A data-driven method for identifying subway swaying problems

Abstract

A data-driven subway car swaying disease identification method is disclosed, including 1) collecting subway car swaying disease data; 2) pre-processing the collected subway car swaying disease data; 3) performing a first merging process on the pre-processed subway car swaying disease data; 4) performing a second merging process on the subway car swaying disease data after the first merging process; and 5) re-determining the over-limit level and determining the main car swaying disease data based on the data obtained after the second merging process. The present invention can merge and process data, eliminate false positive car swaying data, and identify car swaying diseases; effectively improve the credibility of the added ride instrument detection data, make it have a higher data value, reduce the workload of on-site review of car swaying data, improve efficiency, and save costs.

Description

Technical Field

[0001] The present invention relates to the field of rail transit technology, and in particular to a data-driven subway train swaying hazard identification method. Background Art

[0002] The subway portable line inspection instrument (hereinafter referred to as the "add-on instrument") is an instrument that is easy for personnel to carry to conduct "add-on" inspections of subway lines. By measuring the lateral and vertical vibration acceleration of the body of the EMU, locomotive or passenger car during operation, it can detect and locate the bad parts of the line and realize dynamic detection of the entire line. The "add-on instrument" is small in size, low in detection cost, short in detection cycle, and strong in detection timeliness. It is a real-time on-the-way monitoring method commonly used by the railway system and can serve as an effective supplement to the detection of large-scale track inspection vehicles.

[0003] When the lateral and vertical acceleration detected by the track meter exceeds the limit, it is called track shaking. It is generally divided into four levels (level I, level II, level III, and level IV) according to the management threshold. The higher the limit level, the more serious the track shaking. Due to the short detection cycle and flexible detection time, the frequency of using the track meter for detection is high, which can generate rich track shaking data and has become one of the key data for comprehensive assessment of track irregularity.

[0004] However, due to the limitations of the instrument itself and the influence of the external environment (such as vehicle conditions), the vehicle shaking data has a serious measurement deviation problem and false alarms are relatively frequent.

[0005] In the maintenance of subway tracks, staff are usually arranged to go to the site during the night window period to check the swaying data one by one with the help of track inspection instruments, electronic track gauges and other detection instruments, determine the exact mileage location of the swaying problem, and diagnose the cause of the swaying problem. However, the on-site review of false reports of swaying takes a lot of time and energy, resulting in reduced work efficiency and increased maintenance costs, which brings great trouble to maintenance work. Summary of the invention

[0006] In order to effectively utilize the train swaying data detected by the metro ride meter to monitor the line quality status reliably and efficiently, the present invention proposes a data-driven subway train swaying disease identification method, which combines the train swaying data at the same mileage position and the same acceleration type, eliminates false positive train swaying data, and identifies train swaying diseases; improves the credibility of the metro ride meter detection data to make it have higher data value, reduces the workload of on-site review of the train swaying data, improves efficiency and saves costs.

[0007] More specifically, according to one aspect of the present invention, a data-driven subway swaying disease identification method is provided, comprising:

[0008] 1) Collect subway car shaking disease data;

[0009] 2) Preprocess the collected subway swaying disease data;

[0010] 3) Merge the pre-processed subway swaying disease data;

[0011] 4) Perform a second merge process on the subway swaying disease data after the first merge process; and

[0012] 5) Based on the data obtained after the secondary merging process, determine the over-limit level and the main swaying vehicle disease data.

[0013] According to the implementation scheme of the present invention, in step 1), the data is collected using the ride meter, and the collected data set A includes the type of line, mileage, acceleration type, acceleration peak, overload level and vehicle number. The data set A is represented by the following formula:

[0014] A=(a1 a2 … a y … a s ) T (1)

[0015] a y =(d y m y t y p y l y n y ) (2)

[0016] Among them, A is the initially collected vehicle shaking disease dataset; a y is the yth piece of car shaking disease data, where y = 1, 2, …, s, and there are s pieces of car shaking disease data in total; d y is the line type, divided into uplink and downlink; m y is the mileage; t y is the acceleration type, which is divided into lateral (horizontal) and vertical (vertical); p y is the peak acceleration; l y It is the over-limit level, usually 1, 2, 3, 4; n y It's the car number.

[0017] According to the embodiment of the present invention, in step 2), the initially collected vehicle shaking disease data set is divided into four parts according to the line type and acceleration type, namely, the upward lateral A UH , Upward vertical A UV , Downward Horizontal A DH , Downward vertical A DV , and then preprocessing is performed in each part.

[0018] According to the embodiment of the present invention, the preprocessing includes: in each part, the line type d and the acceleration type t in each data are removed, and then each data is sorted from small to large according to mileage and vehicle number, and a sequence number is added to the beginning and a mark is added to the end of each data to obtain a data set containing sequence number i, mileage m, i , peak acceleration p i , Over-limit level i 、Vehicle No.n i and the mark x i Data strip b i , and a preprocessed data set B consisting of the data strips:

[0019] B=(b1 b2 … b i … b r ) T (3)

[0020] b i =(im i p i l i n i x i ) (4)

[0021] Among them, b i is the i-th preprocessed car swaying disease data, where i = 1, 2, …, r, r ≤ s;

[0022] The preprocessed data set B satisfies the following formula:

[0023]

[0024] According to the embodiment of the present invention, the data merging processing rules in steps 3) and 4) include: route length Within the range of z, if the train swaying disease data of the same acceleration type are detected on at least two trains with different car numbers, the train swaying disease data of the same acceleration type can be merged into the same group of train swaying disease data; the train swaying disease data that cannot be merged will be judged as false alarms and will be eliminated; the value range of z can be 10-30 meters, for example 20 meters.

[0025] According to an embodiment of the present invention, in step 3), a merging process includes:

[0026] ① If the following equations (7) and (8) are satisfied, let xi = 1; otherwise, let x i =0:

[0027] m i+1 -m i ≤z (7)

[0028] ni+1 ≠n i (8)

[0029] ② If the following equations (9), (10) and (11) are satisfied, then let x j =1:

[0030] m i+1 -m j ≤z (9)

[0031] n j ≠n i+1 (10)

[0032] x j =0 (11)

[0033] ③ If the following equations (12), (13) and (14) are satisfied, then let x j =2:

[0034] m j -m i ≤z (12)

[0035] n j ≠n i (13)

[0036] x j =0 (14)

[0037] ④ If the following formula (15) is satisfied and x i-1 ={0,2},x i =1, x j ={1,2},x k =2, x k+1 ={0,1}, then b i To b k Initially merged into the vehicle shaking disease data group (b i … b k ) T , where k ≤ r:

[0038] x i ≤x j ≤x k (15)

[0039] In this way, all pre-processed data sets B are processed through the above ①②③④ to form multiple car shaking disease data sets, which are recorded as a merged car shaking disease data set;

[0040] ⑤ b i Since it was detected on only one train, it was determined to be a false alarm.

[0041] According to an embodiment of the present invention, in step 4), the secondary merging process includes:

[0042] In a combined car sway disease data set, if there is a car sway disease data set (b c … b e ) T (c≤d≤e) and the vehicle shaking disease data group (b f … b h ) T (f≤g≤h), f=e+1 and satisfy the following equations (16) and (17), then the above two vehicle swaying disease data sets are merged into the same vehicle swaying disease data set, which is recorded as the secondary merged vehicle swaying disease data set (b c … b h ) T , where i≤c, d, e, f, g, h≤k,

[0043] m g -m d ≤z (16)

[0044] n g ≠n d (17).

[0045] According to the implementation plan of the present invention, in step 5), in each secondary combined car swaying disease data group, the value with the largest over-limit level l is selected as the car swaying disease over-limit level; and the car swaying disease data strip with the largest acceleration peak value p is selected as the main car swaying disease data, and the number of car swaying disease data strips in the secondary combined car swaying disease data group where the main car swaying disease data is located minus 1 is recorded as the number of repeated car swaying diseases.

[0046] According to an embodiment of the present invention, if the maximum over-limit level l in the secondary combined vehicle shaking disease data set is level 1, then m h -m c Should not exceed 3Z, the maximum over-limit level l is 2, then m h -m c Should not exceed 5Z.

[0047] According to an embodiment of the present invention, the data-driven subway swaying disease identification method further includes step 6), conducting on-site review of the determined main swaying disease data. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a flowchart of a data-driven subway swaying hazard identification method according to an implementation scheme of the present invention. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings. The shown contents are used to fully illustrate the contents of the present invention, but are not used to limit the present invention.

[0050] Figure 1 The figure is a flow chart of a data-driven subway swaying disease identification method according to an embodiment of the present invention. Figure 1 According to an embodiment of the present invention, the data-driven subway swaying disease identification method may include:

[0051] First, collect subway swaying disease data. For example, for a certain track, multiple data collections can be performed on different trains using a data collector, such as a ride-adding instrument.

[0052] For a specific subway line, the train swaying data detected by the train meter each time usually has several items, and each item generally includes at least: line type, mileage, acceleration type, acceleration peak, over-limit level, train number, etc. After several tests using the train meter, the initially collected train swaying disease data set A can be expressed as follows:

[0053] A=(a1 a2 … a y … a s ) T (1)

[0054] a y =(d y m y t y p y l y n y ) (2)

[0055] Where A is the initial collected vehicle shaking disease dataset

[0056] a y ——The yth car shaking data, where y=1,2,…,s,

[0057] Total s pieces of car shaking disease data

[0058] d y ——Line type, divided into upward and downward

[0059] m y ——Mileage, unit: km

[0060] t y ——Acceleration type, divided into lateral (horizontal) and vertical (vertical)

[0061] p y ——Peak acceleration, unit: g

[0062] l y——Overload level (generally divided into 1, 2, 3, and 4 levels, the higher the level, the more severe the vehicle shaking)

[0063] n y ——Vehicle number (consider only the numerical sequence part, such as 1, 2, ...)

[0064] Second, after the data of the vehicle shaking disease is collected, the data will be pre-processed, which may include classification, deletion / supplementation (insufficient data), and sorting.

[0065] More specifically, the initially collected vehicle shaking data set A can be divided into four parts according to the line type (upward and downward) and acceleration type (lateral and vertical), namely, the upward lateral part A UH , Upward vertical part A UV , Downward Horizontal Part A DH , Downward vertical part A DV Each part is preprocessed in the same way, and each part processes the data separately internally.

[0066] Below is the horizontal part of the upper row of data set A UH Let’s take this as an example. Dataset A UH It is obtained by classifying according to line type and acceleration type. Therefore, for the purpose of simplification, the line type d and acceleration type t parameters in each data in this part are removed, and then each data is sorted from small to large by mileage (m) and vehicle number (n), and the serial number i is added to the beginning of each data and the mark x is added to the end to obtain the data containing serial number i, mileage m i , peak acceleration p i , Over-limit level i 、Vehicle No.n i and the mark x i Data strip b i All processed data strips constitute the preprocessed data set B UH , as shown in equations (3) and (4):

[0067] B UH =(b1 b2 L b i L r ) T (3)

[0068] b i =(im i p i l i n i x i ) (4)

[0069] Among them, b iis the i-th pre-processed uplink lateral swaying disease data, where i = 1, 2, …, r, r ≤ s; there are r uplink lateral swaying data in total; B UH It is the preprocessed upward lateral sway disease dataset.

[0070] And preprocess the dataset B UH Satisfy the following formula:

[0071]

[0072] Similarly, for the upward vertical part A UV , Downward Horizontal Part A DH , Downward vertical part A DV After preprocessing, we can get the preprocessed data set B UV , Downward Horizontal Part B DH , Downward vertical part B DV .

[0073] Third, data merging processing is performed, and the merging processing includes primary merging processing and secondary merging processing. After many years of application practice of self-positioning and adding passengers instruments on multiple subway lines in China, the inventors have proposed rules suitable for merging and processing subway car data and judging them. Based on the rules, data processing and judgment can be realized scientifically and reasonably, as follows:

[0074] The first step is to determine the merge range, that is, determine the data within a certain route length z to be merged. The length of the merge range z is affected by the accuracy of the data collector (additional instrument), the maintenance level of the track equipment, etc., and can generally be 10-30 meters. For example, for self-positioning additional instrument detection, it can usually be 20 meters.

[0075] Within the length z of the line, if the train swaying disease data of the same acceleration type are detected on at least two trains with different car numbers, the train swaying disease data of the same acceleration type can be combined into the same group of train swaying disease data; in each group of train swaying disease data, one is selected as the main train swaying disease data to be reviewed on site, and the others are repeated train swaying disease data for reference.

[0076] The vehicle shaking disease data that cannot be merged will be judged as false alarms and will be eliminated.

[0077] As shown above, the data processing is carried out in each part. UH Take this as an example to explain the data merging process:

[0078] B UH =(b1 b2 L b i L r ) T(3)

[0079] b i =(im i p i l i n i x i ) (4)

[0080] Make judgments based on the parameters mileage (m) and vehicle number (n) in each data strip:

[0081] ① If the data strip satisfies the following equations (7) and (8), then let the label x i =1, otherwise let x i =0:

[0082] m i+1 -m i ≤z (7)

[0083] n i+1 ≠n i (8)

[0084] Then② If the data strip satisfies the following equations (9), (10) and (11), then let x j =1:

[0085] m i+1 -m j ≤z (9)

[0086] n j ≠n i+1 (10)

[0087] x j =0 (11)

[0088] ③ If the data strip satisfies the following equations (12), (13) and (14), then let x j =2:

[0089] m j -m i ≤z (12)

[0090] n j ≠n i (13)

[0091] x j =0 (14),

[0092] After assigning a value to the label x in the data strip, further judgment is made:④ If the data strip satisfies the following formula (15) and x i-1 ={0,2},xi =1, x j ={1,2},x k =2, x k+1 ={0,1}, then b i To b k Initially merged into the vehicle shaking disease data group (b i ... b k ) T , where k ≤ r:

[0093] x i ≤x j ≤x k (15)

[0094] In this way, all pre-processed data sets B are processed through the above ①②③④ to form multiple car shaking disease data sets, which are recorded as a merged car shaking disease data set;

[0095] ⑤ b i Since it was detected on only one train, it was determined to be a false alarm.

[0096] After the above-mentioned one-time merging process, a merged vehicle shaking disease data set is obtained, and the data strip marked with x=0 is determined as a false alarm.

[0097] Fourth, secondary merging process. According to an embodiment of the present invention, the secondary merging process may include:

[0098] In a combined car sway disease data set, if there is a car sway disease data set (b c … b e ) T (c≤d≤e) and the vehicle shaking disease data group (b f … b h ) T (f≤g≤h), where f=e+1 and satisfies the following equations (16) and (17), then the above two vehicle swaying disease data sets are merged into the same vehicle swaying disease data set, which is recorded as the secondary merged vehicle swaying disease data set (b c …b h ) T , where i≤c, d, e, f, g, h≤k,

[0099] m g -m d ≤z (16)

[0100] n g ≠n d (17).

[0101] The above process is then repeated until no more data can be merged. Thus, multiple secondary merged swaying disease data sets are obtained, as well as a secondary merged swaying disease data set composed of these secondary merged swaying disease data sets. In addition, if the secondary merged swaying disease data set (b c … b h ) T The maximum over-limit level l is level 1, then m h -m c Should not exceed 3z, the maximum over-limit level l is level 2, then m h -m c Should not exceed 5z.

[0102] Fifth, the over-limit level is determined and the main car swaying disease data is determined. According to the implementation scheme of the present invention, in each secondary combined car swaying disease data group, the value with the largest over-limit level l can be selected as the over-limit level of the car swaying disease; and the car swaying disease data strip with the largest acceleration peak value p can be selected as the main car swaying disease data, and the number of car swaying disease data strips where the main car swaying disease data is located minus 1 is recorded as the number of repeated car swaying disease data strips. If there are multiple over-limit data that reach the maximum acceleration peak value in the secondary combined car swaying disease data group at the same time, it can be considered to select the data with a smaller mileage or a median mileage as the main car swaying disease data.

[0103] Sixth, after determining the main swaying fault data, on-site verification can be conducted to further determine the subway swaying fault. For example, staff can go to the site and use track inspection instruments, electronic track gauges and other testing instruments to verify the main swaying fault data, determine the exact mileage location of the swaying fault, and diagnose the cause of the swaying fault.

[0104] Example

[0105] Taking a subway line in a certain city as an example, the method of the present invention is used to perform actual detection, data processing and judgment to identify subway swaying problems.

[0106] (1) Subway car shaking data collection and preprocessing

[0107] Table 1 is the initially collected vehicle shaking disease dataset A:

[0108] Table 1: Car sway disease dataset A

[0109]

[0110]

[0111]

[0112] Table 2 is the preprocessed car shaking disease dataset BTable 2: Car shaking disease dataset B

[0113]

[0114]

[0115]

[0116] (2) Implementation of data-driven subway car sway disease identification algorithm

[0117] Table 3 is a merged car swaying disease data set (the label x is assigned different values), which includes multiple car swaying disease data groups after one merger (where the label is not equal to 0, and each circled box represents a group).

[0118] Table 3: A combined car sway disease dataset

[0119]

[0120]

[0121]

[0122] As shown in the table above, after one merging process, in B UH Two car shaking disease data sets were obtained in part B, including three car shaking disease data strips numbered 1-3 and two car shaking disease data strips numbered 5-6. Similarly, in B DH Six car shaking disease data groups were obtained in the part, including car shaking disease data strips numbered 2-4, 5-7, 14-16, 17-20, 21-23 and 30-31 respectively.

[0123] Table 4 is a secondary merged car swaying disease data set, which includes multiple car swaying disease data groups after secondary merging (in the column where the mark is located, each box circle represents a group).

[0124] Table 4: Secondary merged car sway disease dataset

[0125]

[0126]

[0127]

[0128] As shown in the table above, after the secondary merging process, in B DH The six vehicle shaking disease data sets in Part B were merged into four vehicle shaking disease data sets. UH The two data sets of car shaking disease in the part remain unchanged. The other data sets of car shaking disease are marked as 0.

[0129] Table 5 below shows the results of over-limit grade determination and main car swaying disease data determination based on the secondary merged car swaying disease data set.

[0130] Table 5: Results of overload level determination and main sway vehicle disease data determination

[0131]

[0132]

[0133]

[0134] As shown in the above table, in a total of 6 secondary merged car swaying disease data groups, the over-limit level judgment and main car swaying disease data were determined (see the underlined serial number data groups).

[0135] Afterwards, the track intelligent operation and maintenance management information system was used to diagnose defects, and the static geometric dimensions of the track were measured using track inspection instruments and electronic track gauge detection instruments. The swaying car data were reviewed one by one on site, and the obtained over-limit levels and main swaying car defect data were consistent with those obtained by the above-mentioned method of the present invention, verifying the scientific nature and reliability of the method of the present invention.

[0136] Compared with the prior art, the present invention can achieve beneficial technical effects:

[0137] (1) It can quickly and effectively integrate the same swaying data, eliminate false alarm factors, more efficiently identify swaying defects, and improve the credibility of the data from the added-train instrument in the comprehensive assessment of track irregularities, making it have higher data value.

[0138] (2) The data processing of the present invention undergoes secondary merging, taking into account the merging scope of different electric passenger cars, which is relatively novel and has not been seen in existing research.

[0139] (3) After the merger, the main swaying data will be used for on-site review, and the repeated swaying data will be used for reference. Compared with the previous review of swaying data one by one, it is more conducive to the "precise repair" and "fine repair" of the track, reducing on-site workload, improving efficiency and saving costs.

[0140] The above description of the embodiments is to facilitate the understanding and application of the present invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the embodiments herein, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A data-driven subway swaying disease identification method, characterized in that: include: 1) Collect subway car shaking disease data; 2) Preprocess the collected subway swaying disease data; 3) Merge the pre-processed subway swaying disease data; 4) Perform a second merge process on the subway swaying disease data after the first merge process; as well as 5) Based on the data obtained after the secondary merging process, determine the over-limit level and the main swaying vehicle disease data.

2. The data-driven subway swaying disease identification method according to claim 1 is characterized in that: In step 1), the data is collected using the ride meter. The collected data set A includes the type of vehicle, mileage, acceleration type, acceleration peak, overload level, and vehicle number. The data set A is represented by the following formula: A=(a1 a2...a y ...a s ) T (1) a y =(d y m y t y p y l y n y ) (2) Among them, A is the initially collected vehicle shaking disease dataset; a y is the yth piece of car shaking disease data, where y = 1, 2, ..., s, and there are s pieces of car shaking disease data in total; d y is the line type, divided into uplink and downlink; m y is the mileage; t y is the acceleration type, which is divided into lateral (horizontal) and vertical (vertical); p y is the peak acceleration; l y It is the over-limit level, usually 1, 2, 3, 4; n y It's the car number.

3. The data-driven subway swaying disease identification method according to claim 2 is characterized in that: In step 2), the initially collected vehicle sway disease dataset is divided into four parts according to the line type and acceleration type, namely, the upward lateral A UH , Upward vertical A UV , Downward Horizontal A DH , Downward vertical A DV , and then preprocessing is performed in each part.

4. The data-driven subway swaying disease identification method according to claim 3 is characterized in that: Preprocessing includes: in each part, remove the line type d and acceleration type t in each data, then sort each data by mileage and vehicle number from small to large, and add a sequence number at the beginning of each data and a mark at the end to obtain a data set containing sequence number i, mileage m i , peak acceleration p i , Over-limit level i 、Vehicle No.n i and the mark x i Data strip b i , and a preprocessed data set B consisting of the data strips: B=(b1 b2...b i ...b r ) T (3) b i =(i m i p i l i n i x i ) (4) Among them, b i is the i-th preprocessed car shaking disease data, where i = 1, 2, ..., r, r ≤ s; The preprocessed data set B satisfies the following formula:

5. The data-driven subway swaying disease identification method according to claim 4 is characterized in that: The data merging and processing rules in steps 3) and 4) include: within a certain line length z, if the train swaying disease data with the same acceleration type are detected on at least two trains with different car numbers, then the train swaying disease data with the same acceleration type can be merged into the same group of train swaying disease data; the train swaying disease data that cannot be merged will be judged as a false alarm and will be eliminated; the value range of z can be 10-30 meters, for example 20 meters.

6. The data-driven subway swaying disease identification method according to claim 5 is characterized in that: In step 3), a merge process includes: ① If the following equations (7) and (8) are satisfied, then let x i =1, otherwise let x i =0: m i+1 -m i ≤z (7) n i+1 ≠n i (8) ② If the following equations (9), (10) and (11) are satisfied, then let x j =1: m i+1 -m j ≤z (9) n j ≠n i+1 (10) x j ＝0 (11) ③ If the following equations (12), (13) and (14) are satisfied, then let x j =2: m j -m i ≤z (12) n j ≠n i (13) x j ＝0 (14) ④ If the following formula (15) is satisfied and x i-1 ={0,2},x i =1, x j ={1,2},x k =2, x k+1 = {0, 1}, then b i To b k Initially merged into the vehicle shaking disease data group (b i ...b k ) 2 , where k ≤ r: x i ≤x j ≤x k (15) In this way, all pre-processed data sets B are processed through the above ①②③④ to form multiple car shaking disease data sets, which are recorded as a merged car shaking disease data set; ⑤ b i Since it was detected on only one train, it was determined to be a false alarm.

7. The data-driven subway swaying disease identification method according to claim 6 is characterized in that: In step 4), the secondary merging process includes: In a combined car shaking disease data set, if there is a car shaking disease data set (b c ...b e ) T (c≤d≤e) and the vehicle shaking disease data group (b f …b h ) T (f≤g≤h), f=e+1 and satisfy the following equations (16) and (17), then the above two vehicle swaying disease data sets are merged into the same vehicle swaying disease data set, which is recorded as the secondary merged vehicle swaying disease data set (b c ...b h ) 2 , where i≤c, d, e, f, g, h≤k, m g -m d ≤z (16) n g ≠n d (17)。 8. The data-driven subway swaying disease identification method according to claim 7 is characterized in that: In step 5), in each secondary combined car swaying disease data group, the over-limit level is selected respectively, and the largest value is used as the over-limit level of the car swaying disease; and the car swaying disease data strip with the largest acceleration peak p is selected respectively as the main car swaying disease data, and the number of car swaying disease data strips in the secondary combined car swaying disease data group where the main car swaying disease data is located minus 1 is recorded as the number of repeated car swaying diseases.

9. The data-driven subway swaying disease identification method according to claim 8, characterized in that: If the maximum over-limit level l in the secondary combined vehicle shaking disease data set is 1, then m h -m c Should not exceed 3Z, the maximum over-limit level l is level 2, then m h -m c Should not exceed 5Z.

10. The data-driven subway swaying disease identification method according to claim 1, characterized in that: The method also includes step 6), conducting on-site review of the determined main vehicle swaying disease data.

Images