Adaptive Weighing System With Fast Nonstationary Filtering and Centrifugal Force Compensation

In this paper, a dynamic fruit sorting machine weighing system has been studied and optimized to improve weight measurement precision. This new approach relies on the combination of a fast nonstationary filtering technique with a centrifugal force compensation method. The proposed architecture uses the distributed approach, where an intelligent sensor captures a raw signal and applies an adaptive filter. Subsequently, the value obtained is transmitted to a central unit that applies a weight compensation function to improve measurement precision. The second stage considers regression models to capture the existing relationship and introduces the correction compensating working principle of the sorting machine. All provided analyses used the real signal acquired under industrial conditions from testing the sorting machine prototype. Moreover, the classical filtering approach has been used for the purposes of comparison and performance evaluation. The application of the developed scheme allows one to achieve an average absolute error of about 3 g and standard deviations do not exceed 1 g for any of the operating modes in the range up to 1 kg.