Improvement of waste water treatment using functional magnetites and knowledge engineering techniques

活性污泥法是生活污水处理中常用的一种工艺，但活性污泥法仍有许多问题有待改进。本研究将知识工程技术与生物技术相结合。首先，应用递归模糊神经网络建立了活性污泥法出水COD的预测模型，并应用遗传算法对运行工况进行了准确估计。制备了壳聚糖磁铁矿聚集体，并将其应用于活性污泥法硝化活性的提高。聚集体对亚硝化细菌等硝化细菌有明显的富集性，对亚硝酸盐的去除效率显著提高。通过将典型的反硝化细菌固定在海藻酸凝胶中，反硝化活性也得到了提高。应用人工神经网络对半封闭水生态系统中赤潮的发生进行了非常准确的预测。该模型可防止对渔民造成严重损害，并可用于循环利用的水处理系统的优化。

Activated sludge process is often applied for domestic waste water treatment/ but many problems are remaining to be solved for improvement of the activated sludge process. This research combined knowledge engineering techniques with biotechnology techniques. At first, a prediction model for effluent COD from an activated sludge process was constructed by applying recursive fuzzy neural network and the operating condition was estimated accurately by applying genetic algorithm. Chitosan magnetite aggregates were made and applied for enhancement of nitrification activity in activated sludge process. The aggregates enriched specifically the nitrifying bacteria such as Nitrosomonas europaea and the removal rate for nitrous compounds was enhanced dramatically. Denitrification activity was also enhanced by the immobilization of typical denitrification bacteria within alginate gels. Occurrence of red tide in a semi-closed aqua-ecosystem was predicted very accurately by applying artificial neural network. This model can prevent a severe damage for fishermen and also be applied for optimization of water treatment system in recycle use.