Kalman 滤波的基础：实用指南，第三版...

简介：
This document offers a comprehensive guide to constructing Kalman filters, detailing the application of these filtering equations to a variety of practical scenarios. A multitude of detailed examples are provided, illustrating the diverse methods available for designing Kalman filters. To facilitate understanding and verification, accompanying computer code in FORTRAN, MATLAB™, and True BASIC accompanies each example, enabling readers to validate concepts and investigate challenges extending beyond the text’s boundaries. Occasionally, the authors deliberately incorporate errors into initial filter designs to demonstrate the consequences of improper filter operation. The text systematically establishes a problem before introducing the Kalman filter formulation, fostering an intuitive comprehension of the issue under consideration. Given that real-world problems rarely appear as differential equations and frequently lack unique solutions, the authors present several distinct filtering approaches. Readers will develop proficiency in evaluating software and performance considerations to determine the optimal filtering strategy. The enhancements incorporated into this edition are directly responsive to inquiries and feedback received from readers. This third edition includes three newly developed chapters exploring less common topics related to Kalman filtering and other filtering techniques rooted in the method of least squares. Chapter 17 examines a specific type of filter – termed the fixed or finite memory filter – which retains only a limited number of past measurements. Chapter 18 demonstrates how the chain rule from calculus can be leveraged for filter initialization or even entirely bypassed when filtering is unnecessary. A realistic three-dimensional GPS example is utilized to illustrate the chain-rule method during filter initialization. Finally, Chapter 19 illustrates how a collection of linear sine-wave Kalman filters, each individually calibrated to a different sine-wave frequency, can be employed to estimate both the actual frequency of noisy sinusoidal measurements and provide estimates of states associated with those sine waves when measurement noise is minimal.