A data-driven fuzzy reasoning method for urban rail transit train operation adjustment

Abstract

The present invention provides a data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains. The method includes: extracting train operation adjustment measures based on the handling records of urban rail transit emergencies, and constructing a decision tree model for train operation adjustment measures; extracting micro-adjustment features of the emergency section based on the actual and planned train operation diagram data, analyzing the influence of the congestion degree of the section trains and the train passenger flow congestion index on the train operation adjustment in actual operation, and constructing a train operation adjustment fuzzy set and fuzzy reasoning rules in combination with the train operation adjustment decision tree model; calibrating the fuzzy reasoning parameters of the fuzzy neural network; traversing the train in reverse from the emergency section, and making train operation adjustment decisions based on the micro-adjustment features according to the calibrated fuzzy neural network. The present invention provides technical support for the decision-making of train operation adjustment strategies in urban rail transit emergencies, and improves the safety and efficiency of operations.

Description

Technical Field

[0001] The present invention relates to the technical field of urban rail transit operation management, and in particular to a data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains. Background Art

[0002] With the rapid development of urban rail transit, the coverage of urban rail transit network is getting wider and wider, and the network density is getting higher and higher. In the actual train operation organization, train operation problems will be caused by various emergencies. Due to the characteristics of network transmission, train accidents in some sections will affect the operation status of the entire line or even the entire network. Failure to adjust the train operation in time will increase the impact. How to make scientific and reasonable decisions on train operation adjustment has become a key issue for rail transit operation managers.

[0003] At present, the technology and theory of train operation adjustment under emergencies are scattered at the macro and micro levels. On the one hand, the operation adjustment at the macro level is reflected by written records and experience exchanges. The occurrence scenarios of emergencies are diverse, and it is difficult to fully reflect the rich practical experience accumulated by dispatchers, and it is also difficult to apply it to the auxiliary decision-making system for train operation adjustment; on the other hand, scholars' research on train operation adjustment is relatively microscopic, mostly focusing on certain scenarios, using limited dispatching measures such as train rushing, train return, and train skipping stations to relieve the problem, which has limitations.

[0004] How to make full use of the objective data of existing emergency train operation adjustments to form a comprehensive and quantitative train operation adjustment reasoning method that combines macro and micro factors to guide train operation adjustment decisions is an urgent problem to be solved. Summary of the invention

[0005] The present invention provides a data-driven fuzzy reasoning method for adjusting the operation strategy of urban rail transit trains, so as to effectively solve the problem of adjusting trains in the event of an emergency in urban rail transit.

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

[0007] A data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains, comprising:

[0008] Based on the handling records of urban rail transit emergencies, train operation adjustment measures are extracted and a decision tree model for train operation adjustment measures is constructed;

[0009] Based on the actual operation diagram and the train plan operation diagram data, the micro-adjustment features of the emergency section are extracted, and the micro-adjustment features include the train stop time, section operation time, section train congestion degree, train passenger flow congestion degree, whether the train turns back, and whether the train stops and returns to the depot;

[0010] Analyze the influence of the actual running section train congestion degree and train passenger flow congestion index on train operation adjustment, and construct the fuzzy set and fuzzy reasoning rules for train operation adjustment;

[0011] A fuzzy neural network is constructed based on the train operation adjustment fuzzy set and fuzzy reasoning rules, and parameter adaptive training and adjustment are performed by inputting corresponding training sets under different fuzzy reasoning to calibrate the fuzzy reasoning parameters of the fuzzy neural network;

[0012] The train is traversed backwards from the emergency section, and the train operation adjustment decision is made based on the micro-adjustment characteristics of the emergency section and the fuzzy neural network after parameter calibration.

[0013] Preferably, the extracting of train operation adjustment measures based on the handling records of urban rail transit emergencies and constructing a train operation adjustment measure decision tree model include:

[0014] According to the handling records of urban rail transit emergencies, the train operation adjustment measures taken under emergencies are extracted using the text segmentation method;

[0015] According to the extracted train operation adjustment measures, the temporal sequence and causal relationship between the measures are analyzed, and the frequency of occurrence of the measures is counted;

[0016] According to the relationship between the train operation adjustment measures and the word frequency statistics, the information entropy and conditional entropy corresponding to the measures are obtained, and the measure features with high information gain rate are selected as the decision tree partition attributes to build a train operation adjustment decision tree model.

[0017] Preferably, the root node of the train operation adjustment decision tree model is the train operation adjustment feature, including: train delays, section forced stops and traffic interruptions. The goal is to make decisions on train adjustments. According to the frequency of occurrence of emergency event handling records, train delays are selected as the best partitioning feature, and the trains are divided into two subsets: late trains and normal train arrivals and departures. Each subset corresponds to a branch; then, a subtree is recursively constructed, and the best partitioning feature is selected and divided for each subset until the samples in the subset belong to the same category. The subtree constructed by each divided subset is connected to the corresponding root node to form a complete decision tree.

[0018] Preferably, the micro-adjustment features of the emergency section are extracted based on the actual operation diagram and the train plan operation diagram data, and the micro-adjustment features include the train stop time, section operation time, section train congestion degree, train passenger flow congestion degree, whether the train turns back and whether the train stops and returns to the depot, including:

[0019] (1) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in train stop time is obtained, and the calculation formula is as follows:

[0020] S i,j =(D i,j -A i,j )-(d i,j -a i,j )

[0021] Among them, S i,j is the change in the stop time of train j at station i, D i,j is the scheduled departure time of train j at station i, d i,j is the actual departure time of train j at station i, A i,j is the planned arrival time of train j at station i, a i,j is the actual arrival time of train j at station i;

[0022] (2) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in interval running time is calculated using the following formula:

[0023] R i,j =(A i,j -D i-1,j )-(a i,j -d i-1,j )

[0024] Among them, R i,j is the change in the running time of train j between station i-1 and station i, A i,j is the planned arrival time of train j at station i, D i-1,j is the scheduled departure time of train j at station i-1, a i,j is the actual arrival time of train j at station i, d i-1,j is the actual departure time of train j at station i-1;

[0025] (3) Calculate the train congestion level in the section based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0026] N=n i,i' / M

[0027] Where N is the train density between stations i and i', n i,i' is the number of trains between stations i and i';

[0028] (4) Calculate the train delay time based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0029]

[0030] Among them, S i,j is the stop time deviation of train j at station i, The interval running time deviation of train j in interval m;

[0031] (5) Calculate the number of trains that were forced to stop based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred:

[0032]

[0033] Among them, N p is the number of trains that make emergency stops in M ​​sections, n m is the number of trains that make emergency stops in section m;

[0034] (6) Calculate the degree of train passenger congestion based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0035]

[0036] Among them, P on is the average number of people on the train, P j is the number of people on the train in section M, N m is the number of trains in section m;

[0037] (7) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the formula is used to determine whether the train should turn back:

[0038]

[0039]

[0040] Among them, Z i,j V' is whether train j turns back at station i. When it is 1, it means that train j turns back at station i. When it is 0, it means that train j does not turn back at station i. i,j is the change in the number of stations that train j passes through at station i, is the number of stations that train j plans to pass through in the direction d of station i, is the number of stations actually passed by train j at station i in the driving direction d, is the number of stations actually passed by train j' in the driving direction -d. The following formula is used to judge the difference between train j and train j':

[0041]

[0042] Among them, L j,j' is whether trains j and j' connect at station i, is the actual arrival time of train j' in the travel direction -d, is the actual departure time of train j in travel direction d, is the lower limit of the train turnaround time, The lower limit of the train turnaround time;

[0043] (8) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs, determine whether the train should be stopped and returned to the depot. The formula is as follows:

[0044]

[0045]

[0046] Among them, T i,j is whether train j stops running at station i. When it is 1, it means that train j stops running at station i. When it is 0, it means that train j does not stop running at station i. The meanings of other symbols are the same as (4);

[0047] (9) For trains with different origins and destinations, when an emergency occurs, it is necessary to detain a train or add another train. For detaining a train, the following formula is used to determine:

[0048]

[0049] in, Whether train j is detained at the starting and ending stations is determined by comparing the actual arrival time of the train at the starting and ending stations with the planned arrival time. When the deviation is greater than the maximum train on-line time, the train is waiting in the vehicle depot, and the detention measure is implemented at this time;

[0050] When the line needs to add cars, the car adding situation of the corresponding train j at station i is obtained according to the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs.

[0051] Preferably, the analysis of the influence of the congestion degree of the section trains and the train passenger flow congestion index on the train operation adjustment in actual operation, and the construction of the train operation adjustment fuzzy set and fuzzy inference rules include:

[0052] (1) The linguistic variable "section train congestion degree" is set to a fuzzy set with a linguistic value of 3, including {no congestion, relatively congested, very congested}, and the domain U = [0,6], where the value refers to the average train density of M sections, and the three linguistic values ​​of the section train congestion degree obey the Gaussian distribution function;

[0053] (2) The fuzzy set with the language value of the language variable "train delay degree" as 3 includes {no delay, slight delay, severe delay}, and the domain U = [0,5000]. The value refers to the deviation between the actual arrival time and the planned arrival time of the train at the current station. The three language values ​​of the train delay degree obey the Gaussian distribution function;

[0054] (3) The fuzzy set with the language value of the language variable "train emergency stop duration" of 4 includes {no emergency stop, slight emergency stop, major emergency stop, severe emergency stop}, and the domain U = [0,1800], where the value refers to the train interval running time deviation, and the three language values ​​of the train emergency stop situation obey the Gaussian distribution function;

[0055] (4) The fuzzy set with the language value of the language variable “number of trains that have to stop” being 3 includes {small number, large number, very large number}, and the domain U = [0,6], where the value refers to the number of trains that have to stop in M ​​intervals, and the three language values ​​of the number of trains that have to stop obey the Gaussian distribution function;

[0056] (5) The fuzzy set with the language value of the language variable "train stop time" as 3 includes {normal, small time deviation, large time deviation}, and the domain U = [0,300]. The value refers to the deviation between the actual stop time of the train and the planned stop time. The three language values ​​of the train stop time obey the Gaussian distribution function;

[0057] (6) The fuzzy set with the language value of the language variable “train passenger flow congestion degree” as 3 includes {no congestion, relatively congested, very congested}, and the domain U = [0, 2000], where the value refers to the average number of passengers on the trains in the M intervals. The three language values ​​of the train passenger flow congestion degree obey the Gaussian distribution function;

[0058] (7) The clear set of the language variable "whether the station has the carriage return condition" with a language value of 2 includes {carriage return condition, no carriage return condition}, and the domain U = [0,1], where the value means: 0 represents that the station does not have the carriage return condition, and 1 represents that the station has the carriage return condition;

[0059] (8) The clear set of the language variable "whether the station has the conditions for reversing" with a language value of 2 includes {with reversing conditions, without reversing conditions}, and the domain U = [0,1], where the value means: 0 represents that the station does not have the conditions for reversing, and 1 represents that the station has the conditions for reversing;

[0060] (9) The language variable “the state of spare vehicles in the vehicle garage” is set to have a clear set of 2, including {sufficient spare vehicles, no spare vehicles}, and the domain U = [0,1], where the value means: 0 means that there are no spare vehicles in the vehicle garage and the condition for adding a vehicle cannot be met, and 1 means that there are sufficient spare vehicles in the vehicle garage;

[0061] Constructing fuzzy inference rules includes:

[0062] (1) For trains stopping at stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion, train delay, train stop duration, number of trains stopping, train stop duration, whether conditions for returning are met, whether conditions for reversing are met, and status of spare vehicles in the vehicle depot} to judge the running status of trains in adjacent sections and make train adjustment decisions;

[0063] (2) For section trains, fuzzy rules are established based on the linguistic variables {section train congestion level, train delay level, train emergency stop duration, train stop duration} in the fuzzy set to judge the running status of trains in adjacent sections and make train adjustment decisions;

[0064] (3) For trains at the origin and destination stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion level, train delay level, train emergency stop duration, number of emergency stop trains, train stop duration, and status of spare vehicles in the depot} to judge the operating status of the entire line and adjacent sections of the train and make train adjustment decisions.

[0065] Preferably, the fuzzy neural network is constructed based on the train operation adjustment fuzzy set and fuzzy inference rules, and the fuzzy inference parameters of the fuzzy neural network are calibrated by inputting the corresponding training sets under different fuzzy inferences to perform parameter adaptive training and adjustment, including:

[0066] (1) Three fuzzy neural networks are constructed for stop trains, section trains and origin-destination station trains. The fuzzy neural network includes input layer, fuzzification layer, fuzzy reasoning layer and clarification layer. The number of nodes in the input layer of the fuzzy neural network is determined by the linguistic variables in the fuzzy set. The fuzzification layer describes the membership relationship between the input node data and the fuzzy set through the membership function, and its formula is:

[0067]

[0068] Among them, α i,j is the output of the node, s i is the input data, c ij is the center point of the membership function, b ij is the width vector of the membership function, and m is the number of fuzzy classifications of the input;

[0069] The fuzzy reasoning layer completes the matching of fuzzy rules according to the connection between fuzzy reasoning rules and fuzzification layer, and realizes fuzzy operation between nodes. The number of nodes in this layer is the number of fuzzy rules. The clarity layer uses weighted average decision method to analyze the output decision, and its formula is:

[0070]

[0071] Among them, a is the output decision, ω j is the connection weight of the output layer;

[0072] (2) The weights and thresholds in the fuzzy neural network are initially calibrated. The features extracted from the actual and planned operation diagram data are used as the sample training set input. The features are passed to the fuzzification layer and the fuzzy reasoning layer through the input layer of the fuzzy neural network. The final output is obtained by weighted processing at the output layer. By comparing the error between the output strategy of the fuzzy neural network and the actual train adjustment strategy in the training set, the parameters of the fuzzy neural network are adjusted and updated to achieve adaptive adjustment of the fuzzy neural network parameters.

[0073] Preferably, the reverse traversal of the train from the emergency section and the train operation adjustment decision making based on the microscopic adjustment characteristics of the emergency section according to the fuzzy neural network after parameter calibration include:

[0074] According to the emergency handling records, the section where the emergency occurred in the line network is determined, and the train is traversed in reverse from the emergency section to calculate the micro-adjustment characteristics of the adjacent section or the opposite section of the train. It is determined whether the train is a stop train, a section train, or a start-end station train. According to the judgment result, the fuzzy neural network with calibrated parameters corresponding to the train is selected. According to the micro-adjustment characteristics of the train, the variables contained in the calibrated fuzzy neural network and their corresponding language values ​​are analyzed, and the train operation is adjusted. It is continuously iterated to achieve real-time adjustment of the train operation.

[0075] It can be seen from the technical solutions provided by the above embodiments of the present invention that the present invention provides technical support for adjusting train operation in the event of an emergency in urban rail transit, improves operational efficiency and ensures safe and efficient travel for passengers. The present invention extracts actual experience in quantitative scheduling in a data-driven manner to obtain train operation adjustment decisions, making them more scientific, objective and effective.

[0076] Additional aspects and advantages of the present invention will be given in part in the following description, which will become obvious from the following description, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0078] Figure 1 A processing flow chart of a data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains provided by an embodiment of the present invention;

[0079] Figure 2 A schematic diagram of a train operation adjustment decision tree model provided by an embodiment of the present invention;

[0080] Figure 3 A schematic diagram of a fuzzy set for train operation adjustment provided by an embodiment of the present invention;

[0081] Figure 4 An example diagram of a membership function of the degree of congestion of a section train provided by an embodiment of the present invention;

[0082] Figure 5 An example diagram of fuzzy inference rules for train operation adjustment provided by an embodiment of the present invention;

[0083] Figure 6 A schematic diagram of a fuzzy neural network for train operation adjustment provided by an embodiment of the present invention;

[0084] Figure 7 A schematic diagram of a train traversal process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0085] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0086] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or coupling. The term "and / or" used herein includes any unit and all combinations of one or more associated listed items.

[0087] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0088] To facilitate understanding of the embodiments of the present invention, several specific embodiments will be further explained below with reference to the accompanying drawings, and each embodiment does not constitute a limitation on the embodiments of the present invention.

[0089] The operation adjustment of urban rail transit trains is complex and uncertain. In order to obtain scientific and reasonable comprehensive decisions, it is necessary to extract the microscopic characteristics of line train operation based on operation data and use scientific methods to process and analyze complex characteristics. Fuzzy reasoning can handle problems with high uncertainty and ambiguity, and flexibly adapt to different problems and needs by adjusting weight parameters. It is widely used in control and decision-making problems.

[0090] The processing flow of a data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains provided by an embodiment of the present invention is as follows: Figure 1 As shown, the processing steps include the following:

[0091] Step S1: based on the handling records of urban rail transit emergencies, extract train operation adjustment measures and construct a train operation adjustment measure decision tree model;

[0092] Step S2: Based on the actual operation diagram and the train plan operation diagram data, extract the micro-adjustment features such as the train stop time, section running time, departure interval and whether to turn back in the emergency section;

[0093] Step S3: Analyze the influence of the actual running section train congestion degree, train passenger flow congestion and other indicators on the train operation adjustment, and construct the train operation adjustment fuzzy set and fuzzy reasoning rules;

[0094] Step S4: construct a fuzzy neural network based on the train operation adjustment fuzzy set and fuzzy reasoning rules, perform parameter adaptive training and adjustment by inputting corresponding training sets under different fuzzy reasoning, and calibrate the fuzzy reasoning parameters of the fuzzy neural network.

[0095] Step S5: traverse the train backward from the emergency section, and make train operation adjustment decisions based on the micro-adjustment characteristics of the emergency section according to the fuzzy neural network after parameter calibration.

[0096] Specifically, the above step S1 includes:

[0097] S11: Based on the handling records of urban rail transit emergencies, the text segmentation method is used to extract the train operation adjustment measures taken under the emergencies, such as emergency braking, train return, train suspension, etc.;

[0098] S12: Analyze the temporal sequence and causal relationship between the measures according to the extracted train operation adjustment measures, and make statistics on the occurrence frequency of the measures;

[0099] S13: According to the relationship between the train operation adjustment measures and the word frequency statistics, the information entropy and conditional entropy corresponding to the measures are obtained, and the measure features with high information gain rate are selected as the decision tree partition attributes to build a train operation adjustment decision tree model. The calculation formula of the information gain rate is as follows:

[0100]

[0101]

[0102] Where Y is the sample set of adjustment measures, a is a certain characteristic measure in the sample set, Gain_r(Y,a) is the information gain rate of the sample set divided by feature a, I(a) is the number of feature a, Gain(Y,a) is the information gain of the sample set divided by feature a, N w is the number of samples of each type of feature in the sample set, Y v Indicates the number of samples contained in the vth branch node in feature a.

[0103] Specifically, the above step S2 includes:

[0104] (1) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in train stop time is obtained, and the calculation formula is as follows:

[0105] S i,j =(D i,j -A i,j )-(d i,j -a i,j )

[0106] Among them, S i,j is the change in the stop time of train j at station i, D i,j is the scheduled departure time of train j at station i, d i,j is the actual departure time of train j at station i, A i,j is the planned arrival time of train j at station i, a i,j is the actual arrival time of train j at station i;

[0107] (2) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in interval running time is calculated using the following formula:

[0108] R i,j =(A i,j -D i-1,j )-(a i,j -d i-1,j )

[0109] Among them, R i,j is the change in the running time of train j between station i-1 and station i, A i,j is the planned arrival time of train j at station i, D i-1,j is the scheduled departure time of train j at station i-1, a i,j is the actual arrival time of train j at station i, d i-1,j is the actual departure time of train j at station i-1.

[0110] (3) Calculate the train congestion level in the section based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0111] N=n i,i' / M

[0112] Where N is the train density between stations i and i', n i,i' is the number of trains between stations i and i';

[0113] (4) Calculate the train delay time based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0114]

[0115] Among them, S i,j is the stop time deviation of train j at station i, The interval running time deviation of train j in interval m;

[0116] (5) Calculate the number of trains that were forced to stop based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred:

[0117]

[0118] Among them, N p is the number of trains that make emergency stops in M ​​sections, n m is the number of trains that make emergency stops in section m

[0119] (6) Calculate the degree of train passenger congestion based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs:

[0120]

[0121] Among them, P on is the average number of people on the train, P j is the number of people on the train in section M, N m is the number of trains in section m;

[0122] (7) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the formula is used to determine whether the train should turn back:

[0123]

[0124]

[0125] Among them, Z i,j V' is whether train j turns back at station i. When it is 1, it means that train j turns back at station i. When it is 0, it means that train j does not turn back at station i. i,j is the change in the number of stations that train j passes through at station i, is the number of stations that train j plans to pass through in the direction d of station i, is the number of stations actually passed by train j at station i in the driving direction d, is the number of stations actually passed by train j' in the driving direction -d. The following formula is used to judge the difference between train j and train j':

[0126]

[0127] Among them, L j,j' is whether trains j and j' connect at station i, is the actual arrival time of train j' in the travel direction -d, is the actual departure time of train j in travel direction d, is the lower limit of the train turnaround time, The lower limit of the train turnaround time;

[0128] (8) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs, determine whether the train should be stopped and returned to the depot. The formula is as follows:

[0129]

[0130]

[0131] Among them, T i,j is whether train j stops running at station i. When it is 1, it means that train j stops running at station i. When it is 0, it means that train j does not stop running at station i. The meanings of other symbols are the same as (4).

[0132] (9) For trains with different origins and destinations, when an emergency occurs, it is necessary to detain a train or add another train. For detaining a train, the following formula is used to determine:

[0133]

[0134] in, Whether train j is detained at the starting and ending stations is determined by comparing the actual arrival time of the train at the starting and ending stations with the planned arrival time. When the deviation is greater than the maximum train online time, the train is waiting in the vehicle depot, and the detention measure is implemented at this time.

[0135] When a line needs to add a train, its train number is significantly different from the planned train number. Therefore, based on the actual train operation diagram and planned operation diagram data during the period when the emergency occurs, the additional train situation of the corresponding train j at station i can be obtained.

[0136] Specifically, S3 includes:

[0137] (1) The fuzzy set with the language value of the language variable "section train congestion degree" of 3 includes {no congestion, relatively congested, very congested}, and the domain U = [0,6]. The value in the domain refers to the value of the average train density of M intervals (M < 6). After calculating the average train density of M intervals, the obtained value is fuzzy processed by the Gaussian distribution function and corresponds to the corresponding language value of the fuzzy set;

[0138] (2) The fuzzy set with the language value of the language variable "train delay degree" as 3 includes {no delay, slight delay, severe delay}, and the domain U = [0,5000]. The value in the domain refers to the deviation between the actual arrival time of the train at the current station and the planned arrival time. After calculating the difference between the actual arrival time of the train at the current station and the planned arrival time, the obtained value is fuzzy processed by the Gaussian distribution function and corresponds to the corresponding language value of the fuzzy set;

[0139] (3) The fuzzy set with the language value of the language variable "train emergency stop duration" set to 4 includes {no emergency stop, slight emergency stop, major emergency stop, severe emergency stop}, and the domain U = [0,1800]. The value in the domain refers to the value of the train interval running time deviation. After calculating the difference between the actual interval running time and the planned interval running time of the train, the obtained value is fuzzy processed by the Gaussian distribution function and corresponds to the corresponding language value of the fuzzy set;

[0140] (4) The fuzzy set with the language value of the language variable “the number of trains that have to stop” being 3 includes {small number, large number, very large number}, and the domain U = [0,6]. The values ​​in the domain refer to the values ​​of the number of trains that have to stop in M ​​intervals (M<6). After calculating the number of trains that have to stop in M ​​intervals, the values ​​obtained are fuzzy processed by the Gaussian distribution function and correspond to the corresponding language values ​​of the fuzzy set;

[0141] (5) The fuzzy set with the language value of the language variable "train stop time" set to 3 includes {normal, small time deviation, large time deviation}, and the domain U = [0,300]. The value in the domain refers to the deviation between the actual stop time of the train and the planned stop time. After calculating the difference between the actual stop time of the train and the planned stop time, the obtained value is fuzzy processed by the Gaussian distribution function and corresponds to the corresponding language value of the fuzzy set;

[0142] (6) The fuzzy set with the language value of 3 for the language variable “train passenger flow congestion degree” includes {no congestion, relatively congestion, very congestion}, and the domain U = [0, 2000]. The values ​​in the domain refer to the values ​​of the average number of passengers on the trains in the M intervals. After calculating the average number of passengers on the trains in the M intervals, the obtained values ​​are fuzzy processed by the Gaussian distribution function and correspond to the corresponding language values ​​of the fuzzy set;

[0143] (7) Set the language variable "whether the station has the carriage return condition" to have a clear set of 2 values, including {carriage return condition, no carriage return condition}, and the domain U = [0,1]. The values ​​in the domain refer to: 0 represents that the station does not have the carriage return condition, and 1 represents that the station has the carriage return condition;

[0144] (8) The language variable “whether the station has the conditions for reversing” is set to have a clear set of 2 values, including {with the conditions for reversing, without the conditions for reversing}, and the domain U = [0,1]. The values ​​in the domain refer to: 0 represents that the station does not have the conditions for reversing, and 1 represents that the station has the conditions for reversing;

[0145] (9) The language variable “the state of spare vehicles in the vehicle garage” is set to have a clear set of 2, including {sufficient spare vehicles, no spare vehicles}, and the domain U = [0,1]. The values ​​in the domain refer to: 0 represents that there are no spare vehicles in the vehicle garage and the conditions for adding vehicles cannot be met; 1 represents that there are sufficient spare vehicles in the vehicle garage;

[0146] Constructing fuzzy inference rules includes:

[0147] (1) For trains that stop at stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion, train delay, train forced stop duration, number of forced stop trains, train stop duration, whether conditions for returning are met, whether conditions for reversing are met, and status of spare vehicles in the vehicle depot} to judge the operating status of trains in adjacent sections, thereby making train adjustment decisions.

[0148] (2) For section trains, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion level, train delay level, train forced stop duration, train stop duration} to judge the operating status of trains in adjacent sections and make train adjustment decisions.

[0149] (3) For trains at the origin and destination stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion level, train delay level, train emergency stop duration, number of emergency stop trains, train stop duration, and status of spare vehicles in the depot} to judge the operating status of the entire line and adjacent sections of the train, thereby making train adjustment decisions.

[0150] Specifically, S4 includes: calibrating fuzzy reasoning parameters based on fuzzy neural network. Specifically including:

[0151] (1) Three fuzzy neural networks are constructed for stop trains, section trains, and origin-destination station trains. The network is divided into input layer, fuzzification layer, fuzzy reasoning layer, and clarification layer. The number of nodes in the input layer of the network is determined by the linguistic variables in the fuzzy set; the fuzzification layer describes the membership relationship between the input node data and the fuzzy set through membership functions such as Gaussian distribution function. The formula is:

[0152]

[0153] Among them, α i,j is the output of the node, s i is the input data, c ij is the center point of the membership function, b ij is the width vector of the membership function, and m is the number of fuzzy classifications of the input; the fuzzy reasoning layer is based on the fuzzy reasoning rule of claim 5, and completes the matching of fuzzy rules by connecting with the fuzzification layer to realize fuzzy operations between nodes. The number of nodes in this layer is the number of fuzzy rules; the clarity layer uses the weighted average judgment method to analyze the output decision, and its formula is:

[0154]

[0155] Among them, a is the output decision, ω j is the connection weight of the output layer;

[0156] (2) The weights and thresholds in the network are initially calibrated. The features extracted from the actual and planned running diagram data are then used as the sample training set input. The features are passed to the fuzzification layer and the fuzzy reasoning layer through the input layer of the network. The final output is obtained by weighted processing at the output layer. The network parameters are adjusted and updated by comparing the error between the output strategy of the fuzzy neural network and the actual train adjustment strategy in the training set to achieve adaptive adjustment of the network parameters.

[0157] Specifically, S5 includes: traversing the train in reverse from the emergency section, and making a train operation adjustment decision according to the train operation adjustment measure decision tree model and the fuzzy neural network after parameter calibration, including:

[0158] According to the emergency handling records, the trains are traversed in reverse from the emergency section to determine whether the train is a stop train, a section train, or a start-end station train. The train operation is adjusted according to the fuzzy neural network after parameter calibration. Continuous iteration is carried out to achieve real-time adjustment of train operation and obtain an effective relief plan.

[0159] A schematic diagram of a train operation adjustment decision tree model proposed in an embodiment of the present invention is as follows: Figure 2 As shown, the root node is the train operation adjustment feature, including: train delay, section forced stop and traffic interruption, and the goal is the decision of train adjustment. According to the occurrence frequency of emergency event handling records, the train delay is selected as the best partitioning feature, and the train is divided into two subsets: late train arrival and normal train arrival and departure. Each subset corresponds to a branch; then recursively construct a subtree, select the best partitioning feature for each subset and divide it until the samples in the subset belong to the same category. For example, for the subset of late train arrival, the section forced stop is selected as the partitioning feature, and it is divided into two subsets: train section forced stop and no forced stop; for the subset of train section forced stop, the traffic interruption is selected as the partitioning feature, and it is divided into two subsets: interrupted driving and uninterrupted driving; for the subset of interrupted driving, the train adjustment decision is selected as the partitioning feature, and it is divided into five subsets: train drop, train return, train suspension, passing without stopping and single-line two-way driving; finally, the subtree constructed by each divided subset is connected to the corresponding root node to form a complete decision tree.

[0160] A schematic diagram of a fuzzy set for train operation adjustment proposed in an embodiment of the present invention is as follows: Figure 3 As shown in the figure, according to the train micro-adjustment characteristics and the decision tree model, the variables contained in the train operation adjustment fuzzy set and their corresponding language values ​​are analyzed. At the same time, different language values ​​are expressed in numbers to characterize the degree of train operation congestion, delay, deviation, etc. Figure 3 The specific variables in the fuzzy set, their corresponding linguistic values ​​and numerical representations are represented in .

[0161] An example of a train congestion degree membership function proposed in an embodiment of the present invention is as follows: Figure 4As shown in the figure, the variables in the fuzzy set are fuzzified by the membership function of the Gaussian distribution, where the horizontal axis is the section train density, the vertical axis represents the degree of membership, and the section train density is obtained by the actual operation diagram data and the calculation formula as x (trains / section). The language value corresponding to the congestion degree of the section train in the current state is obtained through the Gaussian membership function (no congestion: 0; relatively congested: 1; very congested: 2).

[0162] An example of a fuzzy inference rule proposed in an embodiment of the present invention is shown in FIG. Figure 5 As shown, based on the decision tree and the constructed fuzzy sets, fuzzy inference rules are constructed for stop trains, section trains and origin-destination station trains.

[0163] For trains stopping at stations, by considering the train congestion level, train delay level, forced stop duration, number of forced stop trains, train stop duration, whether the conditions for returning are met, whether the conditions for reversing are met, and the variable characteristics of the status of spare vehicles in the vehicle depot, the decision is made whether the train will depart normally, depart in advance, continue to stop at stations, stop and return to the depot, or reversal. The shaded part in Table a) means: when the train in the section ahead of the train is crowded, the train is seriously delayed, the train is forced to stop, and the number of forced stop trains is large; the current station has the conditions for reversal; the train in the opposite section is not crowded, there is no delay, and the train has not made an forced stop, the current train will reversal to relieve the operating pressure of the section ahead.

[0164] For the section train, by considering the variable characteristics of the train stop time, train delay, train emergency stop time, number of emergency stop trains, and section train congestion in the section ahead of the current train, the current section, and the section behind the current train, the decision is made whether the train is in normal operation or in an emergency stop in the section. As shown in Table b), the shaded part means: when the stop time of the section ahead of the train is greatly different from the plan, the train is seriously delayed, the train is seriously emergency stopped, and the number of emergency stop trains is large; when the trains in the current section are crowded, the current train will make an emergency stop in the section to prevent safety accidents.

[0165] For trains at the starting and ending stations, by considering the variable characteristics of the train congestion level, train delay level, forced stop duration, number of forced stop trains, train passenger congestion level, train stop duration, and the status of spare vehicles in the vehicle depot in the section ahead of the current train, the current station, and the section behind the train, the decision is made whether the train will depart normally at the starting and ending stations, implement the measures of detaining trains, or add trains. As shown in Table c), the shaded part means: when the train in the section ahead of the train is severely congested, the train is seriously delayed, the train is forced to stop for a large number of forced stop trains, and the train passenger flow is not congested; the train at the current station is delayed; the train in the section behind the train is not congested or is relatively congested, the train is not delayed or is slightly delayed, the train does not make a forced stop or makes a slight forced stop and the number of forced stop trains is small, the current station implements the measures of detaining trains, and by increasing the departure interval of the train at the departure station, the pressure on the operation of the line network trains is relieved.

[0166] A schematic diagram of a fuzzy neural network proposed in an embodiment of the present invention is as follows Figure 6 As shown in the figure, the fuzzy reasoning rules are adjusted according to the train operation, and the fuzzy reasoning parameters are calibrated based on the fuzzy neural network. The fuzzy neural network consists of an input layer, a fuzzification layer, a fuzzy reasoning layer, and a clarification layer. The number of nodes in the input layer of the network is determined by the linguistic variables in the fuzzy set. According to the actual and planned operation diagram data, the corresponding variable characteristics are calculated and input into the fuzzy neural network as samples; the fuzzification layer describes the membership relationship between the input node data and the fuzzy set, which is determined by the membership function of the fuzzy set; the fuzzy reasoning layer completes the matching of fuzzy rules through the connection with the fuzzification layer according to the fuzzy reasoning rules, and realizes fuzzy operations between each node. The number of nodes in this layer is the number of fuzzy rules; the clarification layer uses the weighted average judgment method to analyze and obtain decisions. By comparing the error between the output strategy of the fuzzy neural network and the actual train adjustment strategy in the training set, the network parameters are adjusted and updated to achieve adaptive adjustment of the network parameters.

[0167] A schematic diagram of a train traversal process proposed in an embodiment of the present invention is as follows Figure 7 As shown, the train is traversed backwards from the emergency section, and the train operation adjustment decision is made according to the fuzzy neural network after parameter calibration. The specific steps are as follows:

[0168] Step 1: Determine the section / station where the network accident occurred;

[0169] Step 2: Traverse the train from the emergency section in the reverse direction of the driving direction and calculate the microscopic characteristics of the adjacent sections or opposite sections of the train;

[0170] Step 3: Determine the state of the train, that is, whether the train is currently a stop train, a section train, or a start-end train, and select the fuzzy neural network with parameter calibration corresponding to the train according to the judgment result. Calculate the variable characteristic value of the train running state in the current state as the input of the fuzzy neural network with parameter calibration, and according to the calibrated weight coefficient, after fuzzification processing by the membership function of the fuzzification layer, the corresponding fuzzy rules in the fuzzy reasoning layer are used, and then the weighted average judgment method is used in the clarity layer to analyze and make a decision, so as to adjust the train operation by using the fuzzy neural network with parameter calibration.

[0171] For example, when a train is determined to be a section train, the train stop time, train delay degree, train forced stop time and number of forced stop trains in the section / station ahead and section / station behind the train are calculated, and the train congestion degree index of the current section is calculated. The above train operation indicators are used as the input of the fuzzy neural network for section train parameter calibration. After fuzzification processing, fuzzy rule correspondence and clarity layer analysis, a decision is made as to whether the train operates normally in the section or makes an forced stop in the section.

[0172] In summary, the present invention constructs basic reasoning rules based on the handling information of emergency records and the planned and actual train operation diagrams before and after adjustment; and performs adaptive adjustment of fuzzy reasoning parameters through fuzzy neural networks. Using a data-driven approach, the actual experience of quantitative scheduling is extracted to obtain accurate quantitative train operation adjustment strategy reasoning results.

[0173] The present invention provides technical support for the decision-making of train operation adjustment strategies in the event of an emergency in urban rail transit, thereby improving the safety and efficiency of operations.

[0174] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0175] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0176] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0177] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A data-driven fuzzy reasoning method for adjusting the operation of urban rail transit trains, characterized in that: include: Based on the handling records of urban rail transit emergencies, train operation adjustment measures are extracted and a decision tree model for train operation adjustment measures is constructed; Based on the actual operation diagram and the train plan operation diagram data, the micro-adjustment features of the emergency section are extracted, and the micro-adjustment features include the train stop time, section operation time, section train congestion degree, train passenger flow congestion degree, whether the train turns back, and whether the train stops and returns to the depot; Analyze the influence of the actual running section train congestion degree and train passenger flow congestion index on train operation adjustment, and construct the fuzzy set and fuzzy reasoning rules for train operation adjustment; A fuzzy neural network is constructed based on the train operation adjustment fuzzy set and fuzzy reasoning rules, and parameter adaptive training and adjustment are performed by inputting corresponding training sets under different fuzzy reasoning to calibrate the fuzzy reasoning parameters of the fuzzy neural network; Traverse the train backward from the emergency section, and make train operation adjustment decisions based on the micro-adjustment characteristics of the emergency section according to the fuzzy neural network after parameter calibration; The method of extracting train operation adjustment measures based on the handling records of urban rail transit emergencies and constructing a train operation adjustment measure decision tree model includes: According to the handling records of urban rail transit emergencies, the train operation adjustment measures taken under emergencies are extracted using the text segmentation method; According to the extracted train operation adjustment measures, the temporal sequence and causal relationship between the measures are analyzed, and the frequency of occurrence of the measures is counted; According to the relationship between the train operation adjustment measures and the word frequency statistics, the information entropy and conditional entropy corresponding to the measures are obtained, and the measure features with high information gain rate are selected as the decision tree division attributes to build the train operation adjustment decision tree model; The calculation formula of information gain is as follows: Where Y is the sample set of adjustment measures, a is a certain characteristic measure in the sample set, Gain_r(Y,a) is the information gain rate of the sample set divided by feature a, I(a) is the number of feature a, Gain(Y,a) is the information gain of the sample set divided by feature a, N w is the number of samples of each type of feature in the sample set, Y v Indicates the number of samples contained in the vth branch node in feature a.

2. The method according to claim 1, characterized in that The root node of the train operation adjustment decision tree model is the train operation adjustment feature, including: train delay, section forced stop and train interruption. The goal is to make a decision on train adjustment. According to the frequency of occurrence of emergency event handling records, the train delay is selected as the best division feature. According to the train delay, it is divided into two subsets: late train arrival and normal train arrival and departure. Each subset corresponds to a branch. Then, the subtrees are recursively constructed, and the best partitioning features are selected for each subset and partitioned until the samples in the subset belong to the same category. The subtrees constructed by each partitioned subset are connected to the corresponding root node to form a complete decision tree.

3. The method according to claim 1, characterized in that The micro-adjustment features of the emergency section are extracted based on the actual operation diagram and the train plan operation diagram data. The micro-adjustment features include the train stop time, section operation time, section train congestion degree, train passenger flow congestion degree, whether the train turns back and whether the train stops and returns to the depot, including: (1) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in train stop time is obtained, and the calculation formula is as follows: S i,j =(D i,j -A i,j )-(d i,j -a i,j ) Among them, S i,j is the change in the stop time of train j at station i, D i,j is the scheduled departure time of train j at station i, d i,j is the actual departure time of train j at station i, A i,j is the planned arrival time of train j at station i, a i,j is the actual arrival time of train j at station i; (2) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the change in interval running time is calculated using the following formula: R i,j =(A i,j -D i-1,j )-(a i,j -d i-1,j ) Among them, R i,j is the change in the running time of train j between station i-1 and station i, A i,j is the planned arrival time of train j at station i, D i-1,j is the scheduled departure time of train j at station i-1, a i,j is the actual arrival time of train j at station i, d i-1,j is the actual departure time of train j at station i-1; (3) Calculate the train congestion level in the section based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs: N=n i,i' / M Where N is the train density between stations i and i', n i,i ' is the number of trains between stations i and i'; (4) Calculate the train delay time based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs: Among them, S i,j is the stop time deviation of train j at station i, The interval running time deviation of train j in interval m; (5) Calculate the number of trains that were forced to stop based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred: Among them, N p is the number of trains that make emergency stops in M ​​sections, n m is the number of trains that make emergency stops in section m; (6) Calculate the degree of train passenger congestion based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs: Among them, P on is the average number of people on the train, P j is the number of people on the train in section M, N m is the number of trains in section m; (7) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurred, the formula is used to determine whether the train should turn back: Among them, Z i,j V is whether train j turns back at station i. When it is 1, it means that train j turns back at station i. When it is 0, it means that train j does not turn back at station i. i ' ,j is the change in the number of stations that train j passes through at station i, is the number of stations that train j plans to pass through in the direction d of station i, is the number of stations actually passed by train j at station i in the driving direction d, is the number of stations actually passed by train j' in the driving direction -d. The following formula is used to judge the difference between train j and train j': in, is whether trains j and j' connect at station i, is the actual arrival time of train j' in the travel direction -d, is the actual departure time of train j in travel direction d, is the lower limit of the train turnaround time, The lower limit of the train turnaround time; (8) Based on the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs, determine whether the train should be stopped and returned to the depot. The formula is as follows: Among them, T i,j is whether train j stops running at station i. When it is 1, it means that train j stops running at station i. When it is 0, it means that train j does not stop running at station i. The meanings of other symbols are the same as (7); (9) For trains with different origins and destinations, when an emergency occurs, it is necessary to detain or add a train. For detaining a train, the following formula is used to determine: in, Whether train j is detained at the starting and ending stations is determined by comparing the actual arrival time of the train at the starting and ending stations with the planned arrival time. When the deviation is greater than the maximum train on-line time, the train is waiting in the vehicle depot, and the detention measure is implemented at this time; When the line needs to add cars, the car adding situation of the corresponding train j at station i is obtained according to the actual train operation diagram and planned train operation diagram data during the period when the emergency occurs.

4. The method according to claim 1, characterized in that: The analysis of the influence of the actual running section train congestion degree and train passenger flow congestion index on the train operation adjustment, and the construction of train operation adjustment fuzzy sets and fuzzy reasoning rules include: (1) The linguistic variable "section train congestion degree" is set to a fuzzy set with a linguistic value of 3, including {no congestion, relatively congested, very congested}, and the domain U = [0,6], where the value refers to the average train density of M sections, and the three linguistic values ​​of the section train congestion degree obey the Gaussian distribution function; (2) The fuzzy set with the language value of the language variable "train delay degree" as 3 includes {no delay, slight delay, severe delay}, and the domain U = [0,5000]. The value refers to the deviation between the actual arrival time and the planned arrival time of the train at the current station. The three language values ​​of the train delay degree obey the Gaussian distribution function; (3) The fuzzy set with the language value of the language variable "train emergency stop duration" of 4 includes {no emergency stop, slight emergency stop, major emergency stop, severe emergency stop}, and the domain U = [0,1800], where the value refers to the train interval running time deviation, and the three language values ​​of the train emergency stop situation obey the Gaussian distribution function; (4) The fuzzy set with the language value of the language variable "number of trains stopping forcibly" is set to 3, including {small number, large number, very large number}, and the domain U = [0,6], where the value refers to the number of trains stopping forcibly in M ​​intervals, and the three language values ​​of the number of trains stopping forcibly obey the Gaussian distribution function; (5) The fuzzy set with the language value of the language variable "train stop time" as 3 includes {normal, small time deviation, large time deviation}, and the domain U = [0,300]. The value refers to the deviation between the actual stop time of the train and the planned stop time. The three language values ​​of the train stop time obey the Gaussian distribution function; (6) The fuzzy set with the language value of the language variable "train passenger flow congestion degree" as 3 includes {no congestion, relatively congested, very congested}, and the domain U = [0, 2000], where the value refers to the average number of train passengers in M ​​intervals. The three language values ​​of the train passenger flow congestion degree obey the Gaussian distribution function; (7) Set the language variable "whether the station has the carriage return condition" to have a clear set of 2 values, including {carriage return condition, no carriage return condition}, and the domain U = [0,1], where the value means: 0 represents that the station does not have the carriage return condition, and 1 represents that the station has the carriage return condition; (8) Set the language variable "whether the station has the conditions for turning around" to have a clear set of 2 values, including {with the conditions for turning around, without the conditions for turning around}, and the domain U = [0,1], where the values ​​are: 0 means that the station does not have the conditions for turning around, and 1 means that the station has the conditions for turning around; (9) The language variable "vehicle pool spare vehicle status" is set to have a clear set of 2, including {sufficient spare vehicles, no spare vehicles}, and the domain U = [0,1], where the value means: 0 means that there are no spare vehicles in the vehicle pool and the vehicle adding condition cannot be met, and 1 means that there are sufficient spare vehicles in the vehicle pool; Constructing fuzzy inference rules includes: (1) For trains stopping at stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion, train delay, train stop duration, number of trains stopping, train stop duration, whether conditions for returning are met, whether conditions for reversing are met, and status of spare vehicles in the vehicle depot} to judge the running status of trains in adjacent sections and make train adjustment decisions; (2) For section trains, fuzzy rules are established based on the linguistic variables {section train congestion level, train delay level, train emergency stop duration, train stop duration} in the fuzzy set to judge the running status of trains in adjacent sections and make train adjustment decisions; (3) For trains at the origin and destination stations, fuzzy rules are established based on the linguistic variables in the fuzzy set {section train congestion level, train delay level, train emergency stop duration, number of emergency stop trains, train stop duration, and status of spare vehicles in the depot} to judge the operating status of the entire line and adjacent sections of the train and make train adjustment decisions.

5. The method according to claim 4, characterized in that The fuzzy neural network is constructed based on the train operation adjustment fuzzy set and fuzzy reasoning rules, and the fuzzy reasoning parameters of the fuzzy neural network are calibrated by inputting the corresponding training sets under different fuzzy reasoning to perform parameter adaptive training and adjustment, including: (1) Three fuzzy neural networks are constructed for stop trains, section trains and origin-destination station trains. The fuzzy neural network includes input layer, fuzzification layer, fuzzy reasoning layer and clarification layer. The number of nodes in the input layer of the fuzzy neural network is determined by the linguistic variables in the fuzzy set. The fuzzification layer describes the membership relationship between the input node data and the fuzzy set through the membership function, and its formula is: Among them, α i,j is the output of the node, s i is the input data, c ij is the center point of the membership function, b ij is the width vector of the membership function, and m is the number of fuzzy classifications of the input; The fuzzy reasoning layer completes the matching of fuzzy rules according to the connection between fuzzy reasoning rules and fuzzification layer, and realizes fuzzy operation between nodes. The number of nodes in this layer is the number of fuzzy rules. The clarity layer uses weighted average decision method to analyze the output decision, and its formula is: Among them, a is the output decision, ω j is the connection weight of the output layer; (2) The weights and thresholds in the fuzzy neural network are initially calibrated. The features extracted from the actual and planned operation diagram data are used as the sample training set input. The features are passed to the fuzzification layer and the fuzzy reasoning layer through the input layer of the fuzzy neural network. The final output is obtained by weighted processing at the output layer. By comparing the error between the output strategy of the fuzzy neural network and the actual train adjustment strategy in the training set, the parameters of the fuzzy neural network are adjusted and updated to achieve adaptive adjustment of the fuzzy neural network parameters.

6. The method according to claim 5, characterized in that The method of traversing the train in reverse from the emergency section and making train operation adjustment decisions based on the microscopic adjustment characteristics of the emergency section according to the fuzzy neural network after parameter calibration includes: According to the emergency handling records, the section where the emergency occurred in the line network is determined, and the train is traversed in reverse from the emergency section to calculate the micro-adjustment characteristics of the adjacent section or the opposite section of the train. It is determined whether the train is a stop train, a section train, or a start-end station train. According to the judgment result, the fuzzy neural network with calibrated parameters corresponding to the train is selected. According to the micro-adjustment characteristics of the train, the variables contained in the calibrated fuzzy neural network and their corresponding language values ​​are analyzed, and the train operation is adjusted. It is continuously iterated to achieve real-time adjustment of the train operation.

Images