针对Hopfield神经网络在求解旅行商问题(TSP)时出现的无效解和局部极小值问题,分析和比较两种改进的求解方法,首先从理论上证明算法的有效性,然后对两种算法分别进行计算机仿真,探讨网络收敛于全局有效解的途径。研究表明,改进的算法都可避免无效解,在求解10城市问题时可获得近乎100%的最优解。 To solve the problem of invalid solution and local minimum when Hopfield neural network (TSP) is solved, two improved methods are analyzed and compared. First, Computer simulation is carried out to explore the ways in which the network converges to a globally valid solution. The research shows that all the improved algorithms can avoid invalid solutions, and nearly 100% of the optimal solutions can be obtained when solving the 10-city problem.