Adaptive approximate dynamic optimization : a slider-crank case study

We present the Adaptive Approximate Dynamic Optimization (AADO) algorithm and elaborate on its application for forward (dynamics as constraints) and inverse (dynamics in the objective) Dynamic Optimization (DO). The method is tailored to dynamic optimization problems where the system model contains high-fidelity (HF) model components of subsystem(s) that slow down the system model's evaluation time. In this contribution we extend prior work where we proposed to replace the system model with a local (i.e. only valid in subregions of the state space) approximate dynamic system model based on sampled model evaluations. The sampling is adaptive and localized to the optimal trajectory in correspondence with the trajectory optimizer. Here we suggest only to replace the high-fidelity subsystem models with an approximate subsystem model set. A comparison is made of the method applied in both the forward and inverse convention on a limit cycle slider-crank trajectory optimization for a high-fidelity nonlinear load model. We found that the AADO algorithm requires only 0.1% system model evaluations compared with conventional DO and only 30% compared with non-adaptive AADO. Our results consolidate the methods potential and invite for further research.

ISBN:9781728124933

Publication year:2019

Accessibility:Closed