application-note · English
Continuous Data Analysis with Analog Computers Using Statistical and Regression Techniques
This EAI bulletin (ALAC 62023) demonstrates how fundamental statistical parameters — mean, variance, autocorrelation, cross-correlation, Fourier transform, and power spectrum — can be computed continuously from analog signals using the Exponentially Mapped Past (EMP) method and simple first-order filter circuits. The EMP approach eliminates the need to reset integrators by weighting recent data more heavily, enabling truly continuous on-line estimation. Linear and higher-order regression analysis is also developed, with a worked example applying EMP mean and variance to automated temperature control in the LD steel-making process.
statistical analysisregression analysisexponentially mapped past (EMP)analog computation techniques