A method and system for multi-point synchronous detection and identification of radioactive substances

Abstract

The application discloses a kind of radioactive material multi-point synchronous detection and identification method and system, it is related to radioactive material monitoring technical field.The application includes: step one, real-time acquisition multi-point radiation data, data pre-processing and storage are carried out;Step two, construct space adjacency matrix, estimate local space gradient, calculate abnormal degree value, take optimization measures, and carry out collaborative anomaly evaluation to monitoring node;Step three, calculate total fitting error value, real-time back-propagation leakage source position and leakage intensity, real-time take corresponding measures;Step four, carry out abnormal trend identification, calculate early warning determination value, and implement hierarchical early warning and multi-node cross confirmation measures;Step five, receive and store various data, carry out the visual analysis of data, support event processing and feedback.The application solves the problem that trace leakage is difficult to be detected by single-point monitor due to good shielding and physical barrier of some equipment, and there is a long-term risk of pollution retention.

Description

Technical Field

[0001] This invention relates to the field of radioactive material monitoring technology, specifically to a method and system for multi-point synchronous detection and identification of radioactive materials. Background Technology

[0002] In scenarios such as public safety, nuclear facility supervision, counter-terrorism, and environmental monitoring, the illegal release or accidental leakage of radioactive materials poses a serious threat to human health, ecosystems, and social stability. With the development of wireless communication, distributed sensing, and data analysis technologies, building an intelligent detection system based on multi-node synchronous sensing, anomaly trend fusion judgment, and source tracing has become a trend.

[0003] For example, invention patent CN108089218B discloses a radioactivity monitoring system for entry and exit from a nuclear power plant's controlled area. This system is based on the following arrangement: a changing room and a dose management room are located within the sanitary passageway; a non-radioactive work area is located outside the sanitary passageway; the sanitary passageway leads to the radioactive control area. The changing room is located in the first area entered within the sanitary passageway, and the dose management room is located after the changing room at the boundary between the non-radioactive work area and the radioactive control area. The changing room is equipped with a self-service locker, and the dose management room is equipped with a self-service check-in machine. Both the locker and the machine have card readers connected to a universal internet network and interconnected with the nuclear power plant's server system. Before entering the changing room, staff must be equipped with a dedicated personal dose management terminal, which must be carried at all times when entering and exiting the sanitary passageway or working in the controlled and non-controlled areas.

[0004] For example, invention patent CN102713676B discloses anomaly detection of radioactive features. In this method, a target gamma-ray spectrum is obtained from the target, and a target data set is prepared using the target gamma-ray spectrum. This data set includes multiple intensity values, each associated with an energy chamber representing the gamma-ray energy or gamma-ray energy range in the target gamma-ray spectrum. The target data set is then preprocessed and projected into a principal component space containing a preprocessed database projected into the principal component space. The distance between the projected preprocessed target data set and one or more clusters of the projected preprocessed database is then determined in the principal component space, and this distance is compared with a predetermined threshold distance to determine whether anomalous radioactive material is present in the target.

[0005] However, in the process of implementing the inventive technical solution in the embodiments of this application, it was found that the above-mentioned technology has at least the following technical problems:

[0006] On the one hand, traditional technologies lack effective suppression of short-term fluctuations and electronic noise in the environment, making it difficult to accurately identify weak abnormal signals and easily leading to false alarms or missed alarms. On the other hand, traditional systems lack time-series trend modeling of historical data, making it impossible to identify abnormal deviations based on periodic background changes. Furthermore, the lack of effective information fusion and spatial collaborative judgment mechanisms among multi-node monitoring leads to isolated detection and judgment with poor spatial consistency. In addition, traditional systems generally lack the ability to invert the location and intensity of radioactive sources, making it difficult to achieve accurate location and coordinated response of leakage sources in complex environments.

[0007] Therefore, in order to address the above problems, there is an urgent need for a method and system for multi-point synchronous detection and identification of radioactive materials. Summary of the Invention

[0008] Technical problems to be solved

[0009] To address the shortcomings of existing technologies, this invention provides a method and system for multi-point synchronous detection and identification of radioactive materials. This solves the problem that due to the good shielding and physical barriers of equipment such as radiopharmaceutical cold storage or shielded storage cabinets, minute leaks are difficult to detect by single-point monitors, posing a risk of long-term contamination.

[0010] Technical solution

[0011] To achieve the above objectives, the present invention provides the following technical solution: a method for multi-point synchronous detection and identification of radioactive materials, comprising: Step 1, real-time acquisition of multi-point radiation data, data preprocessing, and real-time storage of the multi-point radiation data; Step 2, construction of a spatial adjacency matrix, estimation of local spatial gradient, calculation of anomaly degree value, real-time optimization measures based on the anomaly degree value, and collaborative anomaly assessment of monitoring nodes; Step 3, calculation of total fitting error value based on multi-point radiation data, real-time back-inference of leakage source location and leakage intensity based on the total fitting error value, and real-time implementation of corresponding measures; Step 4, collection of multi-point radiation data for real-time anomaly trend identification, calculation of early warning judgment value, and implementation of graded early warning and multi-node cross-confirmation measures; Step 5, receipt and storage of various types of data, data visualization analysis, and support for event processing and feedback.

[0012] Furthermore, the specific process of real-time acquisition of multi-point radiation data, data preprocessing, and real-time storage of multi-point radiation data is as follows: Multiple high-sensitivity monitoring nodes are deployed within the radioactive source storage area based on the site layout, potential leak points, and ventilation paths. Each monitoring node is equipped with a Geiger counter, a scintillation counter, and a semiconductor detector, and is assigned a unique identification code. Real-time acquisition of multi-point radiation data includes: radiation dose rate and count rate data of the surrounding environment, as well as the three-dimensional coordinates and equipment parameters of each monitoring node. Moving average filtering is used to denoise the real-time multi-point radiation data of each monitoring node, correct the background dose, remove stable baselines, and perform standardization and normalization. The multi-point radiation data is uploaded to a multi-point radiation database and sent to edge computing devices via a message queue telemetry transmission protocol. In case of network anomalies, the monitoring nodes automatically cache data and support remote firmware upgrades and periodic calibration.

[0013] Further, the specific process of constructing a spatial adjacency matrix, estimating local spatial gradients, calculating anomaly values, and taking real-time optimization measures based on anomaly values ​​is as follows: A spatial adjacency matrix is ​​constructed based on the three-dimensional coordinates recorded when each monitoring node is deployed; the Euclidean distance between all monitoring nodes is calculated, and when the distance between any two monitoring nodes is less than three meters, they are considered neighbors, and the adjacency relationship between the two neighboring monitoring nodes is recorded; based on the spatial adjacency matrix, the gradient components in the three-dimensional direction are calculated using the difference in radiation dose rate and the difference in three-dimensional coordinates between monitoring nodes; the gradient estimation method adopts a weighted difference form: the influence of each neighboring node on the monitoring node is weighted according to the inverse square of the distance, with a larger weight for closer distances; the gradient result is updated every second based on the current radiation dose rate, and it is determined whether there is a trend of dose increase along a certain direction in a certain area; a sliding filter is applied to the historical radiation dose rate data of each monitoring node, using a fixed window sliding average method to smooth out short-period fluctuations caused by electronic noise, environmental disturbances, and individual instantaneous events, to obtain the filtered dose value at the current moment; a daily background model based on a time period is established for each monitoring node: from the same time period of the past three consecutive days... The corresponding historical filtered dose values ​​are extracted, and a stable background mean is obtained by calculating the mean, which serves as the normal state baseline for the monitoring nodes. The standard deviation is calculated based on the filtered dose values ​​from the same time period over the past three consecutive days to obtain the background standard deviation. The filtered dose value of the i-th monitoring node at time t is obtained, along with its background mean and background standard deviation. The standard deviation is obtained by subtracting the background mean from the filtered dose value of the i-th monitoring node at time t and then dividing by the sum of the background standard deviation and a constant. The anomaly threshold weighting factor is subtracted from the standard deviation to obtain the anomaly level of the i-th monitoring node at time t. Anomaly degree value; Perform a maximum function operation on the anomaly degree value: If the anomaly degree value is less than zero, set the anomaly degree value to zero; If the anomaly degree value is not less than zero, the calculated result is the anomaly degree value; When the anomaly degree values ​​of multiple consecutive nodes in a certain area rise abnormally, shorten the sliding window, automatically send on-site verification prompts, and call the regional ventilation system to dilute potential radioactive materials in advance; If a node has a high anomaly degree value for a long period of time and neighboring nodes do not respond, start the sensor self-diagnosis program; Mark it as a "suspected false alarm" state, initiate remote self-calibration, automatically analyze historical performance and error trends, and automatically assess whether to replace the equipment.

[0014] Furthermore, the specific process for collaborative anomaly assessment of monitoring nodes is as follows: Obtain the anomaly severity values ​​of the i-th monitoring node and the j-th neighbor node at time t; obtain the spatial adjacency matrix, filter the neighbor nodes in the spatial adjacency matrix to obtain the set of neighbor nodes for the i-th monitoring node, and obtain the length of the neighbor node set to obtain the number of neighbor nodes; sum the anomaly severity values ​​of all neighbor nodes at time t based on the number of neighbor nodes, then divide the sum by the number of neighbor nodes to obtain the average anomaly value of the neighborhood; multiply the average anomaly value of the neighborhood by the neighborhood fusion weight factor, and then add it to the anomaly severity value of the i-th monitoring node at time t to obtain the collaborative anomaly assessment value; compare the collaborative anomaly assessment value with the first-level and second-level anomaly thresholds in real time. When the collaborative anomaly assessment value is less than or equal to the first-level anomaly threshold, it is considered that the radiation level is within the normal background fluctuation range, and routine monitoring continues to maintain normal equipment and data operation. Data collection is performed without intervention. When the collaborative anomaly assessment value is greater than the first-level anomaly threshold but less than or equal to the second-level anomaly threshold, a minor leak or equipment malfunction is considered to exist. The collaborative anomaly severity value is uploaded to the backend, and higher-frequency data collection is initiated. The system automatically monitors the status of anomaly monitoring nodes and surrounding equipment and environmental parameters. If the anomaly is confirmed to be valid and located in a poorly ventilated or high-risk area, the ventilation system will be automatically activated, local audible and visual alerts will be issued, and protective warning instructions will be sent to the site. When the collaborative anomaly assessment value is greater than the second-level anomaly threshold, the radiation level is considered abnormal, indicating a leak risk. The collaborative anomaly severity value is uploaded to the backend, and the on-site emergency plan is immediately activated to isolate the suspected leak area. Remote and on-site alarms are synchronized, and anomaly alerts are automatically pushed. Portable high-precision detectors are used for precise location and leak source confirmation. Cleanup and repair instructions are pushed according to the leak severity, while the execution progress and feedback are recorded.

[0015] Furthermore, the specific process of calculating the total fitting error value based on multi-point radiation data and then using the total fitting error value to infer the location and intensity of the leakage source in real time is as follows: The three-dimensional coordinates and corresponding radiation dose rates of each monitoring node are obtained, and anomaly filtering is performed based on equipment parameters; a dose gradient field is constructed using spatial difference based on the three-dimensional coordinates and real-time radiation dose rates of the monitoring nodes, and the location of the monitoring node with the highest radiation dose rate is selected as the initial leakage source location; the radiation dose rate of the leakage source location is inferred from the selected maximum radiation dose rate, the distance between the initial leakage source location and the current monitoring node, and the real-time radiation dose rate of the monitoring node, as the initial leakage intensity; based on the information criterion and spatial clustering fusion algorithm, spatial clustering analysis is performed on the anomaly values ​​of the radiation dose rates of all monitoring nodes. By identifying high radiation dose rate anomalous clustering areas, the number of potential source points is initially estimated. Under different assumed source numbers, leakage source inversion and localization calculations are performed using the Bayesian information criterion algorithm, and the number of source points with the smallest Bayesian information criterion value is selected as the current scenario. The optimal total number of leakage sources is determined by: obtaining the total number of monitoring nodes by acquiring the number of detectors deployed; acquiring the three-dimensional coordinates of the i-th monitoring node and the measured radiation dose rate of the i-th monitoring node; acquiring the initial leakage source location and initial leakage intensity; calculating the square of the Euclidean distance from the three-dimensional coordinates of the i-th monitoring node to the initial leakage source location and adding a constant 1; then dividing the initial leakage intensity by the summation result to obtain the leakage source contribution value; based on the total number of leakage sources, summing the leakage source contribution values ​​of all leakage sources to obtain the estimated dose value of the i-th monitoring node; subtracting the radiation dose rate of the i-th monitoring node from the estimated dose value of the i-th monitoring node and squaring the result to obtain the error square term; based on the total number of monitoring nodes, summing the error square terms of all monitoring nodes to obtain the initial total fitting error value of K leakage sources; based on the initial total fitting error value of K leakage sources, using a nonlinear least squares optimization algorithm to adjust the leakage source location and leakage intensity, gradually reducing the total fitting error value until the total fitting error value is less than the error threshold, and then back-calculating to obtain the current location and leakage intensity of the k-th leakage source.

[0016] Furthermore, the specific process of taking corresponding measures in real time is as follows: The total fitting error value is compared with the error threshold in real time. When the total fitting error value is less than or equal to the error threshold, the leak source location result is considered reliable and can be directly used for rapid response and on-site handling of the leak source; the three-dimensional coordinate position of the leak source, the leak intensity, and the total fitting error value are uploaded to the monitoring platform, triggering the leak emergency plan module in real time; risk levels are automatically assigned, and area containment, ventilation, audible and visual alarms, and protection commands are executed in conjunction, while continuously and dynamically monitoring the leak evolution process; when the total fitting error value is greater than the error threshold, the leak source location result is considered unreliable, a "fitting failure" prompt is output, and automatic push notifications are sent to check data quality and equipment status; abnormal multi-point radiation data and monitoring sensor malfunctions are investigated, and data cleaning and equipment calibration are performed; the existing monitoring node layout is evaluated; the total fitting error value is recalculated after adjustment until the total fitting error value meets the valid judgment.

[0017] Furthermore, the specific process of collecting multi-point radiation data for real-time anomaly trend identification and calculating the early warning judgment value is as follows: The edge computing device periodically collects multi-point radiation data transmitted by each monitoring node at a high frequency, establishes a rolling cache, and saves at least one hour of historical multi-point radiation data; and strictly synchronizes the cached multi-point radiation data using timestamps; obtains the radiation dose rate of the i-th monitoring node and the j-th neighbor node at time t, obtains the background mean of the i-th monitoring node and the j-th neighbor node, obtains the set of neighbor nodes of the i-th monitoring node, and obtains the length of the set of neighbor nodes of the i-th monitoring node to obtain the number of neighbor nodes; calculates the difference between the radiation dose rate of the i-th monitoring node and the background mean of the i-th monitoring node to obtain the normal deviation degree of the monitoring node; calculates the difference between the radiation dose rate of the j-th neighbor node and the background mean of the j-th neighbor node, and sums all differences based on the number of neighbor nodes, divides the sum by the number of neighbor nodes, and multiplies it by the neighbor influence weighting factor to obtain the neighbor influence term, and adds the neighbor influence term to the normal deviation degree to obtain the early warning judgment value.

[0018] Furthermore, the specific process of implementing graded early warning and multi-node cross-confirmation measures is as follows: The early warning judgment value is compared in real time with the first and second-level early warning thresholds. When the early warning judgment value is less than the first-level early warning threshold, it is determined to be normal, continuous monitoring is maintained, and periodic archiving is performed without intervention. When the early warning judgment value is greater than or equal to the first-level early warning threshold but less than the second-level threshold, it is determined to be a slight dose increase, possibly affected by disturbance or initial leakage. Local audible and visual alerts are issued, the radiation dose rate change rate is recorded in real time, and monitoring of key areas is strengthened. When the early warning judgment value is greater than or equal to the second-level threshold, it is determined to be a significant anomaly, suspected leakage source activity. The local buzzer and red light alarm are immediately activated, the early warning judgment value is uploaded, mobile detection equipment is automatically dispatched to check the abnormal area, and detection data is transmitted back in real time. The leakage source location and emergency handling procedures are initiated. If a single node alarms but neighboring nodes do not respond, the remote alarm will be delayed and marked as "pending confirmation." If the early warning judgment values ​​of three or more neighboring nodes are simultaneously greater than the early warning threshold and are spatially consistent, it will be upgraded to a "regional leakage early warning," automatically triggering the leakage source inversion and location module to start location.

[0019] Furthermore, the specific process of receiving and storing various types of data, performing data visualization analysis, and supporting event handling and feedback is as follows: The cloud server receives monitoring node identification codes, radiation dose rates, radiation dose rate change rates, location results, and timestamps in real time, and stores them uniformly in a time-series database; it also performs visualization analysis based on multi-point radiation data from monitoring nodes; it supports multi-level alarms and pushes notifications to relevant personnel through multiple channels such as SMS, email, and WeChat; it supports event handling records and backtracking; and it introduces machine learning models to intelligently identify and assess leakage patterns, and automatically generates risk assessment reports on a regular basis.

[0020] The second aspect of this invention provides a multi-point synchronous detection and identification system for radioactive materials, comprising: a multi-point synchronous monitoring module, a dose trend collaborative modeling module, a leak source inversion and localization module, an edge computing and early warning module, and a cloud monitoring and analysis module; wherein the multi-point synchronous monitoring module is used to collect multi-point radiation data in real time, perform data preprocessing, and store the multi-point radiation data in real time; the dose trend collaborative modeling module is used to construct a spatial adjacency matrix, estimate local spatial gradients, calculate anomaly severity values, take optimization measures in real time based on the anomaly severity values, and perform collaborative anomaly assessment on monitoring nodes; the leak source inversion and localization module is used to calculate the total fitting error value based on the multi-point radiation data, infer the leak source location and leakage intensity in real time based on the total fitting error value, and take corresponding measures in real time; the edge computing and early warning module is used to collect multi-point radiation data for real-time anomaly trend identification, calculate early warning judgment values, and implement graded early warning and multi-node cross-confirmation measures; the cloud monitoring and analysis module is used to receive and store various types of data, perform data visualization analysis, and support event processing and feedback.

[0021] Beneficial effects

[0022] The present invention has the following beneficial effects:

[0023] (1) This invention achieves collaborative anomaly assessment of monitoring nodes and timely detection of potential radioactive leakage trends by real-time collection of multi-point radiation data and construction of a spatial adjacency matrix, estimation of local spatial gradient and calculation of anomaly degree value.

[0024] (2) This invention constructs a dose gradient field using multi-point radiation data, and through the optimization calculation of the total fitting error value, it can reverse the location of the leakage source and the leakage intensity, and link with the emergency response mechanism to improve the efficiency of handling leakage events.

[0025] (3) In this invention, through edge computing and early warning modules, the system can identify real-time abnormal trends, calculate early warning judgment values, and implement hierarchical early warning and multi-node cross-confirmation measures, which significantly reduces false alarms and missed alarms and improves the reliability of alarms.

[0026] (4) This invention can perform visual analysis, event processing and risk assessment on all data through cloud monitoring and analysis module, and identify leakage patterns through machine learning model, automatically generate risk assessment report, and realize closed-loop management and intelligent support of the whole process.

[0027] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0028] Figure 1 Flowchart of a method for multi-point synchronous detection and identification of radioactive substances;

[0029] Figure 2 This is a structural diagram of a multi-point synchronous detection and identification system for radioactive materials.

[0030] Figure 3 This is a comparison chart of collaborative anomaly assessment values. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figures 1-3This invention provides a technical solution: a method and system for multi-point synchronous detection and identification of radioactive materials, comprising: Step 1, real-time acquisition of multi-point radiation data, data preprocessing, and real-time storage of the multi-point radiation data; Step 2, construction of a spatial adjacency matrix, estimation of local spatial gradient, calculation of anomaly degree value, real-time optimization measures based on the anomaly degree value, and collaborative anomaly assessment of monitoring nodes; Step 3, calculation of total fitting error value based on multi-point radiation data, real-time deduction of leakage source location and leakage intensity based on the total fitting error value, and real-time implementation of corresponding measures; Step 4, collection of multi-point radiation data for real-time anomaly trend identification, calculation of early warning judgment value, and implementation of graded early warning and multi-node cross-confirmation measures; Step 5, receiving and storing various types of data, performing data visualization analysis, and supporting event processing and feedback.

[0033] Specifically, the process of real-time acquisition of multi-point radiation data, data preprocessing, and real-time storage of multi-point radiation data is as follows: Multiple high-sensitivity monitoring nodes are deployed within the radioactive source storage area based on site layout, potential leak points, and ventilation paths. This significantly improves the spatial awareness coverage of trace leak signals and reduces the probability of missed detection. Each monitoring node is equipped with a Geiger counter, a scintillation counter, and a semiconductor detector, and each node is assigned a unique identification code to ensure the time accuracy and equipment traceability of the acquired data. Real-time acquisition of multi-point radiation data includes: radiation dose rate and count rate data of the surrounding environment, as well as the three-dimensional coordinates and equipment parameters of each monitoring node. Moving average filtering is used to denoise the real-time multi-point radiation data of each monitoring node, correct the background dose, remove stable baselines, and perform standardization and normalization. The multi-point radiation data is uploaded to a multi-point radiation database and sent to edge computing devices via a message queue telemetry transmission protocol. This facilitates the stability of the multi-node system and edge decision collaboration. In case of network anomalies, the monitoring nodes automatically cache data and support remote firmware upgrades and periodic calibration.

[0034] This implementation plan features high spatial coverage and multimodal detection capabilities, enabling stable acquisition and intelligent preprocessing of radiation data to improve the accuracy of identifying trace leaks. It also incorporates mechanisms such as edge computing collaboration, network outage caching, and remote maintenance to achieve highly reliable data transmission throughout the entire process and automatic steady-state operation under abnormal scenarios.

[0035] Specifically, the process of constructing a spatial adjacency matrix, estimating local spatial gradients, calculating anomaly severity values, and taking real-time optimization measures based on these anomaly severity values ​​is as follows: A spatial adjacency matrix is ​​constructed based on the three-dimensional coordinates recorded during the deployment of each monitoring node. This matrix can be used to construct a regional topology map, enabling spatial correlation modeling and local clustering analysis. The Euclidean distance between all monitoring nodes is calculated. When the distance between any two monitoring nodes is less than three meters, they are considered neighbors, and the adjacency relationship between the two neighboring monitoring nodes is recorded. This helps to form high-density sensitive areas and improves the accuracy of leak trend location through spatial coupling analysis. Based on the spatial adjacency matrix, the monitoring... The gradient components in the three-dimensional direction are calculated by comparing the radiation dose rate difference between monitoring nodes with the three-dimensional coordinate difference. The gradient estimation method uses a weighted difference approach: the influence of each neighboring node on the monitoring node is weighted inversely proportional to the square of the distance, with closer nodes having larger weights. The gradient results are updated every second based on the current radiation dose rate, and it is determined whether there is a trend of dose increase along a certain direction in a certain area, which helps to achieve early identification of weak but continuous leaks. Historical radiation dose rate data for each monitoring node is subjected to sliding filtering, using a fixed-window moving average method to smooth out short-period fluctuations caused by electronic noise, environmental disturbances, and individual instantaneous events, obtaining the filtered value at the current moment. Wave dose value; Establish a daily background model based on time period for each monitoring node: extract corresponding historical filtered dose values ​​from the same time period of the past three consecutive days, calculate the mean to obtain a stable background mean, which serves as the normal state baseline for the monitoring node; calculate the standard deviation based on the filtered dose values ​​of the same time period of the past three consecutive days to obtain the background standard deviation; obtain the filtered dose value of the i-th monitoring node at time t, and simultaneously obtain the background mean and background standard deviation of the i-th monitoring node at time t; subtract the background mean from the filtered dose value of the i-th monitoring node at time t, then divide by the sum of the background standard deviation and a constant to obtain the standard deviation term; subtract the abnormal threshold weight from the standard deviation term. The factor obtains the anomaly level value of the i-th monitoring node at time t; the anomaly level value is subjected to a maximum function operation: if the anomaly level value is less than zero, the anomaly level value is set to zero; if the anomaly level value is not less than zero, the calculated result is the anomaly level value; when the anomaly level values ​​of multiple consecutive nodes in a certain area increase abnormally, the sliding window is shortened, an on-site verification prompt is automatically sent, and the regional ventilation system is invoked to dilute potential radioactive materials in advance; if a node has a high anomaly level value for a long period of time and neighboring nodes do not respond, the sensor self-diagnosis program is started; it is marked as a "suspected false alarm" state, a remote self-calibration is initiated, historical performance and error trends are automatically analyzed, and the need to replace the equipment is automatically assessed.

[0036] The specific formula for the abnormality level value is as follows:

[0037] ;

[0038] In the formula, The abnormality value represents the degree of abnormality of the radiation dose rate of the i-th monitoring node at time t relative to the stable background mean. Let be the filtered dose value of the i-th monitoring node at time t; Let be the background mean value of the i-th monitoring node at time t; Let be the background standard deviation of the i-th monitoring node at time t; This indicates the maximum function operation. If the anomaly level value is less than zero, the anomaly level value is set to zero. If the anomaly level value is not less than zero, the calculation result is the anomaly level value, ensuring that the anomaly level value is not negative. As an anomaly threshold weight factor, a training set is constructed based on historical radiation dose rate data, featuring filtered dose value, background mean, and background standard deviation. Through enumeration search and optimization algorithms, the anomaly threshold weight factor with the best anomaly detection performance in the training set is selected, ranging from 0.01 to 0.06.

[0039] This implementation plan enables the construction of spatial adjacency matrices for multi-point radiation data, estimation of local spatial gradients, calculation of anomaly values, and execution of real-time optimization measures. It also features functions such as sliding filter processing, establishment of daily background models, calculation of standard deviation terms, anomaly threshold judgment, regional verification prompts, ventilation system invocation, sensor self-diagnosis, and remote self-calibration, thereby improving the accuracy of anomaly identification and the intelligence level of system response.

[0040] Specifically, the specific process of collaborative anomaly evaluation for monitoring nodes is as follows: Obtain the anomaly degree values of the i-th monitoring node and the j-th neighbor node at time t; Obtain the spatial adjacency matrix and screen the neighbor nodes of the spatial adjacency matrix to obtain the neighbor node set of the i-th monitoring node, and obtain the neighbors node number by getting the length of the neighbor node set; Sum the anomaly degree values of all neighbor nodes at time t based on the neighbor node number, and then divide the sum result by the neighbor node number to obtain the neighborhood average anomaly value. Multiply the neighborhood average anomaly value by the neighborhood fusion weight factor and add it to the anomaly degree value of the i-th monitoring node at time t to obtain the collaborative anomaly evaluation value, strengthening the ability to identify group leakage signals; Compare the collaborative anomaly evaluation value with the first-level and second-level anomaly thresholds in real time to form a hierarchical response mechanism, which helps to take different levels of disposal measures in a timely manner. When the collaborative anomaly evaluation value is less than or equal to the first-level anomaly threshold, it is considered that the radiation level is within the normal background fluctuation range, and continue with routine monitoring, maintaining normal equipment and data collection without intervention; When the collaborative anomaly evaluation value is greater than the first-level anomaly threshold and less than or equal to the second-level anomaly threshold, it is considered that there is a trace leakage or equipment anomaly. Upload the collaborative anomaly degree value to the background, and start higher-frequency data collection; Automatically monitor the status of abnormal monitoring nodes and surrounding equipment and environmental parameters. If it is confirmed that the anomaly is valid and in a poorly ventilated and high-risk area, automatically start the ventilation system, issue local sound and light prompts, and send a protection warning instruction to the on-site area; When the collaborative anomaly evaluation value is greater than the second-level anomaly threshold, it is considered that the radiation level is abnormal and there is a leakage risk. Upload the collaborative anomaly degree value to the background, immediately start the on-site emergency plan, and isolate the suspected leakage area; Alarm synchronously remotely and on-site, automatically push anomaly prompts, and achieve multi-platform and multi-role linkage response to improve the efficiency of anomaly information arrival; Use a portable high-precision detector to carry out precise positioning and confirmation of the leakage source; And push cleaning and repair instructions according to the leakage degree, and record the execution progress and feedback at the same time to complete the closed-loop management of the disposal task, which helps with post-event evaluation and continuous optimization of the model.

[0041] Among them, the specific formula for the collaborative anomaly evaluation value is:

[0042] ;

[0043] In the formula, is the collaborative anomaly evaluation value of the i-th monitoring node at time t, which is used to judge whether there may be local radioactive leakage; is the anomaly degree value of the i-th monitoring node at time t; is the neighbor node set of the i-th monitoring node; is the neighbor node number; is the anomaly degree value of the j-th neighbor node at time t; The neighborhood fusion weight factor is calculated by analyzing historical multi-point radiation data, calculating the time-series correlation coefficient of radiation dose rate between each monitoring node and its neighbors, calculating the correlation of the abnormality value sequence of the i-th monitoring node and its j-th neighbor node, setting the neighborhood fusion weight factor as the weighted average of the correlation coefficients between the i-th monitoring node and its neighbors, and using a statistical correlation analysis algorithm to obtain the result. The value range is between 0 and 1.

[0044] The neighborhood fusion weighting factor was set to 0.6. Different anomaly levels were calculated based on the varying radiation dose rates of radioactive materials detected by different monitoring nodes. Furthermore, different monitoring nodes obtained different average neighborhood anomaly values ​​based on their number of neighboring nodes. With the neighborhood fusion weighting factor consistent, the collaborative anomaly assessment value of the monitoring nodes was calculated. Table 1 shows the collaborative anomaly assessment value data.

[0045] Table 1. Data Table of Collaborative Anomaly Assessment Values

[0046] ;

[0047] like Figure 3 The figure shown is a comparison chart of the collaborative anomaly evaluation values ​​provided in this application example. Based on Table 1 and... Figure 3 It can be seen that when the neighborhood fusion weight factor is consistent, different monitoring nodes have different anomaly values, the number of neighboring nodes, and the average anomaly value of the neighborhood. Different collaborative anomaly evaluation values ​​are calculated, as well as the current anomaly value of each monitoring node and the degree of collaborative influence of its neighborhood.

[0048] This implementation plan enables the calculation and hierarchical determination of collaborative anomaly assessment values, and has functions such as neighborhood fusion, spatial consistency judgment, real-time threshold comparison, dynamic sampling frequency adjustment, equipment and environment linkage monitoring, automatic ventilation control, local sound and light prompts, early warning command issuance, emergency plan activation, leakage source confirmation, cleanup and repair command push, and execution closed-loop recording, thereby improving the system's accuracy in identifying abnormal events and its level of intelligent response.

[0049] Specifically, the process of calculating the total fitting error value based on multi-point radiation data and then using this total fitting error value to infer the location and intensity of the leakage source in real time is as follows: The three-dimensional coordinates and corresponding radiation dose rates of each monitoring node are obtained, and abnormal point filtering is performed based on equipment parameters to remove interference data such as fault points and high-noise points, thus improving input quality; based on the three-dimensional coordinates and real-time radiation dose rates of the monitoring nodes, a dose gradient field is constructed through spatial difference to reveal the dose change trend in space, providing directional information for preliminary location; the location of the monitoring node with the highest radiation dose rate is selected as the initial leakage source location based on the dose gradient field; and the location of the leak source is determined based on the selected highest radiation dose rate. The system calculates the radiation dose rate, the distance between the initial leak source location and the current monitoring node, and uses the real-time radiation dose rate of the monitoring node to infer the radiation dose rate at the leak source location as the initial leak intensity. Based on a fusion algorithm of information criterion and spatial clustering, spatial clustering analysis is performed on the anomaly values ​​of the radiation dose rates of all monitoring nodes to distinguish between independent leak events and local disturbance clusters, adapting to multi-source scenarios. By identifying abnormally clustered areas of high radiation dose rates, the number of potential source points is initially estimated. Under different assumed source numbers, leak source inversion and localization calculations are performed using the Bayesian information criterion algorithm, and the number of source points with the smallest Bayesian information criterion value is selected as the current leak intensity. The optimal total number of leakage sources in the scenario is determined by: obtaining the total number of monitoring nodes by acquiring the number of detectors deployed; acquiring the 3D coordinates and measured radiation dose rate of the i-th monitoring node; acquiring the initial leakage source location and initial leakage intensity; calculating the square of the Euclidean distance from the 3D coordinates of the i-th monitoring node to the initial leakage source location and adding a constant 1; then dividing the initial leakage intensity by the sum to obtain the leakage source contribution value; based on the total number of leakage sources, summing the leakage source contribution values ​​of all leakage sources to obtain the estimated dose value of the i-th monitoring node; and subtracting the radiation dose rate of the i-th monitoring node from the estimated dose value of the i-th monitoring node before proceeding with the next step. The squared error term is obtained by squaring the rows. The squared error term measures the deviation between the simulated value and the measured value and is the core objective of optimization convergence. Based on the total number of monitoring nodes, the squared error terms of all monitoring nodes are summed to obtain the initial total fitting error value of K leakage sources. Based on the initial total fitting error value of K leakage sources, the location and leakage intensity of the leakage sources are adjusted using a nonlinear least squares optimization algorithm to gradually reduce the total fitting error value. Through iterative adjustment, the estimation result gradually approaches the actual leakage source parameters to achieve a global optimal approximation until the total fitting error value is less than the error threshold. Then, the location and leakage intensity of the current k-th leakage source are deduced.

[0050] The specific formula for the total fitting error is as follows:

[0051] ;

[0052] In the formula, Let K be the total fitting error values ​​for the K leakage sources, where These are the three-dimensional coordinates of the k-th leakage source. Let be the leakage intensity of the k-th leakage source, used to measure the error between the sum of theoretical radiation doses calculated from all monitoring nodes and the actual measured radiation dose value; This represents the total number of leakage sources. This represents the total number of monitoring nodes. Let be the three-dimensional coordinates of the i-th monitoring node; Let be the radiation dose rate measured at the i-th monitoring node; This is the location of the k-th leakage source; Let be the leakage intensity of the k-th leakage source.

[0053] This implementation scheme realizes the calculation of the total fitting error value based on multi-point radiation data and the real-time back-inference of the leakage source location and leakage intensity. It has functions such as outlier filtering, dose gradient field construction, initial leakage source estimation, spatial cluster analysis, Bayesian information criterion judgment, nonlinear least squares optimization, and error dynamic convergence, thereby improving the leakage source location accuracy and intelligent adaptability in multi-source scenarios.

[0054] Specifically, the process of taking corresponding measures in real time is as follows: The total fitting error value is compared with the error threshold in real time. When the total fitting error value is less than or equal to the error threshold, the leak source location result is considered reliable and can be directly used for rapid response and on-site handling of the leak source, significantly shortening the time delay from source identification to response execution and improving overall handling efficiency. The three-dimensional coordinates of the leak source, the leak intensity, and the total fitting error value are uploaded to the monitoring platform, triggering the leak emergency plan module in real time. This realizes an automatic linkage mechanism driven by the identification results, supporting full-process visual monitoring and closed-loop traceability. Risk levels are automatically assigned, and coordinated implementation of area lockdown and ventilation is carried out. The system provides audible and visual alarms and protection commands, and continuously monitors the leakage evolution process to ensure the continuity and integrity of the entire lifecycle management of leakage events. When the total fitting error value exceeds the error threshold, the leakage source location result is considered unreliable, and a "fitting failure" prompt is output. It also automatically pushes reminders to check data quality and equipment status. The system investigates whether there is abnormal multi-point radiation data and monitoring sensor failure, performs data cleaning and equipment calibration, evaluates the existing monitoring node layout, and recalculates the total fitting error value after adjustment until the total fitting error value meets the valid judgment, realizing an automatic closed-loop correction process and enhancing the system's adaptability to complex scenarios.

[0055] This implementation plan enables real-time comparison and reliability determination of the total fitting error value and the error threshold. It has functions such as uploading leakage source location results, triggering the leakage emergency plan module, automatic risk level allocation, linkage of regional lockdown ventilation audible and visual alarms, dynamic evolution monitoring, fitting failure prompts, data quality and equipment status reminders, data cleaning and equipment calibration, monitoring node layout evaluation and adjustment, and error recalculation, thereby improving the reliability, closed-loop nature and intelligence level of the system's leakage identification and response.

[0056] Specifically, the process of collecting multi-point radiation data for real-time anomaly trend identification and calculating early warning judgment values ​​is as follows: Edge computing devices periodically collect multi-point radiation data transmitted from each monitoring node at high frequency, establish a rolling cache, and store at least one hour of historical multi-point radiation data to ensure the system's ability to perceive anomalies in a timely manner and support dynamic identification and tracking of minute-level changes. The cached multi-point radiation data is strictly synchronized using timestamps to ensure that spatially heterogeneous data is comparable and fusionable within a unified time frame, avoiding the accumulation of misjudgment factors. The radiation dose rate of the i-th monitoring node and its j-th neighbor node at time t is obtained, as are the background mean values ​​of the i-th monitoring node and its j-th neighbor node, and the set of neighbor nodes of the i-th monitoring node is obtained. The system calculates the number of neighboring nodes by obtaining the length of the set of neighboring nodes for the i-th monitoring node; it calculates the difference between the radiation dose rate of the i-th monitoring node and the background mean of the i-th monitoring node to obtain the degree of normal deviation of the monitoring node, reflecting the abnormal amplitude of the current node relative to its historical normal state; it calculates the difference between the radiation dose rate of the j-th neighboring node and the background mean of the j-th neighboring node, and sums all the differences based on the number of neighboring nodes, divides the sum by the number of neighboring nodes, and multiplies it by the neighbor influence weighting factor to obtain the neighbor influence term. The neighbor influence term is added to the degree of normal deviation to obtain the warning judgment value. This comprehensively considers the local offset and the spatial propagation trend, and is a key decision indicator for triggering subsequent warning measures.

[0057] The specific formula for the warning judgment value is as follows:

[0058] ;

[0059] In the formula, Let be the warning judgment value of the i-th monitoring node at time t, which is used to assess the degree of radiation anomaly of a certain monitoring node at the current time; Let be the radiation dose rate of the i-th monitoring node at time t; The background mean of the i-th monitoring node; Let be the set of neighboring nodes of the i-th monitoring node; This represents the number of neighboring nodes; Let be the radiation dose rate of the j-th neighbor node at time t; The background mean of the j-th neighbor node; The neighbor influence weighting factor is set based on the radiation dose rate of the monitoring node in multiple historical time periods, the corresponding background mean, the three-dimensional coordinates of the monitoring node, and whether a leakage has occurred. A series of candidate neighbor influence weighting factors are set. Under each candidate neighbor influence weighting factor, the early warning judgment value is calculated and compared with known leakage events. The optimal neighbor influence weighting factor is obtained through cross-validation statistical analysis, which ranges from 0 to 1.

[0060] This implementation plan achieves high-frequency acquisition and rolling caching of multi-point radiation data, strict timestamp synchronization, extraction of neighbor node sets, calculation of normal deviation degree, calculation and fusion of neighbor influence items, and generation of early warning judgment values. It has the functions of trend recognition, spatial collaborative analysis and real-time early warning triggering, which improves the system's sensitivity to radiation anomaly trends and response timeliness.

[0061] Specifically, the process of implementing tiered early warning and multi-node cross-confirmation measures is as follows: Real-time comparison of the early warning judgment value with the first and second-level early warning thresholds. When the early warning judgment value is less than the first-level early warning threshold, it is determined to be normal, with continuous monitoring and periodic archiving, requiring no intervention. This state minimizes false alarm interference and ensures efficient system operation. When the early warning judgment value is greater than or equal to the first-level early warning threshold but less than the second-level threshold, it is determined to be a slight dose increase, possibly affected by disturbances or initial leakage. Local audible and visual alerts are issued, and the radiation dose rate change rate is recorded in real time. Enhanced monitoring of key areas facilitates rapid response to potential leakage trends. When the early warning judgment value is greater than or equal to the second-level threshold, it is determined to be a significant anomaly, suspected to be a leakage. Upon detecting a leak, the system immediately activates local buzzers and red alarm lights, uploads warning judgment values, automatically dispatches mobile detection equipment to check abnormal areas, and transmits detection data back in real time, initiating leak source location and emergency response procedures. If a single node alarms but neighboring nodes do not respond, remote alarms will be delayed and marked as "pending confirmation," effectively filtering isolated false alarms and preventing frequent false triggering of contingency plans. If the warning judgment values ​​of three or more neighboring nodes simultaneously exceed the warning threshold and are spatially consistent, the system will be upgraded to a "regional leak warning," automatically triggering the leak source inversion and location module to begin location analysis. This achieves a fusion judgment of regional clustering characteristics and spatial consistency, ensuring the system has the ability to quickly perceive and locate large-scale leaks.

[0062] This implementation plan enables real-time comparison of early warning judgment values ​​with Level 1 and Level 2 early warning thresholds. It features continuous monitoring and archiving, local audio-visual alerts, radiation dose rate change recording, enhanced monitoring of key areas, buzzer and red light response, early warning value uploading, mobile detection equipment scheduling, real-time data transmission, leak source location and emergency handling, "pending confirmation" status marking, and regional leak early warning identification and inversion location activation. This improves the hierarchical handling capability and spatial consistency judgment level of early warning response.

[0063] Specifically, the process of receiving and storing various types of data, performing data visualization analysis, and supporting event handling and feedback is as follows: The cloud server receives monitoring node identification codes, radiation dose rates, radiation dose rate change rates, location results, and timestamps in real time, and stores them uniformly in a time-series database to ensure the real-time and completeness of historical trend and event tracing analysis; it also performs visualization analysis based on multi-point radiation data of monitoring nodes to improve users' intuitive understanding of abnormal situations in monitoring node coordination and assist command personnel in accurately grasping the availability of monitoring nodes; it supports multi-level alarms and pushes notifications to relevant personnel through multiple channels such as SMS, email, and WeChat; it supports event handling records and retrospection, automatically archives alarm handling operations, response times, and feedback content, and improves the closed-loop controllability of event management; it introduces machine learning models to intelligently identify and assess leakage patterns, realizes automatic classification and judgment of different types of leaks, and automatically generates risk assessment reports on a regular basis.

[0064] This implementation plan enables unified storage of monitoring node identification codes, radiation dose rates, radiation dose rate change rates, location results, and timestamps. It features functions such as displaying heat maps of leak source locations, visual analysis of suspected leak intensity and location, multi-level alarms, multi-channel push notifications, event handling records and backtracking, intelligent identification of leak modes, risk assessment, and report generation, thereby enhancing data management, visual perception, and intelligent decision-making capabilities.

[0065] Reference Figure 2As shown, the second aspect of this invention provides a multi-point synchronous detection and identification system for radioactive materials, applied to the aforementioned multi-point synchronous detection and identification method for radioactive materials. The system includes: a multi-point synchronous monitoring module, a dose trend collaborative modeling module, a leak source inversion and localization module, an edge computing and early warning module, and a cloud monitoring and analysis module. The multi-point synchronous monitoring module is used to collect multi-point radiation data in real time, perform data preprocessing, and store the multi-point radiation data in real time. The dose trend collaborative modeling module is used to construct a spatial adjacency matrix, estimate local spatial gradients, calculate anomaly severity values, take optimization measures in real time based on the anomaly severity values, and perform collaborative anomaly assessment on monitoring nodes. The leak source inversion and localization module is used to calculate the total fitting error value based on the multi-point radiation data, infer the leak source location and leakage intensity in real time based on the total fitting error value, and take corresponding measures in real time. The edge computing and early warning module is used to collect multi-point radiation data for real-time anomaly trend identification, calculate early warning judgment values, and implement graded early warning and multi-node cross-confirmation measures. The cloud monitoring and analysis module is used to receive and store various types of data, perform data visualization analysis, and support event processing and feedback.

[0066] This implementation plan enables real-time acquisition, preprocessing, and storage of multi-point radiation data. It features functions such as spatial adjacency matrix construction, local spatial gradient estimation, anomaly degree value calculation and collaborative evaluation, total fitting error value inversion of leakage source location and intensity, anomaly trend identification, early warning judgment value calculation and graded early warning, data visualization analysis and event processing feedback, thereby enhancing the system's multi-dimensional perception, intelligent diagnosis, and response linkage capabilities for leakage events.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0068] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for multi-point synchronous detection and identification of radioactive materials, characterized in that, include: Step 1: Collect multi-point radiation data in real time, perform data preprocessing, and store the multi-point radiation data in real time. Step 2: Construct a spatial adjacency matrix, estimate the local spatial gradient, calculate the anomaly degree value, take optimization measures in real time based on the anomaly degree value, and conduct collaborative anomaly assessment on the monitoring nodes; The specific process of constructing the spatial adjacency matrix, estimating the local spatial gradient, calculating the anomaly degree value, and taking optimization measures in real time based on the anomaly degree value is as follows: Based on the three-dimensional coordinates recorded when each monitoring node is deployed, a spatial adjacency matrix is ​​constructed; and the Euclidean distance between all monitoring nodes is calculated. When the distance between any two monitoring nodes is less than three meters, they are determined to be neighbors, and the adjacency relationship between the two neighboring monitoring nodes is recorded. Based on the spatial adjacency matrix, the gradient components in the three-dimensional direction are calculated using the difference in radiation dose rate and the difference in three-dimensional coordinates between monitoring nodes. The gradient estimation method adopts a weighted difference form: the influence of each neighboring node on the monitoring node is set with a weight inversely proportional to the square of the distance, and the closer the distance, the greater the weight. The gradient result is updated once per second based on the current radiation dose rate, and it is determined whether there is a trend of dose increase in a certain direction in a certain area. The historical radiation dose rate data of each monitoring node is subjected to sliding filtering. The fixed window sliding average method is used to smooth out short-period fluctuations caused by electronic noise, environmental disturbances and individual instantaneous events, so as to obtain the filtered dose value at the current moment. Establish a daily background model based on time period for each monitoring node: extract the corresponding historical filtered dose values ​​from the same time period of the past three consecutive days, calculate the average value to obtain a stable background mean value, and use it as the normal state baseline of the monitoring node. The background standard deviation was obtained by calculating the standard deviation of the filtered dose values ​​for the same period over the past three consecutive days. Obtain the filtered dose value of the i-th monitoring node at time t, and simultaneously obtain the background mean and background standard deviation of the i-th monitoring node at time t; subtract the background mean from the filtered dose value of the i-th monitoring node at time t and divide by the sum of the background standard deviation and a constant to obtain the standard deviation term; subtract the anomaly threshold weight factor from the standard deviation term to obtain the anomaly degree value of the i-th monitoring node at time t. Perform a maximum function operation on the anomaly severity value: if the anomaly severity value is less than zero, then set the anomaly severity value to zero; if the anomaly severity value is not less than zero, then the calculated result is the anomaly severity value. When the abnormality level values ​​of multiple nodes in a certain area increase abnormally, the sliding window is shortened, an on-site verification prompt is automatically sent, and the area ventilation system is activated to dilute potential radioactive materials in advance. If a node consistently shows a high abnormality value and neighboring nodes do not respond, the sensor self-diagnosis program will be activated. If the device is marked as a "suspected false alarm", a remote self-calibration will be initiated, which will automatically analyze historical performance and error trends and automatically assess whether to replace the device. Step 3: Calculate the total fitting error value based on the multi-point radiation data, and use the total fitting error value to infer the location and intensity of the leakage source in real time, and take corresponding measures in real time. Step 4: Collect radiation data from multiple points to identify real-time abnormal trends, calculate early warning judgment values, and implement tiered early warning and multi-node cross-confirmation measures. Step 5: Receive and store various types of data, perform data visualization analysis, and support event handling and feedback.

2. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process of real-time acquisition of multi-point radiation data, data preprocessing, and real-time storage of multi-point radiation data is as follows: Based on the site layout, potential leak points, and ventilation paths, multiple high-sensitivity monitoring nodes are deployed in the radioactive source storage area. Each monitoring node is equipped with a Geiger counter, a scintillation counter, and a semiconductor detector, and each monitoring node is assigned a unique identification code. Multi-point radiation data is collected in real time, including radiation dose rate and count rate data of the surrounding environment, as well as the three-dimensional coordinates and equipment parameters of each monitoring node. Moving average filtering is used to denoise the real-time multi-point radiation data of each monitoring node, correct the background dose, remove stable baselines, and perform standardization and normalization. Multi-point radiation data is uploaded to a multi-point radiation database and sent to edge computing devices via a message queue telemetry transmission protocol. If a network anomaly occurs, the monitoring node automatically caches the data and supports remote firmware upgrades and periodic calibration.

3. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process of conducting collaborative anomaly assessment of the monitoring nodes is as follows: Obtain the anomaly level values ​​of the i-th monitoring node and the j-th neighbor node at time t; obtain the spatial adjacency matrix, filter the neighbor nodes in the spatial adjacency matrix to obtain the set of neighbor nodes of the i-th monitoring node, and obtain the length of the set of neighbor nodes to obtain the number of neighbor nodes; The anomaly values ​​of all neighboring nodes at time t are summed based on the number of neighboring nodes. The summation result is then divided by the number of neighboring nodes to obtain the average anomaly value of the neighborhood. The average anomaly value of the neighborhood is multiplied by the neighborhood fusion weight factor and then added to the anomaly value of the i-th monitoring node at time t to obtain the collaborative anomaly assessment value. Real-time comparison of the collaborative anomaly assessment value with the first- and second-level anomaly thresholds. When the collaborative anomaly assessment value is less than or equal to the first-level anomaly threshold, the radiation level is considered to be within the normal background fluctuation range. Routine monitoring continues, and equipment and data are collected normally without intervention. When the collaborative anomaly assessment value is greater than the first-level anomaly threshold and less than or equal to the second-level anomaly threshold, it is considered that there is a trace leak or equipment malfunction. The collaborative anomaly degree value is uploaded to the background, and higher frequency data collection is initiated. The status of anomaly monitoring nodes and surrounding equipment and environmental parameters are automatically monitored. If the anomaly is confirmed to be valid and located in a poorly ventilated or high-risk area, the ventilation system will be automatically activated, a local audible and visual alert will be issued, and a protective warning instruction will be sent to the site area. When the collaborative anomaly assessment value exceeds the level 2 anomaly threshold, the radiation level is considered abnormal, indicating a risk of leakage. The collaborative anomaly degree value is uploaded to the backend, and the on-site emergency plan is immediately activated to isolate the suspected leakage area. Remote and on-site alarms are synchronized, and anomaly alerts are automatically pushed. Portable high-precision detectors are used to accurately locate and confirm the leakage source. Cleanup and repair instructions are pushed according to the degree of leakage, while the execution progress and feedback are recorded.

4. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process of calculating the total fitting error value based on multi-point radiation data, and then using the total fitting error value to infer the location and intensity of the leakage source in real time is as follows: The three-dimensional coordinates and corresponding radiation dose rates of each monitoring node are obtained, and anomaly filtering is performed based on equipment parameters. A dose gradient field is constructed by spatial difference based on the three-dimensional coordinates and real-time radiation dose rates of the monitoring nodes. The location of the monitoring node with the highest radiation dose rate is selected as the initial leak source location based on the dose gradient field. The radiation dose rate of the leak source location is obtained by back-calculating the selected highest radiation dose rate, the distance between the initial leak source location and the current monitoring node, and the real-time radiation dose rate of the monitoring node, as the initial leak intensity. Based on the information criterion and spatial clustering fusion algorithm, spatial clustering analysis is performed on the abnormality values ​​of radiation dose rate of all monitoring nodes. By identifying abnormal clustering areas of high radiation dose rate, the number of potential source points is initially estimated. Under different assumed source numbers, leakage source inversion and localization calculation analysis are performed by combining the Bayesian information criterion algorithm. The number of source points with the smallest Bayesian information criterion value is selected as the optimal total number of leakage sources in the current scenario. The total number of monitoring nodes is obtained by acquiring the number of detectors deployed; the three-dimensional coordinates and measured radiation dose rate of the i-th monitoring node are acquired; the initial leakage source location and initial leakage intensity are acquired; the square of the Euclidean distance from the three-dimensional coordinates of the i-th monitoring node to the initial leakage source location is calculated and a constant 1 is added; the initial leakage intensity is then divided by the summation result to obtain the leakage source contribution value; based on the total number of leakage sources, the leakage source contribution values ​​of all leakage sources are accumulated to obtain the estimated dose value of the i-th monitoring node; the estimated dose value of the i-th monitoring node is subtracted from the radiation dose rate of the i-th monitoring node and then squared to obtain the error square term; based on the total number of monitoring nodes, the error square terms of all monitoring nodes are summed to obtain the initial total fitting error value of K leakage sources; Based on the initial total fitting error value of K leakage sources, the location and leakage intensity of the leakage sources are adjusted using a nonlinear least squares optimization algorithm to gradually reduce the total fitting error value until the total fitting error value is less than the error threshold. Then, the location and leakage intensity of the current k-th leakage source are obtained by reverse calculation.

5. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process of taking corresponding measures in real time is as follows: The system compares the total fitting error value with the error threshold in real time. When the total fitting error value is less than or equal to the error threshold, the leak source location result is considered reliable and can be directly used for rapid response and on-site handling of the leak source. The system uploads the three-dimensional coordinate position of the leak source, the leak intensity, and the total fitting error value to the monitoring platform to trigger the leak emergency plan module in real time. The system automatically assigns risk levels, links and executes area lockdown, ventilation, audible and visual alarms, and protection commands, and continuously monitors the leak evolution process. When the total fitting error value is greater than the error threshold, the leakage source location result is considered unreliable, a "fitting failure" prompt is output, and an automatic push notification is sent to check the data quality and equipment status. Investigate for abnormal multi-point radiation data and monitoring sensor malfunctions; perform data cleaning and equipment calibration; assess the existing monitoring node layout. After adjustment, recalculate the total fitting error value until the total fitting error value meets the valid criteria.

6. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process for collecting multi-point radiation data, identifying real-time abnormal trends, and calculating early warning judgment values ​​is as follows: Edge computing devices periodically collect multi-point radiation data transmitted by each monitoring node at high frequency, establish a rolling cache, and store at least one hour of historical multi-point radiation data; and strictly synchronize the cached multi-point radiation data using timestamps. Get the radiation dose rate of the i-th monitoring node and the j-th neighbor node at time t, get the background mean of the i-th monitoring node and the j-th neighbor node, get the set of neighbor nodes of the i-th monitoring node, and get the length of the set of neighbor nodes of the i-th monitoring node to get the number of neighbor nodes. The degree of normal deviation of the monitoring node is obtained by calculating the difference between the radiation dose rate of the i-th monitoring node and the background mean of the i-th monitoring node. Calculate the difference between the radiation dose rate of the j-th neighbor node and the background mean of the j-th neighbor node. Based on the number of neighbor nodes, sum all the differences. Divide the sum by the number of neighbor nodes and multiply it by the neighbor influence weighting factor to obtain the neighbor influence term. Add the neighbor influence term to the normal deviation degree to obtain the warning judgment value.

7. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process for implementing the tiered early warning and multi-node cross-confirmation measures is as follows: The system compares the warning judgment value with the first and second level warning thresholds in real time. When the warning judgment value is less than the first level warning threshold, it is determined that there is no abnormality. The system is continuously monitored, archived periodically, and no intervention is required. When the warning judgment value is greater than or equal to the first-level warning threshold and less than the second-level threshold, it is judged as a slight increase in dose, which may be affected by disturbance or initial leakage. Local audible and visual alerts are issued, the radiation dose rate change rate is recorded in real time, and monitoring of key areas is strengthened. When the warning judgment value is greater than or equal to the secondary threshold, it is judged as an obvious abnormality and suspected leakage source activity. The local buzzer and red light alarm will be activated immediately, the warning judgment value will be uploaded, the mobile detection equipment will be automatically dispatched to check the abnormal area, and the detection data will be transmitted back in real time. The leakage source location and emergency handling procedures will be initiated. If a single node alarms but neighboring nodes do not respond, the remote alarm will be delayed and marked as "pending confirmation". If the warning judgment values ​​of three or more neighboring nodes are simultaneously greater than the warning threshold and are spatially consistent, the alarm will be upgraded to "regional leakage warning" and the leakage source inversion and location module will be automatically triggered to start the location.

8. The method for multi-point synchronous detection and identification of radioactive materials according to claim 1, characterized in that, The specific process of receiving and storing various types of data, performing data visualization analysis, and supporting event processing and feedback is as follows: The cloud server receives monitoring node identification codes, radiation dose rates, radiation dose rate change rates, location results, and timestamps in real time, and stores them uniformly in a time-series database; and performs visual analysis based on the multi-point radiation data of the monitoring nodes. It supports multi-level alarms and pushes notifications to relevant personnel through multiple channels such as SMS, email, and WeChat. It also supports event handling records and backtracking. Machine learning models are introduced to intelligently identify and assess leakage patterns, and risk assessment reports are automatically generated on a regular basis.

9. A multi-point synchronous detection and identification system for radioactive materials, employing the multi-point synchronous detection and identification method for radioactive materials as described in any one of claims 1-8, characterized in that, include: Multi-point synchronous monitoring module, dose trend collaborative modeling module, leak source inversion and localization module, edge computing and early warning module, and cloud monitoring and analysis module; The multi-point synchronous monitoring module is used to collect multi-point radiation data in real time, perform data preprocessing, and store the multi-point radiation data in real time. The dose trend collaborative modeling module is used to construct a spatial adjacency matrix, estimate the local spatial gradient, calculate the anomaly degree value, take optimization measures in real time based on the anomaly degree value, and perform collaborative anomaly assessment on the monitoring nodes. The leakage source inversion and localization module is used to calculate the total fitting error value based on multi-point radiation data, infer the location and leakage intensity of the leakage source in real time based on the total fitting error value, and take corresponding measures in real time. The edge computing and early warning module is used to collect multi-point radiation data for real-time abnormal trend identification, calculate early warning judgment value, and implement hierarchical early warning and multi-node cross-confirmation measures. The cloud-based monitoring and analysis module is used to receive and store various types of data, perform data visualization analysis, and support event processing and feedback.

Images