This paper proposes a modified exponential smoothing algorithm that minimizes noise variance while preserving the structure of the original signal. The suitability of the proposed method was evaluated by smoothing a periodic signal with a trend, corrupted by noise with various distributions.1. INTRODUCTION
Smoothing is a method of digital signal processing that involves estimating signal values at fixed moments based on both previous and subsequent observations.
In practice, this involves replacing the terms of a given series, which has an irregular course, with another one having a smoother course, which can be assumed to better represent the essence of the phenomenon than the original series. The most common method for implementing the smoothing problem is appropriate digital signal filtering, from which the term “smoothing filter” is derived, used to describe various smoothing algorithms.
The exponential smoothing model is one of the classic signal smoothing models developed by R.G. Brown [1]. (…)
The article contains 11956 characters.
Source: Logistyka Magazine