An intelligent fault diagnosis method and system in an intelligent manufacturing process of a mine equipment

Abstract

The application relates to the technical field of component testing, and discloses an intelligent fault diagnosis method and system in an intelligent manufacturing process of a mine equipment. The method comprises the following steps: collecting running state information of the mine equipment in real time through a multi-source sensor; performing space coordinate registration after aligning vibration signals and acoustic emission signals through a time stamp to obtain a three-dimensional working condition atlas; decoupling features based on an equipment dynamics model to obtain a fault sensitive feature set; inputting the fault sensitive feature set into a multi-layer decision network to output a fault diagnosis result and a confidence score; constructing a fault cause-effect diagram based on pre-acquired design parameters and manufacturing process parameters; designing a correction scheme for abnormal process parameters in the fault cause-effect diagram, wherein the traceability confidence score of the abnormal process parameters is lower than a set threshold; and fusing the correction scheme, the fault diagnosis result and the three-dimensional working condition atlas to construct a visual diagnosis report. The application can improve the effect of intelligent fault diagnosis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of component testing, in particular to an intelligent fault diagnosis method and system in the intelligent manufacturing process of mine equipment. BACKGROUND

[0002] In the field of intelligent manufacturing of mine equipment, fault diagnosis relies on sensor collection of exogenous data for analysis. Although the existing technology can realize basic data collection and feature extraction, there are significant defects in the aspect of multi-modal data fusion. Due to the insufficient time synchronization accuracy of different types of data such as vibration signals and acoustic emission signals, and the lack of effective spatial coordinate registration mechanism, the device working condition model constructed is difficult to accurately reflect the real running state, thereby causing incomplete decoupling of fault features, a large amount of redundant information interfering with the diagnosis process, slow fault diagnosis speed, and inability to meet the real-time requirements of intelligent manufacturing.

[0003] Traditional fault diagnosis algorithms are mostly based on single physical models or shallow neural networks, and have limited ability to capture the nonlinear characteristics of complex dynamic systems. For example, it is difficult to effectively separate multi-source fault features such as gear meshing impact and bearing wear by analyzing the frequency components of vibration signals through Fourier transform; the diagnosis system based on rule engine cannot dynamically learn the implicit association between parameter changes and fault patterns during device operation, resulting in low diagnosis accuracy. At the same time, the existing methods lack the ability to deeply trace the causal relationship between manufacturing process parameters and faults, and cannot optimize the diagnosis logic from the whole process dimension of design, assembly, and processing, further restricting the overall efficiency of intelligent fault diagnosis. SUMMARY

[0004] The present application provides an intelligent fault diagnosis method in the intelligent manufacturing process of mine equipment, which mainly aims to solve the problem of poor intelligent fault diagnosis effect in the intelligent manufacturing process of mine equipment.

[0005] To achieve the above purpose, the present application provides an intelligent fault diagnosis method in the intelligent manufacturing process of mine equipment, which comprises:

[0006] S1. Collecting vibration signals, acoustic emission signals and thermal imaging data of mine equipment collected by multi-source sensors in real time as the running state information of the mine equipment;

[0007] S2. Aligning the vibration signals and the acoustic emission signals through time stamping, and then performing spatial coordinate registration to obtain a three-dimensional working condition map of the mine equipment;

[0008] S3. Decoupling the features of the running state information under physical constraints based on the device dynamics model to obtain a fault sensitive feature set of the mine equipment;

[0009] S4. inputting the fault-sensitive feature set into a multi-layer decision network to output a fault diagnosis result and a confidence score;

[0010] S5. constructing a fault causal diagram of the mine equipment based on pre-acquired design parameters and manufacturing process parameters, designing a correction scheme containing process constraint conditions for abnormal process parameters of the mine equipment with a traceability confidence score lower than a set threshold in the fault causal diagram;

[0011] S6. fusing the correction scheme, the fault diagnosis result and a three-dimensional working condition map to construct a visual diagnosis report.

[0012] In a preferred embodiment, the vibration signal, the acoustic emission signal and the thermal imaging data of the mine equipment are collected by a multi-source sensor in real time as the operation state information of the mine equipment, including:

[0013] The vibration acceleration sensor is deployed on the axial mounting surface of the bearing seat of the mine equipment to obtain the vibration signal of the mine equipment under full load working condition;

[0014] The acoustic emission sensor is arranged in a ring array form on the surface of the gear box shell to obtain the acoustic emission signal of the mine equipment;

[0015] The infrared thermal imager is installed in the radial observation area of the equipment friction area to realize dynamic tracking focusing of the thermal imaging data of the mine equipment through the stepping motor drive;

[0016] The vibration signal, the acoustic emission signal and the thermal imaging data are collected as the operation state information of the mine equipment.

[0017] In a preferred embodiment, after the vibration signal and the acoustic emission signal are aligned by time stamp, spatial coordinate registration is performed to obtain a three-dimensional working condition map of the mine equipment, including:

[0018] The vibration acceleration sensor and the acoustic emission sensor are used to collect information of the mine equipment after clock synchronization, when the delay of the collected information exceeds a set threshold, a sliding window interpolation is used to compensate the time sequence deviation of the collected information to obtain the vibration signal and the acoustic emission signal of the mine equipment;

[0019] The vibration signal and the acoustic emission signal are aligned by time stamp to obtain the vibration signal and the acoustic emission signal of the same frequency of the mine equipment;

[0020] A three-dimensional space framework is built based on spatial coordinates, and the vibration signals and acoustic emission signals of the same frequency and the thermal imaging data are arranged to the three-dimensional space framework according to spatial positions and time sequences, so as to obtain a three-dimensional working condition atlas of the mine equipment.

[0021] In a preferred embodiment, the feature decoupling of the operating state information under the physical constraint based on the equipment dynamics model obtains a fault sensitive feature set of the mine equipment, including:

[0022] According to the correlation between the structural characteristics, operating principles and physical parameters of the mine equipment, an equipment dynamics equation of the mine equipment is established;

[0023] Based on the equipment dynamics equation, the vibration signals are decomposed into a time-domain impact component related to gear meshing torque and a centrifugal force modulation component related to bearing wear;

[0024] The amplitude spectrum of the acoustic emission signal is extracted through Morlet wavelet transform, the gear tooth breakage impact frequency and its harmonic components are identified, and are associated to a device load rate change curve, so as to obtain a harmonic component of the mine equipment;

[0025] According to the temperature gradient distribution collected by the infrared thermal imager, unit area heat flow density distribution data of the mine equipment is generated;

[0026] The time-domain impact component, the centrifugal force modulation component, the harmonic component and the unit area heat flow density distribution data are collected as the fault sensitive feature set of the mine equipment.

[0027] In a preferred embodiment, the generation of the unit area heat flow density distribution data of the mine equipment according to the temperature gradient distribution collected by the infrared thermal imager includes:

[0028] The temperature distribution data of the friction contact area of the mine equipment is collected by the infrared thermal imager, and a real-time temperature gradient distribution map of the friction contact area is generated;

[0029] The rotational speed, load pressure and lubricant flow parameters of the rotor in the mine equipment are taken as boundary conditions for temperature field modeling, and the temperature gradient distribution map is mapped to the friction contact area according to the Fourier heat conduction law, so as to obtain a three-dimensional transient temperature field model of the friction contact area;

[0030] The friction force and the relative displacement in the three-dimensional transient temperature field model are multiplied to obtain the friction heat generation power of the friction contact area;

[0031] The friction heat generation power is distributed to the friction contact area, combined with grid division of the three-dimensional transient temperature field model, to generate unit area heat flow density distribution data of the mine equipment.

[0032] In a preferred embodiment, the inputting of the fault-sensitive feature set into the multi-layer decision network and outputting of fault diagnosis results and confidence scores include:

[0033] The time series dependence of the fault-sensitive feature set is analyzed by using a gated recurrent layer to output hidden Markov chain state transition probabilities of the mine equipment operation state.

[0034] The energy distribution characteristics of the vibration signal in the gear meshing line direction are captured by using a convolutional neural network to generate a two-dimensional spatial correlation graph of the mine equipment.

[0035] The hidden Markov chain state transition probabilities and the two-dimensional spatial correlation graph are aggregated to output fault diagnosis results and confidence scores of the mine equipment.

[0036] In a preferred embodiment, the construction of the fault causal diagram of the mine equipment based on the pre-acquired design parameters and manufacturing process parameters includes:

[0037] Bayesian estimation is performed on the cutting depth, tool feed rate, and spindle torque in the pre-acquired design parameters and manufacturing process parameters to obtain conditional probability distributions of the mine equipment in different states, the conditional probability distributions are filled into network parameters, and a Bayesian network model is established.

[0038] A hierarchical aggregation strategy is performed on the process parameters and fault labels in the pre-acquired design parameters and manufacturing process parameters to obtain a process-fault matrix of the mine equipment.

[0039] Markov chain Monte Carlo sampling is performed on the Bayesian network model set to be normally distributed and different types of faults in the process-fault matrix to obtain a fault causal diagram of the mine equipment.

[0040] In a preferred embodiment, the abnormal process parameters of the mine equipment with a traceability confidence score lower than a set threshold in the fault causal diagram are used to design a correction scheme containing process constraints, which includes:

[0041] If the fault causal diagram shows that the cutting depth is strongly related to vibration exceeding the standard, the tool feed rate is adjusted and a lubrication compensation condition is added;

[0042] If there is a causal feedback between the centrifugal force modal parameters and the bearing temperature, an assembly tolerance verification instruction is triggered;

[0043] When the heat loss rate is abnormal, the matching degree of the friction pair geometry is verified by finite element simulation.

[0044] In a preferred embodiment, the design of the correction scheme containing process constraints for the abnormal process parameters includes:

[0045] Limiting the path parameters of the cutter based on the maximum allowable cutting force of the mine equipment;

[0046] According to the gear box precision grade requirement of the mine equipment, the mine equipment assembly gap tolerance threshold is set;

[0047] In combination with the material heat treatment characteristics of the mine equipment, the bearing temperature rise critical value of the mine equipment is limited.

[0048] In a preferred embodiment, the fusion of the correction scheme, the fault diagnosis result and the three-dimensional working condition atlas constructs a visual diagnosis report, including:

[0049] After mapping the axial vibration amplitude of the three-dimensional working condition atlas to the color spectrum gradient, the frequency spectrum feature line in the fault diagnosis result is superimposed to the corresponding area of the thermal imaging diagram, and the influence path of the abnormal process parameters on the equipment performance is marked with dynamic vector arrows, to obtain the visual diagnosis report of the mine equipment.

[0050] Compared with the prior art, the present application has the following beneficial effects:

[0051] 1. The present application significantly improves the fault diagnosis speed through multi-source data space-time alignment and feature decoupling technology. The time-stamped vibration signal and acoustic emission signal are synchronized, and the time sequence deviation is compensated by sliding window interpolation, combined with spatial coordinate registration to construct a three-dimensional working condition atlas, realizing the accurate space-time alignment of multi-modal data. At the same time, based on the device dynamics model, the feature decoupling of the running state information is carried out under the physical constraint, forming a high-dimensional, low-redundancy fault sensitive feature set. This structured feature representation can be directly input into a multi-layer decision network, avoiding the complex feature engineering process in traditional methods, combining the parallel processing mechanism of gated recurrent layer and convolutional neural network, greatly shortening the diagnosis response time, meeting the real-time demand of intelligent manufacturing of mine equipment.

[0052] 2.The application realizes double-precision optimization through the synergistic mechanism of multi-layer decision network and fault causal diagram. The multi-layer decision network analyzes the time-dependent relationship through the gating recurrent layer to generate the hidden Markov chain state transition probability, and simultaneously uses the convolutional neural network to capture the spatial energy distribution characteristics of the vibration signal. The two are aggregated to output the fault diagnosis results containing the confidence score, effectively improving the recognition accuracy of multi-source faults under complex working conditions. In addition, the Bayesian network model and the process-fault matrix constructed based on the design parameters and manufacturing process parameters generate the fault causal diagram through Markov chain Monte Carlo sampling, which can probabilistically model the correlation between process parameters such as cutting depth and tool feed rate and faults, trace the fault root cause from the manufacturing process dimension, and dynamically correct the diagnosis logic, forming a closed-loop optimization link of "data acquisition-feature decoupling-intelligent diagnosis-process tracing", significantly reducing the false positive rate and missed diagnosis rate, and achieving systematic improvement of diagnosis accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 A flowchart of an intelligent fault diagnosis method in an intelligent manufacturing process of a mine equipment is provided for an embodiment of the application.

[0054] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0055] It should be understood that the specific embodiments described herein are merely intended to explain the application and are not intended to limit the application.

[0056] Embodiments of the present application provide an intelligent fault diagnosis method in an intelligent manufacturing process of a mine equipment. The execution subject of the intelligent fault diagnosis method in the intelligent manufacturing process of the mine equipment includes but is not limited to at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiments of the present application. In other words, the intelligent fault diagnosis method in the intelligent manufacturing process of the mine equipment can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to a single server, a server cluster, a cloud server, or a cloud server cluster, etc. The server can be a standalone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms, etc. basic cloud computing services.

[0057] REFERENCE Figure 1As shown, it is a flowchart of the intelligent fault diagnosis method in the intelligent manufacturing process of the mine equipment provided by an embodiment of the application. In this embodiment, the intelligent fault diagnosis method in the intelligent manufacturing process of the mine equipment comprises:

[0058] S1. Collecting vibration signals, acoustic emission signals and thermal imaging data of the mine equipment collected by the multi-source sensor in real time as the operation state information of the mine equipment;

[0059] In the embodiment of the application, the collecting vibration signals, acoustic emission signals and thermal imaging data of the mine equipment collected by the multi-source sensor in real time as the operation state information of the mine equipment comprises:

[0060] Deploying the vibration acceleration sensor on the axial mounting surface of the bearing seat of the mine equipment to obtain the vibration signal of the mine equipment under full load working condition;

[0061] Arranging the acoustic emission sensor in the form of a ring array on the surface of the gear box shell to obtain the acoustic emission signal of the mine equipment;

[0062] Installing the infrared thermal imager on the radial observation area of the equipment friction area to realize dynamic tracking focusing of the thermal imaging data of the mine equipment through the stepping motor drive;

[0063] Collecting the vibration signal, the acoustic emission signal and the thermal imaging data as the operation state information of the mine equipment.

[0064] Specifically, the axial mounting surface of the bearing seat of the mine equipment is used with appropriate installation tools, such as high-strength bolts and nuts, to stably install the vibration acceleration sensor in the specified position.

[0065] Further, after installation, ensure that the sensor is tightly attached to the axial mounting surface of the bearing seat without looseness or gap.

[0066] Further, when the mine equipment is in full load working condition, the vibration acceleration sensor starts to work, and the vibration generated by the bearing seat during operation is sensed by the built-in sensitive element.

[0067] Further, these sensitive elements can convert the physical quantity of vibration into an electrical signal, which is amplified, filtered and processed by the internal signal conditioning circuit of the sensor, and finally outputs the vibration signal that can be used for analysis.

[0068] Further, on the surface of the gear box shell, according to the layout requirement of the ring array form, use the special sensor installation glue to uniformly paste multiple acoustic emission sensors on the surface of the shell to form a ring array.

[0069] Further, the installation position and angle of each sensor are precisely measured and calibrated to ensure that the acoustic emission signals generated during the operation of the gearbox can be fully and accurately collected.

[0070] Further, when the mining equipment is running, the mechanical activities such as gear meshing and friction inside the gearbox will generate elastic waves, i.e. acoustic emission signals.

[0071] Further, these acoustic emission signals propagate in the gearbox housing, and the sensors arranged in the annular array can capture these signals and convert them into electrical signals. After preliminary processing by the signal processing module, the acoustic emission signals are output.

[0072] Further, in the friction area of the mining equipment, a radial observation area is determined, and the infrared thermal imager is installed on a suitable bracket near the area.

[0073] Further, by connecting the control system of the stepping motor and the infrared thermal imager, the tracking and focusing program is set.

[0074] Further, when the mining equipment is running, the friction area will generate different temperature distributions due to frictional heating. The infrared thermal imager uses infrared detectors to receive the infrared radiation emitted by the objects in the area and converts it into electrical signals.

[0075] Further, the electrical signals are converted and enhanced by the image processing algorithm to form thermal imaging data.

[0076] Further, the stepping motor accurately controls the movement and focusing of the lens of the infrared thermal imager according to the preset program and the feedback of the thermal imaging data, realizing dynamic tracking and focusing collection of the thermal imaging data of the friction area of the mining equipment.

[0077] Further, during the operation of the mining equipment, the vibration signals collected by the vibration acceleration sensor, the acoustic emission signals collected by the annular array sensor, and the thermal imaging data collected by the infrared thermal imager are transmitted to the data collection center through their respective transmission lines, such as dedicated data cables or wireless transmission modules.

[0078] Further, in the data collection center, the data acquisition card and data processing software are used to integrate and store these different types of signals and data, ultimately forming complete mining equipment operation status information for subsequent equipment status analysis and fault diagnosis.

[0079] In summary, the vibration signals can directly reflect the dynamic load and wear state of mechanical parts, the acoustic emission signals can capture transient impact events such as gear tooth breakage and crack propagation, and the thermal imaging data can reveal hidden faults such as abnormal heating of contact pairs.

[0080] Overall, the three types of signals respectively cover the typical fault modes of mine equipment from the dimensions of mechanical response, energy release and heat conduction characteristics, forming a fault characterization system of multi-physical field coupling, avoiding the one-sidedness of single signal diagnosis.

[0081] Overall, the directional deployment of the vibration acceleration sensor, the ring array layout of the acoustic emission sensor, and the dynamic focusing of the friction area of the infrared thermal imager are designed for key vulnerable parts of the equipment, ensuring strong correlation between the collected data and the fault characteristics.

[0082] Overall, the vibration signal collection under full load working condition and the dynamic tracking of thermal imaging data mechanism can cover the extreme working conditions of the equipment in actual operation, improve the adaptability of the data to complex working conditions, and provide more representative sample set for subsequent fault feature decoupling and diagnosis model training.

[0083] S2. After aligning the vibration signal and the acoustic emission signal by timestamp, spatial coordinate registration is performed to obtain a three-dimensional working condition atlas of the mine equipment;

[0084] In the embodiment of the present application, after aligning the vibration signal and the acoustic emission signal by timestamp, spatial coordinate registration is performed to obtain a three-dimensional working condition atlas of the mine equipment, including:

[0085] The vibration acceleration sensor and the acoustic emission sensor are used to collect information of the mine equipment after clock synchronization. When the delay of the information collection exceeds the set threshold value, a sliding window interpolation method is used to compensate the time sequence deviation of the collected information, and the vibration signal and the acoustic emission signal of the mine equipment are obtained.

[0086] The vibration signal and the acoustic emission signal are aligned by timestamp to obtain vibration signals and acoustic emission signals of the same frequency of the mine equipment;

[0087] A three-dimensional space frame is built based on spatial coordinates, and the vibration signals and acoustic emission signals of the same frequency and the thermal imaging data are arranged in the three-dimensional space frame according to the spatial position and time sequence to obtain a three-dimensional working condition atlas of the mine equipment.

[0088] Specifically, clock synchronization technology is used to accurately synchronize the vibration acceleration sensor and the acoustic emission sensor.

[0089] Further, during the operation of the mine equipment, the two sensors simultaneously start collecting information of the equipment.

[0090] Further, once it is detected that the signal segment of information collection delays more than the pre-set threshold value, the sliding window interpolation method is started.

[0091] Further, the specific operation is to demarcate a fixed-size window centered on the delay signal segment, and take the adjacent normal signals collected in the window as reference data.

[0092] Further, by calculating the change trend and numerical relationship of the reference data, reasonable intermediate values are inserted in the delay signal segment to compensate for the timing deviation, thereby obtaining accurate mine equipment vibration signals and acoustic emission signals.

[0093] Further, the timestamp technology is used to process the vibration signals and acoustic emission signals after timing deviation compensation.

[0094] Further, during the signal collection process, the system automatically adds corresponding timestamps to each set of vibration signals and acoustic emission signals, which accurately record the specific time of signal collection.

[0095] Further, based on the time information provided by the timestamp, the two sets of signals are compared and adjusted one by one, and the vibration signals and acoustic emission signals at the same time point are matched and aligned to ensure that the vibration signals and acoustic emission signals are completely synchronized in the time dimension, thereby obtaining the same frequency vibration signals and acoustic emission signals of the mine equipment.

[0096] Further, based on the actual spatial coordinates of each component of the mine equipment, a professional three-dimensional modeling software is used to build a three-dimensional space framework.

[0097] Further, in this framework, each spatial coordinate corresponds to the actual position of the equipment. The same frequency vibration signals, acoustic emission signals obtained in the previous steps, and the thermal imaging data collected by the infrared thermal imager are accurately mapped to the corresponding positions in the three-dimensional space framework according to their corresponding equipment space positions.

[0098] Further, according to the time sequence of each data collection, these data are arranged in the three-dimensional space framework in turn, and finally a three-dimensional working condition map that can comprehensively and intuitively reflect the running state of the mine equipment is constructed.

[0099] In summary, through the clock-synchronized sensor collection data, combined with the sliding window interpolation to compensate for the timing deviation, the vibration signals and acoustic emission signals are strictly aligned in the time dimension, solving the problem of timing misalignment caused by the difference in sampling frequency and transmission delay of traditional multi-source data.

[0100] In summary, the same frequency aligned signals can accurately map the mechanical response and energy release characteristics of the equipment at the same time, avoiding misjudgment of fault characteristics caused by time mismatch, and providing a reliable time reference for subsequent feature decoupling and fault location.

[0101] Overall, a three-dimensional framework is built based on spatial coordinates, and the aligned vibration / acoustic emission signals and thermal imaging data are structured and deployed according to spatial positions and time sequence, forming a three-dimensional working condition atlas containing 'time-space-physical characteristics'.

[0102] Overall, the atlas can intuitively present the state evolution process of each component of the device in three-dimensional space, and realize the stereoscopic mapping of fault from 'time domain-frequency domain' analysis to'spatial-temporal domain-energy domain' combined with the temperature field distribution of thermal imaging data, significantly improving the accuracy and visualization degree of early fault positioning, and providing multi-dimensional input basis for subsequent feature decoupling based on the dynamics model and multi-layer decision network diagnosis.

[0103] S3. Decoupling features of the running state information under physical constraints based on a device dynamics model to obtain a fault sensitive feature set of the mining equipment;

[0104] In the embodiment of the present application, the decoupling features of the running state information under physical constraints based on a device dynamics model to obtain a fault sensitive feature set of the mining equipment comprises:

[0105] According to the correlation between the structural characteristics, operating principle and physical parameters of the mining equipment, a device dynamics equation of the mining equipment is established;

[0106] Based on the device dynamics equation, the vibration signal is decomposed into a time-domain impact component related to gear meshing torque and a centrifugal force modulation component related to bearing wear;

[0107] The amplitude spectrum of the acoustic emission signal is extracted by Morlet wavelet transform to identify the gear tooth impact frequency and its harmonic components, and is associated with the device load rate change curve to obtain the harmonic component of the mining equipment;

[0108] According to the temperature gradient distribution collected by the infrared thermal imager, the unit area heat flow density distribution data of the mining equipment is generated;

[0109] The time-domain impact component, centrifugal force modulation component, harmonic component and unit area heat flow density distribution data are collected as the fault sensitive feature set of the mining equipment.

[0110] The unit area heat flow density distribution data of the mining equipment generated according to the temperature gradient distribution collected by the infrared thermal imager comprises:

[0111] The temperature distribution data of the friction contact area of the mining equipment is collected by the infrared thermal imager to generate a real-time temperature gradient distribution map of the friction contact area;

[0112] The rotational speed, load pressure and lubricant flow parameters of the rotor in the mine equipment are taken as boundary conditions for temperature field modeling, and the temperature gradient distribution is mapped to the friction contact area according to the Fourier heat conduction law to obtain a three-dimensional transient temperature field model of the friction contact area;

[0113] The friction force and relative displacement in the three-dimensional transient temperature field model are multiplied to obtain the friction heat generation power of the friction contact area;

[0114] The friction heat generation power is distributed to the friction contact area, and the grid division of the three-dimensional transient temperature field model is combined to generate the unit area heat flux density distribution data of the mine equipment.

[0115] Specifically, the components of the mine equipment are detailedly mapped, the shapes and sizes thereof are recorded, and the connection modes between the components, such as bolt connection and welding, are observed.

[0116] Further, the motion trajectories of the components during equipment operation, such as rotation of gears and rotation of bearings, are understood in depth, the force transmission and interaction relationship between the components are clarified, the influence of mass on motion inertia is analyzed, the degree of deformation of the components is determined by stiffness, and the motion energy is consumed by damping.

[0117] Further, the equipment is regarded as being composed of multiple interrelated subsystems by using the principle of system dynamics, the mechanical analysis of each subsystem is performed, the factors of force, motion and mass are considered, and the equipment dynamics equation capable of completely describing the running state of the mine equipment is constructed.

[0118] Further, the vibration signal is input to the signal analysis software, and the signal decomposition algorithm is adopted.

[0119] Further, the vibration signal is divided into multiple small segments in time sequence, for each small signal segment, the waveform change in the signal is analyzed according to the gear meshing and bearing motion law reflected by the equipment dynamics equation.

[0120] Further, when it is found that there is a sudden and severe waveform fluctuation in the signal, similar to the form of impact, and the occurrence law thereof is consistent with the gear meshing period, the signal is extracted to determine a time-domain impact component related to the gear meshing torque; if there is a periodic waveform modulation phenomenon in the signal, and the frequency thereof is related to the bearing rotation frequency, the signal is separated to be a centrifugal force modulation component related to the bearing wear.

[0121] Further, the acoustic emission signal is introduced into a signal processing system supporting Morlet wavelet transform. The Morlet wavelet transform algorithm is to select a suitable Morlet wavelet function first, and the function is a waveform with specific frequency and time characteristics.

[0122] Further, the Morlet wavelet function is slid at a certain time interval from the starting position on the acoustic emission signal, and at each position, a value is obtained by calculating the similarity between the Morlet wavelet function and the acoustic emission signal at the corresponding position.

[0123] Further, after the Morlet wavelet function slides through the entire acoustic emission signal, these values constitute the amplitude spectrum of the acoustic emission signal.

[0124] Further, by analyzing the amplitude spectrum, the part with higher peak value and specific frequency rule is determined as the gear tooth impact frequency and its harmonic components, and then these frequency components are compared with the load rate change curve recorded during equipment operation to find the correlation between them, thereby obtaining the harmonic components of the mine equipment.

[0125] Further, the temperature gradient distribution image collected by the infrared thermal imager is opened using image processing software.

[0126] Further, the surface of the equipment is first divided into regular unit area regions in the image, like dividing the grid on the map.

[0127] Further, according to the physical relationship between temperature and heat flux density, i.e. temperature change will cause heat flow, and the greater the temperature difference, the greater the heat flux density.

[0128] Further, the average temperature of each unit area region is calculated, and then compared with the temperature of the adjacent region, and according to the temperature difference and the area of the region, the heat flux density of each unit area region is calculated, and finally the unit area heat flux density distribution data of the mine equipment is generated.

[0129] Further, through the data transmission line, the time domain impact component and the centrifugal force modulation component decomposed from the vibration signal, the harmonic component extracted from the acoustic emission signal, and the unit area heat flux density distribution data generated according to the infrared thermal imager data are uniformly transmitted to the data processing center.

[0130] Further, in the data processing center, using data integration software, these different types of data are arranged and converted according to the unified format requirements, the repeated and invalid data are removed, and the effective data are combined together to form a fault sensitive feature set that can accurately reflect the potential faults of the mine equipment, which is used for subsequent equipment fault diagnosis and analysis.

[0131] Specifically, an infrared thermal imager is used to monitor the friction contact area of ​​mining equipment in real time. The infrared detector receives the infrared radiation emitted by the objects in the area. Through signal processing and image processing algorithms, infrared radiation of different intensities is converted into corresponding temperature values. A real-time temperature gradient distribution map of the friction contact area is generated according to the spatial distribution, which intuitively presents the temperature level and trend of the area.

[0132] Furthermore, the rotational speed of the rotor in the mining equipment is determined by measuring and recording the number of rotations per minute of the rotor in real time using a speed sensor; the load pressure is obtained by measuring and recording the pressure applied to the equipment using a pressure sensor installed at the load application point; and the lubricant flow parameters are statistically analyzed by calculating the volume of lubricant flowing through the pipeline per unit time using a flow metering device installed in the lubricant delivery pipeline.

[0133] Furthermore, these three parameters are used as boundary conditions for temperature field modeling. Based on Fourier's law of heat conduction, which states that the heat conduction rate is proportional to the temperature gradient and in the opposite direction to the temperature gradient, the temperature data in the real-time temperature gradient distribution map is mapped point by point to the three-dimensional spatial location of the friction contact area according to the direction and rate of heat conduction, thus constructing a three-dimensional transient temperature field model of the friction contact area. This model reflects the temperature distribution at different times and locations.

[0134] Furthermore, frictional force data and relative displacement data of the frictional contact area are extracted from the three-dimensional transient temperature field model.

[0135] Furthermore, frictional force data is obtained by force sensors installed near the friction pairs, reflecting the force between the friction surfaces that opposes relative motion; relative displacement data is obtained by displacement sensors measuring the relative movement distance between the friction pairs.

[0136] Furthermore, by multiplying these two data points—that is, multiplying the value of the frictional force by the value of the relative displacement—the result is the frictional heat generation power of the frictional contact area, which represents the amount of heat generated per unit time during the friction process.

[0137] Furthermore, the calculated frictional heat generation power is distributed to different locations within the frictional contact area based on the actual contact conditions and heat conduction characteristics of each part of the frictional contact area.

[0138] Furthermore, the three-dimensional transient temperature field model is first meshed, dividing the friction contact area into multiple small mesh units, each representing a tiny unit area.

[0139] Further, according to the friction force, relative displacement and heat conduction capacity of the position where each grid unit is located, the heat power that should be distributed to the grid unit is calculated, and then the heat power is divided by the area of the grid unit to obtain the heat flow density per unit area, and finally the heat flow density distribution data per unit area of the mine equipment is generated, and the heat flow density size and distribution law of each unit area in the friction contact area are clearly displayed.

[0140] In general, based on the structural characteristics and operating principle of the mine equipment, the vibration signal is decomposed into a time-domain impact component and a centrifugal force modulation component, the physical features directly related to a specific fault mode are separated from the mixed signal, and the feature ambiguity problem caused by the mixing of multiple source signals in the traditional frequency domain analysis is avoided.

[0141] In general, the harmonic component of the acoustic emission signal is extracted by Morlet wavelet transform, and the unit area heat flow density distribution generated by the thermal imaging data is combined to form a multi-physical quantity fault sensitive feature set covering mechanics, acoustics and thermodynamics, which significantly reduces the data dimension while retaining the core fault information, and improves the input quality of the subsequent diagnosis model.

[0142] In general, the feature decoupling process is closely related to the principle of equipment dynamics, so that each feature component corresponds to a clear physical meaning.

[0143] In general, this feature engineering method based on prior knowledge not only improves the causal relationship between features and fault modes, but also provides a traceable physical mechanism explanation for the diagnosis result, avoids the "black box" problem of deep learning models, and facilitates engineers to quickly locate the fault source and develop targeted maintenance strategies.

[0144] S4. Input the fault sensitive feature set into a multi-layer decision network to output a fault diagnosis result and a confidence score;

[0145] In the embodiment of the present application, the input of the fault sensitive feature set into the multi-layer decision network to output the fault diagnosis result and the confidence score includes:

[0146] The gating recurrent layer is used for time sequence dependent analysis of the fault sensitive feature set, and the hidden Markov chain state transition probability of the mine equipment operation state is output;

[0147] The convolutional neural network is used to capture the energy distribution characteristics of the vibration signal in the gear meshing line direction, and a two-dimensional spatial correlation graph of the mine equipment is generated;

[0148] The hidden Markov chain state transition probability and the two-dimensional spatial correlation graph are aggregated to output the fault diagnosis result and the confidence score of the mine equipment.

[0149] Specifically, the previously obtained mine equipment failure sensitive feature set is sequentially input into the gated recurrent layer model in chronological order.

[0150] Further, the gated recurrent layer internally includes two control units, an update gate and a reset gate. The update gate determines how much state information from the previous time is retained to the current time, and the reset gate determines the degree of forgetting of the state information from the previous time.

[0151] Further, when processing the failure sensitive feature set data at each time, the gated recurrent layer calculates the hidden state at the current time based on the calculation results of the update gate and the reset gate, combined with the current input data and the state at the previous time.

[0152] Further, after all the time data processing is completed, the gated recurrent layer calculates the hidden Markov chain state transition probability of the mine equipment operating state based on the final obtained hidden state sequence, i.e. the possibility of transitioning from one operating state to another.

[0153] Further, the processed vibration signal is input into the convolutional neural network.

[0154] Further, the convolutional neural network is composed of multiple convolutional layers, pooling layers and fully connected layers. In the convolutional layer, multiple convolutional kernels of different sizes and weights are designed to slide on the vibration signal data according to the set step size, and the data of the vibration signal in the gear meshing line direction is extracted for feature extraction.

[0155] Further, each convolutional kernel performs convolution operation with the data of the corresponding region, i.e. multiplication of corresponding elements and summation, to obtain the feature value of the region.

[0156] Further, after processing by multiple convolutional layers, the energy distribution features of the vibration signal in the gear meshing line direction at different levels are extracted.

[0157] Further, the features are processed by the pooling layer to reduce the dimension, retain the main features and reduce the data volume. Finally, the extracted features are integrated by the fully connected layer to generate a mine equipment two-dimensional spatial correlation graph that can reflect the energy distribution features of the vibration signal in the gear meshing line direction.

[0158] Further, the hidden Markov chain state transition probability calculated by the gated recurrent layer and the two-dimensional spatial correlation graph generated by the convolutional neural network are input into the data aggregation module. In the data aggregation module, the data format of the hidden Markov chain state transition probability and the two-dimensional spatial correlation graph is uniformly converted to make them compatible.

[0159] Further, a weighted fusion method is used to combine and calculate the data of the two according to the pre-set weight.

[0160] Further, the merged data is processed by a classifier, which judges the current fault state of the mine equipment according to a trained classification model and rules, and outputs a fault diagnosis result.

[0161] Further, according to the decision information and data features in the classification process, a confidence score of the fault diagnosis result is calculated to represent the reliability of the diagnosis result.

[0162] In summary, the time sequence dependent analysis of the fault sensitive feature set is performed by using the gated recurrent layer, the hidden Markov chain state transition probability is generated to capture the dynamic evolution law of the equipment operation state over time, and meanwhile, the energy distribution features of the vibration signal in the gear meshing line direction are extracted by the convolutional neural network to generate a two-dimensional spatial correlation graph to depict the fault propagation mode of the mechanical parts in the spatial position.

[0163] In summary, the two network structures process the time sequence features and the spatial features in parallel to realize cross-dimension feature aggregation of the 'time sequence-spatial distribution', avoid the limitation of a single network on the representation of complex working conditions, significantly improve the generalization diagnosis capability of the model on multiple types of faults such as non-stationary vibration and sudden impact, and shorten the time link from feature analysis to diagnosis result output.

[0164] In summary, the multi-layer decision network outputs not only the fault type but also the confidence score, which provides a quantitative reliability index for the diagnosis result.

[0165] For example, when the time domain impact component of the vibration signal and the acoustic emission harmonic component trigger the gear tooth breakage warning at the same time, the model can judge the reliability of the diagnosis result through the confidence score to effectively filter the false positives caused by interference factors such as sensor noise and working condition fluctuations. This diagnosis mechanism with probability output not only meets the real-time requirement of intelligent manufacturing, but also improves the reliability of the diagnosis result through data-driven statistical learning, which provides more reliable decision basis for subsequent fault cause tracing and process correction.

[0166] S5. Constructing a fault cause and effect diagram of the mine equipment based on the pre-acquired design parameters and manufacturing process parameters, designing a correction scheme containing process constraint conditions for abnormal process parameters of the mine equipment in the fault cause and effect diagram whose traceability confidence score is lower than a set threshold value;

[0167] In the embodiment of the present application, the construction of the fault cause and effect diagram of the mine equipment based on the pre-acquired design parameters and manufacturing process parameters comprises:

[0168] Bayesian estimation is performed on the cutting depth, tool feed rate and spindle torque in the pre-acquired design parameters and manufacturing process parameters to obtain the conditional probability distribution of the mining equipment in different states, the conditional probability distribution is filled into the network parameters, and a Bayesian network model is established;

[0169] A hierarchical aggregation strategy is performed on the process parameters and failure labels in the pre-acquired design parameters and manufacturing process parameters to obtain a process-failure matrix of the mining equipment.

[0170] Markov chain Monte Carlo sampling is performed on the Bayesian network model set as a normal distribution and different types of failures in the process-failure matrix to obtain a fault causal diagram of the mining equipment.

[0171] Based on the abnormal process parameters of the mining equipment with a traceability confidence score lower than a set threshold in the fault causal diagram, a correction scheme containing process constraint conditions is designed for the abnormal process parameters, including:

[0172] If the fault causal diagram shows that the cutting depth is strongly related to vibration exceeding the standard, the tool feed rate is adjusted and a lubrication compensation condition is added;

[0173] If there is a causal feedback between the centrifugal force modal parameter and the bearing temperature, an assembly tolerance checking instruction is triggered;

[0174] When the thermal loss rate is abnormal, the matching degree of the friction pair geometry is verified by finite element simulation.

[0175] The correction scheme containing process constraint conditions is designed for the abnormal process parameters, including:

[0176] The path parameters of the tool are limited based on the maximum allowable cutting force of the mining equipment;

[0177] The assembly gap tolerance threshold of the mining equipment is set according to the accuracy level requirement of the gear box of the mining equipment;

[0178] The bearing temperature rise critical value of the mining equipment is limited in combination with the material heat treatment characteristics of the mining equipment.

[0179] Specifically, the pre-acquired design parameters and manufacturing process parameters of the mining equipment are first collected, and the manufacturing process parameters are separated therefrom.

[0180] Further, these process parameters and corresponding failure labels are divided according to a certain hierarchical relationship, such as first-class classification according to the major categories of processes, such as casting process, machining process, assembly process, etc., and second-class classification according to specific process steps or process indicators under each major category, and so on.

[0181] Further, the process parameters and fault labels in each level are aggregated to combine the process parameters and fault labels with the associated relationship, and through the hierarchical aggregation strategy, a mine equipment process-fault matrix is finally constructed to clearly show the corresponding relationship between the process parameters and faults.

[0182] Further, for the Bayesian network model set as a normal distribution and different types of faults in the process-fault matrix, a Markov chain Monte Carlo sampling method is used.

[0183] Further, the initial state of the sampling is determined, which can be based on certain prior knowledge or randomly set.

[0184] Further, starting from the initial state, the next state is generated according to the Markov chain rule based on the current state, and in the generation process, each state is evaluated and screened according to the probability distribution set by the Bayesian network model and the relationship between the faults and the process parameters in the process-fault matrix.

[0185] Further, the probability of being accepted is calculated for each new state generated, and if the acceptance probability meets certain conditions, the state is retained as part of the sampling results, otherwise the state is discarded. By repeatedly generating a large number of sampling states, the causal relationship between different process parameters and faults is analyzed and sorted out, and finally a fault causal diagram is drawn to visually show the causal relationship of mine equipment faults.

[0186] Specifically, when the fault causal diagram shows that there is a strong correlation between the cutting depth and the vibration exceeding the standard, immediate adjustment measures are taken.

[0187] Further, the feed rate of the current tool is first determined, and based on experience and actual processing conditions, the tool feed rate is gradually reduced to reduce the amount of tool cutting into the workpiece each time.

[0188] Further, a lubrication compensation condition is added, and in the cutting contact area between the tool and the workpiece, a special lubricating device is used to increase the amount and frequency of lubricant supply, improve the lubrication condition in the cutting process, reduce the cutting resistance and friction, and thus reduce the vibration exceeding the standard problem caused by the cutting depth.

[0189] Further, if a causal feedback is found between the centrifugal modal parameter and the bearing temperature, an assembly tolerance verification instruction is triggered immediately. Professional assembly personnel are arranged to use high-precision measuring tools such as micrometers and dial gauges to measure the size of the parts related to the bearing assembly in the mine equipment.

[0190] Further, the actual size measured is compared one by one with the standard size required by the design requirement to check whether the size deviation of each component is within the specified tolerance range.

[0191] Further, for components that exceed the tolerance range, reprocessing or replacement is performed to ensure that the assembly tolerance of all components meets the requirements and eliminates the causal feedback of abnormal centrifugal modal parameters and bearing temperature rise caused by assembly problems.

[0192] Further, once the abnormal heat loss rate is detected, finite element simulation technology is immediately used to verify the geometric matching degree of the friction pair.

[0193] Further, first use three-dimensional modeling software to accurately construct a three-dimensional model of the friction pair according to the actual size and shape of the friction pair of the mining equipment.

[0194] Further, the constructed three-dimensional model is imported into the finite element analysis software, and the friction pair is meshed to divide the friction pair into multiple small grid units.

[0195] Further, according to the actual working conditions, set the boundary conditions of the simulation, including the movement mode and stress condition of the friction pair.

[0196] Further, run the finite element simulation program in the software to simulate the stress deformation and heat transfer of the friction pair during the working process, and analyze the simulation results to determine whether the geometric shape of the friction pair is well matched and find out the geometric shape problem causing the abnormal heat loss rate.

[0197] Further, based on the maximum allowable cutting force limit of the mining equipment, the path parameters of the tool are corrected.

[0198] Further, the maximum allowable cutting force value specified by the mining equipment is obtained, and a special tool path planning software is used to import the current tool path data.

[0199] Further, the software analyzes the stress of each cutting point on the tool path and calculates the cutting force size of each point under the current path.

[0200] Further, for points where the calculated cutting force exceeds the maximum allowable cutting force, the software automatically adjusts the tool path of the point, such as changing the cutting angle of the tool, increasing the number of cutting times, etc., to re-plan the motion trajectory of the tool, ensuring that the cutting force of the tool during the entire cutting process does not exceed the maximum allowable cutting force limit.

[0201] Further, according to the precision grade requirement of the gear box in the mining equipment, set the mining equipment assembly gap tolerance threshold.

[0202] Further, refer to the technical documents of the mining equipment to determine the required precision grade standard of the gear box. According to the provisions of the assembly gap in the precision grade standard, determine the appropriate assembly gap tolerance range.

[0203] Further, in the actual assembly process, use gap measuring tools such as feeler gauges to measure the assembly gap between the components of the mining equipment.

[0204] Further, when the measured value exceeds the pre-set assembly gap tolerance threshold, adjust the assembly position of the components or replace the related components to ensure that the assembly gap meets the precision grade requirements of the gear box.

[0205] Further, in combination with the material heat treatment characteristics of the mining equipment, limit the bearing temperature rise critical value of the mining equipment.

[0206] Further, collect detailed heat treatment information of the materials used in the bearings of the mining equipment, and understand the changes in physical properties such as hardness and strength of the materials at different temperatures. Through experiments or by referring to the data provided by the material suppliers, determine the maximum temperature that the material can withstand under normal working conditions.

[0207] Further, set this temperature value as the bearing temperature rise critical value. During the operation of the mining equipment, use temperature sensors to monitor the temperature of the bearings in real time. Once the bearing temperature approaches or reaches the critical value, take measures to reduce the temperature immediately, such as increasing the power of the cooling device, to ensure that the temperature of the bearings does not exceed the limited temperature rise critical value.

[0208] In summary, based on the design parameters and manufacturing process parameters, a Bayesian network model and a process-failure matrix are constructed, a fault causal diagram is generated through Markov chain Monte Carlo sampling, and the association between process parameters and failure modes is probabilistically modeled.

[0209] In summary, this data-driven and physical model combined traceability method breaks through the limitations of traditional diagnosis, which only focuses on operation data, and can trace the root cause of failure from the design, processing, and assembly process dimensions, achieving "equipment failure-process parameter anomaly" reverse tracking, solving the problem of disconnection between fault diagnosis and manufacturing process in the prior art, and improving the depth and systematicness of the diagnosis results.

[0210] In summary, for abnormal process parameters with low traceability confidence in the fault causal diagram, a correction scheme is designed in combination with process constraints:

[0211] In summary, based on the maximum allowable cutting force limit of the equipment, the tool path parameters are corrected to avoid mechanical overload caused by blind adjustment.

[0212] In summary, set the assembly gap tolerance threshold according to the precision grade of the gear box to ensure that the corrected assembly process meets the design standards;

[0213] In general, the bearing temperature rise critical value is defined in combination with the material heat treatment characteristics to prevent thermal deformation from causing secondary failures.

[0214] In general, the mechanism directly maps the diagnosis results to process parameter optimization, realizes the upgrade from failure handling to preventive optimization of the manufacturing process through the closed loop of "probability modeling-threshold triggering-constraint correction", not only eliminates the current failure hidden danger, but also reduces the recurrence probability of similar failures through dynamic optimization of process parameters, forms the synergistic effect of failure diagnosis and process improvement in intelligent manufacturing, and significantly improves the reliability and production efficiency of the equipment throughout its life cycle.

[0215] S6. Fuse the correction scheme, the failure diagnosis result and the three-dimensional working condition map to construct a visual diagnosis report.

[0216] In the embodiment of the present application, the fusion of the correction scheme, the failure diagnosis result and the three-dimensional working condition map to construct a visual diagnosis report comprises:

[0217] After mapping the axial vibration amplitude of the three-dimensional working condition map to the color spectrum gradient, the frequency spectrum feature lines in the failure diagnosis result are superimposed to the corresponding area of the thermal imaging map, and the influence path of the abnormal process parameters on the equipment performance is marked with dynamic vector arrows to obtain the visual diagnosis report of the mine equipment.

[0218] Specifically, the three-dimensional working condition map of the mine equipment is processed to extract the axial vibration amplitude data therein.

[0219] Further, a fixed mapping rule is established to correspond different sizes of axial vibration amplitudes to different colors of color spectrum gradients.

[0220] For example, a lower axial vibration amplitude is set to correspond to a cool color, and a higher axial vibration amplitude is set to correspond to a warm color. According to this rule, the axial vibration amplitude of each position in the three-dimensional working condition map is converted to the corresponding color to form a visual effect with a color spectrum gradient.

[0221] Further, frequency spectrum feature line data is obtained from the failure diagnosis result. The previously generated thermal imaging map is found, and according to the equipment position information corresponding to the frequency spectrum feature lines, these frequency spectrum feature lines are accurately superimposed on the corresponding area of the thermal imaging map.

[0222] Further, during the superimposition process, the position, shape and direction of the frequency spectrum feature lines are ensured to match the actual structure and position of the equipment in the thermal imaging map, so that the two can be clearly fused together for observation and analysis.

[0223] Further, for the abnormal process parameters that have been determined, the specific path of their influence on the performance of the mine equipment is analyzed.

[0224] Further, the starting point and the ending point of the influence of each abnormal process parameter on the equipment performance are determined, and the specific route of the influence transmission in the equipment structure is determined.

[0225] Further, dynamic vector arrows are drawn on the visualized graph using drawing software, the starting point of the arrow is the position where the abnormal process parameter occurs, the direction of the arrow is along the route of the image transmission, and the ending point of the arrow is the part of the equipment performance that is affected.

[0226] Further, in this way, the influence path of the abnormal process parameter on the equipment performance is intuitively marked.

[0227] Further, the thermal imaging graph with the color spectrum gradient, the superimposed spectrum feature line, and the marked dynamic vector arrow after the above processing are integrated, necessary legends, text instructions, and title elements are added, and a complete visualized diagnosis report of the mine equipment is formed, which can visually and intuitively show the running state of the mine equipment, the fault feature, and the influence of the abnormal process parameter.

[0228] In general, the axial vibration amplitude of the three-dimensional working condition graph is mapped to the color spectrum gradient, the spectrum feature in the fault diagnosis result is superimposed on the corresponding area of the thermal imaging graph, and the spatio-temporal correlation display of the vibration energy distribution-temperature field abnormality-fault frequency feature is realized.

[0229] For example, the high-temperature area in the thermal imaging graph is superimposed with the high-frequency vibration spectrum line at the same position, the overheat and vibration abnormal coupling fault of the friction pair caused by insufficient lubrication can be quickly located, the information fragmentation problem in the traditional single chart analysis is avoided, and the time cost of the engineering personnel from data interpretation to fault positioning is greatly shortened.

[0230] In general, the influence path of the abnormal process parameter on the equipment performance is marked by the dynamic vector arrow, and the abstract fault cause-and-effect relationship is converted into an intuitive physical link display.

[0231] In general, in combination with the process constraint condition in the correction scheme, the visualized diagnosis report can synchronously present the complete logical chain of the fault phenomenon-process trace-correction strategy.

[0232] For example, the downward trend of the vibration amplitude after the assembly gap correction is dynamically demonstrated by an arrow in the three-dimensional working condition graph, an integrated decision-making interface that combines the diagnosis result, the trace basis, and the improvement scheme is provided for the operator, the transparency and the collaborative efficiency of the fault handling in the intelligent manufacturing process are improved, and traceable historical data reference is provided for subsequent process optimization.

[0233] In the several embodiments provided in the present application, it should be understood that the disclosed method can be implemented in other ways.

[0234] It is apparent for a person skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, but that the present application can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application.

[0235] Embodiments of the present application can acquire and process related data based on artificial intelligence technology. Among them, artificial intelligence is to use digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0236] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A method for intelligent fault diagnosis in the intelligent manufacturing process of mining equipment, characterized in that, The method includes: S1. Collect vibration signals, acoustic emission signals and thermal imaging data of mining equipment in real time through vibration acceleration sensors, acoustic emission sensors and infrared thermal imagers as the operating status information of the mining equipment; S2. After aligning the vibration signal and the acoustic emission signal using timestamps, spatial coordinate registration is performed to obtain a three-dimensional working condition map of the mining equipment; S3. Based on the equipment dynamics model, the operating state information is decoupled under physical constraints to obtain the fault-sensitive feature set of the mining equipment, including: Based on the structural characteristics, operating principles, and correlations between physical parameters of the mining equipment, the equipment dynamics equations of the mining equipment are established. Based on the equipment dynamics equations, the vibration signal is decomposed into a time-domain impact component related to gear meshing torque and a centrifugal force modulation component related to bearing wear. The amplitude spectrum of the acoustic emission signal is extracted by Morlet wavelet transform, the impact frequency of gear tooth breakage and its harmonic components are identified, and correlated with the equipment load rate change curve to obtain the harmonic components of the mining equipment. The heat flux density distribution data per unit area of ​​the mining equipment is generated based on the temperature gradient distribution collected by the infrared thermal imager, including: The infrared thermal imager is used to collect temperature distribution data of the friction contact area of ​​the mining equipment, and a real-time temperature gradient distribution map of the friction contact area is generated. The rotor speed, load pressure, and lubricant flow rate parameters in the mining equipment are used as boundary conditions for temperature field modeling. At the same time, the temperature gradient distribution map is mapped to the friction contact area according to Fourier's law of heat conduction to obtain a three-dimensional transient temperature field model of the friction contact area. The frictional heat generation power of the frictional contact region is obtained by multiplying the frictional force and relative displacement in the three-dimensional transient temperature field model. The frictional heat generation power is distributed to the frictional contact area, and combined with the mesh division of the three-dimensional transient temperature field model, the heat flux density distribution data per unit area of ​​the mining equipment is generated. The time-domain impact component, centrifugal force modulation component, harmonic component, and heat flux density distribution per unit area are collected to form the fault-sensitive feature set of the mining equipment. S4. Input the fault-sensitive feature set into a multi-layer decision network and output the fault diagnosis results and confidence scores; S5. Construct a fault cause-effect graph of the mining equipment based on the pre-acquired design parameters and manufacturing process parameters. Based on the abnormal process parameters of the mining equipment whose traceability confidence scores are lower than a set threshold in the fault cause-effect graph, design a correction scheme including process constraints for the abnormal process parameters. S6. Integrate the aforementioned correction scheme, the fault diagnosis results, and the three-dimensional operating condition map to construct a visual diagnostic report.

2. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 1, characterized in that, After aligning the vibration signal and the acoustic emission signal using timestamps, spatial coordinate registration is performed to obtain a three-dimensional operating condition map of the mining equipment, including: The vibration acceleration sensor and the acoustic emission sensor, after being synchronized with the clock, are used to collect information from the mining equipment. When the delay of the information collection exceeds the signal segment of a set threshold, sliding window interpolation is used to compensate for the timing deviation of the collected information to obtain the vibration signal and acoustic emission signal of the mining equipment. By aligning the vibration signal and the acoustic emission signal using timestamps, the same frequency vibration signal and acoustic emission signal of the mining equipment are obtained; A three-dimensional spatial framework is constructed based on spatial coordinates. The vibration signal and acoustic emission signal of the same frequency, as well as the thermal imaging data, are arranged in the three-dimensional spatial framework according to spatial location and time sequence to obtain a three-dimensional working condition map of the mining equipment.

3. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 1, characterized in that, The step of inputting the fault-sensitive feature set into a multi-layer decision network and outputting fault diagnosis results and confidence scores includes: A gated loop layer is used to perform time-series dependency analysis on the fault-sensitive feature set, and the hidden Markov chain state transition probabilities of the operating state of the mining equipment are output. A two-dimensional spatial correlation diagram of the mining equipment is generated by capturing the energy distribution characteristics of the vibration signal in the direction of the gear meshing line using a convolutional neural network. By aggregating the hidden Markov chain state transition probabilities and the two-dimensional spatial correlation graph, the fault diagnosis results and confidence scores of the mining equipment are output.

4. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 1, characterized in that, The construction of the fault cause-effect graph of the mining equipment based on pre-acquired design parameters and manufacturing process parameters includes: Bayesian estimation is performed on the cutting depth, tool feed rate and spindle torque in the pre-acquired design parameters and manufacturing process parameters to obtain the conditional probability distribution of the mining equipment under different states. The conditional probability distribution is then filled into the network parameters to establish a Bayesian network model. A hierarchical aggregation strategy is applied to the process parameters and fault labels in the pre-acquired design parameters and manufacturing process parameters to obtain the process-fault matrix of the mining equipment; Markov chain Monte Carlo sampling is performed on the Bayesian network model set to a normal distribution and the different types of faults in the process-fault matrix to obtain the fault cause-effect graph of the mining equipment.

5. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 1, characterized in that, The abnormal process parameters of mining equipment whose source tracing confidence scores are lower than a set threshold based on the fault cause-effect graph are addressed by designing a correction scheme that includes process constraints, including: If the cause-and-effect diagram shows a strong correlation between the cutting depth and excessive vibration, adjust the tool feed rate and add lubrication compensation conditions. If there is a causal feedback between the centrifugal force modal parameters and the bearing temperature, an assembly tolerance verification command will be triggered. When the heat loss rate is abnormal, the geometric matching degree of the friction pair is verified by finite element simulation.

6. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 5, characterized in that, The proposed correction scheme for the abnormal process parameters, which includes process constraints, includes: The tool path parameters are modified based on the maximum permissible cutting force limit of the mining equipment; The assembly clearance tolerance threshold of the mining equipment is set according to the gearbox precision level requirements in the mining equipment; The critical temperature rise value of the bearings of the mining equipment is determined based on the material heat treatment characteristics of the mining equipment.

7. The intelligent fault diagnosis method in the intelligent manufacturing process of mining equipment as described in claim 1, characterized in that, The process of integrating the correction scheme, the fault diagnosis results, and the three-dimensional operating condition map to construct a visual diagnostic report includes: The axial vibration amplitude of the three-dimensional working condition spectrum is mapped to the chromatographic gradient and then superimposed with the spectral feature line in the fault diagnosis result to the corresponding area of ​​the thermal imaging image. The influence path of abnormal process parameters on equipment performance is marked with dynamic vector arrows to obtain a visual diagnostic report of the mining equipment.

Images