CN112884233B - Multimode fusion late prediction method for high-speed railway system - Google Patents

Multimode fusion late prediction method for high-speed railway system Download PDF

Info

Publication number
CN112884233B
CN112884233B CN202110232791.0A CN202110232791A CN112884233B CN 112884233 B CN112884233 B CN 112884233B CN 202110232791 A CN202110232791 A CN 202110232791A CN 112884233 B CN112884233 B CN 112884233B
Authority
CN
China
Prior art keywords
factor
factors
late
sub
fault
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110232791.0A
Other languages
Chinese (zh)
Other versions
CN112884233A (en
Inventor
刘英舜
王松
陆弘毅
刘嘉明
刘昭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Science and Technology
Original Assignee
Nanjing University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Science and Technology filed Critical Nanjing University of Science and Technology
Priority to CN202110232791.0A priority Critical patent/CN112884233B/en
Publication of CN112884233A publication Critical patent/CN112884233A/en
Application granted granted Critical
Publication of CN112884233B publication Critical patent/CN112884233B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40—Business processes related to the transportation industry

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Educational Administration (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Train Traffic Observation, Control, And Security (AREA)

Abstract

The invention discloses a method for predicting the late point of a high-speed railway system by multi-mode fusion. The method comprises the following steps: analyzing and determining factors influencing the later time of the high-speed railway operation, and dividing the influencing factors into two categories, namely natural factors and driving fault factors; performing attribution degree calculation analysis on each sub-factor by combining a Fuzzy Analytic Hierarchy Process (FAHP) to obtain an attribution degree value of each sub-factor; determining a late penalty factor; and (3) establishing a delay prediction method by data fusion of the sub-factors of different data sources to obtain a high-speed rail delay time prediction value between stations, and combining with an actual train operation schedule to obtain delay time. The method can predict the time of the high-speed train at a later time, provides a basis for the relevant department of high-speed rail to make the deployment plan of the operation time of the high-speed rail in advance, and ensures the safety and convenient travel of residents.

Description

High-speed railway system multi-mode fusion delay prediction method
Technical Field
The invention relates to the technical field of traffic service, in particular to a method for predicting the multi-mode fusion delay of a high-speed railway system.
Background
At present, urban rail transit in China is in a rapid development stage, and rail transit lines are increasingly in ground and overhead forms. The influence of natural conditions around a train running line on rail transit operation is more and more obvious, and meanwhile fault factors of the train play an important role in the train operation. The method has certain practicability for predicting the high-speed railway system late, and in recent years, each scientific research author makes certain contribution to the high-speed railway late field through some machine learning theoretical algorithms such as a Support Vector Machine (SVM), an artificial neural network, a random forest model and the like.
Nevertheless, a large number of scientific researchers do not reasonably classify various factors influencing the normal operation of the high-speed rail, for example, the scholars anecdotal bear publicly published a "support vector machine regression model for predicting the late time of the high-speed rail fault", although the late cause is taken into account, in order to facilitate the model solution, 7 0-1 classification virtual variables are introduced to replace 7 classified late cause classifications, no reasonable classification of external factors such as weather is involved, and the influence degree of different sub-factors on the late time of the high-speed rail is not considered, so that the actual late time of the high-speed rail is difficult to accurately and reasonably predict.
Disclosure of Invention
The invention aims to provide an efficient and accurate high-speed railway system multi-mode fusion late prediction method.
The technical solution for realizing the purpose of the invention is as follows: a high-speed railway system multi-mode fusion late prediction method comprises the following steps:
step 1, analyzing and determining factors influencing the running time late point of the high-speed railway, and dividing the influencing factors into two categories, namely natural factors and driving fault factors;
step 2, combining a Fuzzy Analytic Hierarchy Process (FAHP), calculating and analyzing the attribution degree of each sub-factor, and obtaining the attribution degree value of each sub-factor;
step 3, determining a late penalty factor C γ ；
And 4, establishing a delay prediction method by data fusion of the sub-factors of different data sources, obtaining a high-speed rail delay time prediction value between stations, and combining the high-speed rail delay time prediction value with an actual train operation schedule to obtain delay time.
Further, the factors influencing the high-speed railway operation time late point are analyzed and determined in the step 1, and the influencing factors are divided into two categories, namely natural factors and driving fault factors, and specifically the following factors are adopted:
and (3) natural factors: according to the wind power grades, the wind and the weather are divided into 5 grades of wind, strong wind and strong wind; according to the rainfall, the rain can be divided into 4 grades of light rain, medium rain, heavy rain and heavy rain; 4 grades of violent snow, big snow, middle snow and small snow can be divided according to the snowfall amount in 1 hour; according to the visibility analysis, the types of fog are divided into heavy fog, thick fog, medium fog and light fog;
driving fault factors are as follows: dividing the current failure mode of the vehicle-mounted equipment into two stages, namely a primary failure mode and a secondary failure mode; the primary fault mode represents the range of a module or unit in which a fault occurs, is called as the fault type of the vehicle-mounted equipment and is divided into 8 levels of ATPCU related fault, BTM related fault, STM related fault, speed and distance measuring unit fault, infinite communication fault, train interface unit fault, DMI fault and other faults; the secondary failure mode represents the specific cause of failure for the failed module or unit.
Further, the step 2 is combined with a Fuzzy Analytic Hierarchy Process (FAHP) to calculate and analyze the attribution degree of each sub-factor, so as to obtain the attribution degree value of each sub-factor, which specifically comprises the following steps:
establishing fuzzy judgment matrixes of all levels according to the structural model of the risk factors; quantitatively comparing the importance of one factor to another factor by making a judgment between the two factors to obtain a fuzzy judgment matrix A = (a) ij ) p×p ，a ij An element representing the ith row and jth column in a matrix, i =1, 2.. And p, j =1, 2.. And p, when the matrix satisfies 2 conditions: (1) a is a ii ＝0.5，②a ij +a ji =1, then the matrix is a fuzzy complementary judgment matrix; when quantitative comparison among 2 factors is carried out, a numerical scale of the importance of the risk factors is carried out by adopting a 0.1-0.9 scale method, and the following steps are specifically carried out according to the fuzzy analytic hierarchy process in the step 2:
(1) Constructing a priority relation matrix: the matrix form is used for expressing the relative importance of each factor in each layer to a factor in the previous layer, and the determination method of the matrix factors is as follows: scaling with 0, 0.5 and 1 to determine factor value, and under q factors, establishing fuzzy priority relation matrix for q single factors
Figure GDA0003723149000000021
Figure GDA0003723149000000022
Is a under factor k i To A j K =1,2, \8230, q, the value is:
Figure GDA0003723149000000023
(2) Establishing a first-level fuzzy matrix: reconstructing the priority relation matrix into a fuzzy consistent matrix and checking; the evaluation of each factor in the same class is called primary fuzzy comprehensive evaluation, and the jth factor in the class i is subjected to U ij Judging, the membership degree of the judgment object is r jk J =1,2, \8230;, p, k =1,2, \8230;, q, i.e., representing factor U ij With comment V k The degree of (d); the single-factor evaluation matrix R of the first-level fuzzy comprehensive evaluation k Comprises the following steps:
Figure GDA0003723149000000031
Figure GDA0003723149000000032
Figure GDA0003723149000000033
Figure GDA0003723149000000034
wherein, the first and the second end of the pipe are connected with each other,
Figure GDA0003723149000000035
is the degree of membership of the jth column sub-factor in the ith row under the factor k,
Figure GDA0003723149000000036
The membership of all sub-factors in row i under factor k,
Figure GDA0003723149000000037
representing the degree of membership of all sub-factors in column j under factor k, p being the size of the matrix dimension, r i Is the sum of the membership of the i-th column sub-factor,r j is the sum of membership of the sub-factors in the jth column, b il Representing the membership of the column I sub-factor in the row i, b lj Representing the membership degree of the ith column of the child factors in the jth column;
(3) And (3) hierarchical single ordering: a method of calculating the priority weight of the importance of the factor of the current level to the factor of the previous level according to the judgment matrix is called level list ordering; dominance value of single factor and index weight
Figure GDA0003723149000000038
The calculation method of (2) is as follows: calculation at scheme A i Dominance value of k factor
Figure GDA0003723149000000039
Adopting a square root method:
Figure GDA00037231490000000310
Figure GDA00037231490000000311
in the formula (I), the compound is shown in the specification,
Figure GDA00037231490000000312
a square root value representing the degree of membership of all the sub-factors in the ith row under the k-factor,
Figure GDA00037231490000000313
representing the membership of all the sub-factors in the ith row under the k factor;
(4) And (3) overall hierarchical ordering: the so-called total hierarchical ranking is the ranking weight of the factors contained in the same layer to the importance of the highest layer; calculating the weight of the importance of all factors of the hierarchy relative to the previous hierarchy by using the result of single sequencing of all the hierarchies in the same hierarchy; the total hierarchical sequencing needs to be performed layer by layer from top to bottom, and for a second layer below the highest layer, the single hierarchical sequencing of the second layer is the total sequencing;
according to the system separability principle, when a scheme is optimized, considered factors are divided into a plurality of subsystems according to attributes, the bisection systems adopt a single-level-multi-factor model to calculate the optimal attribute value of each scheme, and the formula is as follows:
Figure GDA0003723149000000041
in the formula, S i A goodness value, W, representing factor i k The weight matrix of the uniform matrix is blurred for all factors of the highest level scheme,
Figure GDA0003723149000000042
shown in scheme A i The dominance value of the factor k, q is the number of the factors;
to S i Sorting results in a superior membership ranking of p solutions under the influence of q factors, i =1,2,3 \8230, p.
Further, step 3 determines a late penalty factor C γ The method comprises the following steps:
classifying different conditions faced by the high-speed train, collecting train late data between various classified stations, and utilizing data processing software SPSS and combining with a model
Figure GDA0003723149000000043
Fitting the data to obtain different C γ A value; wherein T represents the site late prediction time, T γ Represents the fixed running time of the train in the gamma-th running interval,
Figure GDA0003723149000000044
represents the weight coefficient, beta, of the nth type of sub-factor under the mth natural factor δ The weighting coefficient is the weight coefficient of the delta-th fault sub-factor in the driving fault factors.
Further, in the step 4, a late prediction method is established by data fusion of the sub-factors of different data sources, a high-speed rail late time prediction value between stations is obtained, and the late time is obtained by combining with an actual train operation schedule, which specifically comprises the following steps:
step 4.1, combining attribution degree of various influence factors and late penalty factor C under different conditions γ Establishing a method for predicting the late points among the sites;
the fixed driving time value between the stations is not changed and is T γ Predicting the late time as t; considering the difference of the influence degree of the weather factor and the equipment fault factor on the train at the late time, the attribution degree presents a decreasing trend, the factor with large influence degree is large in attribution degree, the more the time at the late time is caused, and the fixed running time T is combined γ Should be that
Figure GDA0003723149000000045
Therefore, the inter-site late prediction method comprises the following steps:
Figure GDA0003723149000000046
wherein t is the predicted time of the site at the late point, C γ As late penalty factor, T γ The fixed running time of the train in the gamma-th running interval is set,
Figure GDA0003723149000000047
is the weight coefficient, beta, of the nth type of sub-factor under the mth natural factor δ The weight coefficient is the delta-th fault sub-factor in the driving fault factors;
4.2, selecting 3/4 of the number of the stations of the high-speed rail line as a prediction object, and acquiring the latest high-speed rail late time in advance for later-stage driving planning, wherein the high-speed train late time prediction method comprises the following steps:
Figure GDA0003723149000000051
in the formula, T is the total train late prediction time.
Compared with the prior art, the invention has the remarkable advantages that: (1) Extracting text data of different data sources, converting non-structural data into structural data, and fusing various data indexes under the background of multi-source data to realize high-precision prediction of the high-speed train late time; (2) High-precision high-speed rail late time prediction is adopted, and reliable information is provided for reasonable train operation deployment made by relevant railway departments and timely trip arrangement made by residents.
Drawings
FIG. 1 is a flow chart of the method for predicting the late point of the multi-mode fusion of the high-speed railway system.
Detailed Description
The invention relates to a multi-mode fusion late prediction method for a high-speed railway system, wherein multi-mode conversion is to convert text unstructured data of various external and internal factors influencing the operation of the high-speed railway into structured data which can be identified and processed by a computer, and when the characteristics of the text data are extracted, the redundancy of the data is reduced and the high dimension of the data is reduced so as to reflect the characteristics of the original text data; and data fusion is carried out between different converted data sources in a specific mode, the actual running condition of the high-speed rail under the influence of various factors is comprehensively reflected, and the data fusion method is used for predicting the possible conditions in the future time. The method comprises the following steps: firstly, analyzing and determining main factors influencing the high-speed railway operation late point, dividing the influence factors into two categories of natural factors and driving fault factors, then combining a Fuzzy Analytic Hierarchy Process (FAHP) to calculate and analyze the attribution degree of each sub-influence factor, obtaining the attribution degree value of each sub-factor, and adding a late point punishment factor C γ And fusing the sub-factors of different data sources to obtain the high-speed rail late time predicted value between stations, and combining the predicted value with an actual train operation schedule to obtain the late time.
The invention discloses a high-speed railway system multi-mode fusion late prediction method, which comprises the following steps of:
step 1, analyzing and determining factors influencing the running time late point of the high-speed railway, and dividing the influencing factors into two categories, namely natural factors and driving fault factors;
step 2, combining a Fuzzy Analytic Hierarchy Process (FAHP), calculating and analyzing the attribution degree of each sub-factor, and obtaining the attribution degree value of each sub-factor;
step 3, determining a late penalty factor C γ ；
And 4, establishing a delay prediction method by data fusion of the sub-factors of different data sources to obtain a high-speed rail delay time prediction value between stations, and combining the high-speed rail delay time prediction value with an actual train operation schedule to obtain delay time.
Further, the factors influencing the high-speed railway operation time late point are analyzed and determined in the step 1, and the influencing factors are divided into two categories, namely natural factors and driving fault factors, and specifically the following factors are adopted:
1. natural factors of the world
Inclement weather includes: wind, fog, rain, snow, sand storms, hailstones, etc. Severe weather has no direct influence on the operation of urban rail transit underground lines, and mainly causes the increase of passenger flow and the increase of operation demand. For an open-air line, severe weather directly acts on ground equipment on the line and a train running on the line, and great influence is generated on the running safety, so that the influence of the severe weather on the running of the open-air line is mainly analyzed. In severe weather, wind, rain, snow, fog and extreme weather are common severe weather, the annual occurrence frequency is higher, other severe weather such as sand storms, hailstones and the like do not occur frequently, and therefore, the influence of 5 common severe weather such as wind, rain, snow, fog and extreme temperature weather on open-air lines of urban rail transit is only considered.
(1) Wind factors. For an urban rail transit system, wind weather can be divided into the following ways according to the wind power level: grade 10 and above: the system has the advantages that the system has great influence on the stability of the train, has high overturning risk of the train, can cause serious damage to urban rail infrastructure, train vehicles and the like, and seriously threatens the life safety of passengers and workers; and 9, stage: the stability of the train is greatly influenced, and urban rail infrastructure can be damaged to different degrees; 6-8 stages: the stability of the train is influenced to a certain extent, and the urban rail infrastructure is generally not damaged; grade 6 is as follows: the influence on the stability of the train is small, and the urban rail transit infrastructure cannot be damaged. Since the type of grade 9 or more has a serious influence on the train operation, the high-speed rail operation department door may directly take the shutdown measures, and thus the shutdown measures are not taken into consideration. Meanwhile, the influence of breeze below three levels on the running of the vehicle is not large, and the method is not considered in the range.
(2) Rain factor. According to the rainfall, the rain can be divided into 4 grades of light rain, medium rain, heavy rain and heavy rain.
(3) Snow factors. The types of snow can be divided into 4 grades of heavy snow, big snow, medium snow and small snow according to the precipitation amount in 1 hour.
(4) Fog factor. The requirements of emergency plan (temporary) regulations on severe weather such as railway storm, rain, snow, fog and the like require that when the visibility is less than 50 meters, 100 meters and 200 meters, I-level emergency state, II-level emergency state and III-level emergency state are respectively entered. Practice proves that fog (including heavy fog and dense fog) with visibility of more than 200 m has little influence on the high-speed railway. The strong dense fog and the extra strong dense fog with visibility below 200 m have influence on the train operation and need to be operated at a limited speed. The types of fog are classified into 4 levels of heavy fog, thick fog, medium fog and light fog according to visibility.
(5) Extreme temperature factors. Extreme temperatures can cause problems for typical electrical and electronic components such as main control switches, transformers, frequency converters, batteries and on-board electrical devices. According to statistics, the extreme lowest temperature of each city in China is not lower than minus 45 ℃, the extreme highest temperature is not higher than plus 45 ℃, and the high temperature resistance and the low temperature resistance of the trains and ground equipment of urban rail transit are better in the temperature range, so that the influence of the extreme temperature on the running of the trains on the open-air line is not considered.
2. Fault factor of running equipment
In the operation of a high-speed train, the failure of the on-board equipment is various. Referring to data such as 'train control vehicle-mounted equipment typical fault case' and 'signal equipment fault one-point connection', and combining long-term summary of railway field personnel, the current vehicle-mounted equipment fault mode is divided into two stages, namely, a first-stage fault mode (FFP) and a second-stage fault mode (SFP). The primary failure mode represents the general extent of the module or unit in which the failure occurred and may be referred to as the failure type of the in-vehicle device. Such as ATPCU-related failures, BTM-related failures, wireless communication failures, speed measurement and ranging unit failures, and the like. The secondary failure mode represents a specific failure cause of a failed module or unit and is described in greater detail than the primary failure mode. For example, in BTM related failures (primary failure modes), the primary failure modes mainly include: the BTM port is invalid, BSA related faults, an all-zero transponder and BTM test timeout; the faults of the speed measuring and distance measuring unit (primary fault mode) comprise two secondary fault modes of speed sensor fault and Doppler radar fault. In the maintenance log, there is a large difference between the number of samples of different failure modes, i.e. a large-scale failure mode and a small-scale failure mode.
Further, the step 2 is combined with a Fuzzy Analytic Hierarchy Process (FAHP) to calculate and analyze the attribution degree of each sub-factor, so as to obtain the attribution degree value of each sub-factor, which specifically comprises the following steps:
the fuzzy analytic hierarchy process has the characteristic of consistency with human thinking, so the fuzzy analytic hierarchy process is more applied to the schemes with more evaluation indexes and fuzziness evaluation. The following is a detailed description of the calculation steps of the fuzzy analytic hierarchy process: (1) The fuzzy complementary judgment matrix is constructed aiming at different evaluation objects, the importance of each index is different, and different weights are given. According to the structural model of the risk factors, it is necessary to establish fuzzy judgment matrixes of all levels. By making a judgment between two factors, quantitatively comparing the importance of one factor to the other, a fuzzy judgment matrix A = (a) ij ) p×p ，a ij An element representing the ith row and jth column in a matrix, i =1, 2.. And p, j =1, 2.. And p, when the matrix satisfies 2 conditions: (1) a is a ii ＝0.5，②a ij +a ji And =1, the matrix is a fuzzy complementary judging matrix. When quantitative comparisons between 2 factors were made, most of the data were assigned factor importance using the 9-scale method by looking up the literature. Some documents improve the fuzzy analytic hierarchy process, but propose to replace the 9-scale process with the 3-scale process, directly establish the fuzzy complementary judgment matrix, this kind of method can solve the difficulty of conformance test, but only construct the judgment matrix through the evaluation of a few experts, the subjective factor is too big, may not accord with the actual situation. Therefore, the design adopts the method of 0.1-0.9 scaleCarrying out numerical scaling on the importance of the risk factors, and specifically developing the following according to the fuzzy analytic hierarchy process in the step 2:
(1) Constructing a priority relation matrix: the relative importance of each factor in each layer to a factor in the previous layer can be expressed in a matrix form, and the determination method of the matrix factors is as follows: scaling with 0, 0.5 and 1 can be used to determine the factor value, which is simple and easy to implement. Under q factors, p evaluation schemes constitute an optimal problem, firstly, a fuzzy priority relation matrix is established for q single factors
Figure GDA0003723149000000081
Figure GDA0003723149000000082
Is a under factor k i To A j K =1,2, \ 8230;, q, with the value:
Figure GDA0003723149000000083
(2) Establishing a first-level fuzzy matrix: and modifying the priority relation matrix into a fuzzy consistent matrix, and checking to ensure the consistency of the fuzzy consistent matrix. The evaluation of each factor in the same class is called primary fuzzy comprehensive evaluation, and the jth factor in the class i is subjected to U ij Judging, the membership degree of the judgment object is r jk J =1,2, \8230;, p, k =1,2, \8230;, q, i.e., the expression factor U ij With comment V k The degree of (d); the single-factor evaluation matrix R of the first-level fuzzy comprehensive evaluation k Comprises the following steps:
Figure GDA0003723149000000084
Figure GDA0003723149000000085
Figure GDA0003723149000000086
Figure GDA0003723149000000087
wherein the content of the first and second substances,
Figure GDA0003723149000000088
is the degree of membership of the jth column sub-factor in the ith row under the factor k,
Figure GDA0003723149000000089
The membership of all sub-factors in row i under factor k,
Figure GDA00037231490000000810
representing the degree of membership of all sub-factors in column j under factor k, p being the size of the matrix dimension, r i Is the sum of the membership of the i-th column sub-factor, r j Is the sum of membership of the j-th column of sub-factors, b il Representing the membership of the ith column of the sub-factor, b lj Representing the membership grade of the first column of the second column;
(3) And (3) hierarchical single ordering: the method of calculating the priority of the importance of the factor at the current level to the factor at the previous level based on the judgment matrix is called hierarchical single ranking. Merit figure of single factor and index weight
Figure GDA00037231490000000811
The calculation method of (2) is as follows: calculation at scheme A i Dominance value of k factor
Figure GDA00037231490000000812
Adopting a square root method:
Figure GDA0003723149000000091
Figure GDA0003723149000000092
in the formula (I), the compound is shown in the specification,
Figure GDA0003723149000000093
a square root value representing the degree of membership of all the sub-factors in the ith row under the k-factor,
Figure GDA0003723149000000094
representing the membership of all the sub-factors in the ith row under the k factor;
(4) And (3) overall hierarchical ordering: the so-called total hierarchical ranking is to calculate the ranking weight of the factors contained in the same layer to the importance of the highest layer. By using the result of the single ordering of all levels in the same level, the importance weight of all factors of the level can be calculated for the previous level. The total hierarchical sorting needs to be performed layer by layer from top to bottom, and for the second layer below the highest layer, the single hierarchical sorting is the total sorting.
Referring to the principle of system separability, when a scheme is preferred, factors to be considered are divided into a plurality of subsystems according to attributes of the schemes, and the binary system calculates the goodness value of each scheme by adopting a single-level multi-factor model, which can be represented by a formula:
Figure GDA0003723149000000095
in the formula, S i A dominance value, W, representing factor i k The weight matrix of the uniform matrix is blurred for all factors of the highest level scheme,
Figure GDA0003723149000000096
shown in scheme A i The dominance value of the factor k, q is the number of the factors;
to S i Sorting results in a superior membership ranking of p solutions under the influence of q factors, i =1,2,3 \8230, p.
Further, the step 3 of determining the late penalty factor C γ The method comprises the following steps:
late timePenalty factor C γ Presenting diversity during train travel. Because high-speed trains are exposed to different conditions during running, various weather conditions and vehicle equipment problems can occur in the process, and certain influence factors occur together. For different conditions, a penalty factor C for the late γ The values of the data are diversified, the different conditions faced by the high-speed train are classified, the late train data among various classified stations are collected, and data processing software SPSS is utilized and combined with a model
Figure GDA0003723149000000097
Fitting the data to obtain different C γ The value is obtained. Wherein T represents the site late prediction time, T γ Represents the fixed running time of the train in the gamma-th running interval,
Figure GDA0003723149000000098
represents the weight coefficient, beta, of the nth type of sub-factor under the mth natural factor δ The weighting coefficient is the weight coefficient of the delta-th fault sub-factor in the driving fault factors.
Further, the establishment of the multi-mode late prediction method for the high-speed railway system is described in step 4.
Step 4.1, combining the attribution degree of various influence factors given in the steps 1,2 and 3 with the late penalty factor C under different conditions γ A method for predicting the late points among the sites can be established.
The fixed running time value between the stations is unchanged and is T γ Predicting the late time as t; considering the difference of the influence degree of the weather factor and the equipment fault factor on the train at the late time, the attribution degree presents a decreasing trend, the factor with large influence degree is large in attribution degree, the more the time at the late time is caused, and the fixed running time T is combined γ Should be as follows
Figure GDA0003723149000000101
Therefore, the inter-site delay prediction method comprises the following steps:
Figure GDA0003723149000000102
where t is the predicted time of the site at a later time, C γ A penalty factor for late, T γ The fixed running time of the train in the gamma-th running interval,
Figure GDA0003723149000000103
is the weight coefficient, beta, of the nth type of sub-factor under the mth natural factor δ And the weight coefficient is the delta-th fault sub-factor in the driving fault factors.
And 4.2, in order to predict the high-speed rail late time, making reasonable train operation deployment and safe and convenient travel of residents for the high-speed rail relevant department, selecting 3/4 of the number of the high-speed rail line stations as prediction objects so as to obtain the latest high-speed rail late time in advance and make full preparation for later-stage driving planning. Therefore, the method for predicting the late of the high-speed train comprises the following steps:
Figure GDA0003723149000000104
in the formula, T is the total train late prediction time.
The method is used for predicting the time of the high-speed train at the late point, provides a basis for the relevant department of high-speed rail to make the deployment plan of the operation time of the high-speed rail in advance, and ensures the safe and convenient travel of residents.
The invention is described in further detail below with reference to the figures and the embodiments.
Examples
With reference to fig. 1, the method for predicting the late point of the multi-mode fusion of the high-speed railway system comprises the following steps:
step 1: step 1.1, natural factors. The main inclement weather includes: wind, fog, rain, snow, sand storms, hailstones, etc. Severe weather has no direct influence on the operation of urban rail transit underground lines, and mainly causes the increase of passenger flow and the increase of operation demand. For an open-air line, severe weather directly acts on ground equipment on the line and a train running on the line, and great influence is generated on the running safety, so that the influence of the severe weather on the running of the open-air line is mainly analyzed. In severe weather, wind, rain, snow, fog and extreme weather are common severe weather, the annual occurrence frequency is higher, other severe weather such as sand storms, hailstones and the like do not occur frequently, and therefore, the influence of 5 common severe weather such as wind, rain, snow, fog and extreme temperature weather on open-air lines of urban rail transit is only considered.
(1) Wind factor
For an urban rail transit system, wind weather can be divided into the following ways according to the wind power level:
grade 10 and above: the stability of the train is greatly influenced, the overturning risk of the train is high, urban rail infrastructure, train vehicles and the like can be seriously damaged, and the life safety of passengers and workers is seriously threatened;
and 9, stage: the stability of the train is greatly influenced, and urban rail infrastructure can be damaged to different degrees;
6-8 level: the stability of the train is influenced to a certain extent, and the infrastructure of the urban rail transit is generally not damaged;
grade 6 is as follows: the influence on the stability of the train is small, and the urban rail transit infrastructure cannot be damaged.
Since the type of the grade 9 or more has a serious influence on the operation of a train running at a high speed, the operation stop measure may be directly taken by the high-speed railway operation department door, so that the type of the wind speed of the grade 9 or more is not considered. Meanwhile, the influence of breeze below three levels on the running of the vehicle is not large, the influence on the vehicle body is not large, and the influence is not considered within the range. The wind type division criteria are shown in table 1.1:
TABLE 1.1 wind type division Standard
Figure GDA0003723149000000111
(2) Rain factor
According to the size of the rainfall, the rainfall can be divided into 4 grades of light rain, medium rain, heavy rain and heavy rain, as shown in a table 1.2:
TABLE 1.2 rain type division criteria
Figure GDA0003723149000000112
(3) Snow factor
The snow types can be divided according to the precipitation amount in 1 hour as shown in table 1.3.
TABLE 1.3 snow type division criteria
Figure GDA0003723149000000121
(4) Fog factor
The requirements of emergency plan (temporary) regulations on severe weather such as railway storm, rain, snow, fog and the like require that when the visibility is less than 50 meters, 100 meters and 200 meters, I-level emergency state, II-level emergency state and III-level emergency state are respectively entered. Practice proves that fog (including heavy fog and dense fog) with visibility of more than 200 m has little influence on the high-speed railway. The strong dense fog and the extra strong dense fog with the visibility of less than 200 meters have influence on the running of the train and need to be operated at a limited speed. The types of fog were divided according to visibility, as shown in table 1.4:
TABLE 1.4 fog type division Standard
Figure GDA0003723149000000122
(5) Extreme temperature factor
Extreme temperatures can cause problems for typical electrical and electronic components such as main control switches, transformers, frequency converters, batteries and on-board electrical devices. According to statistics, the extreme lowest temperature of each city in China is not lower than minus 45 ℃, the extreme highest temperature is not higher than plus 45 ℃, and the high temperature resistance and the low temperature resistance of the trains and ground equipment of urban rail transit are better in the temperature range, so that the influence of the extreme temperature on the running of the trains on the open-air line is not considered.
And step 1.2, fault factors of the traveling equipment. In the operation of a high-speed train, the failure of the on-board equipment is various. Referring to data such as 'train control vehicle-mounted equipment typical fault case' and 'signal equipment fault one-point communication', and combining long-term summary of railway field personnel, the current vehicle-mounted equipment fault mode is divided into two stages, namely a first-stage fault mode (FFP) and a second-stage fault mode (SFP).
The primary failure mode represents the general range of modules or units in which the failure occurs and may be referred to as the failure type of the in-vehicle device. Such as ATPCU-related failures, BTM-related failures, wireless communication failures, speed measurement and ranging unit failures, and the like. The secondary failure mode represents a specific cause of failure for the failed module or unit, and is described in greater detail than the primary failure mode. For example, in BTM related failures (primary failure modes), the primary failure modes mainly include: invalid BTM port, BSA related fault, all-zero transponder, BTM test timeout; the faults of the speed and distance measuring unit (primary fault mode) comprise two secondary fault modes of speed sensor fault and Doppler radar fault. In the maintenance log, the sample numbers of different failure modes have larger difference, namely a large-proportion failure mode and a small-proportion failure mode, and tables 1.5 and 1.6 respectively and specifically give each failure type in the primary failure mode and the secondary failure mode and the proportion of the failure type in the primary failure mode and the secondary failure mode in the whole sampling data.
TABLE 1.5 Primary Fault ratios and sample ratios
Figure GDA0003723149000000131
TABLE 1.6 Secondary Fault ratios and sample ratios
Figure GDA0003723149000000132
Figure GDA0003723149000000141
Step 2: the fuzzy analytic hierarchy process has the characteristic of consistency with human thinking, therefore, the fuzzy analytic hierarchy process is applied to the human thinkingThe evaluation indexes are more, and the evaluation scheme with ambiguity is more in application. The following is a detailed description of the computation steps of the fuzzy analytic hierarchy process: (1) The fuzzy complementary judgment matrix is constructed aiming at different evaluation objects, the importance of each index is different, and different weights are given. According to the structural model of the risk factors, it is necessary to establish fuzzy judgment matrixes of all levels. By making a decision between two factors, quantitatively comparing the importance of one factor to the other, a fuzzy decision matrix A = (a) is obtained ij ) p×p . When the matrix satisfies 2 conditions: (1) a is ii ＝0.5，②a ij +a ji And =1, the matrix is a fuzzy complementary judging matrix. When quantitative comparisons between 2 factors were made, most of the data were assigned factor importance using the 9-scale method by consulting the literature. Some documents improve the fuzzy analytic hierarchy process, but propose to replace the 9 scale process with the 3 scale process, directly establish the fuzzy complementary judgment matrix, this kind of method can solve the difficulty of conformance test, but only construct the judgment matrix through the evaluation of a few experts, the subjective factor is too big, may not accord with the actual conditions. Therefore, the design adopts a 0.1-0.9 scaling method to carry out numerical scaling on the importance of the risk factors.
TABLE 2 numerical Scale of importance of factors
Figure GDA0003723149000000142
(1) Constructing a priority relation matrix: the relative importance of each factor in each layer to a factor in the previous layer can be expressed in a matrix form, and the matrix factors are determined by the following method: scaling with 0, 0.5 and 1 can be used to determine the factor value, which is simple and easy to implement. Under q factors, p evaluation schemes constitute an optimization problem, firstly, a fuzzy priority relation matrix is established for q single factors
Figure GDA0003723149000000151
(is A under factor k i To A j Priority coefficient) having a value of:
Figure GDA0003723149000000152
(2) Establishing a first-level fuzzy matrix: and (5) transforming the priority relation matrix into a fuzzy consistent matrix, and checking to ensure the consistency of the fuzzy consistent matrix. The evaluation of each factor in the same class is called primary fuzzy comprehensive evaluation, and the jth factor in the class i is subjected to U ij Judging, the membership degree of the judgment object is r jk (j =1,2, \8230;, p; k =1,2, \8230;, q), i.e., representing the factor U ij With comment V k (k =1,2, \8230;, q). Then the single-factor evaluation matrix of the first-level fuzzy comprehensive evaluation:
Figure GDA0003723149000000153
Figure GDA0003723149000000154
Figure GDA0003723149000000155
Figure GDA0003723149000000156
(3) And (3) hierarchical single ordering: the method of calculating the priority of the importance of the factor at the current level to the factor at the previous level based on the judgment matrix is called hierarchical single ranking. Dominance value of single factor and index weight
Figure GDA0003723149000000157
The calculation method of (2) is as follows: calculation at scheme A i Dominance value of k factor
Figure GDA0003723149000000158
Adopting a square root method:
Figure GDA0003723149000000159
Figure GDA00037231490000001510
(4) Overall hierarchical ordering: the so-called total hierarchical ranking is a ranking weight for calculating the importance of the factors contained in the same layer to the highest layer. By using the result of single ordering of all levels in the same level, the weight of the importance of all factors of the level can be calculated for the previous level. The total hierarchical sequencing needs to be performed layer by layer from top to bottom, and for the second layer below the highest layer, the single hierarchical sequencing is the total sequencing. According to the system separability principle, when the scheme is optimized, the considered factors are divided into a plurality of subsystems according to the attributes of the factors, and the subsystem adopts a single-level multi-factor model to calculate the optimal attribute value of each scheme. Can be represented by the formula:
Figure GDA0003723149000000161
to S i (i =1,2,3 \8230;, p) the ranking results in a ranking of the dominance of the p solutions under the influence of q factors.
And (3) solving the weight matrix w of the fuzzy consistent matrix of each risk factor by using Matlab software, wherein the calculation result is shown in the following table:
TABLE 3 weight coefficient table for each factor
Figure GDA0003723149000000162
Figure GDA0003723149000000171
Table 3 (continuation)
Figure GDA0003723149000000172
And step 3: late penalty factor C γ Presenting diversity during train travel. Since the high-speed train is exposed to different conditions during running, various weather conditions and vehicle equipment problems can occur during the running process, and certain influence factors occur together. For different conditions, a penalty factor C for the late γ The values of the data are diversified, different conditions faced by the high-speed train need to be classified, the train delay data among stations under various classifications is collected, and data processing software SPSS is utilized and combined with a model
Figure GDA0003723149000000173
Fitting the data to obtain different C γ The value is obtained. Theoretically, the more complex the train faces the influence factor, C γ The larger the value.
And 4, step 4: step 4.1, combining the attribution degree of various influence factors given in the steps 1,2 and 3 with the late penalty factor C under different conditions γ A method for predicting the late point between sites can be established. The fixed running time value between the stations is unchanged and is T γ Predicting the late time as t; considering the difference of the influence degree of the weather factor and the equipment fault factor on the train at the late time, the attribution degree presents a decreasing trend, the factor with large influence degree is large in attribution degree, the more the time at the late time is caused, and the fixed running time T is combined γ Should be that
Figure GDA0003723149000000174
Therefore, the inter-site late prediction method comprises the following steps:
Figure GDA0003723149000000175
γ＝1,2,3…；m＝1,2,3,4；
n＝1,2,3,4,5；δ＝1,2…,8
and 4.2, in order to predict the high-speed rail late time and customize reasonable train operation deployment and safe and convenient travel of residents for the high-speed rail related department, 3/4 of the number of the high-speed rail line stations is selected as a prediction object. Therefore, the method for predicting the late of the high-speed train comprises the following steps:
Figure GDA0003723149000000181
γ＝1,2,3…；m＝1,2,3,4；
n＝1,2,3,4,5；δ＝1,2…,8
in conclusion, the invention aims to provide a multi-source data fusion method, and establishes a multi-mode fusion delay prediction method for a high-speed railway system. In the running process of the train, the type of the running fault and the natural condition of the train are fused in a form of converting text information into data information, the time used by the running section is predicted, the late time is obtained by combining a running schedule, the late time of partial sections is added, the condition of the high-speed rail of the line at about late can be predicted, and powerful support is provided for the high-speed rail relevant department to customize reasonable train running deployment and safe and convenient travel of residents.

Claims (4)

1. A method for predicting the late point of a high-speed railway system through multi-mode fusion is characterized by comprising the following steps of:
step 1, analyzing and determining factors influencing the running time late point of the high-speed railway, and dividing the influencing factors into two categories, namely natural factors and driving fault factors;
step 2, combining a Fuzzy Analytic Hierarchy Process (FAHP), calculating and analyzing the attribution degree of each sub-factor, and obtaining the attribution degree value of each sub-factor;
step 3, determining a late penalty factor C γ ；
Step 4, establishing a delay prediction method by data fusion of the sub-factors of different data sources to obtain a high-speed rail delay time prediction value between stations, and combining the high-speed rail delay time prediction value with an actual train operation schedule to obtain delay time;
and 2, combining a Fuzzy Analytic Hierarchy Process (FAHP) to calculate and analyze the attribution degree of each sub-factor to obtain the attribution degree value of each sub-factor, wherein the method specifically comprises the following steps:
establishing fuzzy judgment matrixes of all levels according to the structural model of the risk factors; quantitatively comparing the importance of one factor to another factor by making a judgment between the two factors to obtain a fuzzy judgment matrix A = (a) ij ) p×p ，a ij Represents the element in the ith row and jth column of the matrix, i =1, 2.. P, j =1, 2.. P, p represents p scoring schemes, when the matrix satisfies 2 conditions: (1) a is a ii ＝0.5，②a ij +a ji If =1, the matrix is a fuzzy complementary judgment matrix; when quantitative comparison among 2 factors is carried out, a numerical scale of the importance of the risk factors is carried out by adopting a 0.1-0.9 scale method, and the following steps are specifically carried out according to the fuzzy analytic hierarchy process in the step 2:
(1) Constructing a priority relation matrix: the matrix form is used for expressing the relative importance of each factor in each layer to a factor in the previous layer, and the determination method of the matrix factors is as follows: scaling with 0, 0.5 and 1 to determine factor value, and under q factors, establishing fuzzy priority relation matrix for q single factors
Figure FDA0003723148990000011
Figure FDA0003723148990000012
Is a under factor k i To A j K =1,2, \ 8230, q, taking the value:
Figure FDA0003723148990000013
(2) Establishing a first-level fuzzy matrix: reconstructing the priority relation matrix into a fuzzy consistent matrix and checking; the evaluation of each factor in the same class is called primary fuzzy comprehensive evaluation, and the jth factor U in the ith class is set ij Performing judgment, wherein the membership degree of the judgment object is r jk J =1,2, \8230;, p, k =1,2, \8230;, q, i.e., representing factor U ij With comment V k Is not limited toDegree; the single-factor evaluation matrix R of the first-level fuzzy comprehensive evaluation k Comprises the following steps:
Figure FDA0003723148990000021
Figure FDA0003723148990000022
Figure FDA0003723148990000023
Figure FDA0003723148990000024
wherein the content of the first and second substances,
Figure FDA0003723148990000025
is the degree of membership of the jth column sub-factor in the ith row under the factor k,
Figure FDA0003723148990000026
The membership of all sub-factors in row i under factor k,
Figure FDA0003723148990000027
representing the degree of membership of all sub-factors in column j under factor k, p being the size of the matrix dimension, r i Is the sum of the membership of the i-th column sub-factor, r j Is the sum of membership of the sub-factors in the jth column, b il Representing the membership of the ith column of the sub-factor, b lj Representing the membership degree of the ith column of the child factors in the jth column;
(3) Hierarchical list sorting: a method of calculating the priority weight of the importance of the factor of the current level to the factor of the previous level according to the judgment matrix is called level list ordering; dominance value of single factor and index weight
Figure FDA0003723148990000028
The calculation method of (2) is as follows: calculation at scheme A i Dominance value of k factor
Figure FDA0003723148990000029
Adopting a square root method:
Figure FDA00037231489900000210
Figure FDA00037231489900000211
in the formula (I), the compound is shown in the specification,
Figure FDA00037231489900000212
a square root value representing the degree of membership of all the sub-factors in the ith row under the k-factor,
Figure FDA00037231489900000213
representing the membership of all the sub-factors in the ith row under the k factor;
(4) And (3) overall hierarchical ordering: the so-called total hierarchical ranking is the ranking weight of the factors contained in the same layer to the importance of the highest layer; calculating the weight of the importance of all factors of the hierarchy relative to the previous hierarchy by using the result of single sequencing of all the hierarchies in the same hierarchy; the total hierarchical sequencing needs to be performed layer by layer from top to bottom, and for a second layer below the highest layer, the single hierarchical sequencing of the second layer is the total sequencing;
according to the system separability principle, when a scheme is optimized, considered factors are divided into a plurality of subsystems according to attributes, the bisection systems adopt a single-level-multi-factor model to calculate the optimal attribute value of each scheme, and the formula is as follows:
Figure FDA0003723148990000031
in the formula, S i A goodness value, W, representing factor i k The weight matrix of the uniform matrix is blurred for all factors of the highest level scheme,
Figure FDA0003723148990000032
shown in scheme A i The dominance value of the factor k, q is the number of the factors;
to S i The ranking is carried out to obtain the superior attribute ranking of p schemes under the influence of q factors, i =1,2,3, \8230, p.
2. The method for predicting the high-speed railway system multi-mode fusion late point as claimed in claim 1, wherein the factors affecting the high-speed railway operation time late point are analyzed and clarified in step 1, and the affecting factors are divided into two categories, namely natural factors and driving fault factors, and specifically as follows:
and (3) natural factors: according to the wind power grades, the wind and the weather are divided into 5 grades of wind, strong wind and strong wind; according to the rainfall, the rain can be divided into 4 grades of light rain, medium rain, heavy rain and heavy rain; 4 grades of violent snow, big snow, middle snow and small snow can be divided according to the snowfall amount in 1 hour; according to the visibility analysis, the types of fog are divided into heavy fog, thick fog, medium fog and light fog;
driving fault factors are as follows: dividing the current failure mode of the vehicle-mounted equipment into two stages, namely a primary failure mode and a secondary failure mode; the primary fault mode represents the range of a module or unit in which a fault occurs, is called as the fault type of the vehicle-mounted equipment and is divided into 8 levels of ATPCU related fault, BTM related fault, STM related fault, speed and distance measuring unit fault, infinite communication fault, train interface unit fault, DMI fault and other faults; the secondary failure mode represents the specific cause of failure for the failed module or unit.
3. The method for predicting the late time of multi-modal fusion of high-speed railway system according to claim 1Characterised in that the determination of the late penalty factor C in step 3 γ The method comprises the following steps:
classifying different conditions faced by the high-speed train, collecting train late data among stations under various classifications, utilizing data processing software SPSS and combining models
Figure FDA0003723148990000033
Fitting the data to obtain different C γ A value; wherein T represents the site late prediction time, T γ Represents the fixed running time of the train in the gamma-th running interval,
Figure FDA0003723148990000034
weight coefficient, beta, representing the nth class of sub-factors under the mth natural factor δ The weighting coefficient is the weight coefficient of the delta-th fault sub-factor in the driving fault factors.
4. The high-speed railway system multi-mode fusion delay prediction method according to claim 1, wherein in step 4, a delay prediction method is established by data fusion of the sub-factors of different data sources, a high-speed railway delay time prediction value between stations is obtained, and delay time is obtained by combining with an actual train operation schedule, and the delay time prediction method specifically comprises the following steps:
step 4.1, combining the attribution degree of various influence factors and late penalty factor C under different conditions γ Establishing a method for predicting the late points among the sites;
the fixed driving time value between the stations is not changed and is T γ Predicting the late time as t; considering the difference of the influence degree of the weather factor and the equipment fault factor on the train at the late time, the attribution degree presents a decreasing trend, the factor with large influence degree is large in attribution degree, the more the time at the late time is caused, and the fixed running time T is combined γ Should be that
Figure FDA0003723148990000041
Therefore, the inter-site late prediction method comprises the following steps:
Figure FDA0003723148990000042
where t is the predicted time of the site at a later time, C γ A penalty factor for late, T γ The fixed running time of the train in the gamma-th running interval,
Figure FDA0003723148990000043
is the weight coefficient, beta, of the nth type of sub-factor under the mth natural factor δ The weight coefficient is the delta-th fault sub-factor in the driving fault factors;
4.2, selecting 3/4 of the number of the stations of the high-speed rail line as a prediction object, and acquiring the latest high-speed rail late time in advance for later-stage driving planning, so that the high-speed train late prediction method comprises the following steps:
Figure FDA0003723148990000044
in the formula, T is the total delay forecast time of the train.
CN202110232791.0A 2021-03-03 2021-03-03 Multimode fusion late prediction method for high-speed railway system Active CN112884233B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110232791.0A CN112884233B (en) 2021-03-03 2021-03-03 Multimode fusion late prediction method for high-speed railway system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110232791.0A CN112884233B (en) 2021-03-03 2021-03-03 Multimode fusion late prediction method for high-speed railway system

Publications (2)

Publication Number Publication Date
CN112884233A CN112884233A (en) 2021-06-01
CN112884233B true CN112884233B (en) 2022-10-21

Family

ID=76055275

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110232791.0A Active CN112884233B (en) 2021-03-03 2021-03-03 Multimode fusion late prediction method for high-speed railway system

Country Status (1)

Country Link
CN (1) CN112884233B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114971073A (en) * 2022-06-24 2022-08-30 北京交通大学 Method and device for predicting arrival late of multiple stations of over-the-line train

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109359788A (en) * 2018-12-06 2019-02-19 西南交通大学 A kind of initial late method for building up for influencing prediction model of bullet train
CN109377024A (en) * 2018-09-30 2019-02-22 北京航空航天大学 A kind of recovery capability appraisal procedure comprehensive based on step analysis and grey fuzzy
CN109508751A (en) * 2018-12-06 2019-03-22 西南交通大学 The deep neural network model modeling method of the late time prediction of High Speed Railway Trains
CN112232553A (en) * 2020-09-27 2021-01-15 同济大学 Bayesian network-based high-speed rail train late influence factor diagnosis method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109377024A (en) * 2018-09-30 2019-02-22 北京航空航天大学 A kind of recovery capability appraisal procedure comprehensive based on step analysis and grey fuzzy
CN109359788A (en) * 2018-12-06 2019-02-19 西南交通大学 A kind of initial late method for building up for influencing prediction model of bullet train
CN109508751A (en) * 2018-12-06 2019-03-22 西南交通大学 The deep neural network model modeling method of the late time prediction of High Speed Railway Trains
CN112232553A (en) * 2020-09-27 2021-01-15 同济大学 Bayesian network-based high-speed rail train late influence factor diagnosis method

Also Published As

Publication number Publication date
CN112884233A (en) 2021-06-01

Similar Documents

Publication Publication Date Title
CN110766298B (en) Multi-standard multi-service rail transit simulation method and system
CN106845371B (en) A kind of city road network automotive emission remote sensing monitoring system
CN112622990A (en) Intelligent operation and maintenance management system for urban rail metro vehicle
CN107180534B (en) The express highway section average speed estimation method of support vector regression fusion
CN112884233B (en) Multimode fusion late prediction method for high-speed railway system
CN104991549A (en) Track circuit red-light strip default diagnosis method based on FTA and multilevel fuzzy-neural sub-networks
CN111507879A (en) Urban space structure identification method
Li Predicting short-term traffic flow in urban based on multivariate linear regression model
CN113344408A (en) Processing method for multi-scale situation perception process of civil aviation traffic control operation
CN114363316A (en) Intelligent networking monitoring and supervision system for cross-regional road infrastructure
CN115938124A (en) Urban rail transit emergency influence prediction method, system and product
CN106327867A (en) Bus punctuality prediction method based on GPS data
CN115099646A (en) Power transmission line maintenance method based on fuzzy probability mixed reliability evaluation and risk analysis
CN116597642A (en) Traffic jam condition prediction method and system
CN113409578B (en) Traffic network health portrait method and system based on fuzzy clustering
CN117995027A (en) Flight plan drop-preparation recommendation method, system, equipment and medium
CN112232553B (en) Bayesian network-based high-speed rail train late influence factor diagnosis method
CN116862234A (en) Risk assessment method for track traffic self-consistent energy system
CN116434361A (en) Highway congestion identification method based on ETC transaction data
CN115713184A (en) Bus route operation service evaluation method
Yuan et al. A gray prediction method for economic loss of road traffic accidents based on Logistic model.
CN114662304A (en) Dynamic event trigger filtering estimation method for urban road traffic flow
CN112183597A (en) Unbalanced traffic accident severity analysis method based on spectral clustering
Li et al. A Multisource Data Fusion Modeling Prediction Method for Operation Environment of High‐Speed Train
Bohan et al. Short-term traffic flow prediction based on bus floating car

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
CB03 Change of inventor or designer information
CB03 Change of inventor or designer information

Inventor after: Liu Yingshun

Inventor after: Wang Song

Inventor after: Lu Hongyi

Inventor after: Liu Jiaming

Inventor after: Liu Zhao

Inventor before: Wang Song

Inventor before: Lu Hongyi

Inventor before: Liu Yingshun

Inventor before: Liu Jiaming

Inventor before: Liu Zhao

GR01 Patent grant
GR01 Patent grant