Bp Neural Network Implementation On Real-time Reconfigurable FPGA System For A Soft-sensing Process

In this paper, the algorithm and structure of BP NN (backpropagation neural network) and its training process used for a soft-sensing process are described and its implementation on real-time reconfigurable FPGA (field programmable field array) system is introduced. The whole system is totally controlled by a microprocessor chip which can completely manage to in system reconfigure FPGA between the two models: BP neural network model and its training model. That is, only one single FPGA is configured with multifunction. This technique can be widely applied into the field such as we measurement of human-body internal state, space traffic equipment, deep-sea exploration, where small size of measurement equipment is required and only one microcomputer system used for multifunction is allowed to be installed