A novel Adaptive Filtering approach to Drive File Identification for Service Environment Replication

Service Environment Replication refers to the process of using test machines to apply controlled dynamic loads to test articles in order to replicate operating conditions that the article was designed for. Such test machines hence require the development of dynamic time series commands that drive the actuators in order to replicate the responses of the actual dynamic system measured separately in its service environment.

A novel adaptive filtering approach, called the Pulse Train Filtered-x Least Mean Square algorithm for waveform generation and drive file identification is proposed in this thesis based on methods developed for Active Noise and Vibration Control. Simulation studies are considered using various test benches with varying degrees of nonlinearity to validate the performance of the proposed algorithm to rapidly converge to a dynamic solution in a small number of iterations. The PT-Fx-LMS algorithm is also shown to enable targeted iteration over isolated time slices within the data set, which challenge conventional iterative DFID techniques. Further modifications to the algorithm are proposed that uses a completely offline workflow using the estimated dynamics of the plant and an empirical termination criteria to improve performance and ensure stability of the adaptive process. The architecture developed is applicable for a wide array of dynamic systems with single or multiple actuators and sensors. Experimental validation of the proposed algorithm is conducted using an acoustic setup to replicate target sound fields for a wide array of configurations.