Operating data-based nuclear power sensor reliability assessment method and storage medium

Abstract

The present invention relates to an operating data-based nuclear power sensor reliability assessment method and a storage medium. The method comprises the following steps: S10, acquiring operating data of a plurality of sensors used for measuring the same parameter; S20, on the basis of the operating data, calculating a measurement deviation value corresponding to each sensor; S30, extracting points at a preset time interval from amongst the measurement deviation values, so as to determine corresponding second data to be fitted y and a time sampling serial number x within an operating cycle, wherein x=0, 1, 2, 3,..., n; S40, on the basis of said second data y and the time sampling serial number x, determining a cumulative failure distribution model of each sensor; and S50, on the basis of the cumulative failure distribution model, obtaining a reliability model corresponding to each sensor. Implementing the present invention can realize assessment of performance degradations and reliability of sensors on the basis of historical operating data of nuclear power sensors.

Description

Nuclear power sensor reliability evaluation method and storage medium based on operation data Technical Field

[0001] The present invention relates to the technical field of nuclear power sensors, and in particular to a nuclear power sensor reliability assessment method based on operating data and a storage medium. Background Art

[0002] During the daily operation of nuclear power plants, instrumentation and control systems rely on sensors to collect information on various physical quantities (such as temperature, pressure, flow, and liquid level) to determine the unit's operating status and implement appropriate process control and protection actions. Therefore, sensor reliability has a crucial impact on reactor safety. If a sensor fails and the measurement data is inaccurate, it could lead to incorrect control and protection actions by the control system or power plant operators, resulting in serious consequences.

[0003] A sensor is a device or apparatus that senses a specified measurand and converts it into a usable signal according to a specific pattern. It typically consists of a sensitive element and a transducer element. Drift refers to the unintended change in a sensor's measurement accuracy over a specific time interval, independent of changes in the measured input. Sensor drift is a functional degradation or degradation, the rate of which depends on many factors, including but not limited to aging of the measuring diaphragm and electronic components, or changes in the resistance of the transducer element. This can cause the sensor's accuracy to decrease over time, leading to measurement errors and potential hazards in the system.

[0004] Nuclear power sensors are highly reliable and long-life devices with a very low probability of failure. Drift is the most common failure mode. Nuclear power plant control systems typically rely on regular calibration and replacement to ensure sensor performance and reliability. Another approach involves using redundant data-based sensor monitoring, which uses cross-comparison to assess sensor failure or degradation.

[0005] However, there is currently no mature method to conduct reliability assessment of field sensors.

[0006] Traditionally, a key method for evaluating equipment reliability involves testing a batch of samples, obtaining product failure data through extensive life or accelerated life testing. However, for high-reliability, long-life products, performance degradation data is often used for evaluation. By recording product degradation data at different test times and building degradation trajectory models, product reliability can be analyzed.

[0007] However, degradation tests are generally destructive tests. Due to the high price of important sensors in nuclear power plants, it is difficult to invest resources to carry them out. In addition, the accuracy, reliability and lifespan of sensors are getting higher and higher with technological advancements. Under limited time constraints, this method can only obtain a small amount of failure life information or even no failure life information. Insufficient test conditions have led to the inability to conduct reliability assessments of important nuclear power sensors. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a nuclear power sensor reliability assessment method and storage medium based on operation data.

[0009] The technical solution adopted by the present invention to solve the technical problem is: a nuclear power sensor reliability assessment method based on operation data, the method comprising the following steps:

[0010] S10, acquiring operating data of multiple sensors for measuring the same parameter;

[0011] S20, calculating a measurement deviation value corresponding to each sensor according to the operating data;

[0012] S30, sampling the measured deviation value at a preset time interval to determine the corresponding second data y to be fitted and the time sampling sequence number x within the operation cycle, where x=0, 1, 2, 3, ..., n;

[0013] S40, determining a cumulative failure distribution model for each sensor according to the second to-be-fitted data y and the time sampling sequence number x;

[0014] S50: Obtain a reliability model corresponding to the sensor according to the cumulative failure distribution model.

[0015] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S30 includes:

[0016] S301, sampling the measured deviation value at a preset time interval to obtain first data y1 to be fitted and a time sampling sequence number x within an operation cycle, where x=0, 1, 2, 3, ..., n;

[0017] S302 : Determine, for each sensor, an optimal fitting regular function that meets a first preset goodness of fit from a plurality of preset fitting regular functions according to the first data to be fitted y1 and the time sampling sequence number x.

[0018] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S302 includes:

[0019] S3021. Determine corresponding coefficients of each preset fitting regular function according to the first to-be-fitted data y1 and the time sampling sequence number x, thereby obtaining a plurality of fitting regular functions and a goodness of fit corresponding to each of the fitting regular functions;

[0020] S3022: Compare the goodness of fit corresponding to each of the fitting regular functions with the first preset goodness of fit to screen out the best fitting regular function.

[0021] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S40 includes:

[0022] S401, determining the corresponding second data to be fitted y and the time sampling sequence number x according to the best fitting regular function of each sensor;

[0023] S402 : Determine, for each sensor, an optimal cumulative failure distribution model that meets a second preset goodness of fit from a plurality of preset cumulative failure distribution models according to the second data to be fitted y and the time sampling sequence number x.

[0024] Furthermore, in the nuclear power sensor reliability assessment method based on operation data of the present invention, step S40 further includes:

[0025] S403: When there is more than one cumulative failure distribution model that meets the second preset goodness of fit, and one of the cumulative failure distribution models is an exponential distribution model, the exponential distribution model is used as the optimal cumulative failure distribution model.

[0026] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S402 includes:

[0027] S4021. Determine the failure rate of each preset cumulative failure distribution model based on the second to-be-fitted data y and the time sampling sequence number x, thereby obtaining a plurality of cumulative failure distribution models and a goodness of fit corresponding to each of the cumulative failure distribution models;

[0028] S4022: Compare the goodness of fit corresponding to each of the cumulative failure distribution models with the second preset goodness of fit to screen and obtain the optimal cumulative failure distribution model.

[0029] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, before step S20 and after step S10, the method further includes:

[0030] S12. Eliminate outliers in the operating data based on the box plot.

[0031] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S10 includes:

[0032] Obtaining operating data of a plurality of sensors for measuring the same parameter over a preset number of operating cycles;

[0033] Step S20 includes:

[0034] The operation data of each operation cycle are processed based on preset rules to obtain the measurement deviation value corresponding to each sensor in each operation cycle.

[0035] Furthermore, in the nuclear power sensor reliability assessment method based on operating data of the present invention, step S20 includes:

[0036] Taking the average value of the operating data of other sensors in the same operating cycle as the first sensor as a reference value, performing subtraction processing on the operating data of the first sensor and the reference value to obtain a measurement deviation value corresponding to each operating cycle of the first sensor; or

[0037] Taking an average value of the operating data of all sensors in the same operating cycle as a reference value, performing subtraction processing on the operating data of the first sensor and the reference value to obtain a measurement deviation value corresponding to each operating cycle of the first sensor;

[0038] Step S30 includes:

[0039] Sampling the measured deviation value of each operation cycle at a preset time interval to determine the second data to be fitted y and the time sampling sequence number x corresponding to each operation cycle;

[0040] Step S40 includes:

[0041] A cumulative failure distribution model corresponding to each operating cycle is determined for each sensor according to the second to-be-fitted data y corresponding to each operating cycle and the time sampling sequence number x.

[0042] In addition, the present invention also provides a computer-readable storage medium, which stores a computer program, and the computer program is suitable for loading by a processor to execute the steps of the nuclear power sensor reliability assessment method based on operating data as described above.

[0043] The implementation of the nuclear power sensor reliability assessment method and storage medium based on operating data of the present invention has the following beneficial effects: the present invention can implement performance degradation and reliability assessment of the sensor based on the historical operating data of the nuclear power sensor. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0045] FIG1 is a flow chart of a method for evaluating the reliability of a nuclear power sensor based on operating data provided by an embodiment of the present invention;

[0046] 2 is a flow chart of a method for evaluating the reliability of nuclear power sensors based on operating data according to an embodiment of the present invention;

[0047] 3 is a flow chart of a method for evaluating the reliability of nuclear power sensors based on operating data according to an embodiment of the present invention;

[0048] 4 is a flow chart of a method for evaluating the reliability of nuclear power sensors based on operating data according to an embodiment of the present invention;

[0049] 5 is a flow chart of a method for evaluating the reliability of nuclear power sensors based on operating data provided by an embodiment of the present invention;

[0050] 6 is a flow chart of a method for evaluating the reliability of nuclear power sensors based on operating data according to an embodiment of the present invention;

[0051] FIG7 is a box plot provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.

[0053] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0054] In a preferred embodiment, referring to FIG1 , the nuclear power sensor reliability assessment method based on operation data of this embodiment includes the following steps:

[0055] S10, obtaining operating data of multiple sensors for measuring the same parameter. Preferably, the operating data of multiple sensors for measuring the same parameter over a preset number of operating cycles are obtained.

[0056] Specifically, critical sensors in nuclear power plants are typically used to measure physical quantities such as pressure, temperature, flow, and water level. Due to the plant's redundancy design, two or more sensors are typically used to simultaneously measure the same parameter. Therefore, multiple sensors measuring the same set of parameters are considered a group, and as much historical operating data as possible is collected from this group of sensors to serve as an analytical basis for evaluating their performance and reliability.

[0057] That is to say, the sensors applicable to the present invention are preferably redundant sensors, that is, for the same physical quantity, there are two or more relatively independent sensors, so that the data measured by them are also redundant data. For example, for the same physical quantity, relatively independent data can be obtained by measuring two or more sensors of the same model or models from different manufacturers for verification.

[0058] S20. Calculate a measurement deviation value corresponding to each sensor based on the operating data. That is, compare the deviations of the measurement data from the same group of sensors to obtain a measurement deviation value corresponding to each sensor. Specifically, process the operating data for each operating cycle based on preset rules to obtain a measurement deviation value corresponding to each sensor in each operating cycle.

[0059] Alternatively, there are two ways to determine the measurement deviation value: The average value of the operating data of other sensors in the same operating cycle as the first sensor is used as a reference value, and the operating data of the first sensor is subtracted from the reference value to obtain the measurement deviation value of the first sensor corresponding to each operating cycle. Alternatively, the average value of the operating data of all sensors in the same operating cycle is used as a reference value, and the operating data of the first sensor is subtracted from the reference value to obtain the measurement deviation value of the first sensor corresponding to each operating cycle.

[0060] S30 , sampling the measurement deviation value at a preset time interval to determine the corresponding second data y to be fitted and the time sampling sequence number x within the operation cycle, where x=0, 1, 2, 3, . . . , n.

[0061] Specifically, the measurement deviation value of each operation cycle is sampled at preset time intervals to determine the second data to be fitted y and the time sampling sequence number x corresponding to each operation cycle.

[0062] For example, if a point is taken every 5 minutes for the measured deviation value, then a set of sampled data columns can be determined by taking a point every 5 minutes during the operation cycle. That is, the first deviation value is taken in the first 5 minutes, then x=1, y=the first deviation value. The second deviation value is taken in the second 5 minutes, then x=2, y=the second deviation value, ..., and so on. Of course, a point can also be taken every 1 minute, or every 3 minutes. In other words, the present invention does not limit the specific time interval for taking points, and it can be set according to actual needs.

[0063] S40 : Determine a cumulative failure distribution model for each sensor according to the second to-be-fitted data y and the time sampling sequence number x.

[0064] Specifically, a cumulative failure distribution model corresponding to each operating cycle is determined for each sensor according to the second to-be-fitted data y and the time sampling sequence number x corresponding to the operating cycle.

[0065] S50: Obtain a reliability model corresponding to the sensor according to the cumulative failure distribution model.

[0066] Specifically, obtaining the reliability model corresponding to the sensor based on the cumulative failure distribution model includes but is not limited to the following two methods:

[0067] The first method is to derive a reliability model for each sensor in each operating cycle based on the cumulative failure distribution model corresponding to each operating cycle. In other words, since each sensor has a corresponding failure rate in each operating cycle, each sensor has a corresponding reliability model in each operating cycle. For example, if there are four sensors and data is collected for three operating cycles, each sensor will have three cumulative failure distribution models, meaning three failure rates. Consequently, each sensor will have three reliability models, bringing the total number of reliability models for the four sensors to 12.

[0068] The second method treats multiple sensors in each operating cycle as a whole, or a sensor group. Specifically, the cumulative failure distribution models of each sensor in the sensor group during the same operating cycle are multiplied to obtain the cumulative failure distribution model for the sensor group in each operating cycle. A corresponding reliability model is then derived based on the cumulative failure distribution model for the sensor group in each operating cycle. For example, if data is collected for three operating cycles and there are four sensors in the sensor group, three reliability models will be derived for the sensor group.

[0069] In this embodiment, degradation modeling is carried out based on the historical and operating data of the sensor to evaluate the reliability status of the sensor after any period of time on-site at the nuclear power plant. By analyzing the operating characteristics of the sensor under conditions of drift or degradation, sensor reliability assessment based on operating data is achieved, which helps nuclear power plants to conduct online status monitoring and real-time risk assessment of sensors.

[0070] In some embodiments of the nuclear power sensor reliability assessment method based on operating data, when the deviation value is unstable or there is disturbance, in order to make the failure rate and goodness of fit characterizing the sensor performance obtained based on the operating data more accurate and closer to the actual situation, it is preferred to perform regular fitting on the operating data.

[0071] Specifically, referring to FIG2 , step S30 includes:

[0072] S301 , sampling the measurement deviation value at a preset time interval to obtain first data to be fitted y1 and a time sampling sequence number x within an operation cycle, where x=0, 1, 2, 3, ..., n.

[0073] S302 : Determine, for each sensor, the best fitting regular function that meets a first preset goodness of fit from a plurality of preset fitting regular functions according to the first data to be fitted y1 and the time sampling sequence number x.

[0074] Specifically, referring to FIG3 , step S302 includes:

[0075] S3021 : Determine corresponding coefficients of each preset fitting regular function according to the first data to be fitted y1 and the time sampling sequence number x, thereby obtaining multiple fitting regular functions and the goodness of fit corresponding to each fitting regular function.

[0076] S3022: Compare the goodness of fit corresponding to each fitting regular function with a first preset goodness of fit to screen out the best fitting regular function.

[0077] Based on the above embodiment, referring to FIG2 , step S40 includes:

[0078] S401: Determine the corresponding second data y to be fitted and the time sampling sequence number x according to the best fitting regular function of each sensor. Specifically, the corresponding second data y to be fitted and the time sampling sequence number x are obtained by sampling points at preset time intervals for the best fitting regular function.

[0079] Alternatively, since the constant term does not change over time and does not affect sensor drift degradation, when the determined best-fitting regular function contains a constant term, the obtained best-fitting regular function can be subjected to secondary processing, including but not limited to removing the constant term, thereby forming a new best-fitting regular function. Based on this new best-fitting regular function, the corresponding second data to be fitted y and time sampling sequence x are determined. Of course, when considering positive and negative characteristics, the best-fitting regular function can be further transformed, such as removing the common negative sign in the coefficients, to form a new best-fitting regular function to represent the regularity of the sensor operating data.

[0080] S402 : Determine, for each sensor, an optimal cumulative failure distribution model that meets a second preset goodness of fit from a plurality of preset cumulative failure distribution models according to the second data to be fitted y and the time sampling sequence number x.

[0081] Specifically, referring to FIG3 , step S402 includes:

[0082] S4021 : Determine the failure rate of each preset cumulative failure distribution model according to the second to-be-fitted data y and the time sampling sequence number x, thereby obtaining multiple cumulative failure distribution models and the goodness of fit corresponding to each cumulative failure distribution model.

[0083] S4022: Compare the goodness of fit corresponding to each cumulative failure distribution model with a second preset goodness of fit to screen out the best cumulative failure distribution model.

[0084] Specifically, the operating data within each operating cycle is first fitted with a regularity. A goodness-of-fit test is then performed to identify the optimal regularity function for the time-varying deviation data. This optimal regularity function is then used as a secondary data benchmark to sample data at equal intervals, forming a new data column. Next, a model hypothesis test is performed based on the cumulative failure distribution model commonly used in electronic products to obtain the optimal failure probability model, also known as the optimal cumulative failure distribution model. By analyzing the degradation patterns of this group of sensors and establishing a failure probability model, the reliability of the sensors under current or historical refueling cycles can be calculated and evaluated.

[0085] Optionally, referring to FIG4 , step S40 further includes:

[0086] S403: When there is more than one cumulative failure distribution model that meets the second preset goodness of fit, and one of the cumulative failure distribution models is an exponential distribution model, the exponential distribution model is used as the optimal cumulative failure distribution model.

[0087] In this embodiment, compared with the traditional accelerated life test method, the reliability assessment is no longer based on the failure life information of the sensor. Instead, when the test is difficult to carry out, only the historical operating data of the important sensors of the nuclear power plant can be used to evaluate the performance degradation of the sensor. By calculating the reliability value of the sensor at different times in any working cycle based on the collected operating data of the sensor, the online real-time assessment of the sensor reliability can be achieved. For nuclear power units of different design types, as long as the sensor has sufficient operating data, this method can be used to carry out the assessment.

[0088] It should be noted that the goodness of fit (also known as fitting accuracy) in the present invention refers to the degree of fit of the regression curve to the observed values. The statistic used to measure goodness of fit is the coefficient of determination (R²). The maximum value of R² is 1. As a general rule, the closer the R² value is to 1, the better the fit of the regression curve to the observed values; conversely, the smaller the R² value, the worse the fit of the regression curve to the observed values. The closer the goodness of fit is to 1, the better the fit. Generally speaking, a goodness of fit of 0.5-0.7 or above is considered a good fit, close to the actual situation. In other words, after calculating the failure rate through fitting, the lower the failure rate, the higher the reliability of the sensor.

[0089] For the present invention, the optimal cumulative failure distribution model is not necessarily the one with the highest goodness of fit. This is typically determined based on a combination of the object's physical characteristics. Generally, the exponential distribution model is preferred as the optimal cumulative failure distribution model. For example, if multiple preset cumulative failure distribution models meet a goodness of fit of 0.6, and the exponential distribution model achieves a goodness of fit of 0.63, while other cumulative failure distribution models achieve a goodness of fit of 0.7 or higher, the exponential distribution model is generally preferred as the optimal cumulative failure distribution model.

[0090] In some embodiments of the nuclear power sensor reliability assessment method based on operating data, referring to FIG5 , before step S20 and after step S10, the method further includes:

[0091] S12. Eliminate outliers in the operating data based on the box plot. In other words, pre-process the operating data to eliminate obvious abnormal data.

[0092] Generally speaking, outliers significantly impact data regression fitting results. Therefore, before performing degradation analysis (and therefore determining the cumulative failure distribution model), it is important to first remove significantly outliers from the measured data. Ideally, a boxplot can be used for data preprocessing, which is particularly effective when the actual data distribution is unknown.

[0093] Referring to Figure 7, the specific steps for the box plot are as follows:

[0094] ① For each set of data, find the first quantile Q1, median M, and third quantile Q3;

[0095] ② Set the inner distance iqr=K*Q3-Q1 and calculate the upper and lower thresholds respectively:

[0096] Lower threshold = Q1-iqr

[0097] Upper threshold = Q3 + iqr

[0098] Wherein, K is the coefficient of the box line, and the value of K is determined according to the characteristics of different data sources. In this embodiment, the coefficient K can be preferably set to 3; data exceeding the upper and lower thresholds is considered abnormal data.

[0099] In this embodiment, before the degradation analysis, the obtained measurement source data are subjected to removal of obviously abnormal data, thereby reducing the influence of the abnormal data on the fitting result.

[0100] In a specific embodiment, referring to FIG6 , a water level sensor of a refueling water tank in a nuclear power plant is taken as an example. The refueling water tank has four redundant water level sensors, each in m, with a measurement range of 0.5 to 16.0 m. The water level is usually around 15.5 m during power plant operation.

[0101] (1) Data preprocessing

[0102] The four water level sensors are grouped together to export historical operating data for the corresponding sensor points in the power plant. In this example, three refueling cycles, or three operating cycles, representing approximately 4.5 years of operating data, are exported, with data recorded at one measuring point every 5 minutes.

[0103] As equipment runs longer, sensors may drift. Raw sensor data is segmented and organized according to different refueling cycles, and outliers are removed using a boxplot. The boxplot coefficients are determined based on the characteristics of the data source; in this example, a coefficient of 3 is chosen. Data exceeding the upper and lower thresholds is considered abnormal.

[0104] (2) Deviation value calculation

[0105] Analyze the characteristics and overall trends of each set of data recorded by each sensor to determine whether the sensor has experienced performance degradation. One method for determining performance degradation is to analyze the deviation of each sensor's measured data from the reference value to see if there is any cumulative change over time.

[0106] Since sensors used for pressure, flow, and water level measurement are not designed with other parameters or indicators for online monitoring of sensor performance degradation, they can only be determined through instrument calibration during equipment installation or maintenance. (Calibration involves using maintenance tools to provide the sensor with analog input signals over several application ranges and checking the deviation between the instrument output and the ideal output.) Therefore, further investigation is conducted by measuring the deviation of sensor measurement values.

[0107] Specifically, the pre-processed measurement data is calculated and analyzed to obtain the deviation value. Depending on the number of redundant sensors, the following different deviation value calculation methods can be used:

[0108] ① When the number of the aforementioned redundant sensors is two, the specific steps are: select one sensor from this group of sensors, use the measurement value of the other sensor as the reference value of the sensor, or use the average of the measurement values ​​of the two sensors as the reference value; subtract the reference value from the measurement value of the sensor to obtain the measurement deviation value of the sensor.

[0109] ② If there are three or more redundant sensors, the average value of the sensor measurement values ​​​​outside the selected sensor object is used as the reference value, or the average value of the measurement values ​​​​of all sensors in this group is used as the reference value; subtract the reference value from the measurement value of the sensor to obtain the measurement deviation value of the sensor.

[0110] (3) Data pattern fitting

[0111] That is, regular function fitting is performed to select the best fitting regular function. Specifically, the measurement deviation values ​​of the four redundant sensors in the same group of refueling tank water level sensors are fitted with the general regularity of the data according to the periodic characteristics reflected in each operating cycle. For example, multiple preset fitting regular functions can be used, including but not limited to fitting of first-order functions, second-order functions, exponential functions, Gaussian functions, and other forms. Some fitting formulas are expressed as follows:

[0112]

[0113] In this embodiment, the overall fitting accuracy of the second-order function obtained by the refueling tank water level sensor is the highest. Therefore, the second-order function is used to regularly fit the data of the entire working cycle of the sensor, and the fitting expression is:

[0114]

[0115] Where A, B, and C are coefficients, and x represents the corresponding time sampling sequence number or the number of time points in each operation cycle (that is, every 5 minutes is defined as 1 point), with values ​​of 0, 1, 2, 3, ..., n.

[0116] In order to obtain the common drift degradation law of the sensor itself, the fitting accuracy of the data function used in the analysis is required to be no less than a certain standard. In this embodiment, a fitting goodness of fit of no less than 70% is selected to obtain the second-order fitting function of the deviation value of each refueling tank water level sensor in each operating cycle through fitting.

[0117] (4) Secondary data processing

[0118] In the second-order function formed by fitting the sensor operation data, it has been found through analysis that: in each working cycle, the constant term C does not change with time, so the first-order and second-order variables in the second-order function may affect the parameter drift.

[0119] From the fitting results, coefficient A and coefficient B are both negative. When considering the positive and negative, let A = -a, B = -b, then formula (1) is transformed into:

[0120]

[0121] Therefore, Represents the degradation drift of sensor parameters. Assuming that in each working cycle, the cumulative failure function of the four redundant sensors obeys the single-parameter exponential distribution, the two-parameter Weibull distribution, the normal distribution, and the log-normal distribution, the fitting function of the corresponding operating cycle is respectively The points are taken at preset time intervals. In this embodiment, a point can be taken every 5 minutes.

[0122] Then the fitting and hypothesis testing of the above four types of distributions are carried out.

[0123] (5) Fitting and calculation

[0124] After fitting multiple preset cumulative failure distribution models including but not limited to single-parameter exponential distribution, two-parameter Weibull distribution, normal distribution and lognormal distribution, according to R 2 The hypothesis test shows that all four distributions meet the requirements. From a practical application perspective, since the failure rate of the exponential distribution has only one constant λ, the exponential distribution is preferred as the cumulative failure distribution model for parameter drift in each operating cycle.

[0125] If the integer i (1, 2, 3, …) represents the number of working cycles, then the cumulative failure distribution of the parameter drift of each water level sensor in the water tank during each working cycle can be expressed as:

[0126]

[0127] Its reliability model is:

[0128]

[0129] Where λ is the failure rate; x is the time sampling number, unit: hour; j = 1, 2, 3, ..., represents the sensor number, sensor 1, sensor 2, sensor 3, sensor 4, ....

[0130] Alternatively, multiple sensors or all sensors in each operating cycle may be taken as a whole, that is, a sensor group. Then, the reliability model of the sensor group in each operating cycle of this embodiment is:

[0131]

[0132] In this embodiment, the failure rate calculation results under different material replacement cycles are shown in Table 1. From the data in the table, it can be seen that in the first cycle of the sensor operation, the failure rate The order of magnitude is basically in the E-7 power, and the failure rate in the second working cycle is The order of magnitude of begins to increase to E-6. It can be seen that as the sensor operation time accumulates, the failure rate of the sensor degradation also increases.

[0133] Table 1 Calculation results of index distribution of water level sensor in refueling tank

[0134]

[0135] It should be noted that once the data to be fitted and its mean as well as the fitted data are determined, the coefficient of determination R² is also determined. For details, please refer to the existing formula and will not be repeated here.

[0136] In this embodiment, when life tests are difficult to carry out, historical and operating data of important sensors in the nuclear power plant are used to analyze the data change patterns to obtain a relationship model between parameter degradation and reliability, thereby realizing reliability assessment under sensor degradation or degradation. The embodiment can also be widely used in commercial nuclear power plants or test reactors to help power plant equipment managers or researchers understand and analyze the reliability of highly reliable and long-life sensors, and contribute to the online status monitoring and real-time risk assessment of nuclear power plants.

[0137] It should be noted that for some sensors, if the fitting accuracy of multiple forms of data functions is not high enough, or the fitting accuracy varies greatly in different periods, this means that based on the analysis of the existing data, the overall degradation trend of these sensors does not meet the characteristics of stable degradation. The changes and fluctuations in the data source of the measurement object far exceed the degradation trend of the sensor itself. The regularity of the degradation data cannot be analyzed through data analysis methods, so the next step of reliability assessment cannot be carried out.

[0138] Therefore, for sensors in this type of situation, an analog sensor monitoring method based on redundant data is adopted, that is, a cross-threshold comparison method is used to monitor whether the function of the power plant on-site sensors is degraded or failed.

[0139] Specifically, the average value of the redundant sensor measurements or the external test results are used as a reference value. For details, please refer to the existing technology. For example, assuming that the reference value used for verification is R and the output value of the verified instrument is V, then the deviation B of the verified instrument is:

[0140] B=VR

[0141] For cross-comparison verification between n redundant instruments, when the verified instrument is the jth instrument, the reference value is the arithmetic mean of the outputs of the other n-1 instruments except the jth instrument:

[0142]

[0143] The measurement deviation of the verified sensor is:

[0144]

[0145] The deviation B is compared with a pre-set criterion (a different value standard is established for different sensor devices). If the deviation is greater than the criterion (threshold), it indicates that there may be an abnormality in the sensor measurement.

[0146] In another preferred embodiment, the computer-readable storage medium of this embodiment stores a computer program, and the computer program is suitable for loading by a processor to execute the steps of the above-mentioned nuclear power sensor reliability assessment method based on operating data.

[0147] This embodiment can evaluate the performance degradation and reliability of the sensor based on the historical operating data of the nuclear power sensor.

[0148] The computer-readable storage medium of the present invention may be any computer-readable storage medium that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk.

[0149] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0150] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0151] It is understandable that the above embodiments only express the preferred implementation modes of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the patent scope of the present invention. It should be pointed out that for ordinary technicians in this field, without departing from the concept of the present invention, the above technical features can be freely combined, and several deformations and improvements can be made, all of which fall within the scope of protection of the present invention. Therefore, all equivalent changes and modifications made to the scope of the claims of the present invention should fall within the scope of coverage of the claims of the present invention.

Claims

1. A nuclear power sensor reliability assessment method based on operation data, characterized in that: The method comprises the following steps: S10, obtaining operation data of multiple sensors for measuring the same parameter; S20, calculating a measurement deviation value corresponding to each sensor according to the operating data; S30, sampling the measured deviation value at a preset time interval to determine the corresponding second data y to be fitted and the time sampling sequence number x within the operation cycle, where x=0, 1, 2, 3, ..., n; S40, determining a cumulative failure distribution model of each sensor according to the second to-be-fitted data y and the time sampling sequence number x; S50: Obtain a reliability model corresponding to the sensor according to the cumulative failure distribution model.

2. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 1 is characterized in that: Step S30 includes: S301, sampling the measured deviation value at a preset time interval to obtain the first data y1 to be fitted and the time sampling sequence number x within the operation cycle, where x=0, 1, 2, 3, ..., n; S302 , determining the best fitting regular function that meets a first preset goodness of fit for each sensor from a plurality of preset fitting regular functions according to the first data to be fitted y1 and the time sampling sequence number x.

3. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 2 is characterized in that: Step S302 includes: S3021, determining the corresponding coefficient of each preset fitting regular function according to the first to-be-fitted data y1 and the time sampling sequence number x, thereby obtaining a plurality of fitting regular functions and the goodness of fit corresponding to each of the fitting regular functions; S3022: Compare the goodness of fit corresponding to each of the fitting regular functions with the first preset goodness of fit to screen out the best fitting regular function.

4. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 2 is characterized in that: Step S40 includes: S401, determining the corresponding second to-be-fitted data y and the time sampling sequence number x according to the best fitting regular function of each sensor respectively; S402: Determine, for each sensor, an optimal cumulative failure distribution model that meets a second preset goodness of fit from a plurality of preset cumulative failure distribution models according to the second data to be fitted y and the time sampling sequence number x.

5. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 4 is characterized in that: Step S40 also includes: S403: When there is more than one cumulative failure distribution model that meets the second preset goodness of fit, and one of the cumulative failure distribution models is an exponential distribution model, the exponential distribution model is used as the optimal cumulative failure distribution model.

6. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 4 is characterized in that: Step S402 includes: S4021, determining the failure rate of each preset cumulative failure distribution model according to the second to-be-fitted data y and the time sampling sequence number x, thereby obtaining a plurality of cumulative failure distribution models and a goodness of fit corresponding to each of the cumulative failure distribution models; S4022: Compare the goodness of fit corresponding to each of the cumulative failure distribution models with the second preset goodness of fit to screen out the optimal cumulative failure distribution model.

7. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 1, characterized in that: Before step S20 and after step S10, the method further includes: S12. Eliminate outliers in the operating data based on the box plot.

8. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 1, characterized in that: Step S10 includes: Obtaining operating data of a plurality of sensors for measuring the same parameter over a preset number of operating cycles; Step S20 includes: The operation data of each operation cycle are processed respectively based on preset rules to obtain the measurement deviation value corresponding to each sensor in each operation cycle.

9. The method for evaluating the reliability of nuclear power sensors based on operating data according to claim 8, characterized in that: Step S20 includes: Taking the average value of the operating data of other sensors in the same operating cycle as the first sensor as a reference value, performing a difference process between the operating data of the first sensor and the reference value to obtain a measurement deviation value corresponding to each operating cycle of the first sensor; or Taking the average value of the operating data of all sensors in the same operating cycle as a reference value, performing a difference process between the operating data of the first sensor and the reference value to obtain a measurement deviation value corresponding to each operating cycle of the first sensor; Step S30 includes: The measured deviation value of each operation cycle is sampled at preset time intervals to determine the second to-be-fitted data y and the time sampling sequence number x corresponding to each operation cycle; Step S40 includes: A cumulative failure distribution model corresponding to each operating cycle is determined for each sensor according to the second to-be-fitted data y corresponding to each operating cycle and the time sampling sequence number x.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which is suitable for being loaded by a processor to execute the steps of the nuclear power sensor reliability assessment method based on operating data as described in any one of claims 1 to 9.

Images