【摘要】 为了便于钻机自动送钻系统模糊最优控制的离线仿真寻优 ,以实际钻井现场测试的送钻过程数据为依据 ,建立了三层BP神经网络 ,拟合钻井送钻过程模型。以系统输入与输出之间的误差平方积分最小为优化目标 ,以模糊逻辑控制的量化因子、比例因子、修正因子、模糊量隶属函数的类型和模糊量论域的界限等参数为优化变量 ,以送钻过程神经网络模型为被控系统 ,进行离线仿真寻优 ,得出模糊控制器的最优参数 ,并以这些参数为基础进行自动送钻系统控制实验。研究结果表明 ,用三层BP神经网络构建送钻系统模型 ,拟合效果好 ,收敛快 ;先离线仿真寻求控制参数的最优值 ,以此为基础再进行在线控制是可行的 ;对于钻机自动送钻系统 ,模糊最优控制比常规模糊控制的控制效果更好。更多还原

【Abstract】 In order to search optimum parameters of fuzzy logic controller as off-line simulating, a three-layers BP neural network was established based on the data from practical drilling, which was used for model of bit feed process on rig. The optimum target of the controller integral was minimum of error’s square with the adjusting parameters including fuzzy factors, scale factor, modifying factor and limits of the region in which fuzzy values were taken, and so on. The optimum parameters were obtained as off-line simulating again and again usedcalculating method that combined random searching with complex shape method in BP network model of bit feed process. Experiments were done with the parameters and normal parameters. Results of simulation and experiment show ed that the model of bit feed processwas well imitated hy using three-layers BP network, and the method was feasible in which off-line simulating was made when optimum parameters were searched, and on-line control was taken after optimum values were gotten.And the fuzzy optimum controller has better control effects thannormal fuzzy controller for automatic bit feed system on rig.更多还原