【摘要】 针对曝气池污泥体积指数(SVI)难于在线测量的情况,提出了一种基于改进型的BP神经网络对SVI软测量的模型,该模型利用改进粒子群优化算法对BP神经网络进行权值调整。为了提高BP神经网络的学习性能,使粒子群的惯性权值按照对数规律变化,同时引入自适应变异算子对局部最优的粒子进行变异,然后利用粗糙集理论消除冗余信息,得到输入变量。该模型对实际污水厂的SVI值进行软测量预测,并与其它软测量模型进行比较,结果表明了该软测量模型具有较好的精度。更多还原

【Abstract】 Due to the aeration tank sludge volume index( SVI) is difficult to measure online. A BP neural network is used for SVI soft sensor in this paper. To improve the predictive accuracy of the BP neural network,BP neural network based on the improved particle swarm optimization ways has been put forward,in which the inertia weight varies according to a logarithmic scale,while the adaptive mutation operator is introduced for the variation of the local optimum particle to improve the particle swarm algorithm optimization ability. The rough set theory is used to eliminate redundant information and BP neural network based soft sensor model is established. Then the model is used to simulate the actual wastewater treatment plant SVI,and it is compared with other soft sensor. The results prove the reliability and validity of the soft sensor model.更多还原