Self-Recovery of Motor Control Circuit Based on MFNNVRC

To solve the problems of the unknown of ordinary FPGA configuration bitstreams and long time for download configuration bitstreams into ordinary FPGA in evolvable hardware, multilayer feed forward neural network was combined with virtual reconfigurable circuit to build MFNNVRC in an ordinary FPGA. This MFNNVRC structure was used to implement the control circuit of three phase brushless DC motor. In order to validate the effectiveness of this circuit design, a self-recovery experiment was carried out in the FPGA with some given partial faults. The successful result of experiment argues that the proposed circuit is a befitting evolutionary reconfigurable platform utilizing ordinary FPGAs for combinational logic circuits.