A health state evaluation and intelligent operation and maintenance method and system for key equipment of a physical protection system

Abstract

The present application relates to a kind of health state evaluation and intelligent operation method of physical protection system key equipment and system thereof.The method includes the following steps:S1, obtain the multi-source operation data of nuclear facility physical protection system key equipment, generate equipment health evaluation core data set.S2, calculate the dynamic weight of each fault type on equipment reliability, and based on dynamic weight and equipment own fault record, calculate equipment dynamic reliability coefficient.S3, based on fault maintenance data, calculate the cumulative performance attenuation coefficient of equipment, and based on equipment dynamic reliability coefficient and cumulative performance attenuation coefficient, calculate equipment actual health state value.S4, predict the remaining life of equipment, output equipment health and life two-dimensional evaluation result.S5, based on equipment health and life two-dimensional evaluation result, the health grade of equipment is divided by preset threshold value determination rule, and the health grade and preventive maintenance priority are quantitatively bound, and the equipment classification result is output to generate classification operation strategy.

Description

Technical Field

[0001] This invention relates to the cross-technical field of physical protection and equipment operation and maintenance management of nuclear facilities, and in particular to a health status assessment and intelligent operation and maintenance method and system for key equipment in a physical protection system. Background Technology

[0002] Physical protection systems are a core component of nuclear facility security, and the stable operation of their key equipment (including servers, switches, cameras, detectors, door magnetic switches, etc.) is fundamental to ensuring the security capabilities of nuclear facilities. Equipment lifecycle management and health status assessment are core technical aspects of intelligent operation and maintenance of physical protection systems. Currently, the operation and maintenance management of nuclear facility physical protection system equipment largely relies on manual records, using corrective and fault-based maintenance methods. However, this approach has the following technical drawbacks: (1) The lack of a quantitative equipment health status evaluation model means that the equipment operating status is judged solely by human experience, which makes it impossible to accurately and objectively assess the equipment health status and guide preventive maintenance work. This results in a high rate of equipment failure and affects the overall reliability of the physical protection system.

[0003] (2) Existing equipment reliability assessment methods simply count the number of failures without considering the severity of the failure levels or the objective decay of reliability after equipment failure repair. The evaluation results deviate greatly from the actual operating status of the equipment.

[0004] (3) The management of equipment life cycle information (basic information, installation information, fault and maintenance records, running time, etc.) is scattered and lacks integrated information management methods. It is impossible to achieve unified collection, storage and analysis of equipment data, resulting in low operation and maintenance efficiency.

[0005] (4) Existing operation and maintenance strategies lack a linkage mechanism with equipment health status and adopt a "one-size-fits-all" maintenance approach, which results in over-maintenance leading to resource waste and under-maintenance leading to equipment failures and defects, and cannot achieve intelligent hierarchical operation and maintenance.

[0006] Among existing related patented technologies, some solutions propose methods for health status assessment or full life cycle management of industrial equipment, but most are aimed at general industrial equipment and do not provide customized designs based on the operational characteristics and security requirements of nuclear facility physical protection system equipment. Some solutions involve calculating equipment failure probability, but do not introduce reliability attenuation correction after failure repair and equipment resilience coefficient, thus failing to accurately reflect the actual health status of key equipment in the nuclear facility physical protection system. Furthermore, existing technologies have not achieved deep integration of equipment health status assessment results with intelligent operation and maintenance strategies, failing to form a closed-loop management system encompassing health assessment, tiered operation and maintenance, intervention feedback, and data mining. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method and system for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system, in order to address the above-mentioned deficiencies.

[0008] The technical solution adopted by this invention to solve its technical problem is: a method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system, comprising the following steps: S1. Obtain multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate a core dataset for equipment health assessment; the multi-source operation and maintenance data includes at least static basic data, dynamic operation data, and fault repair data. S2. Based on the core dataset of equipment health evaluation and big data of similar equipment failures, calculate the dynamic weight of each failure type on equipment reliability, and calculate the dynamic reliability coefficient of the equipment based on the dynamic weight and the equipment's own failure records. S3. Based on the fault repair data, calculate the cumulative performance degradation coefficient of the equipment, and based on the dynamic reliability coefficient of the equipment and the cumulative performance degradation coefficient, calculate the actual health status value of the equipment. S4 inputs the core dataset of equipment health evaluation and the actual health status value of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output a dual-dimensional evaluation result of equipment health and lifespan. S5. Based on the dual-dimensional evaluation results of equipment health and lifespan, the equipment is classified into health levels by a preset threshold judgment rule, and the health level is quantitatively bound to the priority of preventive maintenance. The equipment classification results are output to generate a graded operation and maintenance strategy.

[0009] Further, in step S2, the dynamic reliability coefficient of the computing device includes: The quantitative calculation formula for the dynamic reliability coefficient of the equipment is as follows: in, The dynamic reliability coefficient of the equipment; The dynamic weight of the j-th type of fault for the same type of equipment; This represents the cumulative number of various types of malfunctions occurring in a single device. The total number of units put into operation for the same type of equipment; j=1 indicates a serious fault type; j=2 indicates an important fault type; j=3 indicates a general fault type.

[0010] Furthermore, the range of the dynamic reliability coefficient of the device is as follows: Step S2 also includes: If the calculation result obtained from the quantitative calculation formula of the dynamic reliability coefficient of the equipment If it is less than 0, then take =0%.

[0011] Further, in step S3, the cumulative performance degradation coefficient of the computing device includes: The formula for calculating the cumulative performance degradation coefficient of the device is as follows: in, This is the cumulative performance degradation coefficient of the equipment, and If the equipment has no maintenance records, then ; Let be the single-time performance degradation coefficient after repair of type j fault, and ; The performance recovery rate after repairing various faults, and .

[0012] Further, in step S3, the actual health status value of the computing device includes: The quantitative calculation formula for the actual health status value of the equipment is as follows: in, This represents the actual health status value of the device, and If the calculation result <0, then Set = 0%; The dynamic reliability coefficient of the equipment; This is the cumulative performance degradation coefficient of the equipment; The toughness coefficient of the equipment, and >0, if If it approaches 0, then take... =0.1.

[0013] Furthermore, the core dataset for equipment health evaluation includes a highly correlated core feature set. In step S4, predicting the remaining lifespan of the equipment includes: A training sample set was constructed using a set of highly correlated core features of similar devices and their actual operating life. A lightweight long short-term memory neural network with two hidden layers and 32 neurons was used for model training. The model was then solidified after training. The highly correlated core feature set of the device to be evaluated and the actual health status value of the device are input into the solidified lightweight long short-term memory neural network to predict the remaining lifespan of the device. Specifically, when the actual health status value of the device is lower than the preset failure threshold, the device is directly determined to be faulty, and the remaining lifespan of the device is zero.

[0014] Further, in step S5, the preset threshold determination rule includes a health value threshold and a remaining lifespan threshold. The health level classification jointly determines the actual health status value of the device and the remaining lifespan of the device, wherein: When the actual health status value of the device is greater than or equal to the first health value threshold, and the remaining lifespan of the device is greater than or equal to the first remaining lifespan threshold, it is judged as excellent. When the actual health status value of the device is between the second health value threshold and the first health value threshold, and the remaining lifespan of the device is between the second remaining lifespan threshold and the first remaining lifespan threshold, it is judged to be of good quality. When the actual health status value of the device is between the third health value threshold and the second health value threshold, and the remaining lifespan of the device is between the third remaining lifespan threshold and the second remaining lifespan threshold, it is determined to be intermediate. When the actual health status value of the device is less than the third health value threshold, or when the remaining lifespan of the device is less than the third remaining lifespan threshold, it is judged as poor. The first health value threshold is greater than the second health value threshold, the second health value threshold is greater than the third health value threshold, the first remaining lifespan threshold is greater than the second remaining lifespan threshold, and the second remaining lifespan threshold is greater than the third remaining lifespan threshold.

[0015] Furthermore, in step S5, the preventive maintenance priority is divided into four levels, and the health level is bound to the preventive maintenance priority in reverse one-to-one correspondence; the superior level corresponds to the lowest priority, and the poor level corresponds to the highest priority.

[0016] Further, in step S1, the multi-source operation and maintenance data undergoes quantitative preprocessing, including: The dynamic running data is subjected to time-series alignment quantization, time-domain core feature extraction quantization, outlier cleaning and completion quantization, feature normalization quantization, and high-correlation core feature set screening quantization.

[0017] In addition, the present invention also provides a health status assessment and intelligent operation and maintenance system for key equipment in a physical protection system, comprising: The multi-source data preprocessing module is used to acquire multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate a core dataset for equipment health assessment; the multi-source operation and maintenance data includes at least static basic data, dynamic operation data, and fault repair data. The equipment dynamic reliability calculation module is used to calculate the dynamic weight of each fault type on equipment reliability based on the core dataset of equipment health evaluation and big data of faults of the same type of equipment, and to calculate the equipment dynamic reliability coefficient based on the dynamic weight and the equipment's own fault records. The actual health status calculation module is used to calculate the cumulative performance degradation coefficient of the equipment based on the fault maintenance data, and to calculate the actual health status value of the equipment based on the dynamic reliability coefficient of the equipment and the cumulative performance degradation coefficient. The dual-dimensional quantification module for health and lifespan is used to input the core dataset of equipment health evaluation and the actual health status value of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output the dual-dimensional evaluation results of equipment health and lifespan. The dual-dimensional grading module, based on the dual-dimensional evaluation results of equipment health and lifespan, classifies the equipment into health levels through preset threshold judgment rules, and quantitatively binds the health levels with the priority of preventive maintenance, outputting the equipment grading results to generate a grading operation and maintenance strategy.

[0018] The health status evaluation and intelligent operation and maintenance method and system for key equipment in the physical protection system of this invention have the following beneficial effects: This invention breaks through the limitations of existing technologies that simply count the number of failures without considering the differences in the severity of failure levels and the objective degradation of reliability after repair. It achieves accurate quantitative assessment of the health status of key equipment in the physical protection system of nuclear facilities. At the same time, by quantitatively and strongly binding the equipment health level with the priority of preventive maintenance, a hierarchical operation and maintenance strategy is formed, which solves the problem of over-maintenance or under-maintenance caused by the one-size-fits-all approach of existing operation and maintenance methods. It realizes the transformation from passive fault maintenance to proactive preventive maintenance, and significantly improves the operation and maintenance efficiency and overall reliability of key equipment in the physical protection system. Attached Figure Description

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart illustrating the health status assessment and intelligent operation and maintenance method for key equipment in a physical protection system provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the health status evaluation and intelligent operation and maintenance system for key equipment of the protection system provided in the physical embodiment of this invention. Detailed Implementation

[0020] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the invention will now be described in detail with reference to the accompanying drawings. The terms "first," "second," "third," etc., are used only for the convenience of describing the technical solution and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0022] refer to Figure 1 In a preferred embodiment, this embodiment provides a method for health status assessment and intelligent operation and maintenance of key equipment in a physical protection system, including the following steps: S1. Acquire multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate a core dataset for equipment health assessment. The multi-source operation and maintenance data includes at least static basic data, dynamic operation data, and fault repair data.

[0023] In some embodiments, step S1 performs quantitative preprocessing on multi-source operation and maintenance data, including: time-series alignment quantization of dynamic operation data, extraction and quantization of time-domain core features, outlier cleaning and completion quantization, feature normalization quantization, and screening and quantization of highly correlated core feature sets. This outputs a core dataset for equipment health evaluation. Furthermore, the calculated scalar sequence is bound to the unique equipment identifier ID, achieving spatiotemporal unification of static basic data, dynamic operation data, and fault maintenance data. This step primarily provides a unique standardized input for all subsequent calculations.

[0024] Specifically, step S1 includes the following process: S11. Data Acquisition Definition: Extracting three types of raw data from individual devices from the nuclear power plant's physical protection equipment operation and maintenance system and sensor acquisition system, which are field-acquisitionable scalar / time-series scalar: Static basic data This includes the device's unique identifier (ID) and device type (T) code. (These correspond to physical protection devices such as cameras, microwave Doppler sensors, door magnetic sensors, and servers, respectively) and commissioning time. (Unit / month), Design life (Unit / Month), Equipment Rated Parameters (e.g., detector sensitivity, server CPU rated utilization). Dynamic running data This refers to a time-series scalar. Specifically, it includes the sampling time t and the sampling frequency. Dynamic operating indicators collected by the equipment's own sensors For example, temperature, frame rate, sensitivity, CPU usage, etc.

[0025] Fault Repair Data This refers to the quantitative record of equipment malfunctions and repairs. Fault and repair data. Including fault types (1 = Critical fault, 2 = Important fault, 3 = General fault), Total number of occurrences of each type of fault (j=1,2,3), performance recovery rate after various fault repairs (j=1,2,3). This was determined through on-site evaluation tests. .

[0026] S12, Dynamic Data Spatiotemporal Alignment Quantization: For dynamic running data By fixed time window ( (After on-site calibration) the data is sliced, transforming the time series data into a scalar sequence with equal time intervals, and the statistical characteristics within each time window are calculated: in, The mean of the window index; Number the time windows; The number of sampling points in the k-th window ( ≥3); The variance of the index within the window; For the first Peak values ​​of operating indicators within a time window.

[0027] S13. Temporal core feature extraction and quantization: For the spatiotemporally aligned scalar sequence, global temporal quantization features of the device's operating status are extracted. The specific algorithm formula is as follows: in, This is the global time-domain average operating indicator for the equipment, used to reflect the overall operating level. K represents the total number of time windows. The standard deviation is denoted as . The coefficient of variation is used to eliminate the influence of dimensions. This represents the maximum value of the peak performance metrics across all time windows.

[0028] Output single-device time-domain feature set: S14, Outlier Cleaning and Completion Quantization: The lightweight isolated forest algorithm is used to calculate the outlier scores of dynamic metrics. Outliers are identified using a threshold and then imputed using linear interpolation to ensure data validity. The specific algorithm formula is as follows: in, This is the anomaly detection threshold set on-site, and the default value of 0.7 can be used. =1 indicates that the data is an outlier and needs to be removed. This is the padded value after removing outliers. for The dynamic operating indicators at the next moment. for The dynamic operating indicators of the previous moment.

[0029] S15, Feature Normalization and Quantization: For all time-domain features, Normalization is mapped to [0,1] to eliminate the dimensional differences between different indicators and ensure the consistency of subsequent calculations. The specific algorithm formula is as follows: in, These are the features after normalization. This is the minimum reasonable value for the equipment itself. This indicator represents the maximum reasonable value for the equipment itself, and can be based on the equipment's rated parameters. Based on on-site operational experience, the calibration is applicable to similar equipment. Specific calculation constraints are: if = ,but =0.5.

[0030] S16. Quantification of High-Correlation Core Feature Set Screening: The mutual information method is used to calculate the correlation between each normalized feature and the equipment fault state Y, eliminating redundant features and retaining the core feature set with high correlation that is significant for equipment health assessment. The specific algorithm formula is as follows: in, The joint probability of feature X and fault state Y; Marginal probability (based on big data statistics of faults in similar equipment); This is the correlation threshold calibrated on-site; the default value can be 0.3. This is a set of core features with high correlation. Represents the empty set. Mutual information value. It is an indicator that quantifies the correlation between two random variables, and in this method, it is used to measure the set of equipment operating characteristics. To what extent can changes in mutual information reflect changes in the equipment fault state Y? The closer the mutual information value is to 1, the better. and The stronger the correlation, the more accurately the changes in operating characteristics can reflect the equipment failure patterns; the closer the mutual information value is to 0, the more accurate it is to indicate that the equipment failure pattern is being reflected. and The weaker the correlation, the less obvious the change in operating characteristics is with equipment failure.

[0031] S17. Output Results: Core dataset for equipment health assessment. Specifically: , S2. Based on the core dataset of equipment health assessment and big data on faults of similar equipment, calculate the dynamic weight of each fault type on equipment reliability, and calculate the dynamic reliability coefficient of the equipment based on the dynamic weight and the equipment's own fault records. This step, by determining the equipment's dynamic reliability coefficient, reflects the basic operational reliability of the equipment based on its own fault records and the fault patterns of similar equipment.

[0032] Specifically, step S2 includes the following process: First, define (j=1,2,3...) represents the cumulative number of various types of faults occurring in a single device; This refers to the total number of similar equipment put into operation. This represents the number of times the j-th type of fault occurs on the m-th device of the same type. Let j be the total number of type j failures for the same type of equipment, and ; This represents the average number of months of operation for similar equipment. The total operating time for similar equipment, and .

[0033] S21. Dynamic weights are calculated using the entropy weight method: Based on large datasets of faults in similar equipment, the dynamic weights of different fault types on equipment reliability are calculated. These dynamic weights reflect the actual impact of each fault type. The specific algorithm formula is as follows: in, Let be the fault frequency of the j-th type of fault; The information entropy of the j-th type of fault; Let be the difference coefficient for the j-th type of fault; These are normalization coefficients used to ensure... This method categorizes faults into three types (critical fault type / important fault type / general fault type). In the entropy weight method, the denominator of the normalization coefficient is usually the natural logarithm of the number of evaluation indicators. Since this method uses fault type as the evaluation indicator, it takes... , =0.9102. This represents the dynamic weight for type j faults in the same type of equipment. For ease of calculation, this embodiment amplifies the dynamic weight by a factor of 100. The specific calculation constraints are: Dynamic Weight It can be updated every six months based on field data.

[0034] S22. Calculation of equipment dynamic reliability coefficient: Based on the dynamic weights of the number of equipment failures and the types of failures, a formula for calculating the dynamic reliability coefficient of equipment without redundant parameters is constructed.

[0035] The formula for quantifying the dynamic reliability coefficient of equipment is as follows: in, This represents the dynamic reliability coefficient of the equipment. j=1 indicates a severe failure type, j=2 indicates a significant failure type, and j=3 indicates a general failure type.

[0036] The specific calculation constraints are as follows: If the calculation result If < 0, then take =0%; if the equipment is fault-free ( = ), then take =100%.

[0037] S3. Based on fault repair data, calculate the cumulative performance degradation coefficient of the equipment, and based on the equipment dynamic reliability coefficient and the cumulative performance degradation coefficient, calculate the actual health status value of the equipment. This step reflects the true health status of the equipment after multiple fault repairs through the actual health status value.

[0038] Specifically, step S3 includes the following process: S31. Quantitative calculation of performance degradation coefficient: Based on the performance recovery rate after equipment failure repair, the single-failure performance degradation coefficient and cumulative performance degradation coefficient are calculated to reflect the degree of performance degradation after multiple repairs. The specific algorithm formula is as follows: in, This is the cumulative performance degradation coefficient of the equipment, and If the equipment has no maintenance records, then . Let be the single-time performance degradation coefficient after repair of type j fault, and . The performance recovery rate after repairing various faults, and .

[0039] S32. Calculate the actual health status value of the equipment. The specific algorithm formula is as follows: in, This represents the actual health status value of the device, and If the calculation result <0, then Set to 0%. This represents the dynamic reliability coefficient of the equipment. This is the cumulative performance degradation coefficient of the equipment. The toughness coefficient of the equipment, and >0, if If it approaches 0, then take... =0.1. ≈2.71828, which is the natural constant.

[0040] It is understandable that the equipment resilience coefficient reflects the equipment's ability to withstand failures. In this step, the equipment resilience coefficient can be calculated based on the average number of failures in a large dataset of similar equipment failures and the failure severity coefficient calibrated on-site. The specific algorithm formula is as follows: in, , These represent the average number of occurrences of general and critical faults before a serious fault occurs in the same type of equipment (which can be obtained from field data statistics). , This refers to the fault severity coefficient calibrated on-site (severe faults directly lead to equipment failure and are not included in the toughness calculation). Preferably, Take 0.1, 0.5 is acceptable.

[0041] S4. Input the core dataset for equipment health assessment and the actual health status values ​​of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output a dual-dimensional evaluation result of equipment health and lifespan. Specifically, the core dataset for equipment health assessment includes a set of highly correlated core features. This step inputs the highly correlated core features from the core dataset for equipment health assessment. The actual health status value H obtained from the above calculation is used as the input of a lightweight long short-term memory neural network to achieve quantitative prediction of the remaining lifespan of the equipment.

[0042] It can be understood that the dual-dimensional evaluation result of equipment health and lifespan refers to the quantitative evaluation result {H, Rul} consisting of the actual health status value H of the equipment and the remaining usable lifespan Rul of the equipment. Among them, H reflects the current operational health level of the equipment, and Rul reflects the remaining usable lifespan of the equipment. Together, they constitute the dual-dimensional quantitative evaluation index of equipment health and lifespan.

[0043] In some embodiments, step S4, predicting the remaining lifespan of the device, includes: constructing a training sample set using a set of highly relevant core features of similar devices and their actual operating lifespan; training the model using a lightweight long short-term memory neural network with two hidden layers and 32 neurons; and then solidifying the model after training. The set of highly relevant core features of the device to be evaluated and the actual health status value of the device are input into the solidified lightweight long short-term memory neural network to predict the remaining lifespan of the device. Specifically, when the actual health status value of the device is lower than a preset failure threshold, the device is directly determined to have failed, and the remaining lifespan of the device is set to zero.

[0044] Specifically, the training sample set is: in, The actual operating life of the m-th piece of equipment of the same type, in months.

[0045] Preferably, a preset failure threshold is used. The value is 30%. If... < If so, the device is directly determined to be faulty, and the following steps are taken: =0.

[0046] Final output device remaining lifespan ∈ , Unit: Month.

[0047] S5. Based on the dual-dimensional evaluation results of equipment health and lifespan, the equipment is classified into health levels according to preset threshold judgment rules, and the health level is quantitatively bound to the preventive maintenance priority, outputting the equipment classification result to generate a graded operation and maintenance strategy. This embodiment, by quantitatively binding health level to preventive maintenance priority, provides a clear quantitative basis for generating operation and maintenance strategies, and provides a direct quantitative basis for formulating intelligent and graded preventive maintenance strategies for key equipment in the physical protection system. Of course, in some other embodiments, this step can also classify the equipment health level according to preset threshold judgment rules, and determine the corresponding preventive maintenance priority based on the pre-stored mapping relationship between health level and preventive maintenance priority, thereby outputting the equipment classification result.

[0048] In some embodiments, the preset threshold determination rules are based on big data from on-site maintenance of similar equipment. Preventive maintenance priorities are divided into four levels, with health levels and preventive maintenance priorities linked in reverse order. The highest priority level corresponds to the lowest priority, and the lowest priority level corresponds to the highest priority.

[0049] Specifically, the preset threshold determination rules include a health value threshold and a remaining lifespan threshold. The health level classification jointly determines the actual health status value of the equipment and the remaining lifespan of the equipment, wherein: When the actual health status value of the device is greater than or equal to the first health value threshold, and the remaining lifespan of the device is greater than or equal to the first remaining lifespan threshold, it is judged as excellent.

[0050] When the actual health status value of the equipment is between the second health value threshold and the first health value threshold, and the remaining lifespan of the equipment is between the second remaining lifespan threshold and the first remaining lifespan threshold, it is judged to be of good quality.

[0051] When the actual health status value of the device is between the third health value threshold and the second health value threshold, and the remaining lifespan of the device is between the third remaining lifespan threshold and the second remaining lifespan threshold, it is judged as intermediate.

[0052] When the actual health status value of the device is less than the third health value threshold, or the remaining lifespan of the device is less than the third remaining lifespan threshold, it is judged as poor.

[0053] The first health threshold is greater than the second health threshold, the second health threshold is greater than the third health threshold, the first remaining lifespan threshold is greater than the second remaining lifespan threshold, and the second remaining lifespan threshold is greater than the third remaining lifespan threshold.

[0054] For example, the first health value threshold 90% is acceptable, the second health threshold. 70% is acceptable, the third health value threshold. A value of 50% can be used. The first remaining lifetime threshold can be 12 months, and the second remaining lifetime threshold... A 6-month period is acceptable, which is the third remaining lifespan threshold. A 3-month grace period is acceptable. Preventive maintenance priorities are: Level IV (lowest), Level III, Level II, and Level I (highest).

[0055] Finally, the equipment classification results are output. It can be understood that the equipment classification results refer to the results derived from the health level. Prioritizing preventive maintenance The resulting binary quantification classification .

[0056] This embodiment overcomes the limitations of existing technologies that simply count the number of failures without considering the differences in the severity of failure levels and the objective degradation of reliability after repair. It achieves a precise quantitative assessment of the health status of key equipment in the physical protection system of nuclear facilities. At the same time, by quantitatively and strongly binding the equipment health level with the priority of preventive maintenance, a hierarchical operation and maintenance strategy is formed, which solves the problem of over-maintenance or under-maintenance caused by the one-size-fits-all approach of existing operation and maintenance methods. It realizes the transformation from passive fault maintenance to proactive preventive maintenance, and significantly improves the operation and maintenance efficiency and overall reliability of key equipment in the physical protection system.

[0057] The following detailed description of the specific implementation of the present invention is based on actual operating data from two typical key devices (cameras and servers) in the physical protection system of nuclear power plants.

[0058] This embodiment is implemented based on the technical solution of the present invention, and provides a detailed calculation process and judgment results. Those skilled in the art can reproduce the technical solution of the present invention based on this embodiment. The protection scope of the present invention is not limited to the following embodiment.

[0059] The following is a health assessment and classification of the cameras (equipment type T=1) of the physical protection system of nuclear power plants.

[0060] Basic Equipment Information: Equipment Unique Identifier ID=DSI001TV, Commissioning Time =42 months, design life =60 months, camera video frame rate 25fps, rated operating temperature -10℃~50℃; fault repair data : =0 (no serious fault) =1 (1 major fault, lens malfunction) =2 (2 general faults, network disconnection); Post-repair performance recovery rate =0.92、 =0.96; Big data on similar devices: =60 units, =45 months, =60×45=2700; Total number of failures of the same type of equipment =0、 =32、 =68; Statistics on the toughness coefficient of similar equipment =9、 =4.

[0061] After quantitative preprocessing including spatiotemporal alignment, feature extraction, anomaly cleaning, normalization, and mutual information filtering, a set of core features with high device correlation was obtained: Mutual information value =0.45> =0.3, the filter is effective.

[0062] Output the core dataset for device health assessment: D={DSI001TV,1,{42,60,25},{0.88,0.03,0.95,0.04},{0,1,2,0.92,0.96}}.

[0063] Calculate the failure frequency : =32 / (0+32+68)=0.32 Calculate single-type information entropy (k≈0.9102): =0 Calculate the coefficient of variation for a single type : Calculate dynamic weights Dynamic weights Magnified 100 times, retaining 2 decimal places: = =41.09% = =27.51% = =31.40% Weight verification: 41.09% + 27.51% + 31.40% = 100%, the calculation is valid.

[0064] Calculate the dynamic reliability coefficient of the equipment: =96.66% The quantification of the actual health status and remaining lifespan of equipment includes the following calculation steps: Calculate the single-cycle performance degradation coefficient : In this embodiment, This indicates that the corresponding device is fault-free.

[0065] Calculate the cumulative performance degradation coefficient : =0.0533 Calculate the equipment toughness coefficient: Calculate the actual health status value of the device: =96.66% 0.9818 = 94.9% Quantitative prediction of remaining equipment lifespan: After correction of the lightweight model input, a core feature set with H=94.90% and high correlation was obtained. ={0.88,0.03,0.95,0.04}, because the actual health status value of the equipment is much higher than the preset failure threshold. =30%, no risk of equipment failure. Based on the operating patterns of similar equipment, the remaining lifespan is predicted. =17 months.

[0066] Therefore, the dual-dimensional evaluation result of the DSI001TV device's health and lifespan is 17 months.

[0067] Equipment health and lifespan are quantitatively graded in two dimensions: (1) Hierarchical threshold matching Actual equipment health status value: H≈94.90%≥ =90%; Remaining lifespan: 17 months.

[0068] (2) Grading results Health rating (excellent), preventive maintenance priority IV (lowest).

[0069] Traditional solutions calculate equipment reliability coefficients by manually setting fixed weights for fault types (e.g., 0.6 for severe faults, 0.3 for important faults, and 0.1 for general faults) and combining this with the number of faults, without dynamically assigning weights based on big data analysis of faults in similar equipment. Drawbacks include: manually set fixed weights are highly subjective and cannot reflect the actual fault patterns of physical protection equipment in nuclear power plants. For example, network disconnection faults of cameras are frequent among similar equipment, and their actual impact deviates from the manually set weights; once weights are set, they cannot be updated based on on-site fault data, resulting in poor adaptability and a large deviation between the calculated reliability coefficient and the actual operating state of the equipment. In contrast, this invention uses the entropy weight method to calculate dynamic weights based on big data analysis of faults in similar equipment. This method objectively reflects the actual impact of fault types based entirely on data and can be updated every 6 months based on on-site data, adapting to changes in the fault patterns of physical protection equipment in nuclear power plants. For example, in this embodiment, important and general faults of cameras are each weighted 50% based on big data analysis, which is more closely aligned with actual on-site conditions.

[0070] Traditional solutions directly use the equipment reliability coefficient as the health status value, without incorporating a performance degradation model after fault repair and an equipment resilience coefficient, thus ignoring the performance loss after multiple repairs. Drawbacks include: nuclear power plant physical protection equipment cannot be restored to a brand-new state after fault repair; performance degradation is an objective reality. Ignoring performance degradation leads to an overestimation of the actual equipment health status, resulting in flawed maintenance strategies. Furthermore, the lack of a resilience coefficient makes it impossible to differentiate the fault tolerance capabilities of different equipment, resulting in a lack of specificity in the evaluation model. In contrast to traditional solutions, this invention constructs a degradation quantification model based on performance recovery rate, combining it with an equipment resilience coefficient to correct the reliability coefficient, accurately reflecting the true state of the equipment after multiple repairs and avoiding evaluation bias.

[0071] Traditional solutions classify equipment based solely on a single indicator: health value or remaining lifespan. For example, a health value ≥80% is considered excellent, and 60%-80% is considered good, failing to achieve a two-dimensional collaborative assessment. A drawback is that single-dimensional classification cannot comprehensively reflect the equipment's status. For instance, a camera with a health value of 95% but only two months of remaining lifespan might be classified as excellent and subject to low-priority maintenance strategies if only the health value is considered excellent. This could lead to sudden equipment failures impacting security, and since physical protection equipment is a core component of security in nuclear power plants, this deficiency could pose serious safety hazards. In contrast to traditional solutions, this invention employs a dual-dimensional quantitative classification of health and lifespan, combined with logical rules for precise judgment. The four health levels are strongly linked to four maintenance priorities, fully adapting to the high security requirements of nuclear facilities and avoiding the risks of overlooking aspects of single-dimensional classification.

[0072] This invention achieves the optimal balance between nuclear facility adaptability, quantitative accuracy, lightweight deployment, and intelligent hierarchical management. It effectively solves the core problems of subjective evaluation, blind maintenance, and low efficiency in the operation and maintenance of physical protection systems in nuclear power plants, and provides a practical technical means for the intelligent operation and maintenance of physical protection systems for nuclear facilities.

[0073] refer to Figure 2 In another preferred embodiment, the health status evaluation and intelligent operation and maintenance device for key equipment of the physical protection system in this embodiment includes: The multi-source data preprocessing module is used to acquire multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate a core dataset for equipment health assessment. The multi-source operation and maintenance data includes at least static basic data, dynamic operational data, and fault repair data.

[0074] The equipment dynamic reliability calculation module is used to calculate the dynamic weight of each fault type on equipment reliability based on the core dataset of equipment health evaluation and big data of faults of similar equipment, and to calculate the dynamic reliability coefficient of the equipment based on the dynamic weight and the equipment's own fault records.

[0075] The actual health status calculation module is used to calculate the cumulative performance degradation coefficient of the equipment based on fault maintenance data, and to calculate the actual health status value of the equipment based on the dynamic reliability coefficient and the cumulative performance degradation coefficient.

[0076] The dual-dimensional quantification module for health and lifespan is used to input the core dataset of equipment health evaluation and the actual health status value of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output the dual-dimensional evaluation results of equipment health and lifespan.

[0077] The dual-dimensional grading module is used to classify the health level of equipment based on the evaluation results of equipment health and lifespan, and to quantitatively bind the health level with the priority of preventive maintenance, outputting the equipment grading results to generate a grading operation and maintenance strategy.

[0078] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0079] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, or any other form of storage medium known in the art.

[0080] It is understood that the above embodiments only illustrate preferred embodiments of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can freely combine the above technical features without departing from the concept of the present invention, and can also make several modifications and improvements, all of which fall within the protection scope of the present invention. Therefore, all equivalent transformations and modifications made with respect to the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for health status assessment and intelligent operation and maintenance of key equipment in a physical protection system, characterized in that, Includes the following steps: S1. Obtain multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate core dataset for equipment health assessment. The multi-source operation and maintenance data includes at least static basic data, dynamic operation data, and fault repair data; S2. Based on the core dataset of equipment health evaluation and big data of similar equipment failures, calculate the dynamic weight of each failure type on equipment reliability, and calculate the dynamic reliability coefficient of the equipment based on the dynamic weight and the equipment's own failure records. S3. Based on the fault repair data, calculate the cumulative performance degradation coefficient of the equipment, and based on the dynamic reliability coefficient of the equipment and the cumulative performance degradation coefficient, calculate the actual health status value of the equipment. S4 inputs the core dataset of equipment health evaluation and the actual health status value of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output a dual-dimensional evaluation result of equipment health and lifespan. S5. Based on the dual-dimensional evaluation results of equipment health and lifespan, the equipment is classified into health levels by a preset threshold judgment rule, and the health level is quantitatively bound to the priority of preventive maintenance. The equipment classification results are output to generate a graded operation and maintenance strategy.

2. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S2, the dynamic reliability coefficient of the computing device includes: The quantitative calculation formula for the dynamic reliability coefficient of the equipment is as follows: in, The dynamic reliability coefficient of the equipment; The dynamic weight of the j-th type of fault for the same type of equipment; This represents the cumulative number of various types of malfunctions occurring in a single device. The total number of units put into operation for the same type of equipment; j=1 indicates a serious fault type; j=2 indicates an important fault type; j=3 indicates a general fault type.

3. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 2, characterized in that, The range of values ​​for the dynamic reliability coefficient of the device is as follows: ; Step S2 also includes: If the calculation result obtained from the quantitative calculation formula of the dynamic reliability coefficient of the equipment If it is less than 0, then take =0%.

4. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S3, the cumulative performance degradation coefficient of the computing device includes: The formula for calculating the cumulative performance degradation coefficient of the device is as follows: in, This is the cumulative performance degradation coefficient of the equipment, and If the equipment has no maintenance records, then ; Let be the single-time performance degradation coefficient after repair of type j fault, and ; The performance recovery rate after repairing various faults, and .

5. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S3, the actual health status value of the computing device includes: The quantitative calculation formula for the actual health status value of the equipment is as follows: in, This represents the actual health status value of the device, and If the calculation result <0, then Set = 0%; The dynamic reliability coefficient of the equipment; This is the cumulative performance degradation coefficient of the equipment; The toughness coefficient of the equipment, and >0, if If it approaches 0, then take... =0.

1.

6. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, The core dataset for equipment health assessment includes a highly correlated core feature set. In step S4, predicting the remaining lifespan of the equipment includes: A training sample set was constructed using a set of highly correlated core features of similar devices and their actual operating life. A lightweight long short-term memory neural network with two hidden layers and 32 neurons was used for model training. The model was then solidified after training. The highly correlated core feature set of the device to be evaluated and the actual health status value of the device are input into the solidified lightweight long short-term memory neural network to predict the remaining lifespan of the device. Specifically, when the actual health status value of the device is lower than the preset failure threshold, the device is directly determined to be faulty, and the remaining lifespan of the device is zero.

7. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S5, the preset threshold determination rule includes a health value threshold and a remaining lifespan threshold. The health level classification jointly determines the actual health status value of the device and the remaining lifespan of the device, wherein: When the actual health status value of the device is greater than or equal to the first health value threshold, and the remaining lifespan of the device is greater than or equal to the first remaining lifespan threshold, it is judged as excellent. When the actual health status value of the device is between the second health value threshold and the first health value threshold, and the remaining lifespan of the device is between the second remaining lifespan threshold and the first remaining lifespan threshold, it is judged to be of good quality. When the actual health status value of the device is between the third health value threshold and the second health value threshold, and the remaining lifespan of the device is between the third remaining lifespan threshold and the second remaining lifespan threshold, it is determined to be intermediate. When the actual health status value of the device is less than the third health value threshold, or when the remaining lifespan of the device is less than the third remaining lifespan threshold, it is judged as poor. The first health value threshold is greater than the second health value threshold, the second health value threshold is greater than the third health value threshold, the first remaining lifespan threshold is greater than the second remaining lifespan threshold, and the second remaining lifespan threshold is greater than the third remaining lifespan threshold.

8. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S5, the preventive maintenance priority is divided into four levels, and the health level is bound to the preventive maintenance priority in reverse one-to-one correspondence; the superior level corresponds to the lowest priority, and the poor level corresponds to the highest priority.

9. The method for health status evaluation and intelligent operation and maintenance of key equipment in a physical protection system according to claim 1, characterized in that, In step S1, the multi-source operation and maintenance data undergoes quantitative preprocessing, including: The dynamic running data is subjected to time-series alignment quantization, time-domain core feature extraction quantization, outlier cleaning and completion quantization, feature normalization quantization, and high-correlation core feature set screening quantization.

10. A health status assessment and intelligent operation and maintenance system for key equipment in a physical protection system, characterized in that, include: The multi-source data preprocessing module is used to acquire multi-source operation and maintenance data of key equipment in the physical protection system of nuclear facilities, perform quantitative preprocessing on the multi-source operation and maintenance data, and generate a core dataset for equipment health assessment. The multi-source operation and maintenance data includes at least static basic data, dynamic operation data, and fault repair data; The equipment dynamic reliability calculation module is used to calculate the dynamic weight of each fault type on equipment reliability based on the core dataset of equipment health evaluation and big data of faults of the same type of equipment, and to calculate the equipment dynamic reliability coefficient based on the dynamic weight and the equipment's own fault records. The actual health status calculation module is used to calculate the cumulative performance degradation coefficient of the equipment based on the fault maintenance data, and to calculate the actual health status value of the equipment based on the dynamic reliability coefficient of the equipment and the cumulative performance degradation coefficient. The dual-dimensional quantification module for health and lifespan is used to input the core dataset of equipment health evaluation and the actual health status value of the equipment into a lightweight long short-term memory neural network to predict the remaining lifespan of the equipment and output the dual-dimensional evaluation results of equipment health and lifespan. The dual-dimensional grading module, based on the dual-dimensional evaluation results of equipment health and lifespan, classifies the equipment into health levels through preset threshold judgment rules, and quantitatively binds the health levels with the priority of preventive maintenance, outputting the equipment grading results to generate a grading operation and maintenance strategy.

Images