Implementation of Facility Models Using Neural Networks for Improved Control of Wind Tunnels

Test facilities such as wind tunnels require stringent control of test parameters such as wind speed, temperature, pressure, etc., in order to satisfy the objectives of the simulation. Control of test facilities can be significantly improved using mathematical models for the facility behavior that are developed from physical principles. A shortcoming of mathematical control models is that they typically contain parameters that must be directly measured in order to match the performance of the actual facility. As a result, it is necessary to “tune” the mathematical models to match the actual facility response. Tuning of the mathematical models can be a very complex and time-intensive activity, particularly for complex facility operations requiring simultaneous control of multiple systems or actuators. The objective of the proposed research is to investigate methods of automating the data collection and analysis associated with tuning mathematical facility models, so that these activities are run simultaneously with regular facility operations and require minimal involvement of facility staff. BENEFIT: This program will result in improvements and automation in the operation of wind tunnel facilities that may be applied to government and industrial facilities.