The weighing algorism using notch filters (On the improvement of the ability of weighing for checkweighers)

Abstract

A checkweigher is an automatic machine to measure the weight of in-motion products. We propose the mass estimation method for checkweighers only employs an adaptive FIR filter and a notch FIR filter. To minimize calculation load, the adaptive filter is composed of two coefficients to be updated every sampling period. It estimates the mass of object to be weighed and the natural frequency of the vibration of the mechanical structure for weighing in order to check the validity of the estimated mass. However the actual vibration of the mechanical structure for weighing often includes parasitic vibratory components such as the components generated by the collision of the object and weigh belt and caused by the micro-tremors of the floor where a checkweigher is installed. The notch filter is composed of notch zeros which cancel some of components of the tiny vibrations previously known and zeros for shaping which suppress the gain of the high frequency band. The filter significantly attenuates the vibration components of the weight signal filtered by the adaptive filter from the observation of its output signals. If the filtered weight signal is sufficiently smooth and its part available for estimating the mass of a product is fitted to a straight line, the slope of the line is almost equivalent to zero. By setting upper and lower thresholds for this slope and the natural frequency estimated using the adaptive filter mentioned above, the timing and duration for estimating the mass can be determined without any hardware device. In the present paper, with the detail observation of results obtained by applying the mentioned above methods, which are combined into one algorithm for weighing, to an actual weight signal, we demonstrate the availability and know-how for the implementation of the algorithm.