Railway knowledge graph data update frequency evaluation method based on life cycle prediction

Abstract

The present invention relates to the technical field of knowledge graph management, and in particular to a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction. The method comprises: obtaining target importance values ​​of each monitoring data sequence subset corresponding to each railway facility category according to each monitoring data subsequence in the monitoring data subsequence set, obtaining weight values ​​corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility according to the target importance values ​​of each monitoring data sequence subset; obtaining the predicted remaining life corresponding to each railway facility according to the weight value corresponding to each monitoring data sequence, and obtaining the future update frequency corresponding to each railway facility according to the predicted remaining life corresponding to each railway facility. The present invention can not only reduce resource consumption when updating the railway knowledge graph, but also ensure the efficiency and accuracy of railway management.

Description

Technical Field

[0001] The present invention relates to the field of knowledge graph management technology, and in particular to a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction. Background Art

[0002] The railway knowledge graph is an integrated information management system, also known as the railway infrastructure graph. It combines various data on railway facilities, such as data on railway facility performance, data on railway facility maintenance, and data on railway facility failures. Moreover, the railway knowledge graph makes railway management more efficient and accurate by integrating various data on railway facilities into a structured knowledge system or by associating various data on railway facilities.

[0003] Because data on railway facility performance, railway facility maintenance, and railway facility failures will continue to increase over time, after the railway infrastructure knowledge graph is constructed, the multi-dimensional data of each railway facility in the railway knowledge graph needs to be updated to ensure the efficiency and accuracy of real-time railway management based on the railway knowledge graph. The process of updating the multi-dimensional data of each railway facility in the railway knowledge graph belongs to the management process of the railway knowledge graph. Currently, when updating the multi-dimensional data of each railway facility in the railway knowledge graph, the update frequency of each railway facility is usually set to an empirical value. However, during the update process, if a higher update frequency is set for some railway facilities with a longer remaining service life and a lower update frequency is set for some railway facilities with a shorter remaining service life, not only will a large amount of resources, including computing resources, human resources, time resources, etc., be wasted or consumed, but the efficiency and accuracy of railway management will also be reduced. Therefore, in the process of updating the multi-dimensional data of each railway facility in the railway knowledge graph, how to obtain the update time or update frequency of various railway facilities is an urgent problem to be solved. Summary of the Invention

[0004] In order to solve the above problems, the present invention provides a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction. The technical solutions adopted are as follows:

[0005] One embodiment of the present invention provides a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction, comprising the following steps:

[0006] Obtain a monitoring data sequence set corresponding to each railway facility in each railway facility category in the railway facility category set corresponding to the railway knowledge graph, a monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set, and a tag value corresponding to each monitoring data sequence;

[0007] obtaining, based on the tag values ​​corresponding to the monitoring data sequences, subsets of the monitoring data sequences corresponding to the railway facility categories; obtaining, based on the monitoring data sequences in the monitoring data sequence subsets and the monitoring time periods corresponding to the monitoring data sequences, sets of monitoring data subsequences corresponding to the monitoring data sequences in the monitoring data sequence subsets;

[0008] Obtaining, according to each monitoring data subsequence in the monitoring data subsequence set, a target importance value of each monitoring data sequence subset corresponding to each railway facility category;

[0009] Obtaining, according to the target importance values ​​of the monitoring data sequence subsets, a weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility;

[0010] According to the weight values ​​corresponding to the monitoring data sequences, the predicted remaining life of the railway facilities is obtained, and according to the predicted remaining life of the railway facilities, the future update frequency of the railway facilities is obtained.

[0011] Beneficial effects: The present invention obtains each monitoring data sequence subset corresponding to each railway facility category based on the tag value corresponding to each acquired monitoring data sequence, and obtains a monitoring data subsequence set corresponding to each monitoring data sequence in each monitoring data sequence subset based on each monitoring data sequence in the acquired monitoring data sequence subset and the monitoring time period corresponding to each monitoring data sequence; then, according to each monitoring data subsequence in the monitoring data subsequence set, the target importance value of each monitoring data sequence subset corresponding to each railway facility category is obtained, and according to the target importance value of each monitoring data sequence subset, the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility is obtained; finally, according to the weight value corresponding to each monitoring data sequence, the predicted remaining life corresponding to each railway facility is obtained, and according to the predicted remaining life corresponding to each railway facility, the future update frequency corresponding to each railway facility is obtained. The present invention can more reliably predict the remaining life of each railway facility by obtaining the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility through the target importance value of each monitoring data sequence subset, that is, it can more reliably obtain the predicted remaining life of each railway facility. The future update frequency of each railway facility obtained based on the predicted remaining life of each railway facility can not only reduce the resource consumption when updating the railway knowledge graph, but also ensure the efficiency and accuracy of railway management. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. 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.

[0013] Figure 1 This is a flowchart of a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction according to the present invention. DETAILED DESCRIPTION

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field fall within the scope of protection of the embodiments of the present invention.

[0015] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0016] This embodiment provides a method for evaluating the update frequency of railway knowledge graph data based on life cycle prediction, which is described in detail as follows:

[0017] like Figure 1 As shown in FIG, the railway knowledge graph data update frequency evaluation method based on life cycle prediction includes the following steps:

[0018] Step S001: obtain a monitoring data sequence set corresponding to each railway facility in each railway facility category in the railway facility category set corresponding to the railway knowledge graph, a monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set, and a label value corresponding to each monitoring data sequence.

[0019] The main purpose of this embodiment is to determine the update time or update frequency of each railway facility based on the predicted remaining life of each railway facility during the process of updating the multi-dimensional data of each railway facility in the railway knowledge graph, thereby reducing the resource consumption when updating the railway knowledge graph while ensuring or improving the efficiency and accuracy of real-time railway management; in addition, the railway knowledge graph in this embodiment is a knowledge graph that has been constructed, that is, this embodiment does not involve the construction process of the railway knowledge graph, but is aimed at the update management process of the railway knowledge graph after the construction of the railway knowledge graph, that is, the process of determining the update frequency of the multi-dimensional data of each railway facility in the railway knowledge graph is the update management process of the railway knowledge graph.

[0020] In addition, for the convenience of analysis, this embodiment will subsequently analyze the update process of any completed railway knowledge graph as an example, that is, the railway knowledge graphs that appear subsequently in this embodiment all refer to the same railway knowledge graph, and all railway facilities that appear subsequently belong to the railway facilities in the same railway knowledge graph; therefore, this embodiment first records the set constructed by all railway facilities in the railway knowledge graph as a railway facility set, and there are railway facilities of repeated types in the railway facility set, such as there are multiple railway signal lights in the railway facility set; in addition, the railway facilities in this embodiment are all facilities that can be monitored.

[0021] Then, the types of all railway facilities appearing in the railway facility set are counted and recorded as the railway facility categories corresponding to the railway facility set. The set constructed by all railway facility categories corresponding to the railway facility set is recorded as the railway facility category set corresponding to the railway knowledge graph, and the types of all railway facilities in the railway facility category are the same. For example, all railway signal lights in the railway facility set can form a railway facility category.

[0022] This embodiment next needs to obtain the current monitoring time period corresponding to each railway facility in the railway facility set, and the current monitoring time period corresponding to each railway facility refers to the time period formed from the beginning of the corresponding railway facility being put into use to the current moment; then, in the current monitoring time period corresponding to each railway facility, all monitoring data corresponding to each railway facility and the collection time corresponding to all monitoring data are collected, and among all monitoring data corresponding to the railway facilities, there will be monitoring data with repeated collection time but inconsistent monitoring data types; therefore, in the current monitoring time period corresponding to each railway facility, all monitoring data corresponding to each railway facility and the collection time corresponding to all monitoring data can be obtained.

[0023] For any railway facility a0 in any railway facility category B0 in the railway facility category set: a set constructed by all monitoring data corresponding to the railway facility a0 obtained in the current monitoring time period corresponding to the railway facility a0 is recorded as the monitoring data set corresponding to the railway facility a0; and the types of all monitoring data in the monitoring data set corresponding to the railway facility a0 are statistically obtained and recorded as the monitoring data type corresponding to the railway facility a0, that is, the monitoring data type corresponding to the railway facility a0 monitored and collected is multidimensional; then, in the monitoring data set corresponding to the railway facility a0, a sequence constructed by monitoring data of the same monitoring data type is recorded as the monitoring data sequence corresponding to the railway facility a0, and the number of monitoring data sequences corresponding to the railway facility a0 is the same as the number of types of monitoring data types corresponding to the railway facility a0, that is, one monitoring data type can obtain one monitoring data sequence, and the monitoring data in the monitoring data sequence are arranged in chronological order of collection time; then, a set constructed by all monitoring data sequences corresponding to the railway facility a0 is recorded as the monitoring data sequence set corresponding to the railway facility a0; in addition, in this embodiment, the current monitoring time period corresponding to the railway facility a0 is used as the monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to the railway facility a0.

[0024] Therefore, according to the above method of obtaining the monitoring data sequence set corresponding to the railway facility a0, the monitoring data sequence set corresponding to each railway facility can be obtained.

[0025] In this embodiment, for any railway facility category, the number of monitoring data sequences in the monitoring data sequence set corresponding to all railway facilities in the railway facility category is the same, and the types of monitoring data represented by the monitoring data sequences located at the same position are consistent, that is, the types of monitoring data in the monitoring data sequences located at the same position are the same. For example, if the railway facility category refers to a traffic light, and the type of monitoring data in the xth monitoring data sequence in the monitoring data sequence set corresponding to any railway facility in the railway facility category refers to the luminous intensity of the railway facility, then the type of monitoring data in the xth monitoring data sequence in the monitoring data sequence set corresponding to the remaining railway facilities in the railway facility category all refers to the luminous intensity of the railway facility.

[0026] In addition, the monitoring data types in this embodiment include, but are not limited to, performance data, fault record data, and maintenance record data of railway facilities; and the performance data of facilities mainly refers to the technical performance indicators exhibited by the facilities during operation or use. These indicators reflect the efficiency and effectiveness of the facilities during use, and different types of railway facilities can monitor and collect different performance data. For example, the luminous intensity and operating voltage of signal lights on the railway belong to the performance data of signal lights, and the gauge, plane and elevation deviation of the tracks on the railway belong to the performance data of the tracks. This embodiment also requires that the monitoring data types monitored and collected by all railway facilities in the same railway facility category are the same. For example, the monitoring data types corresponding to all signal lights on the railway include luminous intensity, operating voltage, fault record data, and maintenance record data.

[0027] In this embodiment, the performance data of railway facilities are usually collected and acquired based on detection equipment, and the collection frequency is usually set to an empirical value, while the fault record data and maintenance record data are usually acquired manually.

[0028] In addition, based on the above description, it can be seen that the monitoring data types obtained in this embodiment also include the fault record data and maintenance record data of the railway facilities, and these record data are usually in the form of codes. For example, for any railway facility, all fault types and all maintenance types existing in the railway facility are obtained, and then all fault types existing in the railway facility are coded. Similarly, all maintenance types existing in the railway facility are also coded. If there are three types of all faults existing in the railway facility, then the codes of the three fault types existing in the railway facility are 1, 2, and 3 respectively. Similarly, if there are four types of all maintenance types existing in the railway facility, then the codes of the three fault types existing in the railway facility are 1, 2, 3, and 4 respectively. If a fault type coded as 1 occurs at a certain moment in the monitoring time period corresponding to the railway facility, then the fault record data at that moment is 1. Similarly, if maintenance is performed at a certain moment in the monitoring time period corresponding to the railway facility and the maintenance type is coded as 2, then the maintenance record data at that moment is 2.

[0029] Therefore, based on the above process, this embodiment can obtain the monitoring data sequence set corresponding to each railway facility in each railway facility category and the monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set. Next, this embodiment will obtain the tag value corresponding to each monitoring data sequence. The specific acquisition process is as follows:

[0030] For railway facility category B0: the monitoring data sequence sets corresponding to each railway facility in railway facility category B0 are all recorded as feature sequence sets corresponding to the railway facility category B0, the label values ​​corresponding to the c-th monitoring data sequence in each feature sequence set corresponding to railway facility category B0 are all recorded as c, and the types of monitoring data in the c-th monitoring data sequence in each feature sequence set are the same; that is, for all feature sequence sets corresponding to railway facility category B0, the label values ​​of monitoring data sequences at the same position in all feature sequence sets are the same, and the types of monitoring data in all monitoring data sequences with the same label values ​​are the same, that is, the monitoring data in the monitoring data sequences with the same label values ​​belong to data of the same dimension.

[0031] At this point, based on the above process, this embodiment can obtain the monitoring data sequence set corresponding to each railway facility in each railway facility category in the railway facility category set corresponding to the railway knowledge graph, the monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set, and the label value corresponding to each monitoring data sequence.

[0032] Step S002: obtaining, based on the tag value corresponding to each monitoring data sequence, each monitoring data sequence subset corresponding to each railway facility category; obtaining, based on each monitoring data sequence in the monitoring data sequence subset and the monitoring time period corresponding to each monitoring data sequence, each monitoring data sequence set corresponding to each monitoring data sequence in the monitoring data sequence subset; obtaining, based on each monitoring data subsequence in the monitoring data sequence set, a target importance value for each monitoring data sequence subset corresponding to each railway facility category; obtaining, based on the target importance value for each monitoring data sequence subset, a weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility.

[0033] Since the update frequency of each railway facility is usually set to an empirical value during the process of updating the multi-dimensional data of each railway facility in the railway knowledge graph, if a higher update frequency is set for some railway facilities with a longer remaining service life and a lower update frequency is set for some railway facilities with a shorter remaining service life during the process of updating the multi-dimensional monitoring data of each railway facility in the railway knowledge graph, not only a large amount of resources, including computing resources, human resources, time resources, etc., will be wasted or consumed, but the efficiency and accuracy of real-time railway management will also be reduced. Because in order to reduce resource consumption during updates and to ensure the efficiency and accuracy of railway management, this embodiment will next predict the remaining service life of each railway facility based on the monitoring data sequence set corresponding to each railway facility in each railway facility category obtained above, that is, determine the predicted remaining service life of each railway facility, and determine the next update time or future update frequency of each railway facility based on the predicted remaining service life of each railway facility. The next update time or future update frequency of each railway facility determined based on the predicted remaining service life of each railway facility can not only reduce resource consumption when updating the railway knowledge graph, but also ensure the efficiency and accuracy of railway management.

[0034] Because the monitoring data corresponding to each railway facility in this implementation is multidimensional, when predicting the remaining life of each railway facility in the future, the remaining life is usually predicted based on all the monitoring data sequences corresponding to the railway facilities, and currently, the remaining life of each railway facility is usually predicted based on the weighted linear regression remaining life prediction model. In addition, because in the current process of training the weighted linear regression remaining life prediction model, the weight values ​​of all monitoring data sequences are usually set to the same, but the importance or influence of monitoring data of different dimensions on the remaining life prediction is different, that is, the importance or influence of monitoring data sequences of different monitoring data types on the remaining life prediction is different, so in the process of training the weighted linear regression remaining life prediction model, if all the monitoring data sequences participating in the training are used as the weights of the remaining life prediction model, the remaining life of each railway facility is usually predicted based on the weighted linear regression remaining life prediction model. The weight values ​​of the trained monitoring data sequences are all the same, which may result in the monitoring data information corresponding to the monitoring data types with greater importance or influence on the remaining life prediction being ignored, thereby resulting in poor performance or generalization ability of the trained model. Therefore, in order to further ensure the accuracy of the subsequent remaining life prediction of railway facilities, this embodiment will analyze the importance or influence of monitoring data sequences of different monitoring data types on the remaining life prediction to obtain the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility. Subsequently, the weighted linear regression remaining life prediction model will be trained based on the weight value and the monitoring data sequence. Therefore, the specific process of obtaining the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in this embodiment is as follows:

[0035] This embodiment first needs to obtain each monitoring data sequence subset corresponding to each railway facility category based on the tag value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in each railway facility category. The purpose of obtaining each monitoring data sequence subset is to facilitate the subsequent acquisition of the target importance value of the monitoring data sequence subset. The specific acquisition process is as follows:

[0036] For railway facility category B0 in the railway facility category set:

[0037] The new set constructed from the set of monitoring data sequences corresponding to all railway facilities in railway facility category B0 is recorded as the comprehensive monitoring data sequence set corresponding to railway facility category B0; the sets constructed from all monitoring data sequences with the same tag value in the comprehensive monitoring data sequence set are all recorded as monitoring data sequence subsets corresponding to railway facility category B0, and the tag values ​​of all monitoring data sequences in any monitoring data sequence subset corresponding to railway facility category B0 are the same, and the number of monitoring data sequence subsets corresponding to railway facility category B0 is the same as the number of types of tag values ​​in the tag values ​​corresponding to all monitoring data sequences in the comprehensive monitoring data sequence set.

[0038] Therefore, through the above process, each monitoring data sequence subset corresponding to each railway facility category can be obtained.

[0039] Then, based on each monitoring data sequence in each monitoring data sequence subset corresponding to each railway facility category and the monitoring time period corresponding to each monitoring data sequence, a monitoring data subsequence set corresponding to each monitoring data sequence in each monitoring data sequence subset corresponding to each railway facility category is obtained. The monitoring data subsequence set is an important basis for subsequently obtaining the target importance value of each monitoring data sequence subset. Therefore, the specific method for obtaining the monitoring data subsequence set corresponding to each monitoring data sequence is:

[0040] A preset time length is set, and in specific applications, the implementer needs to set the preset time length according to actual conditions. Since the monitoring data types in this embodiment include fault record data and maintenance record data, and the fault record data and maintenance record data are usually sparse, this embodiment requires that the preset time length be set longer in order to facilitate more reliable acquisition of the target importance value in the future, that is, the value of the preset time length is larger. For example, the value of the preset time length can be set to 1 month in this embodiment.

[0041] The monitoring time periods corresponding to the monitoring data sequences are then divided using preset time lengths. Based on the division results, a monitoring data subsequence set corresponding to each monitoring data sequence is obtained. For ease of understanding, this embodiment will now describe the process of obtaining a monitoring data subsequence set corresponding to the monitoring data sequence A0 in the monitoring data sequence subset C0 corresponding to the railway facility category B0 as an example. Therefore, the specific method for obtaining the monitoring data subsequence set corresponding to the monitoring data sequence A0 is as follows:

[0042] For monitoring time period T0 corresponding to monitoring data sequence A0, first, starting from the start time of monitoring time period T0, monitoring time period T0 is evenly divided into preset time lengths to obtain all monitoring sub-time periods corresponding to monitoring time period T0. For any monitoring sub-time period t0 corresponding to monitoring time period T0: In monitoring data sequence A0, the sequence constructed by all monitoring data with the collection time within monitoring sub-time period t0 is recorded as the monitoring data subsequence corresponding to monitoring sub-time period t0.

[0043] In addition, when the monitoring time period T0 is evenly divided using the preset time length, it may happen that the time length of the sub-time period including the last moment in the monitoring time period T0 is less than the preset time length. In this embodiment, the sub-time period with a time length less than the preset time length is used as the monitoring sub-time period corresponding to the monitoring time period T0.

[0044] Therefore, based on the above process, the monitoring data subsequences corresponding to all monitoring sub-time periods corresponding to the monitoring time period T0 can be obtained. In this embodiment, the subsequence set constructed by the monitoring data subsequences corresponding to all monitoring sub-time periods corresponding to the monitoring time period T0 is recorded as the monitoring data subsequence set corresponding to the monitoring data sequence A0.

[0045] In this embodiment, the monitoring data subsequence sets corresponding to the monitoring data sequence A0 can be obtained according to the above method. After obtaining the monitoring data subsequence sets corresponding to the monitoring data sequences, this embodiment will analyze each monitoring data subsequence in the monitoring data subsequence set to determine the target importance value of each monitoring data sequence subset corresponding to each railway facility category. The target importance value is an important indicator in the subsequent determination of the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility. Therefore, the specific process of obtaining the target importance value of each monitoring data sequence subset in this embodiment is as follows:

[0046] For the monitoring data sequence subset C0 corresponding to the railway facility category B0:

[0047] First, any two monitoring data sequences in the monitoring data sequence subset C0 are randomly permuted and combined without repetition to obtain all monitoring data sequence combinations corresponding to the monitoring data sequence subset C0, and the correlation representation value corresponding to each monitoring data sequence combination corresponding to the monitoring data sequence subset C0 is obtained. The non-repetition permutation and combination is a well-known technique and will not be described in detail. In this embodiment, the specific process of obtaining the correlation representation value corresponding to each monitoring data sequence combination corresponding to the monitoring data sequence subset C0 is as follows:

[0048] For any monitoring data sequence combination F corresponding to the monitoring data sequence subset C0:

[0049] First, the two monitoring data sequences in the monitoring data sequence combination F are respectively recorded as the first sequence and the second sequence, and it is determined whether the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the first sequence is not greater than the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the second sequence. If so, the first sequence is recorded as the first monitoring data sequence of the monitoring data sequence combination F, and the second sequence is recorded as the second monitoring data sequence of the monitoring data sequence combination F. Otherwise, the second sequence is recorded as the first monitoring data sequence of the monitoring data sequence combination F, and the first sequence is recorded as the second monitoring data sequence of the monitoring data sequence combination F, that is, the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the first monitoring data sequence is not greater than the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the second monitoring data sequence.

[0050] Then, based on the monitoring data subsequence set corresponding to the first monitoring data sequence of the monitoring data sequence combination F and the monitoring data subsequence set corresponding to the second monitoring data sequence of the monitoring data sequence combination F, the monitoring data subsequence pairs corresponding to each monitoring data subsequence in the monitoring data subsequence set corresponding to the first monitoring data sequence are obtained. Thereafter, the monitoring data subsequence pairs corresponding to all monitoring data subsequences in the monitoring data subsequence set corresponding to the first monitoring data sequence are recorded as the monitoring data subsequence pairs corresponding to the monitoring data sequence combination F.

[0051] This embodiment will then obtain, based on the monitoring data subsequence set corresponding to the first monitoring data sequence of the monitoring data sequence combination F and the monitoring data subsequence set corresponding to the second monitoring data sequence of the monitoring data sequence combination F, monitoring data subsequence pairs corresponding to each monitoring data subsequence in the monitoring data subsequence set corresponding to the first monitoring data sequence. For ease of understanding, this embodiment will be described below using the process of obtaining the monitoring data subsequence pair corresponding to the i-th monitoring data subsequence in the monitoring data subsequence set D1 corresponding to the first monitoring data sequence as an example. Therefore, the specific process of obtaining the monitoring data subsequence pair corresponding to the i-th monitoring data subsequence is as follows:

[0052] First, the monitoring data subsequence set corresponding to the second monitoring data sequence of the monitoring data sequence combination F is recorded as the monitoring data subsequence set to be matched corresponding to the i-th monitoring data subsequence, and all monitoring data subsequences in the monitoring data subsequence set to be matched are recorded as monitoring data subsequences to be matched.

[0053] Then, a target difference value corresponding to each to-be-matched monitoring data subsequence in the to-be-matched monitoring data subsequence set is obtained, and the target difference value corresponding to each to-be-matched monitoring data subsequence can reflect the difference between the i-th monitoring data subsequence and the corresponding to-be-matched monitoring data subsequence. Subsequently, a monitoring data subsequence pair corresponding to the i-th monitoring data subsequence is determined based on the target difference value. For ease of understanding, this embodiment will be described by taking the process of obtaining the target difference value corresponding to the j-th to-be-matched monitoring data subsequence in the to-be-matched monitoring data subsequence set as an example. Therefore, the specific process of obtaining the target difference value corresponding to the j-th to-be-matched monitoring data subsequence in this embodiment is as follows:

[0054] If it is determined that the types of monitoring data in all monitoring data sequences in the monitoring data sequence combination F all belong to performance data of railway facilities, then the DTW distance between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is used as the target difference value corresponding to the j-th monitoring data to be matched; and the DTW distance between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched can reflect the difference or similarity between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched, and a larger DTW distance value indicates that the difference between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is greater, and also indicates that the target difference value corresponding to the j-th monitoring data subsequence to be matched is greater, and the larger the target difference value corresponding to the j-th monitoring data subsequence to be matched is, the greater the possibility that the j-th monitoring data subsequence to be matched is the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence, and also indicates that the possibility that the combination of the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is a monitoring data subsequence pair is greater.

[0055] In addition, the process of obtaining the DTW distance between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched using the DTW algorithm is a well-known technology, and thus will not be described in detail in this embodiment.

[0056] If it is determined that the monitoring data types in all monitoring data sequences in the monitoring data sequence combination F do not belong to the performance data of railway facilities, the first eigenvalue corresponding to the i-th monitoring data subsequence and the first eigenvalue corresponding to the j-th monitoring data to be matched are obtained, and then the absolute value of the difference between the first eigenvalue corresponding to the j-th monitoring data subsequence to be matched and the first eigenvalue corresponding to the i-th monitoring data subsequence is calculated and recorded as the characteristic index value, and then the value obtained by performing negative correlation mapping on the characteristic index value is recorded as the difference between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched. Mapping value, then calculate the difference between the preset first constant and the mapping value, and record it as the first difference value between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched, then calculate the absolute value of the difference between the total number of monitoring data in the j-th monitoring data subsequence to be matched and the total number of monitoring data in the i-th monitoring data subsequence, and record it as the quantity difference value between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched; then obtain the product of the first difference value and the quantity difference value, and record it as the target difference value corresponding to the j-th monitoring data subsequence to be matched.

[0057] In this embodiment, the types of monitoring data in all monitoring data sequences in the monitoring data sequence combination F do not belong to the performance data of railway facilities, which means that the types of monitoring data in all monitoring data sequences in the monitoring data sequence combination F may all belong to fault record data or maintenance record data.

[0058] And when the quantity difference value and the first difference value between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched are larger, it indicates that the target difference value corresponding to the j-th monitoring data subsequence to be matched is larger, and the target difference value corresponding to the j-th monitoring data subsequence to be matched is larger, it indicates that the possibility that the j-th monitoring data subsequence to be matched is the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence is larger, and it also indicates that the possibility that the combination of the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is a monitoring data subsequence pair is larger; and when the monitoring data types in all monitoring data sequences in the monitoring data sequence combination F are When the types do not belong to the performance data of railway facilities but to fault record data or maintenance record data, the reason for considering the quantity difference value and the first difference value when obtaining the target difference value corresponding to the subsequence of monitoring data to be matched is that the time when two railway facilities of the same type start to be put into use or the frequency of use may be different, and then the number of monitoring data in the j-th subsequence of monitoring data to be matched and the i-th subsequence of monitoring data may be the same, but the location distribution of the monitoring data may be different. Therefore, in order to ensure the reliability of the target difference value corresponding to the obtained subsequence of monitoring data to be matched, it is necessary to consider the quantity difference value and the first difference value.

[0059] In addition, in specific applications, the implementer may set the value of the preset first constant according to actual conditions. For example, in this embodiment, the value of the preset first constant is set to be 1.

[0060] The method for obtaining the first eigenvalue corresponding to the i-th monitoring data subsequence is: obtaining the characteristic difference value sequence corresponding to the i-th monitoring data subsequence, and the h-th characteristic difference value in the characteristic difference value sequence is the time difference between the acquisition time corresponding to the h+1-th monitoring data in the i-th monitoring data subsequence and the acquisition time corresponding to the h-th monitoring data in the i-th monitoring data subsequence, and then obtaining the product of the mean of the characteristic difference value sequence and the variance of the characteristic difference value sequence, and recording it as the first eigenvalue corresponding to the i-th monitoring data subsequence.

[0061] In this embodiment, the method for obtaining the first characteristic value corresponding to the jth monitoring data to be matched is the same as the method for obtaining the first characteristic value corresponding to the i-th monitoring data subsequence, and thus will not be described in detail.

[0062] In addition, in this embodiment, the specific expression for calculating the first difference value between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is:

[0063]

[0064] in, is the first difference between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched, exp() is an exponential function with a constant e as the base, is the variance of the characteristic difference sequence corresponding to the i-th monitoring data subsequence, is the total number of characteristic difference values ​​in the characteristic difference sequence corresponding to the i-th monitoring data subsequence, is the nth characteristic difference value in the characteristic difference value sequence corresponding to the i-th monitoring data subsequence, is the variance of the characteristic difference sequence corresponding to the jth monitoring data subsequence to be matched, is the total number of characteristic difference values ​​in the characteristic difference value sequence corresponding to the jth monitoring data subsequence to be matched, is the mth feature difference value in the feature difference value sequence corresponding to the jth monitoring data subsequence to be matched, is the first eigenvalue corresponding to the i-th monitoring data subsequence, is the first eigenvalue corresponding to the jth monitoring data subsequence to be matched.

[0065] In the above formula, when the difference between the first eigenvalue corresponding to the i-th monitoring data subsequence and the first eigenvalue corresponding to the j-th monitoring data subsequence to be matched is greater, it indicates that The larger the value of The larger the value of , the more dissimilar the position distribution of the monitoring data in the i-th monitoring data subsequence is to the j-th monitoring data subsequence to be matched. On the contrary, the smaller the difference between the first eigenvalue corresponding to the i-th monitoring data subsequence and the first eigenvalue corresponding to the j-th monitoring data subsequence to be matched is, the smaller the difference between the first eigenvalue corresponding to the j-th monitoring data subsequence to be matched is. The smaller the value of The smaller the value of , the more similar the position distribution of the monitoring data in the i-th monitoring data subsequence is to that in the j-th monitoring data subsequence to be matched.

[0066] Therefore, through the above process, the target difference values ​​corresponding to all the to-be-matched monitoring data subsequences in the to-be-matched monitoring data subsequence set can be obtained, and then the set constructed by the target difference values ​​corresponding to all the to-be-matched monitoring data subsequence sets in the to-be-matched monitoring data subsequence set is recorded as the target difference value set corresponding to the to-be-matched monitoring data subsequence set; and the to-be-matched monitoring data subsequence corresponding to the minimum target difference value in the target difference value set is selected and recorded as the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence, and the combination of the i-th monitoring data subsequence and the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence is recorded as the monitoring data subsequence pair corresponding to the i-th monitoring data subsequence.

[0067] Therefore, this embodiment obtains the monitoring data subsequence pair corresponding to the i-th monitoring data subsequence in the monitoring data subsequence set D1 corresponding to the first monitoring data sequence of the monitoring data sequence combination F through the above process. Therefore, according to the above method for obtaining the monitoring data subsequence pair corresponding to the i-th monitoring data subsequence, it is possible to obtain the monitoring data subsequence pairs corresponding to all monitoring data subsequences in the monitoring data subsequence set D1 corresponding to the first monitoring data sequence. In this embodiment, the monitoring data subsequence pairs corresponding to all monitoring data subsequences in the monitoring data subsequence set D1 corresponding to the first monitoring data sequence are recorded as the monitoring data subsequence pairs corresponding to the monitoring data sequence combination F.

[0068] Therefore, this embodiment obtains all monitoring data subsequence pairs corresponding to the monitoring data sequence combination F through the above process. After obtaining all monitoring data subsequence pairs corresponding to the monitoring data sequence combination F, this embodiment also needs to obtain the time difference value corresponding to each monitoring data subsequence pair. The time difference value corresponding to the monitoring data subsequence pair is an important indicator for subsequently determining the correlation representation value corresponding to the monitoring data sequence combination F. Therefore, the specific method for obtaining the time difference value corresponding to each monitoring data subsequence pair corresponding to the monitoring data sequence combination F in this embodiment is:

[0069] For any monitoring data subsequence pair G corresponding to the monitoring data sequence combination F, the monitoring sub-time periods corresponding to the two monitoring data subsequences in the monitoring data subsequence pair G are respectively recorded as the first monitoring sub-time period and the second monitoring sub-time period, and the absolute value of the difference between the starting time in the first monitoring sub-time period and the starting time in the second monitoring sub-time period is recorded as the time difference value corresponding to the monitoring data subsequence pair G.

[0070] Therefore, through the above process, the time difference values ​​corresponding to each monitoring data subsequence pair corresponding to the monitoring data sequence combination F can be obtained, and then a set of the time difference values ​​corresponding to all monitoring data subsequence pairs corresponding to the monitoring data sequence combination F is constructed, and the constructed set is recorded as the time difference value set corresponding to the monitoring data sequence combination F. After that, the normalized value of the information entropy of the time difference value set is calculated and recorded as the correlation representation value corresponding to the monitoring data sequence combination F; and in this embodiment, when the value of the information entropy of the time difference value set is larger, it indicates that the correlation representation value corresponding to the monitoring data sequence combination F is larger, and the larger the correlation representation value corresponding to the monitoring data sequence combination F, the smaller the correlation between the two monitoring data sequences in the monitoring data sequence combination F, and vice versa, it indicates that the correlation between the two monitoring data sequences in the monitoring data sequence combination F is larger.

[0071] Therefore, through the above process, the correlation characterization values ​​corresponding to each monitoring data sequence combination corresponding to the monitoring data sequence subset C0 corresponding to the railway facility category B0 can be obtained, and then the mean of the correlation characterization values ​​corresponding to all monitoring data sequence combinations corresponding to the monitoring data sequence subset C0 can be obtained, and recorded as the initial importance value corresponding to the monitoring data sequence subset C0. The larger the value of the correlation characterization value corresponding to all monitoring data sequence combinations corresponding to the monitoring data sequence subset C0, the greater the initial importance value corresponding to the monitoring data sequence subset C0. The larger the initial importance value corresponding to the monitoring data sequence subset C0, the greater the weight of the monitoring data sequence in the monitoring data sequence subset C0 in the subsequent training process of the weighted linear regression prediction model.

[0072] Therefore, according to the above process, this embodiment can obtain the initial importance value corresponding to each monitoring data sequence subset corresponding to the railway facility category B0. Next, the initial importance value corresponding to each monitoring data sequence subset corresponding to the railway facility category B0 is normalized, and the normalized value is recorded as the target importance value. That is, the size of the target importance value is related to the weight of the monitoring data sequence in the monitoring data sequence subset, specifically:

[0073] For the railway facility category B0, the cumulative sum of the initial importance values ​​corresponding to all monitoring data sequence subsets corresponding to the railway facility category B0 is recorded as the comprehensive representation value corresponding to the railway facility category B0, and the ratio of the initial importance value corresponding to each monitoring data sequence subset corresponding to the railway facility category B0 to the comprehensive representation value is recorded as the target importance value of the corresponding monitoring data sequence subset.

[0074] Based on the above process, it can be seen that the size of the target importance value is related to the weight of the monitoring data sequence in the monitoring data sequence subset. Therefore, this embodiment will obtain the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility based on the target importance value of each monitoring data sequence subset corresponding to the railway facility category B0. The specific acquisition method is:

[0075] For the wth monitoring data sequence in the monitoring data sequence set corresponding to the sth railway facility in the railway facility category B0, if the wth monitoring data sequence is included in the rth monitoring data sequence subset corresponding to the railway facility category B0, the target importance value of the rth monitoring data sequence subset is used as the weight value corresponding to the wth monitoring data sequence.

[0076] Therefore, through the above process, the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in each railway facility category can be obtained.

[0077] Step S003, obtaining the predicted remaining life of each railway facility according to the weight value corresponding to each monitoring data sequence, and obtaining the future update frequency corresponding to each railway facility according to the predicted remaining life of each railway facility.

[0078] Next, this embodiment will obtain the predicted remaining life of each railway facility based on the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in each railway facility category, and obtain the future update frequency corresponding to each railway facility based on the predicted remaining life of each railway facility. For ease of understanding, this embodiment will be described below using the process of obtaining the predicted remaining life of each railway facility in the railway facility category B0 as an example. Therefore, the specific process of obtaining the predicted remaining life of each railway facility in the railway facility category B0 is as follows:

[0079] First, a weighted linear regression remaining life prediction model corresponding to the railway facility category B0 is constructed, and the weighted linear regression remaining life prediction model corresponding to the railway facility category B0 is trained according to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0 and the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0, to obtain a trained prediction model, which is recorded as the target remaining life prediction model corresponding to the railway facility category B0, and each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0 is input into the target remaining life prediction model corresponding to the railway facility category B0, and the predicted remaining life corresponding to each railway facility in the railway facility category B0 is output.

[0080] In this embodiment, the training process of the weighted linear regression remaining life prediction model is a well-known technology, so it will not be described in detail in this embodiment.

[0081] Then, based on the predicted remaining life of each railway facility in the railway facility category B0, the future update frequency of each railway facility in the railway facility category B0 is obtained, specifically:

[0082] For the s-th railway facility in the railway facility category B0: obtain the rated life corresponding to the s-th railway facility, and then obtain the ratio of the predicted remaining life corresponding to the s-th railway facility to the rated life corresponding to the s-th railway facility, and record it as the judgment index corresponding to the s-th railway facility.

[0083] In this embodiment, the rated life of the railway facilities refers to the service life of the corresponding railway facilities under normal use, and usually the rated life of the railway facilities is set by the railway facility manufacturer; in addition, this embodiment requires that the rated life be consistent with the time unit of the predicted remaining life.

[0084] Based on the acquisition process of the judgment index, it can be seen that the value interval of the judgment index is [0,1], which is recorded as the judgment index value interval. Then, the judgment index value interval is divided to obtain the judgment index value sub-interval corresponding to the judgment index value interval. Then, the minimum value in the judgment index value sub-interval is used as the sorting value of the corresponding judgment index value sub-interval. Then, all judgment index value sub-intervals corresponding to the judgment index value interval are sorted in order from small to large according to the sorting value to obtain a judgment index value sub-interval sequence. Then, the update frequency of each judgment index value sub-interval sequence in the judgment index value sub-interval sequence is obtained, and in the judgment index value sub-interval sequence, the earlier the judgment index value sub-interval is, the smaller the update frequency is.

[0085] In this embodiment, the implementer needs to set the division rules for dividing the judgment indicator value interval according to actual conditions, for example, equal division or unequal division can be performed, and the update frequency of each judgment indicator value sub-interval sequence needs to be set according to actual conditions. For example, in this embodiment, the judgment indicator value interval [0,1] can be divided into three judgment indicator value sub-intervals, namely [0,0.3), [0.3,0.7) and [0.7,1], and the update frequency of the judgment indicator value sub-interval [0,0.3) can be set to 10 days, the update frequency of the judgment indicator value sub-interval [0.3,0.7) can be set to 30 days, and the update frequency of the judgment indicator value sub-interval [0.7,1] can be set to 40 days.

[0086] Then, in the preset judgment indicator interval sequence, the update frequency of the judgment indicator value sub-interval containing the judgment indicator corresponding to the s-th railway facility is obtained and used as the future update frequency corresponding to the s-th railway facility, and the future update frequency corresponding to the s-th railway facility refers to the next update time corresponding to the s-th railway facility; for example, if the judgment indicator corresponding to the s-th railway facility belongs to the judgment indicator value sub-interval [0.3, 0.7), then the future update frequency corresponding to the s-th railway facility is 30 days, and the time interval between the most recent update time of the s-th railway facility and the next update time of the s-th railway facility is 30 days, and when updating, all monitoring data of the s-th railway facility collected in the time period between the most recent update time and the next update time of the s-th railway facility are updated to or stored in the railway knowledge graph.

[0087] Therefore, this embodiment can obtain the future update frequencies corresponding to all railway facilities in the railway facility category B0 based on the above process of obtaining the future update frequency corresponding to the s-th railway facility, and can also obtain the future update frequencies corresponding to each railway facility in other railway facility categories based on the above method of obtaining the future update frequencies corresponding to all railway facilities in the railway facility category B0.

[0088] To summarize, based on the tag value corresponding to each acquired monitoring data sequence, each monitoring data sequence subset corresponding to each railway facility category is obtained, and based on each monitoring data sequence in the acquired monitoring data sequence subset and the monitoring time period corresponding to each monitoring data sequence, a monitoring data subsequence set corresponding to each monitoring data sequence in each monitoring data sequence subset is obtained; then, based on each monitoring data subsequence in the monitoring data subsequence set, the target importance value of each monitoring data sequence subset corresponding to each railway facility category is obtained, and based on the target importance value of each monitoring data sequence subset, the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility is obtained; finally, based on the weight value corresponding to each monitoring data sequence, the predicted remaining life corresponding to each railway facility is obtained, and based on the predicted remaining life corresponding to each railway facility, the future update frequency corresponding to each railway facility is obtained. This embodiment can more reliably predict the remaining life of each railway facility by obtaining the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility through the target importance value of each monitoring data sequence subset, that is, can more reliably obtain the predicted remaining life of each railway facility. The future update frequency of each railway facility obtained based on the predicted remaining life of each railway facility can not only reduce the resource consumption when updating the railway knowledge graph, but also ensure the efficiency and accuracy of railway management.

[0089] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A railway knowledge graph data update frequency evaluation method based on life cycle prediction, characterized in that: The method comprises the following steps: Obtain a set of monitoring data sequences corresponding to each railway facility in each railway facility category in the railway facility category set corresponding to the railway knowledge graph, a monitoring time period corresponding to each monitoring data sequence in the monitoring data sequence set, and a tag value corresponding to each monitoring data sequence, wherein the monitoring data in all monitoring data sequences with the same tag value are of the same type; obtaining, based on the tag values ​​corresponding to the monitoring data sequences, subsets of the monitoring data sequences corresponding to the railway facility categories; obtaining, based on the monitoring data sequences in the monitoring data sequence subsets and the monitoring time periods corresponding to the monitoring data sequences, sets of monitoring data subsequences corresponding to the monitoring data sequences in the monitoring data sequence subsets; Obtaining, according to each monitoring data subsequence in the monitoring data subsequence set, a target importance value of each monitoring data sequence subset corresponding to each railway facility category; Obtaining, according to the target importance values ​​of the monitoring data sequence subsets, a weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility; Obtaining, based on the weight values ​​corresponding to the monitoring data sequences, the predicted remaining life of each railway facility, and obtaining, based on the predicted remaining life of each railway facility, the future update frequency of each railway facility; The method for obtaining the target importance value of each monitoring data sequence subset corresponding to each railway facility category includes: For any monitoring data sequence subset C0 corresponding to any railway facility category B0: perform non-repeated permutations and combinations on any two monitoring data sequences in the monitoring data sequence subset C0 to obtain all monitoring data sequence combinations corresponding to the monitoring data sequence subset C0, and obtain the correlation representation value corresponding to each monitoring data sequence combination corresponding to the monitoring data sequence subset C0; record the average of the correlation representation values ​​corresponding to all monitoring data sequence combinations corresponding to the monitoring data sequence subset C0 as the initial importance value corresponding to the monitoring data sequence subset C0; For the railway facility category B0, the cumulative sum of the initial importance values ​​corresponding to all monitoring data sequence subsets corresponding to the railway facility category B0 is recorded as the comprehensive representation value corresponding to the railway facility category B0, and the ratio of the initial importance value corresponding to each monitoring data sequence subset corresponding to the railway facility category B0 to the comprehensive representation value is recorded as the target importance value of the corresponding monitoring data sequence subset; The method for obtaining the correlation characterization value corresponding to each combination of monitoring data sequences includes: For any monitoring data sequence combination F corresponding to the monitoring data sequence subset C0: The two monitoring data sequences in the monitoring data sequence combination F are respectively recorded as the first monitoring data sequence and the second monitoring data sequence of the monitoring data sequence combination F, wherein the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the first monitoring data sequence is not greater than the number of monitoring data subsequences in the monitoring data subsequence set corresponding to the second monitoring data sequence; obtaining, based on the monitoring data subsequence set corresponding to the first monitoring data sequence and the monitoring data subsequence set corresponding to the second monitoring data sequence, monitoring data subsequence pairs corresponding to each monitoring data subsequence in the monitoring data subsequence set corresponding to the first monitoring data sequence, and recording the monitoring data subsequence pairs corresponding to all monitoring data subsequences in the monitoring data subsequence set corresponding to the first monitoring data sequence as monitoring data subsequence pairs corresponding to the monitoring data sequence combination F; For any monitoring data subsequence pair G corresponding to the monitoring data sequence combination F, the monitoring sub-time periods corresponding to the two monitoring data subsequences in the monitoring data subsequence pair G are respectively recorded as the first monitoring sub-time period and the second monitoring sub-time period, and the absolute value of the difference between the start time of the first monitoring sub-time period and the start time of the second monitoring sub-time period is recorded as the time difference value corresponding to the monitoring data subsequence pair G; a set constructed from the time difference values ​​corresponding to all monitoring data subsequence pairs corresponding to the monitoring data sequence combination F is recorded as the time difference value set corresponding to the monitoring data sequence combination F, and the normalized value of the information entropy of the time difference value set is recorded as the correlation representation value corresponding to the monitoring data sequence combination F; The method for obtaining the railway facility category set corresponding to the railway knowledge graph includes: recording a set constructed by all railway facilities in the railway knowledge graph as a railway facility set, obtaining statistics of all railway facility categories corresponding to the railway facility set, and recording a set constructed by all railway facility categories corresponding to the railway facility set as a railway facility category set corresponding to the railway knowledge graph; A method for obtaining a set of monitoring data sequences corresponding to any railway facility a0 includes: obtaining all monitoring data corresponding to the railway facility a0 in a current monitoring time period corresponding to the railway facility a0, and recording a set constructed from all monitoring data corresponding to the railway facility a0 as the monitoring data set corresponding to the railway facility a0; obtaining statistically the types of all monitoring data in the monitoring data set corresponding to the railway facility a0, and recording them all as the monitoring data types corresponding to the railway facility a0; in the monitoring data set corresponding to the railway facility a0, recording a sequence constructed from monitoring data of the same monitoring data type as the monitoring data sequence corresponding to the railway facility a0, the monitoring data in the monitoring data sequence being arranged in chronological order of collection time; and recording a set constructed from all monitoring data sequences corresponding to the railway facility a0 as the monitoring data sequence set corresponding to the railway facility a0.

2. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 1 is characterized in that: The method for obtaining subsets of monitoring data sequences corresponding to the railway facility categories according to the tag values ​​corresponding to the monitoring data sequences includes: For the railway facility category B0: a new set constructed by the set of monitoring data sequences corresponding to all railway facilities in the railway facility category B0 is recorded as the comprehensive monitoring data sequence set corresponding to the railway facility category B0; a set constructed by all monitoring data sequences with the same label value in the comprehensive monitoring data sequence set is recorded as a subset of the monitoring data sequence corresponding to the railway facility category B0.

3. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 1 is characterized in that: The method for obtaining a monitoring data subsequence set corresponding to each monitoring data sequence in each monitoring data sequence subset comprises: For the monitoring time period T0 corresponding to any monitoring data sequence A0 in any monitoring data sequence subset C0 corresponding to the railway facility category B0: The monitoring time period T0 is evenly divided to obtain all monitoring sub-time periods corresponding to the monitoring time period T0; for any monitoring sub-time period t0 corresponding to the monitoring time period T0, in the monitoring data sequence A0, a sequence constructed by all monitoring data whose collection time falls within the monitoring sub-time period t0 is recorded as the monitoring data subsequence corresponding to the monitoring sub-time period t0; For the monitoring data sequence A0, a subsequence set constructed by monitoring data subsequences corresponding to all monitoring sub-time periods corresponding to the monitoring time period T0 is recorded as the monitoring data subsequence set corresponding to the monitoring data sequence A0.

4. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 1 is characterized in that: The method for obtaining a monitoring data subsequence pair corresponding to each monitoring data subsequence in a monitoring data subsequence set corresponding to the first monitoring data sequence includes: For the i-th monitoring data subsequence in the monitoring data subsequence set D1 corresponding to the first monitoring data sequence: Recording the set of monitoring data subsequences corresponding to the second monitoring data sequence as the set of monitoring data subsequences to be matched corresponding to the i-th monitoring data subsequence; Obtain a target difference value corresponding to each to-be-matched monitoring data subsequence in the to-be-matched monitoring data subsequence set, and record a set constructed from the target difference values ​​corresponding to all to-be-matched monitoring data subsequences in the to-be-matched monitoring data subsequence set as a target difference value set; select the to-be-matched monitoring data subsequence corresponding to the minimum target difference value in the target difference value set as the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence; and record a combination of the i-th monitoring data subsequence and the matching monitoring data subsequence corresponding to the i-th monitoring data subsequence as a monitoring data subsequence pair corresponding to the i-th monitoring data subsequence.

5. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 4 is characterized in that: The method for obtaining the target difference value corresponding to each to-be-matched monitoring data subsequence in the to-be-matched monitoring data subsequence set includes: For the jth monitoring data subsequence to be matched in the set of monitoring data subsequences to be matched: If it is determined that the monitoring data types in all monitoring data sequences in the monitoring data sequence combination F belong to the performance data of railway facilities, the DTW distance between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched is used as the target difference value corresponding to the j-th monitoring data to be matched; If it is determined that the types of monitoring data in all monitoring data sequences in the monitoring data sequence combination F do not belong to performance data of railway facilities, then a first eigenvalue corresponding to the i-th monitoring data subsequence and a first eigenvalue corresponding to the j-th monitoring data to be matched are obtained, and a method for obtaining the first eigenvalue corresponding to the j-th monitoring data to be matched is the same as a method for obtaining the first eigenvalue corresponding to the i-th monitoring data subsequence. The absolute value of the difference between the first eigenvalue corresponding to the j-th monitoring data subsequence to be matched and the first eigenvalue corresponding to the i-th monitoring data subsequence is recorded as a characteristic index value. A value obtained by performing negative correlation mapping on the characteristic index value is recorded as a mapping value between the i-th monitoring data subsequence and the j-th monitoring data subsequence to be matched. The difference between a preset first constant and the mapping value is recorded as a first difference value. The absolute value of the difference between the total number of monitoring data in the j-th monitoring data subsequence to be matched and the total number of monitoring data in the i-th monitoring data subsequence is recorded as a quantity difference value. The product of the first difference value and the quantity difference value is recorded as a target difference value corresponding to the j-th monitoring data subsequence to be matched. The method for obtaining the first eigenvalue corresponding to the i-th monitoring data subsequence includes: obtaining a feature difference value sequence corresponding to the i-th monitoring data subsequence, the h-th feature difference value in the feature difference value sequence being the time difference between the acquisition time corresponding to the h+1-th monitoring data in the i-th monitoring data subsequence and the acquisition time corresponding to the h-th monitoring data in the i-th monitoring data subsequence; and recording the product of the mean of the feature difference value sequence and the variance of the feature difference value sequence as the first eigenvalue corresponding to the i-th monitoring data subsequence.

6. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 3 is characterized in that: The method for obtaining the weight value corresponding to each monitoring data sequence in the set of monitoring data sequences corresponding to each railway facility includes: For the wth monitoring data sequence in the monitoring data sequence set corresponding to the sth railway facility in the railway facility category B0, if the wth monitoring data sequence is included in the rth monitoring data sequence subset corresponding to the railway facility category B0, the target importance value of the rth monitoring data sequence subset is used as the weight value corresponding to the wth monitoring data sequence.

7. The railway knowledge graph data update frequency evaluation method based on life cycle prediction according to claim 6 is characterized in that: The method of obtaining the predicted remaining life of each railway facility according to the weight value corresponding to each monitoring data sequence, and obtaining the future update frequency of each railway facility according to the predicted remaining life of each railway facility, includes: For the railway facility category B0: according to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0 and the weight value corresponding to each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0, a weighted linear regression remaining life prediction model corresponding to the railway facility category B0 is trained to obtain a trained weighted linear regression remaining life prediction model corresponding to the railway facility category B0, and recorded as a target remaining life prediction model corresponding to the railway facility category B0; each monitoring data sequence in the monitoring data sequence set corresponding to each railway facility in the railway facility category B0 is input into the target remaining life prediction model corresponding to the railway facility category B0, and a predicted remaining life corresponding to each railway facility in the railway facility category B0 is output; For the s-th railway facility in the railway facility category B0: obtain the rated life corresponding to the s-th railway facility, and record the ratio of the predicted remaining life corresponding to the s-th railway facility to the rated life corresponding to the s-th railway facility as the determination index corresponding to the s-th railway facility; obtain a determination index value interval, divide the determination index value interval, obtain all determination index value sub-intervals corresponding to the determination index value interval, use the minimum value in the determination index value sub-interval as the ranking value of the corresponding determination index value sub-interval, and sort all determination index value sub-intervals corresponding to the determination index value interval in ascending order of the ranking values ​​to obtain a determination index value sub-interval sequence and an update frequency of each determination index value sub-interval in the determination index value sub-interval sequence, wherein the earlier the determination index value sub-interval in the determination index value sub-interval sequence, the faster the update frequency; in the determination index value sub-interval, obtain the update frequency of the determination index value sub-interval containing the determination index corresponding to the s-th railway facility, and use it as the future update frequency corresponding to the s-th railway facility.

Images