摘要
对于多输入多输出系统,针对如何根据系统模型和期望输出反求系统输入的问题,提出了一种基于过程神经网络和量子遗传算法相结合的方法,并给出了具体的实现方法。首先根据实际系统的领域知识和学习样本集,建立满足系统实际输入输出映射关系的正向过程神经网络;然后按照系统在过程区间的某一期望输出,用过程神经网络的输出误差构造适应度函数,用量子遗传算法逆向确定系统的过程输入信号,使该输入信号满足已建立的正向过程映射关系,从而完成系统的逆向过程控制。油藏采收率参量的逆向求解结果证明了该方法的有效性。
An optimization algorithm of process neural networks and quantum genetic algorithm （ PNN-QGA ） was proposed to ascertain the input of multiple-input and multiple-output （MIMO） system from both system model and hope output. And the general realization approach was presented. Firstly, the process neural network （PNN） that represents the mapping relation between input and output of system is founded according to system field knowledge and training samples sets. Secondly, the fitness function of quantum genetic algorithm （QGA） is constructed by using PNN output error based on the hope output of process interval. The system input information is ascertained by QGA according to a certain hope output of system, and it accords with the PNN mapping relation that is founded. Hence, the converse process solution of the system is accomplished. Finally, a converse-solving example of oil recovery ratio was given to illustrate the availability of the approach.
出处
《中国石油大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第6期120-126,共7页
Journal of China University of Petroleum(Edition of Natural Science)
基金
国家自然科学基金重点项目(50643020)
关键词
过程神经网络
量子遗传算法
油藏采收率
逆向求解
process neural network
quantum genetic algorithm
oil recovery ratio
converse solution
