Print Circuit Board (PCB) is a basic and important element for electronic aircraft system. Considering aviation safety, it is necessary to regularly evaluate the PCB to ensure the system stably and reliably. The best solution is to exchange a new PCB when system has any error, but it is hard to do due to the economic problem of the high cost. Since we cannot replace the PCB arbitrarily, we endeavor to find the failure unit and try to repair it. That is a challenge to find the problem for the professional and senior electrical engineer. According to this situation, we hope to design an automatic system to support PCB checking for general engineer. However, checking the PCB is a recognizing technology, it is similar to construct a software agent for electrical system. In order to construct a recognizing system for PCB checking, it is a flexible way to adopt the neural network technique to design that automatic system. In this paper, we propose an automatic system to support PCB testing and checking with neural network. Finally, we have shown some recognition examples to demonstrate the feasibility of our proposed system.