Quantitative detection of mixed gases by method of sensor array and neural networks

摘要

The operation reliability of power transformer is related to the safety and stability of power system directly. The method, which analyses the characteristic gases in transformer oil, is an effective means. The characteristic gases include always H sub 2, Ch sub 4, C sub 2H sub 6; C sub 2 H sub 4, C sub 2 H sub 2 and CO ice. In this paper, we have tesed and analyzed these mixed gases concentration using to gas sensor array to judge the running state of transformer, and based on novel neural networks to enhance the selection of gas sensors. An improved adaptive genetic algorithm has been used in these neural networks. Experimental esult shows this method can obtain satisfactory effect successfully for decide transformer-running state.