Research And Design Of Portable Mine Gas Detector

The traditional gas detection equipment is usually put in a certain place underground mines to detect gas concentration because of their great volumn and difficulty to move. A large number of detection equipments is required if the mines are large and the number of the detection points are excessive, which increase not only the cost but also the difficulty to maintain. Also, these equipments have the following shortcomings: low detection speed, inaccurate measurement accuracy, and unable to read the result directly. Currently the gas detectors widely used in china mainly adopt 8/16 bits microcontrollers as the processors and make use of simple digitron to display data. The advantage of the system is its low-cost, and the disadvantage is its limited function. The function of storage, graphic display and data transmission can not be accomplished using this system. With the development of semiconductor technology, the function of embedded processors is increasingly powerful and the stable and reliable features have been enlarged, which make it more promising in the application field of the coal mine gas detection.A new portable ARM-based coal mine gas detector was designed for gas detection in the present work. ARM processors have features of large storage capacity and fast calculating speed. The small size of the designed detector would make it convenient to measure data movably underground mines. The detection data could be real-time displayed on LCD by list and graphic. Moreover, the system also has the functions of data storage and history data inquiry, so the data would be analyzed and handled underground mines, which would overcome the problem that the data must be transmitted to the host to be handled. There the results can be feedbacked to the miners timely, and thus the accuracy of gas detection would be improved.Meanwhile, prediction function of gas concentration was added to the detector. The genetic neural network model was adopted. An improved dual population genetic algorithm based on individual similarity(DGAIS) was proposed. Simulation experiments and results analysis had been done on DGAIS and improved neural network model. It was proved that DGAIS is an efficient and effective improved genetic algorithm according to the four indexes (the average evolutionary generation、the average maximum fitness、the searching times of the optimal value and the accuracy of the mean value). Compared to the standard BP neural network model, the iterations and the absolute errors had been significantly decreased for improved genetic neural network model.