Research Of On-chip Intelligent Self-recovery Technology Based On Evolutionary Hardware

During the operation of the electronic system,due to the influence of the external environment,a variety of circuit failure problems may occur in the electronic system.Although the traditional system fault recovery scheme can achieve system fault recovery under certain conditions,it will also bring a lot of unnecessary hardware resource consumption and time consumption to the chip circuit system.The unique adaptive,self-organizing,and self-repairing characteristics of evolutionary hardware can well meet the self-repairing requirements of chip circuit systems.In order to improve the safety and reliability of the chip circuit,this paper proposes a chip self-recovery scheme based on evolutionary hardware,so that the chip can automatically sense the changes of the external environment during operation,and make real-time self-diagnosis of the faults that occur in the system.According to the current environment,the circuit structure is adaptively adjusted to realize local fault self-recovery.The self-detection and self-recovery of faults are the two most important parts of chip self-repair.By analyzing the defects of traditional fault detection technology,and combining virtual reconfigurable circuits and Cartesian genetic coding,this paper designs a hierarchical and fine-grained system fault detection scheme.After the fault injection is performed on the target circuit in the offline state,this paper uses the multi-label classification algorithm to train the fault circuit dataset to obtain the fault model of the target circuit,and establishes a fault model matching library for real-time fault detection of the on-chip system,which significantly improves the real-time performance of online fault location.In the fault recovery phase,to maintain the normal functions of the circuit during the fault recovery,a fault compensation recovery mechanism is adopted to simulate the fault circuit output by dynamically loading the truth table to ensure the real-time performance of the system recovery.At the same time,in order to improve the evolutionary recovery efficiency of the chip system,two improved adaptive genetic algorithm are proposed.The improved genetic algorithms optimize the selection,crossover and mutation stages of the evolutionary recovery process,so that the algorithms continuously adjust the evolutionary algorithm during the evolution process,which improves the search ability and evolutionary convergence efficiency in the global solution space of the evolutionary algorithm.This article builds a chip intelligent self-recovery system based on evolutionary hardware on the Zedboard development board.This system comprehensively considers the fault location and fault recovery performance of the chip system's fault tolerance,and combines the proposed hierarchical fine-grained fault location and fault recovery scheme based on evolved hardware.When a circuit fault occurs,the fault source is detected and located through the fault model matching library.The fault compensation recovery mechanism is used to isolate the circuit fault and ensure the normal operation of the system.The fault configuration library is used to improve the recovery efficiency of fault circuit.For unknown circuit fault types,an improved adaptive evolution algorithm is used to perform online evolution recovery on the fault circuit,which improves the evolution efficiency and success rate of the recovery algorithm,and ultimately improves the reliability and stability of the chip system.