In order to improve fault-tolerant ability of circuit system in high voltage and strong electromagnetic environment

reduce the fault rate effectively

and improve the robustness of the system

we proposed a redundancy fault-tolerant circuit design method based on feature mapping and heterogeneous evaluation. The method was based on evolvable hardware design ideas

and combined with the feature mapping algorithm of neural network. By analyzing

the characteristics of excellent sub circuits were obtained and input to the neural network for mapping clustering learning

and several different circuit sets were obtained by clustering. The heterogeneous circuit evaluation and selection strategy was designed

the heterogeneous degree correlation among sub circuits were researched

and the best individual in the sets of high heterogeneous degree was selected

which was composed of the heterogeneous redundancy circuit system. The results show that the newly constructed system can still maintain high stability and fault-tolerant ability under affected by interference or local damage. In the simulation experiment of single point short fault

single point disconnect fault

and random fault

the optimal fault-free rate of the new heterogeneous redundancy system designed in this paper is 92.5% and average fault-free rate is 90.5%. Compared with the random redundancy circuit system

the average fault-free rate is improved by 6.5%

and compared with the homogeneous redundancy circuit system

the average fault-free rate is improved by 26.5%

which proves that the high heterogeneous redundancy circuit system has better fault-tolerant ability and reliability. The research results provide a new way to solve the problem for the robust circuit under complex electromagnetic environment.