Steel rail overhaul section prediction method based on probability density evolution algorithm

Abstract

The invention is suitable for the technical field of rail transit operation and maintenance, and provides a steel rail overhaul section prediction method based on a probability density evolution algorithm, and the method comprises the steps: collecting steel rail damage data through on-site detection, storing the data in a relational database, and carrying out the cleaning and standardization processing of the data; analyzing the data, including correlation analysis and modeling of a steel rail state evolution model; predicting a steel rail overhaul section based on the constructed steel rail state evolution model; and outputting and visualizing a result. According to the invention, the steel rail section needing to be replaced can be accurately predicted, so that the scientificity and efficiency of subway operation and maintenance management are improved.

Description

Technical Field

[0001] The present invention relates to the field of rail transit operation and maintenance technology, and in particular to a rail overhaul section prediction method based on a probability density evolution algorithm. Background Art

[0002] With the acceleration of urbanization, subways, as a vital component of urban public transportation, are crucial for their operational safety and efficiency. Rails, a key component of subway track structures, have a direct impact on train safety and passenger comfort. However, due to the long-term impact of train loads and environmental factors, subway rails inevitably suffer from wear, cracks, and damage, which can lead to rail failure and even serious safety accidents.

[0003] Currently, subway rail maintenance and management primarily rely on regular inspections and manual judgment. Common inspection methods include ultrasonic testing and magnetic particle testing. While these methods can detect rail defects, they cannot fully reflect the rail's overall condition and future development trends. Furthermore, traditional maintenance strategies often rely on fixed rail replacement intervals, lacking accurate prediction of rail condition evolution, leading to wasted resources and potential safety hazards.

[0004] In recent years, with the development of big data technology, some research has utilized data analysis methods to optimize rail operation and maintenance management. For example, by analyzing data such as the cumulative total weight of rails, operating hours, and damage density, the remaining life of rails can be predicted. However, existing methods are mostly based on single-dimensional data, fail to fully consider the complexity of rail state evolution, and have limited prediction accuracy. The existing technology lacks a method that can comprehensively consider multiple types of operation and maintenance data and predict rail overhaul sections based on the evolution of probability density distribution. Therefore, developing a rail overhaul section prediction method based on big data and probability density algorithms has important practical significance and application value. Summary of the Invention

[0005] The present invention provides a rail overhaul section prediction method based on a probability density evolution algorithm to solve the technical problems mentioned in the background technology.

[0006] A method for predicting rail overhaul sections based on a probability density evolution algorithm, the method comprising:

[0007] S1. Collect rail damage data through on-site inspections, including the total weight of rails, the number of damaged rails per kilometer, the year the rails were put into operation, the maximum allowable speed of the line, the proportion of curves with a radius of less than 800 meters, and the total weight of rails per year. This data is then stored in a relational database and cleaned and standardized.

[0008] S2. Analyze the data, including correlation analysis and rail state evolution modeling;

[0009] S3. Predict rail overhaul areas based on the constructed rail state evolution model;

[0010] S4. Result output and visualization.

[0011] As a further technical solution of the present invention, in step S1, rail damage data is collected through on-site testing and stored in a relational database. The steps of cleaning and standardizing the data include:

[0012] Data Collection:

[0013] Rail damage data is collected through on-site inspection;

[0014] Data Storage:

[0015] The collected data is stored in an Oracle database; multiple data tables are created to store rail operation and maintenance data of different dimensions; key and index technologies are used to establish associations between different data tables to optimize data query efficiency and ensure that data of different dimensions accurately corresponds to the same rail;

[0016] Data processing:

[0017] The data was cleaned and standardized. Outliers were identified and eliminated using the quartile method. Duplicate uploaded data was queried and unique values ​​were retained. For records with missing items, if the total weight of the rails or the number of damaged items per kilometer was missing, the corresponding records were deleted. If other items were missing, the previous and next rails were matched using mileage information, and missing values ​​were filled using linear interpolation. The Min-Max standardization method was used to convert the data to a uniform scale.

[0018] As a further technical solution of the present invention, in step S2, the steps of analyzing the data, including correlation analysis and rail state evolution model building, include:

[0019] Correlation analysis:

[0020] Based on the random forest model, the correlation between different rail damage data and overhaul and replacement is explored, and the cumulative total weight of rails and the number of serious damages per kilometer are given higher weight coefficients;

[0021] Modeling of rail state evolution model:

[0022] Based on the collected rail damage data, the probability density function of the rail state is constructed using multidimensional kernel density estimation (KDE). The joint probability density function is defined as:

[0023]

[0024] Where:

[0025] is a multidimensional Gaussian kernel function: ,in ;

[0026] is the bandwidth matrix, which controls the degree of smoothing in each dimension;

[0027] is the determinant of the bandwidth matrix;

[0028] is the sample size;

[0029] is the i-th sample vector.

[0030] By analyzing the changing trend of the probability density function, a rail state evolution model is established to describe the transition process of the rail from the normal state to the damaged state.

[0031] As a further technical solution of the present invention, in step S3, the step of predicting the rail overhaul section based on the constructed rail state evolution model includes:

[0032] Based on the constructed rail state evolution model, the probability distribution of rail damage within a certain time period in the future is predicted; based on the prediction results, high-risk areas where the damage probability exceeds the set threshold (such as 75%) are located, and combined with actual operation and maintenance needs, the rail sections that need major repairs or replacement are determined.

[0033] As a further technical solution of the present invention, in step S4, the step of outputting and visualizing the results includes:

[0034] Provides a detailed prediction report, including prediction results: rail sections are divided into high, medium, and low risk levels, and the rail sections that need major repairs and their priorities are listed; damage probability distribution: displays the probability density distribution of rail damage; confidence interval: provides the confidence interval of the prediction results to reflect the uncertainty of the prediction;

[0035] Provide visual data reports to show the risk level or priority of different sections, or the changing trend of rail damage probability over time, providing a scientific basis for operation and maintenance decisions.

[0036] Beneficial effects achieved by the present invention:

[0037] Technically, by integrating multidimensional operation and maintenance data (such as cumulative total weight, damage density, and rail time online) and combining it with a probability density evolution algorithm, the model can more comprehensively reflect the changing patterns of rail condition. The model supports real-time data input and dynamic updates, adapting predictions based on the latest operation and maintenance data to complex operating environments. Compared to traditional single-metric prediction methods, this model can quantify the uncertainty of rail condition and significantly improve prediction accuracy.

[0038] In terms of economic benefits, accurate prediction of rail overhaul locations can avoid unnecessary rail replacement and maintenance, concentrating limited maintenance resources in high-risk areas. This reduces resource waste caused by excessive or untimely maintenance and increases rail lifespan. The model's predictions can provide data support for long-term planning and budgeting for subway systems, reducing long-term operation and maintenance costs. Scientific planning can avoid unexpected large expenditures caused by rail problems.

[0039] In terms of social benefits, accurate prediction and timely maintenance can effectively reduce the incidence of rail failures, improve the safety and reliability of subway transportation, reduce the risk of accidents caused by rail failures, and ensure public safety. More intelligent and refined rail health management will also improve the quality of public services and enhance public trust in rail transportation. This model, based on big data and probability density algorithms, aligns with the development direction of smart cities and promotes the intelligent upgrade of rail transit systems. Through technological innovation, we can improve the overall level of urban public transportation systems and contribute to the construction of smart cities.

[0040] In summary, the rail overhaul section prediction model proposed in this invention can not only improve the scientificity and efficiency of subway rail operation and maintenance, but also provide strong support for the sustainable development of urban public transportation systems. It has broad application prospects and social value. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of a rail overhaul section prediction method based on a probability density evolution algorithm of the present invention. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0043] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0044] See also Figure 1 The embodiment of the present invention provides a method for predicting rail overhaul sections based on a probability density evolution algorithm, the method comprising:

[0045] S1. Collect rail damage data through on-site inspections, including the total weight of rails, the number of damaged rails per kilometer, the year the rails were put into operation, the maximum allowable speed of the line, the proportion of curves with a radius of less than 800 meters, and the total weight of rails per year. This data is then stored in a relational database and cleaned and standardized.

[0046] S2. Analyze the data, including correlation analysis and rail state evolution modeling;

[0047] S3. Predict rail overhaul areas based on the constructed rail state evolution model;

[0048] S4. Result output and visualization.

[0049] In this embodiment, in step S1, rail damage data is collected through on-site testing and stored in a relational database. The steps of cleaning and standardizing the data include:

[0050] Data Collection:

[0051] Rail damage data is collected through on-site inspections using ultrasonic flaw detectors and entered into the database through a "manual and automatic" approach. Dedicated data personnel upload and enter indicators such as the cumulative total weight of rails and the time they were online into the database based on the rail operation and maintenance records of the line work area.

[0052] Data Storage:

[0053] The collected data is stored in an Oracle database, leveraging its high performance and reliability to manage massive amounts of data. Multiple data tables are created to store rail operation and maintenance data of different dimensions, facilitating partition management and improving query performance. Key and indexing techniques are used to establish relationships between different data tables, optimizing data query efficiency and ensuring that data of different dimensions accurately corresponds to the same rail.

[0054] Data processing:

[0055] The data is cleaned and standardized. Outliers are identified and eliminated using the quartile method. Duplicate uploaded data is queried and unique values ​​are retained. If the total weight of the rails or the number of damaged rails per kilometer is missing, the corresponding records are deleted. If other items are missing, the previous and next rails are matched using mileage information, and linear interpolation is used to fill in the missing values. The Min-Max standardization method is used to convert the data to a uniform scale.

[0056] In this embodiment, in step S2, the steps of analyzing the data, including correlation analysis and rail state evolution model building, include:

[0057] Correlation analysis:

[0058] The random forest model was used to explore the correlation between different rail damage data and overhaul and replacement, assigning higher weight coefficients to the cumulative total weight of rails and the number of serious damages per kilometer, providing decision support for the subsequent probability density evolution algorithm.

[0059] Modeling of rail state evolution model:

[0060] Based on the collected rail damage data, the probability density function of the rail state is constructed using multidimensional kernel density estimation (KDE). The joint probability density function is defined as:

[0061]

[0062] Where:

[0063] is a multidimensional Gaussian kernel function: ,in ;

[0064] is the bandwidth matrix, which controls the degree of smoothing in each dimension;

[0065] is the determinant of the bandwidth matrix;

[0066] is the sample size;

[0067] is the i-th sample vector.

[0068] By analyzing the changing trend of the probability density function, a rail state evolution model is established to describe the transition process of the rail from the normal state to the damaged state.

[0069] In this embodiment, in step S3, the step of predicting the rail overhaul section based on the constructed rail state evolution model includes:

[0070] Based on the constructed rail state evolution model, the probability distribution of rail damage within a certain time period in the future is predicted; based on the prediction results, high-risk areas where the damage probability exceeds the set threshold (such as 75%) are located, and combined with actual operation and maintenance needs, the rail sections that need major repairs or replacement are determined.

[0071] In this embodiment, in step S4, the step of outputting and visualizing the results includes:

[0072] Provides a detailed prediction report, including prediction results: rail sections are divided into high, medium, and low risk levels, and the rail sections that need major repairs and their priorities are listed; damage probability distribution: displays the probability density distribution of rail damage; confidence interval: provides the confidence interval of the prediction results to reflect the uncertainty of the prediction;

[0073] Provide visual data reports to show the risk level or priority of different sections, or the changing trend of rail damage probability over time, providing a scientific basis for operation and maintenance decisions.

[0074] It should be noted that, in this document, the term "comprises" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or apparatus that includes a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0075] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A rail overhaul section prediction method based on probability density evolution algorithm, characterized in that: The method comprises: S1. Collect rail damage data through on-site inspections, including the total weight of rails, the number of damaged rails per kilometer, the year the rails were put into operation, the maximum allowable speed of the line, the proportion of curves with a radius of less than 800 meters, and the total weight of rails per year. This data is then stored in a relational database and cleaned and standardized. S2. Analyze the data, including correlation analysis and rail state evolution modeling; S3. Predict rail overhaul areas based on the constructed rail state evolution model; S4. Result output and visualization.

2. The method for predicting rail overhaul sections based on a probability density evolution algorithm according to claim 1, characterized in that: In step S1, rail damage data is collected through on-site inspection and stored in a relational database. The steps of cleaning and standardizing the data include: Data Collection: Rail damage data is collected through on-site inspection; Data Storage: The collected data is stored in an Oracle database; multiple data tables are created to store rail operation and maintenance data of different dimensions; key and index technologies are used to establish associations between different data tables to optimize data query efficiency and ensure that data of different dimensions accurately corresponds to the same rail; Data processing: The data was cleaned and standardized. Outliers were identified and eliminated using the quartile method. Duplicate uploaded data was queried and unique values ​​were retained. For records with missing items, if the total weight of the rails or the number of damaged items per kilometer was missing, the corresponding records were deleted. If other items were missing, the previous and next rails were matched using mileage information, and missing values ​​were filled using linear interpolation. The Min-Max standardization method was used to convert the data to a uniform scale.

3. The method for predicting rail overhaul sections based on a probability density evolution algorithm according to claim 1, characterized in that: In step S2, the data is analyzed, including correlation analysis and rail state evolution modeling, including: Correlation analysis: Based on the random forest model, the correlation between different rail damage data and overhaul and replacement is explored, and the cumulative total weight of rails and the number of serious damages per kilometer are given higher weight coefficients; Modeling of rail state evolution model: Based on the collected rail damage data, the probability density function of the rail state is constructed using multidimensional kernel density estimation. The joint probability density function is defined as: Where: is a multidimensional Gaussian kernel function: ,in ; is the bandwidth matrix, which controls the degree of smoothing in each dimension; is the determinant of the bandwidth matrix; is the sample size; is the i-th sample vector. By analyzing the changing trend of the probability density function, a rail state evolution model is established to describe the transition process of the rail from the normal state to the damaged state.

4. The method for predicting rail overhaul sections based on a probability density evolution algorithm according to claim 1, characterized in that: In step S3, the steps of predicting rail overhaul sections based on the constructed rail state evolution model include: Based on the constructed rail state evolution model, the probability distribution of rail damage within a certain time period in the future is predicted; based on the prediction results, high-risk areas where the damage probability exceeds the set threshold are located, and combined with actual operation and maintenance needs, the rail sections that need major repairs or replacement are determined.

5. The method for predicting rail overhaul sections based on probability density evolution algorithm according to claim 1, characterized in that: In step S4, the steps of outputting and visualizing the results include: The provided prediction report includes prediction results: rail sections are divided into high, medium and low risk levels, and the rail sections that need major repairs and their priorities are listed; damage probability distribution: showing the probability density distribution of rail damage; confidence interval: providing the confidence interval of the prediction results to reflect the uncertainty of the prediction; Provide visual data reports to show the risk level or priority of different sections, or the changing trend of rail damage probability over time, providing a scientific basis for operation and maintenance decisions.

Images