The site selection of refuse incineration power plant has many influencing factor and is of severe social impact. In this paper,an optimizing algorithm for site selection of refuse incineration power plant is provided based on principal component analysis and back propagation (BP) neural network. The samples are selected by principal component analysis,as a result that a few samples can reflect as many sample characteristics as possible. The neural network is trained by Levenberg-Marquardt BP algorithm,and the training speed is improved.