The iterative least squares parameter estimation technique for intermittent systems with applications to intelligent building control
Abstract
A new on-line parameter estimation technique is proposed for real-time system modelling problems where input/output data is sparse. The algorithm generates the coefficients for an "n"th order ARMAX model that best firs the input/output data in a least-squares sense. The algorithm estimates the system dead time as well. A theoretical analysis is developed to prove the algorithm's remarkable stability and convergence properties. Simulation results are presented to demonstrate the algorithm's ability to produce the desired coefficient estimates if and only if the input/output data is sufficient to define a system model. Experimental results are also presented to demonstrate the algorithm's application to an adaptive HVAC control system for regulating comfort conditions in commercial office space.
Degree
Ph.D.
Advisors
Shoureshi, Purdue University.
Subject Area
Systems design|Mechanical engineering|Artificial intelligence
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.