A leak detection instrument leak rate prediction method based on function fitting and correlation analysis

Abstract

The application discloses a leak rate prediction method of a leak detector based on function fitting and correlation analysis, relates to the technical field of helium mass spectrum leak detectors, and comprises the following steps: collecting a sequence of leak rates corresponding to a leak hole and time, and performing function fitting processing on the sequence of leak rates and time to obtain a leak rate attenuation characteristic model; collecting a real-time data fitting model corresponding to the leak hole, and performing correlation analysis on the real-time data fitting model and the leak rate attenuation characteristic model corresponding to the real-time data fitting model to determine a correlation analysis result; and finally, predicting the leak rate of the leak hole based on the correlation analysis result to determine a leak rate prediction result, so that the determination time can be greatly shortened within any leak rate range, and the leak detection efficiency is greatly improved.

Description

Technical Field

[0001] This application relates to the field of helium mass spectrometry leak detector technology, specifically to a leak detector leak rate prediction method based on function fitting and correlation analysis. Background Technology

[0002] A helium mass spectrometer leak detector is a highly sensitive detection device used to detect and quantify minute leaks in systems. It utilizes helium as a tracer gas and detects the amount of helium leakage through mass spectrometry. Specifically, helium is injected into the exterior of the object being tested. If a leak exists, the helium enters the object and is extracted by a vacuum pump. The mass spectrometer identifies and measures the leak by analyzing the concentration ratio of helium molecules. This instrument boasts advantages such as high sensitivity, non-destructive testing, speed and efficiency, precise leak location, and compatibility with various leak detection methods. It is widely used in industries with extremely high sealing requirements, such as aerospace, electronics, pharmaceuticals, chemicals, and semiconductors. Currently, in the initial stage of leak detection, helium mass spectrometer leak detectors require evacuating the machine and the tested component to a high vacuum. During this process, the leak rate gradually decreases and tends to stabilize. Because the signal from the helium ions is inherently very small, the industry typically calculates the leak rate by integrating the helium ion flow. This results in a long time to stabilize to the lowest leak rate and requires filtering, which introduces a lag. Summary of the Invention

[0003] The purpose of this application is to provide a leak rate prediction method for leak detectors based on function fitting and correlation analysis, which solves the problems of long time required for existing technologies to stabilize to the lowest leak rate and the need for filtering. This application utilizes the correlation between leak rate curves of the same device under the same conditions after integral filtering, which greatly accelerates the convergence speed and improves the leak detection efficiency.

[0004] This application is achieved through the following technical solution:

[0005] A leak detector leak rate prediction method based on function fitting and correlation analysis includes:

[0006] For any magnitude of leak in the leak detector, the sequence of leak rate versus time is collected, and the sequence of leak rate versus time is subjected to function fitting to obtain a leak rate decay characteristic model.

[0007] Collect real-time data fitting models corresponding to leaks, and perform correlation analysis between the real-time data fitting models and their corresponding leak rate decay characteristic models to determine the correlation analysis results.

[0008] Based on the correlation analysis results, the leakage rate of the leak is predicted, and the leakage rate prediction result is determined; wherein, the leakage rate prediction result includes whether the predicted leakage rate has reached the minimum leakage rate or the predicted leakage rate has not reached the minimum leakage rate.

[0009] In one possible implementation, the leakage rate decay characteristic model is:

[0010] ;

[0011] in, Represents the logarithmic function. Let represent the leakage rate decay characteristic model corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, I represents the total number of orders of magnitude, and t represents the sampling time. This represents the zeroth coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the first coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the second coefficient corresponding to the leakage rate decay characteristic model at the i-th order of magnitude. This represents the third coefficient corresponding to the leakage rate decay feature model for the i-th order of magnitude.

[0012] In one possible implementation, the real-time data fitting model is:

[0013] ;

[0014] Where lg represents the logarithmic function. Let represent the real-time data fitting model corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, I represents the total number of orders of magnitude, and t represents the sampling time. This represents the zeroth coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the first coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the second coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the third coefficient of the real-time data fitting model corresponding to the i-th order of magnitude.

[0015] In one possible implementation, a correlation analysis is performed on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results. This includes: using the Pearson correlation analysis method or the Spearman correlation analysis method to perform a correlation analysis on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results.

[0016] In one possible implementation, based on the correlation analysis results, the leakage rate of the leak is predicted, and the leakage rate prediction result is determined, including:

[0017] If the correlation in the correlation analysis result is greater than a preset correlation threshold and the duration is greater than a preset time threshold, then the leak rate prediction result is determined to be the minimum leak rate; otherwise, the leak rate is determined to be the minimum leak rate.

[0018] In one possible implementation, it also includes:

[0019] If the predicted leakage rate is the minimum leakage rate, then the real-time data fitting model is subjected to fitting index analysis to obtain the fitting index corresponding to the real-time data fitting model.

[0020] Based on the fitting index corresponding to the real-time data fitting model, the leakage rate decay feature model is detected and updated to achieve model calibration.

[0021] In one possible implementation, the real-time data fitting model is subjected to fitting index analysis to obtain the fitting index corresponding to the real-time data fitting model, including:

[0022] Goodness-of-fit analysis and root mean square loss error analysis are performed on the real-time data fitting model to obtain the goodness-of-fit and root mean square loss error values ​​corresponding to the real-time data fitting model, and the goodness-of-fit and root mean square loss error values ​​are used together as fitting index.

[0023] In one possible implementation, the goodness of fit is:

[0024] ;

[0025] in, Indicates the goodness of fit. This indicates that the leakage rate decay feature model corresponding to the i-th order of magnitude is at sampling point t. n The logarithmic value with base 10, This indicates that the real-time data fitting model is at sampling point t. n The above are logarithmic values ​​with base 10, n = 1, 2, ..., N, where N represents the total number of sampling points. express The average value across all sampling points.

[0026] In one possible implementation, the root mean square loss error value is:

[0027] ;

[0028] RMSE represents the root mean square error value.

[0029] In one possible implementation, the leakage rate decay feature model is detected and updated based on the fitting index corresponding to the real-time data fitting model, including:

[0030] If the goodness of fit and / or root mean square loss error value in the fitting index do not meet the preset numerical requirements, the leakage rate decay feature model will be calibrated and updated.

[0031] Compared with the prior art, this application has the following advantages and beneficial effects:

[0032] This application provides a leak rate prediction method for leak detectors based on function fitting and correlation analysis. The method involves collecting a time-dependent sequence of the leak rate corresponding to a leak, performing function fitting on this sequence to obtain a leak rate decay characteristic model, then collecting the real-time data fitting model corresponding to the leak, and performing correlation analysis between the real-time data fitting model and its corresponding leak rate decay characteristic model to determine the correlation analysis results. Finally, based on the correlation analysis results, the leak rate of the leak is predicted, and the leak rate prediction result is determined. This method can significantly shorten the judgment time and greatly improve leak detection efficiency within any leak rate range. Attached Figure Description

[0033] To more clearly illustrate the technical solutions of the exemplary embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0034] Figure 1 A flowchart illustrating a leak detector leak rate prediction method based on function fitting and correlation analysis, provided for an embodiment of this application;

[0035] Figure 2 10 provided for embodiments of this application -13 The convergence plots corresponding to Pa·m³ / s are shown in (a) for the logarithmic leak rate of the test data, (b) for the real-time correlation with the calibration function, and (c) for the real-time leak rate.

[0036] Figure 3 This is a schematic diagram of the correlation matrix of different leakage rate data provided in the embodiments of this application.

[0037] Figure 4The figures provided in this application are experimental results of goodness-of-fit comparison. (a) is a comparison of the goodness-of-fit convergence process for four magnitudes of leakage rate, (b) is a comparison of the goodness-of-fit change process for four magnitudes of leakage rate, and (c) is a comparison of the goodness-of-fit stability for four magnitudes of leakage rate. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this application are only for explaining this application and are not intended to limit this application.

[0039] like Figure 1 As shown in the figure, this application provides a leak detector leak rate prediction method based on function fitting and correlation analysis, including:

[0040] S101. For any type of leak in the leak detector, collect the sequence of leak rate versus time corresponding to the leak, and perform function fitting on the sequence of leak rate versus time to obtain a leak rate decay characteristic model.

[0041] The leak detector described in this application embodiment is officially called a helium mass spectrometer leak detector, which is a highly sensitive detection device used to detect and quantify minute leaks in a system. Leak detectors typically have a sensitivity of 10⁻⁻⁶. 8 Pa·m³ / s, 10⁻ 9 Pa·m³ / s, 10⁻ 10 Pa·m³ / s, 10 -13 The leak detector operates on four scales: Pa·m³ / s. For any scale, the leak rate can be injected and stopped using the corresponding built-in standard leak orifice, while simultaneously acquiring a time-dependent sequence. Then, logarithmic transformation and polynomial fitting can be performed on the time-dependent sequence for each scale to obtain the leak rate decay characteristic model for each scale. This model can then be permanently stored. It is worth noting that for 10... -13 For leaks of Pa·m³ / s, no leak signal injection is required. The leak detector's leak rate will reach the machine's baseline leak rate, which is also the machine's minimum detectable leak rate.

[0042] In the industry, standard leaks are generally used for calibration. These standard leaks are factory-calibrated, and the leak rate values ​​corresponding to them are generally considered correct. As long as the operating conditions of the standard leak, such as temperature and humidity, are met, the leak rate model collected will be correct. Therefore, standard leaks can be used to obtain a leak rate attenuation characteristic model.

[0043] For example, after collecting the time-related sequence of the leakage rate corresponding to the leak, the leakage rate value at each sampling point can be logarithmically transformed first, and then a third-order least squares fit can be performed to obtain the leakage rate decay feature model.

[0044] In one possible implementation, the leakage rate decay characteristic model is:

[0045] ;

[0046] in, Represents the logarithmic function. Let i represent the leakage rate decay feature model corresponding to the i-th order of magnitude, i=1,2,…,I, where I represents the total number of order of magnitude and t represents the sampling time (e.g., leakage rate sampling is performed once every 10ms; if it is the 10th sampling, then t=100ms). This represents the zeroth coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the first coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the second coefficient corresponding to the leakage rate decay characteristic model at the i-th order of magnitude. This represents the third coefficient corresponding to the leakage rate decay feature model for the i-th order of magnitude.

[0047] S102. Collect the real-time data fitting model corresponding to the leak, and perform correlation analysis on the real-time data fitting model and its corresponding leak rate decay characteristic model to determine the correlation analysis results.

[0048] The real-time data fitting model continuously matches the model in the system during the leak detection process, while the standard leak rate decay characteristic model only has a few fixed orders of magnitude. If the leak rate has not yet reached the corresponding order of magnitude, the real-time data fitting model will be matched.

[0049] In one possible implementation, the real-time data fitting model is:

[0050] ;

[0051] Where lg represents the logarithmic function. Let represent the real-time data fitting model corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, I represents the total number of orders of magnitude, and t represents the sampling time. This represents the zeroth coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the first coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the second coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the third coefficient of the real-time data fitting model corresponding to the i-th order of magnitude.

[0052] In one possible implementation, a correlation analysis is performed on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results. This includes: using the Pearson correlation analysis method or the Spearman correlation analysis method to perform a correlation analysis on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results.

[0053] For example, the correlation analysis results obtained using the Pearson correlation analysis method are as follows:

[0054] ;

[0055] in, This indicates the correlation analysis results obtained using the Pearson correlation analysis method. express The average value across all sampling points.

[0056] The correlation analysis results obtained using the Spearman correlation analysis method are as follows:

[0057] ;

[0058] in, This indicates the correlation analysis results obtained using the Spearman correlation analysis method. express and The difference between function values, where N represents the total number of sampling points.

[0059] S103. Based on the correlation analysis results, predict the leakage rate of the leak and determine the leakage rate prediction result; wherein, the leakage rate prediction result includes whether the predicted leakage rate has reached the minimum leakage rate or the predicted leakage rate has not reached the minimum leakage rate.

[0060] Optionally, once data that meets the criteria is obtained, it will be merged into the existing data. To save storage space, the merged data will not be saved in its original form, but will be fitted into a polynomial using the least squares method, storing only the four coefficients of the third-order polynomial.

[0061] If the predicted leak rate does not reach the minimum leak rate, it may be due to machine aging or other reasons that cause the model to fail to fit, or it may be that there is no model of the corresponding magnitude. Therefore, abnormal feedback or prompts can be given.

[0062] In one possible implementation, based on the correlation analysis results, the leakage rate of the leak is predicted, and the leakage rate prediction result is determined, including:

[0063] If the correlation in the correlation analysis result is greater than a preset correlation threshold and the duration is greater than a preset time threshold, then the leak rate prediction result is determined to be the minimum leak rate; otherwise, the leak rate is determined to be the minimum leak rate.

[0064] For example, the correlation threshold can be set to 0.9. If the correlation analysis result is r, then if the duration exceeds the preset time threshold when r is greater than or equal to 0.9, the leak rate prediction result can be determined to be the leak rate prediction, which has reached the minimum leak rate. This can speed up the convergence speed and improve the leak detection efficiency.

[0065] The time threshold is variable and can be adjusted using the corresponding leak rate decay feature model. This allows the leak rate decay feature model to quickly converge to the minimum value for different magnitudes of leaks. For example, the time threshold can be adjusted based on the minimum leak rate corresponding to different magnitudes of leaks; for instance, for a -8 leak, the time from the start of leak detection to the -8 leak rate stabilization is very fast and does not require a long time, so the time threshold can be set to 3 seconds.

[0066] In one possible implementation, it also includes:

[0067] If the predicted leakage rate is the minimum leakage rate, then the real-time data fitting model is subjected to fitting index analysis to obtain the fitting index corresponding to the real-time data fitting model.

[0068] Based on the fitting index corresponding to the real-time data fitting model, the leakage rate decay feature model is detected and updated to achieve model calibration.

[0069] In one possible implementation, the real-time data fitting model is subjected to fitting index analysis to obtain the fitting index corresponding to the real-time data fitting model, including:

[0070] Goodness-of-fit analysis and root mean square loss error analysis are performed on the real-time data fitting model to obtain the goodness-of-fit and root mean square loss error values ​​corresponding to the real-time data fitting model, and the goodness-of-fit and root mean square loss error values ​​are used together as fitting index.

[0071] In one possible implementation, the goodness of fit is:

[0072] ;

[0073] in, Indicates the goodness of fit. This indicates that the leakage rate decay feature model corresponding to the i-th order of magnitude is at sampling point t. nUsing a base-10 logarithmic value, we need to calculate the leakage rate decay feature model for each order of magnitude, and select the leakage rate decay feature model with the highest goodness of fit as the model corresponding to the real-time data fitting model. This indicates that the real-time data fitting model is at sampling point t. n The above are logarithmic values ​​with base 10, n = 1, 2, ..., N, where N represents the total number of sampling points. express The average value across all sampling points.

[0074] In one possible implementation, the root mean square loss error value is:

[0075] ;

[0076] RMSE represents the root mean square error value.

[0077] In one possible implementation, the leakage rate decay feature model is detected and updated based on the fitting index corresponding to the real-time data fitting model, including:

[0078] If the goodness of fit and / or root mean square loss error value in the fitting index do not meet the preset numerical requirements, the leakage rate decay feature model will be calibrated and updated; otherwise, no calibration and update will be performed.

[0079] Optionally, it can be done in R 2 If the error rate is ≤80% and / or the root mean square loss exceeds a preset error threshold, a curve update is triggered to prevent model mismatch caused by machine aging or temperature drift. Correlation reflects the degree of association between real-time data and the model, but goodness of fit reflects the fitting effect between the function fitted to the real-time data and the model function. Therefore, it is possible to have a good correlation but a poor goodness of fit, in which case the leak rate decay feature model needs to be calibrated and updated.

[0080] For example, if the real-time data meets the model update conditions, the built-in model can be updated, and the new model feature model is:

[0081]

[0082] Where lg represents the logarithmic function. Let represent the fusion fitting model corresponding to the i-th order of magnitude, i = 1, 2, ..., I, where I represents the total number of orders of magnitude, j = 1, 2, ..., J, where J represents the order, and t represents the sampling time. This represents the j-th coefficient of the fusion fitting model corresponding to the i-th order of magnitude. This represents the weights for different orders, with values ​​ranging from [0,1]. During the fusion process, different fusion strategies are employed for coefficients of different orders. The constant term reflects the baseline leakage rate and uses a conservative fusion approach. The value is set to 0.8; the linear term reflects the decay rate, and a balanced fusion is used. Taking 0.5, the dynamic characteristics of the attenuation of higher-order terms are analyzed, and adaptive fusion is employed. and Take 0.3;

[0083] like Figure 2 As shown in (a) through (c), the test data was collected using the same leak detector to reach 1.0*10. -13 The data on the leakage rate in Pa·m³ / s shows that, in real-time correlation, the correlation value r exceeds 0.95 at t=100, indicating a strong correlation. The real-time leakage rate change curve represents the data curve displayed in real time after algorithm optimization. In the real-time leakage rate change curve, green indicates that the correlation r has not reached the set condition, while red indicates the minimum leakage rate of the model after reaching the set condition. When the test curve has not yet reached the -13 leakage rate (baseline leakage rate), the function fit has reached 0.9, indicating a strong correlation. To avoid misjudgment, the fit needs to meet a certain duration or other set conditions, and then it is determined that the measurement can reach the minimum leakage rate of the model. Similarly, as long as there is a standard leak, decay curves of various orders of magnitude can be collected, and multiple decay results can be fitted, allowing the machine to converge to the minimum leakage rate of the model more quickly.

[0084] For 1.0*10 -8 Convergence at the Pa·m³ / s level is inherently short. If model prediction is required, the order of the fitting function needs adjustment. Due to its fast convergence speed, the curve is closer to a first or quadratic function; a second-order fit is more suitable. Furthermore, using Pearson correlation to measure the relationship with the calibration curve can accelerate model convergence. For 1.0*10... -9 Pa·m³ / s, 1.0*10 -10 Pa·m³ / s, 1.0*10 -11 Pa·m³ / s and background leakage rate 1.0*10 -13 Because Pa·m³ / s has a relatively long convergence time, fitting the curve with a third-order polynomial and measuring the relationship with the calibration curve using Spearman correlation or Kendall correlation can more accurately predict the convergence value.

[0085] For 1.0*10 -8Convergence at the Pa·m³ / s level is inherently short. If model prediction is required, the order of the fitting function needs adjustment. Due to its fast convergence speed, the curve is closer to a first or quadratic function; a second-order fit is more suitable. Furthermore, using Pearson correlation to measure the relationship with the calibration curve can accelerate model convergence. For 1.0*10... -9 Pa·m³ / s, 1.0*10 -10 Pa·m³ / s, 1.0*10 -11 Pa·m³ / s and background leakage rate 1.0*10 -13 Because Pa·m³ / s has a relatively long convergence time, fitting the curve with a third-order polynomial and measuring the relationship with the calibration curve using Spearman correlation or Kendall correlation can more accurately predict the convergence value.

[0086] In practical use of leak detectors, the leak rate to be measured is unknown. Therefore, it is necessary to fit the test data with the built-in model of each leak detector for calculation. Figure 3 As shown, tests were conducted using test data of various magnitudes. It can be seen that the model at each magnitude has the highest good of fit, thus demonstrating that the technical solution described in this application can be applied in practical use. To further optimize the algorithm, goodness of fit is needed. Since correlation is merely a direct trend of the evaluation function, and it generally decreases for the leak rate curve, relying solely on correlation may not be sufficient in certain scenarios. Goodness of fit more intuitively reflects the correlation between the model and the test data. For example... Figure 3 As shown, it can be seen that for 1.0*10 -8 Pa·m³ / s, 1.0*10 -9 Pa·m³ / s, 1.0*10 -10 Pa·m³ / s and 1.0*10 -13 Data in the Pa·m³ / s range exhibits a strong correlation with the corresponding model. Further analysis allows for a better assessment of whether the data closely matches the model at a given time point. This can be achieved by calculating the ΔR² value (the rate of change of goodness of fit) to determine if the R² value is stabilizing. The model with the largest R² value among all models should be considered. If the ΔR² value approaches 0, it indicates a similar trend to the model. Then, it's determined whether the R² value corresponding to this model is the maximum value within the model. If it is the maximum value, it means the data will converge to the model's minimum value.

[0087] Specific implementation results are as follows: Figure 4 As shown in (a) to (c), the trends of R² and ΔR² values ​​for four different leak rate models as the number of test points increases are illustrated. Figure 4In (a) of the model, the closer the R² value is to 1, the better the model fits the test data; the faster the curve rises, the faster the model converges; the smaller the fluctuation after the curve stabilizes, the more stable the convergence. Therefore, it can be seen that using 1.0*10 -13 Data on the order of Pa·m³ / s were fitted to a value of 1.0*10. -13 The model with a Pa·m³ / s has the best fit and converges very quickly. Figure 4 (b) shows the trend of ΔR² value. The ΔR² value approaching zero is the key condition for judging convergence. It can be seen that in the initial stage, the ΔR² value fluctuates greatly and the model fitting is still being adjusted. As the number of test points increases, the ΔR² value gradually decreases and approaches zero. Figure 4 (c) shows a scatter plot comparing the R² value and its corresponding ΔR² value on a two-dimensional plane. The horizontal axis represents the R² value (range 0-1). The scatter plots move from the left (low R² value) to the right (high R² value). As the R² value increases, the ΔR² value gradually approaches the zero line. The scatter plots within the convergence region are concentrated in areas with higher R² values ​​and ΔR² values ​​close to zero. The convergence points of each model are located at or near the convergence region. However, the convergence criterion is that both the R² value and the ΔR² value must be sufficiently high and sufficiently small. Experiments confirm that only data of the same magnitude and models of the same magnitude can achieve convergence. Based on 1.0*10 -13 Experimental results using data on the order of Pa·m³ / s confirm that the technical solution described in this application reduces convergence time by 78%.

[0088] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0089] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0090] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0092] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0093] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A leak rate prediction method for leak detectors based on function fitting and correlation analysis, characterized in that, include: For any magnitude of leak in the leak detector, the sequence of leak rate versus time is collected, and the sequence of leak rate versus time is subjected to function fitting to obtain a leak rate decay characteristic model. Collect real-time data fitting models corresponding to leaks, and perform correlation analysis between the real-time data fitting models and their corresponding leak rate decay characteristic models to determine the correlation analysis results. During the leak detection process, the real-time data fitting model continuously matches the standard leak rate decay characteristic model in the system. The standard leak rate decay characteristic model only has a few fixed orders of magnitude. If the leak rate has not yet reached the corresponding order of magnitude, the real-time data fitting model is matched. Based on the correlation analysis results, the leakage rate of the leak is predicted, and the leakage rate prediction result is determined; wherein, the leakage rate prediction result includes whether the predicted leakage rate has reached the minimum leakage rate or the predicted leakage rate has not reached the minimum leakage rate. The leakage rate decay characteristic model is as follows: ； in, Represents the logarithmic function. Let represent the leakage rate decay characteristic model corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, I represents the total number of orders of magnitude, and t represents the sampling time. This represents the zeroth coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the first coefficient of the leakage rate decay feature model corresponding to the i-th order of magnitude. This represents the second coefficient corresponding to the leakage rate decay characteristic model at the i-th order of magnitude. This represents the third coefficient corresponding to the leakage rate decay feature model for the i-th order of magnitude. The real-time data fitting model is as follows: ； Where lg represents the logarithmic function. Let represent the real-time data fitting model corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, I represents the total number of orders of magnitude, and t represents the sampling time. This represents the zeroth coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the first coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the second coefficient of the real-time data fitting model corresponding to the i-th order of magnitude. This represents the third coefficient of the real-time data fitting model corresponding to the i-th order of magnitude; Also includes: If the predicted leakage rate is the minimum leakage rate, then the real-time data fitting model is subjected to fitting index analysis to obtain the fitting index corresponding to the real-time data fitting model. This includes: performing goodness-of-fit analysis and root mean square loss error analysis on the real-time data fitting model to obtain the goodness-of-fit and root mean square loss error values ​​corresponding to the real-time data fitting model, and using the goodness-of-fit and root mean square loss error values ​​together as the fitting index. If the goodness of fit and / or root mean square loss error value in the fitting indices do not meet the preset numerical requirements, the leakage rate decay characteristic model will be calibrated and updated as follows: ； Where lg represents the logarithmic function. Let represent the leakage rate decay characteristic model after calibration update corresponding to the i-th order of magnitude, where i = 1, 2, ..., I, and I represents the total number of orders of magnitude. This represents the j-th coefficient corresponding to the leakage rate decay feature model at the i-th order of magnitude. This represents the j-th coefficient of the real-time data fitting model corresponding to the i-th order of magnitude, where j = 0, 1, 2, 3, and t represents the sampling time. This represents the j-th coefficient of the fusion fitting model corresponding to the i-th order of magnitude. This represents the weights of different orders, and the value range is [0,1]. Take 0.8; Take 0.5, and Take 0.

3.

2. The leak rate prediction method for leak detectors based on function fitting and correlation analysis according to claim 1, characterized in that, The correlation analysis is performed on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results. This includes using the Pearson correlation analysis method or the Spearman correlation analysis method to perform correlation analysis on the real-time data fitting model and its corresponding leakage rate decay characteristic model to determine the correlation analysis results.

3. The leak rate prediction method for leak detectors based on function fitting and correlation analysis according to any one of claims 1-2, characterized in that, Based on the correlation analysis results, the leakage rate of the leak is predicted, and the leakage rate prediction result is determined, including: If the correlation in the correlation analysis result is greater than a preset correlation threshold and the duration is greater than a preset time threshold, then the leak rate prediction result is determined to be the minimum leak rate; otherwise, the leak rate is determined to be the minimum leak rate.

4. The leak rate prediction method for leak detectors based on function fitting and correlation analysis according to claim 1, characterized in that, The goodness of fit is: ； in, Indicates the goodness of fit. This indicates that the leakage rate decay feature model corresponding to the i-th order of magnitude is at sampling point t. n The logarithmic value with base 10, This indicates that the real-time data fitting model is at sampling point t. n The above are logarithmic values ​​with base 10, n = 1, 2, ..., N, where N represents the total number of sampling points. express The average value across all sampling points.

5. The leak rate prediction method for leak detectors based on function fitting and correlation analysis according to claim 4, characterized in that, The root mean square loss error value is: ； RMSE represents the root mean square error value.

Images