This paper's main task is to apply fault diagnosis method of artificial intelligence to thermal power generating units on-site equipment, mainly research on fault diagnosis theories and methods based on neural network, and applied it to thermal power plant condensing steam system. Based on analysis-based model of condensing steam in thermal power systems, summarizing the typical fractional condenser fault of halving style condenser, extracting the knowledge of fault samples to get a practical engineering knowledge base. This paper gets a enhanced of multiple bi-directional associative memory neural network based on the original BAM neural networks. Taking advantage of the existing perfect fault characteristics database, using the enhanced bi-directional associative memory neural networks to get the fault diagnosis knowledge model, it shows the availability and validity of the fault diagnosis method using simulation experiments. |