本文提出了一种基于遗传算法的数字电路测试图形生成方法，首先把被测电路的门级描述转化为易于计算的非线性网络，然后用遗传算法找到网络能量函数的最优解，从而得到被测电路的测试集．这种方法对可测故障都能生成测试，能方便地产生多故障的测试图形，同时具有较好的并行性，易于在多处理机上实现． In this paper, we present a digital circuit test pattern generation method based on genetic algorithm. Firstly, the gate-level description of the circuit under test is transformed into an easy-to-calculate nonlinear network. Then the optimal solution of network energy function is found by genetic algorithm, Test circuit test set. This method can generate test for measurable fault, can easily generate multi-fault test pattern, and has better parallelism and is easy to implement on multiprocessor.