| As a key component of Chinese energy power,the coal plays an important role in promoting sustainable economic and social development.Coal mine underground tram locomotives have the characteristics of frequent operation,large transportation volume and long running distance because their transportation tasks,such as coal gangue,equipment and personnel,and it has become an important part of coal mine auxiliary transportation.Most of the traditional coal mine electric locomotives rely on the driver’s control.Due to the narrow underground roadway,insufficient lighting conditions,and driver fatigue,safety transportation accidents often occur,and the electric locomotive accidents rank first in the roadway transportation accidents.Therefore,it is necessary to adopt an unmanned driving system for coal mines electric locomotives to free drivers from the heavy work underground,because it can reduce the probability of safety accidents,improve the efficiency of coal mine transportations,and it can also further ensure safe and efficient production of coal mines.The paper makes in-depth research on the key technologies of unmanned electric locomotive system in coal mine,such as image noise reduction enhancement algorithm,multi-target detection algorithm,obstacle recognition algorithm and unmanned electric locomotive control system.The main research work is as follows:(1)Research on high noises image of noise reduction which based on improved wavelet threshold functions combined with bilateral filter.Aiming at the disadvantages of the traditional image noise reduction,especially the discontinuity of function,constant deviation and inability to achieve self-adaptation in the traditional wavelet threshold function,a coal mine underground high-noise image noises reduction model based on the improved wavelet threshold function combined with bilateral filters were established.That is,the wavelet threshold function is improved and an adaptive weight factor is introduced,and the image after wavelet threshold noise reduction is subjected to bilateral filtering for secondary noise reduction.Compared with the traditional image noise reduction model,the proposed model has a higher noise suppression ability,which not only improves the clarity of the noise image,but also further preserves the image detail informations,and it has superiority and universality in removing noise in coal mines.(2)Research on image enhancement technology after noise reduction based onγ-CLAHE algorithm.Aiming at the inability of conventional image enhancement algorithms to solve the problems of image noise suppression,the technical research on enhancement after image noise suppression was carried out,and an image enhancement model after noise reduction was established in HSV space through gamma transformation combined with CLAHE algorithm.The results showed that compared with the traditional image enhancement algorithm,the proposed model had significantly improved the visualization effect of the denoised image,the color saturation has been further improved,the edge information of the image details was better preserved,and the amount of information had been further improved compared with the original image improvement.(3)Research on multi-objective detection technology for coal tunnel based on improved two-stage instance segmentation algorithm.Aiming at the low detection accuracy of small targets and missed detection and wrong detection in unmanned electric locomotive driving scenes in coal mines,a two-stage instance segmentation network Mask R-CNN is proposed,which combines compression-excitation network and hybrid atrous convolution and reduces Methods for baseline network depth to improve models.Compared with the model before improvement,the proposed method has improved the features of utilization and information relevance,it has further reduced the probability of missed detection and wrong detection of targets,and it can also effectively realize multi-target detection in different driving scenarios of coal road electric locomotives.(4)Research on obstacle recognition and ranging technology based on improved single-stage instance segmentation algorithm.Aiming at the problem of how to realize the obstacle identification and ranging of unmanned electric locomotives in coal mine underground roadways,a technical framework for obstacle identification and ranging of underground electric locomotive rails based on the improved single-stage instance segmentation algorithm YOLACT++is proposed.The strategy of path enhancement and replacement of classification loss function is adopted to make full use of the underlying detailed feature informations and to improve the model learning ability.Further,we construct the effective driving area to form an obstacle identification and ranging technology framework,and realize the obstacle identification and ranging functions in the effective driving area of the electric locomotive.(5)Research on active control strategy technology of electric locomotive based on edge detection equipment.Aiming at the problem of how to realize active obstacle identification and autonomous control operation of coal mine electric locomotives,and further realize the unmanned driving of electric locomotives,it is proposed to compress and prune the instance segmentation network model and transplant it to edge detection equipment,and further improve obstacle identification and measurement.distance plan.By designing the experimental system platform and using the running video of the coal mine electric locomotive to conduct experiments,the results have showed that the proposed electric locomotive control strategy can effectively identify the obstacles in front of the electric locomotive driving rail and beside the rail and calculate their distance in real time,it can also effectively control the driving state of the electric locomotive,and provide a feasible solution for the autonomous control operation of unmanned electric locomotives in coal mines.Figure [65] Table [17] Reference [108]... |
PDF Full Text Request