Chee Khiang Pang

Intelligent Diagnosis and Prognosis of Tool Wear Using Dominant Feature Identification

Identification and prediction of a lifetime of industrial cutting tools using minimal sensors is crucial to reduce production costs and downtime in engineering systems. In this paper, we provide a formal decision software tool to extract the dominant features enabling tool wear prediction. This decision tool is based on a formal mathematical approach that selects dominant features using the singular value decomposition of real-time measurements from the sensors of an industrial cutting tool. Selection of dominant features is important, as retaining only essential features allows reduced signal processing or even reduction in the number of required sensors, which cuts costs. It is shown that the proposed method of dominant feature selection is optimal in the sense that it minimizes the least-squares estimation error. The identified dominant features are used with the recursive least squares (RLS) algorithm to identify parameters in forecasting the time series of cutting tool wear. Experimental results on an industrial high-speed milling machine show the effectiveness in predicting the tool wear using only the dominant features.

Tool Wear Monitoring Using Acoustic Emissions by Dominant-Feature Identification

Identification and online prediction of lifetime of cutting tools using cheap sensors is crucial to reduce production costs and downtime in industrial machines. In this paper, we use the acoustic emission from an embedded sensor for computation of features and prediction of tool wear. Acoustic sensors are cheap and nonintrusive, coupled with fast dynamic responses as compared with conventional force measurements using dynamometers. A reduced feature subset, which is optimal in both estimation and clustering least squares errors, is then selected using a new dominant-feature identification algorithm to reduce the signal processing and number of sensors required. Tool wear is then predicted using an Auto-Regressive Moving Average with eXogenous inputs model based on the reduced features. Our experimental results on a ball nose cutter in a high-speed milling machine show the effectiveness in predicting the tool wear using only the dominant features. A reduction in 16.83% of mean relative error is observed when compared to the other methods proposed in the literature.

Intelligent Diagnosis and Prognosis of Industrial Networked Systems

In an era of intensive competition where plant operating efficiencies must be maximized, downtime due to machinery failure has become more costly. To cut operating costs and increase revenues, industries have an urgent need to predict fault progression and remaining lifespan of industrial machines, processes, and systems. An engineer who mounts an acoustic sensor onto a spindle motor wants to know when the ball bearings will wear out without having to halt the ongoing milling processes. A scientist working on sensor networks wants to know which sensors are redundant and can be pruned off to save operational and computational overheads. These scenarios illustrate a need for new and unified perspectives in system analysis and design for engineering applications. Intelligent Diagnosis and Prognosis of Industrial Networked Systems proposes linear mathematical tool sets that can be applied to realistic engineering systems. The book offers an overview of the fundamentals of vectors, matrices, and linear systems theory required for intelligent diagnosis and prognosis of industrial networked systems. Building on this theory, it then develops automated mathematical machineries and formal decision software tools for real-world applications. The book includes portable tool sets for many industrial applications, including: Forecasting machine tool wear in industrial cutting machines Reduction of sensors and features for industrial Fault Detection and Isolation (FDI) Identification of critical resonant modes in mechatronic systems for system design of R&D Probabilistic small-signal stability in large-scale interconnected power systems Discrete event command and control for military applications The book also proposes future directions for intelligent diagnosis and prognosis in energy-efficient manufacturing, life cycle assessment, and systems of systems architecture. Written in a concise and accessible style, it presents tools that are mathematically rigorous but not involved. Bridging academia, research, and industry, this reference supplies the know-how for engineers and managers making decisions about equipment maintenance, as well as researchers and students in the field.

Singular Perturbation Control for Vibration Rejection in HDDs Using the PZT Active Suspension as Fast Subsystem Observer

Currently, position sensors other than the read/write head are not embedded into current hard disk drives (HDDs) due to signal-to-noise ratio and nanometer resolution issues. Moreover, a noncollocated sensor fusion creates nonminimum phase zero dynamics which degrades the tracking performance. In this paper, the singular perturbation theory is applied to decompose the voice coil motors (VCMs) and induced PZT active suspensions dynamics into fast and slow subsystems, respectively. The control system is decomposed into fast and slow time scales for controller designs, and control effectiveness is increased to tackle more degrees-of-freedom via an inner loop vibration suppression with measured high-frequency VCMs and PZT active suspensions dynamics from the piezoelectric elements in the suspension. Experimental results on a commercial HDD with a laser doppler vibrometer show the effective suppression of the VCM and PZT active suspensions flexible resonant modes, as well as an improvement of 39.9% in 3sigma position error signal during track following when compared to conventional notch-based servos

Design, Fabrication and Control of a Micro X-Y Stage with Large Ultra-Thin Film Recoding Media Platform

We report a design of the micro X-Y stage with 6 mm × × × × 6 mm recording media platform, which is actuated by comb-drives. The fabrication process including the integration of the 40 nm thickness PMMA (polymethyl methacrylate) recording media is presented. The prototype of the micro X-Y stage is fabricated by micromachining techniques. The FEA (finite element analysis) results show that the first two modes of the X-Y stage are in plane modes at 440 Hz. The displacement of the media platform can achieve 20 µm with the driving voltage of 55 V. A control scheme is designed and simulated. It shows that the closed-loop system has strong error and vibration rejection capabilities.

Classification of energy consumption patterns for energy audit and machine scheduling in industrial manufacturing systems

To reduce energy consumption for sustainable and energy-efficient manufacturing, continuous energy monitoring and process tracking of industrial machines are essential. In this paper, we introduce a novel approach to reduce the number of required sensors in process tracking by identifying the operational states based on real-time energy data. Finite-state machines are used to model the engineering processes, and a two-stage framework for online classification of real-time energy measurement data in terms of machine operational states is proposed for energy audit and machine scheduling. The first stage uses advanced signal processing techniques to reduce noise while preserving important features, and the second stage uses intelligent pattern recognition algorithms to cluster energy consumption patterns. Our proposed two-stage framework is evaluated on an industrial injection moulding system using a Savizky–Golay filter and a neural network, and our experimental results show a 95.85% accuracy in identification of machine operational states.

Data-driven approaches in health condition monitoring — A comparative study

In this paper, four data-driven classification approaches, that is, K-nearest neighbors (K-NN), self-organizing map (SOM), multi-layer perceptron (MLP), and Bayesian Network classifier (BNC), are applied to a health condition monitoring problem for the wearing cutter. The dataset is produced from a cutting machine using force sensing. A genetic algorithm (GA) based search is performed to select 3 dominant features from a 16-dimensional feature space, which is computed directly from the real dataset. Subsequently K-NN, SOM, MLP, and BNC algorithms are trained to predict the wearing status of the cutter, respectively. The suitability of the four data-driven approaches for the health condition monitoring are investigated and compared.

Nonrepeatable Run-out Rejection Using Online Iterative Control for High-Density Data Storage

The spectra of disturbances and noises affecting precise servo positioning for ultrahigh-density storage in future hard disk drives are time-varying and remain unknown. In this paper, we propose an online iterative control algorithm that sets the measured position error signal (PES) into the servo system to achieve high track densities by minimizing the square of the H2-norm of the transfer function from nonrepeatable run-out (NRRO) disturbances to the true PES. It is not necessary to solve any algebraic Riccati equations and linear matrix inequalities. The algorithm constructs an online repeatable run-out estimator to extract NRRO components for gradient estimates, thereby preventing the controller parameters from being trapped in a local minima. Experimental results on a PC-based servo system for a spinstand show an improvement of 22% in 3sigma NRRO and suppression of baseline NRRO spectrum

Design and Modeling of a Six-Degree-of-Freedom Magnetically Levitated Positioner Using Square Coils and 1-D Halbach Arrays

This paper presents a novel design of six-degree-of-freedom (6-DOF) magnetically levitated (maglev) positioner, where its translator and stator are implemented by four groups of 1-D Halbach permanent-magnet (PM) arrays and a set of square coils, respectively. By controlling the eight-phase square coil array underneath the Halbach PM arrays, the translator can achieve 6-DOF motion. The merits of the proposed design are mainly threefold. First, this design is potential to deliver unlimited-stroke planar motion with high power efficiency if additional coil switching system is equipped. Second, multiple translators are allowed to operate simultaneously above the same square coil stator. Third, the proposed maglev system is less complex in regard to the commutation law and the phase number of coils. Furthermore, in this paper, an analytical modeling approach is established to accurately predict the Lorentz force generated by the square coil with the 1-D Halbach PM array by considering the corner region, and the proposed modeling approach can be extended easily to apply on other coil designs such as the circular coil, etc. The proposed force model is evaluated experimentally, and the results show that the approach is accurate in both single- and multiple-coil cases. Finally, a prototype of the proposed maglev positioner is fabricated to demonstrate its 6-DOF motion ability. Experimental results show that the root-mean-square error of the implemented maglev prototype is around 50 nm in planar motion, and its velocity can achieve up to 100 mm/s.

PDF and Breakdown Time Prediction for Unobservable Wear Using Enhanced Particle Filters in Precognitive Maintenance

Machine health prognosis is crucial to reduce unexpected downtime, maintenance costs, and safety hazards in industrial systems. In this paper, a novel methodology to predict probability density function (pdf) and breakdown time of unobservable degradation processes is proposed. A transition-based autoregressive moving average model and an enhanced particle filter (EPF) are developed at the prognosis stage for the pdf prediction of industrial wear. The strictly monotonic increasing behavior of degradation is ensured by executing a monotonic resampling scheme in EPF, and the number of particles is chosen to be time-varying to reduce computation costs. The effectiveness of our proposed framework is tested on the tool wear in an industrial milling machine, and achieves the predicted bounds with accuracies of at least 90.3% as well as saves more than 50% calculation time without loss of accuracy.