Energy consumption monitoring of the steam pipe network based on affinity propagation clustering

The steam system is an important part of chemical utility system, but there are widespread phenomenon about lack of testing information, energy consumption configuration depend on given experience and wasting energy. So this paper puts forward a method about the steam pipe network system´s status identification of different energy consumption based on the steam pipe network´s characteristics of complex structure, much steam equipment, lack of testing information and difficult to build accurate mathematical model. The method based on affinity propagation clustering that can solve big set of data´s clustering problem quickly and effective. As it is hard to find preference parameters and damping factor, this paper uses PSO to find the most optimal parameters in order to achieve the best clustering effect. This method is applied test both in classic data set and the steam pipe network of ethylene plant´s status identification, the results show the effectiveness of this method.