Search

The fault diagnosis method of planetary gearbox based on machine learning relies on the artificial selection of the eigenvectors. The quality of eigenvectors selection largely determines the accuracy of the diagnosis method. The convolutional neural network (CNN) can extract characteristics automatically. But it is difficult to accurately diagnose the fault from a single vibration signal when it is used for the planetary gearbox fault diagnosis. To solve the above problems, a fault diagnosis method of planetary gearbox based on multi-information fusion and CNN is proposed. The method performs data layer fusion on three-dimensional (horizontal radial direction, vertical radial direction and axial direction) vibration signals and sound signals of the planetary gearbox. The one-dimensional vibration signals and sound signals are integrated into two-dimensional signals in a parallel connection mode. The two-dimensional signals are used as the input of CNN. The multiple convolutional layers and maximum pooling layers are used for depth characteristic extraction and information filtering. Finally, the Softmax classifier is used to achieve fault classification. The fault diagnosis experiment platform of the planetary gearbox is built. The vibration signals and sound signals of normal and fault states of the planetary gearbox under different speed and load conditions are collected and input into CNN for training and verification. Four single-source information of horizontal radial vibration signal, vertical radial vibration signal, axial vibration signal and sound signal are selected under the same conditions and combined with CNN respectively for comparison. The experiment is used to verify the superiority of the fault diagnosis method for planetary gearbox based on multi-information fusion and CNN. The experimental results show that the fault identification accuracy of the two methods of axial vibration signal+CNN and sound signal+CNN is 74.07% and 75.13% respectively. The fault identification accuracy of the two methods of horizontal radial vibration signal+CNN and vertical radial vibration signal+CNN is 89.70% and 87.09% respectively. The method based on multi-information fusion and CNN has the fastest convergence speed and the highest fault identification accuracy, which is 93.33%.

The coal mine massive data has problems such as 'data island', weak correlation, poor data quality due to lack of data management system. It is difficult to make full use of the data and provide analysis and decision-making support for coal mine intelligence. The data warehouse can meet the requirements of multi-source heterogeneous data integration in coal mine, and provide data basis for intelligent application in coal mine. By analyzing the coal mine data types, characteristics and intelligent application requirements of actual data, the intelligent coal mine data warehouse modeling method is studied. Firstly, the layered architecture of intelligent coal mine data warehouse is constructed, and the characteristics of data model of original data layer, detailed data layer, basic index layer, service data layer and public dimension layer are analyzed. Secondly, taking the data of fully mechanized working face as an example, the modeling process of data warehouse is expounded from the aspects of business data analysis, application demand analysis and layered architecture design. Thirdly, the construction method of data model in coal mine data warehouse is introduced. The original data is transformed into data warehouse dimensional model through dimension alignment, dimension association and dimensional index aggregation. The method solves the application problem of coal mine data association in different dimensions. Finally, in order to solve the problem of portability of coal mine data warehouse, the design idea of coal mine parametric data warehouse based on general data warehouse in coal mine industry + parametric ETL (extraction-transformation-load) method is proposed. The platform of coal mine data warehouse in the laboratory environment is set up to process the data of fully mechanized working face of Shanxi Tiandi Wangpo Coal Industry Co., Ltd. The auxiliary mechanism model analysis and visual management cockpit are realized based on the processing data, which verifies the practicability of intelligent coal mine data warehouse. The performance indexes of the original data model and the intelligent coal mine data warehouse are compared. The results show that the data organization, model reuse and iteration difficulty of the intelligent coal mine data warehouse are better than those of the original data model, and the data query response time is shortened by more than 50%.

In order to solve the problems of low precision, poor real-time performance and easy missing detection of small targets in the current deep learning-based coal and gangue target detection methods, the SSD model is improved by using lightweight network, self-attention mechanism and anchor frame optimization method to construct Ghost-SSD model, and then a lightweight coal and gangue target detection method is proposed. The Ghost-SSD model is based on the SSD model, and the GhostNet lightweight characteristic extraction network is used to replace the main network layer VGG16 so as to improve the detection speed of coal and gangue targets. In order to solve the problem that the shallow characteristic map contains more background noise and insufficient semantic information, the self-attention module is introduced to enhance the characteristics of the shallow characteristic map and increase the focus on the foreground region. Moreover, the dilated convolution is applied to increase the receptive field of the shallow characteristic maps and enrich the semantic information of the shallow characteristic maps. The K-means algorithm is used to cluster the anchor frames, optimize the size of the anchor frame, and further improve the precision of coal and gangue target detection. The experimental results show that when the Ghost-SSD model is applied in coal and gangue target detection, the mean average precision is 3.6% higher than that of the SSD model, the detection speed is increased by 75 frames/s, and the detection precision and speed are better than that of the Faster-RCNN and Yolov3 models. Moreover, the model has a good detection effect on small coal and gangue targets.

In order to solve the problem of slow detection speed of existing deep learning based belt conveyor foreign object detection methods, an improved YOLOv3 model is proposed and applied to coal mine belt conveyor foreign object detection. The model uses the lightweight network DarkNet22-DS as the backbone feature extraction network. DarkNet22-DS replaces the standard convolution with depthwise separable convolution, which reduces the network parameters significantly and improves the feature utilization efficiency by composite residual blocks. By introducing weighted bi-directional feature pyramid networks and dual-scale output, the model improves the feature fusion network and enhances the model's detection efficiency of large foreign objects. The complete intersection ratio loss function is used as the target box regression loss function, and the correlation between the target box information is fully utilized to improve the convergence speed and detection accuracy of the model. The improved YOLOv3 model is deployed on the embedded platform Jetson Xavier NX for coal mine belt conveyor foreign object detection experiments. The results show that compared with the YOLOv3 model, the weight file size of the improved YOLOv3 model is reduced by 91.4%, and the amount of model parameters is reduced significantly. The detection speed is increased by 16 times, reaching 30.7 frames/s. The performance meets the real-time detection requirements of foreign objects in coal mine belt conveyors.

In order to solve the problem that a large amount of shearer operating state data cannot be processed in time, the distributed real-time prediction model of shearer operating state data based on Storm is proposed. Combined with the actual operating state data of the shearer, the Hadoop distributed storage database is used to simulate the real-time data flow of the operating state of the shearer. The Storm distributed real-time big data processing framework is used to process a large number of time series data of the operating state of the shearer. And Gate Recurrent Unit (GRU) is adopted as the prediction model to achieve real-time prediction of shearer operating state data. Combined with the threshold setting of various data, the model can realize fault warning. Taking the data of MG400930-WD electric traction shearer in fully mechanized working face as an example, eight kinds of monitoring data are used as experimental data to train and test the prediction model. The data include cutting part motor current, cutting part motor temperature, traction part motor current, traction part motor speed, height adjustment pump working pressure, height adjustment pump working speed, cooling water pressure and inverter current. The results show that the prediction model converges quickly, and the goodness of fit is above 0.9. Except for the cooling water pressure, the early warning accuracy rate of the remaining data is above 95%. The processing speed is fast and the whole early warning process is about 10 s in total, which can meet the application requirements.

For problems of wired while drilling device (MWD) which used central cable drilling rig as signal transmission channel such as complex drilling rig structure, limited mechanical behavior, poor anti-interference performance and so on, a mine-used mud pulse wireless MWD device was developed. The device is composed of explosion-proof computer, explosion-proof keyboard, explosion-proof data memory, explosion-proof measuring tube, explosion-proof pressure transmitter, etc. It works in intermittent mode. When shutdown signal of mud pump is detected by measuring section of the explosion-proof measuring tube, static data including drilling trajectory parameter and state parameter of directional drilling tool is collected and coded. When opening signal of mud pump is detected, following works are done by driving section of the explosion-proof measuring tube including sending the static data, measuring and sending dynamic data, and controlling the pulse generator to adjust flow channel area of hydraulic conduit. The explosion-proof pressure transmitter collects pressure signal of mud pump and sends it to the explosion-proof computer. When data transmission is completed, the explosion-proof measuring tube stops working, flow channel inside the pulse generator is restored, mud pump pressure is changed to normal value, then directional drilling is started. The test result shows that the device runs stably with correct measurement data and improves efficiency and safety of directional drilling, which can meet directional drilling need of all kinds of coal-seam or rock-seam drilling holes and realize measurement while drilling of traditional drilling trajectory.

In order to solve problems of short usage time, poor signal stability, high fault ratio and high maintenance cost of existing cable measurement while drilling device for coal mine drilling which was powered by battery in hole, a technology scheme was proposed which used explosion-proof computer to power supply measuring tube in hole. By researching design of constant voltage source and signal demodulation of explosion-proof computer, design of parameter measurement, secondary power supply and signal carrier transmission work of measuring tube and development of measurement software, a cable measurement while drilling device was developed which was powered by explosion-proof computer. Measurement accuracy test of the device is taken in a professional organizations as well as work stability test in Shizuishan No.2 Coal Mine and depth test in Sihe Coal Mine. The tests results show that the device is feasible with higher measuring accuracy than the design value, and has strong working stability with drilling transmission distance of 1 881 m. The device can meet need of directional drilling with different types in coal mine.

In view of problems that manual checking ore-pass method cannot observe wall of mine ore-pass clearly and has hidden danger, a panoramic scanning device of mine ore-pass was designed, which was composed of data collection system, hoisting system, positioning system of ore-pass depth and data processing and video monitoring system. The data collection system collects image data of wall of ore-pass through four symmetrically arranged cameras, the image data is transmitted to the data processing and video monitoring system through gigabit network line after processed by gigabit network exchanger. The data processing and video monitoring system expanses the image of the wall and displays it on video monitoring interface. The positioning system of ore-pass depth collects depth information of scanning probe in ore-pass, and positioning program of ore-pass depth calculates depth of the scanning probe in the ore-pass according to circumference of number axis and the depth information. Actual test result shows that the device can realize panoramic scanning for ore-pass with high-depth, large-diameter and lightless.

In view of problem of higher electricity consumption of air compressor, the paper proposed an optimal control scheme of air compressor unit, namely building dispatching model of air compressor unit to realize goal of constant-pressure and energy-saving. Optimal control principle of the model is as follows: when pipe network pressure changes, making constant-pressure decision, energy-saving decision and distribution decision for situation of each air compressor in air compressor unit to maintain network pressure in 0.58-0.65 MPa. The test result validated feasibility of the model.

In view of the problem that initial control signal is defined by experience in traditional excitation controller of synchronous generator based on iterative learning control algorithm, the paper proposed a design scheme of excitation controller of synchronous generator based on lazy learning type iterative learning control algorithm. The scheme introduces lazy learning algorithm into iterative learning control algorithm and uses lazy learning algorithm to calculate initial control signal, so as to reduce error between initial control signal and the ideal one. The simulation result showed that the controller has quick convergence speed and higher ability of maintaining terminal voltage.

By strict theoretical analysis and calculation, a mathematical model of synchronous motor of E-L equation in synchronous rotating coordinates was established and nonlinear model of inverter was given. Based on the models, a passivity asymptotical tracking controller of synchronous motor was designed in case of time-varying and unknown load torque(rotation rate and location). The controller realizes mechanical control target in meanwhile realizes control of magnetic-field direction and magnetic-field amplitude, namely realizes magnetic-field vector control. The simulation result showed that the passivity controller can realize tracking control of rotation rate and location, which has high dynamic control performance and robust.

The paper introduced hardware and software design of intelligent controller based on CAN bus and its application scheme in mine safety monitoring system. The controller takes STC10F08 microcontroller as control core, uses UDP protocol and Ethernet to communicate with host computer, and uses CAN bus to communicate with underground monitoring points, so as to achieve communication between host computer and underground monitoring points. Practical application showed that the controller has good stability, strong anti-jamming capability and fast data transmission.

Taking the design of monitoring and control system of compressor house of north ventilation shaft in No.8 Coal Mine of Pingdingshan Tia'an Mining Co., Ltd. as an example, the paper introduced characteristics and parameters of TPC1063E touch screen and MCGS configuration software, and gave general steps of configuration of the touch screen. It also gave implementation of project analysis, engineering framework, data object, graphic making, flow control, program debugging, driver programm of connection device and integration testing according to concrete conditions. The practical application showed that the touch screen works well and has easy operation because of friendly interface, which provides a guarantee for safe equiblast underground.

For the problem of power consumption caused by delay of control mode of current air-compressed system used in mine, a control model of air compressors based on Multi-Agent was proposed by comparing with advantages and disadvantages of several control modes. Model framework of Multi-Agent was given, function and interaction mode of each Agent in the framework was described, feasibility of the model simulation was discussed, and a simulation way was put forward which uses object-oriented technology and ACL to simulate model.

The paper introduced working principle of power balance technology in multi-motor driving system of coal mine and gave application of the technology in belt conveying system of a coal mine. The application showed that using power balance technology in belt conveying system can make currents of master motor and slave motor change with loads synchronously, so as to reduce ratio of fault shutdown caused by power unbalance of frequency converter.