论文部分内容阅读
结合改进的遗传算法（ＩＧＡ）和误差反传算法（ＢＰＡ）训练人工神经元网络，用以对三元不对称有机磷构效关系进行研究。作者保留了ＢＰＡ作为权值基本训练方法简捷的优点，又利用遗传算法的全局搜索性，以克服ＢＰＡ陷入局部极小点的缺陷，使两者集成在一起，通过因子α调整遗传算法和误差反传算法的结合程度，发挥各自的长处，达到训练过程的优化。通过实例表明建立了更准确的数学模型，具有较好的实用价值。
The artificial neural network was trained with improved genetic algorithm (IGA) and error back propagation algorithm (BPA) to study the structure-activity relationship of ternary asymmetric organophosphorus. The author retains BPA as a simple method of weight training, and also uses the global search of genetic algorithm to overcome the defect of BPA getting into a local minimum, so as to integrate the two together. Genetic algorithm and error correction are adjusted by factor α Communication method of combining the degree of play their respective strengths to achieve the optimization of the training process. The example shows that a more accurate mathematical model is established and has good practical value.