Abstract:  Burning is a very complex process in a rotary kiln, for lack of temperature detection in the past. A three-color temperature measurement formula was established on the basis of three primary color principles and the plank law. Combined with digital image processing and optical technology, a method of temperature detection based on the BP neural network was put forth. The software and hardware design of this system were presented, and the engineering realization was also given. The results show that this system is simple and convenient for practical application, which has higher measurement accuracy and better suitability than the traditional detection method. This technology can be used to realize the on-line detection of the temperature field, and evaluate the combustion situation in the furnace in conformity with reality.

Key words: three-color temperature detection, rotary kiln, digital image processing, BP neural network