Research On Data Processing Of Coal Mine Water-inrush Monitoring System Based On Bp Neural Network

In product and consume coal, our country is the most leading of the world, At present, accident occurred so frequently during exploiting coal mine, especially for gas outburst potential and Water-inrush. Every year, the Personal Casualty and property damaged caused by this accident could not be reduced. The state and the people is always focus on the question of the safety development in coal industry. Nowadays, there are still a lot of deficiencies in the system of coal mine water-inrush monitoring and early warning in our country, which is caused by a lock of real-time dynamic in the method of monitoring. This paper is focus to research the situation and the tendency of coal mine water-inrush monitoring, design a system of coal mine water-inrush monitoring with multi-parameter and real-time dynamic. The system is constituted of the data acquisition the data processing and the medium of Data transmission between the acquisition and processing system.In the part of the data acquisition, I choose the Field Programmable Gate Array (FPGA) as the kernel of the part of the data acquisition. This chapter focuses to explain the basic structure of FPGA and design flow. Choose the EP1C6Q240C8chip of the Cyclone series which made in Altera Company as the core of the system. And then, in order to contort to the FPGA, choose the hardware circuit such as power circuit、clock circuit and AD converter circuit to prepare for the data acquisition system. Also, this paper has researched each part of the hardware circuit and the design of software will be introduced.In the part of the data processing, I choose the BP neural network as the kernel. The data will be processed and forecasted by means of Mixing programming form LabVIEW and Matlab, which means that calling a Matlab Script node in LabVIEW, whose powerful data processing function has been used. In the soft of Matlab, calling a feed-forward neural network by programming, and according actual situation, make sure that input and output layers have4neurons in this BP neural network. And then train each network which contains3,5,10,15or20neuron on hide layer. Finally, I choose the structure of BP network is4×10×4which having the optimal error curve according to all error curves. Testing the network by test sample, generated a BP neural network which has high stability, high precision and could predict data. Complete data access module by programming on LabVIEW.