An intelligent analysis method for rail transit vibration and noise based on multi-source data fusion

Abstract

The present invention relates to the field of data processing technology, and in particular to a method for intelligent analysis of rail transit vibration and noise based on multi-source data fusion. The method comprises the following steps: S1: acquiring a train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file; S2: extracting a train track operation and maintenance data file based on the train operation and maintenance data file; extracting a segmented environment data file of the train from the departure station to the terminal station based on the train environment data file; extracting a train rail damage analysis data file based on the train vibration and noise data file; S3: performing an intelligent analysis of train vibration and noise based on the train track operation and maintenance data file, the segmented environment data file, and the train rail damage analysis data file to obtain the results of the intelligent analysis. The present invention ensures safe and efficient maintenance of rail transit through multi-source data fusion and dynamic correction.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a multi-source data fusion intelligent analysis method for rail transit vibration noise. Background Art

[0002] With the continued expansion of rail transit networks, line coverage has extended to remote areas and complex terrain. However, short nighttime maintenance windows and limited maintenance resources make real-time health monitoring of large-scale rail networks difficult. Existing vibration and noise analysis technologies often rely on single-sensor data and fail to fully consider the correlation between environmental interference and historical maintenance records, leading to inaccurate damage location. Especially on long lines, where trains traverse diverse geographical environments, vibration signals are easily contaminated by external factors. Traditional methods cannot effectively distinguish inherent rail damage from environmental noise, resulting in false alarms or missed detections. Furthermore, existing systems lack the ability to dynamically correct auxiliary data such as maintenance coordinates and sections with no environmental noise, making it difficult to quantify the relationship between damage severity and maintenance requirements. While some technologies have attempted to incorporate spectral analysis, their models fail to incorporate dynamic weighting of multi-source data, resulting in rigid correction logic and inability to adapt to the changing nighttime maintenance scenarios. Therefore, an intelligent vibration and noise analysis method for long lines and complex environments is urgently needed. Through the collaborative correction and dynamic verification of multi-source data, precise rail damage location and quantitative assessment can be achieved. Summary of the Invention

[0003] In order to overcome the disadvantage of large vibration and noise interference in complex environments, the present invention provides a rail transit vibration and noise intelligent analysis method based on multi-source data fusion.

[0004] The technical implementation scheme of the present invention is: a rail transit vibration noise intelligent analysis method based on multi-source data fusion, comprising the following steps:

[0005] S1: Acquire a train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file;

[0006] S2: Based on the train operation and maintenance data file, extract the train track operation and maintenance data file; based on the train environment data file, extract the train segment environment data file from the departure station to the terminal station; based on the train vibration and noise data file, extract the train rail damage analysis data file;

[0007] S3: Based on the train track operation and maintenance data file, the segmented environment data file and the train rail damage analysis data file, intelligent analysis is performed on the train vibration noise to obtain the results of the intelligent analysis.

[0008] Preferably, the acquiring of the train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file, includes:

[0009] The train operation data file is standardized, and based on the standardized train operation data file, a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file are extracted.

[0010] Preferably, extracting the train track operation and maintenance data file based on the train operation and maintenance data file includes:

[0011] According to the arrival order of each station from the departure station to the terminal station, each stop is divided into station 1, station 2, station 3, ..., station N;

[0012] Based on the site 1, site 2, site 3, ..., site N, the geometric midpoint coordinates of the adjacent sites are obtained and divided into midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1;

[0013] If the train arrives at midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1, the train track operation and maintenance data file in the train operation and maintenance data file is extracted.

[0014] Preferably, if the train arrives at midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1, extracting the train track operation and maintenance data file from the train operation and maintenance data file includes:

[0015] Based on the train track operation and maintenance data file, extract the actual coordinates and operation and maintenance time in the historical maintenance records of the train track, and classify them according to the absolute difference between the actual coordinates and the geometric midpoint coordinates of the adjacent stations;

[0016] Based on the geometric midpoint coordinates of the adjacent site and the actual coordinates, the absolute difference between the geometric midpoint coordinates of the adjacent site and the actual coordinates is used as the first analysis data of vibration noise, and the operation and maintenance time is used as the second analysis data of vibration noise.

[0017] Preferably, extracting the segmented environment data file of the train from the departure station to the terminal station based on the train environment data file includes:

[0018] Based on the segmented environmental data file, the environmental interference area is divided according to the real-time travel time of the train, and the timestamp and mileage coordinates of each environmental parameter are obtained;

[0019] Based on the timestamp and mileage coordinates, the data of the section without environmental interference during the train's travel is obtained. The data of the section without environmental interference refers to an independent monitoring section that is not affected by external environmental factors during the train's travel. The data of the section without environmental interference includes the starting mileage mark, the ending mileage mark and the corresponding travel time of the independent monitoring section, wherein the starting mileage mark and the ending mileage mark are defined as the positions of the section without environmental interference, and the travel time is defined as the time of the section without interference.

[0020] Preferably, obtaining the environmental interference-free road section data during the train's travel based on the timestamp and mileage coordinates includes:

[0021] Based on the position of the road section without environmental interference, extracting the starting milepost and the ending milepost, and calculating the ratio of the length of the road section without interference to the total length of the monitored road section as the third analysis data of the vibration noise;

[0022] The ratio of the number of environmental interference switching boundaries to the number of non-interference road sections in the environmental interference area is used as the fourth analysis data of vibration noise, where the environmental interference switching boundary refers to a physical boundary where the environmental interference range suddenly changes.

[0023] Preferably, extracting a train rail damage analysis data file based on the train vibration noise data file comprises:

[0024] Extracting the train rail damage location, rail damage time, and damage extent based on the train rail damage analysis data file;

[0025] If the rail damage degree exceeds the preset rail damage degree threshold, the absolute difference between the rail damage location and the geometric midpoint coordinates of the adjacent station is used as the first comparative analysis data; the absolute difference between the rail damage time and the operation and maintenance time of the actual coordinates in the operation and maintenance record is used as the second comparative analysis data;

[0026] Analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection indicator;

[0027] Analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection indicator;

[0028] The third analysis data and the fourth analysis data are used as correction data for the first vibration noise abnormality detection flag and the second vibration noise abnormality detection flag.

[0029] Preferably, the analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection flag; and analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection flag include:

[0030] using an absolute difference between the first analysis data and the first comparative analysis data as a first vibration noise anomaly detection indicator;

[0031] An absolute difference between the second analysis data and the second comparative analysis data is used as a second vibration noise abnormality detection indicator.

[0032] Preferably, the intelligent analysis of train vibration noise based on the train track operation and maintenance data file, the segmented environment data file, and the train rail damage analysis data file to obtain the results of the intelligent analysis includes:

[0033] Obtaining a rail damage amount using a rail damage amount formula based on the first vibration noise anomaly detection flag and the second vibration noise anomaly detection flag;

[0034] Based on the rail damage amount, the third analysis data and the fourth analysis data, the final rail damage amount is obtained using the rail damage amount verification formula; the rail damage amount formula is as follows:

[0035]

[0036] in, is the initial damage amount, is the vibration amplitude, is the main frequency offset, 、 They are the first and second vibration noise abnormality detection marks, 、 is the modified weight coefficient, is the position matching function, is the sampling point index in the time series, indicating the subsampling, It is the total number of sampling times when the train travels from the departure station to the terminal station.

[0037] Preferably, the method of obtaining the final rail damage amount by using a rail damage amount verification formula based on the rail damage amount, the third analysis data and the fourth analysis data comprises: the rail damage amount verification formula is as follows:

[0038]

[0039] in, is the final damage amount, is the initial damage amount, For the third analysis data, For the fourth analysis data, 、 It is a correction factor calibrated based on the intensity of environmental interference.

[0040] Beneficial effects: The present invention significantly improves the accuracy and reliability of rail damage detection through multi-source data fusion and dynamic correction mechanism; based on the spatiotemporal error correction of the track midpoint coordinates and historical operation and maintenance data, it solves the problem of positioning inaccuracy caused by coordinate deviation. Combined with the length of the interference-free section and the density of the environmental switching boundary, the noise component in the vibration signal is dynamically corrected, which is suitable for complex scenes such as tunnels and bridges. Through the dynamic fusion calculation of the vibration amplitude, main frequency offset and multi-source correction parameters, the quantitative damage amount is output to avoid the false detection or missed detection caused by the single criterion of the traditional threshold method. Based on the weight distribution of abnormal identification and environmental parameters, high-risk sections are accurately located, the intensity of manual inspections is reduced, and the operation and maintenance efficiency is improved. The present invention provides reliable technical support for the full-time monitoring and safe operation of long-line rail transit. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of the rail transit vibration and noise intelligent analysis method based on multi-source data fusion of the present invention;

[0042] Figure 2 The figure is a schematic diagram of the process of extracting the train track operation and maintenance data file of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] An intelligent analysis method for rail transit vibration noise based on multi-source data fusion, such as Figure 1 As shown, the following steps are included:

[0045] S1: Acquire a train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file;

[0046] S2: Based on the train operation and maintenance data file, extract the train track operation and maintenance data file; based on the train environment data file, extract the train segment environment data file from the departure station to the terminal station; based on the train vibration and noise data file, extract the train rail damage analysis data file;

[0047] S3: Based on the train track operation and maintenance data file, the segmented environment data file and the train rail damage analysis data file, intelligent analysis is performed on the train vibration noise to obtain the results of the intelligent analysis.

[0048] Obtain a train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file, including:

[0049] The train operation data file is standardized, and based on the standardized train operation data file, a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file are extracted.

[0050] Further explanation: Train operation data from different sources (maintenance, environmental, vibration, and noise) is unified into a unified format to ensure data compatibility. Technical Effect: Eliminates data heterogeneity, facilitating subsequent fusion analysis. A brief example: Aligning data timestamps from different sensors, for example, by standardizing the sampling frequency to 1 kHz. Coordinate data is converted to a unified geographic coordinate system (such as WGS84). Additional Note: Standardization is a fundamental step in data fusion. Common methods in existing technologies include time synchronization, unit unification, and missing value filling.

[0051] Extracting the train track operation and maintenance data file based on the train operation and maintenance data file includes:

[0052] According to the arrival order of each station from the departure station to the terminal station, each stop is divided into station 1, station 2, station 3, ..., station N;

[0053] Based on the site 1, site 2, site 3, ..., site N, the geometric midpoint coordinates of the adjacent sites are obtained and divided into midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1;

[0054] If the train arrives at midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1, the train track operation and maintenance data file in the train operation and maintenance data file is extracted.

[0055] A further explanation is that Figure 2As shown, the midpoint coordinates of adjacent stations are generated in the order of train stops as the spatial reference nodes of track operation and maintenance data. Technical effect: Divide long lines into small sections to facilitate the refined matching of historical operation and maintenance records and real-time data. Operation and maintenance records usually mark the maintenance location with specific coordinates, but the actual train trajectory may deviate from the theoretical coordinates (such as the influence of curves and slopes). Logical segments are established through midpoint coordinates to associate operation and maintenance data with actual positions and reduce positioning errors. Simple example: The midpoint of station A (mileage 0km) and station B (mileage 10km) is C (mileage 5km). When the train arrives at C, the operation and maintenance data extraction is triggered. Supplementary explanation: The midpoint coordinates are the reference points for dynamic correction to ensure the spatial and temporal alignment of data.

[0056] If the train arrives at midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1, then the train track operation and maintenance data file in the train operation and maintenance data file is extracted, including:

[0057] Based on the train track operation and maintenance data file, extract the actual coordinates and operation and maintenance time in the historical maintenance records of the train track, and classify them according to the absolute difference between the actual coordinates and the geometric midpoint coordinates of the adjacent stations;

[0058] Based on the geometric midpoint coordinates of the adjacent site and the actual coordinates, the absolute difference between the geometric midpoint coordinates of the adjacent site and the actual coordinates is used as the first analysis data of vibration noise, and the operation and maintenance time is used as the second analysis data of vibration noise.

[0059] Further explanation: Positioning deviation is quantified by calculating the difference between the actual coordinates in the maintenance records and the theoretical midpoint coordinates. Technical Effect: Spatiotemporal error correction parameters (primary analysis data) are generated to eliminate vibration noise misinterpretations caused by track geometry deviation or measurement errors. Historical maintenance coordinates may deviate from the actual midpoint coordinates due to construction errors or equipment accuracy. Difference analysis can identify systematic positioning errors and dynamically correct position distortion in the vibration signal. A simple example: If the historical maintenance coordinates are C1 (5.2 km) and the midpoint coordinates are C2 (5.0 km), the difference |5.2 - 5.0| = 0.2 km is used as the primary analysis data. Additional explanation: Difference classification (e.g., grading by threshold) can distinguish between random and systematic errors, improving the targeted nature of corrections. Example classification results: Category 1: Difference ≤ 0.1 km → Minor error (no correction required); Category 2: 0.1 km < Difference ≤ 0.3 km → Moderate error (triggering an alert); Category 3: Difference > 0.3 km → Severe error (mandatory correction). Technical effect: Error hierarchical management is achieved through classification, improving correction efficiency.

[0060] Based on the train environment data file, extracting the segmented environment data file of the train from the departure station to the terminal station, including:

[0061] Based on the segmented environmental data file, the environmental interference area is divided according to the real-time travel time of the train, and the timestamp and mileage coordinates of each environmental parameter are obtained;

[0062] Based on the timestamp and mileage coordinates, the data of the section without environmental interference during the train's travel is obtained. The data of the section without environmental interference refers to an independent monitoring section that is not affected by external environmental factors during the train's travel. The data of the section without environmental interference includes the starting mileage mark, the ending mileage mark and the corresponding travel time of the independent monitoring section, wherein the starting mileage mark and the ending mileage mark are defined as the positions of the section without environmental interference, and the travel time is defined as the time of the section without interference.

[0063] A further explanation is that the technical logic for dividing sections without environmental interference is as follows: based on timestamps and mileage coordinates, independent monitoring sections that are not affected by external interference are screened. Technical effect: Distinguish the inherent damage signal of the rail from the environmental interference signal in the vibration noise. Sections without environmental interference: such as independent monitoring sections such as the inside of a tunnel (no wind and rain interference) and closed viaducts (no traffic vibration). Technical principle: Environmental interference (such as bridge vibration and tunnel wind noise) has temporal and spatial mutation characteristics, and the interference boundary can be located through time-mileage mapping. Brief example: The train travels in the tunnel from 10:00 to 10:05 (mileage 50-55km), which is marked as a section without interference. Supplementary explanation: The section without interference is a "pure" data source for vibration signal analysis and is used to calibrate the environmental noise model.

[0064] Based on the timestamp and mileage coordinates, data of a section without environmental interference during the train's travel is obtained, including:

[0065] Based on the position of the road section without environmental interference, extracting the starting milepost and the ending milepost, and calculating the ratio of the length of the road section without interference to the total length of the monitored road section as the third analysis data of the vibration noise;

[0066] The ratio of the number of environmental interference switching boundaries to the number of non-interference road sections in the environmental interference area is used as the fourth analysis data of vibration noise, where the environmental interference switching boundary refers to a physical boundary where the environmental interference range suddenly changes.

[0067] Further explanation: The technical logic for calculating the proportion of non-interference road sections is: quantify the proportion of non-interference road sections in the total monitored road sections (third analysis data). Technical effect: reflects the global intensity of environmental interference and is used to correct the damage weight. Simple example: total monitored road section is 100km, non-interference road section is 60km → third analysis data =0.6. Supplementary Note: The higher the ratio, the smaller the environmental interference, and the more the damage calculation relies on the vibration signal itself.

[0068] The technical logic for the ratio of environmental interference switching boundaries to non-interference sections is to calculate the ratio of the number of environmental interference mutation boundaries (such as tunnel entrances and bridge junctions) to the number of non-interference sections (the fourth analysis data). Technical effect: Reflects environmental complexity and dynamically adjusts the correction factor. A simple example: Number of switching boundaries = 5 (e.g., 3 tunnel entrances + 2 bridges), Number of non-interference sections = 10 → =5 / 10=0.5. Supplementary explanation: The higher the ratio, the more frequent the environmental interference, and the more correction efforts are needed. ) and the fourth analysis data ( ) not only characterizes the environmental interference characteristics, but also serves as a correction parameter for subsequent vibration and noise analysis to achieve quantitative compensation of environmental interference.

[0069] Based on the train vibration and noise data file, a train rail damage analysis data file is extracted, including:

[0070] Extracting the train rail damage location, rail damage time, and damage extent based on the train rail damage analysis data file;

[0071] If the rail damage degree exceeds the preset rail damage degree threshold, the absolute difference between the rail damage location and the geometric midpoint coordinates of the adjacent station is used as the first comparative analysis data; the absolute difference between the rail damage time and the operation and maintenance time of the actual coordinates in the operation and maintenance record is used as the second comparative analysis data;

[0072] Analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection indicator;

[0073] Analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection indicator;

[0074] The third analysis data and the fourth analysis data are used as correction data for the first vibration noise abnormality detection flag and the second vibration noise abnormality detection flag.

[0075] Further explanation is the technical logic of generating anomaly detection mark: compare the difference between the rail damage location / time and the operation and maintenance record to generate an anomaly mark ( 、 ). Technical effect: Identify potential damage points and reduce false alarm rate. If the damage location deviates too much from the operation and maintenance record, it may be a false detection (such as environmental interference); if the damage time is close to the operation and maintenance time, it may be a new damage that is not covered by historical maintenance. Brief example: If the absolute difference between the actual detection location of the rail damage (C1) and the geometric midpoint coordinates of the adjacent station (C2) |C1-C2|=0.3km (the preset threshold is 0.2km, and the threshold is determined based on the historical statistical analysis of the track geometry accuracy. For example: if the standard deviation of the track positioning error is 0.05km, the threshold is set to 3 times the standard deviation (ie 0.15km), and 0.2km is taken after reserving a safety margin.), then the first anomaly detection flag =0.3-0.2=0.1, indicating that there is a significant spatial deviation. Supplementary note: The abnormal identification needs to be combined with the third analysis data ( ) and the fourth analysis data ( )fix( 、 ) correction to avoid misjudgment by a single threshold. The third analysis data ( ) represents the global proportion of non-interference road sections, and the fourth analysis data ( ) reflects the frequency of environmental mutations, and the two are used together as correction parameters. 、 ) Dynamically adjust the damage weight, the specific relationship is as follows: shown.

[0076] Analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection flag; analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection flag, including:

[0077] using an absolute difference between the first analysis data and the first comparative analysis data as a first vibration noise anomaly detection indicator;

[0078] An absolute difference between the second analysis data and the second comparative analysis data is used as a second vibration noise abnormality detection indicator.

[0079] Further explanation is that the first abnormality detection flag ( ): By calculating the absolute difference between the rail damage location and the geometric midpoint coordinates of the adjacent station (i.e. the difference between the first analysis data and the first comparative analysis data), the spatial positioning deviation is quantified. The second anomaly detection mark ( ): By calculating the absolute difference between the time of rail damage and the operation and maintenance time of the actual coordinates in the operation and maintenance records (i.e. the difference between the second analysis data and the second comparative analysis data), the time series deviation is quantified. Technical effect: Directly quantify the degree of abnormality of positioning and time, avoiding the limitations of subjective threshold setting. For subsequent correction parameters ( 、 ) provides a dynamic weight basis to improve the robustness of the algorithm. Example: If the damage location deviation is 0.3km (the preset threshold is 0.2km), then =0.3-0.2=0.1, indicating a significant spatial anomaly. If the operation and maintenance time difference is 2 hours (the preset threshold is 1 hour), then =2-1=1, indicating a time anomaly. Additional note: Difference calculations must be based on a unified data unit (e.g., mileage in kilometers, time in hours). The larger the absolute value of the anomaly indicator, the more severe the deviation, and the more important it is to trigger the correction logic.

[0080] Based on the train track operation and maintenance data file, the segmented environment data file, and the train rail damage analysis data file, intelligent analysis is performed on the train vibration noise to obtain the results of the intelligent analysis, including:

[0081] Obtaining a rail damage amount using a rail damage amount formula based on the first vibration noise anomaly detection flag and the second vibration noise anomaly detection flag;

[0082] Based on the rail damage amount, the third analysis data and the fourth analysis data, the final rail damage amount is obtained using the rail damage amount verification formula; the rail damage amount formula is as follows:

[0083]

[0084] in, is the initial damage amount, is the vibration amplitude, is the main frequency offset, 、 They are the first and second vibration noise abnormality detection marks, 、 is the modified weight coefficient, is the position matching function, is the sampling point index in the time series, indicating the subsampling, It is the total number of sampling times when the train travels from the departure station to the terminal station.

[0085] Further explanation is that the technical logic of the rail damage formula is: integrating vibration amplitude ( ), main frequency offset ( ), abnormal identification ( 、 ) and position matching function ( ), calculate the initial damage amount ( ). Technical effect: Achieve quantitative assessment of damage degree. Formula analysis: Molecule ( · ): The product of the vibration amplitude and the main frequency offset, reflecting the strength of the damage signal.

[0086] ): Weighted square root of anomaly identification, suppressing the influence of positioning or time deviation. Position matching function : Ensure that vibration data is aligned with spatial position (e.g. Gaussian weighting function). Simple example: If =10mm, =5Hz, =0.1, =0.05, =0.5, =0.5→denominator= = ≈0.274, numerator = 10×5 = 50 → single point contribution = 50 / 0.274≈182.5. Supplementary explanation: The formula is calculated by dynamic weight ( 、 ) Balance spatiotemporal errors and vibration characteristics to avoid over-correction or under-correction. is the spatial weighting function, The location of the damage, is the sampling point mileage, such as Gaussian function: ,in, is the attenuation coefficient, which is used to enhance the weight of the vibration signal at the adjacent position. 、 Calibration through historical data regression analysis, for example: =0.6 means that the spatial positioning error accounts for 60% of the damage amount; =0.4 means that the time series error has a 40% impact on the damage amount. and Normalization is required to ensure Dimensionless and normalized are well known techniques to those skilled in the art and will not be introduced here.

[0087] Based on the rail damage amount, the third analysis data and the fourth analysis data, a rail damage amount verification formula is used to obtain a final rail damage amount, including: the rail damage amount verification formula is as follows,

[0088]

[0089] in, is the final damage amount, is the initial damage amount, For the third analysis data, For the fourth analysis data, 、 It is a correction factor calibrated based on the intensity of environmental interference.

[0090] Further explanation is that the technical logic of the rail damage verification formula is: introducing the environmental correction factor ( 、 ) for the initial damage amount ( ) for secondary correction. Technical effect: suppress the influence of environmental interference on the final result and improve reliability. Formula analysis: (1- · ): The higher the proportion of non-interference road sections ( The larger the environmental interference is, the smaller the correction coefficient is, and the closer it is to 1. · ): The higher the switching boundary ratio ( The larger the environmental interference, the more complex it is, and the correction coefficient approaches 0. A brief example: If =1000, =0.6, =0.5, =0.1, =0.2→ =1000×(1-0.1×0.6)×(1-0.2×0.5)=1000×0.94×0.9=846. Supplementary note: The correction coefficient needs to be calibrated according to the measured data, for example: (Percentage of non-interference road length) increases by 0.1 (i.e. 10%), =0.1 means the damage amount is reduced by 1%. 、 Dynamic adjustment through real-time environmental monitoring data, for example: when the bridge vibration intensity is detected to be greater than 5mm / s², Increase by 0.2 to increase the correction strength. is a dimensionless score, The calculation is based on the dimensionless initial damage and dimensionless environmental correction factors.

[0091] The above is a detailed introduction to the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. At the same time, for those skilled in the art, based on the idea of ​​the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A multi-source data fusion rail transit vibration noise intelligent analysis method, characterized by: The following steps are involved: S1: Acquire a train operation data file, wherein the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file; S2: extracting a train track operation and maintenance data file based on the train operation and maintenance data file; According to the arrival order of each station from the departure station to the terminal station, each stop is divided into station 1, station 2, station 3, ..., station N; Based on the site 1, site 2, site 3, ..., site N, the geometric midpoint coordinates of the adjacent sites are obtained and divided into midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1; If the train arrives at midpoint coordinate 1, midpoint coordinate 2, midpoint coordinate 3, ..., midpoint coordinate N-1, extract the train track operation and maintenance data file from the train operation and maintenance data file; based on the train track operation and maintenance data file, extract the actual coordinates and operation and maintenance time from the train track historical maintenance record, and classify them according to the absolute difference between the actual coordinates and the geometric midpoint coordinates of the adjacent stations; Based on the geometric midpoint coordinates of the adjacent stations and the actual coordinates, the absolute difference between the geometric midpoint coordinates of the adjacent stations and the actual coordinates is used as the first analysis data of vibration noise, and the operation and maintenance time is used as the second analysis data of vibration noise; based on the train environment data file, the segmented environment data file of the train from the departure station to the terminal station is extracted; Based on the segmented environmental data file, the environmental interference area is divided according to the real-time travel time of the train, and the timestamp and mileage coordinates of each environmental parameter are obtained; Based on the timestamp and mileage coordinates, data on a section without environmental interference during the train's travel is obtained, wherein the data on the section without environmental interference refers to an independent monitoring section that is not affected by external environmental factors during the train's travel, and the data on the section without environmental interference includes a starting mileage mark, an ending mileage mark, and a corresponding travel time of the independent monitoring section, wherein the starting mileage mark and the ending mileage mark are defined as the location of the section without environmental interference, and the travel time is defined as the time of the section without environmental interference; based on the location of the section without environmental interference, the starting mileage mark and the ending mileage mark are extracted, and a ratio of the length of the section without environmental interference to the total length of the monitoring section is calculated as third analysis data of vibration noise; The ratio of the number of environmental interference switching boundaries to the number of non-interference sections in the environmental interference area is used as fourth analysis data for vibration noise, where the environmental interference switching boundary refers to a physical boundary where a sudden change occurs in the environmental interference range; based on the train vibration noise data file, a train rail damage analysis data file is extracted; based on the train rail damage analysis data file, a train rail damage location, rail damage time, and damage extent are extracted; If the rail damage degree exceeds the preset rail damage degree threshold, the absolute difference between the rail damage location and the geometric midpoint coordinates of the adjacent station is used as the first comparative analysis data; the absolute difference between the rail damage time and the operation and maintenance time of the actual coordinates in the operation and maintenance record is used as the second comparative analysis data; Analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection indicator; Analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection indicator; using the third analysis data and the fourth analysis data as correction data for the first vibration and noise abnormality detection flag and the second vibration and noise abnormality detection flag; S3: Based on the train track operation and maintenance data file, the segmented environment data file and the train rail damage analysis data file, intelligent analysis is performed on the train vibration noise to obtain the results of the intelligent analysis.

2. The rail transit vibration noise intelligent analysis method based on multi-source data fusion according to claim 1 is characterized in that: The train operation data file is obtained, and the train operation data file includes a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file, including: The train operation data file is standardized, and based on the standardized train operation data file, a train operation and maintenance data file, a train environment data file, and a train vibration and noise data file are extracted.

3. The rail transit vibration noise intelligent analysis method based on multi-source data fusion according to claim 1 is characterized in that: analyzing the first analysis data and the first comparative analysis data to obtain a first vibration noise anomaly detection indicator; Analyzing the second analysis data and the second comparative analysis data to obtain a second vibration noise anomaly detection flag includes: using an absolute difference between the first analysis data and the first comparative analysis data as a first vibration noise anomaly detection indicator; An absolute difference between the second analysis data and the second comparative analysis data is used as a second vibration noise abnormality detection indicator.

4. The rail transit vibration noise intelligent analysis method based on multi-source data fusion according to claim 1 is characterized in that: The intelligent analysis of train vibration noise based on the train track operation and maintenance data file, the segmented environment data file, and the train rail damage analysis data file is performed to obtain the results of the intelligent analysis, including: Obtaining a rail damage amount using a rail damage amount formula based on the first vibration noise anomaly detection flag and the second vibration noise anomaly detection flag; Obtaining a final rail damage amount using a rail damage amount verification formula based on the rail damage amount, the third analysis data, and the fourth analysis data; The formula for rail damage is as follows: ， in, is the initial damage amount, is the vibration amplitude, is the main frequency offset, 、 They are the first and second vibration noise abnormality detection marks, 、 is the modified weight coefficient, is the position matching function, is the sampling point index in the time series, indicating the subsampling, It is the total number of sampling times when the train travels from the departure station to the terminal station.

5. The rail transit vibration noise intelligent analysis method based on multi-source data fusion according to claim 4 is characterized in that: The method of obtaining the final rail damage amount based on the rail damage amount, the third analysis data, and the fourth analysis data using a rail damage amount verification formula includes: ， in, is the final damage amount, is the initial damage amount, For the third analysis data, For the fourth analysis data, 、 It is a correction factor calibrated based on the intensity of environmental interference.

Images