ON THE MEASUREMENT OF RANDOMLY TIMEVARYING LINEAR SYSTEMS,
Abstract
It is shown what measurements to perform on a system that is randomly time-varying and linear, but otherwise unknown, to completely characterize it. The success of these measurements depends on a representation of a randomly time-varying linear system different from those commonly used. It is shown what measurements to perform on a randomly time-varying linear system to predict the auto-correlation function of the output signal from the autocorrelation function of the input signal when the input is a stationary random process. It is shown what measurements must be performed on a randomly time-varying linear system to design a linear open-loop controller which is optimum in the mean square sense. These measurements are simpler than the measurement of the autocorrelation function transfer relation. Experimental results concerning measurement and system performance were accomplished by applying a white Gaussian test signal to the unknown system and measuring statistical averages of quantities obtained by jointly manipulating the input and resulting output. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1965
- Accession Number
- AD0624693
Entities
People
- Gerald E. Pollon
Organizations
- Massachusetts Institute of Technology