Fault detection method and system for mine-used single-track double-driving light narrow transport equipment

Abstract

This application relates to the field of transportation equipment fault detection technology, and discloses a fault detection method and system for a mine monorail dual-driver light narrow-type transportation equipment. The method includes: real-time acquisition of operating data from multiple sensors deployed on the mine monorail dual-driver light narrow-type transportation equipment; calculation of the signal correlation function of vibration and acoustic signals in the frequency domain, filtering out frequency bands with signal correlation greater than a preset correlation threshold to construct a first fault set; calculation of the correlation coefficient between current changes and vibration signal changes, identifying parts where the correlation coefficient deviates from a preset correlation range to construct a second fault set; analysis of gear wear on the transportation equipment to obtain fault results; dynamic allocation of the transportation equipment's movement control process to one of the two driver's cabs, and fault warning; this application, through signal fusion analysis, can effectively filter out environmental interference, improve the accuracy of fault detection results, and accurately locate the fault position.

Description

Technical Field

[0001] This application relates to the field of transportation equipment fault detection technology, and more specifically to a fault detection method and system for mining monorail dual-drive light narrow-type transportation equipment. Background Technology

[0002] Currently, mining monorail dual-cab lightweight narrow-type transport equipment operates for extended periods in complex underground environments. Failures of key mechanical components are a major risk factor for safety accidents and production interruptions. Existing fault detection technologies rely on threshold alarms from single sensors or periodic manual inspections. These methods suffer from high false alarm rates, difficulty in identifying early, subtle faults, and a disconnect between diagnostic results and control decisions, especially under the interference of strong noise and varying loads underground. For equipment with a dual-cab design, current technology treats this merely as simple operational redundancy, failing to link fault diagnosis information with the dual-cab control logic and thus unable to dynamically select the safest operating mode when a fault occurs.

[0003] The existing technology has the following problems: relying on regular manual inspections or single vibration monitoring makes it difficult to achieve early fault warning; early wear defects of key transmission components such as gearboxes can only be discovered after the fault expands, causes abnormal noise, or stops; the underground environment is noisy, and the signal of a single sensor is easily interfered with, leading to false alarms or missed alarms; in order to solve at least one of the above problems, this application proposes a fault detection method and system for mine monorail dual-drive light narrow-type transportation equipment. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide a fault detection method and system for mine monorail dual-driver light narrow-type transportation equipment, which can effectively solve the problems in the background technology. The specific technical solution of this application is as follows:

[0005] Fault detection methods for mine monorail dual-operator light narrow-type transportation equipment include:

[0006] Based on multiple sensors deployed on the mining monorail dual-drive light narrow-type transport equipment, the operating data of the transport equipment is collected in real time, including vibration signals, acoustic signals, current signals and image data;

[0007] Calculate the signal correlation function of vibration signal and acoustic signal in the frequency domain, filter out the frequency band part with signal correlation greater than the preset correlation threshold, and construct the first fault set;

[0008] Based on the analysis of electrical load fluctuations using current signals, the correlation coefficient between current changes and vibration signal changes is calculated, the portion of the correlation coefficient that deviates from the preset correlation range is identified, and a second fault set is constructed.

[0009] By combining the first fault set, the second fault set, and image data, the gear wear of the transportation equipment is analyzed to obtain fault results.

[0010] A dual-control collaborative control mechanism is configured to dynamically allocate the movement control process of the transportation equipment to one of the two driver's cabs according to the fault result, and to provide fault warnings in order to detect faults in the transportation equipment.

[0011] Specifically, the calculation of the signal correlation function between the vibration signal and the acoustic signal in the frequency domain, the selection of frequency bands with signal correlation greater than a preset correlation threshold, and the construction of a first fault set include:

[0012] The vibration signal and acoustic signal are synchronously Fourier transformed and compared with the corresponding reference data under the stable unloaded operation of the transportation equipment to identify the first region with energy higher than the reference.

[0013] The signal correlation function of vibration signal and acoustic signal in the frequency domain is calculated based on the first region. The frequency bands with signal correlation greater than the preset correlation threshold are selected to construct the first fault set.

[0014] Specifically, the step of calculating the signal correlation function of the vibration signal and the acoustic signal in the frequency domain based on the first region, filtering out the frequency bands with signal correlation greater than a preset correlation threshold, and constructing a first fault set includes:

[0015] The cross-spectral coherence function of the vibration signal and the acoustic signal in the frequency domain is calculated in the first region to obtain the first spectral function;

[0016] The vibration signal and acoustic signal are demodulated and the corresponding envelope signals are extracted. The coherence function between the envelope signals is calculated to obtain the second spectral function.

[0017] By fusing the first spectral function and the second spectral function, a signal correlation function is obtained. Frequency bands with signal correlation greater than a preset correlation threshold are selected to construct the first fault set.

[0018] Specifically, the step of analyzing electrical load fluctuations based on current signals, calculating the correlation coefficient between current changes and vibration signal changes, identifying the portion of the correlation coefficient that deviates from a preset correlation range, and constructing a second fault set includes:

[0019] Based on the analysis of electrical load fluctuations using current signals, and combined with the real-time operating conditions of the transportation equipment, the correlation coefficient between current changes and vibration signal changes under each operating condition is calculated.

[0020] Identify the parts where the correlation coefficient deviates from the preset correlation range and construct a second fault set.

[0021] Specifically, the step of analyzing electrical load fluctuations based on current signals, combined with the real-time operating conditions of the transportation equipment, calculates the correlation coefficient between current changes and vibration signal changes under each operating condition, including:

[0022] According to the real-time operating conditions of the transportation equipment, the current signal and vibration signal are divided into signal segments under the corresponding operating conditions.

[0023] For each operating condition, the reference data under normal conditions are removed from the current signal and vibration signal in the signal segment to obtain the current difference sequence and vibration difference sequence;

[0024] Based on the analysis of electrical load fluctuations using current difference sequences, the mutual information value and correlation coefficient between the current difference sequence and the vibration difference sequence are calculated.

[0025] Based on the mutual information value and correlation coefficient, cross-validation and correlation analysis are performed on the envelope spectrum characteristics of current characteristics and vibration signals to obtain the corresponding correlation coefficients.

[0026] Specifically, by combining the first fault set, the second fault set, and image data, the gear wear of the transportation equipment is analyzed to obtain fault results, including:

[0027] By combining the first fault set, the second fault set, and image data, the data is mapped to the real-time operating location and track section of the transportation equipment to obtain a fault feature sequence after position alignment.

[0028] The failure results are obtained by analyzing the gear wear of the transportation equipment based on the failure characteristic sequence.

[0029] Specifically, the step of analyzing the gear wear of the transportation equipment based on the fault characteristic sequence to obtain the fault results includes:

[0030] Based on the fault feature sequence analysis, the consistency between the first fault set and the second fault set at the corresponding position is analyzed. The correspondence between the characteristic frequency and the fault position and the gear is analyzed. The fault positions with consistent faults and corresponding gears are selected to obtain the fault sequence.

[0031] Defect identification is performed on the image data corresponding to the fault sequence to obtain the corresponding defect information;

[0032] By combining fault sequences and defect information, the gear wear of the transportation equipment is analyzed to obtain fault results.

[0033] Specifically, the dual-control collaborative control mechanism includes:

[0034] Based on the fault results and combined with the real-time driving information of the transportation equipment, the movement control process of the transportation equipment is dynamically allocated to one of the two driver's cabs;

[0035] Fault warnings are issued based on the fault results to detect faults in transportation equipment.

[0036] Specifically, the process of dynamically allocating the movement control process of the transportation equipment to one of the two driver's cabs based on the fault result and in conjunction with the real-time driving information of the transportation equipment includes:

[0037] Based on the fault type and location in the fault results, analyze the scope of the fault's impact, calculate the risk of the fault's impact on the front and rear cabs respectively, and obtain the corresponding risk levels.

[0038] The driver's cab with the lower risk level is set as the main driver's cab between the front and rear cabs. Based on the real-time driving information of the transportation equipment and the current risk level of the driver's cab, driving switching instructions are set.

[0039] According to the driving switching command, the movement control process of the transport equipment is dynamically assigned to one of the two driver's cabs.

[0040] A fault detection system for mine monorail dual-operator light narrow-type transportation equipment is used to implement the aforementioned fault detection method for mine monorail dual-operator light narrow-type transportation equipment, including:

[0041] The data acquisition module collects real-time operating data of the mining monorail dual-drive light narrow-type transport equipment based on multiple sensors deployed on the equipment. The operating data includes vibration signals, acoustic signals, current signals, and image data.

[0042] The first fault analysis module calculates the signal correlation function of vibration signal and acoustic signal in the frequency domain, filters out the frequency band portion where the signal correlation is greater than the preset correlation threshold, and constructs the first fault set.

[0043] The second fault analysis module analyzes the electrical load fluctuation based on the current signal, calculates the correlation coefficient between the current change and the vibration signal change, identifies the part of the correlation coefficient that deviates from the preset correlation range, and constructs a second fault set.

[0044] The fault result analysis module combines the first fault set, the second fault set, and image data to analyze the gear wear of the transportation equipment and obtain the fault results.

[0045] The fault warning module is equipped with a dual-control collaborative control mechanism. According to the fault result, it dynamically allocates the movement control process of the transportation equipment to one of the two driver's cabs and issues a fault warning to detect faults in the transportation equipment.

[0046] The beneficial effects of this application are as follows: By analyzing the spectrum of vibration and acoustic signals and calculating their correlation function in the frequency domain, highly correlated frequency bands are selected, effectively eliminating random environmental noise and highlighting characteristic frequencies excited by the same fault source and simultaneously reflected in vibration and acoustic signals, thus improving the signal-to-noise ratio and reliability of mechanical fault feature extraction; by analyzing current signals reflecting electrical load and dynamically calculating the correlation coefficient between current changes and vibration signal changes, and analyzing the deviation of the correlation coefficient from the preset normal range, abnormal vibrations caused by non-load factors are identified; by aligning the first fault set, the second fault set, and image data, the consistency of fault features is analyzed, and defect identification verification is performed using images, improving the accuracy of fault results; through a dual-control collaborative control mechanism, the vehicle movement control is dynamically switched to the lower-risk cab based on the fault type and location, enabling fault warning and timely response; and through signal fusion analysis, environmental interference can be effectively filtered out, improving the accuracy of fault detection results and accurately locating the fault location, reducing false alarm and false negative rates. Attached Figure Description

[0047] Figure 1 This is an overall structural diagram of the mining monorail dual-driver light narrow-type transportation equipment in Embodiment 1 of this application;

[0048] Figure 2 This is a flowchart illustrating the fault detection method for a mine monorail dual-driver light narrow-type transportation equipment in Embodiment 2 of this application.

[0049] Figure 3 This is a flowchart illustrating the process of calculating the correlation coefficient between current changes and vibration signal changes in Embodiment 2 of this application.

[0050] Figure 4 This is a schematic diagram of the fault detection system for a single-rail, dual-driver, light and narrow-type transportation equipment used in mining, as described in Embodiment 2 of this application. Detailed Implementation

[0051] The present application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0052] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0053] Hereinafter, the terms "first," "second," and other generic terms are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0054] Example 1:

[0055] In traditional auxiliary transportation at tunneling faces, manual labor and conveyor belts are the primary methods used. Manual transportation involves workers carrying materials by hand, while with conveyor belts, auxiliary materials are typically transported to the tail of the conveyor and then manually moved to the tunneling face. Both methods waste significant manpower and resources, hinder tunneling progress, and pose substantial safety hazards. Issues such as material detachment and loss, and the potential for material entanglement in the conveyor belt can damage the equipment. Therefore, a new auxiliary transportation approach is needed to address the material transportation difficulties in coal mine tunneling.

[0056] Current tunneling roadways are designed with long runs, making the transportation of personnel and materials difficult, labor-intensive, and time-consuming. This is especially true when the tunnel slope is steep, hindering the overall improvement of mechanization and automation in tunneling. Therefore, this application provides a mine-use monorail, dual-drive, lightweight, narrow-type transportation device, such as... Figure 1 As shown, the vehicle structure of the mining monorail dual-driver light narrow-type transportation equipment includes: a front and rear passenger car section and a power car section; the driving and power sections are separated, with a spacing greater than or equal to 30% of the vehicle length; the vehicle is mainly used for transporting personnel, materials, and tools, and the vehicle structure dimensions are: total length 7.3 meters, width 0.6 meters, height 1.4 meters, simple and lightweight, with a weight of 1200 kg; the power system uses an explosion-proof lithium battery to power the motor (72V, 230Ah), the motor power is 7.5-11kW, and the empty vehicle running speed is ≤1.5 m / s.

[0057] Example 2:

[0058] refer to Figure 2 The image shows a specific implementation method for the fault detection method of the mine monorail dual-driver light narrow-type transportation equipment of this application, including:

[0059] S101. Based on multiple sensors installed on the mining monorail dual-drive light narrow-type transportation equipment, real-time operation data of the transportation equipment is collected. The operation data includes vibration signals, acoustic signals, current signals and image data.

[0060] S102. Calculate the signal correlation function of vibration signal and acoustic signal in the frequency domain, filter out the frequency bands with signal correlation greater than the preset correlation threshold, and construct the first fault set.

[0061] S103. Analyze the electrical load fluctuation based on the current signal, calculate the correlation coefficient between the current change and the vibration signal change, identify the part of the correlation coefficient that deviates from the preset correlation range, and construct a second fault set.

[0062] S104. Combining the first fault set, the second fault set, and image data, analyze the gear wear of the transportation equipment to obtain fault results;

[0063] S105. Configure a dual-control collaborative control mechanism. According to the fault result, the movement control process of the transportation equipment is dynamically allocated to one of the two driver's cabs, and fault warning is given to detect faults in the transportation equipment.

[0064] In this embodiment, high-sensitivity triaxial vibration acceleration sensors are installed on key mechanical transmission components of the transportation equipment, such as the drive motor housing and gearbox bearing housing, to collect vibration signals of the structure. Weather-resistant acoustic sensors are installed near the gearbox to collect airborne noise signals during equipment operation. Hall effect current sensors are installed in the motor drive circuit to collect real-time operating current signals. Small industrial cameras with protective covers are installed facing the key gear meshing area or drive shaft to collect local image data during equipment operation. All sensors are connected to a central data processing unit via an industrial bus on the equipment. This unit has a built-in high-precision clock that records the timestamps of all sensor data streams, ensuring strict temporal synchronization of vibration, acoustic, current, and image frames, guaranteeing temporal consistency in the correlation analysis between different physical quantities. Data is digitized and buffered at a fixed sampling frequency; for example, the sampling frequency for vibration and acoustic signals is set to 12.8 kHz, the sampling frequency for current signals is set to 1 kHz, and images are acquired at a rate of 5 frames per second.

[0065] It should be noted that by acquiring data from multiple sensors, accurate data support is provided for the fault detection process, overcoming the shortcomings of incomplete information from a single signal dimension. Subsequent analysis can simultaneously consider mechanical structure vibration, acoustic radiation characteristics caused by the fault, load and energy consumption changes of the drive system, and physical morphological changes of the component surface. Through synchronous acquisition of multi-dimensional data, it is ensured that fault features from different sources can be aligned and correlated, effectively avoiding misjudgments caused by data asynchrony or missing dimensions.

[0066] Specifically, after preprocessing the synchronously acquired vibration and acoustic signals, a Fast Fourier Transform (FFT) is performed on each signal to convert it from the time domain to the frequency domain. The cross-spectral coherence function (CFC) between these two frequency domain signals is calculated to measure the linear correlation between the vibration and acoustic signals at each frequency point. The CFC value ranges from 0 to 1; the closer the value is to 1, the stronger the correlation between the two signals at that frequency point, and the more likely that frequency component originates from the same physical excitation source. Simultaneously, the vibration and acoustic signals are demodulated, and the extracted envelope signal is subjected to spectral analysis to obtain the envelope spectrum. The CFC between the two envelope spectra is then calculated. The results of the cross-spectral coherence function and the envelope spectrum coherence function are fused to obtain the signal correlation function. By statistically analyzing a large amount of normal state data and setting a correlation threshold (e.g., 0.8), all frequency bands with CFC values ​​exceeding the threshold are selected. The center frequencies corresponding to these highly correlated frequency bands are considered to be characteristic frequencies caused by potential faults and exhibiting strong correlation in both vibration and acoustic modes, thus constructing the first fault set.

[0067] It should be noted that frequency domain coherence analysis technology is used to cross-validate signals from two different propagation paths: vibration and acoustic. Random noise in the environment is usually uncorrelated or weakly correlated in vibration and acoustic signals, while periodic impacts caused by internal equipment faults will generate correlated frequency components in both structural vibration and radiated noise. By screening high coherence frequency bands, background noise interference unrelated to the fault can be suppressed, improving the signal-to-noise ratio and reliability of fault feature extraction. This allows for the accurate capture of weak fault features and effectively reduces the risk of false alarms caused by specific interference from a single sensor.

[0068] Furthermore, based on the equipment's speed, gear, or instructions from the control system, different real-time operating conditions are defined, including but not limited to no-load constant speed, heavy-load acceleration, and uphill traction. For each stable signal segment under different operating conditions, current and vibration signals are extracted. To eliminate differences in reference values ​​under different operating conditions, the pre-calibrated reference current curve and reference vibration curve for that operating condition under healthy conditions are subtracted from the current signal segment, resulting in a current difference sequence and a vibration difference sequence reflecting the current instantaneous fluctuations.

[0069] Preferably, the relationship between the current difference sequence and the vibration difference sequence is analyzed, and the mutual information value and correlation coefficient between the current difference sequence and the vibration difference sequence are calculated. The correlation coefficient includes, but is not limited to, the Pearson correlation coefficient. Based on the mutual information value and the correlation coefficient, the fluctuation characteristics of the current and the envelope spectrum characteristics of the vibration signal are cross-compared and correlated to calculate the correlation coefficient used to describe the tightness of the correlation between current change and vibration change under the current operating condition. Under the healthy state of the equipment, this correlation coefficient under different operating conditions will stabilize within a specific preset correlation range. The correlation range is set by historical health data. For example, for uniform speed operating conditions, the normal range is between 0.7 and 0.9. When the correlation coefficient calculated under a certain operating condition deviates from the correlation range, it indicates that the normal coupling relationship between current and vibration has been disrupted, which may be caused by abnormal vibration due to mechanical failure. This abnormal operating condition and its corresponding characteristic information are recorded to form a second fault set.

[0070] It should be noted that by analyzing the current signal, which reflects the system's input load, and examining the abnormal dynamic correlation between this current signal and the output mechanical vibration signal, internal mechanical faults can be identified, effectively decoupling the system. When increased vibration is caused by a normal increase in external load, the current will also increase proportionally, and the correlation coefficient will remain within the normal range. However, when increased vibration is caused by mechanical faults such as internal gear wear, the current shows no significant abnormal change, leading to a decrease in the correlation coefficient. This method can effectively distinguish between load fluctuations and actual mechanical faults, improving the system's adaptability to operating conditions and the accuracy of fault detection results.

[0071] Specifically, the first and second fault sets are mapped to the specific track positions of the equipment according to their corresponding timestamps, resulting in a fault feature sequence arranged by position. At the same position, it is checked whether the faulty component corresponding to the characteristic frequency indicated by the first fault set and the abnormality indicated by the second fault set mechanically point to the same component. Positions where both fault indications are consistent and point to a specific gear are selected, thus obtaining the fault sequence. A pre-trained deep learning image recognition model is then used to analyze the image data at the corresponding positions in the fault sequence.

[0072] For example, deep learning image recognition models include, but are not limited to, object detection and classification models based on convolutional neural networks. The training dataset for such models includes: synthetic images with various gear wear textures generated using simulation software, and actual images acquired after applying different degrees of wear to real gears on an experimental bench. The model takes a local image of the gear as input and outputs the probability of a defect in the image and a defect type classification, including normal, pitting, and spalling. During training, cross-entropy is used as the loss function, and stochastic gradient descent is employed for optimization. The initial learning rate is set to 0.001, the batch size to 32, and training is iterative until the loss converges. By combining the fault location and probability information provided by the fault sequence with the visual defect evidence and type information provided by the deep learning image recognition model, a fault result is obtained, such as moderate spalling wear on the gear.

[0073] It should be noted that by using spatial alignment and data fusion analysis to verify the first and second fault sets with the results based on image recognition, the reliability and accuracy of fault detection results can be improved. This provides clear information for maintenance, enabling predictive maintenance, reducing unnecessary downtime for inspections, and ensuring precise maintenance of the right components at the right time, thereby optimizing maintenance resources and improving equipment availability.

[0074] Specifically, based on the fault type, location, and severity in the fault results, combined with the real-time equipment status, including driving direction, speed, and whether there are drivers in the front and rear cabs, a safety risk assessment is conducted. For example, if the fault location is near the drive axle under the front cab, the analysis suggests that the fault may cause transmission jamming, abnormal noise, or vibration, resulting in a high safety risk level for the front cab and a low risk level for the rear cab. The analysis process can consider factors such as fault propagation path, physical distance, and the degree of driver impact for quantitative scoring. The cab with the low risk level is designated as the primary cab, and a driving switch command is generated, which is sent to the vehicle controller via the vehicle control network.

[0075] Preferably, the transport equipment controller dynamically allocates the highest decision-making authority for the vehicle's power, braking, steering, and other movement control processes to the designated primary driver's cab based on the driver switching command; simultaneously, the original driver's cab's control is temporarily restricted or switched to follow mode. The entire switching process is completed within milliseconds, ensuring control continuity. While dynamically allocating control authority, the driver is alerted via audible and visual alarms and a display screen within the driver's cab with warning information including fault details, impacts, and the control switching status, and is advised to take actions such as slowing down or heading to a repair shop.

[0076] It should be noted that by intelligently switching control to the cab, which is less affected by the fault, the impact of the fault on the driver's operating environment can be isolated to the greatest extent, avoiding operational errors caused by excessive vibration or direct mechanical impact on the driver, ensuring the safe departure of the equipment from the operating area, improving fault early warning and response efficiency, and enhancing the driving safety and stability of the transportation equipment.

[0077] This application utilizes spectral analysis of vibration and acoustic signals, calculating their correlation function in the frequency domain to filter out highly correlated frequency bands. This effectively eliminates random environmental noise, highlighting characteristic frequencies excited by the same fault source and simultaneously reflected in both vibration and acoustic signals, thus improving the signal-to-noise ratio and reliability of mechanical fault feature extraction. By analyzing current signals reflecting electrical loads and dynamically calculating the correlation coefficient between current changes and vibration signal changes, the deviation of the correlation coefficient from the preset normal range is analyzed to identify abnormal vibrations caused by non-load factors. Aligning the first and second fault sets with image data and analyzing the consistency of fault features, along with image-based defect identification verification, improves the accuracy of fault results. Through a dual-control collaborative control mechanism, based on the fault type and location, vehicle movement control is dynamically switched to the lower-risk driver's cab for fault warning and timely response. Signal fusion analysis effectively filters out environmental interference, improving the accuracy of fault detection results and accurately locating fault positions, reducing false alarm and false negative rates.

[0078] Furthermore, the signal correlation function of the vibration signal and the acoustic signal in the frequency domain is calculated, and the frequency bands with signal correlation greater than a preset correlation threshold are selected to construct a first fault set, including:

[0079] S201. Perform synchronous Fourier transform on the vibration signal and acoustic signal, and compare the energy with the corresponding reference data under the stable no-load operation state of the transportation equipment to identify the first region with energy higher than the reference.

[0080] S202. Calculate the signal correlation function of vibration signal and acoustic signal in the frequency domain according to the first region, filter out the frequency band part where the signal correlation is greater than the preset correlation threshold, and construct the first fault set.

[0081] In this embodiment, when the transportation equipment is in a stable and fault-free operating state, it undergoes a period of unloaded stable operation on a standard test track, simultaneously collecting multiple sets of vibration and acoustic signal sample data. The sample data are then subjected to a Fast Fourier Transform (FFT) to calculate the average amplitude at each frequency point, resulting in a pair of reference spectra, which serve as the vibration reference spectrum and the acoustic reference spectrum. During fault detection, the synchronously collected vibration and acoustic signals are subjected to the same FFT to obtain the real-time vibration spectrum and the real-time acoustic spectrum. The real-time spectra are then compared point-to-point with their corresponding reference spectra for energy.

[0082] Specifically, the difference between the real-time spectrum amplitude and the reference spectrum amplitude at each frequency point is calculated. Frequency points whose real-time spectrum amplitude exceeds 1.5 times the reference spectrum amplitude are selected. The frequency bands corresponding to these frequency points are constructed as the first region. The first region reflects the set of frequency components in the current signal that have significantly abnormal energy relative to the healthy no-load state, which may be caused by faults, load changes or strong interference.

[0083] It should be noted that by comparing with the benchmark of a specific working condition of no-load stability, background components such as inherent vibration of the equipment and normal meshing frequency can be effectively filtered out, and frequency components caused by faults such as wear and impact can be accurately identified; providing accurate analysis direction for subsequent fault detection, avoiding correlation search in the entire frequency band, and improving the efficiency and accuracy of the fault detection process.

[0084] Specifically, based on the first region, the correlation function of vibration and acoustic signals in the frequency domain is calculated, and frequency bands with correlation greater than a preset correlation threshold are selected to construct a first fault set. Frequency domain coherence analysis improves the anti-interference capability and accuracy of the fault detection process. Random environmental noise or non-fault-related structural resonances downhole can cause energy spikes in a single vibration or acoustic signal, but will not simultaneously exhibit highly correlated waveform characteristics in both heterogeneous signals. By selecting locations that simultaneously meet the conditions of high energy and high correlation, abnormal indications caused by the environment can be effectively eliminated, and periodic components excited by internal faults and simultaneously coupled to structural vibration and airborne sound can be identified, improving the accuracy of mechanical fault screening results.

[0085] Furthermore, based on the first region, the signal correlation function of the vibration signal and the acoustic signal in the frequency domain is calculated, and the frequency bands with signal correlation greater than a preset correlation threshold are selected to construct a first fault set, including:

[0086] S301. Calculate the cross-spectral coherence function of the vibration signal and the acoustic signal in the frequency domain in the first region to obtain the first spectral function;

[0087] S302. Demodulate the vibration signal and acoustic signal respectively and extract the corresponding envelope signal. Calculate the coherence function between the envelope signals to obtain the second spectral function.

[0088] S303. By fusing the first spectral function and the second spectral function, a signal correlation function is obtained. Frequency bands with signal correlation greater than a preset correlation threshold are selected to construct a first fault set.

[0089] In this embodiment, the cross-spectral coherence function measures whether the components of two signals at a specific frequency originate from the same physical source. A higher value indicates that the frequency component is more likely to exhibit a fixed amplitude ratio and stable phase relationship in both signals, rather than being independent random noise. Based on the first region, the time series of the synchronously acquired vibration and acoustic signals are processed. The same windowing function is applied to both signals to reduce spectral leakage. The windowing function includes, but is not limited to, the Hanning window. Fast Fourier transforms are performed to obtain the corresponding complex spectra. The cross-power spectrum is obtained by calculating the conjugate product of the complex spectra of the vibration and acoustic signals. Simultaneously, the auto-power spectra of the vibration and acoustic signals are calculated separately. For each frequency point, the square of the cross-power spectrum amplitude is divided by the product of the auto-power spectra of the two signals to obtain the cross-spectral coherence function value, resulting in the first spectral function. The output value of the function at each frequency point is a real number between 0 and 1, reflecting the strength of the linear correlation at that frequency point.

[0090] It should be noted that by calculating the cross-spectral coherence function, background noise unrelated to the fault can be effectively suppressed. For example, random gas flow noise downhole can affect the acoustic signal but will not generate relevant frequency components in the vibration signal, and the coherence value at these frequency points will be very low. Vibrations with specific characteristics caused by gear meshing faults will generate relevant acoustic radiation, showing a high coherence value at the fault characteristic frequency. The first spectral function can screen out those frequency components that exist in both heterogeneous signals and have strong correlation from the perspective of statistical correlation, providing data basis for fault feature extraction.

[0091] Specifically, early localized damage to bearings or gears generates periodic impacts, which modulate at high-frequency resonant frequencies such as gear meshing frequencies. By demodulating the signal and extracting the envelope, clear low-frequency fault characteristics can be obtained. The vibration and acoustic signals are demodulated separately, and their envelope signals are extracted. The analytic signal is calculated using Hilbert transform, and the modulus of the analytic signal is used as the envelope of the original signal. After obtaining the vibration and acoustic envelope signals, Fourier transforms are performed on the two envelope signals to obtain the envelope spectra. The coherence function between the envelope spectra is calculated, and the corresponding coherence function curve is obtained as the second spectral function, reflecting the correlation between the vibration and acoustic signals at the envelope spectrum level.

[0092] It should be noted that for minor faults such as early pitting of bearing raceways or slight cracks in gear tooth roots, the resulting impact energy is low and easily obscured by the main vibration components in the original spectrum. By extracting and analyzing the coherence of the envelope spectrum, fault information can be separated from the high-frequency carrier and correlation analysis can be performed on the fault characteristic frequencies. The second spectral function provides an analytical perspective that complements the first spectral function, which can identify amplitude-modulated fault modes and improve the sensitivity and range of fault types in fault detection.

[0093] Specifically, different fault modes or different stages of the same fault exhibit varying degrees of severity in linear spectral correlation and envelope spectral correlation. By fusing the two coherence functions, missed detections due to the limitations of a single criterion can be avoided. For each frequency point within the first region, the first and second spectral functions are fused. Weights are assigned based on the sensitivity of the two coherence functions to different fault types in historical data. The values ​​of the corresponding points of the first and second spectral functions are then weighted and summed according to these weights to obtain the signal correlation function.

[0094] Preferably, based on the statistical analysis of the comprehensive signal correlation function calculated from the equipment operating under numerous fault-free conditions, a correlation threshold is set, for example, 0.8. The first frequency band is scanned, and frequency bands where the signal correlation function value continuously exceeds the preset correlation threshold are selected. These bands correspond to characteristic frequency bands where the vibration and acoustic signals are highly correlated at both the spectral and envelope levels, and are highly likely to be generated by the same internal fault source. The center frequency, bandwidth, and corresponding peak correlation values ​​of these characteristic frequency bands are integrated to obtain the first fault set. Information fusion combines fault correlation to improve detection sensitivity, and threshold judgment can suppress the randomness in single coherent analysis, improving the reliability of fault detection.

[0095] Furthermore, based on the analysis of electrical load fluctuations using current signals, the correlation coefficient between current changes and vibration signal changes is calculated. The portion of the correlation coefficient that deviates from the preset correlation range is identified, and a second fault set is constructed, including:

[0096] S401. Analyze the electrical load fluctuation based on the current signal, and calculate the correlation coefficient between the current change and the vibration signal change under each operating condition, in conjunction with the real-time operating conditions of the transportation equipment.

[0097] S402. Identify the parts where the correlation coefficient deviates from the preset correlation range and construct a second fault set.

[0098] In this embodiment, the electrical load fluctuation is analyzed based on the current signal, and the correlation coefficient between the current change and the vibration signal change under each operating condition is calculated in conjunction with the real-time operating conditions of the transportation equipment. By analyzing each operating condition for fault detection, the accuracy and reliability of fault detection under variable load conditions are improved. By combining the current signal analysis with the load and establishing dynamic correlation benchmarks for different operating conditions, it is possible to quickly distinguish between normal vibration increase caused by increased load and abnormal vibration increase caused by mechanical failure. When the mechanical system is healthy, even if the load changes drastically, the current difference and vibration difference remain highly correlated. When a mechanical failure occurs, the vibration will increase abnormally or a specific impact will occur, causing the calculated correlation coefficient to decrease.

[0099] Specifically, during the monitoring period when the equipment is in a healthy state, a large amount of data is collected under various operating conditions. The correlation coefficient of the operating conditions under the healthy state is calculated, and these data are statistically analyzed to calculate the mean and standard deviation, setting a preset correlation range for each defined operating condition. For example, for the heavy-load constant-speed operating condition, the health correlation coefficient is set between 0.75 and 0.95. In real-time fault detection, after each operating condition segment is analyzed and the current correlation coefficient is calculated, it is compared with the preset correlation range of the corresponding operating condition. If the current correlation coefficient falls outside the correlation range, it is determined to be a correlation deviation, reflecting that the normal coupling relationship between current and vibration is disrupted during this period, and there are non-load factors interfering with the normal operation of the system. The timestamp, operating condition type, degree of deviation, and corresponding current and vibration characteristics of the operating condition segment where the correlation deviation occurs are recorded to construct a second fault set.

[0100] It should be noted that the first fault set is based on the correlation of acoustic and vibration signals, identifying faults from the perspective of signal pattern consistency; the second fault set is based on the correlation deviation of motor signals, identifying faults from the perspective of abnormal system input-output interaction. These two methods, based on different physical principles, complement and verify each other. When a certain time period or location is simultaneously marked by both the first and second fault sets, the confidence level of the existence of a mechanical fault at that location will be greatly increased, improving the accuracy of fault detection results.

[0101] like Figure 3 As shown, based on the analysis of electrical load fluctuations using current signals and combined with the real-time operating conditions of the transportation equipment, the correlation coefficient between current changes and vibration signal changes under each operating condition is calculated, including:

[0102] S501. According to the real-time operating conditions of the transportation equipment, the current signal and vibration signal are divided into signal segments under the corresponding operating conditions.

[0103] S502. For each working condition, remove the reference data under normal conditions from the current signal and vibration signal in the signal segment to obtain the current difference sequence and vibration difference sequence.

[0104] S503. Analyze electrical load fluctuations based on current difference sequences, and calculate the mutual information value and correlation coefficient between the current difference sequence and the vibration difference sequence;

[0105] S504. Based on the mutual information value and correlation coefficient, cross-validation and correlation analysis are performed on the envelope spectrum characteristics of the current characteristics and vibration signals to obtain the corresponding correlation coefficients.

[0106] In this embodiment, the real-time operating conditions of the transportation equipment are determined based on the vehicle speed, gear position, throttle opening, braking status signals, and direct measurement signals from sensors in the transportation equipment control system. By analyzing whether the vehicle speed is stable, the throttle opening, and the gradient information, the operating conditions are divided into unloaded constant speed driving, heavy-load constant speed driving, acceleration, and deceleration. The system continuously monitors these state variables. When a state combination is detected to meet the definition of a preset operating condition and remain stable for more than a preset time window, such as 2 seconds, the equipment is considered to have entered a stable operating condition stage. The current and vibration signals synchronously collected during this time period are divided into corresponding signal segments, and each signal segment is labeled with a corresponding operating condition tag.

[0107] It should be noted that by identifying operating conditions and segmenting data, the operating states of transportation equipment are distinguished, ensuring the correlation between current and vibration in subsequent calculations. This enables the establishment of independent and accurate health correlation benchmarks for each operating state, improving the adaptability of the fault detection process to changes in the actual operating state of the equipment.

[0108] Specifically, during the healthy operation phase of the equipment, a large number of signal segments are collected for each operating condition. Through statistical analysis, the average value at each time point is used to construct a standard current reference curve and vibration reference curve for each operating condition. The reference curve represents the baseline shape of the current and vibration signals of healthy equipment changing over time under that operating condition. In real-time fault detection, for each newly acquired signal segment tagged with the operating condition, the corresponding reference curve for that operating condition is invoked. The data points of the current current signal segment are subtracted point by point from the values ​​at the corresponding time points on the current reference curve to obtain a current difference sequence, reflecting the degree to which the current load fluctuation deviates from the typical healthy state. Similarly, the current vibration signal segment is subtracted point by point from the vibration reference curve to obtain a vibration difference sequence, reflecting the degree to which the current mechanical vibration response deviates from the typical healthy state. The difference sequence filters out the inherent static characteristics of the operating condition and can reflect the deviation between the current operating transient and the healthy standard transient.

[0109] It should be noted that by introducing a baseline operating condition and performing interpolation calculations, the interference of baseline drift caused by slow system aging or mild environmental changes between different operating conditions and within the same operating condition can be eliminated from the correlation analysis. When the mechanical system is healthy, instantaneous load fluctuations will immediately cause proportional fluctuations in the mechanical response, and the current difference and vibration difference should be highly correlated. If a fault exists in the mechanical system, the vibration response will contain additional components unrelated to current fluctuations, leading to a decrease in the correlation between the two interpolation sequences; the current difference sequence and vibration difference sequence can be used to detect whether electromechanical coupling is detuned.

[0110] Specifically, mutual information measures the amount of information shared between two sequences, i.e., knowing the value of one sequence reduces the uncertainty about the value of the other. Correlation coefficient measures the strength and direction of the linear relationship between two sequences. To calculate mutual information, the continuous current difference and vibration difference sequences are discretized into several small intervals based on their numerical range. The probability that the values ​​of both sequences simultaneously fall within two specific intervals is calculated, yielding a joint probability distribution and their respective marginal probability distributions. Based on the joint and marginal probability distributions, the mutual information value is calculated; a larger value indicates a stronger statistical dependence between the two sequences. The correlation coefficient is obtained by calculating the ratio of the product of the covariance of the two difference sequences to their respective standard deviations. Mutual information reflects the overall statistical dependence, while the correlation coefficient reflects the consistency of the linear trend.

[0111] It should be noted that by calculating mutual information values ​​and correlation coefficients, it is possible to combine the nonlinear dependencies and linear trend information between data to comprehensively analyze the normality or abnormality of the relationship between current fluctuations and vibration fluctuations, thereby improving the accuracy and robustness of the correlation analysis results. For example, under healthy conditions, both exhibit high mutual information values ​​and high correlation coefficients; under fault conditions, mutual information values ​​may change due to the presence of nonlinear components, and correlation coefficients may decrease due to phase misalignment or the presence of irrelevant vibration components.

[0112] Specifically, envelope spectrum features are extracted from the vibration difference sequence. A Hilbert transform is performed on the vibration difference sequence to extract the envelope, and a Fourier transform is performed on the envelope signal. The analysis examines whether components in the envelope spectrum exhibit frequencies consistent with the main fluctuation frequencies in the current difference sequence, and calculates the envelope spectrum consistency score. For example, if the current fluctuation has a significant low-frequency period, a healthy transmission system will generate the same modulation frequency in the vibration envelope. Based on historical data analysis of the sensitivity of each indicator to different early faults, weights are assigned. The mutual information value, correlation coefficient, and envelope spectrum consistency score are weighted and summed according to these weights to obtain the correlation coefficient, which reflects the overall health of the correlation between current and vibration fluctuations, assessed from multiple aspects including statistical dependence, linear relationship, and characteristic frequency matching.

[0113] It should be noted that by integrating information from three dimensions—mutual information, correlation coefficient, and envelope spectrum consistency—the correlation coefficient can more comprehensively and stably reflect the electromechanical coupling state, improve the detection capability of potential mechanical fault modes, and enhance the coverage and reliability of fault detection.

[0114] Furthermore, by combining the first fault set, the second fault set, and image data, the gear wear of the transportation equipment is analyzed to obtain fault results, including:

[0115] S601. Combining the first fault set, the second fault set, and the image data, the data is mapped to the real-time operating location and track section of the transportation equipment to obtain a fault feature sequence after position alignment.

[0116] S602. Analyze the gear wear of the transportation equipment based on the fault characteristic sequence to obtain the fault results.

[0117] In this embodiment, based on the timestamp and location of each data point, the mileage encoder on the transportation equipment can output the coordinates of the equipment on the track in real time. For the first fault set and the second fault set, the equipment mileage location at the corresponding time is queried in the location time series according to the corresponding timestamp, and a track location is added for each fault record. The image data is mapped to the corresponding track location, and the first fault set, the second fault set, and the image data are integrated according to the track segments they are mapped to to obtain a fault feature sequence after position alignment. Each element of this sequence corresponds to a track segment.

[0118] It should be noted that by aligning the positions, it is possible to clearly observe whether the fault characteristics recur at specific locations on the track, distinguish between vibrations caused by track excitation and vibrations caused by damage to the equipment itself, and provide spatial position support for the subsequent analysis process to associate signal characteristics with specific physical components.

[0119] Specifically, the wear of gears in transportation equipment is analyzed based on fault characteristic sequences to obtain fault results. By combining acoustic and vibration characteristics, electromechanical correlations, and visual defects, misjudgments caused by individual sensor failures, instantaneous strong interference, or limitations of single analysis methods can be effectively eliminated. Fault detection results with clear location, type, severity, and visual images can be obtained, reducing unplanned downtime, optimizing spare parts inventory, and ensuring accurate and efficient maintenance work.

[0120] Furthermore, based on the fault characteristic sequence analysis of the gear wear of the transportation equipment, the fault results are obtained, including:

[0121] S701. Analyze the consistency of the first fault set and the second fault set at the corresponding positions based on the fault feature sequence, analyze the correspondence between the fault position and the gear corresponding to the feature frequency, filter out the fault positions that are consistent with the gear, and obtain the fault sequence.

[0122] S702. Perform defect identification on the image data corresponding to the fault sequence to obtain the corresponding defect information;

[0123] S703. By combining the fault sequence and defect information, analyze the gear wear of the transportation equipment to obtain the fault results.

[0124] In this embodiment, for each track position segment in the fault feature sequence, a first fault set and a second fault set are analyzed. A position is considered to have passed the consistency check only when both a valid first fault set indication exists (e.g., at least one characteristic frequency coherence value exceeds a high threshold) and a valid second fault set indication exists (e.g., the motor correlation coefficient is significantly lower than the lower limit of the healthy operating range). Fault source mapping is performed on the positions that pass the consistency check. The characteristic frequencies extracted from the first fault set are analyzed and matched with the meshing characteristic frequencies of each gear in the transmission chain of the transport equipment. For example, if the number of teeth of gear G1 in the gearbox is Z1 and the real-time rotational speed of its shaft is n, then its meshing characteristic frequency is (Z1×n) / 60. If a certain characteristic frequency is close to the meshing characteristic frequency or its harmonics, or if there is a sideband with the meshing characteristic frequency as the carrier frequency and a certain low frequency as the modulation edge, then the characteristic frequency is considered to correspond to a fault in gear G1. The fault positions that can be clearly mapped to specific gears are recorded. All positional information that has passed multi-source consistency verification and can be clearly mapped to specific gears is arranged in order of track position to obtain a fault sequence, including the faulty gear identifier and its precise track position.

[0125] It should be noted that by requiring two diagnostic methods based on different physical mechanisms to issue alarms simultaneously, false alarms caused by temporary failure of a single sensor, instantaneous strong environmental interference, or limitations of a single analysis method can be effectively filtered out. Matching the characteristic frequency with the meshing characteristic frequency of gear theory improves the accuracy and comprehensiveness of fault sequences, thereby improving the efficiency and accuracy of the fault detection process.

[0126] Specifically, based on the track positions recorded in the fault sequence, high-resolution images of the gear meshing area at the corresponding locations, either pre-collected or triggered in real-time, are retrieved. A pre-trained deep learning image recognition model analyzes the image data at the corresponding locations in the fault sequence. This deep learning image recognition model includes, but is not limited to, object detection and classification models based on convolutional neural networks. The training dataset for this model includes: synthetic images with various gear wear textures generated using simulation software, and actual images collected after applying different degrees of wear to real gears on an experimental bench. The model takes a local image of the gear as input and outputs the probability of a defect in the image and a defect type classification, including normal, pitting, and spalling. Cross-entropy is used as the loss function during training, and stochastic gradient descent is employed for optimization. The initial learning rate is set to 0.001, the batch size to 32, and training is iterative until the loss converges. By combining the fault component location and probability information provided by the fault sequence with the visual defect evidence and type information provided by the deep learning image recognition model, the fault result is obtained, such as moderate spalling wear on the gear.

[0127] Furthermore, for each entry in the fault sequence, for example, location P, suspected fault in gear G1, the defect information output by the image recognition model at location P is retrieved and fused for analysis. If image recognition confirms a defect in the gear G1 region, and the defect type matches the fault mode corresponding to the characteristic frequency in signal analysis, the corresponding fault result is obtained, including the faulty component, fault location, fault type, severity level, and confidence level. If image recognition does not find an obvious defect, but the signal consistency is extremely strong, the confidence level in the fault result will be lowered accordingly, and it will be marked as a suspected fault, requiring further manual review. All generated fault results are sorted by severity level and confidence level to obtain the final fault result.

[0128] It should be noted that by integrating signal anomalies with image information, the accuracy and comprehensiveness of fault results are improved. This allows equipment maintenance personnel to develop precise repair or replacement plans directly based on reports without having to perform complex analysis and guesswork. This reduces unnecessary downtime caused by misjudgment, avoids the risk of fault escalation due to missed diagnoses, and improves equipment availability, safety, and management efficiency.

[0129] Furthermore, the dual-control collaborative control mechanism includes:

[0130] S801. Based on the fault results and combined with the real-time driving information of the transport equipment, the movement control process of the transport equipment is dynamically allocated to one of the two driver's cabs.

[0131] S802. Based on the fault results, issue a fault warning to detect faults in the transportation equipment.

[0132] In this embodiment, based on the fault result and combined with the real-time driving information of the transportation equipment, the movement control process of the transportation equipment is dynamically allocated to one of the two cabs; through intelligent cab switching, the driver is placed in a relatively safe operating environment, which can effectively isolate the direct interference of the fault to the driver, prevent secondary accidents caused by improper human emergency response, and enhance the stability and safety of the transportation equipment under complex working conditions.

[0133] Specifically, simultaneously with the generation of control commands in step S801, the system immediately triggers a multimodal warning. The warning actions are categorized according to the severity of the fault. For low-severity faults, a text prompt appears on the driver's cab display screen along with an audible alert. For medium-severity faults, or when control is switched, in addition to highlighting the information on the display screen, a clear audible and visual alarm is triggered to attract the driver's attention. For high-severity faults, the highest level alarm is activated, including a continuous, strong audible and visual alarm. The warning information is automatically transmitted to the ground remote monitoring center via the wireless communication module. Through fault warnings, the driver can immediately understand the current system status and take timely countermeasures, ensuring timely fault response, guaranteeing safe human-machine collaborative operation, and improving the safety and stability of the transportation equipment's operation.

[0134] Furthermore, based on the fault results and combined with the real-time driving information of the transport equipment, the movement control process of the transport equipment is dynamically allocated to one of the two driver's cabs, including:

[0135] S901. Based on the fault type and fault location in the fault results, analyze the scope of the fault's impact, calculate the risk of the fault's impact on the front cab and the rear cab respectively, and obtain the corresponding risk level.

[0136] S902. Set the cab with the lower risk level between the front cab and the rear cab as the main cab, and set driving switching instructions based on the current risk level of the cab, combined with the real-time driving information of the transportation equipment.

[0137] S903. According to the driving switching command, the movement control process of the transport equipment is dynamically allocated to one of the two driver's cabs.

[0138] In this embodiment, the system has a pre-built fault impact knowledge base, which includes the potential impact attributes of various faults. For example, for faults like gear peeling, the main impacts are periodic impact vibrations and high-frequency abnormal noises. Based on the fault type, physical location, and severity level in the fault results, the spatial distances from the fault point to the front and rear cabs are calculated. Combining the reciprocal of the spatial distance with the severity level, the severity level is mapped to a corresponding severity intensity value. The reciprocal of the spatial distance and the severity intensity value are summed to calculate the risk score of the fault on the front and rear cabs, respectively. The risk score is then mapped to a preset risk level, divided into low, medium, and high levels: a risk score below 10 is low, 10 to 30 is medium, and above 30 is high. By comprehensively considering the fault type, location, and severity, the impact of the fault can be accurately analyzed, providing accurate data support for the subsequent selection of the optimal driver's cab and avoiding suboptimal decisions or ineffective switching due to coarse evaluation.

[0139] Specifically, the construction process of the fault impact knowledge base includes: collecting equipment fault maintenance records accumulated from years of mining operations and destructive test data from simulated test benches to classify and statistically analyze various common faults (such as gear spalling, bearing pitting, gear tooth breakage, and motor stator short circuits); using finite element simulation and multibody dynamics analysis to simulate the propagation laws of physical fields such as vibration, noise, and temperature rise caused by different faults under different operating conditions, quantifying the degree of impact on the cab position; and weighting and correcting the data to obtain the risk scores of the faults on the front and rear cabs. The risk scores range from 0 to 50, with below 10 indicating low risk, 10 to 30 indicating medium risk, and above 30 indicating high risk.

[0140] Specifically, the risk levels of the front and rear cabs for the same fault are compared, and the cab with the lower risk level is selected as the candidate primary cab. Verification is performed based on safety margins and real-time status. When the risk level of the target cab is at least one full level lower than the current cab's risk level, a switch is considered to have a clear safety benefit and is necessary to trigger. If both risk levels are the same, the current control is maintained to avoid unnecessary switching disturbances. Simultaneously, the control process is verified using real-time driving information from the transport equipment. The presence of a driver in the target primary cab is determined by the cab seat pressure sensor or the door opening / closing status. If no driver is present, the target cab cannot be designated as the primary cab. If the vehicle is undergoing critical maneuvers such as emergency braking or high-speed cornering, the switch command is delayed until the vehicle's condition stabilizes before execution. The final driving switch command is generated, including the target cab, switch type, and relevant fault information. By analyzing the risk difference and the real-time status of the transport equipment, it is ensured that each control switch is necessary and safe, preventing frequent or ineffective switches due to minor risk differences or temporary driver absence. This ensures the continuity of driving operations and improves the safety and stability of the transport equipment's operation.

[0141] Specifically, the vehicle controller receives the driver switching command and broadcasts the command over the network to notify all subsystems of the impending change of control. It then authorizes the control input interface of the target driver's cab, allowing its input signals to be received and processed. Simultaneously, it revokes the permissions of the original driver's cab's corresponding control interface, causing its input signals to be ignored. The power system, braking system, and other systems switch their internal control command sources to the corresponding driver's cab controller channel based on the new control assignment. After each subsystem completes the switch, it sends a confirmation signal to the vehicle controller. Upon receiving confirmation from all critical subsystems, the vehicle controller determines the switch was successful and updates its internal current master driver's cab state variables. The entire switching process should be completed within a very short time to ensure the continuity of vehicle dynamic control. During the switch, for critical systems such as power and braking, a hold strategy can be adopted, such as maintaining the current motor torque request unchanged at the moment of switching to avoid sudden power interruptions or increases.

[0142] It should be noted that through the centralized coordination of vehicle network protocols and vehicle controllers, multiple control input sources can be managed safely and orderly. The rapid switching speed minimizes system state interruption time, ensuring the stability of vehicle operation. The smooth transition strategy avoids the introduction of new driving shocks or safety hazards due to control switching, thus improving the safety and stability of the transportation equipment's operation.

[0143] like Figure 4 As shown, a fault detection system for a single-rail, dual-operator, light and narrow-type mining transport equipment is used to implement a fault detection method for such equipment, including:

[0144] The data acquisition module collects real-time operating data of the mining monorail dual-drive light narrow-type transport equipment based on multiple sensors deployed on the equipment. The operating data includes vibration signals, acoustic signals, current signals, and image data.

[0145] The first fault analysis module calculates the signal correlation function of vibration signal and acoustic signal in the frequency domain, filters out the frequency band portion where the signal correlation is greater than the preset correlation threshold, and constructs the first fault set.

[0146] The second fault analysis module analyzes the electrical load fluctuation based on the current signal, calculates the correlation coefficient between the current change and the vibration signal change, identifies the part of the correlation coefficient that deviates from the preset correlation range, and constructs a second fault set.

[0147] The fault result analysis module combines the first fault set, the second fault set, and image data to analyze the gear wear of the transportation equipment and obtain the fault results.

[0148] The fault warning module is equipped with a dual-control collaborative control mechanism. According to the fault result, it dynamically allocates the movement control process of the transportation equipment to one of the two driver's cabs and issues a fault warning to detect faults in the transportation equipment.

[0149] Example 3:

[0150] This embodiment describes the overall process of the method in conjunction with a specific application scenario. Taking a single-rail, dual-driver, light and narrow-type transportation equipment used in an underground mine as the implementation object, after completing the transportation task for the shift, the system initiates a periodic self-check process on the way back to the yard empty.

[0151] When the equipment is running at a constant speed under no-load conditions, vibration acceleration sensors located on the drive motor bearing housing and gearbox housing collect vibration signals in real time at a sampling frequency of 16,000 times per second. Simultaneously, a weather-resistant microphone installed near the gearbox collects acoustic signals, a current sensor in the motor circuit collects the operating current, and an endoscope camera installed near the gearbox observation port periodically captures images of the gear meshing area. All data is transmitted to the onboard central processing unit after being stamped with a uniform timestamp and location information from the onboard odometer encoder. For example, within the mileage calibration range of K253+150 meters to K253+200 meters, the system collects and caches all sensor data corresponding to this range.

[0152] The onboard central processing unit processes the vibration and acoustic signals within this section. By performing a Fast Fourier Transform on both signals and comparing them with pre-stored reference spectra under constant speed and no-load conditions when the equipment is healthy, it was found that the spectral amplitudes of both vibration and acoustic signals consistently exceeded twice their corresponding reference amplitudes near 1250 Hz and 2500 Hz. These frequency bands were identified as candidate regions for in-depth analysis. The cross-spectral coherence functions of the vibration and acoustic signals were calculated within these candidate frequency bands, along with the coherence functions of their envelope signals. The calculation results showed that at 1250 Hz, both coherence function values ​​exceeded 0.92, and at 2500 Hz, the coherence function value also exceeded 0.88. With a preset coherence threshold of 0.8, the 1250 Hz and 2500 Hz frequencies were identified as abnormally relevant frequencies and recorded, forming the first fault set indicating abnormal mechanical conditions. Based on the design parameters of the transmission system, the system calculations show that 1250 Hz is exactly the theoretical value of the meshing characteristic frequency of the first-stage reduction gear of the drive axle.

[0153] Simultaneously, the system analyzes the current signal within the same section. Under the stable operating condition of no-load uniform speed, the current signal should be very stable. Comparing the current signal with the health baseline curve, its fluctuation amplitude is found to be within a reasonable range, indicating that there is no abnormal fluctuation in the electrical load. When calculating the correlation coefficient between current fluctuation and vibration fluctuation within this section, the obtained value is only 0.3. According to historical health data statistics, under the no-load uniform speed condition, the normal range of this correlation coefficient should be between 0.7 and 0.9. The calculated value of 0.3 is significantly lower than the lower limit of the normal range, reflecting that the normal coupling relationship between mechanical vibration and driving current has broken down. This abnormal event is recorded, forming a second fault set, reflecting from an electromechanical correlation perspective that there is a fault in the mechanical system independent of the load.

[0154] The system performs information fusion and localization, mapping the abnormal frequency (1250 Hz) found in the first fault set to the electromechanical correlation anomaly events recorded in the second fault set, based on their timestamps, to a specific track location between K253+150 meters and K253+200 meters. The two locations completely overlap spatially, and the abnormal frequency points to a specific gear component. This cross-dimensional consistency of evidence enhances the credibility of a genuine fault at this location. The system automatically retrieves gear images taken at this location and inputs them into a trained deep learning image recognition model for analysis. This model, previously trained on a large amount of image data of healthy and various defective gears, can identify defects such as pitting and spalling on the gear surface. The model's analysis output for the current image is: spalling defects exist on the gear meshing surface, with a confidence level of 96%. The triple evidence from acoustic vibration analysis, electromechanical correlation analysis, and visual analysis forms a complete chain of evidence.

[0155] Based on the above fusion analysis, the corresponding fault result was generated: At track location K253+175 meters, the first-stage reduction gear of the drive axle experienced moderate spalling wear. The dual-control collaborative control mechanism was activated. The system assessed that the fault location was adjacent to the front cab of the equipment, and the risk of vibration and abnormal noise caused by the gear spalling affecting the operating environment of the front cab was rated as high, while the risk affecting the rear cab was rated as medium. Combining real-time information confirming that there were drivers in both the front and rear cabs, control was switched to the lower-risk rear cab. The vehicle controller executed the switching command within milliseconds, transferring core control authority such as drive and braking to the rear cab control panel. Simultaneously, a clear warning message popped up on the displays in both the front and rear cabs: "Drive gear spalling detected; control has automatically switched to the rear cab. Please decelerate smoothly and proceed to the repair point." The entire process, from data acquisition, multi-dimensional analysis, fusion diagnosis to active safety control, forms a complete closed loop, realizing a complete process from condition monitoring to predictive maintenance and then to active safety protection.

[0156] The above description is merely a preferred embodiment of this application. The scope of protection of this application is not limited to the above embodiments. All technical solutions falling within the scope of this application's concept are within the scope of protection of this application. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this application should also be considered within the scope of protection of this application.

Claims

1. A fault detection method for a single-rail, dual-operator, light and narrow-type mining transport equipment, characterized in that, include: Based on multiple sensors deployed on the mining monorail dual-drive light narrow-type transport equipment, the operating data of the transport equipment is collected in real time, including vibration signals, acoustic signals, current signals and image data; Calculate the signal correlation function of vibration signal and acoustic signal in the frequency domain, filter out the frequency band part with signal correlation greater than the preset correlation threshold, and construct the first fault set; Based on the analysis of electrical load fluctuations using current signals, the correlation coefficient between current changes and vibration signal changes is calculated, the portion of the correlation coefficient that deviates from the preset correlation range is identified, and a second fault set is constructed. By combining the first fault set, the second fault set, and image data, the gear wear of the transportation equipment is analyzed to obtain fault results. A dual-control collaborative control mechanism is configured to dynamically allocate the movement control process of the transportation equipment to one of the two driver's cabs according to the fault result, and to provide fault warnings in order to detect faults in the transportation equipment.

2. The fault detection method for a single-rail, dual-operation, light and narrow-type mining transport equipment according to claim 1, characterized in that, The calculation of the signal correlation function between the vibration signal and the acoustic signal in the frequency domain, the selection of frequency bands with signal correlation greater than a preset correlation threshold, and the construction of a first fault set include: The vibration signal and acoustic signal are synchronously Fourier transformed and compared with the corresponding reference data under the stable unloaded operation of the transportation equipment to identify the first region with energy higher than the reference. The signal correlation function of vibration signal and acoustic signal in the frequency domain is calculated based on the first region. The frequency bands with signal correlation greater than the preset correlation threshold are selected to construct the first fault set.

3. The fault detection method for a mine monorail dual-operator light narrow-type transportation equipment according to claim 2, characterized in that, The step involves calculating the signal correlation function of the vibration signal and the acoustic signal in the frequency domain based on the first region, filtering out the frequency bands with signal correlation greater than a preset correlation threshold, and constructing a first fault set, including: The cross-spectral coherence function of the vibration signal and the acoustic signal in the frequency domain is calculated in the first region to obtain the first spectral function; The vibration signal and acoustic signal are demodulated and the corresponding envelope signals are extracted. The coherence function between the envelope signals is calculated to obtain the second spectral function. By fusing the first spectral function and the second spectral function, a signal correlation function is obtained. Frequency bands with signal correlation greater than a preset correlation threshold are selected to construct the first fault set.

4. The fault detection method for a single-rail, dual-operation, light and narrow-type mining transport equipment according to claim 1, characterized in that, The process involves analyzing electrical load fluctuations based on current signals, calculating the correlation coefficient between current changes and vibration signal changes, identifying portions of the correlation coefficient that deviate from a preset range, and constructing a second fault set, including: Based on the analysis of electrical load fluctuations using current signals, and combined with the real-time operating conditions of the transportation equipment, the correlation coefficient between current changes and vibration signal changes under each operating condition is calculated. Identify the parts where the correlation coefficient deviates from the preset correlation range and construct a second fault set.

5. The fault detection method for a mine monorail dual-operator light narrow-type transportation equipment according to claim 4, characterized in that, The process involves analyzing electrical load fluctuations based on current signals, combining this with the real-time operating conditions of the transportation equipment, and calculating the correlation coefficient between current changes and vibration signal changes under each operating condition, including: According to the real-time operating conditions of the transportation equipment, the current signal and vibration signal are divided into signal segments under the corresponding operating conditions. For each operating condition, the reference data under normal conditions are removed from the current signal and vibration signal in the signal segment to obtain the current difference sequence and vibration difference sequence; Based on the analysis of electrical load fluctuations using current difference sequences, the mutual information value and correlation coefficient between the current difference sequence and the vibration difference sequence are calculated. Based on the mutual information value and correlation coefficient, cross-validation and correlation analysis are performed on the envelope spectrum characteristics of current characteristics and vibration signals to obtain the corresponding correlation coefficients.

6. The fault detection method for a mine monorail dual-operator light narrow-type transportation equipment according to claim 1, characterized in that, The method combines the first fault set, the second fault set, and image data to analyze the gear wear of the transportation equipment and obtain fault results, including: By combining the first fault set, the second fault set, and image data, the data is mapped to the real-time operating location and track section of the transportation equipment to obtain a fault feature sequence after position alignment. The failure results are obtained by analyzing the gear wear of the transportation equipment based on the failure characteristic sequence.

7. The fault detection method for a mine monorail dual-operator light narrow-type transportation equipment according to claim 6, characterized in that, The analysis of gear wear in the transportation equipment based on the fault characteristic sequence yields the following fault results: Based on the fault feature sequence analysis, the consistency between the first fault set and the second fault set at the corresponding position is analyzed. The correspondence between the characteristic frequency and the fault position and the gear is analyzed. The fault positions with consistent faults and corresponding gears are selected to obtain the fault sequence. Defect identification is performed on the image data corresponding to the fault sequence to obtain the corresponding defect information; By combining fault sequences and defect information, the gear wear of the transportation equipment is analyzed to obtain fault results.

8. The fault detection method for a single-rail, dual-operation, light and narrow-type mining transport equipment according to claim 1, characterized in that, The dual-control collaborative control mechanism includes: Based on the fault results and combined with the real-time driving information of the transportation equipment, the movement control process of the transportation equipment is dynamically allocated to one of the two driver's cabs; Fault warnings are issued based on the fault results to detect faults in transportation equipment.

9. The fault detection method for a single-rail, dual-operation, light and narrow-type mining transport equipment according to claim 8, characterized in that, The process of dynamically allocating the movement control process of the transportation equipment to one of the two driver's cabs based on the fault result and in conjunction with the real-time driving information of the transportation equipment includes: Based on the fault type and location in the fault results, analyze the scope of the fault's impact, calculate the risk of the fault's impact on the front and rear cabs respectively, and obtain the corresponding risk levels. The driver's cab with the lower risk level is set as the main driver's cab between the front and rear cabs. Based on the real-time driving information of the transportation equipment and the current risk level of the driver's cab, driving switching instructions are set. According to the driving switching command, the movement control process of the transport equipment is dynamically assigned to one of the two driver's cabs.

10. A fault detection system for a single-rail, dual-operator, light and narrow-type mining transport equipment, characterized in that: The method for detecting faults in a mine monorail dual-driver light narrow-type transportation equipment as described in any one of claims 1 to 9 includes: The data acquisition module collects real-time operating data of the mining monorail dual-drive light narrow-type transport equipment based on multiple sensors deployed on the equipment. The operating data includes vibration signals, acoustic signals, current signals, and image data. The first fault analysis module calculates the signal correlation function of vibration signal and acoustic signal in the frequency domain, filters out the frequency band portion where the signal correlation is greater than the preset correlation threshold, and constructs the first fault set. The second fault analysis module analyzes the electrical load fluctuation based on the current signal, calculates the correlation coefficient between the current change and the vibration signal change, identifies the part of the correlation coefficient that deviates from the preset correlation range, and constructs a second fault set. The fault result analysis module combines the first fault set, the second fault set, and image data to analyze the gear wear of the transportation equipment and obtain the fault results. The fault warning module is equipped with a dual-control collaborative control mechanism. According to the fault result, it dynamically allocates the movement control process of the transportation equipment to one of the two driver's cabs and issues a fault warning to detect faults in the transportation equipment.

Images