Abnormal monitoring data diagnosis method and system combined with prior physical information

Abstract

The invention provides an abnormal monitoring data diagnosis method and system combined with prior physical information, and belongs to the technical field of railway operation and maintenance monitoring. For a sensor combination with high correlation, a correlation prior hypothesis is put forward, the correlation among multiple sensors is calculated in real time with a fixed time window based on the hypothesis, and abnormity is judged; establishing an observation regression model based on prior physical information for the sensors without correlativity; and in combination with the Bayesian theory and a prior regression model, carrying out prior assumption on the number and positions of abnormal points in an observation area, obtaining a value combination with the highest possibility through Markov chain sampling, and determining the number and positions of the abnormal points. According to the method, a prior correlation assumption model and a prior physical regression model are respectively provided based on prior physical information of the monitoring data, and strong dependence of a traditional anomaly detection algorithm on data characteristics and distribution is avoided, so that rapid and accurate capture of the abnormal monitoring data is realized, and a powerful guarantee is provided for data reliability of a railway monitoring system.

Description

Technical Field

[0001] The present invention relates to the technical field of railway operation and maintenance monitoring, and in particular to a method and system for diagnosing abnormal monitoring data combined with prior physical information. Background Art

[0002] Objective and effective monitoring data is fundamental for condition assessment and safety early warning of railway infrastructure. While monitoring systems are highly reliable, data anomalies caused by temperature fluctuations, rain, snow, strong winds, vehicle vibration, sensor aging, structural expansion and contraction, and human factors cannot be ruled out during long-term monitoring. Research by the US Center for Intelligent Maintenance Systems indicates that over 40% of automated system fault alerts are false alarms due to faults in the sensor system itself. Abnormal monitoring data often distorts the results of numerical analysis. Therefore, pre-processing steps such as anomaly diagnosis and correction are necessary before analysis and processing. This ensures accurate and effective monitoring data by deriving effective information that truly reflects the service status of seamless railway lines and facilitates analysis and evaluation.

[0003] When there is an unacceptable deviation between the measured value and the true value, a sensor is considered to have failed. Sensor failure is generally defined from two perspectives: a system-centric perspective and a data-centric perspective. From the system-centric perspective, sensor failures can be categorized as calibration failures, connection failures, and low battery. From the data-centric perspective, sensor failures are analyzed based on the characteristics of the sensor measurement data. Sensor failures can be divided into six categories: deviation, gain, drift, accuracy degradation, stuckness, and outliers (sensor tripping).

[0004] In the prior art, a common practice is to statistically describe existing historical data, assign a confidence probability, determine a corresponding confidence interval, and then identify errors that exceed the confidence interval as gross errors, which are then eliminated. Commonly used methods for identifying gross errors include the 3σ criterion (Laida criterion), box plots, Chauvenet's criterion, Grubbs's criterion, and fine-grained criteria. However, these methods make certain assumptions about the data distribution (such as normal distribution) and have relatively strict definitions of anomalies. Using fixed static thresholds to diagnose anomalies ignores the physical significance of the actual monitoring data, resulting in significant limitations. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for diagnosing abnormal monitoring data that combines prior physical information and can fully utilize the physical characteristics of historical monitoring data to effectively guide the abnormal data diagnosis process and provide reliable data support for railway monitoring operation safety, so as to solve at least one technical problem existing in the above-mentioned background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a method for diagnosing abnormal monitoring data in combination with prior physical information, comprising:

[0008] Perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation;

[0009] For highly correlated sensor combinations, a correlation priori hypothesis is proposed, which states that data anomalies are accidental and it is impossible for multiple sensors to be abnormal at the same time. Based on this correlation priori hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and anomalies are determined based on the calculation results.

[0010] For sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed and an observation regression model based on prior physical information is established;

[0011] Combining Bayesian theory with the observation regression model, a priori assumptions are made on the number and location of outliers in the observation area, and the highest probability value combination is obtained through Markov chain sampling to determine the number and location of outliers.

[0012] As a further limitation of the first aspect of the present invention, historical monitoring data is collected and recorded according to the sensor sequence number as {s1, s2, ....s n};

[0013] Starting from s1, the correlation between each sensor's observation data and all other sensors is calculated and recorded as {(s1,s2),(s1,s3),......(s1,s n ),(s2,s1),(s2,s3),......(s n ,s n-1 )}, retain the combination of each sensor with the highest correlation with other sensors, and record it as {(s1,s i ),(s2,s j ),......(s n ,s k )};

[0014] The combination that satisfies high correlation is included in the combination of highly correlated sensors, which is {(s i ,s j )......} corr , the sensor records that do not have high correlation are {s k ......} incorr .

[0015] As a further limitation of the first aspect of the present invention, for a highly correlated sensor combination {(s i ,sj )......} corr , proposes a priori assumptions about correlation, including: the abnormality of monitoring data is an accidental event, and it is impossible for multiple sensors to be abnormal at the same time; after prior calculation, {(s i ,s j )......} corr There is a high correlation between the sensor combination. When the correlation decreases significantly, it means that a sensor has an abnormality. Based on the prior assumption, a fixed time window is used to calculate {(s i ,s j )......} corr The correlation between a certain two sensor combinations is calculated. When the high correlation assumption is not met, it is considered that one of the sensors in the sensor combination is abnormal.

[0016] As a further limitation of the first aspect of the present invention, for sensors that do not have a high degree of correlation, k ......} incorr , analyze its historical data in time and frequency domains, clarify its periodic characteristics, and establish an observation regression model based on prior physical information:

[0017]

[0018] Among them, ε represents the regression error, ω q , Represents the frequency and phase information of a specific periodic component, a q The regression coefficient under the corresponding periodic component, M is the number of components that have a significant impact on the observed value Y;

[0019] According to the frequency domain signal, determine its main wavelength component ω q To avoid solving the phase information, the above formula is converted into the sum of sine and cosine components using trigonometric functions:

[0020]

[0021] Among them, the wavelength component is determined by prior analysis, and the initial values of other error terms and coefficients are taken as 1. Then, the initial parameters are dynamically improved based on the measured data to make the regression model better fit the observed sequence Y.

[0022] As a further limitation of the first aspect of the present invention, based on the prior regression model, the monitoring data with a total length of N is calculated in its subset Y i:j ={Y i ,Y i+1 ,......Y j}, (1≤i≤j≤N), that is:

[0023] f(Yi:j )=f(Y i ,Y i+1 ,…,Y j-1 ,Y j |X)

[0024] The actual observed value of the monitoring data is taken as the predicted value Y i:j The joint probability of; Assume that there are K position change points, and record the position of the change point as C K ={c1,c2,......c k}, then under the prior conditions of the number and location of change points, the likelihood probability is:

[0025] P k (Y 1:j )=f(Y1,Y2,…,Y j |K=k,C K ,X)

[0026] When the observed monitoring data changes state K times, and the corresponding change time is C K Under the premise of i:j The joint probability of .

[0027] As a further limitation of the first aspect of the present invention, the number of change points K and the position of the change points C are obtained by combining the Bayesian inference theory. K The posterior probability is obtained by repeated sampling to determine the most likely combination:

[0028]

[0029] Where, f(Y 1:N |K=k,C K ) is P k (Y 1:j ), f(K=k), f(C K |K=k) represents the prior estimate of the number of change points and the probability density of the change point positions. Through Markov chain sampling, the most likely value of K is obtained. When K increases compared with the previous observation time, the latest observation position is considered to be a new change point detected, that is, a sensor anomaly has occurred at this time.

[0030] In a second aspect, the present invention provides an abnormal monitoring data diagnosis system combined with prior physical information, comprising:

[0031] A calculation module is used to perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation;

[0032] The judgment module is used to propose a priori correlation hypothesis for highly correlated sensor combinations, namely, data anomalies are accidental events and it is impossible for multiple sensors to be abnormal at the same time. Based on the prior correlation hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and the anomaly is judged based on the calculation results.

[0033] A construction module is used to analyze the typical characteristics of historical monitoring data of sensors that do not have a correlation relationship and establish an observation regression model based on prior physical information;

[0034] The determination module is used to combine Bayesian theory with the observation regression model to make prior assumptions about the number and location of abnormal points in the observation area, and obtain the most likely value combination through Markov chain sampling to determine the number and location of abnormal points.

[0035] In a third aspect, the present invention provides a non-transitory computer-readable storage medium, which is used to store computer instructions. When the computer instructions are executed by a processor, the abnormal monitoring data diagnosis method combined with prior physical information as described in the first aspect is implemented.

[0036] In a fourth aspect, the present invention provides a computer device comprising a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the abnormal monitoring data diagnosis method combined with prior physical information as described in the first aspect.

[0037] In a fifth aspect, the present invention provides an electronic device comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the abnormal monitoring data diagnosis method combined with prior physical information as described in the first aspect.

[0038] The beneficial effects of the present invention are as follows: the abnormality monitoring of sensor monitoring data is carried out by the correlation coefficient method, which is different from the traditional abnormal data detection method and can calculate the change point diagnosis results of each sensor in real time; the regression observation model is obtained by combining prior physical information, and the model parameters are dynamically adjusted according to the actual observation values. In combination with the Bayesian inference theory, the change point information within the observation range is obtained by repeated sampling, which has high detection accuracy and low calculation amount, and can quantify the uncertainty of the number of change points and the position of the change points.

[0039] Additional advantages of the present invention will be more clearly given in the following description or learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 This is a flow chart of the abnormality detection data diagnosis method according to an embodiment of the present invention.

[0042] Figure 2 This is a linear fitting effect diagram of the highly correlated sensor according to an embodiment of the present invention.

[0043] Figure 3 This is a linear fitting effect diagram of the non-height-correlated sensor according to an embodiment of the present invention.

[0044] Figure 4 This is a graph showing the calculation results of the correlation coefficient according to an embodiment of the present invention. Figure 4 (a) shows the fluctuation of three correlation coefficients: Spearman, Kendall, and Maximum Mutual Information Coefficient (MIC) over time; Figure 4 (b) shows the fluctuation of frequency along with the correlation coefficient.

[0045] Figure 5 This is a diagram showing the effect of abnormal diagnosis based on sensor correlation according to an embodiment of the present invention.

[0046] Figure 6 This is a diagram showing the effect of local abnormality diagnosis based on sensor correlation according to an embodiment of the present invention.

[0047] Figure 7 This is a diagram of the frequency domain component analysis results of sensor monitoring data according to an embodiment of the present invention.

[0048] Figure 8 This is a diagram showing the effect of anomaly recognition of monitoring data based on Bayesian theory according to an embodiment of the present invention.

[0049] in, Figure 8 (a) is the temperature monitoring result diagram; Figure 8 (b) is the posterior probability monitoring diagram. DETAILED DESCRIPTION

[0050] The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and are not to be construed as limiting the present invention.

[0051] Those skilled in the art will understand that unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this invention belongs.

[0052] It should also be understood that terms, such as those defined in commonly used dictionaries, should be understood to have a meaning consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless as defined herein.

[0053] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a," "an," "said," and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, and / or groups thereof.

[0054] In the description of this specification, reference to the terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless otherwise contradictory.

[0055] To facilitate understanding of the present invention, the present invention is further explained below with reference to specific embodiments in conjunction with the accompanying drawings. However, the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0056] Those skilled in the art should understand that the drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily necessary for implementing the present invention.

[0057] Example 1

[0058] In this embodiment 1, first, a system for diagnosing abnormal monitoring data combined with prior physical information is provided, including: a calculation module for performing correlation calculation on historical monitoring data to obtain a sensor combination with high correlation. A judgment module for proposing a correlation prior hypothesis for a sensor combination with high correlation, that is, data anomalies are accidental events, and it is impossible for multiple sensors to be abnormal at the same time; based on the correlation prior hypothesis, the correlation between multiple sensors is calculated in real time with a fixed time window, and the anomaly is judged according to the calculation result. A construction module is used to analyze the typical characteristics of the historical monitoring data of sensors that do not have a correlation relationship, and establish an observation regression model based on prior physical information. A determination module is used to combine Bayesian theory with the observation regression model, make a prior hypothesis on the number and location of abnormal points in the observation area, and obtain the most likely value combination through Markov chain sampling to determine the number and location of abnormal points.

[0059] In this embodiment, the above-mentioned system is used to implement a method for diagnosing abnormal monitoring data combined with prior physical information, including the following process steps:

[0060] S1. Calculate the correlation of historical monitoring data to obtain sensor combinations with high correlation;

[0061] S2. For a highly correlated sensor combination, a correlation a priori hypothesis is proposed, that is, data anomalies are accidental events, and it is impossible for multiple sensors to be abnormal at the same time;

[0062] S3, based on the S2 assumption, calculates the correlation between multiple sensors in real time using a fixed time window, and determines anomalies based on the calculation results;

[0063] S4. For sensors that do not have a correlation relationship, analyze the typical characteristics of their historical monitoring data and establish an observation regression model based on prior physical information;

[0064] S5. Combining Bayesian theory with the observation regression model, we make a priori assumptions about the number and location of outliers in the observation area, and obtain the most likely value combination through Markov chain sampling to determine the number and location of outliers.

[0065] Step S1 collects historical monitoring data of the monitoring system or similar monitoring systems, and records them as {s1, s2, ....s n}:

[0066] Starting from s1, the correlation between each sensor's observation data and all other sensors is calculated and recorded as {(s1,s2),(s1,s3),......(s1,s n ),(s2,s1),(s2,s3),......(s n ,s n-1)}, retain the combination of each sensor with the highest correlation with other sensors, and record it as {(s1,s i ),(s2,s j ),......(s n ,s k )}, at this time i,j...can be repeated.

[0067] Then the combination that satisfies high correlation (correlation coefficient greater than 0.6) is included in the highly correlated sensor combination, which is {(s i ,s j )......} corr , the sensor records that do not have high correlation are {s k ......} incorr .

[0068] In S2, there is a highly correlated sensor combination {(s i ,s j )......} corr , making a priori assumptions about the correlation, including:

[0069] (1) Abnormal monitoring data is an accidental event, and it is impossible for multiple sensors to be abnormal at the same time;

[0070] (2) According to a priori calculation, {(s i ,s j )......} corr There is a high correlation between the sensor combinations. When the correlation drops significantly, it means that a sensor has an abnormality.

[0071] In S3, based on the prior assumption of S2, a fixed time window (e.g., 24 hours) is used to calculate {(s i ,s j )......} corr The correlation between two sensor combinations in the system is calculated. When the high correlation assumption is not met (i.e., the correlation coefficient is less than 0.6), it is considered that one of the sensors in the sensor combination is abnormal. This process is based on the sensor correlation information for diagnosis and has sufficient physical significance.

[0072] In S4, for sensors that do not have a high correlation {s k ......} incorr , analyze its historical data in time and frequency domains, clarify its periodic characteristics, and establish an observation regression model based on prior physical information:

[0073]

[0074] Among them, ε represents the regression error, ω q, Represents the frequency and phase information of a specific periodic component, a q is the regression coefficient under the corresponding periodic component, and M is the number of components that have a significant impact on the observed value Y.

[0075] According to the frequency domain signal, determine its main wavelength component ω q To avoid solving the phase information, the above formula is converted into the sum of sine and cosine components using trigonometric functions:

[0076]

[0077] The wavelength component is determined a priori, and the other error terms and coefficients are initially set to 1. These initial parameters are then dynamically refined based on measured data to better fit the observed sequence Y. The observation regression model, derived from time- and frequency-domain analysis of the measured data, is a comprehensive standard paradigm that considers the sensor monitoring data period, errors, and regression coefficients. The regression model, derived from the phase-shift properties of trigonometric functions, is more suitable for fitting measured data.

[0078] In said S5, based on the prior regression model established in S4, the monitoring data with a total length of N is calculated in its subset Y i:j ={Y i ,Y i+1 ,......Y j}, (1≤i≤j≤N), that is:

[0079] f ( Y i:j ) =f ( Y i ,Y i+1 ,…,Y j-1 ,Y j |X )

[0080] This means that the actual observed value of the monitoring data is exactly taken as the predicted value Y i:j The joint probability of . Assume that there are K position change points, and record the position of the change point as C K ={c1,c2,......c k}, then under the prior conditions of the number and location of change points, the likelihood probability is:

[0081] P k (Y 1:j )=f ( Y1,Y2,…,Y j |K=k,C K ,X)

[0082] Its meaning is that when the observed monitoring data changes state K times, and the corresponding change time is C K Under the premise of i:j The joint probability of , can be obtained by iterative calculation based on the chain reaction rule.

[0083] On this basis, combined with Bayesian inference theory, we can further obtain the number of change points K and the position of change points C K The posterior probability is obtained by repeated sampling to determine the most likely combination. The specific formula is:

[0084]

[0085] Where, f(Y 1:N |K=k,C K ) is P k (Y 1:j ), f(K=k), f(C K |K=k) represents the prior estimate of the number of change points and the probability density of the change point locations. Markov chain sampling is used to obtain the most likely value of K. When K increases compared to the previous observation, the latest observation location is considered a new change point, indicating a sensor anomaly.

[0086] In summary, combining S3 and S5, we can respectively i ,s j )......} corr With {s k ......} incorr Real-time anomaly detection by two types of sensors can meet the abnormal data diagnosis requirements of all sensor devices in the monitoring system. The method described in this embodiment has a simple data algorithm and high accuracy, and can provide a reliable diagnosis method for railway monitoring abnormal value data processing.

[0087] Example 2

[0088] like Figure 1 As shown, this embodiment 2 provides a method for diagnosing abnormal monitoring data combined with prior physical information, which specifically includes the following steps:

[0089] S1. Calculate the correlation of historical monitoring data to obtain sensor combinations with high correlation. In a specific example, historical monitoring data of temperature sensors and rail temperature sensors at corresponding locations of stations on a certain line are collected, and the correlation of each sensor with all other sensor observation data is calculated, and the combination with the highest correlation of each sensor with other sensors is retained. Combinations that meet high correlation (correlation coefficient greater than 0.6) are then included in the highly correlated sensor combination for processing in steps S2-S3, and sensors that do not have high correlation are included in the non-highly correlated sensor combination for processing in steps S4-S5.

[0090] S2. For sensor combinations with high correlation, a priori assumption of correlation is made, that is, data anomalies are accidental events, and it is impossible for multiple sensors to be abnormal at the same time. In a specific example, it is obvious that different sensor combinations have different correlations, such as Figure 2 As shown in Figure 2, the sensor monitoring data points in a highly correlated sensor are densely distributed around the fitting curve, with an R2 score of 0.993. The x-axis and y-axis data show a strong correlation. Figure 3 As shown in the figure, the sensor monitoring data points of a non-highly correlated sensor are discretely distributed around the fitting curve. Although there is a certain positive correlation, the correlation is very low.

[0091] S3, based on the assumption of S2, calculate the correlation between multiple sensors in real time with a fixed time window (for example, 24 hours), and judge the abnormality based on the calculation results. In a specific example, based on the prior assumption of S2, the correlation between a certain two sensor combinations among the highly correlated sensors is calculated in real time online with a 24-hour time window, such as Figure 4 Figure 2 shows the temporal fluctuations of the three correlation coefficients: Spearman, Kendall, and Maximum Mutual Information Coefficient (MIC). The Spearman and Kendall correlation coefficients remained above 0.8 for most of the time, indicating a strong positive correlation between air temperature and rail temperature.

[0092] In the real-time operation process, when the high correlation assumption is not met (i.e. the correlation coefficient is less than 0.6), it is considered that a sensor in the sensor combination is abnormal. The data abnormality diagnosis results obtained by the three correlation coefficients are as follows: Figure 5 As shown in Figure 3, for the abnormal data distribution of these three correlation coefficients, the distribution of Spearman and Kendall correlation coefficients is relatively concentrated. In contrast, the distribution of the maximum mutual information coefficient is more dispersed.

[0093] Local anomaly diagnosis using measured temperature calculations such as Figure 6As shown in the figure, which includes both Spearman and Kendall correlation coefficients, it can be seen that the Kendall correlation coefficient is more effective in diagnosing abnormal monitoring data and can more sensitively diagnose abnormal data. This method can effectively utilize the correlation between highly correlated sensors to identify and process abnormal data, providing strong data support for railway operation safety.

[0094] S4. For sensors that do not have a correlation relationship, analyze the typical characteristics of their historical monitoring data and establish an observation regression model based on prior physical information. In a specific example, the observation regression model is established and adjusted as shown in the following steps:

[0095] (1) Analyze the physical information of the sensor historical monitoring data that does not have high correlation in time and frequency domains, clarify its periodic characteristics, and establish an observation regression model;

[0096] (2) By Figure 7 The frequency domain distribution shows that the temperature change signal frequency peak is more concentrated at 1d (24 hours), so its main wavelength component is determined to be 1, and then the observation regression model represented by the sum of sine and cosine components is transformed using the trigonometric function relationship;

[0097] (3) The initial parameters are then dynamically improved based on the measured data to make the regression model better fit the observed sequence.

[0098] S5. Combining Bayesian theory with the observation regression model, make a priori assumptions about the number and location of outliers in the observation area, and obtain the most likely value combination through Markov chain sampling to determine the number and location of outliers. Based on the prior regression model established in S4, the steps for determining the change point information of the monitoring data are as follows:

[0099] (1) Calculate the likelihood probability of the monitoring data with a total length of N on its subset window, that is, the joint probability that the actual observation value of the monitoring data is exactly the predicted value;

[0100] (2) The joint probability that the actual observed value is exactly the predicted value is obtained by iterative calculation based on the chain reaction law;

[0101] (3) Finally, combined with Bayesian inference theory, the posterior probability of the number of change points and the location of the change points is further obtained, and the most likely combination is determined by repeated sampling. Combined with Markov chain repeated sampling, the most likely value of the number of change points is obtained.

[0102] In a specific example, taking the rail temperature monitored by a sensor as an example, the change points of the sensor's measured data during the observation period are dynamically analyzed, and the posterior probability of a state mutation at each observation moment is calculated. The results are as follows: Figure 8As shown in the figure, the black line represents the monitoring data, the red line represents the regression results of the prior model, and the blue scattered points represent the posterior probability of each observation being a change point. A zoomed-in image shows that the development trend of the rail axial force before and after the peak of the posterior probability of the change point differs significantly. Using the change point monitoring method, when the number of change points obtained by the observation regression model and repeated sampling increases compared to the previous observation time, abnormal data is diagnosed, and the sampling window continues to move forward, guiding the completion of the diagnosis of all monitoring data.

[0103] Example 3

[0104] This embodiment 3 provides a non-transitory computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the abnormal monitoring data diagnosis method combined with prior physical information is implemented as described above. The method includes:

[0105] Perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation;

[0106] For highly correlated sensor combinations, a correlation priori hypothesis is proposed, which states that data anomalies are accidental and it is impossible for multiple sensors to be abnormal at the same time. Based on this correlation priori hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and anomalies are determined based on the calculation results.

[0107] For sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed and an observation regression model based on prior physical information is established;

[0108] Combining Bayesian theory with the observation regression model, a priori assumptions are made on the number and location of outliers in the observation area, and the highest probability value combination is obtained through Markov chain sampling to determine the number and location of outliers.

[0109] Example 4

[0110] This embodiment 4 provides a computer device, including a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the above-mentioned abnormal monitoring data diagnosis method combined with prior physical information, the method comprising:

[0111] Perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation;

[0112] For highly correlated sensor combinations, a correlation priori hypothesis is proposed, which states that data anomalies are accidental and it is impossible for multiple sensors to be abnormal at the same time. Based on this correlation priori hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and anomalies are determined based on the calculation results.

[0113] For sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed and an observation regression model based on prior physical information is established;

[0114] Combining Bayesian theory with the observation regression model, a priori assumptions are made on the number and location of outliers in the observation area, and the highest probability value combination is obtained through Markov chain sampling to determine the number and location of outliers.

[0115] Example 5

[0116] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to cause the electronic device to execute instructions for implementing the above-mentioned abnormal monitoring data diagnosis method combining prior physical information. The method includes:

[0117] Perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation;

[0118] For highly correlated sensor combinations, a correlation priori hypothesis is proposed, which states that data anomalies are accidental and it is impossible for multiple sensors to be abnormal at the same time. Based on this correlation priori hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and anomalies are determined based on the calculation results.

[0119] For sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed and an observation regression model based on prior physical information is established;

[0120] Combining Bayesian theory with the observation regression model, a priori assumptions are made on the number and location of outliers in the observation area, and the highest probability value combination is obtained through Markov chain sampling to determine the number and location of outliers.

[0121] In summary, the abnormal monitoring data diagnosis method combined with prior physical information described in the embodiment of the present invention. First, the historical monitoring data is correlated to obtain a sensor combination with high correlation; then, for the sensor combination with high correlation, a correlation prior hypothesis is proposed, that is, the data anomaly is an accidental event, and it is impossible for multiple sensors to be abnormal at the same time; based on the hypothesis, the correlation between multiple sensors is calculated in real time with a fixed time window (for example, 24 hours), and the anomaly is judged according to the calculation result; for sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed, and an observation regression model based on prior physical information is established; finally, combining Bayesian theory with the prior regression model, a prior hypothesis is made on the number and location of abnormal points in the observation area, and the combination with the highest possibility is obtained through Markov chain sampling to determine the number and location of abnormal points. The present invention proposes a prior correlation hypothesis and a prior physical regression model based on the prior physical information of the monitoring data, avoiding the strong dependence of the traditional anomaly detection algorithm on data characteristics and distribution, thereby realizing the rapid and accurate capture of abnormal monitoring data, and providing a strong guarantee for the data reliability of the railway monitoring system.

[0122] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0123] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts 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, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0124] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0125] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0126] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solutions disclosed in the present invention without the need for creative work should be included in the scope of protection of the present invention.

Claims

1. A method for diagnosing abnormal monitoring data in combination with prior physical information, characterized in that: include: Perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation; For highly correlated sensor combinations, a correlation priori hypothesis is proposed, which states that data anomalies are accidental and it is impossible for multiple sensors to be abnormal at the same time. Based on this correlation priori hypothesis, the correlation between multiple sensors is calculated in real time within a fixed time window, and anomalies are determined based on the calculation results. For sensors that do not have a correlation, the typical characteristics of their historical monitoring data are analyzed and an observation regression model based on prior physical information is established; Combining Bayesian theory with the observation regression model, a priori assumptions are made on the number and location of outliers in the observation area, and the highest probability value combination is obtained through Markov chain sampling to determine the number and location of outliers.

2. The abnormal monitoring data diagnosis method combined with prior physical information according to claim 1 is characterized in that: Collect historical monitoring data and record them according to sensor serial number {s1,s2,....s n }; Starting from s1, the correlation between each sensor's observation data and all other sensors is calculated and recorded as {(s1,s2),(s1,s3),......(s1,s n ),(s2,s1),(s2,s3),......(s n ,s n-1 )}, retain the combination of each sensor with the highest correlation with other sensors, and record it as {(s1,s i ),(s2,s j ),......(s n ,s k )}; The combination that satisfies high correlation is included in the combination of highly correlated sensors, which is {(s i ,s j )......} corr , the sensor records that do not have high correlation are {s k ......} incorr .

3. The abnormal monitoring data diagnosis method combined with prior physical information according to claim 2 is characterized in that: For highly correlated sensor combinations {(s i ,s j )......} corr , proposes a priori assumptions about correlation, including: the abnormality of monitoring data is an accidental event, and it is impossible for multiple sensors to be abnormal at the same time; after prior calculation, {(s i ,s j )......} corr There is a high correlation between the sensor combination. When the correlation decreases significantly, it means that a sensor has an abnormality. Based on the prior assumption, a fixed time window is used to calculate {(s i ,s j )......} corr The correlation between a certain two sensor combinations is calculated. When the high correlation assumption is not met, it is considered that one of the sensors in the sensor combination is abnormal.

4. The abnormal monitoring data diagnosis method combined with prior physical information according to claim 3 is characterized in that: For sensors that do not have high correlation k ......} incorr , analyze its historical data in time and frequency domains, clarify its periodic characteristics, and establish Observation regression model based on prior physical information: Among them, ε represents the regression error, ω q , Represents the frequency and phase information of a specific periodic component, a q The regression coefficient under the corresponding periodic component, M is the number of components that have a significant impact on the observed value Y; According to the frequency domain signal, determine its main wavelength component ω q To avoid solving the phase information, the above formula is converted into the sum of sine and cosine components using trigonometric functions: Among them, the wavelength component is determined by prior analysis, and the initial values of other error terms and coefficients are taken as 1. Then, the initial parameters are dynamically improved based on the measured data to make the regression model better fit the observed sequence Y.

5. The abnormal monitoring data diagnosis method combined with prior physical information according to claim 4 is characterized in that: Based on the prior regression model, calculate the monitoring data with a total length of N in its subset Y i:j ={Y i ,Y i+1 ,......Y j }, (1≤i≤j≤N), that is: f(Y i:j )=f(Y i ,AND i+1 ,…,AND j-1 ,AND j |X) The actual observed value of the monitoring data is taken as the predicted value Y i:j The joint probability of; Assume that there are K position change points, and record the position of the change point as C K ={c1,c2,......c k }, then under the prior conditions of the number and location of change points, the likelihood probability is: P k (Y 1:j )=f(Y1,Y2,…,Y j ∣K=k,C K ,X) When the observed monitoring data changes state K times, and the corresponding change time is C K Under the premise of i:j The joint probability of .

6. The abnormal monitoring data diagnosis method combined with prior physical information according to claim 5 is characterized in that: Combined with Bayesian inference theory, the number of change points K and the position of change points C are obtained K The posterior probability is obtained by repeated sampling to determine the most likely combination: Where, f(Y 1:N |K=k,C K ) is P k (Y 1:j ), f(K=k), f(C K |K=k) represents the prior estimate of the number of change points and the probability density of the change point positions. Through Markov chain sampling, the most likely value of K is obtained. When K increases compared with the previous observation time, the latest observation position is considered to be a new change point detected, that is, a sensor anomaly has occurred at this time.

7. An abnormal monitoring data diagnosis system combined with prior physical information, characterized in that: include: A calculation module is used to perform correlation calculation on historical monitoring data to obtain sensor combinations with high correlation; The judgment module is used to propose a priori assumptions about the correlation of sensor combinations with high correlation, that is, data anomalies are accidental events and it is impossible for multiple sensors to be abnormal at the same time; Based on the prior assumption of correlation, the correlation between multiple sensors is calculated in real time with a fixed time window, and anomalies are judged based on the calculation results; A construction module is used to analyze the typical characteristics of historical monitoring data of sensors that do not have a correlation relationship and establish an observation regression model based on prior physical information; The determination module is used to combine Bayesian theory with the observation regression model to make prior assumptions about the number and location of abnormal points in the observation area, and obtain the most likely value combination through Markov chain sampling to determine the number and location of abnormal points.

8. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by the processor, the abnormal monitoring data diagnosis method combined with prior physical information according to any one of claims 1 to 6 is implemented.

9. A computer device, characterized in that: It includes a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the abnormal monitoring data diagnosis method combined with prior physical information as described in any one of claims 1 to 6.

10. An electronic device, characterized in that: include: A processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to execute instructions for implementing the abnormal monitoring data diagnosis method combined with prior physical information as described in any one of claims 1 to 6.

Images