Automatic adjustment and fault diagnosis method and system for intelligent belt cleaner

Abstract

The invention relates to an automatic adjustment and fault diagnosis method and system for an intelligent belt sweeper, and the method comprises the steps: scanning the surface of a scraper through a multispectral laser array, separating a reflected signal, filtering the interference of ambient light, and generating original point cloud data representing a wear contour; performing deformation error correction on the point cloud data based on a dynamic curvature compensation algorithm to generate thickness distribution in a non-uniform wear state; the fuzzy PID control technology is adopted to dynamically distribute pressure on the two sides of the scraper, and the abrasion rate is balanced; fusing wear thickness distribution and pressure adjustment data, and judging and outputting an eccentric wear fault code and an adjustment deviation value through a dynamic threshold value; and carrying out fusion prediction on the multi-source heterogeneous data in combination with a transfer learning technology, and outputting the residual life. The problems of measurement distortion, adjustment lag and low maintenance efficiency caused by attachment interference, deformation errors and data splitting are solved, and accurate monitoring, dynamic adjustment and predictive maintenance of the abrasion state of the scraper can be achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent sweeper technology, specifically to an automatic adjustment and fault diagnosis method and system for an intelligent belt sweeper. Background Technology

[0002] In industrial material conveying systems, belt cleaners are key components whose core function is to remove residual material adhering to the conveyor belt, preventing equipment wear and operational failures caused by material accumulation. Traditional belt cleaners typically rely on manual inspection and experience-based adjustments, periodically stopping the machine to check the wear condition of the scrapers and adjusting the cleaning pressure according to fixed thresholds. However, as conveying systems develop towards higher speeds and continuous operation, the shortcomings of traditional methods in dynamic wear monitoring and real-time adjustment capabilities have become increasingly apparent. Especially in complex operating conditions such as high humidity and dusty environments, they struggle to effectively address issues such as uneven scraper wear and lag in pressure regulation, leading to shortened equipment lifespan and increased maintenance costs.

[0003] While existing wear detection schemes based on single sensors such as laser ranging or pressure feedback can achieve basic condition monitoring, they have the following limitations: First, interference from adhering substances and dynamic deformation on the scraper surface can easily lead to measurement errors, making it difficult to accurately reflect the true wear thickness. Second, fixed parameter adjustment strategies cannot adapt to the difference in wear rates on both sides, exacerbating uneven wear. Third, the lack of integrated analysis of historical data and real-time conditions means that fault warnings and life predictions rely on human experience, posing a risk of misjudgment. Furthermore, the fragmented processing of data dimensions such as wear morphology, pressure deviation, and fault characteristics in traditional methods further limits the efficiency of generating intelligent maintenance decisions. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide an intelligent belt sweeper automatic adjustment and fault diagnosis method that can sense the multi-dimensional wear state of the scraper in real time, dynamically optimize adjustment parameters, and realize predictive maintenance.

[0005] The objective of this invention is achieved through the following solution:

[0006] In a first aspect, the present invention provides an automatic adjustment and fault diagnosis method for an intelligent belt sweeper, comprising the following steps:

[0007] S1: The surface of the scraper of the smart belt cleaner is scanned by a multispectral laser array integrated on the bracket of the smart belt cleaner, and the reflection signals of different wavelengths are separated to eliminate ambient light interference and generate raw point cloud data. The raw point cloud data is used to characterize the surface contour of the scraper.

[0008] S2: Perform dynamic curvature compensation processing on the original point cloud data, dynamically adjust the wear depth calculation according to the change of the scraper surface shape, and generate the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state.

[0009] S3: Based on fuzzy PID control technology, dynamic pressure weight allocation is performed on the wear thickness distribution. The wear compensation pressure difference between the two sides of the scraper is calculated in real time, an asymmetric pressure adjustment command is generated, and the asymmetric pressure adjustment command is sent to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment command is used to control the intelligent belt sweeper to balance the wear rate on both sides.

[0010] S4: Based on the feedback-based update of wear thickness distribution and asymmetric pressure regulation commands, perform dynamic wear rate fusion processing and threshold judgment to identify unilateral excessive wear anomalies, output the wear fault code that triggers maintenance alarm and the compensation pressure deviation value to evaluate the regulation effect;

[0011] S5: Based on transfer learning technology, multi-source data fusion processing is performed on the wear fault code, updated wear thickness distribution and compensation pressure deviation value to generate and send wear prediction results. The wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

[0012] In one embodiment, S2 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0013] S21: Based on wavelength division filtering technology, the original point cloud data is processed to remove ambient light interference, and near-infrared and short-wave infrared reflection signals are separated to generate effective reflection point data. The effective reflection point data is used to extract the effective contour of the scraper surface.

[0014] S22: Based on the dynamic curvature compensation algorithm, the effective reflection point data is subjected to non-uniform B-spline surface fitting processing, and the wear compensation depth is generated by combining the viscoelastic parameters of the scraper material and the belt speed feedback value.

[0015] S23: Based on data fusion technology, the compensation wear depth and the original point cloud data are weighted and superimposed to generate the wear thickness distribution.

[0016] In one embodiment, the compensation wear depth is calculated as follows:

[0017]

[0018] in, To compensate for the depth of wear, The viscoelastic coefficient, As an edge wear acceleration factor, The radius of curvature of the scraper surface is calculated using a non-uniform B-spline surface fitting algorithm. Let E be the belt speed, E be the elastic modulus of the scraper material, w be the scraper surface displacement in the original point cloud data, and x be the scraper position in the original point cloud data. is the time interval for the k-th scan.

[0019] In one embodiment, S1 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0020] S11: Based on the multispectral laser array integrated on the bracket of the intelligent belt sweeper, the scraper surface is simultaneously scanned in multiple bands to separate near-infrared and short-wave infrared reflection signals, penetrate the scraper surface attachments and enhance the detection of low reflectivity areas, and generate multispectral scanning data.

[0021] S12: Perform wavelength frequency division filtering on the multispectral scanning data to separate the effective reflection signal from the environmental noise and generate filtered point cloud data;

[0022] S13: Perform polarization filtering on the filtered point cloud data to remove specular reflection interference from the metal surface and generate the original point cloud data.

[0023] In one embodiment, S3 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0024] S31: The wear thickness distribution is calculated based on the fuzzy PID algorithm, the difference between the maximum wear zone thickness and the preset safety thickness is calculated, and the unilateral pressure increment is generated.

[0025] S32: Based on the single-sided pressure increment, the dynamic weight coefficient allocation is performed on both sides of the scraper. The weight coefficient is assigned to the severely worn side to generate the initial asymmetric pressure adjustment command.

[0026] S33: Based on the thin-film pressure sensor, the actual execution pressure of the initial asymmetric pressure regulation command is monitored and processed in real time. When the deviation between the actual pressure and the target pressure exceeds the preset threshold, the PID parameter self-correction algorithm is triggered to generate the final asymmetric pressure regulation command.

[0027] In one embodiment, S4 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0028] S41: Based on the updated wear thickness distribution and asymmetric pressure regulation command, spatiotemporal alignment processing is performed, and the correlation data between wear rate and pressure regulation are fused to generate a dynamic wear rate matrix;

[0029] S42: Perform anomaly detection processing on the dynamic wear rate matrix. When the wear rate on one side exceeds the safety threshold range and the compensation pressure deviation continues to increase, it is determined to be an irreversible uneven wear anomaly, and the anomaly determination result is output.

[0030] S43: Based on the determination result of irreversible uneven wear, match the fault characteristics in the preset fault type database to generate a unique uneven wear fault code. The uneven wear fault code is used to associate the solution in the maintenance knowledge base.

[0031] S44: Based on the dynamic difference between the pressure regulation data and the target pressure in the dynamic wear rate matrix, a compensation pressure deviation value is generated by calculating using the sliding window mean algorithm.

[0032] In one embodiment, S5 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0033] S51: Based on heterogeneous data fusion technology, the wear fault codes, updated wear thickness distribution and compensation pressure deviation values ​​are standardized and encoded. Discrete fault codes are converted into feature vectors, continuous wear data are converted into grayscale matrices, and dynamic deviation values ​​are normalized into scalar sequences to generate a fault input vector in a unified format.

[0034] S52: Based on a pre-trained neural network model, multi-source feature extraction processing is performed on the fault input vector. By freezing the parameters of the underlying network and fine-tuning the fully connected layer, the wear prediction task under the current working condition is adapted to generate a fused feature tensor.

[0035] S53: Perform spatiotemporal sequence prediction processing on the fused feature tensor to generate and send wear prediction results containing scraper wear status and remaining lifespan.

[0036] Secondly, the present invention provides an intelligent belt sweeper automatic adjustment and fault diagnosis system, which is configured with the following modules:

[0037] The data acquisition and processing module is used to scan the scraper surface of the smart belt cleaner using a multispectral laser array integrated on the bracket of the smart belt cleaner, separate the reflected signals of different wavelengths to eliminate ambient light interference, and generate raw point cloud data. The raw point cloud data is used to characterize the scraper surface contour.

[0038] The wear thickness calculation module is used to perform dynamic curvature compensation processing on the original point cloud data. It dynamically adjusts the wear depth calculation according to the change of the scraper surface shape and generates the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state.

[0039] The adjustment command generation module is used to perform dynamic pressure weight allocation processing on the wear thickness distribution based on fuzzy PID control technology, calculate the wear compensation pressure difference on both sides of the scraper in real time, generate asymmetric pressure adjustment commands, and send the asymmetric pressure adjustment commands to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment commands are used to control the intelligent belt sweeper to balance the wear rate on both sides.

[0040] The fault diagnosis module is used to perform dynamic wear rate fusion processing and threshold judgment based on the feedback-based updated wear thickness distribution and asymmetric pressure adjustment commands, identify unilateral excessive wear anomalies, and output the wear-bias fault code that triggers maintenance alarms and the compensation pressure deviation value for evaluating the adjustment effect.

[0041] The life prediction module is used to perform multi-source data fusion processing on wear fault codes, updated wear thickness distribution and compensation pressure deviation values ​​based on transfer learning technology, generate and send wear prediction results, and the wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

[0042] Thirdly, this application provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the above-mentioned methods for automatic adjustment and fault diagnosis of intelligent belt sweepers.

[0043] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-mentioned methods for automatic adjustment and fault diagnosis of intelligent belt sweepers.

[0044] In summary, the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by this invention combines multispectral laser array scanning with dynamic curvature compensation, which can effectively eliminate interference from scraper surface deposits and measurement errors caused by dynamic deformation, achieving non-contact high-precision wear thickness detection. Based on the dynamic pressure weight allocation technology of fuzzy PID control, it generates asymmetric adjustment commands by sensing the wear difference on both sides of the scraper in real time, thereby achieving adaptive balance control of cleaning pressure and avoiding the problem of uneven wear accumulation under the traditional fixed threshold adjustment strategy. By fusing updated wear thickness distribution and pressure adjustment feedback data, a dynamic wear rate matrix is ​​constructed and multi-dimensional threshold judgment is performed, which can accurately identify unilateral excessive wear anomalies and simultaneously output uneven wear fault codes and compensation pressure deviation values, providing a quantitative basis for maintenance decisions. Furthermore, by combining transfer learning-driven multi-source data fusion processing, discrete fault codes, continuous wear data, and dynamic deviation values ​​are mapped into a unified feature space. Based on cross-scenario knowledge transfer and local parameter fine-tuning, it can realize the prediction of remaining life and generation of graded maintenance strategies under small sample working conditions, thereby supporting the construction of a predictive maintenance system. The overall solution forms a closed-loop technical path of "precise perception - dynamic adjustment - intelligent diagnosis - predictive decision-making", which significantly improves the reliability of condition monitoring, real-time adjustment response and scientific maintenance decision-making of belt sweepers in complex industrial scenarios.

[0045] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0046] Figure 1 A flowchart illustrating an automatic adjustment and fault diagnosis method for an intelligent belt sweeper provided in an embodiment of this application;

[0047] Figure 2 A schematic diagram of the process for generating the wear thickness distribution at various locations of the scraper, provided in an embodiment of this application;

[0048] Figure 3 This is a schematic diagram of the structure of an intelligent belt sweeper automatic adjustment and fault diagnosis system provided in another embodiment of this application. Detailed Implementation

[0049] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0051] In one embodiment, such as Figure 1 As shown, an automatic adjustment and fault diagnosis method for an intelligent belt sweeper is provided. This embodiment illustrates the method applied to a terminal. It is understood that this method can also be applied to a server, and to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0052] S1: The surface of the scraper of the smart belt sweeper is scanned by a multispectral laser array integrated on the bracket of the smart belt sweeper. The reflected signals of different wavelengths are separated to eliminate ambient light interference and generate raw point cloud data. The raw point cloud data is used to characterize the surface contour of the scraper.

[0053] Specifically, the intelligent belt sweeper integrates a high-precision multispectral laser array system on its bracket. This array is equipped with laser emitters of multiple wavelengths and high-sensitivity receivers. The multispectral laser array comprehensively covers the scraper surface using a specific scanning pattern. The emitted lasers of different wavelengths are reflected off the scraper surface and captured by the receiver. The system uses a built-in signal processing algorithm to perform fine wavelength separation on the returned reflected signals, effectively filtering out interference from ambient light and ensuring the accuracy of the reflected signals. These processed signals are converted into high-precision raw point cloud data, which contains the three-dimensional coordinate information of each sampling point on the scraper surface, covering the microscopic texture features of the scraper. The system uses this rich three-dimensional data to construct a detailed contour model of the scraper surface, laying a solid data foundation for subsequent wear analysis. This model not only clearly shows the overall shape of the scraper but also accurately reflects subtle changes on the scraper surface, providing a precise initial state record for monitoring the wear of the scraper.

[0054] S2: Perform dynamic curvature compensation processing on the original point cloud data, dynamically adjust the wear depth calculation according to the change of the scraper surface shape, and generate the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state.

[0055] Specifically, during the operation of the intelligent belt sweeper, the system performs in-depth and complex dynamic curvature compensation processing on the raw point cloud data. This process fully considers the surface shape changes that may occur on the scraper during long-term use, such as curvature changes caused by local wear due to material impact and compression, as well as minor deformations that may occur overall. Preferably, the system can employ a dynamic curvature compensation algorithm to analyze and process the point cloud data in real time, dynamically adjusting the calculation method for wear depth. By accurately calculating the wear depth at different locations on the scraper surface, the system can accurately reconstruct the actual wear thickness at each location on the scraper. During the calculation process, the algorithm comprehensively considers multi-dimensional information such as the scraper's material characteristics, working environment factors, and historical wear data to ensure the accuracy and reliability of the calculation results. Finally, the system generates a wear thickness distribution map at each location on the scraper, intuitively displaying the non-uniform wear state of the scraper surface, clearly identifying severely worn areas and lightly worn areas, providing a crucial basis for subsequent precise adjustments. Wear thickness distribution map not only reflects the current wear condition of the scraper, but also provides important reference data for predicting future wear trends, enabling maintenance personnel to formulate corresponding maintenance plans in advance and effectively prevent equipment failures.

[0056] S3: Based on fuzzy PID control technology, dynamic pressure weight allocation is performed on the wear thickness distribution. The wear compensation pressure difference between the two sides of the scraper is calculated in real time, and an asymmetric pressure adjustment command is generated. The asymmetric pressure adjustment command is sent to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment command is used to control the intelligent belt sweeper to balance the wear rate on both sides.

[0057] Specifically, fuzzy PID control technology is an advanced control strategy that combines fuzzy logic reasoning with traditional PID control algorithms, effectively handling uncertainties and nonlinear factors in the system. Specifically, the system first divides the scraper surface into multiple small control regions based on the wear thickness distribution map, each region corresponding to a different degree of wear. Then, the system uses a fuzzy logic rule base to fuzzify the wear thickness of each region, mapping it to predefined fuzzy sets such as "slight wear," "moderate wear," and "severe wear." Next, based on these fuzzy sets and combined with the proportional, integral, and derivative parameters in the PID control algorithm, the system dynamically calculates the pressure weight for each control region. These weights reflect the pressure adjustment range required for different regions during wear compensation. The system integrates the pressure weights of all control regions in real time to calculate the wear compensation pressure difference between the two sides of the scraper.

[0058] To ensure the accuracy of the calculation results, the system employs a redundant calculation mechanism to iteratively calculate and verify the pressure difference multiple times. Ultimately, the system generates an asymmetric pressure adjustment command, which specifies in detail the pressure value and direction to be adjusted by the hydraulic or pneumatic actuators on both sides of the scraper. The system rapidly transmits these commands to the control terminal of the intelligent belt sweeper via a high-speed communication interface. Upon receiving the command, the control terminal immediately drives the corresponding actuator to precisely adjust the pressure on both sides of the scraper, achieving a dynamic balance in the wear rate on both sides. This process not only effectively reduces uneven wear on the scraper and extends its service life but also ensures that the belt sweeper maintains high-efficiency cleaning performance throughout its entire working cycle.

[0059] S4: Based on the feedback-based update of wear thickness distribution and asymmetric pressure regulation commands, perform dynamic wear rate fusion processing and threshold judgment to identify unilateral excessive wear anomalies, output the wear fault code that triggers maintenance alarm and the compensation pressure deviation value to evaluate the regulation effect.

[0060] Specifically, the system compares and analyzes newly generated wear thickness distribution data with historical data, and dynamically tracks the changing trend of scraper wear rate by establishing a time series model. Simultaneously, the system incorporates the execution feedback information of asymmetric pressure regulation commands into the analysis to assess the actual impact of pressure regulation measures on the wear rate. During the dynamic wear rate fusion processing, the system can employ a multi-dimensional data fusion algorithm to comprehensively consider various factors such as the wear characteristics of the scraper material, the pressure regulation range, belt speed, and material properties, constructing a comprehensive wear rate evaluation model.

[0061] Based on this model, the system can calculate the current wear rate of the scraper in real time and compare it with a preset safety threshold. When the wear rate on one side of the scraper exceeds the set threshold, indicating abnormal unilateral excessive wear, the system initiates a fault diagnosis program. Through in-depth analysis of the wear characteristics and pressure regulation history of the abnormal area, the system accurately identifies the specific cause of the uneven wear fault, such as improper pressure regulation, uneven material distribution, or scraper installation deviation. Subsequently, the system outputs an uneven wear fault code that triggers a maintenance alarm according to predefined fault code generation rules. These codes are unique and resolvable, allowing maintenance personnel to quickly understand the fault type and severity by consulting the fault manual provided by the system. Simultaneously, the system calculates and outputs a compensation pressure deviation value, which intuitively reflects the difference between the current pressure regulation and the ideal state, providing a quantitative basis for maintenance personnel to evaluate the adjustment effect. Based on this information, maintenance personnel can take timely and precise maintenance measures, such as adjusting pressure parameters, clearing blockages, or recalibrating the scraper position, thereby effectively preventing further deterioration of equipment failure and ensuring the stable operation of the conveying system.

[0062] S5: Based on transfer learning technology, multi-source data fusion processing is performed on the wear fault code, updated wear thickness distribution and compensation pressure deviation value to generate and send wear prediction results. The wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

[0063] Specifically, transfer learning technology effectively transfers historical data and existing knowledge from the system to the current wear prediction model, thereby improving the model's accuracy and adaptability. During data fusion, the system first performs standardized preprocessing on various types of data to eliminate differences in units and formats between different data sources. Then, the system uses feature extraction algorithms to mine key feature information from the data, such as typical wear patterns corresponding to fault codes, key regional features in the wear thickness distribution, and the changing trends of compensation pressure deviation values. This feature information is integrated into a multi-dimensional feature vector, which serves as the input to the transfer learning model. The system maintains a vast historical data warehouse containing a large amount of labeled wear case data, including wear thickness distribution under different operating conditions, fault codes, and corresponding equipment operating status. The transfer learning model is trained on this historical data to learn the inherent laws and characteristic patterns of scraper wear. When processing current data, the model can transfer this learned knowledge to new scenarios to accurately assess the current scraper wear state.

[0064] Based on the model's output, the system generates wear prediction results. These results include not only an assessment of the scraper's current wear level, such as wear percentage and remaining effective thickness, but also a prediction of its remaining lifespan. The remaining lifespan prediction employs probabilistic statistical methods, combining the scraper's current wear rate, material fatigue characteristics, and future operating condition predictions to provide a reasonable expected lifespan range, presented to the user in the form of visual charts. The system transmits these wear prediction results to user terminal devices, such as computers in the monitoring center or mobile terminals of maintenance personnel, in a timely manner via a high-speed communication network. Based on these predictions, users can develop scientifically sound maintenance plans in advance, rationally allocate equipment repair time and resources, minimize unplanned downtime, improve production efficiency, and ensure that the equipment is always in a safe and reliable operating state.

[0065] In summary, the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by this invention combines multispectral laser array scanning with dynamic curvature compensation, which can effectively eliminate interference from scraper surface deposits and measurement errors caused by dynamic deformation, achieving non-contact high-precision wear thickness detection. Based on the dynamic pressure weight allocation technology of fuzzy PID control, it generates asymmetric adjustment commands by sensing the wear difference on both sides of the scraper in real time, thereby achieving adaptive balance control of cleaning pressure and avoiding the problem of uneven wear accumulation under the traditional fixed threshold adjustment strategy. By fusing updated wear thickness distribution and pressure adjustment feedback data, a dynamic wear rate matrix is ​​constructed and multi-dimensional threshold judgment is performed, which can accurately identify unilateral excessive wear anomalies and simultaneously output uneven wear fault codes and compensation pressure deviation values, providing a quantitative basis for maintenance decisions. Furthermore, by combining transfer learning-driven multi-source data fusion processing, discrete fault codes, continuous wear data, and dynamic deviation values ​​are mapped into a unified feature space. Based on cross-scenario knowledge transfer and local parameter fine-tuning, it can realize the prediction of remaining life and generation of graded maintenance strategies under small sample working conditions, thereby supporting the construction of a predictive maintenance system. The overall solution forms a closed-loop technical path of "precise perception - dynamic adjustment - intelligent diagnosis - predictive decision-making", which significantly improves the reliability of condition monitoring, real-time adjustment response and scientific maintenance decision-making of belt sweepers in complex industrial scenarios.

[0066] In one embodiment, S2 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0067] S21: Based on wavelength division filtering technology, the original point cloud data is processed to remove ambient light interference, and near-infrared and short-wave infrared reflection signals are separated to generate effective reflection point data. The effective reflection point data is used to extract the effective contour of the scraper surface.

[0068] Specifically, the raw point cloud data contains reflection signals of various wavelengths, including not only useful signals from the near-infrared and short-wave infrared bands, but also ambient light interference signals of other wavelengths, such as stray light in the visible light range. The core of wavelength division filtering technology lies in the precise frequency domain segmentation and filtering of signals of different wavelengths. The system pre-sets filtering parameters for the near-infrared and short-wave infrared bands, and uses mathematical tools such as Fourier transform to convert the raw point cloud data from the time domain to the frequency domain, thereby clearly distinguishing the signal amplitude and phase information corresponding to each wavelength component. Subsequently, the system uses a pre-designed bandpass filter to allow only near-infrared and short-wave infrared signals to pass through, while filtering out ambient light interference signals of other wavelengths, effectively separating the useful reflection signals and generating effective reflection point data, laying a solid foundation for subsequent extraction of the effective contour of the scraper surface. In this process, the system fully considers the differences in the characteristics of different wavelength signals during transmission and reflection, as well as the complexity and variability of ambient light interference, ensuring the accuracy and reliability of the effective reflection point data.

[0069] S22: Based on the dynamic curvature compensation algorithm, the effective reflection point data is subjected to non-uniform B-spline surface fitting processing. Combined with the viscoelastic parameters of the scraper material and the belt speed feedback value, the compensation wear depth is generated.

[0070] Specifically, the calculation method for the compensation wear depth is as follows:

[0071]

[0072] in, To compensate for the depth of wear, The viscoelastic coefficient, As an edge wear acceleration factor, The radius of curvature of the scraper surface is calculated using a non-uniform B-spline surface fitting algorithm. Let E be the belt speed, E be the elastic modulus of the scraper material, w be the scraper surface displacement in the original point cloud data, and x be the scraper position in the original point cloud data. is the time interval for the k-th scan.

[0073] After acquiring valid reflection point data, the system first processes it using a non-uniform B-spline surface fitting algorithm. This algorithm dynamically adjusts the position and weight of control points based on the distribution characteristics of the valid reflection point data, thus accurately constructing a surface model of the scraper. Building upon this, the system incorporates the viscoelastic parameters of the scraper material, which reflect its deformation and recovery characteristics under stress, as well as the belt speed feedback value. Since belt speed directly affects the relative motion and wear between the scraper and the belt, a dynamic curvature compensation algorithm is used to further optimize and adjust the fitted surface to more realistically reflect the actual wear state of the scraper surface. This dynamic curvature compensation algorithm comprehensively considers the curvature change of the scraper surface, the viscoelastic effect of the material, and the influence of belt speed on wear, thereby generating a compensated wear depth that accurately quantifies the degree of scraper wear, providing crucial data support for subsequent wear analysis and compensation adjustments.

[0074] S23: Based on data fusion technology, the compensation wear depth and the original point cloud data are weighted and superimposed to generate the wear thickness distribution.

[0075] Specifically, data fusion technology fully utilizes the scraper surface wear information contained in the compensated wear depth and the rich geometric details provided by the original point cloud data. By rationally allocating weights, it achieves a weighted superposition of the two. During the weighted superposition process, the system assigns corresponding weight coefficients to each based on the reliability and importance of the compensated wear depth, and the advantages of the original point cloud data in reflecting the geometry of the scraper surface. These weight coefficients are determined based on extensive experimental data and the system's self-learning capabilities to ensure that the fused result can reproduce the true wear condition of the scraper surface to the greatest extent possible. After weighted superposition, the system finally generates a wear thickness distribution. This wear thickness distribution presents the wear thickness variations in different areas of the scraper surface in an intuitive and visual form, clearly reflecting the wear morphology and severity of the scraper, providing a comprehensive and accurate basis for subsequent fault diagnosis, wear compensation, and maintenance decisions. In this way, the system can more accurately grasp the wear state of the scraper, promptly identify potential wear problems, and thus take effective measures to extend the scraper's service life and ensure the stable operation of the belt conveyor system.

[0076] The aforementioned intelligent belt sweeper automatic adjustment and fault diagnosis method uses wavelength frequency division filtering technology to process the original point cloud data generated by multispectral laser scanning, effectively separating near-infrared and short-wave infrared reflection signals and filtering out ambient light and specular reflection interference from metal surfaces to achieve high-precision extraction of the effective contour of the scraper surface. Based on the dynamic curvature compensation algorithm, non-uniform B-spline surface fitting is performed on the effective reflection point data. Combined with the viscoelastic properties of the scraper material and real-time belt speed feedback, the measurement error caused by material deformation and motion inertia can be dynamically corrected to generate compensated wear depth data. Furthermore, through data fusion technology, the compensated wear depth and the original laser measurement value are weighted and superimposed to achieve the complementary advantages of dynamic error correction and static measurement data, generating a thickness distribution heat map that reflects the true wear state. This process overcomes the measurement distortion problems caused by environmental interference, dynamic deformation and data uniformity in traditional single-point laser ranging technology through multi-dimensional data collaborative processing. It provides high-confidence basic data support for subsequent wear rate analysis, pressure regulation and life prediction, and significantly improves the robustness and accuracy of scraper wear status perception under complex working conditions.

[0077] In one embodiment, S1 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0078] S11: Based on the multispectral laser array integrated on the bracket of the intelligent belt sweeper, the surface of the scraper is scanned in multiple bands simultaneously to separate near-infrared and short-wave infrared reflection signals, penetrate the attachments on the scraper surface and enhance the detection of low reflectivity areas, and generate multispectral scanning data.

[0079] Specifically, the multispectral laser array consists of multiple high-precision laser emitters capable of synchronously emitting lasers in different wavelengths, including near-infrared and short-wave infrared bands. Near-infrared lasers possess strong penetrating power, allowing them to penetrate adhering substances such as dust and moisture on the scraper surface, thus obtaining the true reflection signal from the scraper surface. Short-wave infrared lasers are highly sensitive to materials with low reflectivity, enhancing the detection capability in these areas and ensuring the system acquires accurate detection data when facing scrapers of different materials and surface conditions. Synchronous scanning processing of the multispectral laser array is achieved by precisely controlling the emission frequency and phase of each laser emitter, ensuring comprehensive and high-quality multispectral scanning data is acquired during the scanning process. This data includes the reflection characteristics of the scraper surface at different wavelengths, providing a rich information foundation for subsequent signal processing. During scanning, the system also dynamically adjusts the laser emission intensity and scanning speed according to the scraper's movement and surface characteristics to ensure the accuracy and completeness of the scanning data.

[0080] S12: Perform wavelength frequency division filtering on the multispectral scanning data to separate the effective reflection signal from the environmental noise and generate filtered point cloud data.

[0081] Specifically, the system performs frequency domain transformation on the multispectral scanning data, decomposing the complex mixed signal into different frequency components. By analyzing the characteristics of each frequency component, it identifies frequency bands corresponding to environmental noise, such as interference light from surrounding lighting equipment and electromagnetic interference from electronic devices. For these noise frequency bands, the system applies band-stop filters to suppress environmental noise interference. During the filtering process, the system fully considers the characteristics of different wavelength laser signals. For near-infrared and short-wave infrared reflected signals, the system can employ adaptive filtering algorithms, dynamically adjusting filtering parameters based on signal intensity, frequency, and phase characteristics to ensure the integrity and accuracy of the effective reflected signal. For example, when strong ambient light interference is detected in a certain area, the system automatically optimizes the frequency response of the filter, enhancing the ability to extract the effective signal from that area. The filtered point cloud data generated after wavelength-division filtering significantly reduces the impact of environmental noise and improves the signal-to-noise ratio. This filtered point cloud data not only preserves the key reflection characteristics of the scraper surface but also enhances the stability and reliability of the signal, laying a solid foundation for subsequent polarization filtering. The system uses wavelength division filtering to effectively remove environmental noise interference and extract high-quality effective reflection signals, ensuring that subsequent signal processing can be carried out on a clean and reliable data basis, thereby improving the performance and accuracy of the entire detection system.

[0082] S13: Perform polarization filtering on the filtered point cloud data to remove specular reflection interference from the metal surface and generate the original point cloud data.

[0083] Specifically, in industrial environments, scrapers are typically made of metal, and their surfaces are prone to specular reflection during use. This specular reflection can interfere with the system's normal detection, leading to data distortion. The system addresses this issue by analyzing the polarization characteristics of light. When a laser beam strikes a metal surface, the polarization state of the reflected light changes; specularly reflected light typically exhibits strong linear polarization. The system utilizes this characteristic by employing a polarizing filter to process the filtered point cloud data. A polarizing filter allows light with a specific polarization direction to pass through while blocking light from other directions. By precisely adjusting the polarization direction of the polarizing filter, the system can effectively remove interference from specularly reflected light. For example, setting the transmission axis of the polarizing filter perpendicular to the polarization direction of the specularly reflected light maximizes the blocking of specularly reflected light from entering the detection system.

[0084] In actual processing, the system dynamically adjusts the angle and parameters of the polarization filter based on the geometry and material properties of the scraper surface. For different areas of the scraper surface, the system optimizes the polarization filtering effect in real time according to their reflection characteristics and polarization information. The raw point cloud data generated after polarization filtering eliminates the interference of specular reflection from the metal surface, accurately reflecting the true contour and features of the scraper surface. This raw point cloud data provides high-quality data support for subsequent wear analysis and fault diagnosis, ensuring the high accuracy and reliability of the system's detection results. Through polarization filtering, the system effectively removes the interference of specular reflection from the metal surface, generating raw point cloud data that accurately reflects the true condition of the scraper surface, providing a solid foundation for subsequent analysis and processing, and ensuring the efficient operation and accurate detection of the entire detection system.

[0085] In one embodiment, S3 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0086] S31: The wear thickness distribution is calculated based on the fuzzy PID algorithm. The difference between the maximum wear zone thickness and the preset safety thickness is calculated to generate a unilateral pressure increment.

[0087] Specifically, the system performs in-depth analysis on the collected wear thickness distribution data to identify the thickness value of the area with the maximum wear. This maximum wear thickness value is then compared with a preset safe thickness threshold within the system, and the difference between the two is calculated. The preset safe thickness is not static but a dynamic reference value determined through extensive experiments and data analysis, taking into account the material properties of the scraper, its design strength, and various factors in the actual working environment. When the wear thickness exceeds the safe range, the system uses the rules of the fuzzy PID algorithm to fuzzify the calculated thickness difference, converting it into a series of fuzzy linguistic variables. These fuzzy variables better simulate the decision-making logic of human experts during the adjustment process, enabling the system to make effective pressure adjustment decisions even when facing complex and changing working conditions. The fuzzy PID controller infers and makes decisions based on these fuzzy variables and preset fuzzy rules, generating a one-sided pressure increment to adjust the scraper's cleaning pressure. In this way, the system can dynamically fine-tune the scraper's cleaning pressure, ensuring that even after wear, the scraper can still maintain effective cleaning performance, while avoiding excessive wear due to excessive pressure, thus extending its service life.

[0088] S32: Based on the single-sided pressure increment, the dynamic weight coefficient allocation is performed on both sides of the scraper, and the weight coefficient is assigned to the side with severe wear to generate the initial asymmetric pressure adjustment command.

[0089] Specifically, the two sides of the scraper often experience varying degrees of wear during actual use. To balance the wear rate on both sides, the system needs to adjust the pressure on the more severely worn side. The system uses the pressure increment on one side as its core basis, combined with real-time wear data from both sides of the scraper, to rationally distribute the cleaning pressure. Specifically, the system assigns a larger weighting coefficient to the more severely worn side based on the wear difference, allowing that side to receive more cleaning pressure and thus compensate for the reduced cleaning efficiency caused by wear. After determining the weighting coefficients, the system generates an initial asymmetric pressure adjustment command based on these coefficients. This command contains specific cleaning pressure parameters for both sides of the scraper, providing clear guidance for subsequent cleaning pressure adjustments. In this way, the system can effectively address the uneven wear on both sides of the scraper, ensuring that the cleaning pressure on both sides is always at its optimal level, improving the overall performance and efficiency of the sweeper.

[0090] S33: Based on the thin-film pressure sensor, the actual execution pressure of the initial asymmetric pressure regulation command is monitored and processed in real time. When the deviation between the actual pressure and the target pressure exceeds the preset threshold, the PID parameter self-correction algorithm is triggered to generate the final asymmetric pressure regulation command.

[0091] Specifically, a thin-film pressure sensor is tightly fitted to both sides of the scraper, accurately sensing the actual cleaning pressure and feeding the pressure signal back to the system in real time. The system analyzes this real-time pressure data and compares it with the target pressure in the initial asymmetric pressure adjustment command. If the deviation between the actual pressure and the target pressure exceeds a preset threshold, it indicates that the current cleaning pressure adjustment is ineffective and requires further optimization. At this point, the system triggers a PID parameter self-correction algorithm. This algorithm automatically adjusts the parameters of the PID controller, such as the proportional coefficient, integral coefficient, and derivative coefficient, based on the magnitude and direction of the deviation to improve the system's adjustment accuracy and stability. The system continuously monitors pressure changes through the thin-film pressure sensor and recalculates the asymmetric pressure adjustment command based on the corrected PID parameters, ensuring that the cleaning pressure on both sides of the scraper remains stable near the target value, achieving precise cleaning results. Through this real-time monitoring and parameter self-correction mechanism, the system can dynamically adapt to scraper wear changes and operating condition fluctuations, ensuring that the sweeper is always in optimal working condition, effectively extending the equipment's service life and reducing maintenance costs.

[0092] In one embodiment, S4 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0093] S41: Based on the updated wear thickness distribution and asymmetric pressure regulation command, spatiotemporal alignment processing is performed, and the correlation data between wear rate and pressure regulation are fused to generate a dynamic wear rate matrix.

[0094] Specifically, the system performs spatiotemporal alignment processing based on updated wear thickness distribution and asymmetric pressure regulation commands. This precisely aligns the updated wear thickness distribution data with the asymmetric pressure regulation commands in both time and space dimensions to ensure data synchronization and consistency. Through this processing, the system can accurately correlate wear thickness at different times and locations with corresponding pressure regulation commands, providing a reliable data foundation for subsequent analysis. The system integrates the correlation data between wear rate and pressure regulation to establish a dynamic wear rate matrix. This matrix not only contains wear rate information at various locations on the scraper surface but also reflects the impact of pressure regulation on the wear rate. By analyzing the relationship between wear rate and pressure regulation, the system can understand the effectiveness of different pressure regulation strategies in practical applications, providing a basis for optimizing the sweeper's operating parameters. The system utilizes data analysis algorithms to deeply mine and analyze wear thickness distribution and pressure regulation commands, extracting key correlation features and integrating them into the dynamic wear rate matrix. This matrix can reflect the wear status of the scraper and the effect of pressure regulation in real time, providing an accurate reference for subsequent anomaly detection.

[0095] S42: Perform anomaly detection processing on the dynamic wear rate matrix. When the wear rate on one side exceeds the safety threshold range and the compensation pressure deviation continues to increase, it is determined to be an irreversible uneven wear anomaly, and the anomaly determination result is output.

[0096] Specifically, the system monitors the data in the dynamic wear rate matrix in real time. When it detects that the wear rate on one side exceeds the safety threshold range and the compensation pressure deviation continues to increase, the system determines it to be an irreversible uneven wear anomaly. The safety threshold range is set comprehensively based on factors such as the material properties, design strength, and actual working conditions of the scraper, aiming to ensure the normal operation of the scraper and the safety of the equipment. The system analyzes the dynamic wear rate matrix through complex algorithms to identify abnormal wear patterns and pressure regulation deviations. When an anomaly is detected, the system immediately outputs the anomaly determination result so that corresponding maintenance measures can be taken in a timely manner. This process relies not only on the system's real-time monitoring capabilities but also on its analysis and learning of historical data, enabling the system to accurately identify abnormal situations and issue timely alarms, preventing equipment failure due to excessive wear and ensuring the stable operation of the intelligent belt sweeper.

[0097] S43: Based on the determination result of irreversible uneven wear anomaly, match the fault characteristics in the preset fault type database to generate a unique uneven wear fault code. The uneven wear fault code is used to associate the solution in the maintenance knowledge base.

[0098] Specifically, during the operation of the intelligent belt sweeper, the system, based on the determination of irreversible uneven wear anomalies, matches fault characteristics in a preset fault type database. The system compares the determination result with various fault characteristics stored in the database, and through a complex pattern recognition algorithm, identifies the fault type that best matches the current anomaly, thus generating a unique uneven wear fault code. This uneven wear fault code is the system's identifier for a specific uneven wear fault, and it can quickly and accurately link to solutions in the maintenance knowledge base. Maintenance personnel can quickly find the corresponding solution based on this fault code, thereby improving maintenance efficiency, shortening downtime, and ensuring the normal operation of the equipment. In this way, the system can not only promptly detect equipment anomalies but also provide effective solutions, helping users quickly restore the equipment to normal operation and improving its reliability and availability.

[0099] S44: Based on the dynamic difference between the pressure regulation data and the target pressure in the dynamic wear rate matrix, a compensation pressure deviation value is generated by calculating using the sliding window mean algorithm.

[0100] Specifically, the system monitors the pressure regulation data in the dynamic wear rate matrix in real time and compares it with the set target pressure, calculating the difference between the two. The sliding window averaging algorithm calculates the average value of the pressure difference data within a certain time window, thus obtaining the compensated pressure deviation value. This deviation value reflects the gap between the current pressure regulation and the ideal state, providing a basis for further optimization of pressure regulation. The system uses the compensated pressure deviation value to adjust the pressure regulation system in real time, ensuring that the pressure on both sides of the scraper remains within the optimal range, thereby effectively reducing wear, extending the service life of the equipment, and improving its operating efficiency and reliability.

[0101] In one embodiment, S5 of the intelligent belt sweeper automatic adjustment and fault diagnosis method provided by the present invention specifically includes the following steps:

[0102] S51: Based on heterogeneous data fusion technology, the wear fault codes, updated wear thickness distribution and compensation pressure deviation values ​​are standardized and encoded. Discrete fault codes are converted into feature vectors, continuous wear data are converted into grayscale matrices, and dynamic deviation values ​​are normalized into scalar sequences to generate fault input vectors in a unified format.

[0103] Specifically, the system converts discrete wear fault codes into feature vectors. Through a specific encoding algorithm, it maps the fault code's category information and severity to a vector space, ensuring each fault code corresponds to a unique feature vector, facilitating subsequent analysis and processing. For continuous wear thickness distribution data, the system converts it into a grayscale matrix, representing the wear thickness distribution as grayscale values. The magnitude of the grayscale value is proportional to the wear thickness, thus simplifying the complex wear distribution information into a easily processed matrix form. Simultaneously, the system normalizes dynamic deviation values ​​into a scalar sequence. This normalization process unifies deviation values ​​with different dimensions and ranges into a fixed range, improving data comparability and processing efficiency. After these processes, the system generates a fault input vector in a unified format, providing a standardized data foundation for subsequent wear prediction.

[0104] S52: Based on a pre-trained neural network model, multi-source feature extraction processing is performed on the fault input vector. By freezing the underlying network parameters and fine-tuning the fully connected layer, the wear prediction task under the current working conditions is adapted to generate a fused feature tensor.

[0105] Specifically, the pre-trained neural network model can extract hidden feature patterns from complex input data. The system freezes the underlying network parameters and fine-tunes the fully connected layers, enabling the model to better adapt to the wear prediction task under current operating conditions. Freezing the underlying network parameters preserves the model's general feature extraction capabilities learned during pre-training, while fine-tuning the fully connected layers allows the model to learn and adjust specifically according to the characteristics of the current task, thereby maintaining the model's generalization ability while improving its adaptability and prediction accuracy for specific tasks. Through this process, the system generates a fused feature tensor, which integrates multi-source feature information from the fault input vector, providing rich feature input for subsequent spatiotemporal sequence prediction. The fused feature tensor contains data that has undergone feature extraction and adaptation, more accurately reflecting the wear state and changing trends of the scraper, providing strong support for the prediction model.

[0106] S53: Perform spatiotemporal sequence prediction processing on the fused feature tensor to generate and send wear prediction results containing scraper wear status and remaining lifespan.

[0107] Specifically, the system analyzes and fuses feature information in the feature tensor, combining it with the changing patterns of time and space dimensions to predict the future wear and remaining lifespan of the scraper. Spatiotemporal sequence prediction processing comprehensively considers the wear characteristics of the scraper at different locations and times, thus more accurately predicting its future trends. Based on the prediction results, the system generates and sends wear prediction results containing the scraper's wear status and remaining lifespan. These wear prediction results are presented to users in an intuitive way, helping them understand the equipment's operating status in a timely manner, formulate maintenance plans in advance, rationally allocate maintenance resources, avoid production interruptions caused by sudden equipment failures, improve equipment operating efficiency and reliability, and reduce maintenance costs. In this way, the system provides a scientific basis for decision-making regarding the maintenance and management of intelligent belt sweepers, ensuring that the equipment operates in a highly efficient and stable state.

[0108] The aforementioned intelligent belt sweeper automatic adjustment and fault diagnosis method utilizes heterogeneous data fusion technology to standardize the encoding of discrete fault codes, continuous wear thickness distribution, and dynamic pressure deviation values. This enables unified format and feature alignment of multi-source heterogeneous data, mapping fault codes to feature vectors, converting wear thickness into grayscale matrices, and normalizing dynamic deviations into scalar sequences. This generates a fusion vector adapted to the input of a deep learning model, addressing the inconsistency in feature representation caused by fragmented multimodal data in traditional methods. Based on a pre-trained neural network model, multi-source feature extraction is performed on the fusion vector. By freezing the underlying network parameters to retain common features across scenarios and fine-tuning the fully connected layer to adapt to the current equipment operating conditions, efficient transfer learning under small sample data is achieved, generating a fusion feature tensor representing the wear evolution law and fault correlation. Furthermore, by combining a spatiotemporal sequence prediction algorithm to perform temporal modeling and spatial dimension analysis on the fusion feature tensor, the method can simultaneously capture the progressive degradation trend and local abrupt change features of scraper wear, generating prediction results including the remaining life probability distribution and wear heatmap. This supports dynamic optimization of maintenance strategies and precise scheduling of spare parts resources. This technical approach breaks through the limitations of traditional single-dimensional prediction models by progressively processing data fusion, transfer learning, and spatiotemporal prediction, thereby achieving cross-modal collaborative assessment and full lifecycle management of scraper health status.

[0109] Preferably, such as Figure 3 As shown, the present invention provides an intelligent belt sweeper automatic adjustment and fault diagnosis system 600, which is configured with the following modules:

[0110] The data acquisition and processing module 610 is used to scan the scraper surface of the smart belt cleaner by a multispectral laser array integrated on the bracket of the smart belt cleaner, separate the reflected signals of different wavelengths to eliminate ambient light interference, and generate raw point cloud data. The raw point cloud data is used to characterize the scraper surface contour.

[0111] The wear thickness calculation module 620 is used to perform dynamic curvature compensation processing on the original point cloud data, dynamically adjust the wear depth calculation according to the change of the scraper surface shape, and generate the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state.

[0112] The adjustment command generation module 630 is used to perform dynamic pressure weight allocation processing on the wear thickness distribution based on fuzzy PID control technology, calculate the wear compensation pressure difference on both sides of the scraper in real time, generate an asymmetric pressure adjustment command, and send the asymmetric pressure adjustment command to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment command is used to control the intelligent belt sweeper to balance the wear rate on both sides.

[0113] The fault diagnosis module 640 is used to perform dynamic wear rate fusion processing and threshold judgment based on the feedback-based updated wear thickness distribution and asymmetric pressure adjustment command, identify unilateral excessive wear anomalies, and output the wear fault code that triggers the maintenance alarm and the compensation pressure deviation value for evaluating the adjustment effect.

[0114] The life prediction module 650 is used to perform multi-source data fusion processing on the wear fault code, the updated wear thickness distribution and the compensation pressure deviation value based on transfer learning technology, generate and send the wear prediction results, and the wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

[0115] In summary, the intelligent belt sweeper automatic adjustment and fault diagnosis system provided by this invention combines multispectral laser array scanning with dynamic curvature compensation to effectively eliminate interference from scraper surface deposits and measurement errors caused by dynamic deformation, achieving non-contact high-precision wear thickness detection. Based on fuzzy PID control and dynamic pressure weight allocation technology, it senses the wear difference between the two sides of the scraper in real time and generates asymmetric adjustment commands to achieve adaptive balance control of cleaning pressure, avoiding the problem of uneven wear accumulation under traditional fixed threshold adjustment strategies. By fusing updated wear thickness distribution and pressure adjustment feedback data, a dynamic wear rate matrix is ​​constructed and multi-dimensional threshold judgment is performed, accurately identifying unilateral excessive wear anomalies and simultaneously outputting uneven wear fault codes and compensation pressure deviation values, providing quantitative basis for maintenance decisions. Furthermore, by combining transfer learning-driven multi-source data fusion processing, discrete fault codes, continuous wear data, and dynamic deviation values ​​are mapped into a unified feature space. Based on cross-scenario knowledge transfer and local parameter fine-tuning, it can achieve remaining life prediction and generation of graded maintenance strategies under small sample working conditions, supporting the construction of a predictive maintenance system. The overall solution forms a closed-loop technical path of "precise perception - dynamic adjustment - intelligent diagnosis - predictive decision-making", which significantly improves the reliability of condition monitoring, real-time adjustment response and scientific maintenance decision-making of belt sweepers in complex industrial scenarios.

[0116] Preferably, the data acquisition and processing module 610 provided in the above embodiments is configured with the following units:

[0117] The scanning data generation unit is used to perform multi-band synchronous scanning processing on the scraper surface based on the multispectral laser array integrated on the bracket of the intelligent belt sweeper, separate near-infrared and short-wave infrared reflection signals, penetrate the scraper surface attachments and enhance the detection of low reflectivity areas, and generate multispectral scanning data.

[0118] The filtered point cloud generation unit is used to perform wavelength frequency division filtering on multispectral scanning data, separate the effective reflection signal from environmental noise, and generate filtered point cloud data.

[0119] The raw point cloud generation unit is used to perform polarization filtering on the filtered point cloud data to remove specular reflection interference from the metal surface and generate raw point cloud data.

[0120] Preferably, the wear thickness calculation module 620 provided in the above embodiment is configured with the following units:

[0121] The effective reflection generation unit is used to filter out ambient light interference from the original point cloud data based on wavelength frequency division filtering technology, separate near-infrared and short-wave infrared reflection signals, and generate effective reflection point data. This data is used to extract the effective contour of the scraper surface.

[0122] The wear compensation generation unit is used to perform non-uniform B-spline surface fitting on the effective reflection point data based on the dynamic curvature compensation algorithm, and generate the wear compensation depth by combining the viscoelastic parameters of the scraper material and the belt speed feedback value.

[0123] The thickness distribution generation unit is used to generate a wear thickness distribution by weighted superposition of the compensated wear depth and the original point cloud data based on data fusion technology.

[0124] Preferably, the adjustment instruction generation module 630 provided in the above embodiments is configured with the following units:

[0125] The single-sided pressure increment generation unit is used to calculate the wear thickness distribution based on the fuzzy PID algorithm, calculate the difference between the maximum wear zone thickness and the preset safety thickness, and generate the single-sided pressure increment.

[0126] The asymmetric pressure weight adjustment unit is used to dynamically allocate weight coefficients to both sides of the scraper based on the single-sided pressure increment, allocate weight coefficients to the severely worn side, and generate an initial asymmetric pressure adjustment command.

[0127] The pressure regulation command generation unit is used to monitor and process the actual execution pressure of the initial asymmetric pressure regulation command in real time based on the thin-film pressure sensor. When the deviation between the actual pressure and the target pressure exceeds a preset threshold, the PID parameter self-correction algorithm is triggered to generate the final asymmetric pressure regulation command.

[0128] Preferably, the fault diagnosis module 640 provided in the above embodiments is configured with the following units:

[0129] The dynamic wear rate generation unit is used to perform spatiotemporal alignment processing based on the updated wear thickness distribution and asymmetric pressure regulation command, and to fuse the correlation data between wear rate and pressure regulation to generate a dynamic wear rate matrix.

[0130] The anomaly judgment result output unit is used to perform anomaly detection processing on the dynamic wear rate matrix. When the wear rate on one side exceeds the safety threshold range and the compensation pressure deviation continues to increase, it is judged as an irreversible uneven wear anomaly, and the anomaly judgment result is output.

[0131] The uneven wear fault code generation unit is used to generate a unique uneven wear fault code by matching the fault characteristics in the preset fault type database based on the judgment result of the irreversible uneven wear anomaly. This code is used to associate the solution in the maintenance knowledge base.

[0132] The compensation pressure deviation value generation unit generates the compensation pressure deviation value by calculating the dynamic difference between the pressure adjustment data in the dynamic wear rate matrix and the target pressure through a sliding window mean algorithm.

[0133] Preferably, the lifetime prediction module 650 provided in the above embodiments is configured with the following units:

[0134] The fault input vector generation unit is used to standardize the encoding of uneven wear fault codes, updated wear thickness distribution and compensation pressure deviation values ​​based on heterogeneous data fusion technology. It converts discrete fault codes into feature vectors, continuous wear data into grayscale matrices, and dynamic deviation values ​​into scalar sequences to generate fault input vectors in a unified format.

[0135] The fusion feature tensor generation unit is used to perform multi-source feature extraction processing on the fault input vector based on the pre-trained neural network model. By freezing the parameters of the underlying network and fine-tuning the fully connected layer, it adapts to the wear prediction task under the current working conditions and generates a fusion feature tensor.

[0136] The wear prediction result generation and transmission unit is used to perform spatiotemporal sequence prediction processing on the fused feature tensor, and generate and transmit wear prediction results containing scraper wear status and remaining life.

[0137] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method for automatic adjustment and fault diagnosis of an intelligent belt sweeper.

[0138] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for automatic adjustment and fault diagnosis of an intelligent belt sweeper.

[0139] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0140] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0141] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for automatic adjustment and fault diagnosis of an intelligent belt sweeper, characterized in that, Includes the following steps: S1: The surface of the scraper of the smart belt cleaner is scanned by a multispectral laser array integrated on the bracket of the smart belt cleaner, and the reflected signals of different wavelengths are separated to eliminate ambient light interference and generate raw point cloud data. The raw point cloud data is used to characterize the surface contour of the scraper. S2: Perform dynamic curvature compensation processing on the original point cloud data, dynamically adjust the wear depth calculation according to the change of the scraper surface shape, and generate the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state. S3: Based on fuzzy PID control technology, the wear thickness distribution is dynamically weighted and processed. The wear compensation pressure difference between the two sides of the scraper is calculated in real time, an asymmetric pressure adjustment command is generated, and the asymmetric pressure adjustment command is sent to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment command is used to control the intelligent belt sweeper to balance the wear rate on both sides. S4: Based on the feedback-updated wear thickness distribution and the asymmetric pressure adjustment command, perform dynamic wear rate fusion processing and threshold judgment to identify unilateral excessive wear anomalies, output the wear fault code that triggers the maintenance alarm and the compensation pressure deviation value for evaluating the adjustment effect; S5: Based on transfer learning technology, multi-source data fusion processing is performed on the wear fault code, the updated wear thickness distribution, and the compensation pressure deviation value to generate and send wear prediction results. The wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

2. The method according to claim 1, characterized in that, S2 includes: S21: Based on wavelength division filtering technology, the original point cloud data is subjected to ambient light interference filtering to separate near-infrared and short-wave infrared reflection signals and generate effective reflection point data. The effective reflection point data is used to extract the effective contour of the scraper surface. S22: Based on the dynamic curvature compensation algorithm, the effective reflection point data is subjected to non-uniform B-spline surface fitting processing, and the wear compensation depth is generated by combining the viscoelastic parameters of the scraper material and the belt speed feedback value. S23: Based on data fusion technology, the compensated wear depth and the original point cloud data are weighted and superimposed to generate a wear thickness distribution.

3. The method according to claim 2, characterized in that, The calculation method for the compensation wear depth is as follows: in, To compensate for the depth of wear, The viscoelastic coefficient, As an edge wear acceleration factor, The radius of curvature of the scraper surface is calculated using a non-uniform B-spline surface fitting algorithm. Let E be the belt speed, E be the elastic modulus of the scraper material, w be the scraper surface displacement in the original point cloud data, and x be the scraper position in the original point cloud data. is the time interval for the k-th scan.

4. The method according to claim 1, characterized in that, S1 includes: S11: Based on the multispectral laser array integrated on the bracket of the intelligent belt sweeper, the scraper surface is simultaneously scanned in multiple bands to separate near-infrared and short-wave infrared reflection signals, penetrate the scraper surface attachments and enhance the detection of low reflectivity areas, and generate multispectral scanning data. S12: Perform wavelength frequency division filtering on the multispectral scanning data to separate the effective reflection signal from the environmental noise and generate filtered point cloud data; S13: Perform polarization filtering on the filtered point cloud data to remove specular reflection interference from the metal surface and generate the original point cloud data.

5. The method according to claim 1, characterized in that, S3 includes: S31: Calculate the wear thickness distribution based on the fuzzy PID algorithm, calculate the difference between the maximum wear zone thickness and the preset safety thickness, and generate a unilateral pressure increment; S32: Based on the single-sided pressure increment, perform dynamic weight coefficient allocation processing on both sides of the scraper, allocate weight coefficient to the severely worn side, and generate an initial asymmetric pressure adjustment command. S33: Based on the thin-film pressure sensor, the actual execution pressure of the initial asymmetric pressure regulation command is monitored and processed in real time. When the deviation between the actual pressure and the target pressure exceeds a preset threshold, the PID parameter self-correction algorithm is triggered to generate the final asymmetric pressure regulation command.

6. The method according to claim 1, characterized in that, S4 includes: S41: Based on the updated wear thickness distribution and asymmetric pressure regulation command, perform spatiotemporal alignment processing, fuse the correlation data between wear rate and pressure regulation, and generate a dynamic wear rate matrix; S42: Perform anomaly detection processing on the dynamic wear rate matrix. When the wear rate on one side exceeds the safety threshold range and the compensation pressure deviation continues to increase, it is determined to be an irreversible uneven wear anomaly, and the anomaly determination result is output. S43: Based on the determination result of the irreversible uneven wear abnormality, match the fault features in the preset fault type database to generate a unique uneven wear fault code, and the uneven wear fault code is used to associate the solution in the maintenance knowledge base; S44: Based on the dynamic difference between the pressure adjustment data and the target pressure in the dynamic wear rate matrix, a compensation pressure deviation value is generated by calculating using a sliding window mean algorithm.

7. The method according to any one of claims 1-6, characterized in that, S5 includes: S51: Based on heterogeneous data fusion technology, the wear fault code, the updated wear thickness distribution and the compensation pressure deviation value are standardized and encoded. Discrete fault codes are converted into feature vectors, continuous wear data are converted into grayscale matrices, and dynamic deviation values ​​are normalized into scalar sequences to generate a fault input vector in a unified format. S52: Based on the pre-trained neural network model, multi-source feature extraction processing is performed on the fault input vector. By freezing the bottom network parameters and fine-tuning the fully connected layer, the wear prediction task under the current working condition is adapted to generate a fused feature tensor. S53: Perform spatiotemporal sequence prediction processing on the fused feature tensor to generate and send wear prediction results containing scraper wear status and remaining lifespan.

8. An intelligent belt sweeper automatic adjustment and fault diagnosis system, characterized in that, The system includes: The data acquisition and processing module is used to scan the scraper surface of the intelligent belt cleaner using a multispectral laser array integrated on the bracket of the intelligent belt cleaner, separate the reflected signals of different wavelengths to eliminate ambient light interference, and generate raw point cloud data, which is used to characterize the surface contour of the scraper. The wear thickness calculation module is used to perform dynamic curvature compensation processing on the original point cloud data, dynamically adjust the wear depth calculation according to the change of the scraper surface shape, and generate the wear thickness distribution at each position of the scraper. The wear thickness distribution is used to reflect the non-uniform wear state. The adjustment command generation module is used to perform dynamic pressure weight allocation processing on the wear thickness distribution based on fuzzy PID control technology, calculate the wear compensation pressure difference on both sides of the scraper in real time, generate an asymmetric pressure adjustment command, and send the asymmetric pressure adjustment command to the control terminal of the intelligent belt sweeper. The asymmetric pressure adjustment command is used to control the intelligent belt sweeper to balance the wear rate on both sides. The fault diagnosis module is used to perform dynamic wear rate fusion processing and threshold judgment based on the updated wear thickness distribution and the asymmetric pressure adjustment command, identify unilateral excessive wear anomalies, and output the wear fault code that triggers the maintenance alarm and the compensation pressure deviation value for evaluating the adjustment effect. The life prediction module is used to perform multi-source data fusion processing on the wear fault code, the updated wear thickness distribution and the compensation pressure deviation value based on transfer learning technology, generate and send wear prediction results, and the wear prediction results are used to indicate the wear status and remaining life of the scraper of the intelligent belt sweeper.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images