The Design And Implementation Of Flame State Detection System Based On FPGA

Burning is a complicated process, including severe physical change and chemical change. At the same time, burning is necessary in our life, it makes the material change into energy used by us. How to control burning to be in the reasonable scope, make burning to be in the highest efficiency is design principle of combustion state detection system. The flame combustion state detection system after several generations of development, now the technology like the analysis of the dynamic image for the flame burning flame state has appeared. The flame images as a flame burns direct form of expression, with more information and easy to be got, are very suitable for analysis of the state of flame. The FPGA can meet the needs of high speed image processing with its hardware features. SOPC technology can make the design flexible, software and hardware in-system programmable and update.We design a flame combustion state detection system based on FPGA, with the DE2development board of Altera company for hardware platform, using CMOS (Complementary Metal Oxide Semiconductor) image sensor to collect the flame image, VGA (Video Graphic Array) to show the flame temperature field, and the real-time characteristics and the burning flame state are displayed through the LCD (Liquid Crystal Display,). According to flame images from CMOS, the flame combustion state detection system using the FPGA to realize several characteristic features like the center point of the fire, color, temperature calculation and area calculation algorithm of image will be extracted by the flame. They will be display by LCD,at the same time, they will be the input amount of BP neural network for getting the current state of the fire burning. Neural networks is the core module of the system, bear the system’s main state judge function. The system description of a complete system by Verilog language in Quartus Ⅱ to complete, including equipment initialization, data collection and storage, pin description and do on. Soft core design and software algorithm is completed in Nios Ⅱ programming environment by C. By testing in burning environment, CMOS cameras by setting can get relatively real flame images. Through the analysis of the images of fire, the fire burning flame area and the center, the temperature distribution of color parameters are accurate. They can really express burning characteristics. The neural network can accord the characteristic values of flame judge the flame combustion state after more than200times training. It is high accuracy, and can real-time judgment and display.