【作者】 缴春景；

【导师】 朱小平； 职黎光；

【作者基本信息】 西安石油大学 ， 机械电子工程， 2008， 硕士

【摘要】 交流变频驱动抽油机是近年来在游梁式抽油机的基础上应用变频调速技术发展起来的新技术,对其各部分及总体的适用控制方法正在形成。在工程实际中,需控制的系统常常存在着不确定性、高度非线性、时变、时滞等特性,要建立精确的数学模型往往较为困难。因此本文首次将基于遗传算法的模糊神经网络控制应用于交流变频驱动抽油机系统,以求获得良好的计算机实时控制效果。本文的硬件系统是由原有的游梁式抽油机模拟实验装置改造成的交流变频驱动抽油机控制系统实验装置。本文使用的控制算法是:模糊逻辑控制与神经网络控制相结合,利用遗传退火算法对模糊神经网络控制器参数进行综合设计、优化,通过控制器性能指标离线寻找最优的控制器结构和参数。目的是将该算法应用于交流变频驱动的抽油机控制系统中,同时与一般模糊控制进行对比研究,以验证该控制算法的可行性和优越性。通过实验得出:一般模糊控制稳态误差为3%,基于遗传算法的模糊神经网络最优控制的稳态误差1.8%。通过不同控制算法的实验结果得出,基于遗传算法的模糊神经网络最优控制在系统中控制效果更好,表明应用于交流变频驱动抽油机控制系统是可行的。本文通过理论分析与实验研究,将基于遗传算法的模糊神经网络最优控制应用于交流变频驱动抽油机系统是非常有意义的,为该课题的进一步深入研究奠定了基础。更多还原

【Abstract】 The beam pumping unit driven by AC frequency conversion is newly developed technology based on the conventional beam pumping unit. Owing to the existence of undetermined, nonlinear, time-varied and time-lag factor in Engineering contol system, it is difficult to build the precise mathematic model. In this paper, a fuzzy-neural network besed genetic algorithm is applied to the well pumping unit driven AC frequency conversion in order to obtain the better real time effect.The hardware system that is experiment installment of AC control system in well pumping unit is refitted based on the simulating experiment installment of well pumping unit. The control algorithm is described as follows: With a controller of fuzzy logic combined with neural network, parameters of the controller are optimized by genetic algorithm. Then optimal structure and parameters of the controller are obtained by offline. The main purpose is that the control algorithm can be applied to the pumping unit driven AC frequency conversion. In addition, the feasibility and advantage of this algorithm had been verified by compared with the general fuzzy controller. The results of experiment show that stable error of general fuzzy controller is 3% and the stable error of optimal controller of fuzzy-neural network based on the genial algorithm is 1.8%.The experiment results of different control algorithm show that the control effect is better in the system with the optimal control of fuzzy-neural network based on the genetic algorithm. So the control system of AC frequency conversion driven of well pumping unit is feasible.Through theoretical analysis and experiments, it is significant that the optimal control of fuzzy-neural network based on the generic algorithm, the proposed scheme established a basis of advanced study.更多还原