毕业论文_完善版.docx

编号南京航空航天大学毕业论文题 目基于B样条递推最小二乘的非线性传感器系统建模方法学生姓名周鹏学 号030830222学 院自动化学院专 业测控技术与仪器班 级0308302指导教师刘剑 讲师二〇一二年六月南京航空航天大学本科毕业设计（论文）诚信承诺书本人郑重声明：所呈交的毕业设计（论文）（题目：基于B样条递推最小二乘的非线性传感器系统建模方法）是本人在导师的指导下独立进行研究所取得的成果。尽本人所知，除了毕业设计（论文）中特别加以标注引用的内容外，本毕业设计（论文）不包含任何其他个人或集体已经发表或撰写的成果作品。作者签名：年月日（学号）：基于B样条递推最小二乘的非线性传感器系统建模方法摘 要本文首先简要介绍了人工神经网络、支持向量机、加权递推最小二乘等建模方法，并分别分析它们各自的优缺点，最后通过比较发现B样条函数具有良好的低阶光滑特性和逼近性能，能够保证建模的精度,并且B样条基函数的递推定义及最小二乘的无矩阵求逆递推使建模简单易行，当数据呈网格分布时，运用降维处理降低了运算复杂度。理论分析、仿真及实际传感器数据结果表明，B样条递推最小二乘方法能够有效地提高非线性传感器系统建模的精度、稳定性以及运算效率。所以综于以上论述，提出了基于B样条递推最小二乘的非线性传感器系统建模方法，即使用基于B样条递推最小二乘的方法进行曲线和曲面的拟合。文章的主要内容包括：1.B样条曲线的含义及特点2.二维曲线控制系数的递推最小二乘估计3.乘积型B样条曲面及其控制系数的降维递推求解4.对实际电路给出数据的软件仿真关键词：B样条，递推，最小二乘，降维处理Modeling approach to nonlinear sensor system based on B-spline recursive least-squaresAbstractThis paper firstly introduces the artificial neural network, SVM, the weighted recursive least squares and modeling methods, and analyzes the advantages and disadvantages of each respectively .Finally by comparison we conclude that the excellent approximation property of B-Spline function can ensure the modeling accuracy, recursive definition of B-Spline basis function and recursive implementation of Least-Squares without matrix inversion can make the modeling convenient,and dimension-reduced processing to the data on a rectangular mesh distribution can decreases the computing complexity. Theoretical analysis, simulation and practical sensor data processing illustrate that the B-spline recursive Least-Squares modeling approach to nonlinear MISO sensor system can improve accuracy,stability and computing efficiency effectively The article includes:the meaningand characteristicsoftheB-spline curvethe recursiveleast squares estimation of control coefficientthe product typeB-splinesurface and itscontrol coefficientdimensionality reductionrecursivesolving the software simulation ofthedatagivenon the actualcircuitKey Words：B-spline; recursive least-squares; reduced-dimension processing第一章 引 言1.1传感器非