A grinding system for contact wires to prevent electrical corrosion

Abstract

The present invention relates to the field of grinding technology, and specifically discloses a grinding system for anti-electric corrosion contact lines, which is used to solve the problem of the existing lack of real-time abnormality detection of contact lines after grinding, resulting in safety hazards. The present invention obtains contact line grinding parameters through laser scanning, ultrasonic detection and sensors, and obtains a standard contact line grinding image as a first grinding image and a real-time contact line grinding image as a second grinding image through a high-pixel camera. First surface feature data is extracted according to the contact line surface data for data processing, and first grinding image features are obtained based on the first grinding image, and second grinding image features are obtained based on the second grinding image. A layered abnormal response model is constructed according to the contact line grinding parameters, the first grinding image features and the second grinding image features to obtain abnormal grinding parameters, and grinding abnormality judgment is performed based on the abnormal grinding parameters to ensure the contact line grinding quality and prevent the occurrence of electric corrosion safety hazards.

Description

Technical Field

[0001] The present invention relates to the technical field of grinding, and more particularly to a grinding system for an anti-electrocorrosion contact wire. Background Art

[0002] Anti-corrosion contact wires are mainly used to ensure a stable electrical connection between the pantograph and the contact wire, while reducing arc erosion and electrochemical corrosion. However, with the rapid development of high-speed railways and urban rail transit, the long-term use of contact wires has led to problems such as surface wear, localized corrosion, metal fatigue, and arc erosion, which have reduced contact performance and affected power supply reliability. Existing literature: In 2009, Wu Jiqin published a doctoral thesis on China National Knowledge Infrastructure, titled "Research on Electrical Contact Characteristics of Pantograph-Catechnical Network System". It mentioned that the contact area when the pantograph slide and the contact line slide relative to each other is composed of a number of scattered tiny contact points. These contact points not only support the load, but also bear the friction work and the heat flow caused by the current flowing through the contact resistance. The sliding contact wear of the existing pantograph-catenary system can be roughly divided into three types: mechanical wear, chemical wear and electrical loss. Mechanical wear is usually divided into adhesive wear, hard particle wear and fatigue wear. Among them, hard particle wear is that during the sliding process, external pollutants (such as dust, metal oxide particles) or detached chips may be embedded in the contact surface, forming a similar "abrasive particle" effect, causing serious scratches on the contact line and the slide, so it is necessary to polish the anti-corrosion contact line in time.

[0003] The invention patent application with publication number CN114211339A discloses a rigid contact line grinding and polishing device for automatically grinding and polishing the contact line, including a frame, two pairs of guide wheels are provided on both sides of the upper end of the frame, and a positioning wheel is provided under each pair of guide wheels, and the wheel surfaces of the guide wheels and the positioning wheels form a clamping part for clamping the bus; at least one grinding wheel is provided at the lower end of the frame, and the grinding wheel is parallel to the rotating axis of the guide wheel, and the grinding wheel is connected to the output shaft of the motor. The grinding wheel grinds the contact line at the lower end of the bus, and the frame is driven to move on the bus by the rotation of the grinding wheel, but it cannot control the grinding force. Excessive grinding causes the thickness of the contact line to decrease, and insufficient grinding cannot effectively repair defects. There is a lack of real-time abnormality detection of the contact line after grinding, resulting in safety hazards. In order to solve the above problems, a technical solution is now provided. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a grinding system for anti-electric corrosion contact lines, which solves the problem of lack of real-time abnormality detection of the polished contact lines through layered abnormality detection, resulting in safety hazards, so as to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A grinding system for anti-electric corrosion contact lines includes a multi-sensor data acquisition module, a data feature extraction module, an intelligent grinding force optimization module and an abnormality detection and adjustment module. The multi-sensor data acquisition module is used to obtain contact line grinding parameters through laser scanning, ultrasonic detection and sensors, and obtain a standard contact line grinding image as a first grinding image and a real-time contact line grinding image as a second grinding image through a high-pixel camera; the data feature extraction module is used to extract first surface feature data based on contact line surface data for data processing, and obtain first grinding image features based on the first grinding image and second grinding image features based on the second grinding image; the intelligent grinding force optimization module is used to construct a hierarchical abnormal response model based on the contact line grinding parameters, the first grinding image features and the second grinding image features to obtain abnormal grinding parameters; the intelligent grinding force optimization module includes a first abnormal grinding analysis layer; the first abnormal grinding analysis layer is used to extract first surface feature data based on the first grinding image features and the second grinding image features And the second polished image features Import into the first level abnormal polishing impact model to obtain the first abnormal polishing parameters. The construction steps of the first level abnormal polishing impact model are as follows: And the second polished image features Get the polishing difference vector ,Based on the polishing difference vector, a first-level abnormal polishing impact model is constructed to calculate the Euclidean norm of the polishing vector, and then normalized to obtain the first abnormal polishing parameter.

[0007] As a further solution of the present invention, the data feature extraction module includes a data processing unit, an image processing unit and an image feature extraction unit; the data processing unit is used to normalize the contact line grinding parameters and perform time domain alignment processing; the image processing unit is used to perform denoising, grayscale and size adjustment processing on the first grinding image and the second grinding image; the image feature extraction unit is used to use deep learning to extract the first grinding image from the first grinding image. Get the first polishing image features , from the second polished image Get the second polishing image features ,in, is the feature extraction function.

[0008] As a further solution of the present invention, the intelligent grinding force optimization module also includes a second abnormal grinding analysis layer and a third abnormal grinding analysis layer; the second abnormal grinding analysis layer is used to establish a secondary abnormal grinding influence model through the first grinding parameters to obtain the second abnormal grinding parameters; the third abnormal grinding analysis layer is used to establish a tertiary abnormal grinding influence model through the second grinding parameters to obtain the third abnormal grinding parameters.

[0009] As a further solution of the present invention, the second abnormal grinding analysis layer is used to establish a secondary abnormal grinding influence model through the first grinding parameter to obtain the second abnormal grinding parameter. The steps of constructing the secondary abnormal grinding influence model are: extracting the first grinding parameter, obtaining the first grinding vector according to the first grinding parameter , get the preset first grinding standard value and get the second grinding vector , obtain the relative deviation values ​​based on the difference between the first grinding parameter and the first grinding standard value:

[0010] ;

[0011] ;

[0012] ;

[0013] Where: is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, is the surface roughness in the first grinding standard value, is the local concave depth in the first grinding standard value, is the remaining thickness in the first grinding standard value, is the surface roughness in the first grinding parameter, is the local concave depth in the first grinding parameter, is the remaining thickness in the first grinding parameter;

[0014] According to the relative deviation values, the second abnormal grinding parameters are obtained by importing them into the formula of the secondary abnormal grinding influence model. The formula of the secondary abnormal grinding influence model is:

[0015] ;

[0016] Where: is the second abnormal polishing parameter, is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, To increase the grinding strength, is the contact line diameter, is the weight coefficient of the relative deviation value of surface roughness, is the weight coefficient of the relative deviation value of the local depression depth, is the weight coefficient of the relative deviation value of the remaining thickness, is the sensitivity factor of the effect of grinding force on the depth of local depression, It is the sensitivity factor of the effect of grinding force on the remaining thickness.

[0017] As a further solution of the present invention, the third abnormal grinding analysis layer is used to establish a three-level abnormal grinding influence model through the second grinding parameter to obtain the third abnormal grinding parameter. The steps of constructing the three-level abnormal grinding influence model are: obtaining the third grinding vector based on the second grinding parameter , get the preset second grinding standard value to get the fourth grinding vector , obtain the relative deviation values ​​based on the difference between the third grinding parameter and the second grinding standard value:

[0018] ;

[0019] ;

[0020] Where: is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the noise parameter in the second polishing standard value, is the vibration parameter in the second grinding standard value, is the noise parameter in the third polishing parameter, is the vibration parameter in the third polishing parameter;

[0021] According to the relative deviation values, the third abnormal grinding parameter is obtained by importing them into the formula of the three-level abnormal grinding influence model. The formula of the three-level abnormal grinding influence model is:

[0022] ;

[0023] Where: is the third abnormal polishing parameter, is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the weight coefficient of the relative deviation value of the noise parameter, is the weight coefficient of the relative deviation value of the vibration parameter.

[0024] As a further solution of the present invention, the anomaly detection and adjustment module includes a primary anomaly determination unit, a secondary anomaly determination unit, and a tertiary anomaly determination unit;

[0025] The first-level abnormality determination unit is used to perform a first-level abnormality determination based on the first abnormal grinding parameter, compare the first abnormal grinding parameter with a preset first abnormal grinding parameter, and if the first abnormal grinding parameter is greater than or equal to the preset first abnormal grinding parameter, determine that there is an abnormality in the contact line grinding force; if the first abnormal grinding parameter is less than the preset first abnormal grinding parameter, determine that there is no abnormality in the contact line grinding force;

[0026] The secondary abnormality determination unit is used to perform secondary abnormality determination on the second abnormal grinding parameter, and compares the second abnormal grinding parameter with a preset second abnormal grinding parameter. If the second abnormal grinding parameter is greater than or equal to the preset second abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the second abnormal grinding parameter is less than the preset second abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force;

[0027] The three-level abnormality judgment unit is used to perform a three-level abnormality judgment on the third abnormal grinding parameter, and compares the third abnormal grinding parameter with the preset third abnormal grinding parameter. If the third abnormal grinding parameter is greater than or equal to the preset third abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the third abnormal grinding parameter is less than the preset third abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force.

[0028] As a further solution of the present invention, in the abnormality detection and adjustment module, after completing the determination of the first-level abnormality determination unit, the second-level abnormality determination unit, and the third-level abnormality determination unit, the abnormality detection and adjustment module sequentially performs the following technical feature steps:

[0029] Step 1: Send the first, second and third level judgment results to the intelligent grinding force optimization module;

[0030] Step 2: The intelligent grinding force optimization module generates a grinding force adjustment instruction according to the determination result, and sends the instruction to the grinding execution device;

[0031] Step 3: The intelligent grinding force optimization module generates a feed rate adjustment instruction according to the determination result, and sends the instruction to the drive unit of the feed grinding device;

[0032] Step 4: Record the determination result, the grinding force adjustment instruction, and the feed rate adjustment instruction into a log.

[0033] The technical effects and advantages of the grinding system for anti-electric corrosion contact lines of the present invention are as follows: the present invention obtains contact line grinding parameters through laser scanning, ultrasonic detection and sensors, and obtains standard contact line grinding images as the first grinding images and real-time contact line grinding images as the second grinding images through a high-pixel camera, extracts first surface feature data according to the contact line surface data for data processing, obtains first grinding image features based on the first grinding image respectively, can capture changes in surface subtle textures and defects, and timely reflect deviations in surface state; obtains second grinding image features based on the second grinding image, constructs a layered abnormal response model according to the contact line grinding parameters, the first grinding image features and the second grinding image features to obtain abnormal grinding parameters, can perform detailed evaluation of the grinding state from multiple levels and angles, and significantly improves the sensitivity and accuracy of abnormal detection; makes grinding abnormality judgment based on abnormal grinding parameters to ensure the quality of contact line grinding and prevent safety hazards of electric corrosion. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A system block diagram of a polishing system for anti-electrical corrosion contact wires provided by the present invention;

[0035] Figure 2 This is a comparison chart of abnormal grinding parameter curves of the intelligent grinding force optimization module provided by the present invention. DETAILED DESCRIPTION

[0036] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings. Obviously, the technical solutions described are only part of the present invention, not the entire invention. Based on the technical solutions of the present invention, all other technical solutions obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0037] like Figure 1As shown, the present invention provides a contact line grinding system for preventing electrical corrosion, comprising a multi-sensor data acquisition module, a data feature extraction module, an intelligent grinding force optimization module, and an anomaly detection and adjustment module. The multi-sensor data acquisition module is connected to the data feature extraction module, which is in turn connected to the intelligent grinding force optimization module, which is in turn connected to the anomaly detection and adjustment module. The multi-sensor data acquisition module is used to acquire contact line grinding parameters through laser scanning, ultrasonic detection, and sensors, and to acquire a standard contact line grinding image as a first grinding image and a real-time contact line grinding image as a second grinding image using a high-pixel camera. The contact line grinding parameters include first and second grinding parameters. The first grinding parameters include the surface roughness, local depression depth, and remaining thickness of the contact line grinding area; the second grinding parameters include noise parameters and vibration parameters during grinding. The data feature extraction module is used to extract first surface feature data based on the contact line surface data for data processing, acquiring first grinding image features based on the first grinding image and second grinding image features based on the second grinding image. The intelligent grinding force optimization module is used to construct a hierarchical abnormal response model based on the contact line grinding parameters, the first grinding image features, and the second grinding image features to obtain abnormal grinding parameters. The abnormality detection and adjustment module is used to determine grinding abnormalities based on the abnormal grinding parameters.

[0038] Specifically, the multi-sensor data acquisition module includes a laser scanning unit, an ultrasonic detection unit, a high-pixel camera, a noise sensor, and a vibration sensor; the fusion of multi-sensor data and multi-level detection make the system independent of a single data source, and can maintain the accuracy of the overall judgment even when some data contains noise or anomalies.

[0039] It should be specifically explained that the laser scanning unit and the ultrasonic detection unit are used to obtain the first polishing parameter; the high-pixel camera is used to obtain the first polishing image and the second polishing image; and the vibration sensor and the noise sensor are used to obtain the second polishing parameter.

[0040] Specifically, the data feature extraction module includes a data processing unit, an image processing unit and an image feature extraction unit; the image processing unit is connected to the image feature extraction unit.

[0041] The data processing unit is used to normalize the contact line grinding parameters and perform time domain alignment processing;

[0042] The image processing unit is used to perform denoising, grayscale conversion and size adjustment on the first polishing image and the second polishing image;

[0043] The image feature extraction unit is used to extract the first polished image using deep learning Get the first polishing image features , from the second polished image Get the second polishing image features ,in, is the feature extraction function.

[0044] In the disclosed invention patents mentioned in the background technology, the contact line is polished only by a mechanically driven grinding wheel (guide wheel + positioning wheel), which does not have the function of detecting the surface state. The present invention integrates five types of sensors: laser scanning, ultrasonic detection, high-pixel camera, noise and vibration sensor, to collect multi-source grinding parameters such as surface roughness, local depression depth, remaining thickness, noise and vibration in real time, and fuse and judge them to ensure information integrity and anti-interference. The disclosed invention patents mentioned in the background technology do not conduct in-depth research on images or grinding parameters, while the present invention normalizes and aligns the parameters in the time domain, denoises, grayscales, and resizes the collected "first grinding image" and "second grinding image" respectively, and then uses a deep learning feature extraction function. Obtaining high-dimensional features 、 , providing rich semantic information for subsequent intelligent decision-making. The existing disclosed invention patents cannot adjust the grinding force. Relying only on constant speed and pressure can easily cause over-grinding or under-grinding. The present invention constructs a "layered abnormal response model" to calculate the "abnormal grinding parameters" in real time based on the fused grinding parameters and image features, and dynamically outputs the optimal grinding force to accurately repair defects and protect the contact line thickness. At the same time, the existing technology lacks online abnormality judgment and feedback adjustment. The present invention sets an abnormality detection and adjustment module to judge the obtained abnormal grinding parameters. Once super-threshold behavior is detected, the grinding force is immediately adjusted or the operation is interrupted to form a closed-loop control, which greatly improves safety and reliability. In summary, the present invention seamlessly integrates five types of sensors, namely laser, ultrasound, image, noise and vibration. Different from using only single visual detection, it introduces dual deep learning features of the first and second polishing images, maps multi-source data to a hierarchical response structure, and realizes more refined polishing force optimization. It solves the deficiency of no feedback adjustment in traditional solutions, dynamically optimizes the polishing force, ensures uniform removal of defects and protects the remaining thickness, multi-sensor redundancy and abnormal closed-loop feedback, reduces the risk of single sensor failure, real-time abnormal judgment and automatic adjustment, interrupts abnormal polishing actions, eliminates safety hazards, has a high degree of automation and rapid response, improves polishing efficiency and finished product consistency, optimizes the polishing process, ensures that the polishing force is always in the best state, not only improves the polishing quality, but also reduces equipment wear and extends service life, thereby effectively preventing electrical corrosion caused by improper polishing.

[0045] Specifically, the abnormality detection and adjustment module includes a first abnormal polishing analysis layer, a second abnormal polishing analysis layer and a third abnormal polishing analysis layer: the first abnormal polishing analysis layer is used to detect the abnormal polishing image according to the first polishing image feature. And the second polished image features Import into the first-level abnormal polishing influence model to obtain the first abnormal polishing parameter; the second abnormal polishing analysis layer is used to establish the second-level abnormal polishing influence model through the first polishing parameter to obtain the second abnormal polishing parameter; the third abnormal polishing analysis layer is used to establish the third-level abnormal polishing influence model through the second polishing parameter to obtain the third abnormal polishing parameter.

[0046] It should be noted that the first abnormal polishing analysis layer is used to analyze the first polishing image features. And the second polished image features Import into the first level abnormal polishing impact model to obtain the first abnormal polishing parameters. The construction steps of the first level abnormal polishing impact model are as follows: And the second polished image features Get the polishing difference vector ,Based on the polishing difference vector, a first-level abnormal polishing impact model is constructed to calculate the Euclidean norm of the polishing vector, and then normalized to obtain the first abnormal polishing parameter.

[0047] The calculation formula of the first-level abnormal grinding impact model is:

[0048] ;

[0049] Where: is the first abnormal grinding parameter, is the normalized polishing difference vector, is the characteristic component of the first polished image in the i-th dimension, is the characteristic component of the second polished image in the i-th dimension, is the minimum difference value set during normalization, The maximum difference value set during normalization.

[0050] From the deep network Extract high-dimensional features of the two images before and after polishing respectively 、 , subtract dimension by dimension to get the polishing difference vector ,High-dimensional semantic features include texture, edge, and shape information, Each component of corresponds to a subtle visual difference, avoiding the omission of details that are difficult to identify with a single grayscale / edge method. Even if some local differences are introduced by illumination or noise, most components of the high-dimensional vector tend to zero. Only real polishing changes will accumulate significantly in multidimensional space. . Calculate the Euclidean norm of the polishing vector , with the preset 、 Linearly map to [0,1], let , then let the first abnormal grinding parameter Normalization eliminates the dimensional differences of the Euclidean norm under different workpieces and different shooting angles, so that the same threshold can be applied to various working conditions without manual repeated calibration. The Euclidean norm gathers multi-dimensional difference information. Once the actual grinding deviates from the normal range, Rapid decline, triggering anomalies; and local Not enough to make the whole Significant, avoid false alarms. Detect the first abnormal grinding parameter In the event of an anomaly, the system can quickly locate excessive changes in image representation rather than mechanical failure or sensor drift, and can also adjust the image layer individually / or The network structure improves the recognition accuracy of the polishing position and number without affecting the parameter layer model. This method accurately quantifies the real morphological changes before and after polishing, and converts the visual difference into the measurable first abnormal polishing parameter. The adaptive threshold avoids the failure of the fixed threshold in different environments. The high robustness takes into account both sensitivity and anti-interference, ensuring early warning of minor defects or over-polishing. The clear and maintainable modular structure provides robust and interpretable input for subsequent second and third level analysis layers.

[0051] Taking the maintenance of anti-corrosion aluminum alloy contact wires in a certain high-speed railway section as an example, after the maintenance trolley slides to the weld, the laser scanning unit and the ultrasonic detection unit first collect the surface roughness of 3.2μm, the local pit depth of 0.15mm and the remaining thickness of 7.35mm. Then the high-pixel camera takes the images before and after grinding respectively. When the grinding wheel starts, the noise sensor and the vibration sensor synchronously obtain the noise level of 68dB and the vibration amplitude of 0.12g; then, the pre-trained deep feature extraction network Extract 5-dimensional high-dimensional features from images before and after polishing =[0.42,0.76,0.31,0.58,0.63] and =[0.40,0.72,0.30,0.55,0.60], and then calculate the difference vector =[0.02,0.04,0.01,0.03,0.03], its Euclidean norm is 0.06245, and then =0 and =0.10 linear normalization to obtain =0.6245, the final first abnormal grinding parameter =1– ≈0.3755. Because This is far below the set safety threshold of 0.60. The system determines the risk of over-grinding in the weld area within milliseconds, immediately reduces the grinding pressure or suspends the operation to ensure that the contact line thickness does not continue to thin, and feeds back the information of this area to the subsequent second and third level models for more detailed parameter layer anomaly analysis.

[0052] Specifically, the second abnormal grinding analysis layer is used to establish a secondary abnormal grinding influence model through the first grinding parameter to obtain the second abnormal grinding parameter. The steps of constructing the secondary abnormal grinding influence model are: extracting the first grinding parameter, obtaining the first grinding vector according to the first grinding parameter , get the preset first grinding standard value and get the second grinding vector , obtain the relative deviation values ​​based on the difference between the first grinding parameter and the first grinding standard value:

[0053] ;

[0054] ;

[0055] ;

[0056] Where: is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, is the surface roughness in the first grinding standard value, is the local concave depth in the first grinding standard value, is the remaining thickness in the first grinding standard value, is the surface roughness in the first grinding parameter, is the local concave depth in the first grinding parameter, is the remaining thickness in the first grinding parameter;

[0057] According to the relative deviation values, the second abnormal grinding parameters are obtained by importing them into the formula of the secondary abnormal grinding influence model. The formula of the secondary abnormal grinding influence model is:

[0058] ;

[0059] Where: is the second abnormal polishing parameter, is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, To polish the strength, is the contact line diameter, is the weight coefficient of the relative deviation value of surface roughness, is the weight coefficient of the relative deviation value of the local depression depth, is the weight coefficient of the relative deviation value of the remaining thickness, is the sensitivity factor of the effect of grinding force on the depth of local depression, It is the sensitivity factor of the effect of grinding force on the remaining thickness.

[0060] Traditional contact line grinding devices rely solely on constant speed and pressure, and are unable to quantify deviations from standard values ​​for surface roughness, pit depth, and remaining thickness in real time, let alone perform a sensitivity-weighted comprehensive assessment based on these deviations. The present invention, on the other hand, maps multi-source grinding parameters into a differentiable mechanical model, accurately characterizing the impact of each deviation on the grinding force through partial derivatives and sensitivity factors, thereby forming a multi-parameter fusion anomaly criterion. This can quantitatively and dynamically convert the degree of deviation from the actual parameters into a fine-tuning signal for the grinding force, achieving refined feedback control of the grinding force, something that conventional technologies cannot achieve through experience or static threshold adjustment alone. Through this mechanism, the system can promptly detect the risk of local over- or under-grinding, effectively preventing contact line thickness loss or residual defects caused by a cut that is too deep or too shallow, and significantly improving grinding quality, safety, and reliability. As the secondary analysis layer of the hierarchical anomaly detection system, this model complements the detection results of the first layer (based on image features) and the third layer (based on dynamic parameters), enabling the entire system to monitor the grinding status from multiple angles, further improving its sensitivity and response speed to abnormal situations. By calculating various relative deviations in real time and dynamically evaluating the current state of the workpiece using sensitivity and weight parameters, the system can detect accumulated deviations more promptly and provide feedback to the intelligent grinding force optimization module to achieve closed-loop control, effectively reducing the probability of problems such as electrical corrosion and uneven grinding.

[0061] Taking the same high-speed railway weld maintenance as an example, after the system detects the risk of over-grinding in the first-level image layer, it enters the second abnormal grinding analysis. The maintenance trolley collects surface roughness of 3.2μm, pit depth of 0.15mm, and remaining thickness of 7.35mm through laser and ultrasonic detection. After comparing with the preset standard values ​​of standard surface roughness of 2.5μm, standard pit depth of 0.10mm, and standard remaining thickness of 8.0mm, the relative deviation is calculated. =|3.2−2.5|3.2≈0.219, =|0.15−0.10|0.15≈0.333, =|7.35−8.00|7.35≈0.088, the partial derivative of the grinding force model with respect to roughness is taken as 1.2N / μm, and the sensitivity factor of the grinding force on the local concave depth is The sensitivity factor of the grinding force on the remaining thickness is 0.8N / mm. The weight coefficient of the relative deviation value of surface roughness is 0.5N / mm. =0.4, weight coefficient of relative deviation value of local depression depth =0.3, weight coefficient of relative deviation value of remaining thickness =0.3, substitute =0.4×1.2×0.219+0.3×0.8×0.333+0.3×0.5×0.088≈0.198. (0.198) > the preset safety threshold of 0.15. The system determines that the grinding force and operating conditions deviate from normal. The control algorithm then reduces the current grinding force by 10% from 15N to 13.5N and slows the feed rate to prevent further excessive wear. Practice has proven that this secondary model can accurately trigger protective action before the contact line thickness loss exceeds the structural limit, significantly reducing the risk of wire breakage caused by excessive wear, while also avoiding the blind grinding and missed detections that occur in traditional static threshold modes.

[0062] Specifically, the third abnormal grinding analysis layer is used to establish a three-level abnormal grinding influence model through the second grinding parameter to obtain the third abnormal grinding parameter. The construction steps of the three-level abnormal grinding influence model are: based on the second grinding parameter, obtain the third grinding vector , get the preset second grinding standard value to get the fourth grinding vector , obtain the relative deviation values ​​based on the difference between the third grinding parameter and the second grinding standard value:

[0063] ;

[0064] ;

[0065] Where: is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the noise parameter in the second polishing standard value, is the vibration parameter in the second grinding standard value, is the noise parameter in the third polishing parameter, is the vibration parameter in the third polishing parameter;

[0066] According to the relative deviation values, the third abnormal grinding parameter is obtained by importing them into the formula of the three-level abnormal grinding influence model. The formula of the three-level abnormal grinding influence model is:

[0067] ;

[0068] Where: is the third abnormal polishing parameter, is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the weight coefficient of the relative deviation value of the noise parameter, is the weight coefficient of the relative deviation value of the vibration parameter.

[0069] The present invention introduces a dynamic monitoring model with noise and vibration as the core in the third-level abnormal grinding analysis layer: traditional grinding equipment usually only focuses on single mechanical parameters such as speed or pressure, and rarely incorporates acoustic and vibration signals generated at the work site into real-time abnormality judgment. However, this solution first collects the noise level of the current grinding process through noise sensors and vibration sensors. and vibration amplitude and with the preset standard value 、 Calculate the relative deviation and , and then follow The weighted formula outputs the third abnormal grinding parameter On the one hand, this technical approach breaks through the limitations of relying solely on visual and mechanical parameters, and directly uses the acoustic-mechanical coupling characteristics to reflect the microscopic contact state between the grinding wheel and the contact line and the equipment wear; on the other hand, through the adjustable weight coefficient 、 By giving noise and vibration flexible priority in abnormality judgment, the model can not only quickly respond to sudden mechanical resonances, but also sensitively capture gradual abnormalities caused by grinding wheel wear or unstable clamping. This solves the problem that conventional technology cannot promptly identify hidden grinding abnormalities caused by grinding wheel wear, loose bearings or loose fixtures. These faults are often manifested through acoustic vibration signals before the image and roughness parameters have changed significantly. They can immediately trigger alarms or autonomous adjustments at the embryonic stage of abnormalities, avoiding further damage to the contact line or equipment, and significantly improving the safety, continuity and maintenance efficiency of grinding operations. According to the technical solution of the present invention, after the maintenance trolley completes a grinding operation at the same weld position every day, the system will calculate the three levels of abnormality parameters respectively. 、 、 And superimposed on the daily trend curve to reflect the "grinding abnormality" phenomenon at different levels. When the grinding wheel begins to wear seriously, the depth feature difference of the image before and after grinding increases sharply, resulting in the first-level parameter A peak value appeared on that day, at which time the surface roughness, concave depth and remaining thickness of the material also deviated from the standard value. The synchronous increase indicates that the actual grinding effect and the set grinding force are gradually unbalanced. Furthermore, the unstable contact between the worn grinding wheel and the workpiece will produce stronger noise and vibration. Also jumped to the warning zone on the same day, such as Figure 2 The abnormal grinding parameter curve comparison diagram of the intelligent grinding force optimization module is shown. The three curves in the figure represent three different levels of abnormal grinding parameters. Polish the parameter curve for the first anomaly in a week, Polishing parameter curves for the second anomaly in a week, This is the third abnormal grinding parameter curve in a week. The vertical axis shows the value of the abnormal grinding parameter (between 0 and 1), and the horizontal axis is the time or date (such as Monday to Sunday). According to the trend of abnormal indicators at each level, the intelligent grinding force optimization module can dynamically adjust the grinding force or trigger the abnormal adjustment module to intervene in time to prevent excessive wear or deterioration of electrical contact performance. Figure 2 Taking a Thursday as an example, the common inflection point of the three curves accurately reflected that the grinding wheel wear had reached critical levels that day, affecting the image layer, altering the mechanical grinding effect, and generating abnormal impacts on the acoustic and vibration layers. This multi-level, cross-domain, synchronized peak promptly prompted maintenance personnel to replace the grinding wheel and calibrate the fixture. The three curves subsequently returned to a safe range on Friday and Saturday, verifying the effectiveness of the intervention. Therefore, the present invention not only allows for intuitive monitoring of the wear-functional imbalance-acoustic and vibration anomaly fault evolution process through these curves, but also enables dynamic maintenance decisions to be made accordingly, significantly improving the safety and stability of contact line grinding.

[0070] Specifically, the surface roughness of the contact line polished area is calculated as:

[0071] ;

[0072] Where: Surface roughness of the polished area for the contact line, is the number of sampling points, that is, the total number of points with measured height in the polishing area, is the height of the contact line at the i-th position, is the average height of the polished area of ​​the contact line.

[0073] The calculation formula for the local depression depth in the contact line grinding area is:

[0074] ;

[0075] Where: is the local depression depth of the contact line grinding area, is the initial height of the contact line before grinding, The height of the contact line after polishing at the current moment.

[0076] The calculation formula for the remaining thickness of the contact line grinding area is:

[0077] ;

[0078] Where: The remaining thickness of the contact line grinding area, is the initial thickness of the contact line before grinding, is the feed rate of the grinding tool, To polish the time, For in time The thickness of the material removed by grinding.

[0079] Specifically, the intelligent grinding force optimization module is used to determine grinding abnormalities based on abnormal grinding parameters, and to determine abnormalities based on the first abnormal grinding parameter, the second abnormal grinding parameter, and the third abnormal grinding parameter, respectively, to grade and determine whether there is an abnormality in the contact line.

[0080] The abnormality detection and adjustment module includes a first-level abnormality determination unit, a second-level abnormality determination unit, and a third-level abnormality determination unit. The first-level abnormality determination unit is used to perform a first-level abnormality determination based on the first abnormal grinding parameter, and compare the first abnormal grinding parameter with the preset first abnormal grinding parameter. If the first abnormal grinding parameter is greater than or equal to the preset first abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the first abnormal grinding parameter is less than the preset first abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force. The second-level abnormality determination unit is used to perform a second-level abnormality determination based on the second abnormal grinding parameter, and compare the second abnormal grinding parameter with the preset second abnormal grinding parameter. If the second abnormal grinding parameter is greater than or equal to the preset second abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the second abnormal grinding parameter is less than the preset second abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force. The three-level abnormality judgment unit is used to perform a three-level abnormality judgment on the third abnormal grinding parameter, and compares the third abnormal grinding parameter with the preset third abnormal grinding parameter. If the third abnormal grinding parameter is greater than or equal to the preset third abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the third abnormal grinding parameter is less than the preset third abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force.

[0081] Taking the grinding of a weld section as an example, we will explain how the three levels of judgment work together to achieve a protective effect:

[0082] The first step is data collection and basic parameter calculation: obtain height data from laser scanning and ultrasonic testing, the number of sampling points =5, measured =[10.02,9.98,9.95,10.05,9.90]mm, average height =10mm, =0.02+0.02+0.05+0.05+0.105=0.048mm, the depth of the depression is obtained by comparing the highest point before and after grinding. =10.50mm, the current highest =9.90mm, =0.60mm, feed rate =0.2mm / s, grinding time =10s, initial thickness =12mm, =12−0.2×10=10mm;

[0083] The second step is to determine the first-level image layer: a high-pixel camera is used to obtain images before and after polishing. The network structure extracts features of the image before polishing =[0.45,0.80,0.35,0.60,0.65], the image features after polishing are =[0.40,0.75,0.30,0.55,0.60], the difference vector is [0.05,0.05,0.05,0.05,0.05], =1-0.559=0.441, and the preset first-level threshold is 0.6. Since 0.441<0.6, the image analysis result is determined to be no layer anomaly;

[0084] The third step is to determine the secondary parameters: =0.167, =0.167, =0.1, sensitivity and weight data are =1N / μm, =0.5N / mm, =0.4N / mm, weight coefficient of relative deviation value of surface roughness =0.5, weight coefficient of relative deviation value of local depression depth =0.3, weight coefficient of relative deviation value of remaining thickness =0.2, =0.5×1×0.167+0.3×0.5×0.167+0.2×0.4×0.100≈0.084+0.025+0.008=0.117. The preset secondary threshold is 0.1. 0.117>0.1, and the second-layer parameters are judged to be abnormal.

[0085] Step 4: Determine the third level of acoustic vibration layer: Collect noise =75dB, vibration =0.18g; the noise standard value is 65dB, the vibration standard value is 0.10g, =10 / 75=0.133, =0.08 / 0.18=0.444, and The weights are 0.4 and 0.6 respectively, then the value of the third abnormal grinding parameter is calculated according to the formula: ,The preset third-level threshold is 0.20, and the acoustic vibration layer is judged to be abnormal because 0.319>0.20;

[0086] Step 5: Closed-loop adjustment: Although the image layer is normal, the abnormality of both the second and third levels indicates that the grinding force and tool wear have begun to become unbalanced and the grinding wheel resonance is abnormal. The system automatically reduces the grinding force by 15% from 15N to 12.75N, the feed rate from 0.2mm / s to 0.15mm / s, and issues a reminder to replace the grinding wheel.

[0087] Through the implementation of the above-mentioned technical means, the present invention can timely adjust the remaining thickness before it continues to decrease, protect the safety of the contact line structure, make multi-level collaborative judgments, and conduct comprehensive verification of images, mechanics and acoustic vibrations, significantly reduce missed reports or false alarms caused by single signal interference, automatically generate maintenance instructions, avoid manual blind maintenance, and thus improve work efficiency and line availability.

[0088] It should be noted that in the anomaly detection and adjustment module, after completing the determinations of the first-level anomaly determination unit, the second-level anomaly determination unit, and the third-level anomaly determination unit, the anomaly detection and adjustment module sequentially performs the following technical feature steps:

[0089] Step 1: Send the first, second and third level judgment results to the intelligent grinding force optimization module;

[0090] Step 2: The intelligent grinding force optimization module generates a grinding force adjustment instruction according to the determination result, and sends the instruction to the grinding execution device;

[0091] Step 3: The intelligent grinding force optimization module generates a feed rate adjustment instruction according to the determination result, and sends the instruction to the drive unit of the feed grinding device;

[0092] Step 4: Record the determination result, the grinding force adjustment instruction, and the feed rate adjustment instruction into a log.

[0093] In summary, the present invention obtains contact line grinding parameters through laser scanning, ultrasonic detection and sensors, and obtains standard contact line grinding images as the first grinding images and real-time contact line grinding images as the second grinding images through a high-pixel camera, extracts first surface feature data based on the contact line surface data for data processing, and obtains first grinding image features based on the first grinding image, which can capture changes in subtle surface texture and defects and timely reflect deviations in surface state; obtains second grinding image features based on the second grinding image, and constructs a layered abnormal response model based on the contact line grinding parameters, the first grinding image features and the second grinding image features to obtain abnormal grinding parameters, which can perform detailed evaluation of the grinding state from multiple levels and angles, significantly improving the sensitivity and accuracy of abnormal detection; performs grinding abnormality judgment based on abnormal grinding parameters to ensure the quality of contact line grinding and prevent safety hazards of electrical corrosion.

[0094] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0095] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A grinding system for anti-corrosion contact wires, comprising a multi-sensor data acquisition module, a data feature extraction module, an intelligent grinding force optimization module, and an abnormality detection and adjustment module, characterized in that: The multi-sensor data acquisition module is used to obtain contact line grinding parameters through laser scanning, ultrasonic detection and sensors, and to obtain a standard contact line grinding image as a first grinding image and a real-time contact line grinding image as a second grinding image through a high-pixel camera; The data feature extraction module is used to extract the first surface feature data according to the contact line surface data for data processing, and obtain the first polishing image feature based on the first polishing image and the second polishing image feature based on the second polishing image; the intelligent polishing force optimization module is used to construct a layered abnormal response model according to the contact line polishing parameters, the first polishing image feature and the second polishing image feature to obtain the abnormal polishing parameters; the intelligent polishing force optimization module includes a first abnormal polishing analysis layer; the first abnormal polishing analysis layer is used to construct a layered abnormal response model according to the first polishing image feature And the second polished image features Import into the first level abnormal polishing impact model to obtain the first abnormal polishing parameters. The construction steps of the first level abnormal polishing impact model are as follows: And the second polished image features Get the polishing difference vector ,Based on the polishing difference vector, a first-level abnormal polishing impact model is constructed to calculate the Euclidean norm of the polishing vector, and then normalized to obtain the first abnormal polishing parameter.

2. The grinding system for anti-corrosion contact wire according to claim 1, characterized in that: The data feature extraction module includes a data processing unit, an image processing unit, and an image feature extraction unit; the data processing unit is used to normalize the contact line grinding parameters and perform time domain alignment processing; the image processing unit is used to perform denoising, grayscale, and size adjustment processing on the first grinding image and the second grinding image; The image feature extraction unit is used to extract the first polished image using deep learning Get the first polishing image features , from the second polished image Get the second polishing image features ,in, is the feature extraction function.

3. The grinding system for anti-corrosion contact wire according to claim 1, characterized in that: The intelligent grinding force optimization module also includes a second abnormal grinding analysis layer and a third abnormal grinding analysis layer; the second abnormal grinding analysis layer is used to establish a second-level abnormal grinding influence model through the first grinding parameters to obtain the second abnormal grinding parameters; the third abnormal grinding analysis layer is used to establish a third-level abnormal grinding influence model through the second grinding parameters to obtain the third abnormal grinding parameters.

4. The grinding system for anti-corrosion contact wire according to claim 3, characterized in that: The second abnormal grinding analysis layer is used to establish a secondary abnormal grinding influence model through the first grinding parameter to obtain the second abnormal grinding parameter. The construction steps of the secondary abnormal grinding influence model are: extracting the first grinding parameter, obtaining the first grinding vector according to the first grinding parameter , get the preset first grinding standard value and get the second grinding vector , obtain the relative deviation values ​​based on the difference between the first grinding parameter and the first grinding standard value: ； ； ； Where: is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, is the surface roughness in the first grinding standard value, is the local concave depth in the first grinding standard value, is the remaining thickness in the first grinding standard value, is the surface roughness in the first grinding parameter, is the local concave depth in the first grinding parameter, is the remaining thickness in the first grinding parameter; According to the relative deviation values, the second abnormal grinding parameters are obtained by importing them into the formula of the secondary abnormal grinding influence model. The formula of the secondary abnormal grinding influence model is: ； Where: is the second abnormal polishing parameter, is the relative deviation value of surface roughness, is the relative deviation value of the local depression depth, is the relative deviation value of the remaining thickness, To polish the strength, is the contact line diameter, is the weight coefficient of the relative deviation value of surface roughness, is the weight coefficient of the relative deviation value of the local depression depth, is the weight coefficient of the relative deviation value of the remaining thickness, is the sensitivity factor of the effect of grinding force on the depth of local depression, It is the sensitivity factor of the effect of grinding force on the remaining thickness.

5. The grinding system for anti-corrosion contact wire according to claim 3, characterized in that: The third abnormal grinding analysis layer is used to establish a three-level abnormal grinding influence model through the second grinding parameter to obtain the third abnormal grinding parameter. The construction steps of the three-level abnormal grinding influence model are: based on the second grinding parameter, obtain the third grinding vector , obtain the fourth grinding vector according to the preset second grinding standard value , obtain the relative deviation values ​​based on the difference between the third grinding parameter and the second grinding standard value: ； ； Where: is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the noise parameter in the second polishing standard value, is the vibration parameter in the second grinding standard value, is the noise parameter in the third polishing parameter, is the vibration parameter in the third polishing parameter; According to the relative deviation values, the third abnormal grinding parameter is obtained by importing them into the formula of the three-level abnormal grinding influence model. The formula of the three-level abnormal grinding influence model is: ； Where: is the third abnormal polishing parameter, is the relative deviation value of the noise parameter, is the relative deviation value of the vibration parameter, is the weight coefficient of the relative deviation value of the noise parameter, is the weight coefficient of the relative deviation value of the vibration parameter.

6. The grinding system for anti-corrosion contact wire according to claim 1, characterized in that: The abnormality detection and adjustment module includes a primary abnormality determination unit, a secondary abnormality determination unit and a tertiary abnormality determination unit.

7. The grinding system for anti-corrosion contact wire according to claim 6, characterized in that: In the abnormality detection and adjustment module, the first-level abnormality judgment unit is used to perform a first-level abnormality judgment based on the first abnormal grinding parameter, and compare the first abnormal grinding parameter with the preset first abnormal grinding parameter. If the first abnormal grinding parameter is greater than or equal to the preset first abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; if the first abnormal grinding parameter is less than the preset first abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force.

8. The grinding system for anti-corrosion contact wire according to claim 7, characterized in that: In the abnormality detection and adjustment module, the secondary abnormality determination unit is used to perform secondary abnormality determination on the second abnormal grinding parameter, and compares the second abnormal grinding parameter with a preset second abnormal grinding parameter. If the second abnormal grinding parameter is greater than or equal to the preset second abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; If the second abnormal grinding parameter is less than the preset second abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force.

9. The grinding system for anti-corrosion contact wire according to claim 8, characterized in that: In the abnormality detection and adjustment module, the three-level abnormality determination unit is used to perform a three-level abnormality determination based on the third abnormal grinding parameter, and compares the third abnormal grinding parameter with a preset third abnormal grinding parameter. If the third abnormal grinding parameter is greater than or equal to the preset third abnormal grinding parameter, it is determined that there is an abnormality in the contact line grinding force; If the third abnormal grinding parameter is less than the preset third abnormal grinding parameter, it is determined that there is no abnormality in the contact line grinding force.

10. The grinding system for anti-corrosion contact wire according to claim 9, characterized in that: In the anomaly detection and adjustment module, after completing the determinations of the first-level anomaly determination unit, the second-level anomaly determination unit, and the third-level anomaly determination unit, the anomaly detection and adjustment module sequentially performs the following technical feature steps: Step 1: Send the first, second and third level judgment results to the intelligent grinding force optimization module; Step 2: The intelligent grinding force optimization module generates a grinding force adjustment instruction according to the determination result, and sends the instruction to the grinding execution device; Step 3: The intelligent grinding force optimization module generates a feed rate adjustment instruction according to the determination result, and sends the instruction to the drive unit of the feed grinding device; Step 4: Record the determination result, the grinding force adjustment instruction, and the feed rate adjustment instruction into a log.

Images