decontamination area, and clean area of ​​the nuclear emergency decontamination station, and to build a digital twin model of the decontamination station;

[0012] An analysis module is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area according to the digital twin model of the decontamination station by using a hierarchical analysis method to obtain a partition evaluation parameter matrix;

[0013] A calculation module, used for inputting the partition evaluation parameter matrix into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters;

[0014] The generation module is used to perform decision analysis and threshold judgment on the radiation dose rate data, contamination range data, personnel quantity data and equipment status data according to the partition management control parameters, and generate emergency disposal control instructions.

[0015] The third aspect of the present invention provides a computer device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the computer device executes the above-mentioned artificial intelligence-based nuclear emergency decontamination station zoning management method.

[0016] A fourth aspect of the present invention provides a computer-readable storage medium, in which instructions are stored, which, when executed on a computer, enable the computer to execute the above-mentioned artificial intelligence-based nuclear emergency decontamination station zoning management method.

[0017] In the technical solution provided by the present invention, the all-round perception and real-time monitoring of the decontamination station are realized by establishing a digital twin model. The fusion of digital twin technology and multi-source sensor data provides an accurate state evaluation basis, making the operation status of each partition clear and controllable. The hierarchical analysis method is used for multi-dimensional analysis and weight calculation, and a scientific partition evaluation system is constructed to achieve a comprehensive evaluation of the pollution degree, personnel capacity, equipment status and environmental parameters, providing a quantitative basis for management decisions. The deep reinforcement learning network is introduced for dynamic optimization calculation, and the partition management strategy is continuously optimized through the Q-learning algorithm, so that the system can learn autonomously and make optimal control decisions. A multi-level emergency response mechanism based on a decision tree is designed, and the intelligent hierarchical response of emergency disposal is realized through threshold judgment and rule matching, which improves the accuracy and timeliness of emergency disposal. An intelligent personnel classification and path planning system is constructed, and the accurate classification of personnel pollution degree is realized through deep learning algorithms, and the optimal selection of decontamination paths is ensured through multi-objective optimization. A parameter optimization method for decontamination process and waste treatment is developed, and the balance between decontamination efficiency and waste treatment is achieved through multi-dimensional optimization operations, ensuring the efficiency and environmental protection of the disposal process. The coordinated linkage of contaminated areas, disinfection areas and cleaning areas is achieved, and information sharing and equipment linkage are achieved between functional areas, forming a complete intelligent management closed loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 It is a schematic diagram of the steps of the nuclear emergency decontamination station zoning management method based on artificial intelligence in an embodiment of the present invention;

[0020] Figure 2 It is a structural schematic diagram of a nuclear emergency decontamination station zoning management device based on artificial intelligence in an embodiment of the present invention;

[0021] Figure 3 It is a schematic block diagram of the structure of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The embodiment of the present invention provides a method and device for zoning management of nuclear emergency decontamination stations based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of a nuclear emergency decontamination station zoning management method based on artificial intelligence in an embodiment of the present invention includes:

[0024] Step S1, performing three-dimensional laser scanning and data collection on the contaminated area, the decontamination area and the clean area of ​​the nuclear emergency decontamination station to construct a digital twin model of the decontamination station;

[0025] It is understandable that the execution subject of the present invention may be a nuclear emergency decontamination station partition management device based on artificial intelligence, or may be a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0026] Specifically, three-dimensional laser scanning is performed on the contaminated area, the decontamination area, and the clean area respectively, and the physical structure information of the decontamination station is captured by using high-precision three-dimensional laser scanning equipment. The scanned data is saved in the form of three-dimensional point clouds, which reflect the geometric characteristics and spatial layout of different partitions of the decontamination station. The three-dimensional point cloud data is gridded, and the scattered point cloud data is converted into a regular grid structure to extract the features related to each functional area of ​​the decontamination station to form a complete three-dimensional structural model. After completing the three-dimensional structural modeling, the sensor monitoring data of different partitions are integrated to supplement the dynamic properties and environmental status information of the three-dimensional model. In the contaminated area, by deploying radiation dose rate monitors, personnel radioactive surface contamination monitors, temperature and humidity sensors and other equipment, the radiation level, pollutant concentration and environmental parameter data in the area are collected in real time. Since these data may come from various sources and have different dimensions, direct use will lead to calculation errors, so data standardization is performed on them. Through standardization, data of different units and ranges are unified to a standard scale to generate a sensor data matrix reflecting the comprehensive status of the contaminated area. At the same time, in the decontamination area, intelligent water pressure sensors, water quality monitors, aerosol radioactivity concentration detectors, waste liquid pool level sensors, and waste liquid radioactivity concentration continuous monitors work together to collect monitoring data on water flow pressure, water quality, aerosol radioactivity, waste liquid pool level, and waste liquid radioactivity concentration. Similarly, these data are standardized to eliminate the impact of different physical meanings and magnitudes of various data, and the sensor data matrix of the decontamination area is obtained. For the clean area, radiation background value monitoring devices and air quality monitors are arranged. These devices are used to record the radiation background value and air quality parameters in the clean area. After the data are standardized accordingly, the sensor data matrix of the clean area is formed. Time series data analysis is performed on the sensor data matrix of the contaminated area, the sensor data matrix of the decontamination area, and the sensor data matrix of the clean area to identify the state change trend and dynamic behavior characteristics of each partition in different time periods, obtain the dynamic monitoring data of the decontamination station, and describe and predict the operating status of each partition of the decontamination station in real time. The three-dimensional structural model of the decontamination station is fused and feature mapped with dynamic monitoring data to form a digital twin mapping relationship that can synchronously reflect the interaction between the physical entity and the virtual model. The data fusion process integrates multi-source information to combine the static three-dimensional structure with the dynamic monitoring data, thereby reconstructing the real decontamination station operation scene in a virtual environment. Feature mapping establishes an association between spatial position and state variables, so that the virtual model can intuitively present the operating status of each area in the decontamination station and its changes over time. Based on the digital twin mapping relationship, a data synchronization mechanism that reflects the interaction between physical entities and virtual models is constructed. Through real-time data collection and updating, it is ensured that the digital twin model can reflect the actual operating status of the physical decontamination station at any time, and at the same time, the state feedback of the virtual model is realized to guide the adjustment and optimization of the physical system.Through two-way interaction, a digital twin model is established that can comprehensively describe the dynamic operating status of the disinfection station.

[0027] Step S2: Based on the digital twin model of the decontamination station, the analytic hierarchy process is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area to obtain a partition evaluation parameter matrix;

[0028] Specifically, the radiation dose rate data, surface contamination data, radionuclide activity concentration data in the air, and radionuclide activity concentration data in the digital twin model of the decontamination station are normalized to obtain a pollution degree assessment matrix. At the same time, the data such as the utilization rate of the partition area, the crowding of personnel, and the efficiency of personnel passage are normalized to generate a personnel capacity assessment matrix to reflect the capacity of each partition in terms of space utilization and personnel flow. The data such as the integrity rate of the decontamination equipment, the operating status of the waste liquid treatment system, and the efficiency of the ventilation system in the digital twin model of the decontamination station are normalized to form an equipment status assessment matrix to reflect the stability and performance of the equipment operation in each functional area. At the same time, the data such as temperature, humidity, air pressure difference, and ventilation frequency are normalized to generate an environmental parameter assessment matrix, which is used to describe the impact and adaptability of environmental conditions on the partition. Based on the pollution degree assessment matrix, the personnel capacity assessment matrix, the equipment status assessment matrix, and the environmental parameter assessment matrix, a hierarchical analysis method judgment matrix is ​​constructed to determine the relative importance of each evaluation indicator. Through comprehensive analysis of the pollution degree assessment matrix, personnel capacity assessment matrix, equipment status assessment matrix and environmental parameter assessment matrix, a judgment matrix describing the pairwise comparison relationship of indicators is established, and its eigenvalues ​​and eigenvectors are calculated to extract the weight coefficients of the indicators. In this process, the judgment matrix is ​​subjected to consistency check to ensure the rationality and reliability of the judgment results. If the consistency check is passed, the obtained weight coefficient can be used for further calculations, otherwise the content of the judgment matrix needs to be adjusted until the consistency requirements are met. The weight coefficient is weighted and fused with the pollution degree assessment matrix, personnel capacity assessment matrix, equipment status assessment matrix and environmental parameter assessment matrix to comprehensively evaluate the operating status of each partition. Each indicator matrix is ​​linearly combined according to the weight to generate a comprehensive evaluation index. The comprehensive evaluation index is processed by the fuzzy membership function to divide the continuous values ​​into different evaluation levels. The design of the fuzzy membership function is based on the empirical data and expert knowledge of the partition operation. For example, the evaluation results are divided into four levels: excellent, good, medium and poor, to intuitively reflect the operating status of the partition. The partition evaluation parameter matrix is ​​obtained by calculating the fuzzy membership and dividing the level of the comprehensive evaluation index.

[0029] Step S3: Input the partition evaluation parameter matrix into the deep reinforcement learning network for dynamic optimization calculation to obtain the partition management control parameters;

[0030] Specifically, the partition evaluation parameter matrix is ​​feature parsed, and the partition status evaluation level, personnel distribution data, equipment operation status data, and environmental parameter data contained therein are converted into state space vectors. The complex partition status information is expressed in a vectorized manner to provide a standardized format for the input of the deep reinforcement learning network. At the same time, in order to construct an action space that can reflect the partition management action, the ventilation system adjustment parameters, personnel diversion parameters, and equipment start and stop parameters are encoded to form an action space vector. The encoding of these parameters reflects the adjustment space and range of the management strategy under different states. Based on the state space vector and the action space vector, a Q-value function is established to describe the expected cumulative reward that can be obtained by taking a specific action in a specific state. The neural network parameters of the Q-value function are initialized to construct a deep reinforcement learning network. By combining the advantages of deep neural networks and reinforcement learning, the network can model high-dimensional and complex state-action mapping relationships. The state space vector is input into the deep reinforcement learning network, and the action value function, i.e., the Q value, is obtained through forward calculation. This action value function assigns a score to each possible action, reflecting the expected effect of taking the action in the current state, and selects the optimal action in the current state based on these Q values. After selecting the optimal action, execute the action and evaluate its execution effect. The reward function is used to quantify the execution results of the action. The design of the reward function is based on the partition management objectives, such as reducing pollution risks, improving personnel traffic efficiency, optimizing equipment operation status, etc. These reward values ​​are stored in the experience replay pool for subsequent network training and optimization. After accumulating enough training samples, training batches are randomly selected from the experience replay pool, and the target Q value is calculated using the temporal difference algorithm. The loss function is obtained by optimizing the difference between the current Q value and the target Q value. The loss function is calculated by the back propagation algorithm, and the parameters of the deep reinforcement learning network are updated in combination with the gradient descent method, so that the Q value function can more accurately reflect the mapping relationship between state, action and reward. In each training iteration, the deep reinforcement learning network is continuously optimized to improve its adaptability to the dynamic management problem of partitions. When the network training is completed, the optimized Q value function is applied to the current state space vector to output the optimal configuration combination of ventilation system adjustment parameters, personnel diversion parameters and equipment start and stop parameters. These optimized parameters can not only reflect the current state of the partition, but also make adjustments in advance for possible dynamic changes in the future, thereby generating control parameters for partition management.

[0031] Step S4: According to the zoning management control parameters, decision analysis and threshold judgment are performed on the radiation dose rate data, contamination range data, number of personnel data and equipment status data to generate emergency disposal control instructions.

[0032] Specifically, the partition management control parameters, radiation dose rate data, contamination range data, personnel number data and equipment status data are input into the decision tree analysis model to construct the decision tree feature nodes. These feature nodes are the basis for model analysis and decision-making, and contain multi-dimensional information closely related to the operation status of the decontamination station, such as real-time radiation dose rate, contamination range distribution, number of personnel in the area and detailed data on the operation status of the equipment. After constructing the feature nodes, the impurity of the data is measured by calculating the Gini coefficient of each node, and the decision tree nodes are split based on the Gini coefficient as the standard. Through the splitting process, a decision branch structure that can cover different situations is formed, thereby gradually decomposing complex decision problems into multiple hierarchical clear paths. On the basis of the decision branch structure, the radiation dose rate data involved is subjected to threshold hierarchical analysis. By dividing the radiation dose rate into multiple hazard level intervals, a dose rate threshold matrix is ​​generated, and the safety levels corresponding to different dose rate intervals are clarified, such as the four levels of safety, warning, danger and extreme danger, which is convenient for the rapid determination of the subsequent emergency response mechanism. At the same time, for the contamination range data in the decision branch structure, the spatial clustering analysis method is combined to identify the distribution characteristics of pollutants in different regions and generate a contamination area distribution matrix. Based on the results of the pollution range distribution, the density of the number of personnel data in the decision branch structure is calculated to measure the degree of crowding in each area and generate a regional crowding matrix. Fault diagnosis and analysis are performed on the equipment status data. By combining historical operation data and real-time monitoring data, the operating status of the equipment is evaluated, and the diagnosis results are constructed as the equipment operation state vector. This vector includes information such as the integrity rate of the decontamination equipment, the load status of the waste liquid treatment system, and the efficiency of the ventilation system. The dose rate threshold matrix, the pollution area distribution matrix, the regional crowding matrix and the equipment operation state vector are input into the multi-layer decision rule library for rule matching and response level determination. The decision rule library combines the preset emergency response rules and historical disposal experience, and quickly determines the current emergency response level by matching the input matrix and vector. Based on the determined emergency response level, multiple key subsystems of the decontamination station are linked and controlled and calculated, including ventilation system parameters, protective door parameters, waste liquid treatment system parameters, and emergency lighting system parameters. For example, in a high-risk situation, the exhaust speed of the ventilation system needs to be increased, the opening and closing logic of the protective door needs to be dynamically adjusted to limit the spread of pollution, the waste liquid treatment system needs to increase the processing load, and the emergency lighting system needs to start the whole area lighting to ensure the safe evacuation of personnel. The calculation and adjustment of these system parameters are completed through the linkage control logic of the model, and finally the emergency disposal control instructions are generated.

[0033] The partition status data, equipment operation data and environmental parameter data contained in the emergency disposal control instructions are feature extracted to generate a multi-dimensional data structure describing the overall operation status of the decontamination station, namely, the decontamination station operation status matrix. According to the decontamination station operation status matrix, the surface contamination level data, radiation dose rate data and physical condition data of the personnel are standardized to eliminate the dimension differences and range inconsistencies between different data sources, and obtain the comprehensive feature matrix of the personnel. The comprehensive feature matrix of the personnel is input into the deep learning classification network. In the network, the convolution layer is used to extract local features of the data, and the pooling layer is used to reduce the dimension of the features to reduce the influence of redundant information and generate the feature vector of the personnel. Through multi-dimensional clustering analysis of the personnel feature vector, the personnel are divided into a heavy pollution group, a moderate pollution group and a light pollution group. The grouping process automatically classifies according to the feature vector of each person to ensure the accuracy and scientificity of the pollution level classification, and generates personnel classification data at the same time. Based on these classification data, a decontamination channel load balancing model is constructed to dynamically evaluate the processing capacity of each decontamination channel. The evaluation of the decontamination channel includes its current load, expected completion time, equipment utilization efficiency, etc., forming a channel capacity state matrix, which reflects the availability and workload of each channel at different time points. A multi-objective path optimization model is established based on the channel capacity state matrix to determine the optimal path planning for personnel decontamination. The optimization model takes decontamination time, channel congestion and radiation protection requirements as constraints, and comprehensively considers the safety and efficiency of the path. After the optimization goal is clear, the model generates path optimization parameters to provide accurate input for path calculation. Based on these parameters, the optimization requirements are input into the intelligent path-finding system, and the decontamination path is calculated using the improved A* algorithm. The A* algorithm can quickly generate multiple candidate paths by improving the heuristic function, and give priority to the goals of minimizing radiation exposure risk and maximizing decontamination efficiency. After the candidate paths are generated, the path set is comprehensively scored to evaluate the decontamination efficiency, congestion and safety of each path. The scoring results are used to select a path with the best comprehensive index as the final optimal decontamination path data. After completing the optimal path selection, multi-dimensional optimization operations are performed based on the personnel classification data and the optimal disinfection path data to determine the disinfection process parameters and waste treatment parameters. The disinfection process parameters include disinfection time, water pressure, disinfectant dosage, etc. These parameters are dynamically adjusted according to the needs of personnel in different pollution groups to ensure the maximum disinfection effect while avoiding waste of resources. The waste treatment parameters are aimed at the waste liquid and solid waste generated during the disinfection process, and their treatment methods are optimized to meet environmental protection requirements.

[0034] Multi-sample analysis was performed on the surface contamination level and radiation dose rate data in the personnel classification data to evaluate the characteristics of personnel with different contamination levels. By statistically analyzing and summarizing the sample characteristics, the data was converted into a decontamination difficulty level matrix, and each cell of the matrix corresponded to a combination of different contamination levels and dose levels in a specific contamination group. Based on the decontamination difficulty level matrix, multi-dimensional cross-calculations were performed on the decontamination time data, water pressure strength data, and decontamination agent dosage data to generate an initial decontamination parameter matrix. This matrix combines the personnel contamination level and the decontamination equipment capacity, and can effectively represent the initial decontamination strategy under different contamination conditions. However, since the decontamination process is also affected by spatial layout and time distribution, it is necessary to combine the optimal decontamination path data to optimize the spatial and temporal distribution of the initial matrix, balance the load of each decontamination channel, minimize congestion and waste of resources, and ensure the improvement of decontamination efficiency, so as to obtain the optimized decontamination process parameter combination sequence. The decontamination process parameter combination sequence is input into the multi-objective optimization model, and the decontamination parameters are optimized through the Pareto optimal calculation of the decontamination efficiency data and the waste generation data. Pareto optimal calculation finds a set of solutions in the multi-objective trade-off, so that other objectives are optimized without sacrificing a certain objective, and the disinfection parameter optimization matrix is ​​generated. According to the disinfection parameter optimization matrix, the concentration multiple data, treatment cycle data and emission data of the waste liquid treatment system are dynamically balanced, and the waste liquid treatment parameter set is generated by balancing the system's operating capacity and the waste liquid treatment demand. The system capacity constraint analysis and multi-objective planning calculation are performed on the waste liquid treatment parameter set, and the waste liquid treatment optimization parameters are generated by optimizing the waste liquid concentration efficiency and reducing the emission. At the same time, according to the waste liquid treatment optimization parameters, a treatment model for solid waste is constructed, and the data such as solid waste compression ratio, storage cycle and disposal volume are optimized. The goal of the solid waste treatment model is to reduce the occupied space by increasing the compression ratio, extend the storage cycle to reduce the transportation and disposal frequency, and optimize the disposal volume to ensure the safe and compliant treatment of solid waste. After the optimization operation is completed, the solid waste treatment parameters are generated, which can accurately guide each link of solid waste treatment and ensure the efficiency and environmental protection of waste treatment. The disinfection parameter optimization matrix, waste liquid treatment optimization parameters and solid waste treatment parameters are multi-dimensionally fused and calculated to comprehensively consider the dynamic coupling relationship between the disinfection process and waste treatment. Through fusion calculation, the final disinfection process parameters and waste treatment parameters are generated. These parameters can adapt to the disinfection needs in different situations in the application, ensuring the safety, efficiency and environmental friendliness of the disinfection process.

[0035] In the embodiment of the present invention, the digital twin model is established to realize the all-round perception and real-time monitoring of the decontamination station. The fusion of digital twin technology and multi-source sensor data provides an accurate state evaluation basis, making the operation status of each partition clear and controllable. The hierarchical analysis method is used for multi-dimensional analysis and weight calculation, and a scientific partition evaluation system is constructed to realize the comprehensive evaluation of pollution degree, personnel capacity, equipment status and environmental parameters, providing a quantitative basis for management decision-making. The deep reinforcement learning network is introduced for dynamic optimization calculation, and the partition management strategy is continuously optimized through the Q-learning algorithm, so that the system can learn autonomously and make optimal control decisions. A multi-level emergency response mechanism based on decision tree is designed, and the intelligent hierarchical response of emergency disposal is realized through threshold judgment and rule matching, which improves the accuracy and timeliness of emergency disposal. An intelligent personnel classification and path planning system is constructed, and the accurate classification of personnel pollution degree is realized through deep learning algorithm, and the optimal selection of decontamination path is ensured through multi-objective optimization. The parameter optimization method of decontamination process and waste treatment is developed, and the balance between decontamination efficiency and waste treatment is achieved through multi-dimensional optimization operation, ensuring the efficiency and environmental protection of the disposal process. The coordinated linkage of contaminated areas, disinfection areas and cleaning areas is achieved, and information sharing and equipment linkage are achieved between functional areas, forming a complete intelligent management closed loop.

[0036] In a specific embodiment, the process of executing step S1 may specifically include the following steps:

[0037] Conduct 3D laser scanning on the contaminated area, decontamination area and clean area to obtain 3D point cloud data of the decontamination station, and perform grid processing and feature extraction on the 3D point cloud data to obtain a 3D structural model of the decontamination station;

[0038] The contaminated area monitoring data collected by the radiation dose rate monitor, the personnel radioactive surface contamination monitor, and the temperature and humidity sensor are processed by data standardization to obtain the contaminated area sensor data matrix;

[0039] The monitoring data of the decontamination area collected by the intelligent water pressure sensor, water quality monitor, aerosol radioactivity concentration detector, waste liquid pool level sensor, and waste liquid radioactivity concentration continuous monitor are processed by data standardization to obtain the decontamination area sensor data matrix;

[0040] The clean area monitoring data collected by the radiation background value monitoring device and the air quality monitor are processed by data standardization to obtain the clean area sensor data matrix;

[0041] Perform time series data analysis on the sensor data matrix of the contaminated area, the sensor data matrix of the decontamination area, and the sensor data matrix of the clean area to obtain dynamic monitoring data of the decontamination station;

[0042] The three-dimensional structural model of the decontamination station and the dynamic monitoring data of the decontamination station are subjected to data fusion and feature mapping to obtain the digital twin mapping relationship of the decontamination station. Based on the digital twin mapping relationship of the decontamination station, a data synchronization mechanism that reflects the interaction between physical entities and virtual models is constructed to obtain the digital twin model of the decontamination station.

[0043] Specifically, three-dimensional laser scanning technology is used to collect comprehensive spatial data of the entire decontamination station. The three-dimensional point cloud data of the decontamination station is generated by high-precision laser scanning equipment. The point cloud data consists of a large number of discrete points, and each point has a spatial coordinate (x, y, z), where x, y, and z represent the lateral, longitudinal, and height positions of the point in space, respectively. At the same time, each point is accompanied by an attribute I that reflects the reflection intensity of the object surface, which is used to identify different materials or surface characteristics. The point cloud data is meshed to convert discrete points into a regular mesh model. The meshing process is completed through a triangulation algorithm, such as the Delaunay triangulation algorithm, which connects the points of the point cloud data into a series of adjacent triangular facets to form a three-dimensional mesh model. After meshing, the three-dimensional model is subjected to feature extraction, such as extracting the surface normal vector n = (n x ,n y ,n z ) to indicate the direction of the surface; or extract the curvature C, which is defined as the degree of curvature of the surface in a local area. Through these geometric features, a three-dimensional structural model describing the spatial structure of the decontamination station is established. After the three-dimensional structure modeling is completed, dynamic attributes are added to the model in combination with sensor monitoring data. In the contaminated area, the radiation dose rate monitor records the radiation level in the area, and the data is expressed as a dose rate R d Indicates; the personnel radioactive surface contamination monitor collects the surface contamination value C s ; The temperature and humidity sensor provides the ambient temperature T and humidity H. These data are converted into dimensionless data through the standardized processing formula, the formula is:

[0044]

[0045] Among them, X represents the original data, X min and X max are the minimum and maximum values ​​of the data, respectively, norm is the standardized data. Through standardization, the influence of data scale is eliminated and the sensor data matrix M of the polluted area is generated. p , where each column represents a monitoring indicator and each row represents multidimensional data at a certain time point. In the decontamination area, the intelligent water pressure sensor records the water pressure P w , the water quality monitor provides the pollutant concentration C w , aerosol radioactivity concentration detector measures aerosol activity A a , the waste liquid pool level sensor records the liquid level height H l, the waste liquid radioactivity continuous monitor records the waste liquid radioactivity A w These data are also processed by standardized formulas to generate the sensor data matrix M of the decontamination area. d In the clean area, the radiation background value monitoring device provides the background radiation value R b , the air quality monitor records the PM2.5 concentration in the air C pm The clean area sensor data matrix M is generated by similar processing c . p 、M d and M c Perform time series data analysis and calculate its time series characteristics, such as the sliding mean μ t and standard deviation σ t :

[0046]

[0047] Where X i is the data value at time point i, and N is the sliding window size. These time series features are used to identify data trends and anomalies to form a dynamic monitoring data matrix M of the decontamination station. s When fusing the 3D structure model with the dynamic monitoring data, the mapping relationship between the point cloud coordinates and the sensor data is established through the feature mapping method. For example, for a point P (x, y, z) in the point cloud, its feature vector is defined as:

[0048] F(P)=[x,y,z,R d ,C s ,T,H];

[0049] The feature vector combines spatial information with dynamic monitoring data to form a digital twin mapping relationship of the decontamination station. Based on the mapping relationship, the physical status of each area of ​​the decontamination station is reflected in real time. A data synchronization mechanism is established to ensure the consistency of interaction between the virtual model and the physical entity. This mechanism uses the difference detection formula:

[0050] ΔF=F real -F virtual ;

[0051] A data synchronization mechanism is established to ensure the consistency of interaction between the virtual model and the physical entity. This mechanism uses the difference detection formula:

[0052] ΔF=F real -F virtual ;

[0053] where F real and F virtualare the feature vectors of the physical entity and the virtual model, respectively. When ΔF exceeds the preset threshold, the model update is triggered to ensure that the digital twin model can reflect the physical operation status in real time.

[0054] In a specific embodiment, the process of executing step S2 may specifically include the following steps:

[0055] The radiation dose rate data, surface contamination data, radionuclide activity concentration data in the air, and radionuclide activity concentration data in the digital twin model of the decontamination station are normalized to obtain a contamination degree assessment matrix;

[0056] The partition area utilization data, personnel congestion data, and personnel traffic efficiency data in the digital twin model of the decontamination station are normalized to obtain the personnel capacity assessment matrix;

[0057] The decontamination equipment integrity data, waste liquid treatment system operation status data, and ventilation system efficiency data in the decontamination station digital twin model are normalized to obtain the equipment status evaluation matrix.

[0058] Normalize the temperature data, humidity data, air pressure difference data, and ventilation frequency data in the digital twin model of the decontamination station to obtain the environmental parameter evaluation matrix;

[0059] Based on the pollution degree assessment matrix, personnel capacity assessment matrix, equipment status assessment matrix and environmental parameter assessment matrix, a hierarchical analysis method judgment matrix is ​​constructed, and the eigenvalue calculation and consistency test of the judgment matrix are performed to obtain the weight coefficient of each evaluation index;

[0060] The weight coefficient is weighted and fused with the pollution degree assessment matrix, personnel capacity assessment matrix, equipment status assessment matrix and environmental parameter assessment matrix to obtain a comprehensive assessment index. The comprehensive assessment index is then fuzzy-membership calculated and graded to obtain a partition assessment parameter matrix.

[0061] Specifically, the multidimensional data contained in the model is normalized to unify the data scale and avoid calculation errors caused by dimensional differences. d , surface contamination data C s 、Data on radionuclide activity concentration in air A a And the activity concentration data of radionuclides in water A w , apply the normalization formula:

[0062]

[0063] Among them, X represents the original data, X min and X max are the minimum and maximum values ​​of the data, respectively,norm is the dimensionless data after normalization. By normalizing these data, we can obtain the pollution degree assessment matrix M reflecting the pollution degree in different regions. p , where each column represents a certain pollution index, and each row represents the multi-dimensional pollution data at a certain moment. At the same time, in order to evaluate the personnel capacity index, the partition area utilization rate data U a , Crowding data D p And the personnel traffic efficiency data E p Perform normalization and apply the same normalization formula to generate the personnel capacity assessment matrix M c The matrix represents the efficiency of each area of ​​the decontamination station in terms of space usage and personnel flow. For equipment status data, including the integrity rate of decontamination equipment W e , Wastewater treatment system operating status S w and ventilation system efficiency E v , the normalization method is also used to generate the equipment status evaluation matrix M e The matrix provides a quantitative description of the equipment operation. In order to evaluate the environmental parameters, the temperature data T, humidity data H, air pressure difference data ΔP and ventilation frequency data N are calculated. v Perform normalization to form the environmental parameter evaluation matrix M e , reflecting the impact of environmental conditions in each partition on operation. After completing the above matrix construction, the pollution degree evaluation matrix M p , Personnel capacity assessment matrix M c , Equipment status evaluation matrix M e and environmental parameter evaluation matrix M e , the judgment matrix A is constructed by applying the analytic hierarchy process (AHP). The element a in the judgment matrix ij Indicates the relative importance of indicator i and indicator j, which is determined by the expert scoring method:

[0064]

[0065] where a ij The experts assign values ​​based on their actual experience. To ensure consistency, the maximum eigenvalue λ of the judgment matrix is ​​calculated. max And the consistency ratio CR:

[0066]

[0067] Among them, CI is the consistency index, n is the matrix order, and RI is the random consistency index. When CR<0.1, the judgment matrix passes the consistency test, and the weight vector w is obtained by normalizing the eigenvector. The weight vector w is weighted and fused with each evaluation matrix, and the comprehensive evaluation index S is expressed as:

[0068] S=w1Mp +w2M c +w3M e +w4M e ;

[0069] Among them, w1, w2, w3, w4 are the weight coefficients of each index, M p ,M c ,M e ,M e is the corresponding evaluation matrix. The fuzzy membership function is applied to the comprehensive evaluation index S to perform grade classification, and the membership function is defined as:

[0070]

[0071] Where S low and S high are the thresholds of the evaluation level. After calculating the membership, the area is divided into four levels: excellent, good, medium, and poor. The partition evaluation parameter matrix M is generated. f .

[0072] In a specific embodiment, the process of executing step S3 may specifically include the following steps:

[0073] Perform feature analysis on the partition assessment parameter matrix, and map the partition status assessment level, personnel distribution data, equipment operation status data, and environmental parameter data into state space vectors;

[0074] Encode the ventilation system adjustment parameters, personnel diversion parameters and equipment start and stop parameters to construct the action space vector;

[0075] A Q-value function is established based on the state space vector and the action space vector, and the neural network parameters of the Q-value function are initialized to obtain a deep reinforcement learning network;

[0076] Input the state space vector into the deep reinforcement learning network, obtain the action value function through forward calculation, and select the optimal action based on the action value function;

[0077] The reward is measured for the optimal action execution result, a reward function is constructed, and the reward function is input into the experience replay pool to obtain a training sample set;

[0078] Randomly extract training batches from the training sample set, use the temporal difference algorithm to calculate the target Q value, obtain the loss function, and perform back propagation operation and gradient descent optimization on the loss function, update the parameters of the deep reinforcement learning network, and obtain the optimized Q value function;

[0079] The optimized Q-value function is applied to the current state space vector to output the optimal configuration combination of ventilation system adjustment parameters, personnel diversion parameters and equipment start-stop parameters to obtain the partition management control parameters.

[0080] Specifically, the partition evaluation parameter matrix is ​​feature parsed, and the partition status evaluation level, personnel distribution data, equipment operation status data, and environmental parameter data contained therein are mapped into a state space vector s. The state space vector is a multidimensional vector, and each dimension corresponds to a key parameter. For example, the partition status evaluation level is represented by l, and the personnel distribution data is represented by ρ p Indicates that the equipment operation status data is represented by η e The environmental parameter data is represented by θ (comprehensively reflecting temperature, humidity, etc.). The state vector is defined as:

[0081] s=[l,ρ p ,η e ,θ];

[0082] At the same time, the controllable parameters are encoded to construct the action space vector a. The dimensions of the action space vector include the ventilation system adjustment parameters v c , Personnel diversion parameter ρ r And the equipment start and stop parameters s (Binary value, 0 means off, 1 means on). The action vector is defined as:

[0083] a=[v c ,ρ r ,e s ];

[0084] Based on the state space vector and the action space vector, a Q-value function Q(s,a) is established, which represents the expected reward obtained by taking action a under a given state s. In order to solve the Q-value function, a deep reinforcement learning network is constructed, and Q(s,a) is fitted through a neural network. Assuming that the network weights and biases are W and b respectively, the Q-value function is expressed as:

[0085] Q(s,a)=f(W·[s,a]+b);

[0086] Where f is a nonlinear activation function, such as ReLU. When the network is initialized, the weights W and biases b are initialized by random distribution to establish the initial structure of the deep reinforcement learning model. The state space vector s is input into the deep reinforcement learning network, and the action value function Q(s,a) is obtained by forward calculation. Each action value output by the network corresponds to a possible action a, and the optimal action in the current state is selected by comparing the action values:

[0087]

[0088] Choose the best action a *After that, it is applied to the actual system and the effect of the action is evaluated. The effect of the action is measured by the reward function r. The design of the reward function depends on the optimization goal, such as reducing the radiation dose of personnel D, improving the partition efficiency E, and reducing energy consumption C. The reward function is defined as:

[0089] r = -αD + βE - γC;

[0090] Among them, α, β, and γ are weight coefficients used to balance the influence of different goals. The action results and rewards r are stored in the experience replay pool to form a training sample set where s ′ It is the new state of the system after executing the action. During the training process, training batches are randomly selected from the experience replay pool, and the target Q value Q is calculated using the temporal difference algorithm (TD algorithm). target :

[0091]

[0092] Among them, γ is a discount factor, which is used to weigh the relative importance of current rewards and future rewards. The loss function L is calculated based on the target Q value:

[0093]

[0094] Among them, N is the number of samples in the training batch, Q pred is the predicted Q value output by the network. Through the back propagation algorithm and gradient descent method, the network parameters W and b are optimized:

[0095]

[0096] Where η is the learning rate. When the deep reinforcement learning network training is completed, the optimized Q value function Q * (s,a) is applied to the current state vector s to output the ventilation system adjustment parameters Personnel diversion parameters and equipment start and stop parameters These parameters are used to guide the zoning management decisions of the decontamination station to ensure that the operation is completed in a safe, efficient and energy-saving manner.

[0097] In a specific embodiment, the process of executing step S4 may specifically include the following steps:

[0098] Input the partition management control parameters, radiation dose rate data, contamination range data, personnel quantity data and equipment status data into the decision tree analysis model to construct the decision tree feature nodes;

[0099] Calculate the Gini coefficient of the feature nodes of the decision tree, and split the decision tree nodes according to the Gini coefficient to obtain the decision branch structure;

[0100] The radiation dose rate data in the decision branch structure are threshold-stratified, the radiation dose rate is divided into different hazard level intervals, and a dose rate threshold matrix is ​​obtained;

[0101] Perform spatial cluster analysis on the pollution range data in the decision branch structure to obtain the pollution area distribution matrix, and perform density calculation based on the number of people data in the decision branch structure to obtain the regional congestion matrix;

[0102] Perform fault diagnosis and analysis on equipment status data and construct equipment operation status vector;

[0103] The dose rate threshold matrix, contaminated area distribution matrix, regional congestion matrix and equipment operation state vector are input into the multi-layer decision rule library to perform rule matching and response level determination to obtain the emergency response level;

[0104] Based on the emergency response level, the ventilation system parameters, protection door parameters, waste liquid treatment system parameters and emergency lighting system parameters are linked and controlled and calculated to obtain emergency disposal control instructions.

[0105] Specifically, the partition management control parameter C, radiation dose rate data R d 、Pollution range data A p 、Number of personnel data p And the equipment status data S e Input the decision tree analysis model and construct the decision tree feature nodes. Each feature node contains specific parameter values ​​and their corresponding status information, such as the ventilation system adjustment parameters, personnel diversion parameters, and equipment start and stop status in C; R d Indicates the radiation level of the partition, A p and N p Represents the dynamic data of pollution range and personnel distribution, S e Indicates the health of the equipment's operating status. By constructing feature nodes and forming the initial structure of the decision tree, the model can extract key information from multidimensional data. The Gini coefficient of the decision tree feature nodes is calculated to measure the impurity of the data. The Gini coefficient G is defined as:

[0106]

[0107] Among them, p i represents the proportion of samples in the i-th category, and k is the number of categories. When G = 0, it means that the data purity is the highest. Based on the Gini coefficient, the feature nodes are split, and the feature that reduces the Gini coefficient the most is selected as the basis for splitting, and the decision branch structure is constructed. For example, for the radiation dose rate R d , choose a split point R t Divide the data into R d ≤Rt and R d >R t After obtaining the decision branch structure, the radiation dose rate data R in the branch d Perform threshold stratification analysis and divide it into different hazard level intervals. These intervals are used to generate the dose rate threshold matrix M r , each element of the matrix corresponds to the radiation level of a partition. At the same time, the pollution range data A in the decision branch p Spatial clustering analysis is performed to identify the distribution characteristics of pollutants. The DBSCAN algorithm (density clustering algorithm) is used to set the distance threshold ε and the minimum number of samples MinPts, and the points that meet the density requirements are divided into a cluster. The pollution range clustering results form the pollution area distribution matrix M p , each row of the matrix represents a contaminated area, including its center point coordinates, coverage area and other information. According to the number of personnel data N in the decision branch p , calculate the regional congestion ρ p :

[0108]

[0109] Among them, A p is the area of ​​the corresponding region, ρ p Represents the density of people per unit area, which is used to generate the regional congestion matrix M d . For device status data S e Perform fault diagnosis analysis. Use a logistic regression model to predict whether the equipment is faulty. The model input is the equipment operation index, such as temperature T e , vibration amplitude V e and running time H e . Equipment operation state vector v e Defined as:

[0110] v e =[T e ,V e ,H e ,S e ];

[0111] After diagnostic analysis, the operating status of the output device is normal (1) or abnormal (0). r , Pollution area distribution matrix M p , regional congestion matrix M d and the device operation status vector v e Input the multi-layer decision rule base and combine it with the predefined response rules to perform rule matching and emergency response level determination. r The radiation level in a certain area is extremely dangerous, and M dIndicates that the area is highly congested, and the response rule is determined to be a level 1 response. e The ventilation system is faulty and the response level is further increased. r , for ventilation system parameters v c , protective door parameters g d (on / off status), waste liquid treatment system parameters w p and emergency lighting system parameters i l Perform linkage control calculation. The linkage control model is defined as:

[0112] [v c ,g d ,w p ,i l ]=f(L r ,M r ,M p ,M d ,v e );

[0113] Function f is a set of control rules that dynamically adjusts system parameters according to the needs of each level of response. For example, in the first level response, the ventilation speed v c Set to the maximum value, the protective gate g d Automatic shut-off, waste liquid treatment flow w p Increase to upper limit, emergency lighting l Set to full area coverage.

[0114] In a specific embodiment, the method for performing zoning management of nuclear emergency decontamination stations based on artificial intelligence further includes the following steps:

[0115] Feature extraction is performed on the partition status data, equipment operation data and environmental parameter data in the emergency disposal control instructions to obtain the decontamination station operation status matrix;

[0116] According to the operation status matrix of the decontamination station, the surface contamination level data, radiation dose rate data and physical condition data of the personnel are standardized to obtain the comprehensive feature matrix of the personnel, and the comprehensive feature matrix of the personnel is input into the deep learning classification network, and the feature dimension reduction and extraction are performed through the convolution layer and the pooling layer to obtain the feature vector of the personnel;

[0117] Perform multi-dimensional cluster analysis on personnel feature vectors, divide personnel into heavy pollution group, medium pollution group and light pollution group, obtain personnel classification data, and build a decontamination channel load balancing model based on the personnel classification data, dynamically evaluate the processing capacity of each decontamination channel, and obtain the channel capacity state matrix;

[0118] A multi-objective path optimization model is established based on the channel capacity state matrix, and the decontamination time, channel congestion and radiation protection requirements are used as constraints to obtain the path optimization parameters.

[0119] The path optimization parameters are input into the intelligent path-finding system, and the improved A* algorithm is used to calculate multiple candidate paths to obtain a set of candidate paths. The set of candidate paths is comprehensively scored, and the path with the highest decontamination efficiency and the lowest radiation dose is selected as the optimal decontamination path data;

[0120] Multi-dimensional optimization calculations are performed based on personnel classification data and optimal disinfection path data to obtain disinfection process parameters and waste treatment parameters.

[0121] Specifically, feature extraction is performed to form the operation status matrix M of the disinfection station s Matrix M s Contains data reflecting the partition status S f , Equipment operating status S e and environmental parameters P e For example, the partition status data S f Including pollution level, partition radiation dose rate R d , equipment operation data S e Including equipment start and stop status E (0 or 1), equipment failure rate F, and environmental parameters P e Including temperature and humidity T, H. These data are dimensionless through the standardized formula:

[0122]

[0123] Where X is the original data, X min , X max are the minimum and maximum values ​​of the data, respectively, norm It is standardized data to ensure that different indicators are calculated on a unified scale. s Data on the level of contamination on the surface of personnel P c , radiation dose rate data R h and physical condition data B s Combined, through standardization processing, the comprehensive feature matrix M of personnel is generated p For example, the pollution level P of a person c and dose rate R h After normalization by formula, combined with physical condition data (such as heart rate H r ) constitutes a complete feature vector. p Input the deep learning classification network, and use the convolution layer and pooling layer to reduce the dimension and extract the features. The convolution layer extracts local features through the convolution kernel W, and the output is:

[0124] Z=f(W·Mp + b);

[0125] Where f is the activation function (such as ReLU), b is the bias term, and Z is the convolution output. The pooling layer further reduces the dimension through the maximum pooling or average pooling operation, reduces redundant data and improves the generalization ability of the model. After multiple layers of convolution and pooling, the obtained personnel feature vector v p It is a low-dimensional representation that retains key features. Based on the person feature vector v p , multi-dimensional cluster analysis is used to group the personnel. K-means clustering algorithm is used to divide the personnel into heavy pollution group, medium pollution group and light pollution group. Assume that the personnel data set is The clustering goal is to minimize the within-group squared error:

[0126]

[0127] Where K is the number of clusters, C k is the sample set of the kth class, μ k is the center of the kth class. The clustering results form the personnel classification data D p , based on D p , construct the load balancing model of the disinfection channel, dynamically evaluate the processing capacity of each disinfection channel, and obtain the channel capacity state matrix M c The dynamic evaluation of channel capacity includes the current processing load L c , Channel availability A c (binary value), and the waiting time W t The channel status is calculated by the following formula:

[0128]

[0129] Among them C t is the channel capacity state. c , establish a multi-objective path optimization model to determine the decontamination time T w , channel congestion ρ c and radiation protection requirements R b As constraints, optimize the path parameters. The objective function is expressed as:

[0130]

[0131] Among them, α, β, γ are weight coefficients, and P is the path planning variable. The path optimization parameters are input into the intelligent path-finding system, and the improved A* algorithm is used to calculate multiple candidate paths. The A* algorithm calculates the path cost through the heuristic function f(n) = g(n) + h(n), where g(n) is the current path cost and h(n) is the estimated minimum cost to the target. Combined with the decontamination efficiency and radiation exposure, the optimal path P is selected after comprehensive scoring. *. Based on personnel classification data D p and the optimal path data P * , and perform multi-dimensional optimization operations. The disinfection process parameters include water pressure P w , Decontamination time T w , waste treatment parameters include waste liquid flow rate F w , Solid waste amount S w Through dynamic adjustment, these parameters ensure the highest decontamination efficiency and optimal use of resources.

[0132] In a specific embodiment, the execution step performs multi-dimensional optimization calculations based on the personnel classification data and the optimal disinfection path data to obtain the disinfection process parameters and waste treatment parameters, which may specifically include the following steps:

[0133] Perform multi-sample analysis on the surface contamination level data and radiation dose rate data in the personnel classification data to obtain the decontamination difficulty level matrix;

[0134] Based on the decontamination difficulty level matrix, multi-dimensional cross-calculations are performed on the decontamination time data, water pressure strength data, and decontamination agent dosage data to obtain the decontamination parameter initial matrix. The decontamination parameter initial matrix is ​​then optimized in time and space distribution according to the optimal decontamination path data to obtain the decontamination process parameter combination sequence.

[0135] The combination sequence of disinfection process parameters is input into the multi-objective optimization model, and the Pareto optimal calculation of disinfection efficiency data and waste generation data is performed to obtain the disinfection parameter optimization matrix;

[0136] According to the decontamination parameter optimization matrix, a dynamic balance calculation is performed on the concentration multiple data, treatment cycle data and discharge volume data of the waste liquid treatment system to obtain a waste liquid treatment parameter set;

[0137] Perform system capacity constraint analysis and multi-objective planning calculation on the waste liquid treatment parameter set to obtain the waste liquid treatment optimization parameters, and build a solid waste treatment model based on the waste liquid treatment optimization parameters, optimize the solid waste compression ratio data, storage cycle data and disposal volume data to obtain the solid waste treatment parameters;

[0138] The disinfection parameter optimization matrix, waste liquid treatment optimization parameters and solid waste treatment parameters are calculated in multiple dimensions to obtain the disinfection process parameters and waste treatment parameters.

[0139] Specifically, the surface contamination level data C in the personnel classification data s and radiation dose rate data R d Perform multi-sample analysis to generate the disinfection difficulty level matrix M d Each cell of this matrix corresponds to the difficulty level of decontamination for a certain type of contaminated personnel, based on C s and R dFor example, by s and R d The area is divided into lightly contaminated, moderately contaminated and heavily contaminated areas, and the decontamination difficulty score for each category is calculated using the following formula:

[0140] D=α·C s +β·R d ;

[0141] Among them, α and β are weight coefficients used to balance the relative contribution of surface contamination and radiation dose to the decontamination difficulty. d , for the decontamination time data T w , Water pressure strength data P w And disinfectant dosage data Q c Perform multi-dimensional cross calculation to generate the initial matrix M of disinfection parameters p For example, assuming that the decontamination difficulty D and decontamination time T w The decontamination time is calculated by the following formula:

[0142] T w = k·D;

[0143] Where k is the proportionality coefficient. Similarly, the water pressure and the amount of disinfectant can also be adjusted according to the difficulty of disinfection, and are calculated as:

[0144] P w =m·D,Q c = n·D;

[0145] Where m and n are the corresponding proportional coefficients. These parameters together make up M p , indicating the decontamination strategies corresponding to different pollution levels. * Applied to the initial matrix M of decontamination parameters p Optimize the time and space distribution of disinfection. Assuming that the disinfection path contains K nodes, each node has different time and space constraints, the optimization goal is to minimize the use of disinfection resources and time conflicts. The optimization formula is:

[0146]

[0147] Among them, λ1 and λ2 are weight factors used to balance the impact of time and resource usage. is the combined sequence of disinfection process parameters after time-space distribution optimization. Input the multi-objective optimization model to the decontamination efficiency data E w and waste generation data W f As the goal, the Pareto optimal calculation method is used to generate the disinfection parameter optimization matrix M oThe optimization goal is to maximize the decontamination efficiency and minimize the amount of waste generated. The optimization problem is defined as:

[0148]

[0149] The Pareto frontier method identifies the optimal solution without sacrificing any objective, forming M o , containing the optimal combination of decontamination parameters. Based on M o , the concentration multiple data C of the waste liquid treatment system f , Processing cycle data T f and emissions data Q f Perform dynamic equilibrium calculations. The concentration factor is calculated using the following formula:

[0150]

[0151] Where Q input and Q output are the volumes of waste liquid entering and discharged respectively. Treatment cycle T f Calculated by waste flow rate:

[0152]

[0153] Where F is the waste liquid treatment flow rate. Dynamic balance calculation generates the waste liquid treatment parameter set P f . f Conduct system capacity constraint analysis and multi-objective planning calculations to optimize wastewater concentration efficiency and discharge volume, and generate wastewater treatment optimization parameters At the same time, based on Construct a solid waste treatment model and calculate the solid waste compression ratio C s , Storage cycle T s and disposal volume Q s The compression ratio is calculated by the following formula:

[0154]

[0155] Where V initial and V final are the volumes before and after compression respectively. The disinfection parameter optimization matrix M o , Optimization parameters of wastewater treatment and solid waste treatment parameters Perform multi-dimensional fusion calculations to generate the final disinfection process parameters and waste treatment parameters.

[0156] The above describes the nuclear emergency decontamination station zoning management method based on artificial intelligence in the embodiment of the present invention. The following describes the nuclear emergency decontamination station zoning management device based on artificial intelligence in the embodiment of the present invention. Figure 2In one embodiment of the present invention, an artificial intelligence-based nuclear emergency decontamination station zoning management device includes:

[0157] The acquisition module is used to perform 3D laser scanning and data acquisition on the contaminated area, decontamination area, and clean area of ​​the nuclear emergency decontamination station, and to build a digital twin model of the decontamination station;

[0158] The analysis module is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area according to the digital twin model of the decontamination station using the hierarchical analysis method to obtain a partition evaluation parameter matrix;

[0159] A calculation module is used to input the partition evaluation parameter matrix into the deep reinforcement learning network for dynamic optimization calculation to obtain the partition management control parameters;

[0160] The generation module is used to perform decision analysis and threshold judgment on radiation dose rate data, contamination range data, personnel quantity data and equipment status data according to the partition management control parameters, and generate emergency disposal control instructions.

[0161] Through the collaboration of the above components, the digital twin model is established to achieve all-round perception and real-time monitoring of the decontamination station. The integration of digital twin technology and multi-source sensor data provides an accurate status assessment basis, making the operation status of each partition clear and controllable. The analytic hierarchy process is used for multi-dimensional analysis and weight calculation, and a scientific partition assessment system is constructed to achieve a comprehensive assessment of the degree of pollution, personnel capacity, equipment status and environmental parameters, providing a quantitative basis for management decisions. The deep reinforcement learning network is introduced for dynamic optimization calculation, and the partition management strategy is continuously optimized through the Q-learning algorithm, so that the system can learn autonomously and make optimal control decisions. A multi-level emergency response mechanism based on decision trees is designed, and the intelligent hierarchical response of emergency disposal is realized through threshold judgment and rule matching, which improves the accuracy and timeliness of emergency disposal. An intelligent personnel classification and path planning system is constructed, and the accurate classification of personnel pollution degree is realized through deep learning algorithms, and the optimal selection of decontamination paths is ensured through multi-objective optimization. The parameter optimization method of decontamination process and waste treatment is developed, and the balance between decontamination efficiency and waste treatment is achieved through multi-dimensional optimization calculation, ensuring the efficiency and environmental protection of the disposal process. The coordinated linkage of contaminated areas, disinfection areas and cleaning areas is achieved, and information sharing and equipment linkage are achieved between functional areas, forming a complete intelligent management closed loop.

[0162] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0163] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0164] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0165] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0166] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0167] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0168] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A nuclear emergency decontamination station zoning management method based on artificial intelligence, characterized in that: The method comprises: Conduct 3D laser scanning and data collection on the contaminated area, decontamination area, and clean area of ​​the nuclear emergency decontamination station to build a digital twin model of the decontamination station; According to the digital twin model of the decontamination station, a hierarchical analysis method is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area to obtain a partition evaluation parameter matrix; Inputting the partition evaluation parameter matrix into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters; According to the zoning management control parameters, decision analysis and threshold judgment are performed on the radiation dose rate data, contamination range data, personnel quantity data and equipment status data to generate emergency disposal control instructions.

2. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 1 is characterized in that: The three-dimensional laser scanning and data collection of the contaminated area, the decontamination area and the clean area of ​​the nuclear emergency decontamination station are carried out to build a digital twin model of the decontamination station, including: Performing three-dimensional laser scanning on the contaminated area, the decontamination area, and the clean area to obtain three-dimensional point cloud data of the decontamination station, and performing grid processing and feature extraction on the three-dimensional point cloud data to obtain a three-dimensional structural model of the decontamination station; The contaminated area monitoring data collected by the radiation dose rate monitor, the personnel radioactive surface contamination monitor, and the temperature and humidity sensor are processed by data standardization to obtain the contaminated area sensor data matrix; The monitoring data of the decontamination area collected by the intelligent water pressure sensor, water quality monitor, aerosol radioactivity concentration detector, waste liquid pool level sensor, and waste liquid radioactivity concentration continuous monitor are processed by data standardization to obtain the decontamination area sensor data matrix; The clean area monitoring data collected by the radiation background value monitoring device and the air quality monitor are processed by data standardization to obtain the clean area sensor data matrix; Performing time series data analysis on the contaminated area sensor data matrix, the decontamination area sensor data matrix, and the clean area sensor data matrix to obtain decontamination station dynamic monitoring data; The three-dimensional structural model of the decontamination station and the dynamic monitoring data of the decontamination station are subjected to data fusion and feature mapping to obtain a digital twin mapping relationship of the decontamination station. Based on the digital twin mapping relationship of the decontamination station, a data synchronization mechanism that reflects the interaction between the physical entity and the virtual model is constructed to obtain a digital twin model of the decontamination station.

3. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 1 is characterized in that: According to the digital twin model of the decontamination station, the analytic hierarchy process is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area and the clean area to obtain a partition evaluation parameter matrix, including: Normalizing the radiation dose rate data, surface contamination data, radionuclide activity concentration data in the air, and radionuclide activity concentration data in the water in the digital twin model of the decontamination station to obtain a contamination degree assessment matrix; Normalizing the partition area utilization data, personnel congestion data, and personnel traffic efficiency data in the digital twin model of the decontamination station to obtain a personnel capacity assessment matrix; Normalizing the decontamination equipment integrity data, the waste liquid treatment system operation status data, and the ventilation system efficiency data in the decontamination station digital twin model to obtain an equipment status evaluation matrix; Normalizing the temperature data, humidity data, air pressure difference data, and ventilation frequency data in the digital twin model of the decontamination station to obtain an environmental parameter evaluation matrix; Based on the pollution degree assessment matrix, the personnel capacity assessment matrix, the equipment status assessment matrix and the environmental parameter assessment matrix, a hierarchy analysis method judgment matrix is ​​constructed, and eigenvalue calculation and consistency test are performed on the judgment matrix to obtain the weight coefficient of each evaluation index; The weight coefficient is weighted and fused with the pollution degree assessment matrix, the personnel capacity assessment matrix, the equipment status assessment matrix and the environmental parameter assessment matrix to obtain a comprehensive assessment index, and the comprehensive assessment index is fuzzy-membered and graded to obtain a partition assessment parameter matrix.

4. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 1 is characterized in that: The partition evaluation parameter matrix is ​​input into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters, including: Performing feature analysis on the partition assessment parameter matrix, mapping the partition status assessment level, personnel distribution data, equipment operation status data and environmental parameter data into a state space vector; Encode the ventilation system adjustment parameters, personnel diversion parameters and equipment start and stop parameters to construct the action space vector; Establishing a Q-value function based on the state space vector and the action space vector, initializing neural network parameters for the Q-value function, and obtaining a deep reinforcement learning network; Inputting the state space vector into the deep reinforcement learning network, obtaining an action value function through forward calculation, and selecting an optimal action based on the action value function; Performing reward measurement on the optimal action execution result, constructing a reward function, and inputting the reward function into an experience replay pool to obtain a training sample set; Randomly extracting a training batch from the training sample set, using a temporal difference algorithm to calculate a target Q value to obtain a loss function, performing back propagation operation and gradient descent optimization on the loss function, updating the parameters of the deep reinforcement learning network, and obtaining an optimized Q value function; The optimized Q value function is applied to the current state space vector, and the optimal configuration combination of the ventilation system adjustment parameters, the personnel diversion parameters and the equipment start and stop parameters is output to obtain the partition management control parameters.

5. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 1 is characterized in that: According to the partition management control parameters, decision analysis and threshold judgment are performed on the radiation dose rate data, contamination range data, personnel quantity data and equipment status data to generate emergency disposal control instructions, including: Input the partition management control parameters, radiation dose rate data, contamination range data, personnel quantity data and equipment status data into a decision tree analysis model to construct a decision tree feature node; Calculating the Gini coefficient for the feature nodes of the decision tree, and splitting the decision tree nodes according to the Gini coefficient to obtain a decision branch structure; Performing threshold stratification on the radiation dose rate data in the decision branch structure, dividing the radiation dose rate into different hazard level intervals, and obtaining a dose rate threshold matrix; Performing spatial cluster analysis on the pollution range data in the decision branch structure to obtain a pollution area distribution matrix, and performing density calculation based on the number of people data in the decision branch structure to obtain a regional congestion matrix; Performing fault diagnosis analysis on the device status data to construct a device operation status vector; Input the dose rate threshold matrix, the contaminated area distribution matrix, the area congestion matrix and the equipment operation state vector into a multi-layer decision rule library, perform rule matching and response level determination, and obtain an emergency response level; Based on the emergency response level, linkage control calculation is performed on ventilation system parameters, protection door parameters, waste liquid treatment system parameters and emergency lighting system parameters to obtain emergency disposal control instructions.

6. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 5 is characterized in that: The artificial intelligence-based nuclear emergency decontamination station zoning management method also includes: Extracting features from the partition status data, equipment operation data, and environmental parameter data in the emergency disposal control instructions to obtain a decontamination station operation status matrix; According to the operation status matrix of the decontamination station, data standardization is performed on the surface contamination level data, radiation dose rate data and physical condition data of the personnel to obtain a comprehensive feature matrix of the personnel, and the comprehensive feature matrix of the personnel is input into the deep learning classification network, and feature dimension reduction and extraction are performed through the convolution layer and the pooling layer to obtain a feature vector of the personnel; Performing multi-dimensional cluster analysis on the personnel feature vectors, dividing the personnel into a heavy pollution group, a moderate pollution group, and a light pollution group, obtaining personnel classification data, and constructing a decontamination channel load balancing model based on the personnel classification data, dynamically evaluating the processing capacity of each decontamination channel, and obtaining a channel capacity state matrix; A multi-objective path optimization model is established according to the channel capacity state matrix, and the decontamination time, channel congestion and radiation protection requirements are used as constraints to obtain path optimization parameters; The path optimization parameters are input into the intelligent path-finding system, and the improved A* algorithm is used to calculate multiple candidate paths to obtain a set of candidate paths, and the set of candidate paths is comprehensively scored to select the path with the highest decontamination efficiency and the lowest radiation dose as the optimal decontamination path data; A multi-dimensional optimization operation is performed based on the personnel classification data and the optimal disinfection path data to obtain disinfection process parameters and waste treatment parameters.

7. The artificial intelligence-based nuclear emergency decontamination station zoning management method according to claim 6 is characterized in that: The multi-dimensional optimization operation is performed based on the personnel classification data and the optimal disinfection path data to obtain disinfection process parameters and waste treatment parameters, including: Performing multi-sample analysis on the surface contamination level data and radiation dose rate data in the personnel classification data to obtain a decontamination difficulty level matrix; Based on the decontamination difficulty level matrix, a multi-dimensional cross-calculation is performed on the decontamination time data, the water pressure strength data and the decontamination agent dosage data to obtain an initial decontamination parameter matrix, and the time-space distribution of the initial decontamination parameter matrix is ​​optimized according to the optimal decontamination path data to obtain a decontamination process parameter combination sequence; The combination sequence of the decontamination process parameters is input into a multi-objective optimization model, and Pareto optimal calculation is performed on the decontamination efficiency data and the waste generation data to obtain a decontamination parameter optimization matrix; According to the decontamination parameter optimization matrix, a dynamic balance calculation is performed on the concentration multiple data, the treatment cycle data and the discharge volume data of the waste liquid treatment system to obtain a waste liquid treatment parameter set; Performing system capacity constraint analysis and multi-objective planning calculation on the waste liquid treatment parameter set to obtain waste liquid treatment optimization parameters, and constructing a solid waste treatment model based on the waste liquid treatment optimization parameters, optimizing solid waste compression ratio data, storage cycle data, and disposal volume data to obtain solid waste treatment parameters; The disinfection parameter optimization matrix, the waste liquid treatment optimization parameters and the solid waste treatment parameters are subjected to multi-dimensional fusion calculation to obtain disinfection process parameters and waste treatment parameters.

8. A nuclear emergency decontamination station zoning management device based on artificial intelligence, characterized in that: Used to execute the artificial intelligence-based nuclear emergency decontamination station zoning management method as described in any one of claims 1 to 7, the device comprises: The acquisition module is used to perform 3D laser scanning and data acquisition on the contaminated area, decontamination area, and clean area of ​​the nuclear emergency decontamination station, and to build a digital twin model of the decontamination station; An analysis module is used to perform multi-dimensional analysis and weight calculation on the contaminated area, the decontamination area, and the clean area according to the digital twin model of the decontamination station by using a hierarchical analysis method to obtain a partition evaluation parameter matrix; A calculation module, used for inputting the partition evaluation parameter matrix into a deep reinforcement learning network for dynamic optimization calculation to obtain partition management control parameters; The generation module is used to perform decision analysis and threshold judgment on the radiation dose rate data, contamination range data, personnel quantity data and equipment status data according to the partition management control parameters, and generate emergency disposal control instructions.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and is characterized in that when the processor executes the computer program, the artificial intelligence-based nuclear emergency decontamination station zoning management method described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor executes the artificial intelligence-based nuclear emergency decontamination station zoning management method as described in any one of claims 1 to 7.

Images