Pantograph carbon contact strip whole life cycle data management system

Abstract

The invention belongs to the technical field of life state monitoring and maintenance management of a pantograph carbon contact strip, and particularly discloses a pantograph carbon contact strip full-life cycle data management system, which comprises a multi-level gradient feature library, a pseudo edge filtering technology, multi-scale feature extraction and radial basis function classification, full-scale identification of defects such as cracks is realized, and the microcrack detection accuracy is improved; multi-source data collaborative acquisition and three-dimensional coordinate mapping are adopted, and self-adaptive sliding window analysis is combined, so that data integrity and timeliness under different working conditions are ensured; through impact energy parameter integration, pressure fluctuation index calculation and a thermal expansion correction coefficient, a partition life prediction model is established, and the wear prediction error of a key area is reduced; high-risk area automatic marking and maintenance work order generation are realized, the RFID tracing technology is combined, the maintenance efficiency is improved, and the service life is prolonged. According to the method, the defects of a traditional method in the aspects of defect detection, dynamic monitoring and service life prediction are effectively overcome.

Description

Technical Field

[0001] The present invention belongs to the technical field of life status monitoring and maintenance management of a pantograph carbon slide plate, and relates to a full life cycle data management system for a pantograph carbon slide plate. Background Art

[0002] As a key component of the electric traction system, the working condition of the pantograph carbon slide is directly related to the reliability and operational safety of the train power supply system. During long-term operation, the carbon slide is subjected to the coupling of multiple physical fields such as mechanical friction, arc erosion, and environmental corrosion, which can easily produce typical damage morphologies such as cracks and spalling. If not identified and treated in a timely manner, it may cause major safety hazards such as pantograph failure or even power outages. With the rapid development of intelligent sensing technology, machine vision, and big data analysis methods, the establishment of an integrated and intelligent carbon slide full-life cycle data management system to achieve accurate identification of damage characteristics, quantitative assessment of operating status, and predictive maintenance decision-making has become an important technical development direction in the field of intelligent operation and maintenance of rail transit.

[0003] Existing technologies mainly adopt periodic offline detection or single parameter online monitoring mode, record defect characteristics through manual visual inspection, or rely on basic image processing algorithms to realize surface damage identification and formulate maintenance strategies based on empirical models.

[0004] Although the above method can realize the basic monitoring function of the carbon slide plate, it still has the following shortcomings in actual application: 1. In terms of defect detection, the existing analysis method is difficult to effectively balance the recognition accuracy of macro damage and micro defects, especially the lack of characterization ability of tiny features, which affects the early warning effect of damage. 2. The real-time monitoring capability in a dynamic operating environment is limited. The traditional method is difficult to adapt to the instantaneous changes in the train operating conditions, resulting in reduced integrity of key data collection and timeliness of analysis. 3. The existing life prediction is not accurate enough in its portrayal of the coupling of complex physical fields, especially the failure to fully consider the interaction between temperature field and stress field, which makes the wear prediction reliability of key areas insufficient. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purpose, the present invention proposes the following technical solutions: a pantograph carbon skateboard full life cycle data management system, comprising the following contents: a defect visual detection module, which collects the pantograph carbon skateboard image and filters the pseudo-edge information, extracts the corresponding multi-scale feature information of the filtered pantograph carbon skateboard image, and identifies surface defect data based on this, including defect location, defect type and defect characteristics.

[0006] The data collaborative acquisition module collects multi-source data of the pantograph carbon slide in real time during operation, including thickness distribution, surface temperature and contact distance with the contact network.

[0007] The data fusion calibration module establishes the three-dimensional coordinate system of the pantograph carbon slide and maps the multi-source data and surface defect data to the preset coordinate grid units.

[0008] The damage modeling and marking module calculates the impact energy parameters and contact pressure fluctuation index of each grid unit to form a fused data set, and divides the pantograph carbon slide into the core area, transition area and edge area according to a preset proportional structure. Combined with the operating conditions and surface defect data of the pantograph carbon slide, the life attenuation value of each grid unit is calculated and high-risk areas are marked accordingly.

[0009] The maintenance decision module generates maintenance work orders based on the surface defect data of the high-risk areas of the pantograph carbon slide, fills, repairs and solidifies the surface defects, and realizes full life cycle traceability based on re-operation data.

[0010] Compared with the existing technology, the beneficial effects of the present invention are as follows: (1) The present invention constructs a multi-level gradient feature library and filters the pseudo-edge information of the carbon skateboard image. It combines multi-scale feature extraction with radial basis function classification to achieve full-scale accurate identification of defects such as cracks and spalling from macro damage to micro defects, which helps to improve the accuracy of micro crack detection and thus improve the reliability of early damage warning.

[0011] (2) The present invention collects multi-source data from the pantograph carbon slide in real time during operation, and combines it with three-dimensional coordinate system mapping to achieve accurate correlation and spatial positioning of defect data, thus solving the one-sidedness problem of traditional single data source. At the same time, a sliding analysis window of variable length is set up. By monitoring the changes in train speed in real time, the spatiotemporal parameters of the analysis window are dynamically adjusted to ensure the integrity of data collection and the timeliness of analysis under different working conditions. This helps to solve the detection blind spot problem caused by the traditional fixed threshold method.

[0012] (3) When predicting life, the present invention integrates impact energy parameters, calculates pressure fluctuation index and predicts partitioned life, while introducing thermal expansion mass correction coefficient to accurately quantify the effect of temperature gradient on material anisotropy, effectively reducing wear prediction error in key areas, thereby significantly improving the accuracy of life assessment in key areas.

[0013] (4) The present invention realizes the active marking of high-risk areas and the automatic generation of maintenance work orders. Combined with RFID full life cycle traceability, it helps to improve maintenance efficiency and thus extend the service life of the carbon slide. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 This is a schematic diagram of the system module connection of the present invention. DETAILED DESCRIPTION

[0016] 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.

[0017] See also Figure 1 As shown, the pantograph carbon slide full life cycle data management system proposed in the present invention includes: a defect visual detection module, a data collaborative acquisition module, a data fusion calibration module, a damage modeling and marking module, and a maintenance decision module.

[0018] The data fusion calibration module is respectively connected to the defect visual detection module, the data collaborative acquisition module, and the damage modeling and marking module; the damage modeling and marking module is connected to the maintenance decision module; and the maintenance decision module is connected to the defect visual detection module.

[0019] The defect visual detection module collects the pantograph carbon slide image and filters the pseudo-edge information, extracts the corresponding multi-scale feature information of the filtered pantograph carbon slide image, and identifies the surface defect data, including the defect location, defect type and defect characteristics.

[0020] In a preferred embodiment, the process of collecting the pantograph carbon slide image and filtering the pseudo-edge information includes: establishing a three-level feature library for defect identification, including a first-level feature of the aspect ratio of the defect area, a second-level feature of the square ratio of the perimeter of the defect area, and a third-level feature of the color histogram distribution of the defect area.

[0021] For example, the construction of a three-level physical feature library is specifically as follows: first-level geometric features: the aspect ratio of the defect area is greater than 3:1 or less than 1:2; second-level morphological features: the square ratio of the perimeter of the defect area is greater than 8.5 or less than 4.0; third-level texture features: the contrast of the grayscale co-occurrence matrix of the defect area is greater than 0.25.

[0022] When a defective area that meets the three-level feature library for defect identification is detected in the pantograph carbon slide image, the video stream containing several seconds before and after the defective area is automatically intercepted.

[0023] The same features are identified on the pantograph carbon slide plate images corresponding to the intercepted video stream. If there is a defect area with the same features in each frame image in the intercepted video stream, the defect area is located.

[0024] The adaptive Canny operator is used to extract the initial edge of the defect area in the pantograph carbon slide image, and the defect area smaller than 100 mm is filtered out through morphological opening and closing operations. Pixel pseudo-edge, To set the constant, the Zernike moment is used to correct the pseudo-edge pixels to locate the real edge. The defect areas include crack areas, wear areas, arc burn areas, and peeling areas.

[0025] Specifically, the adaptive Canny operator dynamically adjusts the threshold according to the local gradient of the image to avoid edge breakage or noise false detection caused by the global threshold; the morphological opening and closing operation filters out areas smaller than Isolated noise of pixels, such as , retaining the real defect edge; Zernike moment correction is to use orthogonal polynomials to fit the edge pseudo-edge pixel position to improve positioning accuracy.

[0026] In a further preferred embodiment, the corresponding multi-scale feature information of the filtered pantograph carbon slide image is extracted and surface defect data is identified based on it, which includes: constructing a Gaussian pyramid to achieve 4-level scale scaling, with each level interval of 1.5 times, and using the Sobel operator to extract horizontal / vertical gradient features, and constructing an 8-directional gradient histogram to capture the directional information of edges and textures.

[0027] Specifically, a four-layer image pyramid is constructed through Gaussian filtering and downsampling, and the resolution of each layer is gradually reduced to 1 / 1.5 times that of the previous layer (e.g., the first layer is 1000×1000 pixels → the second layer is 667×667 pixels → ...).

[0028] The specific operations are as follows: Level 1: original image, scaled to 1.0.

[0029] Level 2: Gaussian blur followed by downsampling, with a scale of 1.5.

[0030] Level 3: Repeat the steps in Level 2, but at a scale of 2.25 times.

[0031] Level 4: Repeat the operation based on Level 3, but the scale is 3.375 times.

[0032] Reducing the amount of subsequent gradient calculations through high-level pyramids helps solve the problem of detecting defects of different sizes. For example, the small scale (4th layer) can detect microcracks with a diameter of 1mm, and the large scale (1st layer) can identify spalling with an area of 50mm², thereby accelerating feature extraction.

[0033] Each pixel in the filtered pantograph carbon slide image is taken as the central pixel, and the grayscale value of each central pixel is compared with the surrounding 8 neighboring pixels. The neighboring pixels with a value greater than the central pixel are marked as 1, otherwise they are marked as 0, generating an 8-bit binary code, such as 11001011.

[0034] Count the number of binary code jumps, that is, the number of 0 / 1 changes. If the number of binary code jumps is less than or equal to 2, it is classified as a uniform pattern, which helps to reduce the feature dimension. Then, the frequency distribution of all uniform patterns in the image is counted to form a texture feature vector.

[0035] The texture feature vector after dimensionality reduction is generated by fusing the 8-directional gradient histogram of the 4-level pyramid with the texture feature.

[0036] The radial basis function is selected as the nonlinear mapping kernel function to output the defect type and confidence level, and then the defect data of the defect area is marked.

[0037] Specifically, the mathematical expression of the nonlinear mapping kernel function is: , where Represent the feature vector of the defect area to be classified and the feature vector of the defect sample of known category, is the kernel parameter, is the kernel width control parameter, which is used to adjust the complexity of the classification boundary and ensure the separability of samples in the high-dimensional feature space. is an exponential function that maps the Euclidean distance to a similarity score in the [0,1] interval. is a vector and The square of the Euclidean distance between the two samples represents the similarity measure between the two samples in the feature space. hour, , indicating complete similarity, when When it increases, Close to 0, indicating complete dissimilarity.

[0038] The defect type is one of crack, spalling, arc burn, and wear, and the confidence level is normalized to a real number in the interval [0, 1], indicating the credibility of the classification result.

[0039] The penalty coefficient C is set to 1.0 to balance the maximization of the classification interval and the minimization of the training error. A confidence level greater than 0.9 is considered a strong defect, such as a crack longer than 10 mm. A confidence level of 0.7 ≤ C ≤ 0.9 is considered a weak defect, such as slight wear.

[0040] The present invention constructs a multi-level gradient feature library and filters the pseudo-edge information of the carbon skateboard image. It combines multi-scale feature extraction with radial basis function classification to achieve full-scale accurate identification of defects such as cracks and spalling, from macro damage to micro defects, which helps to improve the accuracy of microcrack detection and thus improve the reliability of early damage warning.

[0041] The data collaborative acquisition module collects multi-source data of the pantograph carbon slide plate in real time during operation, including thickness distribution, surface temperature and contact distance with the contact network.

[0042] Specifically, the thickness distribution data of the pantograph carbon slide plate is collected by a laser displacement sensor, the surface temperature data is collected by an infrared thermal imager, and the contact distance is collected by a lidar.

[0043] The data fusion calibration module establishes a three-dimensional coordinate system for the pantograph carbon slide plate and maps multi-source data and surface defect data to preset coordinate grid units.

[0044] In a preferred embodiment, the three-dimensional coordinate system of the pantograph carbon slide is established, and the multi-source data and surface defect data are mapped to the preset coordinate grid unit, including: establishing a three-dimensional coordinate system with the length direction of the pantograph carbon slide working surface as the X-axis, the width direction as the Y-axis, and the thickness direction as the Z-axis, dividing the grid units according to the set spacing along the width direction, and dividing the Z-axis scale according to the height change of the contact network wire along the thickness direction, and each scale corresponds to a specific height value of the contact network wire.

[0045] Multi-source data are mapped to coordinate grid cells, where thickness distribution data and contact spacing with the contact network are directly associated with the Z-axis coordinate, and surface temperature data are corrected to the grid center point according to the heat conduction model.

[0046] Specifically, the surface temperature data is corrected to the grid center point according to the heat conduction model, which refers to the original temperature data that may be distributed on the grid boundary, node or volume pixel surface, and is recalculated and mapped to the grid center point through the physical laws of heat conduction to ensure the physical consistency and calculation accuracy of the temperature field.

[0047] Based on the defect positions of different defect regions, the defect features of each surface defect are mapped to corresponding grid cells, where the defect features refer to one of the length, width or area of the defect region.

[0048] The damage modeling and marking module calculates the impact energy parameters and contact pressure fluctuation index of each grid cell to form a fused data set, and divides the pantograph carbon slide into the core area, transition area and edge area according to a preset proportional structure. Combined with the operating conditions and surface defect data of the pantograph carbon slide, the life attenuation value of each grid cell is calculated and high-risk areas are marked accordingly.

[0049] For example, the core area is defined as the first 20% of the pantograph carbon slide along the length direction. This area directly bears the main contact force with the contact network, has the largest force, and may wear faster; the transition zone is defined as the middle 60% of the area, which has moderate force and is between the core area and the edge area; the edge area is defined as the last 20% of the area, which is far away from the main contact point, has less force, and has a relatively low wear rate.

[0050] In a preferred embodiment, the impact energy parameters and contact pressure fluctuation index of each grid unit are calculated to form a fused data set, including: obtaining the three-dimensional acceleration of the pantograph carbon slide during operation in real time through sensors arranged on the surface of the pantograph carbon slide, vector synthesizing the three-dimensional acceleration components, and obtaining a synthetic acceleration amplitude reflecting the actual impact intensity.

[0051] The synthetic acceleration of the pantograph carbon slide at each time point within a set detection time period is obtained, and the synthetic acceleration is numerically integrated along the time axis of the operation process. A thermal expansion mass correction coefficient is introduced in the integration process. The thermal expansion mass correction coefficient is adaptively adjusted according to the surface temperature data of each grid cell of the pantograph carbon slide to compensate for the influence of thermal expansion on mass distribution.

[0052] Specifically, due to thermal expansion, the mass of the pantograph carbon slide material will change, and the thermal expansion mass correction coefficient To compensate for this effect, the calculation formula is: ,in The thermal expansion coefficient preset for the pantograph carbon slide material, Refers to the fact that the volume expansion changes in three directions all contribute to thermal expansion in three-dimensional space. Initial monitoring temperature of the pantograph carbon slide plate, During operation The surface temperature at a time point, To set the number of the running time point within the detection time period, .

[0053] Based on the thickness data of each grid unit of the pantograph carbon slide, the volume of each grid unit is obtained, and combined with the material density of the pantograph carbon slide, the initial mass of the pantograph carbon slide grid unit is obtained. .

[0054] The corresponding time interval of the mark setting detection time period is , for the synthetic acceleration Multiply by the thermal expansion mass correction factor Then, the impact energy parameter of the pantograph carbon slide grid unit in the time interval is obtained by integrating on the time axis. , the calculation formula is: , The first time that the pantograph carbon slide plate is detected within the set detection time period The resultant acceleration of the run at each time point.

[0055] Based on the above impact energy parameter calculation steps, the impact energy parameters of each grid unit of the pantograph carbon slide in the time interval are obtained. , is the number of each grid cell, .

[0056] Get the volume of each grid cell as , and then convert the result of the synthetic acceleration integration into the volume energy density value of each grid unit , represented by the impact energy parameter of each grid unit.

[0057] The contact distance between each grid unit and the contact network on the time axis is obtained to form a continuous contact distance change curve.

[0058] A sliding analysis window of variable length is set, and the corresponding standard deviation of the continuous contact spacing is calculated in each analysis window. The standard deviation reflects the discrete degree of friction contact fluctuation in the window.

[0059] Specifically, the window duration is dynamically adjusted according to the real-time running speed of the train to ensure that each analysis window corresponds to a fixed spatial displacement.

[0060] The contact pressure fluctuation index of each grid cell is obtained by weighted averaging the corresponding standard deviation values of the contact spacing calculated in multiple consecutive analysis windows. .

[0061] The impact energy parameter and contact pressure fluctuation index of each grid unit are weighted and solved to form a fusion data set consisting of the stress coefficient of each grid unit. The calculation formula of the stress coefficient of each grid unit is: ,in Indicates the preset reference volume energy density value, Indicates the preset reference pressure fluctuation index, 、 They respectively refer to the corresponding preset weight ratios of the impact energy parameter and the contact pressure fluctuation index. The preset reference volume energy density value and the reference pressure fluctuation index are set according to the material properties of the pantograph carbon slide plate and the actual operating conditions.

[0062] Specifically, the impact energy parameters are transmitted to the plate surface through stress waves, exacerbating mechanical wear. When the impact energy parameters exceed the fatigue limit of the pantograph carbon plate material, microcracks initiate and propagate. High fluctuation indexes cause the wear rate in concentrated pressure areas to be much higher than in low-pressure areas. For example, the wear rate in the center of the plate is much higher than that at the edges, forming a "saddle-shaped" wear pattern.

[0063] This method collects multi-source data from the pantograph's carbon slide in real time, combining it with three-dimensional coordinate mapping to accurately correlate and spatially locate defect data, addressing the one-sided nature of traditional single data sources. Furthermore, a variable-length sliding analysis window is established. By monitoring train speed changes in real time, the temporal and spatial parameters of the analysis window are dynamically adjusted to ensure data collection integrity and analysis timeliness under varying operating conditions. This helps address the detection blind spots associated with traditional fixed threshold methods.

[0064] In a further preferred embodiment, the life attenuation value of each grid unit is calculated and high-risk areas are marked accordingly, which includes: extracting historical operating condition data of the pantograph carbon slide plate-affiliated train on the current operating line and corresponding historical life loss characteristics of each defect type in each area, the operating condition data includes operating mileage and line conditions, and the life loss characteristics refer to one of the increased wear area or crack extension length or extension width.

[0065] Specifically, different operating lines, such as urban rail transit, high-speed rail, and mountain railways, have different line conditions, including curve radius, slope, and tunnel length. These conditions affect the stress and wear rate of the pantograph carbon plate. For example, on lines with smaller curve radii, the pantograph carbon plate may be subjected to greater lateral forces; on lines with larger slopes, the pantograph carbon plate may be subjected to greater vertical forces.

[0066] Life decay models are established for the core, transition, and edge regions, each containing historical operating condition data and the corresponding historical life loss characteristics of each defect type in each region. Machine training of the models derives model relationship parameters for each defect type in each region. These model relationship parameters represent the relationship constants between the operating condition data and the corresponding life loss characteristics of each defect type in each region. For example, for the core region, due to its high stress and rapid wear rate, the relationship between wear and mileage is established as: Wear = A x Mileage + B, where A and B are the model relationship parameters.

[0067] The operating condition data of the pantograph carbon slide plate-affiliated train that has been running on the line within the set inspection time period is obtained, and the trained life decay model is imported to obtain the corresponding life loss characteristics of each defect type in each area.

[0068] The corresponding defect features of each grid unit are extracted, and the corresponding life loss features of the defect type matching the corresponding area of the grid unit are accumulated and normalized to obtain the life attenuation value of each grid unit. Then, the grid units with life attenuation values higher than the preset life attenuation baseline value of the area are screened as high-risk areas.

[0069] When predicting life, the present invention integrates impact energy parameters, calculates pressure fluctuation index and predicts partitioned life, while introducing thermal expansion mass correction coefficient to accurately quantify the influence of temperature gradient on material anisotropy, effectively reducing wear prediction error in key areas, thereby greatly improving the accuracy of life assessment in key areas.

[0070] The maintenance decision module generates a maintenance work order based on the surface defect data of the high-risk area of the pantograph carbon slide, fills, repairs and solidifies the surface defects, and realizes full life cycle traceability based on the re-operation data to continuously optimize the management process.

[0071] In a preferred embodiment, a maintenance work order is generated based on the surface defect data of the high-risk area of the pantograph carbon slide, and the surface defects are filled, repaired and solidified. The content includes: determining the attenuation level based on the life attenuation value of the high-risk area, and obtaining the corresponding defect type of the high-risk area, and establishing a pantograph carbon slide maintenance work order based on this, which includes maintenance content for different defect types at different attenuation levels, and different attenuation levels correspond to different life attenuation value ranges.

[0072] For example, when the defect depth exceeds 2mm or the area exceeds 50mm², a first-level work order is automatically generated, and the work order content requires processing within 24 hours; when the same type of defect occurs three times in the same area, a second-level work order is generated, and the work order content requires the initiation of special quality traceability.

[0073] Under the pantograph carbon slide maintenance work order, we obtain filling and repair materials appropriate for the defect type in high-risk areas, perform repairs, and perform a curing process to ensure repair quality. The repair process is optimized based on the defect type and location to restore the normal operation of the pantograph carbon slide. For example, the filling and repair material is made of conductive epoxy resin, and the curing process parameters are set as follows: preheating at 60°C for 5 minutes, pressurized curing at 120°C for 30 minutes, and the pressure is controlled at 0.3±0.05MPa.

[0074] In a further preferred embodiment, the full life cycle traceability based on re-operation data includes: continuously monitoring the defect data of high-risk areas through a defect visual detection module during several subsequent consecutive detection time periods.

[0075] When the defect data monitored in the high-risk area reaches a preset ratio threshold of the defect data before the defect is repaired, the early warning mechanism is activated, wherein the preset ratio threshold is set based on the defect recurrence risk assessment and is used to characterize the degree of danger of the defect recurrence.

[0076] Maintenance records automatically generate electronic tags containing maintenance data, including but not limited to operator, process parameters, and material batch numbers. RFID chips are implanted in the pantograph carbon skateboard body to achieve full life cycle traceability, providing data support for subsequent fault analysis and preventive maintenance.

[0077] In a further preferred embodiment, the full life cycle traceability content also includes: building a first-level cache to store recently acquired high-frequency sampling data, including three-dimensional acceleration, surface temperature data, contact spacing and other real-time stream processing cache data.

[0078] A secondary cache is constructed to store key parameters within a certain time span, including cache data for analyzing working condition characteristics such as maximum wear depth and contact pressure fluctuation index trend.

[0079] A three-level cache is built to store full life cycle data, including original sampling data stored by all sensors segmented by mileage, RFID tag data for each maintenance, historical three-dimensional shape scanning data, and other knowledge graph data.

[0080] Different cache access rights are set according to the data access user role, where field operators are authorized to access the first-level cache, maintenance engineers are authorized to access the second-level cache, and the quality management department is authorized to access the third-level cache.

[0081] The present invention realizes the active marking of high-risk areas and the automatic generation of maintenance work orders. Combined with RFID full life cycle traceability, it helps to improve maintenance efficiency and thus extend the service life of the carbon slide.

[0082] It should be noted that the formulas described above, through the principle of dimensional consistency and mathematical standardization (e.g., normalization, dimensionless parameter conversion, or unified unit system), can translate physical quantities of different attributes into unitless standard values or homogeneous, superimposable parameters. This eliminates the interference of different dimensions on operational logic, ensuring that the formulas retain the distribution characteristics of the original data while maintaining mathematical rationality and adaptability to objective laws. These are merely exemplary embodiments of the present invention and are not intended to limit the scope of the invention.

Claims

1. The pantograph carbon slide full life cycle data management system is characterized by: include: The defect visual detection module collects the pantograph carbon slide image and filters the false edge information, extracts the corresponding multi-scale feature information of the filtered pantograph carbon slide image, and identifies the surface defect data, including the defect location, defect type and defect characteristics; The data collaborative acquisition module collects multi-source data of the pantograph carbon slide in real time during operation, including thickness distribution, surface temperature, and contact distance with the contact network; The data fusion calibration module establishes the three-dimensional coordinate system of the pantograph carbon slide and maps the multi-source data and surface defect data to the preset coordinate grid units; The damage modeling and marking module calculates the impact energy parameters and contact pressure fluctuation index of each grid cell to form a fused data set. It then divides the pantograph carbon slide into core, transition, and edge areas according to a preset proportional structure. Combining the pantograph carbon slide's operating conditions and surface defect data, it calculates the life attenuation value of each grid cell and marks high-risk areas accordingly. The maintenance decision module generates maintenance work orders based on the surface defect data of the high-risk areas of the pantograph carbon slide, fills, repairs and solidifies the surface defects, and realizes full life cycle traceability based on re-operation data.

2. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The collecting of the pantograph carbon slide plate image and filtering of the pseudo-edge information includes: Establish a three-level feature library for defect identification, including the first-level feature of the defect area aspect ratio, the second-level feature of the defect area perimeter square ratio, and the third-level feature of the defect area color histogram distribution; When a defective area that meets the three-level feature library for defect identification is detected in the pantograph carbon slide image, the video stream containing several seconds before and after the defective area is automatically intercepted; Perform identical feature recognition on the pantograph carbon slide plate image corresponding to the intercepted video stream. If a defective area with the same features exists in each frame image of the intercepted video stream, locate the defective area. The adaptive Canny operator is used to extract the initial edge of the defect area in the pantograph carbon slide image, and the defect area smaller than 100 mm is filtered out through morphological opening and closing operations. Pixel pseudo-edge, To set the constant and locate the real edge.

3. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The multi-scale feature information corresponding to the filtered pantograph carbon slide plate image is extracted and surface defect data is identified based on the information, including: Construct a Gaussian pyramid to achieve 4-level scale scaling, extract horizontal / vertical gradient features, and construct an 8-directional gradient histogram; Take each pixel in the filtered pantograph carbon slide image as the center pixel, compare the grayscale value of each center pixel with the surrounding 8 neighboring pixels, mark the neighboring pixels with a grayscale value greater than the center pixel as 1, otherwise as 0, and generate an 8-bit binary code; Count the number of binary code transitions. If the number of binary code transitions is less than or equal to 2, it is classified as a uniform pattern. Then, the frequency distribution of all uniform patterns in the image is counted to form a texture feature vector. The texture feature vector after dimensionality reduction is generated by fusing the 8-directional gradient histogram of the 4-level pyramid with the texture feature; The radial basis function is selected as the nonlinear mapping kernel function to output the defect type and confidence level, and then the defect data of the defect area is marked; The defect type is one of crack, spalling, arc burn, and wear, and the confidence level is normalized to a real number in the interval [0, 1].

4. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The method of establishing a three-dimensional coordinate system for the pantograph carbon slide plate and mapping the multi-source data and surface defect data to a preset coordinate grid unit includes: establishing a three-dimensional coordinate system with the length direction of the pantograph carbon slide plate working surface as the X-axis, the width direction as the Y-axis, and the thickness direction as the Z-axis, dividing the grid units along the width direction according to the set spacing, and dividing the Z-axis scale along the thickness direction according to the height change of the contact wire; Multi-source data are mapped to coordinate grid cells, where thickness distribution data and contact spacing with the contact network are directly associated with the Z-axis coordinates, and surface temperature data are corrected to the grid center point according to the heat conduction model; Based on the defect locations in different defect areas, the defect features of each surface defect are mapped to the corresponding grid cells.

5. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The calculation of the impact energy parameters and contact pressure fluctuation index of each grid unit to form a fused data set includes: obtaining the three-dimensional acceleration of the pantograph carbon slide in real time during operation through sensors arranged on the surface of the pantograph carbon slide, performing vector synthesis on the three-dimensional acceleration components, and obtaining a synthetic acceleration amplitude reflecting the actual impact intensity; Obtain the synthetic acceleration of the pantograph carbon slide at each time point within a set detection time period, numerically integrate the synthetic acceleration along the time axis of the operation process, introduce a thermal expansion mass correction coefficient during the integration process, and adaptively adjust the thermal expansion mass correction coefficient based on the surface temperature data of each grid cell of the pantograph carbon slide; Obtain the volume of each grid cell, and then convert the result of the synthetic acceleration integration into the volume energy density value of each grid cell, which is represented as the impact energy parameter of each grid cell; Obtain the contact distance between each grid unit and the contact network on the time axis to form a continuous contact distance change curve; Set a sliding analysis window of variable length and calculate the corresponding standard deviation of continuous contact spacing in each analysis window; The contact pressure fluctuation index of each grid cell is obtained by taking the weighted average of the corresponding standard deviation values of the contact spacing calculated in multiple consecutive analysis windows. The impact energy parameters and contact pressure fluctuation index of each grid cell are weighted and solved to form a fused data set consisting of the stress coefficients of each grid cell.

6. The pantograph carbon slide full life cycle data management system according to claim 5, characterized in that: The calculation of the life attenuation value of each grid unit and the marking of high-risk areas based on the value include: Extract the historical operating condition data of the pantograph carbon slide train on the current operating line and the corresponding historical life loss characteristics of each defect type in each area; Establish life decay models for the core, transition, and edge regions, including the relationship between historical operating condition data and the corresponding historical life loss characteristics of each defect type in each region. Machine training of the models yields corresponding model relationship parameters for each defect type in each region. The model relationship parameters represent the relationship constants between the operating condition data and the corresponding life loss characteristics of each defect type in each region. Obtain the operating condition data of the pantograph carbon slide plate's affiliated train that has been running on the line within the set inspection time period, import the trained life decay model, and derive the corresponding life loss characteristics of each defect type in each area; The corresponding defect features of each grid unit are extracted, and the corresponding life loss features of the defect type matching the corresponding area of the grid unit are accumulated and normalized to obtain the life attenuation value of each grid unit. Then, the grid units with life attenuation values higher than the preset life attenuation baseline value of the area are screened as high-risk areas.

7. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The maintenance work order is generated based on the surface defect data of the high-risk area of the pantograph carbon slide plate, and the surface defects are filled, repaired and cured, including: Determine the attenuation level of high-risk areas based on their life attenuation values, and obtain the corresponding defect types in high-risk areas. Based on this, establish a maintenance work order for the pantograph carbon slide, including maintenance details for different defect types at different attenuation levels. Under the pantograph carbon slide maintenance work order, obtain filling repair materials suitable for the corresponding defect type in the high-risk area for repair, and ensure the quality of the repair through curing treatment.

8. The pantograph carbon slide full life cycle data management system according to claim 1, characterized in that: The full life cycle traceability based on re-operation data includes: During the subsequent consecutive inspection periods, the defect data of high-risk areas are continuously monitored through the defect visual inspection module; When the defect data in the high-risk area reaches the preset ratio threshold of the defect data before the defect is repaired, the early warning mechanism is activated; Maintenance records automatically generate electronic tags containing maintenance data, and full life cycle traceability is achieved by implanting RFID chips into the pantograph carbon skateboard body.

9. The pantograph carbon slide full life cycle data management system according to claim 8, characterized in that: The full life cycle traceability also includes: Build a first-level cache to store recently acquired high-frequency sampling data; Build a secondary cache to store key parameters within a certain time span; Build a three-level cache to store data throughout its life cycle; Set different cache access permissions based on data access user roles.

Images