Turnout power curve state monitoring method and device

Abstract

The invention relates to the technical field of rail transit equipment fault monitoring, in particular to a turnout power curve state monitoring method and device, and the method comprises the steps: obtaining the first feature information of a real-time turnout action power curve; acquiring second characteristic information of the turnout action power curve in different operation states; based on a grey relational degree method and / or a fuzzy clustering method, calculating the similarity between the first feature information and second feature information corresponding to each fault type; and determining the corresponding operation state of the real-time turnout action power curve according to the similarity. The intelligent railway equipment electrical characteristic curve abnormity diagnosis method and system have higher accuracy, automation degree and robustness, fault features of turnout action power curve data are effectively extracted by adopting the fault diagnosis method based on digital signal processing, various fault detection requirements are met, and the fault diagnosis efficiency is improved. The classifier structure is simplified, and the fault diagnosis precision and efficiency are improved.

Description

Technical Field

[0001] The present application relates to the technical field of rail transit equipment fault monitoring, and in particular to a method and device for monitoring the state of a turnout power curve. Background Art

[0002] The electrical characteristic curve of railway equipment is an important indicator to describe the electrical performance of the equipment, and is of great significance for the operation status monitoring and fault diagnosis of the equipment. As one of the three major outdoor parts of railway signal equipment, the turnout is a line connection device that allows locomotives and vehicles to transfer from one track to another. It is also one of the weak links of the track. It is usually laid in large quantities at stations and marshalling yards, which can give full play to the passing capacity of the line and have an important impact on the safety and efficiency of railway transportation. Therefore, it is very important to judge whether the turnout equipment is operating normally, whether there are abnormal conditions, and the possible types of faults by analyzing the action power curve of the turnout. However, the current railway signal monitoring system ensures the safe and reliable operation of the turnout by means of regular inspection and maintenance of the turnout and setting up skylight points for maintenance. This maintenance method is not only inefficient and labor-intensive, but also relies solely on the experience of workers to identify faults, which is prone to misjudgment and missed judgment, endangering driving safety.

[0003] The current popular solution is the abnormal diagnosis of electrical characteristic curves based on machine learning and data mining technology. This method uses a large amount of historical data for training and builds a model to achieve abnormal diagnosis of electrical characteristic curves. The basic process of this method is: data collection and preprocessing; feature extraction and selection; model construction and training; abnormal identification and diagnosis. Although this existing technical solution has improved the accuracy of abnormal diagnosis of electrical characteristic curves to a certain extent, it still has the following shortcomings:

[0004] For new abnormal situations, the model's generalization ability is limited and may not be able to accurately identify and diagnose; during data collection and preprocessing, the handling of noise and interference is not perfect enough, which may affect the reliability of the diagnosis results; the feature extraction and selection methods may be limited by manual experience and lack automated and intelligent feature selection capabilities. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems in the above-mentioned technology to a certain extent, and proposes a method for monitoring the state of a turnout power curve, comprising:

[0006] Acquire first characteristic information of real-time turnout action power curve;

[0007] Acquire second characteristic information of the turnout action power curve under different operating conditions;

[0008] Based on a grey relational method and / or a fuzzy clustering method, calculating the similarity between the first characteristic information and the second characteristic information corresponding to each fault type;

[0009] The operating state corresponding to the real-time turnout action power curve is determined according to the similarity.

[0010] Preferably, obtaining the first characteristic information of the real-time turnout action power curve includes:

[0011] Obtaining the IMF component of the real-time turnout action power curve;

[0012] Obtain fuzzy entropy of the IMF component as the first feature information.

[0013] Preferably, obtaining the IMF component of the real-time turnout action power curve includes: decomposing the real-time turnout action power curve based on CEEMD to obtain the IMF component.

[0014] Preferably, after determining the operating state corresponding to the real-time turnout action power curve according to the similarity, the method further comprises:

[0015] Constructing a similarity set according to the calculated similarities;

[0016] Sorting the similarity set from large to small according to the similarity;

[0017] Determine whether the operating states corresponding to the first several similarities are all normal or all abnormal; if not, determine the difference between the first several similarities;

[0018] When it is determined that the difference between the first several similarities is less than a threshold value, a secondary judgment is made on the operating state corresponding to the real-time turnout action power curve based on at least one of a year-on-year value method, a year-on-year amplitude method and a ratio-on-ratio value method.

[0019] Preferably, judging the operating state corresponding to the real-time turnout action power curve based on the year-on-year value method includes:

[0020] Obtain the maximum value of several normal turnout action power curves at the target time;

[0021] Correcting the maximum value based on a preset static threshold value to obtain a reference maximum value;

[0022] It is determined whether the value of the real-time turnout action power curve at the target time is greater than the reference maximum value; if the determination result is yes, it is determined that the real-time turnout action power curve has a sudden increase at the target time.

[0023] Preferably, judging the operating state corresponding to the real-time turnout action power curve based on the year-on-year amplitude method includes:

[0024] Obtain the maximum amplitude of several normal turnout operation power curves at the target time;

[0025] Correcting the maximum amplitude based on a preset static threshold to obtain a reference maximum amplitude;

[0026] It is determined whether the amplitude of the real-time turnout action power curve at the target time is greater than zero and greater than the reference maximum value. If the determination result is yes, it is determined that the real-time turnout action power curve has a sudden increase at the target time.

[0027] Preferably, judging the operating state corresponding to the real-time turnout action power curve based on the ratio value method includes:

[0028] Obtaining target data of the real-time turnout action power curve at a target time;

[0029] Obtaining a difference between the target data and each data in a preset time window before the target time, and determining the number of differences greater than a preset static threshold;

[0030] When it is determined that the difference number is greater than the threshold, the target data is determined to be abnormal.

[0031] Preferably, the static threshold is determined based on historical turnout action power curve data.

[0032] The present application also proposes a turnout power curve state monitoring system, comprising:

[0033] A first acquisition module is used to acquire first characteristic information of a real-time turnout action power curve;

[0034] A second acquisition module is used to acquire second characteristic information of the turnout action power curve under different operating conditions;

[0035] A first calculation module, used for calculating the similarity between the first characteristic information and the second characteristic information corresponding to each fault type based on a grey relational method and / or a fuzzy clustering method;

[0036] A state determination module is used to determine the operating state corresponding to the real-time turnout action power curve according to the similarity.

[0037] The present application also proposes a computer-readable storage medium, in which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, it is at least used to implement the above method.

[0038] Compared with the prior art, the beneficial effects of the present application are as follows: the present application provides an intelligent method and system for diagnosing abnormalities in electrical characteristic curves of railway equipment, which has higher accuracy, automation and robustness. By adopting a fault diagnosis method based on digital signal processing, the fault characteristics of turnout action power curve data can be effectively extracted, various fault detection requirements can be met, the classifier structure can be simplified, and the accuracy and efficiency of fault diagnosis can be improved.

[0039] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures particularly pointed out in the written description and the drawings.

[0040] The technical solution of the present application is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to provide a further understanding of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings:

[0042] Figure 1 This is a schematic diagram of the turnout power curve state monitoring method provided in this application;

[0043] Figure 2 The flowchart of extracting feature information based on CEEMD is given in the embodiment;

[0044] Figure 3 The flowchart of obtaining IMF components by decomposing CEEMD signals given in the embodiment;

[0045] Figure 4 A flowchart of fault diagnosis based on gray association and fuzzy clustering is provided for an embodiment;

[0046] Figure 5 The present invention is a flowchart of a fault diagnosis based on a grey correlation method provided in an embodiment;

[0047] Figure 6 The flowchart of fault diagnosis based on fuzzy clustering method given in the embodiment.

[0048] Figure 7 A flowchart of enhanced fault diagnosis provided in an embodiment;

[0049] Figure 8 A schematic diagram of a method for monitoring a turnout power curve state according to an embodiment;

[0050] Fig. 9 A schematic diagram of a turnout power curve state monitoring system provided in an embodiment;

[0051] Fig.10 A schematic diagram of a computer-readable storage medium is provided for an embodiment. DETAILED DESCRIPTION

[0052] The present application is described below in conjunction with the accompanying drawings. The preferred embodiments described herein are only used to illustrate and explain the present application and are not used to limit the present application.

[0053] Figure 1 The turnout power curve condition monitoring method provided in this application includes:

[0054] S11 obtains the first characteristic information of the real-time turnout action power curve;

[0055] S12 obtains the second characteristic information of the turnout action power curve under different operating conditions;

[0056] S13. Based on the grey correlation method and / or the fuzzy clustering method, calculating the similarity between the first characteristic information and the second characteristic information corresponding to each fault type;

[0057] S14. Determine the operating state corresponding to the real-time turnout action power curve according to the similarity.

[0058] According to some embodiments of the present application, step S11 obtains first characteristic information of the real-time turnout action power curve, including extracting characteristic information corresponding to the real-time turnout action power curve based on CEEMD multi-scale fuzzy entropy, and the scheme is as follows:

[0059] This feature extraction method uses the complementary set empirical mode decomposition algorithm CEEMD to decompose the turnout action power curve, obtain a set of intrinsic mode functions IMFs with different time scales from high to low frequencies, and realize the adaptive multi-scaling of the original signal. Then the fuzzy entropy of each IMF component is extracted as the fault diagnosis feature parameter to construct the fault diagnosis feature set. This process is as follows Figure 2 shown.

[0060] (1) CEEMD decomposition to obtain IMF components

[0061] For a given original signal x(t), first initialize the number of complementary set empirical mode decompositions M, then add a set of positive and negative paired white noise sequences n(t) to the original signal x(t), perform M empirical mode EMD decompositions on the original signal x(t), sum and average the results of M EMD decompositions to eliminate the influence of white noise, and thus obtain the final IMF component. The above process is as follows: Figure 3 As shown:

[0062] (2) Fuzzy entropy of IMF.

[0063] The definition of fuzzy entropy is as follows:

[0064]

[0065] Among them, m is the model dimension, r is the similarity tolerance parameter, N is the original data length, It is the relationship dimension under n dimension.

[0066] The original signal x(t) is decomposed by CEEMD to obtain the intrinsic mode functions (IMFs) of different time scales from high to low frequencies, realizing the adaptive multi-scaling of the original signal. The fuzzy entropy of each IMF component after decomposition is extracted as the characteristic parameter of the original signal, that is, the multi-scale fuzzy entropy of the original signal, thereby constructing the feature set of turnout fault diagnosis.

[0067] According to some embodiments of the present application, step S12 obtains the second characteristic information of the turnout action power curve under different operating conditions in the same principle as step S11, and extracts the characteristic information of the turnout action power curve under a known operating condition.

[0068] According to some embodiments of the present application, step S13 calculates the similarity between the first feature information and the second feature information corresponding to each fault type based on a grey relational method and / or a fuzzy clustering method, including the following schemes:

[0069] The diagnostic method based on gray association and fuzzy clustering is to draw on the heterogeneous integration idea in ensemble learning, combining two simple and fast weak diagnostics into a strong diagnostic. The gray association method will obtain the "association" of different fault types, and the fuzzy clustering method will obtain the "membership" of different fault types, and the results they obtain are all values ​​between 0 and 1. This system weights the results of the two methods by a certain proportion to obtain a new diagnostic basis "comprehensive similarity", which is used to diagnose whether the turnout action power curve is abnormal, which can greatly improve the accuracy of abnormal diagnosis and reduce the missed alarms of the railway signal centralized monitoring system. The above process is as follows Figure 4 As shown, the program includes:

[0070] (1) Fault identification method based on grey correlation

[0071] like Figure 5 As shown in the figure, the process of the fault identification method based on grey correlation degree includes: establishing a fault diagnosis feature set X, setting the feature quantity of the sample to be identified, calculating the feature quantity of the sample to be identified and the feature vector X in the feature set X. iThe absolute difference between them is calculated, and the minimum and maximum absolute differences are obtained. The grey correlation coefficient between the feature quantity of the sample to be identified and the feature set X is calculated. According to the grey correlation coefficient, the grey correlation values ​​between the feature quantity of the sample to be identified and each diagnostic feature vector are calculated respectively. According to the size of the “grey correlation degree”, it is determined which type of fault in the fault feature set the current turnout fault power curve belongs to.

[0072] (2) Fault identification method based on fuzzy clustering

[0073] like Figure 6 As shown in the figure, the fault identification method needs to first establish the original characteristic pattern matrix X, which is a matrix composed of the characteristic values ​​of the normal power curve, multiple typical fault power curves and the curve to be tested. Then the matrix X is standardized by the translation standard deviation transformation and the translation range transformation, so that the mean of each variable is 0 and the standard deviation is 1, thereby eliminating the influence of different dimensions on the data. The distance method is used to calculate the original matrix X to obtain the fuzzy similarity matrix R, and then the fuzzy equivalent matrix R* with transitivity is calculated. Then, in *R, when λ changes from 1 to 0, the corresponding Boolean matrix is ​​formed, thereby forming a dynamic clustering graph. According to the size of the "degree of membership" in the dynamic graph, it is determined which type of fault in the fault feature set the current turnout fault power curve belongs to.

[0074] According to some embodiments of the present application, when the top two rankings of the "comprehensive similarity" of the diagnostic results are a certain fault type and a normal type (these two types are not distinguished in terms of ranking), and the "comprehensive similarity" is very close, there is a high probability that a misjudgment may occur. In order to avoid misjudgment, when this situation occurs, the diagnostic judgment can be further strengthened, which is equivalent to adding a layer of misjudgment protection, further improving the accuracy of abnormal diagnosis. In this embodiment, three simple and fast weak diagnostic devices, namely, a year-on-year value method diagnostic device, a year-on-year amplitude method diagnostic device, and a ratio value method diagnostic device, are combined into a strong diagnostic device. When the diagnostic system is prone to misjudgment, it helps to further analyze it. The specific process is as follows: Figure 7 shown.

[0075] (1) Comparative Value (CV)

[0076] Select the data of the turnout action power curve corresponding to the past P normal turnout operations. At a certain moment, P points can be obtained as reference values, which we record as D i (t), where i = 1, ..., P. The static threshold T can be used to determine whether the data of the turnout power curve at a certain moment is D i (t) Is there any abnormality (sudden increase / decrease). If D i If (t) is smaller than the minimum value of the same moment in the past P times multiplied by a threshold T, the data at that moment will be considered an abnormal point (sudden decrease); and if D i(t) is greater than the maximum value of the same moment in the past P times multiplied by a threshold T, then the data D at that moment is considered i (t) is the abnormal point (sudden increase), and the specific calculation formula is as follows:

[0077]

[0078] Among them, Result represents the judgment result, max_T represents the maximum threshold T, min_ represents the minimum threshold T, min(·) represents the maximum value of ·, and max(·) represents the minimum value of ·.

[0079] (2) Common Amplitude Method (CA)

[0080] Select the data of the turnout action power curve corresponding to the past P normal turnout operations, and for a certain moment, P amplitude values ​​can be obtained. Using the absolute value of P amplitudes as the standard, if the amplitude Damp(t) at time t is larger than the maximum amplitude at the same time in the past P days multiplied by a threshold T, and the amplitude at time t is greater than 0, then it is considered that an abnormality (sudden increase) occurs at this moment, and if the amplitude Damp(t) at time t is larger than the maximum amplitude at the same time in the past P days multiplied by a threshold T, and the amplitude at time m is less than 0, then it is considered that an abnormality (sudden decrease) occurs at this moment. The specific calculation formula is as follows:

[0081]

[0082]

[0083] (3) Chain value method (SS)

[0084] Set the size of the time window to W, then take the detection value D(t) at time t and the past W (denoted as D j ) is compared with the data value at the time point. If it is greater than the threshold T, we will increase the count by 1. If the count exceeds the count_num we set, the point is considered an abnormal point. The specific calculation formula is as follows:

[0085]

[0086] According to some embodiments of the present application, the setting method of the threshold T refers to the average avg, maximum max and minimum min of the past period of time (such as the size in the window), and then takes the minimum value of max-avg and avg-min. The specific calculation formula is as follows:

[0087] T=min(max-avg,avg-min)

[0088] The above three methods usually select N moments, N ≥ 3n, and these N moments are selected in the same number in the three time periods of the start, the process and the end of the turnout rotation. When one or more moments are abnormal, it is judged as abnormal.

[0089] Figure 8 The present invention provides a method and system for diagnosing abnormality of the electrical characteristic curve (turnout action power curve) of railway signal equipment based on CEEMD multi-scale fuzzy entropy feature extraction and heterogeneous integration ideas. The complete process is as follows:

[0090] Obtain the curves under different fault types from the CSM database of the railway signal centralized monitoring system, obtain the turnout action power curve collected by the current station machine from the railway signal centralized monitoring system station machine, perform CEEMD decomposition on the two respectively, obtain the corresponding IMF components of the two, and construct the standard fault feature set and the feature quantity of the sample to be diagnosed. Based on the grey correlation and fuzzy clustering, calculate the correlation and membership of the standard fault feature set and the feature quantity of the sample to be diagnosed, and weight them to obtain the comprehensive similarity to determine the operating status. After that, determine whether the similarities with the highest comprehensive similarity correspond to the same operating status type. If so, the current judgment is correct, and determine whether there is a fault and the fault type. If not, further analyze the diagnosis results based on the above-mentioned diagnostic device to finally determine the fault type.

[0091] Based on the same inventive concept, Fig. 9 As shown, the present application also proposes a switch power curve state monitoring system, including: a first acquisition module 91, used to obtain the first characteristic information of the real-time switch action power curve; a second acquisition module 92, used to obtain the second characteristic information of the switch action power curve under different operating conditions; a first calculation module 93, used to calculate the similarity between the first characteristic information and the second characteristic information corresponding to each fault type based on the gray correlation method and / or the fuzzy clustering method; a state determination module 94, used to determine the corresponding operating state of the real-time switch action power curve according to the similarity. Fig.10 As shown, the present application also provides a computer-readable storage medium 1100, in which a computer program or instruction is stored. When the computer program or instruction is executed by a processor, it is at least used to implement the above method.

[0092] It is obvious that a person skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for monitoring the state of a turnout power curve, characterized in that: include: Acquire first characteristic information of real-time turnout action power curve; Acquire second characteristic information of the turnout action power curve under different operating conditions; Based on a grey relational method and / or a fuzzy clustering method, calculating the similarity between the first characteristic information and the second characteristic information corresponding to each fault type; The operating state corresponding to the real-time turnout action power curve is determined according to the similarity.

2. The method according to claim 1, characterized in that The first characteristic information of the real-time turnout action power curve is obtained, including: Obtaining the IMF component of the real-time turnout action power curve; Obtain fuzzy entropy of the IMF component as the first feature information.

3. The method according to claim 2, characterized in that Acquiring the IMF component of the real-time turnout action power curve includes: decomposing the real-time turnout action power curve based on CEEMD to obtain the IMF component.

4. The method according to claim 3, characterized in that After determining the operating state corresponding to the real-time turnout action power curve according to the similarity, the method further includes: Constructing a similarity set according to the calculated similarities; Sorting the similarity set from large to small according to the similarity; Determine whether the operating states corresponding to the first several similarities are all normal or all abnormal; if not, determine the difference between the first several similarities; When it is determined that the difference between the first several similarities is less than a threshold value, a secondary judgment is made on the operating state corresponding to the real-time turnout action power curve based on at least one of a year-on-year value method, a year-on-year amplitude method and a ratio-on-ratio value method.

5. The method according to claim 4, characterized in that The corresponding operating state of the real-time turnout action power curve is judged based on the year-on-year value method, including: Obtain the maximum value of several normal turnout action power curves at the target time; Correcting the maximum value based on a preset static threshold value to obtain a reference maximum value; It is determined whether the value of the real-time turnout action power curve at the target time is greater than the reference maximum value; if the determination result is yes, it is determined that the real-time turnout action power curve has a sudden increase at the target time.

6. The method according to claim 4, characterized in that The corresponding operating state of the real-time turnout action power curve is judged based on the year-on-year amplitude method, including: Obtain the maximum amplitude of several normal turnout operation power curves at the target time; Correcting the maximum amplitude based on a preset static threshold to obtain a reference maximum amplitude; It is determined whether the amplitude of the real-time turnout action power curve at the target time is greater than zero and greater than the reference maximum value. If the determination result is yes, it is determined that the real-time turnout action power curve has a sudden increase at the target time.

7. The method according to claim 4, characterized in that The operation state corresponding to the real-time turnout action power curve is judged based on the ratio value method, including: Obtaining target data of the real-time turnout action power curve at a target time; Obtaining a difference between the target data and each data in a preset time window before the target time, and determining the number of differences greater than a preset static threshold; When it is determined that the difference number is greater than the threshold, the target data is determined to be abnormal.

8. The method according to any one of claims 5 to 7, characterized in that: The static threshold is determined based on historical turnout action power curve data.

9. A turnout power curve status monitoring system, characterized in that: include: A first acquisition module is used to acquire first characteristic information of a real-time turnout action power curve; A second acquisition module is used to acquire second characteristic information of the turnout action power curve under different operating conditions; A first calculation module, used for calculating the similarity between the first characteristic information and the second characteristic information corresponding to each fault type based on a grey relational method and / or a fuzzy clustering method; A state determination module is used to determine the operating state corresponding to the real-time turnout action power curve according to the similarity.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program or instruction, and when the computer program or instruction is executed by the processor, it is used to implement at least the method according to any one of claims 1 to 8.

Images