[ Objective ] The study aimed to discuss the establishment and application of Chlorella vu/gar/growth model on base of optimizing the BP neural network by genetic algorithm. [Method] The values of weight and threshold of BP neural network were optimized by using the genetic algorithm ,and with this network model,the culturing time of C. vulgari and the residual glucose as the input and the thalli light density values (OD680) as the output,the growth state of C. valgariin the 500 L multi-function biological reactor was modeled and the its apphcation was discussed. [ Result ] BP neural network optimized by the genetic algorithm had the smaller error squares of generalization value than the one without optimization,so its predicted value was more close to the real value. T test showed that the established model was credible. The verification showed that the model had the good fitting degree and could well described the relationship of the biomass ( OD680 ) of C. vulgari cultured in 500 L multifunction bioreactor with the residual glucose and culturing time. [ Conclusion] The model established in the test could be used to the prediction of the test results,which had the guiding significance for controlling the culture of C. vulga ri control.%[目的]探讨用遗传算法优化BP 神经网络对小球藻生长模型的建立与应用.[方法]使用遗传算法对BP 神经网络的权值和阈值进行优化,并使用该网络模型,以小球藻培养时间和残余葡萄糖为输入,菌体光密度值(OD)为输出,对小球藻在500 L多功能生物反应器中的生长情况进行了建模,还探讨了该模型的应用情况.[结果]经过遗传算法优化的BP 神经网络,其泛化值的误差平方和比BP神经网络的小,因而预测值更加接近实际值.t 检验表明,所建立的模型是可信的.验证表明,该模型具有良好的拟合度,能够很好地描述在500 L 多功能生物反应器中培齐的小球藻的生物量(OD)与残余葡萄糖和培养时间之间的关系.[结论]所建立的模型可用于试验结果的预测,对小球藻的培养控制具有指导意义.
展开▼
