基于改进的粒子群—小波神经网络的固井质量智能评价  

An Intelligent Evaluation Method Based on Improved PSO-WNN for Cement Bond Quality

作　　者：张伟[1] 师奕兵[1] 周龙甫[1] 卢涛[1,2] 

机构地区：[1]电子科技大学自动化工程学院,四川成都611731 [2]中海油田服务股份有限公司技术中心,北京101149

出　　处：《信息与控制》2010年第3期276-283,共8页Information and Control

基　　金：国家863计划资助项目(2006AA06Z222);教育部新世纪人才支持计划资助项目(NCET-05-0804)

摘　　要：为了克服传统的相对幅度法在固井质量评价中识别率低下的缺点,提出了一种基于改进粒子群一小波神经网络的固井质量智能评价方法.首先在应用李亚普诺夫理论分析得到单个粒子收敛条件的基础上,提出一种粒子群改进算法,接着利用该算法来优化小波神经网络权值.应用Iris标准分类数据集对本文算法进行测试,结果表明该改进算法与BP-WNN、PSO-WNN等经典算法相比,网络不仅易于全局收敛,而且迭代次数、函数逼近误差、分类精度等性能特得到提高.最后用训练好的改进粒子群一小波神经网络对某实验井声波固井质量测井实测数据进行分类识别.结果分析表明,该方法极大提高了水泥胶结情况的识别能力,是一种高效、实用的固井质量评价方法.An intelligent evaluation method based on improved PSO-WNN（particle swarm optimization-wavelet neural network） for cement bond quality is presented because the pattern recognition accuracy of traditional relative amplitude methods are very low.Firstly,Lyapunov stability theory is used to discuss the convergence conditions of a single particle.Then, based on the results,a new strategy is introduced to improve the performance of the PSO algorithm.Secondly,the improved PSO algorithm is used to optimize the parameters of the WNN.The Iris benchmark data set is used to test the proposed algorithm. The improved method is compared with classic algorithms like BP-WNN and basic PSO-WNN.Simulation results confirm that the new algorithm not only has the performance of rapid global convergence,but also the iterative number,error of the function approximation and the classification accuracy of the new classifier are highly improved.Finally,the trained IPSO-WNN is used to identify the acoustic cement bond quality logging data.Experimental results show that the recognition ability of cement bond is improved greatly by means of the new method and the new method is high efficient and practicable for evaluation of cement bond quality.

关 键 词：固井质量评价 粒子群算法 小波神经网络 全局收敛 分类识别 

分 类 号：TE256[石油与天然气工程—油气井工程] TP183[自动化与计算机技术—控制理论与控制工程]