METHODS FOR AUTOMATING MODEL VALIDATION: STEADY-STATE IDENTIFICATION APPLIED ON GRIPEN FIGHTER ENVIRONMENTAL CONTROL SYSTEM MEASUREMENTS

摘要

Model Validation and Verification (V&V) has historically often been considered a final step in the model development process. However, to justify model-based design decisions throughout the entire system development process, a methodology for continuous model V&V is essential. That is, model V&V activities should be fast and easy to reiterate as new information becomes available.Using a high fidelity simulation model of the Environmental Control System (ECS) in the Saab Gripen fighter aircraft as a guiding example, this paper further extends to an existing semi-automatic framework for model steady-state validation developed during ECS model validation efforts. Generic methods for identification of steady-state operation are a prerequisite for steady-state validation of industry grade physics based models against in-situ measurements. Four different established methods for steady-state identification are investigated and compared: steady-state conditions on the standard deviation estimated from in-situ measurements, conditions on the variation coefficient, t-test on the slope of a simple regression line, and comparison of differently estimated variances. The methods’ applicability, on ECS measurements in particular, is evaluated utilizing steady-state identification needs defined during Gripen ECS model validation activities.