Operation fault detection method and system for HIAF synchronous timing system

Abstract

The invention relates to the technical field of electronic digital data processing, in particular to an operation fault detection method and system for an HIAF synchronous timing system. The method comprises the following steps: segmenting waveform data to obtain periodic signals, and calculating an initial Euclidean distance between any two periodic signals; calculating a correction coefficient between any two periodic signals; and multiplying the correction coefficient by the initial Euclidean distance to obtain an optimal Euclidean distance, and carrying out LOF anomaly detection by taking the optimal Euclidean distance as a measure. The method and the device have the effect of improving the operation detection precision of the synchronous timing system.

Description

Technical Field

[0001] The present application relates to the technical field of electronic digital data processing, and in particular to a method and system for detecting operation faults of a HIAF synchronous timing system. Background Art

[0002] The Heavy Ion Accelerator Facility (HIAF) uses accelerator cascade to achieve high beam power and high current. The operation of the cascade requires high-precision timing control of the equipment. The high-precision synchronization timing system is the key to achieving precise control of the equipment and determines the accuracy and precision of the hardware equipment operation. However, during the operation of the HIAF system, due to factors such as equipment aging, electromagnetic interference, and environmental changes, the synchronization timing system may experience clock drift, phase error, signal distortion and other faults, resulting in beam loss, experimental data deviation and even equipment damage. Traditional fault detection methods mainly rely on manual inspections and post-analysis, and have problems such as slow response speed, high false alarm rate, and difficulty in real-time monitoring. With the development and progress of intelligent detection technology, related technologies have begun to use anomaly detection algorithms to detect and analyze faults in synchronization timing systems to improve the stability of system operation. A common fault detection algorithm is LOF (Local outlier factor). This algorithm determines whether it is an abnormal point by comparing the density of each point with its neighboring points.

[0003] The LOF algorithm can be applied to the synchronous timing system, and anomaly detection is performed directly based on the characteristic data of each periodic waveform during the detection of the waveform data of the HIFA synchronous timing system. However, the Euclidean distance in the traditional LOF algorithm detection process does not take into account the shape characteristics of the two waveforms, resulting in the characteristic vectors of two periodic signals with large waveform differences being calculated as a small Euclidean distance, making the calculation of the Euclidean distance inaccurate, affecting the accuracy of the final anomaly detection. Summary of the invention

[0004] In order to solve the problem of low accuracy of traditional anomaly detection, the present application provides an operation fault detection method and system for a HIAF synchronous timing system.

[0005] In a first aspect, the present application provides an operation fault detection method for a HIAF synchronization timing system, which adopts the following technical solution: A method for detecting operation faults in a HIAF synchronous timing system comprises the steps of: dividing waveform data to obtain periodic signals, calculating the initial Euclidean distance between any two periodic signals; calculating a correction coefficient between any two periodic signals; taking the product of the correction coefficient and the initial Euclidean distance as the optimal Euclidean distance, and performing LOF anomaly detection using the optimal Euclidean distance as a metric; The calculation steps of the correction coefficient include: setting the sampling quantity, discretely sampling the periodic signal, and obtaining a sampling data set; calculating the phase difference between two periodic signals according to the difference between the sampling values ​​of the same phase in the two sampling data sets; calculating the signal fluctuation degree between any two periods, and multiplying the phase difference degree and the signal fluctuation degree to obtain the period difference; dividing any period difference by the mean of all period differences to obtain the correction coefficient; wherein the calculation formula of the signal fluctuation degree is: ; In the formula, Indicates The periodic signal and The degree of signal fluctuation between periodic signals; Indicates The periodic signal and The sum of the absolute values ​​of the differences of all sampling points in a periodic signal, Indicates the number of sample values, Indicates In the signal cycle The sample value and In the signal cycle The absolute value of the difference between the sample values.

[0006] Beneficial effects: Two periodic signals are sampled to obtain a sampling data set, the sampling values ​​in the sampling data sets corresponding to different periodic signals are compared, the difference in the same-phase sampling values ​​between two different periodic signals is analyzed, and the same-phase difference between the two periodic signals is obtained. The difference between the two periodic signals is reflected according to the difference between the sampling point difference of the same phase and the mean of the sampling point difference. According to the size of the difference value, the reason for the fluctuation between the two periodic signals is reflected. For example, natural fluctuations or phase displacement fluctuations. The waveform difference of the two periodic signals is reflected by combining the fluctuation of the periodic signal and the same-phase difference. The initial Euclidean distance is corrected by the difference to improve the accuracy of LOF anomaly detection.

[0007] Optionally, the calculation formula for the phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

[0008] Beneficial effects: and Subtracting the two periodic signals yields the local difference of the sampling values ​​of the same phase in the two periodic signals, and accumulating the local differences yields the overall difference between the two periodic signals, that is, the degree of the same phase difference between the two periodic signals.

[0009] Optionally, the calculation formula for the phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

[0010] Beneficial effects: Absolute value calculation is simpler and faster, and improves calculation efficiency. For some special scenarios, such as those requiring efficient data processing, this method can ensure detection accuracy while taking into account calculation efficiency to a certain extent.

[0011] Optionally, the step of calculating the initial Euclidean distance between any two periodic signals includes: extracting features from the periodic signals, obtaining multiple eigenvalues ​​of each periodic signal to form a eigenvector; calculating a correction coefficient according to the interval between the two periodic signals; and calculating the initial Euclidean distance between the two eigenvectors based on the correction coefficient. The calculation formula of the correction coefficient is: ; In the formula, For the The periodic signal and Correction coefficient of a periodic signal; and are the indices of the two-period signals respectively; is an exponential function.

[0012] Beneficial effects: Indicates the sampling time interval between two periodic signals. The closer the distance between the two periods is, the more meaningful the comparison between the two periods is. The farther the distance between the two periods is, the greater the Euclidean distance between the two periodic signals is. This provides more references for the calculation of the Euclidean distance, improves the calculation accuracy of the Euclidean distance, and thus improves the accuracy of subsequent anomaly detection.

[0013] Optionally, the initial Euclidean distance is calculated as: ; Where: Indicates The periodic signal and The initial Euclidean distance between periodic signals, Indicates the first The anomaly weight of the feature value of the item, For the The periodic signal and Correction coefficient of a periodic signal; Indicates The first periodic signal Item data; Indicates The first periodic signal Item data; Represents the number of eigenvalues ​​in the eigenvector.

[0014] Beneficial effect: Different eigenvalues ​​extracted from the periodic signal have different effects on the degree of abnormality. Different weights are assigned to different eigenvalues, thereby improving the accuracy of the calculated Euclidean distance.

[0015] Optionally, the anomaly weight is determined by using a hierarchy analysis method according to the degree of influence of different eigenvalues ​​on the data anomaly.

[0016] Optionally, the steps of performing LOF anomaly detection using the optimal Euclidean distance as a measure include: determining the waveform anomaly of the periodic signal; determining the system anomaly; the steps of determining the waveform anomaly of the periodic signal include: setting an anomaly threshold; performing LOF anomaly detection on the periodic signal to determine the degree of anomaly of the periodic signal; and determining that the waveform of the periodic signal is abnormal when the response and the degree of anomaly are greater than the anomaly threshold.

[0017] Optionally, the step of determining system abnormality includes: setting a quantity threshold; calculating the degree of abnormality of each periodic signal, and calculating the number of periodic signals with waveform abnormalities; and issuing an alarm when the ratio of the number of periodic signals corresponding to waveform abnormalities to all periodic signals is greater than the quantity threshold.

[0018] Optionally, the eigenvalues ​​in the eigenvector include mean, variance, zero crossing rate, skewness, and kurtosis.

[0019] In a second aspect, the present application provides an operation fault detection system for a HIAF synchronization timing system, which adopts the following technical solution: An operation fault detection system for a HIAF synchronization timing system comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an operation fault detection method for a HIAF synchronization timing system is implemented.

[0020] The above-mentioned operation fault detection method for the HIAF synchronous timing system is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a system based on the memory and the processor for easy use.

[0021] The present application has the following technical effects: the difference in waveform between two periodic signals is reflected by the phase difference and signal fluctuation degree of the periodic signals. The correction coefficient is determined based on the phase difference and signal fluctuation degree to correct the Euclidean distance between the two periodic signals, thereby improving the accuracy of Euclidean distance calculation and thus improving the accuracy of LOF anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a method flow chart of an operation fault detection method for a HIAF synchronization timing system according to an embodiment of the present application.

[0023] Figure 2 This is a method flow chart of step S2 of a method for detecting an operation fault of a HIAF synchronization timing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The embodiment of the present application discloses an operation fault detection method for a HIAF synchronization timing system, which improves the accuracy of the Euclidean distance by analyzing the waveform shape, calculating the correction coefficient, and correcting the initial Euclidean distance in the LOF algorithm detection process, thereby improving the accuracy of the synchronization system abnormality detection.

[0025] Reference Figure 1 , the operation fault detection method for the HIAF synchronous timing system includes steps S1 to S3; S1: Segment the waveform data to obtain periodic signals and calculate the initial Euclidean distance between any two periodic signals; Step S1 includes steps S11 to S13; S11: Collect and segment the waveform signal of the synchronous timing system; An oscilloscope probe is connected to the output end of the synchronous timing system to capture the waveform signal output by the synchronous timing system. The collected waveform signal is subjected to noise reduction processing to make the waveform signal clearer and more stable. The waveform signal after noise reduction processing is segmented to form multiple periodic signals, and an index is set for each periodic signal to facilitate the retrieval of data in the periodic signal. In this embodiment, the noise reduction processing method is low-pass filtering, and in other embodiments, it can also be median filtering and other methods.

[0026] S12: extracting features from the periodic signal, obtaining multiple feature values ​​of each periodic signal to form a feature vector; For the mean, variance, zero crossing rate, skewness and kurtosis in each periodic signal, multiple eigenvalues ​​of the periodic signal are obtained, and the multiple eigenvalues ​​constitute the eigenvector of the periodic signal; in other embodiments, only the mean, variance, skewness and kurtosis in the periodic signal may be collected.

[0027] S13: Calculate the correction coefficient according to the interval between two periodic signals; In one embodiment, the calculation formula of the correction coefficient is: ; In the formula; For the The periodic signal and Correction coefficient of a periodic signal; and are the indices of the two-period signals respectively; is an exponential function.

[0028] In another embodiment, the calculation formula of the correction coefficient may also be: ; Where; For the The periodic signal and Correction coefficient of a periodic signal; and are the indices of the two-period signals respectively; is an exponential function.

[0029] S14: Calculate the initial Euclidean distance between two feature vectors; The calculation formula of the initial Euclidean distance is: ; In the formula, Indicates The periodic signal and The initial Euclidean distance between periodic signals, Indicates the first The anomaly weight of the feature value of the item, For the The periodic signal and Correction coefficient of a periodic signal; Indicates The first periodic signal Item data; Separate The first periodic signal Item data; Represents the number of eigenvalues ​​in the eigenvector.

[0030] Each eigenvector contains multiple eigenvalues, that is, multidimensional data of the eigenvector. Among the multiple eigenvalues ​​in the eigenvector, different eigenvalues ​​have different influences on the degree of abnormality of the periodic signal. Therefore, in this application, different weights, that is, abnormal weights, are assigned to each eigenvalue, so that the detection accuracy can be improved to a certain extent.

[0031] It represents the difference between the indexes of two periodic signals, which can reflect the time difference between the two periodic signals. In the actual detection process, the closer the acquisition time of the two periodic signals is, the more meaningful the comparison between the two periodic signals is. Therefore, when the difference is large, the initial Euclidean distance between the two periods increases.

[0032] S2: Calculate the correction coefficient between any two periodic signals; Reference Figure 2 , step S2 includes steps S21 to S23.

[0033] S21: Set the sampling quantity, perform discrete sampling on the periodic signal, and obtain a sampling data set; Determine the number of samples in each periodic signal; discretely sample each periodic signal at a certain time interval, and the sampled data together constitute a sampling data set.

[0034] In this embodiment, the number of samples of each periodic signal is 100. In other embodiments, the number of samples of each periodic signal may also be 150, 160, etc.

[0035] S22: Calculate the phase difference between the two periodic signals according to the difference between the sampling values ​​of the same phase in the two sampling data sets; In one embodiment, the calculation formula of the phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

[0036] Indicates the difference between the sampling values ​​of the same phase in two periodic signals; in an ideal state, the value of this part is zero. The larger the value, the greater the difference between the two periodic signals.

[0037] In another embodiment, the calculation formula of the in-phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

[0038] This formula uses absolute value operation, which makes the calculation process simpler and more efficient.

[0039] S23: Calculate the signal fluctuation degree between any two cycles, and multiply the phase difference degree and the signal fluctuation degree to obtain the cycle difference; The sampling values ​​of the sampling points in the sampling data sets of the two periodic signals are compared, and the signal fluctuation degree between the two periodic signals is calculated according to the difference between the sampling values ​​of the two periodic signals.

[0040] In one embodiment, the calculation formula for the signal fluctuation degree is: ; In the formula, Indicates The periodic signal and The degree of signal fluctuation between periodic signals; Indicates The periodic signal and The sum of the absolute values ​​of the differences of all sampling points in a periodic signal, Indicates the number of sample values, Indicates In the signal cycle The sample value and In the signal cycle The absolute value of the difference between the sample values.

[0041] Each sampling data set of a periodic signal includes an equal number of sampling values; and the sampling points in any two sampling data sets of periodic signals correspond to each other in phase. Two sampling values ​​with the same phase in two sampling data sets constitute a matching pair, and the absolute value of the difference between each matching pair is calculated and added to obtain . Indicates the deviation of the local difference from the average difference. By measuring the deviation of the local difference from the average difference, the fluctuation degree of the difference is calculated. The greater the fluctuation degree, the more random the difference between the two periodic signals is, and the more likely the fluctuation is the natural fluctuation of the signal. If the difference between the two periodic signals changes more steadily, it means that the difference between the two signals is more likely to be caused by the phase difference or frequency difference. Considering the fluctuation degree in the calculation process of the periodic difference can accurately reflect the periodic difference between the two periodic signals, thereby improving the accuracy of the initial Euclidean distance correction.

[0042] S24: Divide any period difference by the mean of all period differences to obtain a correction coefficient; Calculate the period difference of any two periodic signal combinations, calculate all period differences, and calculate the average value. Select one of the period differences and divide it by the average value to obtain the correction coefficient of the two periodic signals corresponding to the period difference.

[0043] The calculation method of the correction coefficient can be expressed by the following formula: ; In the formula, Indicates The periodic signal and Correction coefficient between periodic signals; Indicates The periodic signal and The period difference between the periodic signals is Represents the average of all period differences.

[0044] S3: The product of the correction coefficient and the initial Euclidean distance is taken as the optimal Euclidean distance, and LOF anomaly detection is performed using the optimal Euclidean distance as a metric; Determine the waveform abnormality; when performing LOF abnormality detection, perform abnormality detection using the optimal Euclidean distance as a metric; obtain the degree of abnormality of each waveform. Before abnormality detection, set an abnormality threshold, and when the degree of abnormality of the periodic signal in response to LOF abnormality detection is greater than the abnormal threshold, determine that the periodic signal waveform is abnormal. For example, in this embodiment, the abnormal threshold is set to 0.67, and when the degree of abnormality of the waveform detected by the abnormality is greater than 0.67, the waveform of the periodic signal is determined to be abnormal.

[0045] Determine system abnormality; preset a quantity threshold; when the ratio of the number of periodic signals with abnormal waveforms to all periodic signals is greater than the quantity threshold, issue an alarm to determine that the synchronous timing system is operating abnormally.

[0046] An embodiment of the present application also discloses an operation fault detection system for a HIAF synchronization timing system, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, an operation fault detection method for a HIAF synchronization timing system according to the present application is implemented.

[0047] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0048] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for detecting operation failures in a HIAF synchronous timing system, characterized in that: The method comprises the steps of: dividing the waveform data to obtain a periodic signal, calculating the initial Euclidean distance between any two periodic signals; calculating the correction coefficient between any two periodic signals; The product of the correction coefficient and the initial Euclidean distance is taken as the optimal Euclidean distance, and the optimal Euclidean distance is used as the metric for LOF anomaly detection; The calculation steps of the correction coefficient include: setting the sampling quantity, discretely sampling the periodic signal, and obtaining a sampling data set; The phase difference between two periodic signals is calculated based on the difference between the sampling values ​​of the same phase in two sampling data sets; the signal fluctuation degree between any two periods is calculated, and the period difference is obtained by multiplying the phase difference and the signal fluctuation degree; any period difference is divided by the mean of all period differences to obtain the correction coefficient; the calculation formula of the signal fluctuation degree is: ; In the formula, Indicates The periodic signal and The degree of signal fluctuation between periodic signals; Indicates The periodic signal and The sum of the absolute values ​​of the differences of all sampling points in a periodic signal, Indicates the number of sample values, Indicates In the signal cycle The sample value and In the signal cycle The absolute value of the difference between the sample values.

2. The method for detecting operation failures of a HIAF synchronous timing system according to claim 1, characterized in that: The calculation formula for the phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

3. The method for detecting operation failures of a HIAF synchronous timing system according to claim 1, characterized in that: The calculation formula for the phase difference is: ; In the formula, Indicates The periodic signal and The phase difference of periodic signals; Indicates In the periodic signal Sample values; No. In the periodic signal Sample values; Represents the number of samples in a periodic signal.

4. A method for detecting operation failures of a HIAF synchronization timing system according to claim 2 or 3, characterized in that: The steps of calculating the initial Euclidean distance between any two periodic signals include: extracting features from the periodic signals, obtaining multiple eigenvalues ​​of each periodic signal to form a eigenvector; calculating a correction coefficient according to the interval between the two periodic signals; and calculating the initial Euclidean distance between the two eigenvectors based on the correction coefficient. The calculation formula of the correction coefficient is: ; In the formula, For the The periodic signal and Correction coefficient of a periodic signal; and are the indices of the two-period signals respectively; is an exponential function.

5. The method for detecting operation failures of a HIAF synchronization timing system according to claim 4, characterized in that: The calculation formula of the initial Euclidean distance is: ; Where: Indicates The periodic signal and The initial Euclidean distance between periodic signals, Indicates the first The anomaly weight of the feature value of the item, For the The periodic signal and Correction coefficient of a periodic signal; Indicates The first periodic signal Item data; Separate The first periodic signal Item data; Represents the number of eigenvalues ​​in the eigenvector.

6. The method for detecting operation failures of a HIAF synchronous timing system according to claim 5, characterized in that: The anomaly weights are determined by the analytic hierarchy process according to the degree of influence of different eigenvalues ​​on data anomalies.

7. The method for detecting operation failures of a HIAF synchronous timing system according to claim 1, characterized in that: The steps of performing LOF anomaly detection using the optimal Euclidean distance as a measure include: determining the waveform anomaly of a periodic signal; determining the system anomaly; the steps of determining the waveform anomaly of a periodic signal include: setting an anomaly threshold; performing LOF anomaly detection on the periodic signal to obtain the degree of anomaly of the periodic signal; and determining that the waveform of the periodic signal is abnormal when the response and the degree of anomaly are greater than the anomaly threshold.

8. The method for detecting operation failures of a HIAF synchronous timing system according to claim 7, characterized in that: The steps of judging system abnormality include: setting a quantity threshold; calculating the degree of abnormality of each periodic signal, and calculating the number of periodic signals with waveform abnormalities; when the ratio of the number of periodic signals corresponding to waveform abnormalities to all periodic signals is greater than the quantity threshold, an alarm is issued.

9. The method for detecting operation failures of a HIAF synchronous timing system according to claim 4, characterized in that: The eigenvalues ​​in the eigenvector include mean, variance, zero crossing rate, skewness, and kurtosis.

10. An operation fault detection system for a HIAF synchronous timing system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an operation fault detection method for a HIAF synchronization timing system according to any one of claims 1 to 9 is implemented.

Images