Integrating advanced measurement and signal processing for reliability decision-making

Edward Kozłowski,Jarosław Sęp,Tomasz Żabiński,Katarzyna Antosz,Dariusz Mazurkiewicz

doi:10.17531/ein.2021.4.20

Edward Kozłowski, Jarosław Sęp + Show 3 more

Open Access

https://doi.org/10.17531/ein.2021.4.20

Copy DOI

Abstract

An advanced milling machine multi-sensor measurement system as a condition monitoring tool was presented. It was assumed that the data collected from the 3-axis force and torque sensor can be used as a new approach and an alternative to the typical vibration signal based health monitoring and remaining useful life prediction (RUL), when integrated with machine learning techniques that are regarded as a powerful solution. Measurement system integration with the proposed signal processing method based on decision trees with different types and levels of wavelets for the cutter reliability decision-making process was presented together with proving their ability to trace the tool condition accurately. Prediction errors achieved with the use of different signal sources and data processing methods were presented and compared.

Highlights

Innovative technological machines are constructed as advanced mechatronic systems facing extremely high demands with respect to their performance, reliability and product quality
The features of the signals correlating to the tool wear are captured to monitor tool condition and to do this, a mass of signal processing methods were used, such as time series modeling, Fast Fourier Transform and time–frequency analysis, the amount of data gathered and calculation involved in corresponding parameters with tool wear is enormous
Technological machines designed for the Industry 4.0 applications, among which are machine tools, are advanced mechatronic systems equipped with several sensors

Summary

Introduction

Innovative technological machines are constructed as advanced mechatronic systems facing extremely high demands with respect to their performance, reliability and product quality. According to the detailed analysis presented in [49], up to now, many types of sensors and signal processing techniques are used in machine tool and especially in cutting tool condition monitoring and RUL prediction. Most of these sensors are wired, mounted inconveniently on the machine during the machining operations, and the prognostic information is not easy to be integrated into the manufacturing system [28, 31, 45, 49].

Objectives

Methods

Results

Conclusion