Research On Monitoring System Of Coal Mine Substation Running State Based On Machine Vision

Safety of coal industry is the focus of social attention, and security of coal’s power system is still weak. Therefore, coal production needs improving the operation stability to ensure security. Usually traditional monitoring is boring, and it is easily affected by man-made factors. More seriously, this mode may cause some danger. The mathod makes monitoring staff quickly identify monitoring target from a large number of video data by using machine vision, because that the mathod can identify running status of coal substation. At the same time, automatic alarm is achieved in abnormal conditions. The main work is as follows:a. The paper realizes the image enhancement. This paper is under the background of coal environment, and it chooses improved single scale Retinex algorithm for image enhancement by experiments of several algorithms. In fact, the processing of this algorithm is simple and efficiency. Experiments show that the algorithm can reduce background noise, increase contrast and improve operation efficiency.b. The paper realizes the light detection and pointer instrument detection. This paper identifies indicator light color on the principle of RGB color space and it adjusts the segmentation threshold to improve the recognition effect in actual situation, which makes color interferes with each other. The mathod detects the pointer of pointer instrumen by using morphological operation and Hough algorithm. In addition, this paper identifies pointer instrument through the angle. Experiments show that system reading has approached to manual reading and this method has good stability.c. The paper completes the digital display instrument detection. This paper chooses Harr-Like features for the character because of the protective film of bubble and light interference. Besides, fitness function, crossover and mutation function of genetic algorithm are changed in order to increase the overall situation. Actually, the paper uses improved genetic algorithm to optimize support vector machine parameters. Finally, dimension reduction is realized by principal component analysis, and characters are identified by support vector machine. Through the comparison of identification algorithms, experiment shows that this method has good recognition effect.d. The paper realizes the design of coal substation running monitoring software system. This paper completes system flow chart, software architecture and the specific design of various modules. Function modules includ system login, system management, video management, data management, running status detection and alarm management. Finally, the software system is completed.