【摘要】 利用粒子群算法产生的优化解空间来训练BP网络。在优化解空间基础上, 避免粒子群算法在解决复杂多峰搜索问题时可能出现的早熟情况; 实验证明此方法的优越性。基于此训练出来的BP神经网络预测基因工程菌发酵过程中酶的活性参数。通过预测出来的酶活性参数指导补料控制, 提高单位产量。更多还原

【Abstract】 The article use the space which include all the PSO results to train the BP. Based on the space which include all the PSO results, PSO can void the premature when the PSO deal with the problem which contain many peak value,the test testify the superiority;and predict the value of enzyme based on the BP which train by PSO;We can direct the production through the value of enzyme by prediction and increase the production.更多还原