论文部分内容阅读
本文通过神经网络和响应面法建模,研究温度、pH、氯化钠浓度和硒浓度对预测枯草芽孢杆菌的生长,碱性蛋白酶活和富硒的影响。将径向基(Radial Basis Function,RBF)神经网络和反向传播(Back Propagation,BP)神经网络这两种类型的网络,与响应面法作比较。同时,使用遗传算法优化神经网络的内部结构。通过灵敏度分析发现枯草芽孢杆菌的生物量以及碱性蛋白酶活性的变化与氯化钠浓度的变化密切相关,并且温度的变化也影响碱性蛋白酶的活性,而亚硒酸钠的浓度是影响富硒最重要的参数。通过比较实测值和预测值发现:用BP网络所预测的结果比RBF网络和响应面法更加准确。遗传算法-神经网络方法提供了一个可靠的软件工具来预测枯草芽孢杆菌的富硒过程,为实际的规模化生产奠定理论基础。
In this paper, the effects of temperature, pH, sodium chloride concentration and selenium concentration on the prediction of Bacillus subtilis growth, alkaline protease activity and selenium enrichment were studied by neural network and response surface methodology. Two types of networks, Radial Basis Function (RBF) neural network and Back Propagation (BP) neural network, are compared with the response surface method. At the same time, using genetic algorithm to optimize the internal structure of neural network. Through the sensitivity analysis, it was found that the changes of biomass and alkaline protease activity of Bacillus subtilis were closely related to the change of sodium chloride concentration, and the change of temperature also affected the activity of alkaline protease, while the concentration of sodium selenite was the key factor influencing Se The most important parameter. By comparing the measured values with the predicted values, it is found that the results predicted by the BP network are more accurate than the RBF networks and the response surface method. Genetic algorithm - neural network method provides a reliable software tool to predict the Bacillus subtilis selenium enrichment process, laying the theoretical foundation for the actual large-scale production.