A rigid catenary adaptive grinding control method and system

Abstract

The application discloses a rigid contact net self-adaptive polishing control method and system, belongs to the technical field of track traffic power supply equipment maintenance, obtains the pantograph-catenary action historical data of each position along the busbar direction of the contact net, and constructs a contact wire damage index distribution curve; based on the damage index, a high damage interval is identified, roughness, corrugation amplitude and oxidation thickness parameters are extracted, a damage intensity vector is constructed, and a fuzzy comprehensive evaluation method is used to calculate the damage level score of each position; according to the scoring results, a preset polishing strategy library is matched, corresponding polishing wheel combinations, rotating speeds and contact pressures are determined, and a control parameter set is generated; the polishing device dynamically adjusts the working state according to the control parameters in the running process, and the method realizes the whole-process self-adaptive closed-loop control of state recognition, strategy decision, execution control and quality feedback of the contact wire polishing operation.

Description

Technical Field

[0001] This invention relates to the field of maintenance technology for rail transit power supply equipment, specifically to an adaptive grinding control method and system for rigid contact wires. Background Technology

[0002] As a crucial component of urban rail transit power supply systems, the rigid contact wire's operational status directly impacts the quality of power reception for vehicles and system safety. Prolonged exposure to the pantograph and contact wire can cause problems such as arcing, corrugation, and indentation. If not addressed promptly, these issues can easily lead to safety hazards like pantograph arcing and contact wire breakage. Currently, contact wire polishing is primarily performed manually or semi-automatically, resulting in inconsistent work standards, low efficiency, and high dust exposure.

[0003] Especially in track lines with complex operating conditions, such as underground sections, extra-long sections, curves, or transition zones, the contact wire damage patterns are diverse, and the grinding requirements are highly heterogeneous. If a fixed procedure is used for uniform grinding, there is often a phenomenon where slightly worn areas are over-grinded while severely worn areas are not adequately repaired. This not only wastes resources but may also further aggravate fatigue damage to the contact wire. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive grinding control method and system for rigid contact wires to address the shortcomings in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an adaptive grinding control method for rigid contact wires, comprising:

[0006] Historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located are obtained, and a contact wire damage index distribution curve L(x) is established, where x is the position coordinate along the busbar direction;

[0007] Based on the damage index distribution curve L(x), high damage intervals are identified, and the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the interval are extracted.

[0008] Based on the extracted R(x), B(x), and O(x) parameters, a damage intensity vector D(x) = [R(x), B(x), O(x)] is constructed, and a fuzzy comprehensive evaluation method is used to calculate the damage level score V(x) for each x position;

[0009] Based on the score V(x), a preset grinding strategy library is matched to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and a control parameter set C(x) = [T(x), N(x), P(x)] is generated;

[0010] The control grinding device performs variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction;

[0011] The reflectivity F(x) and residual roughness Rr(x) of the contact line surface after polishing are obtained and fed back to the control system.

[0012] Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

[0013] Preferably, establishing the contact line damage index distribution curve L(x) includes:

[0014] Collect current fluctuation values ​​and vertical impact acceleration data generated during the pantograph-catenary interaction process;

[0015] The current fluctuation values ​​and impact acceleration data are normalized to construct a time series of pantograph-catenary action intensity, and then divided into different segments according to the time window.

[0016] By synchronously registering the location information of the rail vehicle with the time series of pantograph-catenary action intensity, the spatial mapping relationship along the busbar direction x is obtained, and a location association dataset M(x) is constructed.

[0017] Based on M(x), the cumulative impact index and energy dissipation parameters per unit length are calculated, fused into the contact line damage index L(x), and fitted as a continuous distribution curve.

[0018] Preferably, identifying high-damage regions based on the damage index distribution curve L(x) includes:

[0019] A sliding window scanning analysis was performed on the contact wire damage index distribution curve L(x). An adaptive threshold discrimination method was used to identify continuous position segments where the L(x) value is greater than the set threshold L_thresh, which were taken as high damage intervals X_H.

[0020] Within the high-damage range X_H, the longitudinal profile height variation data of the contact line surface is obtained, and the roughness parameter R(x) is extracted by spatial spectrum analysis, where R(x) represents the average height deviation per unit length.

[0021] Fourier transform is performed on the contact line profile data to extract the peak amplitude of the ripple characteristics in the main frequency band, which is denoted as ripple amplitude B(x), and this is used to identify the sections with medium and long ripples.

[0022] Within the high-damage range X_H, spectral reflectance data of the contact line surface are collected, and the oxide layer thickness O(x) is calculated by combining the reflectance-oxide layer thickness fitting model.

[0023] Preferably, the step of calculating the damage level score V(x) for each x location using the fuzzy comprehensive evaluation method includes:

[0024] The extracted contact line roughness R(x), ripple amplitude B(x), and oxide thickness O(x) are normalized and mapped to the [0,1] interval to construct a dimensionless damage intensity vector D(x) = [R'(x), B'(x), O'(x)];

[0025] Three membership function sets for the three damage factors are defined, and fuzzy membership functions of “mild”, “moderate” and “severe” are established for R’(x), B’(x) and O’(x) respectively, using triangular or trapezoidal function construction methods.

[0026] Based on the fuzzy evaluation rule matrix, the membership degrees of each term of D(x) are input into the weighted fuzzy inference system, and the comprehensive membership vector U(x) is calculated by combining the preset weight factors.

[0027] The centroid method is used to defuzzify the comprehensive membership vector U(x) and output the final damage level score V(x) at the corresponding position x, which is used to quantify the comprehensive damage degree of the contact line.

[0028] Preferably, the step of matching the score V(x) with a preset polishing strategy library to determine the polishing wheel type combination T(x), rotational speed N(x), and contact pressure P(x) at the corresponding position x includes:

[0029] Based on the obtained damage level score V(x), interval matching is performed in the polishing strategy level table to assign V(x) to one of the preset polishing level intervals, including four level labels L_g(x): "no polishing", "low intensity polishing", "medium intensity polishing" and "high intensity polishing".

[0030] Based on the grinding grade label L_g(x), the corresponding grinding parameter template is retrieved from the grinding strategy library stored in the control system to obtain the standard grinding wheel combination T_ref, recommended speed range N_ref, and contact pressure range P_ref for that grade.

[0031] By combining the fine-tuning weights of the extracted roughness R(x), ripple amplitude B(x), and oxide thickness O(x), the standard parameters T_ref, N_ref, and P_ref are locally personalized through the interpolation correction algorithm to generate a refined grinding parameter combination T(x), N(x), and P(x);

[0032] The T(x), N(x), and P(x) are integrated into a complete set of control parameters C(x) = [T(x), N(x), P(x)] and stored in the task queue.

[0033] Preferably, the controlled polishing device performs a variable-parameter dynamic polishing operation according to the control parameter set C(x), including:

[0034] The multi-axis actuator in the control grinding device receives the control parameter set C(x), analyzes the target values ​​of grinding wheel type T(x), rotation speed N(x) and contact pressure P(x) in real time according to the change of grinding position x, and generates a multi-channel servo control instruction set for the distributed drive module;

[0035] During the movement of the grinding device along the busbar direction, the current displacement coordinate x' is obtained by a position synchronous encoder and matched with the target coordinate x in the control parameter set;

[0036] The embedded controller enables independent speed adjustment of each grinding motor, allowing multiple grinding wheels to achieve differentiated speed adjustment based on the N(x) value of their respective positions x.

[0037] The contact pressure between the grinding wheel and the contact line is controlled by an electric lead screw or pneumatic pressure regulating module, and the wheel arm clamping force is dynamically adjusted according to the P(x) value of the current position point x.

[0038] This invention also proposes an adaptive grinding control system for rigid contact wires, comprising:

[0039] Data acquisition module: acquire historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located, and establish the contact wire damage index distribution curve L(x), where x is the position coordinate along the busbar direction;

[0040] High damage zone identification module: Based on the damage index distribution curve L(x), identify high damage zones and extract the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the zone;

[0041] Damage level determination module: Based on the extracted R(x), B(x), and O(x) parameters, construct the damage intensity vector D(x) = [R(x), B(x), O(x)], and use the fuzzy comprehensive evaluation method to calculate the damage level score V(x) for each x position;

[0042] Control parameter generation module: Based on the score V(x), match the preset grinding strategy library to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and generate the control parameter set C(x) = [T(x), N(x), P(x)];

[0043] Execution module: Controls the grinding device to perform variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction;

[0044] Feedback data acquisition module: acquires the surface reflectivity F(x) and residual roughness Rr(x) of the contact line after polishing, and feeds them back to the control system;

[0045] Threshold comparison module: Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

[0046] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0047] 1. This invention introduces a damage-driven, grinding feedback, and dynamic control closed-loop mechanism to achieve adaptive optimization of the entire contact wire grinding process, significantly improving the accuracy and efficiency of the operation. The system can automatically adjust the grinding wheel type, speed, and pressure according to the actual condition of the contact wire, and perform real-time evaluation and closed-loop compensation of the grinding effect by online monitoring of reflectivity and roughness indicators, ensuring that every position point meets the preset quality standard, thus overcoming the problems of "over-grinding" or "insufficient grinding" that exist in traditional manual or fixed-program grinding.

[0048] 2. This invention has comprehensive advantages such as refined parameter control, intelligent operation process, quantitative quality assessment, and automated compensation mechanism. It can effectively reduce reliance on manual labor, reduce the intensity of high-altitude operations, and improve the overall maintenance safety and efficiency. It is especially suitable for grinding urban rail rigid contact networks in long sections, complex conditions, or high-risk scenarios, and has significant engineering promotion value. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0050] Figure 1 This is a mind map of the method of the present invention.

[0051] Figure 2 This is a mind map of the system modules of the present invention. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Example 1, please refer to Figure 1 As shown in this embodiment, an adaptive grinding control method for rigid contact wires includes:

[0054] Historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located are obtained, and a contact wire damage index distribution curve L(x) is established, where x is the position coordinate along the busbar direction;

[0055] Based on the damage index distribution curve L(x), high damage intervals are identified, and the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the interval are extracted.

[0056] Based on the extracted R(x), B(x), and O(x) parameters, a damage intensity vector D(x) = [R(x), B(x), O(x)] is constructed, and a fuzzy comprehensive evaluation method is used to calculate the damage level score V(x) for each x position;

[0057] Based on the score V(x), a preset grinding strategy library is matched to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and a control parameter set C(x) = [T(x), N(x), P(x)] is generated;

[0058] The control grinding device performs variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction;

[0059] The reflectivity F(x) and residual roughness Rr(x) of the contact line surface after polishing are obtained and fed back to the control system.

[0060] Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

[0061] To obtain historical data on the pantograph-catenary action process, a set of data acquisition and sensing units needs to be deployed on the pantograph of the trains in daily operation, specifically including:

[0062] Current sensor: Installed on the pantograph leads, it collects the microscopic fluctuations in current during pantograph-catenary contact. Abnormal current peaks can reflect potential hazards such as momentary electric arcs and poor contact.

[0063] The triaxial accelerometer is installed on the pantograph slide bracket and mainly collects the Z-axis (vertical direction) acceleration changes to record the mechanical impact process between the pantograph and the contact wire.

[0064] GNSS positioning module or vehicle-mounted track coding and matching system: used to provide high-precision location tags for data streams to achieve spatiotemporal registration.

[0065] Data is collected at a frequency of 20-50 Hz per second and timestamped with the vehicle time synchronization system to ensure the accuracy of subsequent data processing.

[0066] The collected data is characterized by high frequency, drastic fluctuations, and high noise, and therefore requires the following processing:

[0067] Denoising and filtering: Wavelet denoising is used to filter acceleration data and current fluctuation values ​​to remove high-frequency pseudo signals caused by environmental interference;

[0068] Normalization: The current fluctuation value I(t) and the acceleration value A(t) are normalized in a dimensionless manner to distribute them in the interval [0,1], which facilitates subsequent fusion;

[0069] Window segmentation: The continuous data stream is divided into sliding windows of 10 seconds each. The data in each window is recorded as a basic unit for further calculation of statistics.

[0070] The two types of time-series signals obtained after processing are denoted as follows:

[0071] Current fluctuation sequence S_I(t);

[0072] Acceleration impact sequence S_A(t);

[0073] By using GNSS or track position coding systems, the aforementioned time series data is matched with the train's running position to obtain the actual spatial coordinates of each data segment on the busbar.

[0074] Let Δx be the distance the train travels per unit time. Then each window can be mapped to a certain location interval. Thus, a location association dataset M(x) along the x-axis is constructed, which includes the following fields: average current fluctuation value I_avg(x); peak acceleration A_max(x); impact event frequency N_peak(x); and data acquisition timestamp t(x).

[0075] After completing the construction of M(x), this invention defines a damage index L(x) through a multi-index fusion model, which is used to comprehensively evaluate the damage level of the contact line at position x.

[0076] L(x) is defined as follows: Where: α, β, and γ are weighting coefficients, representing the importance of electrical damage, mechanical shock, and fatigue frequency, respectively; recommended values ​​are α = 0.4, β = 0.4, and γ = 0.2 (which can be adjusted according to the actual situation of the operating line); I_avg(x): represents the average current fluctuation amplitude in this range; A_max(x): represents the maximum acceleration impact value in this range; N_peak(x): the number of impacts exceeding the set threshold in this range. The number of impacts N_peak(x) is obtained by counting the number of times A(t) exceeds a certain acceleration threshold A_thresh, for example, A_thresh = 2.5g.

[0077] The aforementioned damage index L(x) may have discrete noise points in its spatial distribution, therefore a fitting method is needed to form a continuous curve. This invention uses spline interpolation (such as cubic splines) to smooth L(x), obtaining the final contact line damage index distribution curve L(x). This curve uses the busbar direction position x as the abscissa and the damage index value as the ordinate, presenting a panoramic view of the health status of the contact line along the entire section.

[0078] After the L(x) curve is generated, this invention employs a sliding window scanning and adaptive threshold discrimination algorithm to identify high-damage segments, specifically including:

[0079] The L(x) curve is divided into a continuous sliding window of length Δx, typically Δx is between 0.5 and 1.0 meters;

[0080] Calculate the mean L_avg(i) of the L(x) values ​​within each window;

[0081] Set the damage threshold L_thresh, typically with a value of 0.6;

[0082] If L_avg(i) ≥ L_thresh, then mark the window as a high-damage window;

[0083] By merging consecutive high-damage windows, a set of high-damage intervals X_H is constructed.

[0084] The above process can be completed automatically by the embedded data analysis module without manual intervention. The threshold L_thresh can be optimized and adjusted based on historical wear statistics to ensure a balance between recognition sensitivity and false positive rate.

[0085] This invention employs laser confocal profilometry to perform microscopic three-dimensional imaging of the contact line surface within the X_H interval, thereby obtaining its microscopic height profile curve Z(x).

[0086] The specific operating procedure is as follows:

[0087] A non-contact laser confocal scanning module is integrated at the front end of the grinding device;

[0088] During the operation of the equipment along the contact line, the laser scanner acquires Z(x) profile data at a 10-micrometer interval in the high-damage zone X_H;

[0089] The arithmetic mean surface roughness Ra, as defined in ISO 4287, is calculated as follows:

[0090] Where L is the scan length and Z(x) is the height deviation function.

[0091] Roughness R(x) is the Ra value, used to characterize the roughness of a material surface after wear. A typical threshold is:

[0092] R(x) < 3μm: No polishing required;

[0093] 3μm ≤ R(x) < 8μm: Light polishing;

[0094] R(x) ≥ 8μm: Medium to high intensity grinding treatment.

[0095] The corrugations are periodic ripple structures formed by the periodic action of the pantograph, and their presence can cause problems such as arc jumping and poor contact.

[0096] After acquiring Z(x) data using laser, this invention employs Fast Fourier Transform (FFT) to perform spectral analysis on the profile data:

[0097] Discretize the Z(x) data;

[0098] The FFT algorithm is used to transform it from the spatial domain to the frequency domain to obtain the spectrum function F(ω);

[0099] The peak amplitude within the dominant frequency band (e.g., 0.5~3Hz) is calculated and defined as B(x);

[0100] If B(x) is higher than the set threshold (e.g., 0.2 mm), it is determined that there is significant corrugation in the section.

[0101] This method can effectively identify different types of ripple morphology, distinguish between short-wave, high-frequency ripples and medium- and long-wave ripples, and provide a basis for subsequent adjustment of the grinding wheel structure.

[0102] The thickness of the oxide layer reflects the degree of chemical degradation on the material surface, which is difficult to accurately identify using traditional polishing methods. Therefore, this invention introduces multi-band infrared reflectance imaging and a modeling inversion algorithm to extract O(x).

[0103] The specific process is as follows:

[0104] An infrared hyperspectral camera is installed in the grinding device to acquire the multi-band reflectance spectrum R(λ,x) of the target section X_H, where λ is the wavelength;

[0105] Select 3 to 5 typical wavelength bands (such as 850nm, 950nm, 1050nm, etc.) as key reflection points;

[0106] Construct a model to fit the relationship between reflectivity and oxide layer thickness: Where f is a multivariate nonlinear regression function obtained from experimental fitting;

[0107] The O(x) value for each location point is obtained in real time based on this model, with the unit being μm.

[0108] The following are the preset polishing recommendations based on the oxide layer thickness:

[0109] O(x) < 1μm: The surface is basically free of oxidation;

[0110] 1μm ≤ O(x) < 5μm: Moderate polishing;

[0111] O(x) ≥ 5μm: requires intensive polishing and grinding.

[0112] The three parameters R(x), B(x), and O(x) mentioned above all have the characteristics of strong real-time performance, clear physical meaning, and continuous measurement. They can be processed locally and quickly by the device edge computing platform (such as FPGA+ARM architecture) to realize closed-loop feedback control.

[0113] The processing results will be used as input to the next stage of the grinding control decision module for the intelligent generation of control variables such as grinding wheel model matching, pressure adjustment, and speed distribution.

[0114] To ensure comparability of physical parameters with different dimensions, this step employs linear normalization to map roughness R(x), ripple amplitude B(x), and oxide thickness O(x) to the [0,1] interval. Specifically:

[0115] The normalization formula is: for any parameter Y(x)∈{R(x), B(x), O(x)}, its normalized value Y′(x) is calculated as follows: ; where Y_min and Y_max are the theoretical minimum and maximum values ​​of the corresponding parameters in the actual application scenario, respectively.

[0116] For example, for roughness R(x), we can set R_min = 0μm and R_max = 12μm; for ripple amplitude B(x), we can set B_min = 0mm and B_max = 0.6mm; for oxide thickness O(x), we can set O_min = 0μm and O_max = 10μm. After normalization, we construct a dimensionless damage vector D(x) = [R′(x), B′(x), O′(x)].

[0117] In order to transform the above normalized parameters into fuzzy linguistic variables, this invention defines three levels of membership functions for each input dimension, namely "mild", "moderate" and "severe" impairment.

[0118] This embodiment uses triangular or trapezoidal membership functions to construct fuzzy sets, which are simple in form, easy to calculate, and can clearly distinguish fuzzy levels.

[0119] A detailed function illustration (taking R′(x) as an example):

[0120] Mild: Trigonometric function, center at 0.2, base span 0~0.4;

[0121] Moderate: Trigonometric function, center at 0.5, base span 0.3~0.7;

[0122] Severity: Trigonometric function, center at 0.8, base span 0.6~1.0.

[0123] Example of a function expression (light membership function μ_Light(y)):

[0124] If y ≤ a, then μ_Light(y) = 0;

[0125] If a < y ≤ b, then μ_Light(y) = (y - a) / (b - a);

[0126] If b < y < c, then μ_Light(y) = (c - y) / (c - b);

[0127] If y ≥ c, then μ_Light(y) = 0;

[0128] Where a, b, and c are the coordinates of the inflection points of the function.

[0129] Each of the three parameters has its own independent set of membership functions, which together form a three-dimensional fuzzy input space.

[0130] To achieve comprehensive judgment among parameters, this invention sets up a set of rule matrix R_rule and weight coefficient vector W for fuzzy reasoning and weighted processing.

[0131] Weighting principle: Considering the different degrees of influence of various factors on the polishing process, a weight vector is set. ,in:

[0132] The weight of roughness R′(x);

[0133] The weight of the ripple amplitude B′(x);

[0134] The weight of the oxidation thickness O′(x);

[0135] The preferred setting is: = 0.4, = 0.4, = 0.2, ensuring the dominance of wear characteristics while taking into account the surface chemical state.

[0136] Example of fuzzy evaluation rules: Each input combination (e.g., R=severe, B=moderate, O=moderate) will be mapped to a damage level output (e.g., severe) through a rule table. Example rules are as follows:

[0137]

[0138] Each rule is used to generate a fuzzy inference result.

[0139] Output membership degree calculation: The fuzzy membership degrees μ_R, μ_B, and μ_O of each input dimension are combined with the weight W to obtain the comprehensive membership degree U(x) = [μ_mild(x), μ_moderate(x), μ_severe(x)]. To transform the fuzzy output vector U(x) into a quantifiable damage level score, this invention employs a centroid method (center method) defuzzification algorithm.

[0140] The severity level mapping value is defined as follows: the severity levels of "mild", "moderate", and "severe" are mapped to three values ​​of 0.2, 0.5, and 0.8, respectively.

[0141] The calculation formula is: V(x) = [μ_mild(x) × 0.2 + μ_moderate(x) × 0.5 + μ_severe(x) × 0.8] / [μ_mild(x) + μ_moderate(x) + μ_severe(x)]; the rating value V(x) ranges from 0 to 1, and the larger the value, the more severe the injury.

[0142] The scoring intervals are defined as follows: V(x) < 0.3: no polishing required; 0.3 ≤ V(x) < 0.6: light polishing; 0.6 ≤ V(x) < 0.8: moderate polishing; V(x) ≥ 0.8: heavy polishing. The scoring results are directly used as the input trigger conditions for the polishing control strategy.

[0143] The damage score V(x) is a continuous real number ranging from 0 to 1. To match the polishing control strategy, this invention divides it into four intervals, each interval mapping to a polishing level label L_g(x). The intervals are defined as follows:

[0144] V(x) < 0.3: No-grind required;

[0145] 0.3≤V(x)<0.6: Light polishing;

[0146] 0.6≤V(x)<0.8: Medium-intensity polishing;

[0147] V(x)≥0.8: Heavy polishing;

[0148] The polishing level label L_g(x) will be used as the main index field for control strategy invocation, thereby converting continuous state inputs into discrete control command trigger conditions.

[0149] To make grinding control decisions more structured and practical for engineering applications, this invention constructs a grinding strategy library. This library is organized by grinding level tags and stores the following three types of core parameter templates:

[0150] Grinding wheel type combination T_ref: indicates the recommended grinding wheel configuration for different grades, such as quantity, grit grade (120 mesh or 240 mesh), and arrangement (red wheel and green wheel combination).

[0151] Speed ​​range N_ref: indicates the recommended operating speed range of the motor (in revolutions per minute, RPM). For example, 600-800 RPM is recommended for low intensity, 800-1000 RPM for medium intensity, and 1000-1200 RPM for high intensity.

[0152] Contact pressure range P_ref: indicates the range of forces between the grinding wheel and the contact line (in Newtons N), typically 3~10N.

[0153] Each polishing grade label L_g(x) corresponds to a set of parameter templates: ;

[0154] The strategy library is stored in a structured data table format, which can be quickly accessed and updated through embedded systems or control software.

[0155] While standard template parameters can cover most situations, in actual operation, there are significant deviations in the physical state of certain contact line sections (such as high roughness but small ripples). In such cases, further fine-tuning is required based on the standard to improve control accuracy.

[0156] Therefore, based on the template parameters, this invention introduces a fine-tuning mechanism, using the three types of feature values ​​extracted in step S200 for correction. The specific method is as follows:

[0157] Define a fine-tuning factor vector W_m(x) = [w_r, w_b, w_o], representing the weights of roughness R(x), ripple amplitude B(x), and oxide thickness O(x) in the fine-tuning. A default value of [0.4, 0.4, 0.2] is recommended.

[0158] Correction algorithm: For N(x), perform linear weighted adjustment based on N_ref and roughness R(x): Where K_n is the adjustment coefficient (e.g., 20 RPM for every 1μm), and R_ref is the reference roughness (e.g., 6μm).

[0159] For P(x), adjust according to the ripple amplitude B(x): Where K_p is the adjustment coefficient (e.g., 1 N for every 0.1 mm), and B_ref is the ripple reference (e.g., 0.3 mm).

[0160] For the grinding wheel combination T(x), the wheel type ratio is adjusted according to the oxidation thickness O(x). For example, when O(x) ≥ 5μm, the proportion of red wheels is increased by 1 unit and the proportion of green wheels is decreased.

[0161] The above algorithm can run in real time in embedded systems, and combine floating-point sensor data to generate parameters, ensuring that the grinding operation has an adaptive response capability to different damage characteristics.

[0162] After the above strategy invocation and parameter fine-tuning, a complete set of control parameters is obtained: C(x) = [T(x), N(x), P(x)]; This parameter set is encapsulated in the system in a structured data frame format, including the following fields: x: current control point position; T(x): array of grinding wheel models, such as [green, green, red, red]; N(x): target motor speed (unit RPM); P(x): application pressure (unit N); timestamp and status identifier.

[0163] This parameter set will be transmitted to the drive module of the device execution layer, where it will be read and responded to in real time by actuators such as the grinding wheel motor controller and the pneumatic pressure regulating module.

[0164] Meanwhile, all generated C(x) will be cached in the control task queue to realize the dynamic loading and execution of the grinding trajectory, meeting the requirements of efficient continuous grinding operations.

[0165] First, before the equipment starts or during the grinding operation, the control system retrieves the generated control parameter set C(x) from the task cache. Wherein:

[0166] T(x): The combination of grinding wheel types required for the current grinding position, such as [green wheel, red wheel, red wheel, green wheel];

[0167] N(x): The target motor speed required at the current point, in RPM (revolutions per minute);

[0168] P(x): The target contact pressure value at the current point, in Newtons (N).

[0169] The control system uses an embedded parsing algorithm to parse the C(x) structure data into a multi-channel instruction set that can be recognized by the execution layer:

[0170] Target speed command for motor channel i: N_i_target = N(x) (used by the speed control module);

[0171] Pressure control unit target value: P_target = P(x) (for pneumatic / servo actuator modules);

[0172] Wheel configuration command: T_i_config = T(x)[i] (used for the grinding wheel selection controller);

[0173] All parameters will be uniformly encapsulated into standard data frames and sent to each execution module via CAN bus or RS-485 industrial protocol.

[0174] Since the grinding device moves along the x-direction, its operating state must be precisely matched with the current spatial coordinates. Therefore, this invention employs a high-precision incremental encoder or laser displacement sensor, in conjunction with a speed sensor, to construct a position synchronization system.

[0175] The specific process is as follows:

[0176] An incremental encoder is installed in the direction of travel of the grinding device to record the cumulative displacement Δx' in real time;

[0177] The control system maintains a set of control position indexes ;

[0178] When the actual displacement (δ represents tolerance, recommended not to exceed 5mm) When this condition is met, the system will automatically activate the corresponding... Parameter set;

[0179] Ensure that the execution of control signals is strictly synchronized with the spatial position to avoid misaligned grinding.

[0180] This mechanism ensures a precise mapping relationship between physical space and digital control, which is the foundation for realizing continuous variable parameter control.

[0181] The grinding device of the present invention generally includes multiple independently driven grinding wheels (e.g., 4), and each motor is controlled by closed-loop speed regulation through an independent drive channel.

[0182] The specific implementation steps are as follows:

[0183] The control system sends the N(x) value from C(x) to each motor driver;

[0184] The motor controller uses a PI (proportional-integral) closed-loop speed control algorithm to sample the real-time speed value;

[0185] Real-time rotational speed is provided by a Hall sensor or encoder, with accuracy error controlled within ±20 RPM;

[0186] If a deviation ΔN is found between the actual rotational speed N_real and the target rotational speed N(x), adjust as follows:

[0187] If ΔN > 20, then increase the voltage or pulse width;

[0188] If ΔN < -20, then reduce the control signal;

[0189] Until ΔN < ε (speed regulation error tolerance, preferably 10 RPM);

[0190] The multiple wheels allow for differentiated speed adjustments. For example, the red wheel is designed for fine polishing with a speed of 900 RPM, while the green wheel is designed for coarse polishing with a speed of 1200 RPM.

[0191] The aforementioned independent speed control mechanism enables different wheel sets on the same equipment to perform differentiated operations for different wear conditions, thereby improving grinding accuracy and equipment flexibility.

[0192] The pressure between the grinding wheel and the contact wire is a crucial factor affecting the grinding effect and equipment lifespan. This invention provides two methods for pressure control: an electric lead screw mechanism and a pneumatic servo cylinder.

[0193] The control process is as follows:

[0194] The control system reads the P(x) value in C(x) and converts it into the corresponding push rod displacement value S_p (unit mm) or cylinder target air pressure value P_air (unit bar).

[0195] If an electric lead screw is used, a servo motor is used to drive and control the displacement of the lead screw, so that the pressure arm pushes the grinding wheel to fit the contact line.

[0196] If pneumatic control is used, a proportional solenoid valve is used to regulate the air pressure to achieve pressure output on demand;

[0197] The real-time pressure feedback value is provided by a force sensor, and the system continuously monitors the actual pressure value P_real;

[0198] If the difference ΔP between P_real and P(x) is greater than 0.3N, the system triggers a correction action until ΔP ≤ 0.1N.

[0199] This closed-loop control system ensures that the mechanical parameters during the grinding process are maintained stably, preventing damage to the contact wire due to excessive pressure or insufficient grinding due to insufficient pressure.

[0200] This invention utilizes a multi-band reflectivity imaging sensor integrated at the tail of the polishing device to perform online optical sampling of the contact line and obtain the surface reflectivity F(x). The specific process is as follows:

[0201] The surface of the contact line is illuminated using a broadband LED light source with a wavelength range of 400~1100 nanometers;

[0202] The reflected signal is received by the imaging sensor of the integrated optical filter lens, and the reflection intensity value I_ref(x) at each location point is output.

[0203] The system performs illumination compensation and normalization on the raw reflectance data, and calculates the reflectance F(x): Where I_incident represents the incident intensity reference value under system calibration. Reflectivity F(x) reflects whether the surface oxide layer has been removed and the degree of metal exposure; the higher the reflectivity, the more thorough the polishing.

[0204] The recommended reference thresholds are as follows: F(x) ≥ 0.75: good reflection, quality qualified; F(x) < 0.75: there is still oxidation residue on the surface, and compensation polishing is required.

[0205] To evaluate surface micro-smoothness, this invention uses a non-contact laser displacement sensor to acquire micro-profile height data Z(x) along the contact line direction at the tail of the grinding device, and calculates the residual roughness Rr(x) according to the international standard ISO 4287. A lower roughness Rr(x) indicates a smoother surface and better grinding quality. Recommended roughness thresholds are as follows: Rr(x) ≤ 3 μm: good smoothness, acceptable quality; Rr(x) > 3 μm: residual grinding marks exist on the surface, requiring compensation grinding.

[0206] The quality detection module at the tail of the grinding device transmits the collected F(x) and Rr(x) data to the embedded control system. The system will synchronously bind the data with the position coordinate x and write it into the quality monitoring cache for the next step of decision-making.

[0207] The feedback process is carried out simultaneously with the polishing process, without interfering with the equipment's operating rhythm, ensuring real-time performance and high efficiency.

[0208] This invention constructs a dual-index logic judgment model to determine whether the current polishing quality meets the standard. The model logic is as follows:

[0209] Let: F_thr = 0.75 (reflectivity threshold); Rr_thr = 3 μm (roughness threshold);

[0210] For any position x, the judgment condition is: if F(x) ≥ F_thr and Rr(x) ≤ Rr_thr, then the grinding at this position is considered qualified and the next control point is entered; otherwise, the compensation grinding operation is triggered.

[0211] The decision function is expressed as:

[0212] Q(x) = 1, if F(x) ≥ F_thr and Rr(x) ≤ Rr_thr;

[0213] Q(x) = 0, otherwise.

[0214] When Q(x) = 0, the system will call the "compensation grinding subroutine", which controls the grinding device to return to the corresponding position x and perform an adjustable grinding action. The specific process is as follows:

[0215] Position retraction: Based on the encoder record, the drive system controls the device to retract to a position within a range of x ± 10mm;

[0216] Compensation parameter correction: The N(x) and P(x) parameters in the current control parameter set C(x) are increased by a certain proportion: rotational speed increases by 10-20%; pressure increases by 1-2 N; the grinding wheel combination T(x) remains unchanged to avoid excessive interference; Grinding execution: The dynamic grinding command is executed again through the control system; Re-detection: F(x) and Rr(x) are re-collected after the compensated grinding; Maximum execution times setting: Compensated grinding can be executed a maximum of 2 times. If it still fails, the system records an alarm and marks the location as a manual re-inspection point. This compensation logic effectively avoids situations where the initial grinding does not meet expectations, improving operational reliability.

[0217] Example 2, please refer to Figure 2 As shown in this embodiment, an adaptive grinding control system for rigid contact wires includes:

[0218] Data acquisition module: acquire historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located, and establish the contact wire damage index distribution curve L(x), where x is the position coordinate along the busbar direction;

[0219] High damage zone identification module: Based on the damage index distribution curve L(x), identify high damage zones and extract the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the zone;

[0220] Damage level determination module: Based on the extracted R(x), B(x), and O(x) parameters, construct the damage intensity vector D(x) = [R(x), B(x), O(x)], and use the fuzzy comprehensive evaluation method to calculate the damage level score V(x) for each x position;

[0221] Control parameter generation module: Based on the score V(x), match the preset grinding strategy library to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and generate the control parameter set C(x) = [T(x), N(x), P(x)];

[0222] Execution module: Controls the grinding device to perform variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction;

[0223] Feedback data acquisition module: acquires the surface reflectivity F(x) and residual roughness Rr(x) of the contact line after polishing, and feeds them back to the control system;

[0224] Threshold comparison module: Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

[0225] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for adaptive grinding control of rigid contact wires, characterized in that: include: Historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located are obtained, and a contact wire damage index distribution curve L(x) is established, where x is the position coordinate along the busbar direction; Based on the damage index distribution curve L(x), high damage intervals are identified, and the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the interval are extracted. Based on the extracted R(x), B(x), and O(x) parameters, a damage intensity vector D(x) = [R(x), B(x), O(x)] is constructed, and a fuzzy comprehensive evaluation method is used to calculate the damage level score V(x) for each x position; Based on the score V(x), a preset grinding strategy library is matched to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and a control parameter set C(x) = [T(x), N(x), P(x)] is generated; The control grinding device performs variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction; The reflectivity F(x) and residual roughness Rr(x) of the contact line surface after polishing are obtained and fed back to the control system. Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

2. The adaptive grinding control method for rigid contact wires according to claim 1, characterized in that: The establishment of the contact line damage index distribution curve L(x) includes: Collect current fluctuation values ​​and vertical impact acceleration data generated during the pantograph-catenary interaction process; The current fluctuation values ​​and impact acceleration data are normalized to construct a time series of pantograph-catenary action intensity, and then divided into different segments according to the time window. By synchronously registering the location information of the rail vehicle with the time series of pantograph-catenary action intensity, the spatial mapping relationship along the busbar direction x is obtained, and a location association dataset M(x) is constructed. Based on M(x), the cumulative impact index and energy dissipation parameters per unit length are calculated, fused into the contact line damage index L(x), and fitted as a continuous distribution curve.

3. The adaptive grinding control method for rigid contact wires according to claim 2, characterized in that: The identification of high-damage intervals based on the damage index distribution curve L(x) includes: A sliding window scanning analysis was performed on the contact wire damage index distribution curve L(x). An adaptive threshold discrimination method was used to identify continuous position segments where the L(x) value is greater than the set threshold L_thresh, which were taken as high damage intervals X_H. Within the high-damage range X_H, the longitudinal profile height variation data of the contact line surface is obtained, and the roughness parameter R(x) is extracted by spatial spectrum analysis, where R(x) represents the average height deviation per unit length. Fourier transform is performed on the contact line profile data to extract the peak amplitude of the ripple characteristics in the main frequency band, which is denoted as ripple amplitude B(x), and this is used to identify the sections with medium and long ripples. Within the high-damage range X_H, spectral reflectance data of the contact line surface are collected, and the oxide layer thickness O(x) is calculated by combining the reflectance-oxide layer thickness fitting model.

4. The adaptive grinding control method for rigid contact wires according to claim 3, characterized in that: The method of calculating the damage level score V(x) for each x location using fuzzy comprehensive evaluation includes: The extracted contact line roughness R(x), ripple amplitude B(x), and oxide thickness O(x) are normalized and mapped to the [0,1] interval to construct a dimensionless damage intensity vector D(x) = [R'(x), B'(x), O'(x)]; Three membership function sets for the three damage factors are defined, and fuzzy membership functions of "mild", "moderate" and "severe" are established for R'(x), B'(x) and O'(x) respectively, using triangular or trapezoidal function construction methods. Based on the fuzzy evaluation rule matrix, the membership degrees of each term of D(x) are input into the weighted fuzzy inference system, and the comprehensive membership vector U(x) is calculated by combining the preset weight factors. The centroid method is used to defuzzify the comprehensive membership vector U(x) and output the final damage level score V(x) at the corresponding position x, which is used to quantify the comprehensive damage degree of the contact line.

5. The adaptive grinding control method for rigid contact wires according to claim 4, characterized in that: The step of matching the score V(x) with a preset grinding strategy library to determine the grinding wheel type combination T(x), rotational speed N(x), and contact pressure P(x) at the corresponding position x includes: Based on the obtained damage level score V(x), interval matching is performed in the polishing strategy level table to assign V(x) to one of the preset polishing level intervals, including four level labels L_g(x): "no polishing", "low intensity polishing", "medium intensity polishing" and "high intensity polishing". Based on the grinding grade label L_g(x), the corresponding grinding parameter template is retrieved from the grinding strategy library stored in the control system to obtain the standard grinding wheel combination T_ref, recommended speed range N_ref, and contact pressure range P_ref for that grade. By combining the fine-tuning weights of the extracted roughness R(x), ripple amplitude B(x), and oxide thickness O(x), the standard parameters T_ref, N_ref, and P_ref are locally personalized through the interpolation correction algorithm to generate a refined grinding parameter combination T(x), N(x), and P(x); The T(x), N(x), and P(x) are integrated into a complete set of control parameters C(x) = [T(x), N(x), P(x)] and stored in the task queue.

6. The adaptive grinding control method for rigid contact wires according to claim 5, characterized in that: The controlled polishing device performs variable parameter dynamic polishing operations according to the control parameter set C(x), including: The multi-axis actuator in the control grinding device receives the control parameter set C(x), analyzes the target values ​​of grinding wheel type T(x), rotation speed N(x) and contact pressure P(x) in real time according to the change of grinding position x, and generates a multi-channel servo control instruction set for the distributed drive module; During the movement of the grinding device along the busbar direction, the current displacement coordinate x' is obtained by a position synchronous encoder and matched with the target coordinate x in the control parameter set; The embedded controller enables independent speed adjustment of each grinding motor, allowing multiple grinding wheels to achieve differentiated speed adjustment based on the N(x) value of their respective positions x. The contact pressure between the grinding wheel and the contact line is controlled by an electric lead screw or pneumatic pressure regulating module, and the wheel arm clamping force is dynamically adjusted according to the P(x) value of the current position point x.

7. A rigid contact wire adaptive grinding control system, used to implement the rigid contact wire adaptive grinding control method according to any one of claims 1-6, characterized in that: include: Data acquisition module: acquire historical data of pantograph-catenary action at various positions on the rigid contact wire busbar where the grinding device is located, and establish the contact wire damage index distribution curve L(x), where x is the position coordinate along the busbar direction; High damage zone identification module: Based on the damage index distribution curve L(x), identify high damage zones and extract the roughness R(x), ripple amplitude B(x), and oxide thickness O(x) state parameters of the contact line within the zone; Damage level determination module: Based on the extracted R(x), B(x), and O(x) parameters, construct the damage intensity vector D(x) = [R(x), B(x), O(x)], and use the fuzzy comprehensive evaluation method to calculate the damage level score V(x) for each x position; Control parameter generation module: Based on the score V(x), match the preset grinding strategy library to determine the grinding wheel type combination T(x), rotation speed N(x), and contact pressure P(x) at the corresponding position x, and generate the control parameter set C(x) = [T(x), N(x), P(x)]; Execution module: Controls the grinding device to perform variable parameter dynamic grinding operation according to the control parameter set C(x), so that the grinding wheel adaptively adjusts its working state as it moves along the x direction; Feedback data acquisition module: acquires the surface reflectivity F(x) and residual roughness Rr(x) of the contact line after polishing, and feeds them back to the control system; Threshold comparison module: Determine whether F(x) and Rr(x) meet the preset quality threshold. If not, automatically perform compensation grinding at the corresponding position x; otherwise, continue grinding the next position.

Images