欢迎访问《环境科学学报》编辑部网站！

快速检索			高级检索

研究报告

Prediction neural network of the flocculation efficiency and bioflocculant yield using mixed carbon sources

基金项目：国家高技术研究发展计划（No.2009AA062906）；国家自然科学基金（No.51108145，51178139）；国家创新研究群体科学基金（No.51121062）；哈尔滨工业大学城市水资源与水环境国家重点实验室自主课题（No.2013TS02）

摘要：为实现混合底物的高效定向转化，以产絮菌根癌农杆菌（Agrobactrium tumefaciens） F2为研究对象，考察不同单一碳源及不同初始浓度对菌体生长、絮凝效能及絮凝剂产量的变化规律，采用BP算法构建絮凝效能及产量预测神经网络.产絮菌F2利用葡萄糖时的絮凝效能和产量分别为88.98%和2.20 g·L^-1，过低的初始浓度将影响产量，不低于7.5 g·L^-1为佳.以D-（+）-葡萄糖、D-半乳糖和D-甘露糖3种单糖为混合碳源，构建网络结构为3-5-2的产絮效能及絮凝剂产量预测模型，对两个输出层的预测误差范围均在4%以内，预测葡萄糖、半乳糖、甘露糖浓度的最优解为6.59 g·L^-1、 1.32 g·L^-1、3.57 g·L^-1，经验证混合碳源发酵产絮可使絮凝效能和产量比单一葡萄糖发酵时分别提高6.87%和26.82%，本文为产絮菌F2利用含糖有机质废液发酵产絮凝剂提供数据参考.

Abstract：In order to improve the directional transformation of the mixed substrates, Agrobactrium tumefaciens F2 was used for bioflocculant production. The bacterial growth, flocculation efficiency and bioflocculant yield were determined in the single carbon source with different initial concentrations. The prediction neural network was carried out by the BP algorithm. All the results indicated that the utilization among the diverse carbon sources by strain F2 was different. The flocculation efficiency and bioflocculant yield produced from glucose were 88.98% and 2.20 g·L^-1, respectively, which was affected by the low initial concentration (lower than 7.5 g·L^-1). The structure of neural network model is 3-5-2 and the prediction errors of two outputs were both within 4%. The optimal concentration of glucose, galactose and mannose were 6.59 g·L^-1, 1.32 g·L^-1, 3.57 g·L^-1. The flocculation efficiency and bioflocculant yield both rose up to 6.87% and 26.82%. The prediction and convergence rate of model is desirable. This study provides data reference for the bioflocculant production using the mixture of the liquid organic wastes.

Key words：bioflocculant mixed carbon sources flocculation efficiency bioflocculant yield BP neural network