A Comparison of Minimizing Strategies for Maximum Likelihood Identification

A well known procedure for minimizing the negative log-likelihood function which appears in maximum likelihood identification is the Newton-Raphson method. Each time a new search direction is chosen, it remains to choose a step size. Several minimization techniques for obtaining an optimum step are compared using the lateral dynamics of a wide body aircraft as an example. Optimization of the step size is shown to be important when many parameters are to be simultaneously identified. However, the multi-sweep method is shown to produce improved estimates without the need for optimizing the step size.