A method for detecting train operation anomalies based on multiple data sources

Abstract

This invention relates to the field of data processing technology, and in particular to a method for detecting train operation anomalies based on multiple data sources. The method includes: data collection; identifying suspicious sections; determining sections of interest; assessing braking anomaly risk and identifying high-risk sections; adjusting preset parameters; and issuing deceleration warnings. This invention collects multiple key data sources in real time, identifies suspicious sections based on the airflow short-circuit index, then determines sections of interest by combining the airflow short-circuit index, water vapor concentration, and track humidity. Next, it assesses braking anomaly risk by combining the operating speed, braking distance, and track humidity of the sections of interest, and calculates the risk index level using relevant data to accurately locate high-risk braking sections. Finally, it adjusts preset parameters based on high-risk braking sections, re-determines high-risk braking situations, and issues deceleration warnings. This effectively solves the problems of low stopping accuracy and low risk identification accuracy caused by continuous heavy rainfall.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for detecting train operation anomalies based on multiple data. Background Technology

[0002] The subway operating environment is becoming increasingly complex, posing numerous challenges to train operations. Increasing operating speeds lead to longer emergency braking distances, placing higher demands on braking system performance. Expanding subway lines result in complex and variable tunnel environments, and aging equipment on older lines all contribute to potential train operation risks. These problems are exacerbated during severe weather events such as continuous heavy rainfall, where water accumulation in tunnels and slippery tracks frequently cause inaccurate train stops, posing serious hazards such as derailment and rear-end collisions. However, existing monitoring methods often rely on single data points, making it difficult to comprehensively assess operational status and braking risks. Therefore, there is an urgent need for train operation anomaly detection methods that integrate multiple data sources for accurate monitoring and early warning, ensuring safe train operation.

[0003] Patent document CN119389272A discloses a train stopping control method, device, electronic equipment, and storage medium. The method includes: acquiring various first deviation values ​​of train stopping within a historical time period, wherein the first deviation value refers to the positional deviation value between the actual stopping point of the train at the platform and the preset target stopping point; determining the average deviation value and the standard deviation of the deviation based on the various first deviation values; determining the predicted deviation value of the train stopping at the next platform based on the average deviation value, the standard deviation of the deviation, and a normal distribution formula; acquiring a first distance between the train and the target stopping point of the next platform for each preset cycle during the process of the train moving to the next platform; determining a second distance based on the first distance and the predicted deviation value; determining a first speed of the train for the cycle based on the second distance; and controlling the train according to the first speed during the cycle.

[0004] Therefore, the train stopping control method has the following problems: it relies heavily on the train stopping deviation value within a historical time period to predict the stopping deviation of the next platform; when changes occur, historical data loses its reference value, and historical data cannot be updated and adjusted in a timely manner, resulting in the prediction deviation not matching the actual situation and affecting the accuracy of train stopping; the prediction deviation value is determined based on the normal distribution formula, but in actual train operation, the stopping deviation does not completely conform to the assumption of a normal distribution. Summary of the Invention

[0005] To address this, the present invention provides a train operation anomaly detection method based on multiple data, which overcomes the problems of low stopping accuracy and low risk identification accuracy caused by continuous heavy rainfall in the prior art through multiple data monitoring and dynamic adjustment mechanisms.

[0006] To achieve the above objectives, the present invention provides a train operation anomaly detection method based on multiple data, comprising:

[0007] Real-time data collection includes the subway's operating speed, airflow short-circuit index, water vapor concentration, and track humidity when the distance to the next stop is at the historical average braking distance after continuous heavy rainfall; and the data also includes the airflow short-circuit index, water vapor concentration, and track humidity in each test section of the subway tunnel divided according to the preset detection length.

[0008] Several suspicious sections are determined based on the airflow short-circuit index and the preset short-circuit index threshold.

[0009] Several sections of interest are determined based on the airflow short-circuit index, water vapor concentration, and track humidity of each of the suspected sections;

[0010] Based on the operating speed, braking distance, and track humidity of each of the aforementioned concern sections, a risk of braking anomaly is determined, and the risk level of braking anomaly is determined to be high braking risk based on the risk index calculated according to the operating speed, track humidity, and airflow short-circuit index.

[0011] The preset risk index threshold is adjusted based on all the attention sections where the high braking risk occurs, to obtain the adjusted risk index threshold, or the preset detection length is adjusted to obtain the adjusted detection length;

[0012] A deceleration warning is issued based on the high braking risk determined by the adjusted risk index threshold or the adjusted detection length.

[0013] Further, determining suspicious sections based on the airflow short-circuit index and a preset short-circuit index threshold includes:

[0014] When the airflow short-circuit index is greater than the preset short-circuit index threshold, the section to be tested is determined to be the suspicious section, thereby identifying several suspicious sections.

[0015] Furthermore, based on the airflow short-circuit index, water vapor concentration, and orbital humidity of each of the suspected sections, several sections of interest are determined, including:

[0016] When the orbital humidity is greater than a preset humidity threshold, the water vapor concentrations within a preset attention period are normalized to obtain a standard concentration dataset, and the orbital humidity within a preset attention period is normalized to obtain a standard humidity dataset.

[0017] Calculate the correlation coefficient between the standard concentration dataset and the standard humidity dataset to obtain the consistency of change;

[0018] When the degree of change is greater than a preset degree of change threshold, several interest sections are determined based on the airflow short-circuit index and the water vapor concentration.

[0019] Furthermore, based on the water vapor concentration and orbital humidity within the preset attention period, several attention zones are determined, including:

[0020] Plot the change curve of the airflow short-circuit index within the preset attention period to obtain the airflow change curve;

[0021] Plot the water vapor concentration change curve within the preset attention period to obtain the concentration change curve;

[0022] Calculate the cosine similarity between the airflow change curve and the concentration change curve to obtain the degree of synchronization of change;

[0023] When the degree of change synchronization is greater than a preset synchronization threshold, the suspicious segment is determined to be the segment of interest, thereby identifying several segments of interest.

[0024] Furthermore, based on the operating speed, braking distance, and track humidity of each of the aforementioned sections of interest, a risk of braking anomaly is determined, including:

[0025] The humidity friction coefficient is calculated based on the track humidity, the preset humidity influence coefficient, and the preset friction coefficient.

[0026] Based on the humidity friction coefficient, the operating speed, and the braking distance, it is determined that there is a risk of brake malfunction.

[0027] Furthermore, based on the humidity friction coefficient, the operating speed, and the braking distance, a risk of braking abnormality is determined, including:

[0028] Calculate the braking acceleration based on the aforementioned coefficient of friction.

[0029] Calculate the braking distance based on the braking acceleration and the operating speed;

[0030] Calculate the relative deviation between the braking distance and the braking distance to obtain the braking deviation;

[0031] When the braking deviation is greater than a preset braking deviation threshold, it is determined that there is a risk of abnormal braking in the section of interest.

[0032] Furthermore, based on the operating speed, track humidity, and the risk index calculated using the airflow short-circuit index, the level of brake abnormality risk is determined to be high brake risk, including:

[0033] The operating speed is normalized to obtain a standard operating speed;

[0034] The track humidity is normalized to obtain the standard track humidity;

[0035] The airflow short-circuit index is normalized to obtain the standard airflow index;

[0036] The risk index is obtained by weighting and summing the standard operating speed, the standard track humidity, the standard airflow index, the preset speed weight, the preset humidity weight, and the preset airflow weight.

[0037] The level of the abnormal braking risk is determined as high braking risk based on the risk index and the preset risk index threshold.

[0038] Further, determining the level of the braking anomaly risk as high braking risk based on the risk index and the preset risk index threshold includes:

[0039] When the risk index is greater than the preset risk index threshold, the level of the abnormal braking risk is determined to be the high braking risk.

[0040] Further, adjusting the preset risk index threshold based on all the attention sections where the high braking risk occurs, to obtain an adjusted risk index threshold, or adjusting the preset detection length, to obtain an adjusted detection length, includes:

[0041] Calculate the standard deviation of the location of each of the interest sections on the subway operating route to obtain the dispersion.

[0042] When the dispersion is greater than the maximum value of the preset dispersion range, the number of times the dispersion is greater than the maximum value of the preset dispersion range within the preset adjustment time is counted to obtain the dispersion count;

[0043] When the number of dispersions exceeds a preset dispersion threshold, the preset risk index threshold is increased based on the relative deviation between the dispersion and the maximum value of the preset dispersion range and a preset threshold adjustment coefficient to obtain the adjusted risk index threshold.

[0044] When the dispersion is less than the minimum value of the preset dispersion range, the preset detection length is adjusted according to the dispersion and the preset dispersion range to obtain the adjusted detection length.

[0045] Furthermore, the preset detection length is adjusted based on the dispersion and the preset dispersion range to obtain the adjusted detection length, including:

[0046] The number of times the dispersion is less than the minimum value of the preset dispersion range within the preset adjustment time is counted to obtain the set number;

[0047] When the number of concentrations exceeds a preset concentration threshold, the preset detection length is increased based on the relative deviation between the dispersion and the minimum value of the preset dispersion range and a preset length adjustment coefficient to obtain an adjusted detection length.

[0048] Compared with existing technologies, the advantages of this invention lie in its ability to determine whether the subway is operating at a reasonable speed by real-time collection of the operating speed when the distance to the stopping station is within the historical average braking distance. The airflow short-circuit index, water vapor concentration, and track humidity in each test section within the tunnel reflect the tunnel ventilation, air moisture content, and track humidity, all of which affect the subway's braking performance. First, suspicious sections are identified based on the airflow short-circuit index and preset thresholds. Then, the airflow short-circuit index, water vapor concentration, and track humidity of the suspicious sections are combined to determine sections of interest, gradually filtering out areas with potential braking risks. Next, the operating speed, braking distance, and track humidity of the sections of interest are combined to determine the risk of abnormal braking, and the risk index level is calculated using relevant data, accurately locating high-risk braking situations. Finally, preset parameters are adjusted based on high-risk braking sections to re-determine the risk and issue a deceleration warning, effectively solving the problems of low stopping accuracy and low risk identification accuracy caused by continuous heavy rainfall.

[0049] Furthermore, by collecting the airflow short-circuit index in real time and comparing it with the preset short-circuit index threshold, the section to be tested is determined to be a suspicious section when the airflow short-circuit index exceeds the preset threshold. The airflow short-circuit index directly reflects the ventilation status in the tunnel. When it is greater than the preset threshold, it indicates that the ventilation performance of the section has deteriorated and there may be problems such as water accumulation and excessive humidity. Suspicious sections that need to be focused on can be quickly and accurately screened out.

[0050] Furthermore, by analyzing the relationship between track humidity, water vapor concentration, and airflow short-circuit index, potential risk areas in subway operation can be identified more accurately. When track humidity exceeds a preset threshold, water vapor concentration and track humidity are normalized to obtain standard concentration and humidity datasets. The correlation coefficient between these two datasets is calculated to obtain the consistency of change, reflecting whether the trends of water vapor concentration and track humidity change over time are consistent. If the consistency of change is greater than the preset threshold, it indicates that the trends of these two factors are highly correlated over time, which may suggest a persistent water accumulation or moisture buildup problem in that section. This, combined with the airflow short-circuit index, may increase the risk of subway operation, effectively identifying areas that require special attention and further treatment.

[0051] Furthermore, by calculating the cosine similarity between the airflow change curve and the concentration change curve, the degree of similarity in shape between the two curves can be quantified, thereby reflecting the synchronicity of the airflow short-circuit index and water vapor concentration changes over time. When the synchronicity of change is greater than a preset synchronicity threshold, it indicates that the two factors have a highly consistent trend over time, suggesting that there are common environmental influencing factors in this section, such as continuous water vapor accumulation or poor ventilation. These problems are closely related to the risks of subway operation, and sections that require special attention can be identified.

[0052] Furthermore, by considering the impact of track humidity on friction, the humidity-friction coefficient is used to quantify the change in braking performance caused by humidity. Track humidity, as an environmental factor, together with the preset humidity influence coefficient and friction coefficient, determines the humidity-friction coefficient, which in turn affects braking performance. Running speed and braking distance are basic parameters of train dynamics, and together with the friction coefficient, they determine whether the train can stop within a specified distance. This can effectively identify the risk of abnormal braking and improve the accuracy of subway operation.

[0053] Furthermore, by calculating braking acceleration, the train's deceleration capability under specific humidity conditions can be precisely quantified. The humidity friction coefficient directly reflects the impact of track humidity on braking performance. Operating speed is the core dynamic parameter determining braking distance; the higher the speed, the longer the required braking distance. Combining the humidity friction coefficient with operating speed to calculate the theoretical braking distance provides an ideal braking performance under current environmental and speed conditions. By comparing this with the actual required braking distance, the braking deviation can be obtained, intuitively reflecting the gap between actual braking performance and theoretical expectations. When the braking deviation exceeds a preset threshold, a risk of braking anomaly is identified, effectively identifying the reliability of the braking system.

[0054] Furthermore, by calculating the risk index using standard operating speed, standard track humidity, standard airflow index, and a preset weighted array, the following parameters are considered: operating speed directly affects the braking distance required; higher speeds result in longer braking distances. Track humidity affects the friction between the wheels and the track; higher humidity may reduce friction, leading to increased braking distance. The airflow short-circuit index reflects the ventilation conditions within the tunnel; poor ventilation may cause moisture accumulation, thus affecting track humidity and braking performance. Therefore, by weighting and summing these three standardized parameters, the combined effects of speed, humidity, and airflow on braking performance can be comprehensively considered, thus providing a more comprehensive and accurate reflection of actual braking risks.

[0055] Furthermore, a risk index is obtained by comprehensively considering multiple key factors such as operating speed, track humidity, and airflow short-circuit index, and by weighting and summing them according to their respective importance. A preset risk index threshold serves as the established benchmark. When the risk index exceeds this preset threshold, it indicates that the current combination of operating conditions has reached a high-risk level, enabling the system to quickly and intuitively determine whether there is a high risk of braking, and thus take timely and appropriate measures.

[0056] Furthermore, the dispersion is determined by calculating the standard deviation of the locations of all high-braking-risk sections of concern along the subway route. If the dispersion exceeds the maximum value of the preset dispersion range, it indicates that these sections are too spatially dispersed. In this case, the number of times this occurs within a preset adjustment period is counted, i.e., the dispersion count. When the dispersion count exceeds the preset dispersion count threshold, the high-braking-risk sections of concern are too frequently dispersed. The preset risk index threshold is increased by calculating the relative deviation between the dispersion and the preset maximum value and combining it with the preset threshold adjustment coefficient, thus obtaining a new adjusted risk index threshold, making the system's risk assessment more stringent. On the other hand, if the dispersion is lower than the minimum value of the preset dispersion range, it indicates that the sections of concern are too concentrated. In this case, the preset detection length is adjusted according to the relationship between the dispersion and the preset range, resulting in an adjusted detection length to optimize the division of detection intervals and improve the targeting and efficiency of detection.

[0057] Furthermore, by counting the number of clusterings when the dispersion is consistently below the preset minimum value and comparing it with a threshold, it can be determined whether the phenomenon of excessive clustering in the area of ​​interest is universal and persistent. When the number of clusterings exceeds the preset threshold, the detection length can be adjusted based on the relative deviation of the dispersion from the preset minimum value and the preset coefficient. This can expand the detection range, make the detection area more representative, and avoid detection blind spots caused by excessive clustering in the area of ​​interest due to an excessively small detection interval, thereby improving the comprehensiveness and accuracy of the detection. Attached Figure Description

[0058] Figure 1 This is a flowchart of the train operation anomaly detection method based on multiple data in this embodiment;

[0059] Figure 2 This is a logic diagram for determining suspicious sections in this embodiment;

[0060] Figure 3 This embodiment defines the logic diagram for determining the area of ​​interest.

[0061] Figure 4 This is a logic diagram for determining the risk of brake malfunction in this embodiment. Detailed Implementation

[0062] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0063] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0064] Please see Figure 1 As shown, it is a flowchart of the train operation anomaly detection method based on multiple data in this embodiment;

[0065] This embodiment provides a train operation anomaly detection method based on multiple data sources, including:

[0066] Real-time data collection includes the subway's operating speed, airflow short-circuit index, water vapor concentration, and track humidity when the distance to the next stop is at the historical average braking distance after continuous heavy rainfall; and the data also includes the airflow short-circuit index, water vapor concentration, and track humidity in each test section of the subway tunnel divided according to the preset detection length.

[0067] Several suspicious sections are determined based on the airflow short-circuit index and the preset short-circuit index threshold.

[0068] Several sections of interest are determined based on the airflow short-circuit index, water vapor concentration, and track humidity of each of the suspected sections;

[0069] Based on the operating speed, braking distance, and track humidity of each of the aforementioned concern sections, a risk of braking anomaly is determined, and the risk level of braking anomaly is determined to be high braking risk based on the risk index calculated according to the operating speed, track humidity, and airflow short-circuit index.

[0070] The preset risk index threshold is adjusted based on all the attention sections where the high braking risk occurs, to obtain the adjusted risk index threshold, or the preset detection length is adjusted to obtain the adjusted detection length;

[0071] A deceleration warning is issued based on the high braking risk determined by the adjusted risk index threshold or the adjusted detection length.

[0072] After periods of heavy and prolonged rainfall, changes in the tunnel environment, such as water accumulation and slippery tracks, often lead to inaccurate train stops. Therefore, real-time monitoring of the airflow short-circuit index, water vapor concentration, and track humidity within the tunnels, combined with historical average braking distances and the metro's real-time operating speed, ensures precise docking of train doors with the platform. The historical average braking distance refers to the average distance required for the metro to come to a complete stop from its current speed under normal conditions; this distance can be calculated based on historical data. "Each monitored section" refers to the metro tunnel being divided into multiple sections according to a pre-defined length for segmented monitoring. The airflow short-circuit index reflects the ventilation conditions within the tunnel and measures airflow; it is detected by ventilation sensors or airflow monitoring equipment uniformly installed within the tunnel. Water vapor concentration refers to the amount of water vapor in the air within the tunnel. After heavy rainfall, water vapor concentration typically increases significantly, potentially causing slippery tracks and affecting train braking performance; it can be measured by humidity sensors uniformly installed within the tunnel. Track humidity refers to the degree of moisture on the track surface, which directly affects the friction between the wheels and the track. After heavy rainfall, track humidity increases and friction decreases, resulting in an increase in the braking distance of the train. It is expressed as a percentage (between 0 and 1) and can be obtained using track humidity sensors that are uniformly installed in the tunnel.

[0073] The preset detection length is pre-set based on factors such as the subway tunnel structure and train operation safety requirements. It serves as a standard length for dividing the tunnel into different test sections, and depends on the curve radius, gradient changes, and station spacing of the subway tunnel. It is typically set between 50 and 100 meters. In this embodiment, it is set to 70 meters to balance detection accuracy and computational efficiency.

[0074] The preset short-circuit index threshold is a boundary value used to determine whether the degree of airflow short-circuiting in a section of a subway tunnel is abnormal. It depends on the design standards of the subway tunnel's ventilation system, the airflow distribution pattern during normal operation, and past operating data, and is usually set between 0.1 and 0.5. In this embodiment, it is set to 0.3, which effectively filters out suspicious sections with relatively obvious airflow short-circuiting phenomena.

[0075] The preset risk index threshold is a boundary value used to determine the risk level of train braking. It depends on the performance of the train braking system, the relationship between the track friction coefficient and humidity, the impact of airflow short circuits on brake heat dissipation and air resistance, and historical operating data, and is usually set between 0.6 and 0.9. In this embodiment, it is set to 0.8 to ensure that a deceleration warning is issued in a timely manner when the risk is high, thus ensuring the safety of the train stopping.

[0076] By collecting real-time data on the subway's operating speed, braking distance, airflow short-circuit index, water vapor concentration, and track humidity in each test section after continuous heavy rainfall, suspicious sections are identified based on the airflow short-circuit index and a preset short-circuit index threshold. Then, sections of concern are determined by combining the airflow short-circuit index, water vapor concentration, and track humidity. Based on the operating speed, braking distance, and track humidity of the sections of concern, it is determined that there is a risk of abnormal braking. A risk index is calculated using the operating speed, track humidity, and airflow short-circuit index, and the risk level is determined to be high braking risk. The preset parameters are adjusted based on the sections of concern with high braking risk, and a deceleration warning is issued based on the adjusted parameters to re-determine the high braking risk.

[0077] By collecting real-time data on the operating speed at which the distance to the stopping station falls within the historical average braking distance, it is possible to determine whether the subway is operating at a reasonable speed. The airflow short-circuit index, water vapor concentration, and track humidity in each test section within the tunnel reflect the tunnel's ventilation, air moisture content, and track humidity, respectively, all of which affect the subway's braking performance. First, suspicious sections are identified based on the airflow short-circuit index and preset thresholds. Then, by combining the airflow short-circuit index, water vapor concentration, and track humidity of these suspicious sections, sections of interest are determined, gradually filtering out areas with potential braking risks. Next, the operating speed, braking distance, and track humidity of the sections of interest are combined to determine the risk of abnormal braking, and the risk index level is calculated using relevant data, accurately locating high-risk braking situations. Finally, preset parameters are adjusted based on high-risk braking sections, the risk is reassessed, and a deceleration warning is issued, effectively solving the problems of low stopping accuracy and low risk identification accuracy caused by continuous heavy rainfall.

[0078] Please continue reading. Figure 2 As shown, it is the logic diagram for determining suspicious sections in this embodiment;

[0079] Suspicious sections are determined based on the airflow short-circuit index and a preset short-circuit index threshold, including:

[0080] When the airflow short-circuit index is greater than the preset short-circuit index threshold, the section to be tested is determined to be the suspicious section, thereby identifying several suspicious sections.

[0081] By comparing the airflow short-circuit index with a preset short-circuit index threshold, the section to be tested is determined to be a suspicious section when the airflow short-circuit index is greater than the preset short-circuit index threshold, so as to identify several suspicious sections.

[0082] By collecting the airflow short-circuit index in real time and comparing it with the preset short-circuit index threshold, the section to be tested is determined to be a suspicious section when the airflow short-circuit index exceeds the preset threshold. The airflow short-circuit index directly reflects the ventilation status in the tunnel. When it is greater than the preset threshold, it indicates that the ventilation performance of the section has deteriorated and there may be problems such as water accumulation and excessive humidity. Suspicious sections that need to be focused on can be quickly and accurately screened out.

[0083] Specifically, based on the airflow short-circuit index, water vapor concentration, and orbital humidity of each of the suspected sections, several sections of interest are determined, including:

[0084] When the orbital humidity is greater than a preset humidity threshold, the water vapor concentration of all items within a preset attention period is normalized to obtain a standard concentration dataset, and the orbital humidity of all items within a preset attention period is normalized to obtain a standard humidity dataset.

[0085] Calculate the correlation coefficient between the standard concentration dataset and the standard humidity dataset to obtain the consistency of change;

[0086] When the degree of change is greater than a preset degree of change threshold, several interest sections are determined based on the airflow short-circuit index and the water vapor concentration.

[0087] The preset humidity threshold is a benchmark value used to determine whether the track humidity is too high. It depends on historical data, the moisture resistance of the track material, and subway operation safety standards, and is usually set between 0.6 and 0.8. In this embodiment, it is set to 0.7, which can more sensitively detect abnormal changes in track humidity and promptly screen out sections that may pose a risk.

[0088] The preset monitoring duration refers to the time range considered when determining the monitoring segment. It depends on the rate of data change, the speed of risk development, and the response capability of the monitoring system, and is usually set between 5 and 15 minutes. In this embodiment, it is set to 10 minutes to ensure that enough effective data is collected, while also guaranteeing the timeliness and accuracy of the analysis.

[0089] The preset consistency threshold is a benchmark value used to measure the correlation between orbital humidity and water vapor concentration. It depends on the principles of statistical correlation analysis and the risk assessment requirements in practical applications, and is typically set between 0.6 and 0.8. In this embodiment, it is set to 0.7, which can effectively filter out highly correlated segments while ensuring the accuracy of data correlation analysis.

[0090] When the orbital humidity exceeds a preset humidity threshold, the water vapor concentration and orbital humidity within a preset monitoring period are normalized to obtain standard concentration and standard humidity datasets. The correlation coefficient between the two datasets is then calculated to determine the consistency of change. Finally, when the consistency of change exceeds a preset consistency threshold, several monitoring sections are further identified based on the airflow short-circuit index and water vapor concentration of the suspected sections.

[0091] By analyzing the relationship between track humidity, water vapor concentration, and airflow short-circuit index, potential risk areas in subway operation can be identified more accurately. When track humidity exceeds a preset threshold, water vapor concentration and track humidity are normalized to obtain standard concentration and humidity datasets. The correlation coefficient between these two datasets is calculated to obtain the consistency of change, reflecting whether the trends of water vapor concentration and track humidity change consistently over time. If the consistency of change is greater than the preset threshold, it indicates that the trends of these two factors are highly correlated over time, potentially suggesting a persistent water accumulation or moisture buildup problem in that section. This, combined with the airflow short-circuit index, may increase the risk of subway operation, effectively identifying areas requiring special attention and further treatment.

[0092] Please continue reading. Figure 3 As shown, it is the logic diagram for determining the area of ​​interest in this embodiment;

[0093] Based on the water vapor concentration and orbital humidity within the preset attention period, several attention zones are determined, including:

[0094] Plot the change curve of the airflow short-circuit index within the preset attention period to obtain the airflow change curve;

[0095] Plot the water vapor concentration change curve within the preset attention period to obtain the concentration change curve;

[0096] Calculate the cosine similarity between the airflow change curve and the concentration change curve to obtain the degree of synchronization of change;

[0097] When the degree of change synchronization is greater than a preset synchronization threshold, the suspicious segment is determined to be the segment of interest, thereby identifying several segments of interest.

[0098] The preset synchronization threshold is a standard value used to determine whether the airflow change curve and the concentration change curve are sufficiently similar. It depends on the complexity of the subway operating environment, the correlation requirements of airflow and water vapor concentration changes in the tunnel, and the statistical results of historical data, and is usually set between 0.5 and 0.9. In this embodiment, it is set to 0.7, which can effectively filter out curve similarity caused by accidental factors while ensuring a certain level of sensitivity, making the identified sections of interest more practical and reliable.

[0099] By plotting the changes in airflow short-circuit index and water vapor concentration within a preset attention period, airflow change curves and concentration change curves are obtained. Then, the cosine similarity between the two curves is calculated to obtain the change synchronicity. Finally, when the change synchronicity is greater than the preset synchronicity threshold, the suspicious segment is determined as a segment of interest, thereby identifying several segments of interest.

[0100] By calculating the cosine similarity between the airflow change curve and the concentration change curve, the degree of similarity in shape between the two curves can be quantified, thus reflecting the synchronicity of the airflow short-circuit index and water vapor concentration changes over time. When the synchronicity is greater than a preset threshold, it indicates that the two factors have a highly consistent trend over time, suggesting that there are common environmental influencing factors in this section, such as continuous water vapor accumulation or poor ventilation. These problems are closely related to the risks of subway operation, and sections that require special attention can be identified.

[0101] Specifically, the risk of brake malfunction is determined based on the operating speed, braking distance, and track humidity of each of the aforementioned sections of interest, including:

[0102] The humidity friction coefficient is calculated based on the track humidity, the preset humidity influence coefficient, and the preset friction coefficient.

[0103] Based on the humidity friction coefficient, the operating speed, and the braking distance, it is determined that there is a risk of brake malfunction.

[0104] The formula for calculating the coefficient of friction due to humidity is: μ w =μ0×(1-H×k);

[0105] Where: μ w is the humidity friction coefficient (dimensionless); μ0 is the preset friction coefficient (dimensionless); H is the track humidity (expressed as a percentage, between 0 and 1, dimensionless); k is the preset humidity influence coefficient (dimensionless).

[0106] The preset humidity influence coefficient is a parameter used to quantify the degree of influence of track humidity on the coefficient of friction. It reflects the changing trend of track friction performance under humidity conditions and depends on factors such as track material, surface condition, and subway operating environment. It is usually set between 0 and 1. In this embodiment, it is set to 0.2, which can accurately reflect the actual impact of humidity on friction, neither being overly conservative nor underestimating the risk.

[0107] The preset friction coefficient refers to the basic friction coefficient between the subway track and wheels under normal dry conditions. It depends on the material, surface roughness, and processing technology of the track and wheels, and is usually set between 0.1 and 0.4. In this embodiment, it is set to 0.3, which conforms to the actual friction level of a typical subway system under dry conditions and can provide a reliable benchmark for subsequent calculations.

[0108] The humidity friction coefficient is calculated by taking into account the track humidity, the preset humidity influence coefficient, and the preset friction coefficient. Then, the braking distance corresponding to the humidity friction coefficient, the current running speed, and the distance to the next stopping station is comprehensively considered to determine whether there is a risk of abnormal braking.

[0109] By considering the impact of track humidity on friction, the humidity-friction coefficient is used to quantify the change in braking performance. Track humidity, as an environmental factor, together with the preset humidity influence coefficient and friction coefficient, determines the humidity-friction coefficient, which in turn affects braking performance. Running speed and braking distance are basic parameters of train dynamics. They, together with the friction coefficient, determine whether the train can stop within a specified distance, which can effectively identify the risk of abnormal braking and improve the accuracy of subway operation.

[0110] Please continue reading. Figure 4 As shown, it is the logic diagram for determining the risk of brake malfunction in this embodiment;

[0111] Specifically, determining the risk of brake malfunction based on the humidity friction coefficient, the operating speed, and the braking distance includes:

[0112] Calculate the braking acceleration based on the aforementioned coefficient of friction.

[0113] Calculate the braking distance based on the braking acceleration and the operating speed;

[0114] Calculate the relative deviation between the braking distance and the braking distance to obtain the braking deviation;

[0115] When the braking deviation is greater than a preset braking deviation threshold, it is determined that there is a risk of abnormal braking in the section of interest.

[0116] The formula for calculating braking acceleration is:

[0117] Where: a w F is the braking acceleration (unit: m / s²); b Braking force provided to the braking system (unit: kg·m / s) 2 ); m is the mass of the subway car (in kg); μ w The coefficient of friction due to humidity (dimensionless); This is the pre-set braking acceleration provided by the train's braking system under normal circumstances.

[0118] The formula for calculating braking distance is:

[0119] Where: d is the braking distance (in meters); v is the train speed (in meters per second); aw Braking acceleration (unit: m / s²) 2 ).

[0120] The preset braking deviation threshold refers to a standard value used to determine the allowable deviation range between the actual braking distance and the expected braking distance of a train. It depends on various factors such as the design braking performance of the subway vehicle, the normal friction conditions of the track, and safety operation standards, and is typically set between 5% and 10%. In this embodiment, it is set to 8% to avoid frequent false alarms due to an excessively low threshold, which could interfere with normal operation, while also tolerating reasonable environmental changes and measurement fluctuations to a certain extent.

[0121] Braking acceleration is calculated using the coefficient of friction based on humidity. Then, the theoretical braking distance is calculated using the braking acceleration and the current operating speed. Subsequently, the calculated braking distance is compared with the actual required braking distance to determine the relative deviation, i.e., the braking deviation. If this braking deviation exceeds a preset braking deviation threshold, the area of ​​concern is deemed to have a risk of abnormal braking.

[0122] By calculating braking acceleration, the deceleration capability of a train under specific humidity conditions can be precisely quantified. The humidity friction coefficient directly reflects the impact of track humidity on braking performance. Operating speed is the core dynamic parameter determining braking distance; the higher the speed, the longer the required braking distance. Combining the humidity friction coefficient with operating speed to calculate the theoretical braking distance provides an ideal braking performance under current environmental and speed conditions. Comparing this to the actual required braking distance yields the braking deviation, which intuitively reflects the gap between actual braking performance and theoretical expectations. When the braking deviation exceeds a preset threshold, a risk of braking anomaly is identified, effectively assessing the reliability of the braking system.

[0123] Specifically, the risk level of brake anomaly is determined as high brake risk based on a risk index calculated according to operating speed, track humidity, and the airflow short-circuit index, including:

[0124] The operating speed is normalized to obtain a standard operating speed;

[0125] The track humidity is normalized to obtain the standard track humidity;

[0126] The airflow short-circuit index is normalized to obtain the standard airflow index;

[0127] The risk index is obtained by weighting and summing the standard operating speed, the standard track humidity, the standard airflow index, the preset speed weight, the preset humidity weight, and the preset airflow weight.

[0128] The level of the abnormal braking risk is determined as high braking risk based on the risk index and the preset risk index threshold.

[0129] The preset speed weight refers to the weight value assigned to operating speed when calculating the risk index. It depends on the degree of influence of operating speed on the risk of brake malfunction and is usually set between 0.4 and 0.6. In this embodiment, it is set to 0.5, which highlights the importance of operating speed in brake risk assessment, giving it a larger weight in the calculation of the risk index, and thus more sensitively reflecting the impact of speed changes on brake risk.

[0130] The preset humidity weight refers to the weight value assigned to track humidity when calculating the risk index. It depends on the degree of influence of track humidity on the coefficient of friction between the wheel and the track, and is usually set between 0.2 and 0.5. In this embodiment, it is set to 0.3, which can reasonably reflect the impact of humidity on braking performance, while avoiding overemphasizing the role of humidity and maintaining the balance of risk assessment.

[0131] The preset airflow weight refers to the weight value assigned to the airflow short-circuit index when calculating the risk index. It depends on the indicative significance of the airflow short-circuit index for tunnel ventilation conditions and potential risks, and is typically set between 0.1 and 0.3. In this embodiment, it is set to 0.2, which considers both the indirect impact of airflow factors on braking risk and avoids overstating the role of other more direct factors.

[0132] Standard operating speed, standard track humidity, and standard airflow index are obtained by normalizing the operating speed, track humidity, and airflow index respectively. Then, these standard values ​​are weighted and summed with their respective preset weights to calculate a comprehensive risk index. Finally, this risk index is compared with a preset risk index threshold to determine whether the braking anomaly risk reaches a high braking risk level.

[0133] The risk index is calculated using standard operating speed, standard track humidity, standard airflow index, and a preset weighted array. Operating speed directly affects the braking distance; higher speeds require longer braking distances. Track humidity affects the friction between the wheels and the track; higher humidity may reduce friction, leading to increased braking distance. The airflow short-circuit index reflects the ventilation conditions within the tunnel; poor ventilation can cause moisture accumulation, further affecting track humidity and braking performance. Therefore, by weighting and summing these three standardized parameters, the combined effects of speed, humidity, and airflow on braking performance can be comprehensively considered, resulting in a more complete and accurate reflection of actual braking risks.

[0134] Specifically, determining the level of the braking anomaly risk as high braking risk based on the risk index and the preset risk index threshold includes:

[0135] When the risk index is greater than the preset risk index threshold, the level of the abnormal braking risk is determined to be the high braking risk.

[0136] By comparing the risk index with a preset risk index threshold, when the risk index is greater than the preset risk index threshold, the level of abnormal braking risk is determined to be high braking risk.

[0137] The risk index is derived by comprehensively considering multiple key factors such as operating speed, track humidity, and airflow short-circuit index, and by weighting and summing them according to their respective importance. A preset risk index threshold serves as the baseline. When the risk index exceeds this preset threshold, it indicates that the current combination of operating conditions has reached a high-risk level, enabling the system to quickly and intuitively determine whether there is a high risk of braking and thus take timely and appropriate measures.

[0138] Specifically, adjusting the preset risk index threshold based on all the attention sections where the high braking risk occurs, to obtain an adjusted risk index threshold, or adjusting the preset detection length, to obtain an adjusted detection length, includes:

[0139] Calculate the standard deviation of the location of each of the interest sections on the subway operating route to obtain the dispersion.

[0140] When the dispersion is greater than the maximum value of the preset dispersion range, the number of times the dispersion is greater than the maximum value of the preset dispersion range within the preset adjustment time is counted to obtain the dispersion count;

[0141] When the number of dispersions exceeds a preset dispersion threshold, the preset risk index threshold is increased based on the relative deviation between the dispersion and the maximum value of the preset dispersion range and a preset threshold adjustment coefficient to obtain the adjusted risk index threshold.

[0142] When the dispersion is less than the minimum value of the preset dispersion range, the preset detection length is adjusted according to the dispersion and the preset dispersion range to obtain the adjusted detection length.

[0143] The preset adjustment duration refers to the length of the time window considered when evaluating and adjusting parameters. It depends on the system's response speed requirements and data stability, and is usually set between 10 and 30 minutes. In this embodiment, it is set to 15 minutes, which can capture risk changes in a relatively short time while avoiding excessive data fluctuations due to an excessively short time window.

[0144] The preset dispersion range refers to the normal fluctuation range of the dispersion of the section of interest. It depends on the layout of the subway line and historical operation data, and is usually set between 0.3 and 1.0. In this embodiment, it is set to [0.5 to 0.8], which can accurately reflect the distribution of the section of interest under normal operating conditions and provide a reasonable benchmark for the detection of anomalies in dispersion.

[0145] The preset dispersion threshold refers to the number of times the dispersion exceeds the maximum value of the preset dispersion range within a preset adjustment period. It depends on the system's tolerance for the frequency of risk events and is typically set between 2 and 5 times. In this embodiment, it is set to 3 times, which allows for timely identification of frequently occurring high-risk dispersion events without being overly sensitive, thereby triggering the parameter adjustment mechanism.

[0146] The preset threshold adjustment coefficient refers to the gain factor used to adjust the preset risk index threshold. It depends on the sensitivity of the risk assessment model and the system stability requirements, and is usually set between 0.1 and 0.3. In this embodiment, it is set to 0.2, which can moderately increase the risk index threshold when the dispersion is large, making the system more sensitive to high-risk situations, while avoiding frequent alarms or false alarms caused by excessive adjustment.

[0147] The dispersion is determined by calculating the standard deviation of the locations of all high-risk braking sections along the subway route. If the dispersion exceeds the maximum value of a preset dispersion range, the number of times this occurs within a preset adjustment period is counted, i.e., the dispersion count. When the dispersion count exceeds a preset dispersion count threshold, the preset risk index threshold is increased by calculating the relative deviation between the dispersion and the preset maximum value and combining it with a preset threshold adjustment coefficient, thus obtaining a new adjusted risk index threshold. On the other hand, if the dispersion is lower than the minimum value of a preset dispersion range, the preset detection length is adjusted according to the relationship between the dispersion and the preset range, resulting in an adjusted detection length. The system can adaptively optimize risk assessment and detection strategies based on the actual risk distribution.

[0148] The dispersion is determined by calculating the standard deviation of the locations of all high-risk braking sections along the subway route. If the dispersion exceeds the maximum value of the preset dispersion range, it indicates that these sections are too spatially dispersed. In this case, the number of times this occurs within a preset adjustment period is counted, i.e., the dispersion count. When the dispersion count exceeds a preset dispersion count threshold, the high-risk braking sections are too frequently dispersed. The preset risk index threshold is increased by calculating the relative deviation between the dispersion and the preset maximum value and combining it with a preset threshold adjustment coefficient, thus obtaining a new adjusted risk index threshold, making the system's risk assessment more stringent. On the other hand, if the dispersion is below the minimum value of the preset dispersion range, it indicates that the sections of concern are too concentrated. In this case, the preset detection length is adjusted according to the relationship between the dispersion and the preset range, resulting in an adjusted detection length to optimize the division of detection intervals and improve the targeting and efficiency of detection.

[0149] Specifically, the preset detection length is adjusted based on the dispersion and a preset dispersion range to obtain the adjusted detection length, including:

[0150] The number of times the dispersion is less than the minimum value of the preset dispersion range within the preset adjustment time is counted to obtain the set number;

[0151] When the number of concentrations exceeds a preset concentration threshold, the preset detection length is increased based on the relative deviation between the dispersion and the minimum value of the preset dispersion range and a preset length adjustment coefficient to obtain an adjusted detection length.

[0152] The preset threshold for the number of times the dispersion is less than the minimum value of the preset dispersion range within a preset adjustment period is a warning value. It depends on the layout, operating conditions, historical data, and safety standards of the subway line, and is typically set between 5 and 15 times. In this embodiment, it is set to 10 times, which can effectively identify situations where the focus area is too concentrated and adjust the detection strategy in a timely manner.

[0153] The preset length adjustment coefficient is a coefficient used to adjust the preset detection length according to the dispersion. It depends on the sensitivity requirements of the detection length adjustment, the line length, and the operating speed, and is usually set between 0.1 and 0.5. In this embodiment, it is set to 0.3, which allows for flexible adjustment of the detection length according to the concentration of the section of interest. When the section is too concentrated, appropriately increasing the detection length can expand the monitoring range.

[0154] The number of clustering occurrences is obtained by counting the number of times the dispersion is less than the minimum value of the preset dispersion range within a preset adjustment period. When the number of clustering occurrences exceeds a preset clustering occurrence threshold, the preset detection length is increased based on the relative deviation between the current dispersion and the minimum value of the preset dispersion range, combined with a preset length adjustment coefficient, thus obtaining the adjusted detection length.

[0155] By counting the number of clusterings when the dispersion is consistently below the preset minimum value and comparing it with a threshold, it can be determined whether the phenomenon of excessive concentration in the area of ​​interest is universal and persistent. When the number of clusterings exceeds the preset threshold, the detection length can be adjusted based on the relative deviation of the dispersion from the preset minimum value and the preset coefficient. This can expand the detection range, make the detection area more representative, and avoid detection blind spots caused by excessive concentration in the area of ​​interest due to an excessively small detection interval, thereby improving the comprehensiveness and accuracy of the detection.

[0156] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A train operation abnormality detection method based on multi-data, characterized by, The method comprises: collecting, in real time, a subway running speed at a braking distance from a next stop site being equal to a historical average, an airflow short circuit index of each to-be-tested section in a subway running tunnel divided according to a preset detection length, a water vapor concentration, and a track humidity; determining suspicious sections according to the airflow short circuit index and a preset short circuit index threshold value; determining attention sections according to the airflow short circuit index, the water vapor concentration, and the track humidity of each suspicious section; determining that there is a brake abnormal risk according to the running speed, the braking distance, and the track humidity of each attention section, and determining a high brake risk level of the brake abnormal risk according to a risk index calculated from the running speed, the track humidity, and the airflow short circuit index; adjusting a preset risk index threshold value according to all the attention sections where the high brake risk occurs to obtain an adjusted risk index threshold value, or adjusting the preset detection length to obtain an adjusted detection length; issuing a deceleration prompt according to the high brake risk re-determined based on the adjusted risk index threshold value or the adjusted detection length.

2. The train operation abnormality detection method based on multiple data according to claim 1, characterized by, The method for determining suspicious sections according to the airflow short circuit index and a preset short circuit index threshold value comprises: when the airflow short circuit index is greater than the preset short circuit index threshold value, determining that the to-be-tested section is the suspicious section, so as to determine the suspicious sections.

3. The train operation abnormality detection method based on multiple data according to claim 2, characterized by, The method for determining attention sections according to the airflow short circuit index, the water vapor concentration, and the track humidity of each suspicious section comprises: when the track humidity is greater than a preset humidity threshold value, performing normalization processing on all the water vapor concentrations in a preset attention time length to obtain a standard concentration data set, and performing normalization processing on all the track humidities in the preset attention time length to obtain a standard humidity data set; calculating a correlation coefficient of the standard concentration data set and the standard humidity data set to obtain a change consistency degree; when the change consistency degree is greater than a preset consistency degree threshold value, determining the attention sections according to the airflow short circuit index and the water vapor concentration.

4. The train operation abnormality detection method based on multiple data according to claim 3, characterized by, The method for determining attention sections according to the water vapor concentration and the track humidity in the preset attention time length comprises: drawing a change curve of the airflow short circuit index in the preset attention time length to obtain an airflow change curve; drawing a change curve of the water vapor concentration in the preset attention time length to obtain a concentration change curve; calculating a cosine similarity of the airflow change curve and the concentration change curve to obtain a change synchronization degree; when the change synchronization degree is greater than a preset synchronization degree threshold value, determining that the suspicious section is the attention section, so as to determine the attention sections.

5. The train operation abnormality detection method based on multiple data according to claim 4, characterized by, The method for determining that there is a brake abnormal risk according to the running speed, the braking distance, and the track humidity of each attention section comprises: calculating a humidity friction coefficient according to the track humidity, a preset humidity influence coefficient, and a preset friction coefficient; determining that there is a brake abnormal risk according to the humidity friction coefficient, the running speed, and the braking distance.

6. The train operation abnormality detection method based on multiple data according to claim 5, characterized by, The method for determining that there is a brake abnormal risk according to the humidity friction coefficient, the running speed, and the braking distance comprises: calculating a brake acceleration according to the humidity friction coefficient; calculating a braking distance according to the running speed and the braking acceleration; calculating a braking deviation according to the braking distance and the braking distance; determining that the concerned section has a braking abnormal risk when the braking deviation is greater than a preset braking deviation threshold.

7. The train operation abnormality detection method based on multiple data according to claim 6, characterized by, determining that the concerned section has a high braking risk according to a risk index calculated according to the running speed, track humidity and the airflow short circuit index, including: normalizing the running speed to obtain a standard running speed; normalizing the track humidity to obtain a standard track humidity; normalizing the airflow short circuit index to obtain a standard airflow index; performing weighted summation on the standard running speed, the standard track humidity, the standard airflow index, a preset speed weight, a preset humidity weight and a preset airflow weight to obtain a risk index; determining the level of the braking abnormal risk according to the risk index and a preset risk index threshold.

8. The train operation abnormality detection method based on multiple data according to claim 7, characterized by, determining the level of the braking abnormal risk according to the risk index and a preset risk index threshold, including: determining that the level of the braking abnormal risk is the high braking risk when the risk index is greater than the preset risk index threshold.

9. The train operation abnormality detection method based on multiple data according to claim 8, characterized by, adjusting a preset risk index threshold to obtain an adjusted risk index threshold or adjusting the preset detection length to obtain an adjusted detection length according to all the concerned sections that have the high braking risk, including: calculating a standard deviation of the positions of the concerned sections on the subway running route to obtain a dispersion degree; counting the number of times that the dispersion degree is greater than the maximum value of the preset dispersion degree range within a preset adjustment time period to obtain a dispersion frequency when the dispersion degree is greater than the maximum value of the preset dispersion degree range; increasing the preset risk index threshold according to the relative deviation between the dispersion degree and the maximum value of the preset dispersion degree range and a preset threshold adjustment coefficient to obtain an adjusted risk index threshold when the dispersion frequency is greater than a preset dispersion frequency threshold; adjusting the preset detection length according to the dispersion degree and the preset dispersion degree range to obtain an adjusted detection length when the dispersion degree is less than the minimum value of the preset dispersion degree range.

10. The train operation abnormality detection method based on multiple data according to claim 9, characterized by, adjusting the preset detection length according to the dispersion degree and the preset dispersion degree range to obtain an adjusted detection length, including: counting the number of times that the dispersion degree is less than the minimum value of the preset dispersion degree range within the preset adjustment time period to obtain a concentration frequency; increasing the preset detection length according to the relative deviation between the dispersion degree and the minimum value of the preset dispersion degree range and a preset length adjustment coefficient to obtain an adjusted detection length when the concentration frequency is greater than a preset concentration frequency threshold.

Images