从舌象的特点和基于舌象的八纲辨证中医知识的参数形式,并结合各证型的主证分析因子,利用改进遗传算法对构建的前馈式神经网络进行优化,是对目前中医诊断神经网络方法运用的一种新发展。该遗传算法结合了实数编码、数学交叉、PSO等思想,对于形成一种中医专家系统的软计算构成技术奠定了基本的方法。仿真实验结果表明,该方法具有较为理想的效果。
This paper put forward an improved genetic algorithm to optimize a feed-back neural network( NN ), which based on the characteristic of tongue feature and the data format of TCM eight principal syndromes. It was a new development of the use of the TCMD NN technology. The improved genetic algorithm combined the thought of real coding, arithmetic recombination, PSO, etc, which laid a foundation to the combined software computing technology of the TCM expert systems. Simulation results show that this algorithm can achieve preferable effect.