Judgment Based On The End Of The Wavelet Neural Network Inversion Mouth Flame Temperature Secondary Steelmaking

A new method for the end-point control of basic oxygen furnace (BOF) is put forward, which makes use of the temperature information of flame in the converter mouth to assist end—point control. Wavelet neural network is used to measure the temperature of the flame in the furnace mouth. The experimental system including flame image capturing system, data analysis and processing system is established. The flame image capturing system is composed of telescope and digital camera construct.In the process of the wavelet neural network calculating temperature, HSV model is used as the pattern character vectors. Information of flame picture is captured and color information in the furnace mouth is converted from RGB space to HSV space. The wavelet neural network is designed with matlab. Using the information of the mean value of hue and saturation in each picture, the wavelet neural network can deduce the corresponding temperature. With the premise that there is correlation between high temperature object and color, the main work is to design appropriate wavelet neural network with software so as to obtain the temperature information from the color information. In the process of designing wavelet neural network, appropriate wavelet function and training method were selected. The choice of wavelet function and training method is important, because appropriate wavelet function can get exact result and suitable training method can improve the training speed of wavelet neural network and avoid that arithmetic get into minimum value. Furthermore, the merits of wavelet neural network in approaching function are introduced and that using wavelet neural network to assist end-point control by calculating temperature is proved to be feasible.