11 3 Fault Diagnosis and Prognosis for Reliability Enhancement3.1 Introduction3.2 Fundamentals3.2.1 The Pattern of Failures with Time for Non-Repairable Items3.2.2 Distribution Functions3.2.3 Confidence in Reliability and Prognosis3.3 Component Reliability3.4 Reliability of Subsystems and Systems3.4.1 Analysis Tools3.5 Lifetime, Reliability Prediction3.6 Fault Management and Mitigation3.7 Design and Manufacturing3.8 Applications and Case Studies3.9 Scheduled Maintenance, Condition-Based Maintenance3.9.1 Reliability and Costs3.10 ConclusionsBibliography
12 Index
List of Illustrations
1 Chapter 1Figure 1.1 General overview of Diagnosis and Prognosis methodsFigure 1.2 Application fields for sensorsFigure 1.3 AccelerometerFigure 1.4 Axial flux distribution in an induction motor and possible distributionFigure 1.5 Lobar polar patternFigure 1.6 Direct current generator @LeroySomerFigure 1.7 Pulse angular speed tachometerFigure 1.8 Incremental encoderFigure 1.9 Numeric Grey absolute encoderFigure 1.10 Synchro-resolverFigure 1.11 The trajectory of the line current: the motor operates under ...Figure 1.12 The spectrum of the line current: the motor operates under one ...Figure 1.13 The Hann windowFigure 1.14 The Hamming windowFigure 1.15 The Blackman windowFigure 1.16 Spectrum without windowFigure 1.17 Spectrum using the Hamming windowFigure 1.18 Spectrum using the Hann windowFigure 1.19 Spectrum using the Blackman windowFigure 1.20 Spectra on the range 40–60HzFigure 1.21 Rules of transformation between the time–frequency and...Figure 1.22 Stationary signal example (Bonnardot [2004])Figure 1.23 Estimation of the parametersFigure 1.24 Estimation of the parametersFigure 1.25 ObserverFigure 1.26 Extended rotor resistance observerFigure 1.27 Unscented Kalman Filter Algorithm ...Figure 1.28 Iris data base (a) initial values (b) after PCA.Figure 1.29 Pearson coefficient Source: Tufféry [2012] and Ben Marzoug [2020]Figure 1.30 Difference between Pearson and Spearman coefficients Source...Figure 1.31 Sequential backward feature selectionFigure 1.32 Sequential forward feature selectionFigure 1.33 Inertia explained by axesFigure 1.34 Confusion matrix after a discriminant analysis of Fisher’s iris dataFigure 1.35 Illustration of Huygens relationFigure 1.36 General process of classificationFigure 1.37 Examples of supervised clusteringFigure 1.38 Examples of unsupervised clusteringFigure 1.39 Isovalues of Euclidean (circles) and Mahalanobis (ellipse) distances...Figure 1.40 kNN rules applied to a new observationFigure 1.41 Class overlapFigure 1.42 Binary SVMFigure 1.43 A basic neural network with one hidden layerFigure 1.44 Optimization process of a basic neural networkFigure 1.45 A basic recurrent neural networkFigure 1.46 A basic neural network for faults detectionFigure 1.47 Example of a dendrogramFigure 1.48 Chain effectFigure 1.49 K-means algorithm for the partitioning of two classesFigure 1.50 Architecture of a self-organizing map (1D network)Figure 1.51 Evolution of the learning rate over timeFigure 1.52 Prognosis methods (Soualhi et al. [2020]) with the permission of ...Figure 1.53 Prognosis processFigure 1.54 Hybrid Prognostics approach. (Pecht and Jaai [2010])...Figure 1.55 Example of time series with monthly counts of airline passengers...Figure 1.56 Example of the trend and seasonability of a time series (MATLAB [2020])Figure 1.57 Example of a time series moving average filtering (MATLAB [2020])Figure 1.58 Example of ARIMA model (MATLAB [2020])Figure 1.59 Naive Bayesian networkFigure 1.60 Augmented naive Bayesian networkFigure 1.61 Different cases for the rainflow algorithmFigure 1.62 Example of the rainflow algorithm (Lee and Tjhung [2012])
2 Chapter 2Figure 2.1 Motor Drive overview.Figure 2.2 Diagnosis Test bench for synchronous machine in...Figure 2.3 Chevrolet Spark Electric Vehicle Electric Induction Motor, GMS.Figure 2.4 Chevrolet Spark Electric Vehicle Electric Permanent Magnet AC...Figure 2.5 Hysteresis curve with minor loops.Figure 2.6 Demagnetization characteristics of rare earth magnets.Figure 2.7 Demagnetization characteristics of NdFeB (Neodymium Iron Boron).Figure 2.8 Types of distributed windings. (a) Single layer distributed stator...Figure 2.9 Types of concentrated windings. (a) Double layer distributed stator...Figure 2.10 Stator slot designs.Figure 2.11 Interior and Surface Permanent Magnet Machines...Figure 2.12 Induction machine rotor.Figure 2.13 Schematic of Field Oriented torque control of an induction machineFigure 2.14 Schematic of flux and voltage sectors for Direct Torque Control.Figure 2.15 Schematic of Direct Torque Control Concept for Induction Machines.Figure 2.16 Typical cross sections of radial flux PMAC machines...Figure 2.17 Failure distribution of Low and Medium Power Motors...Figure 2.18 Failure distribution of High power motors for petrochemistryFigure 2.19 Stray flux and measurement positions Ceban et al. (2012).Figure 2.20 Classification based on TFR by A. Lebaroud, Lebaroud and Clerc (2008).Figure 2.21 Sensor fault indicators, Zhang et al. (2020).Figure 2.22 Integrated Prognosis Algorithm for Electric Drive Systems...Figure 2.23 Bearing geometry.Figure 2.24 Effect of crates in the transient signal,...Figure 2.25 SNRs of the denoised signals from the SOS-based and SOSO based...Figure 2.26 Feature trends of selected features Kim et al. (2012).Figure 2.27 Probability distribution of each health state Kim et al. (2012).Figure 2.28 Comparison of actual RUL and estimated RUL Kim et al. (2012).Figure 2.29 Neural network and HMM prognosis from Soualhi et al..Figure 2.30 Curve fitting to variance and entropy features...Figure 2.31 RUL estimation with different EKF tracking start times Singleton...Figure 2.32 RUL estimation with the variance feature...Figure 2.33 Observations, micro-states and macro states Ma et al. (2020).Figure 2.34 The framework of the proposed GMM-SR-HSMM approach. Ma et al. (2020).Figure 2.35 The calculation of remaining life for each state. Ma et al. (2020).Figure 2.36 RUL prediction using based on the SR feature Ma et al. (2020).Figure 2.37 Overview of materials in a low-voltage insulation system...Figure 2.38 RUL estimate using peak-to-peak value of the switching current...Figure 2.39 Experiment equipment setup, Tsyokhla et al. (2019)...Figure 2.40 Insulation frequency-dependent characteristics...Figure 2.41 Equivalent capacitance and RUL estimation...Figure 2.42 Partial discharges in voids.Figure 2.43 Separation of PD from commutation disturbance in motors using...Figure 2.44 Open circuit faults: a. Parallel coils in a phase...Figure 2.45 First harmonic magnitude analysis after data regularization...Figure 2.46 Analysis of time harmonic magnitudes in space.Figure 2.47 Framework for diagnosis and prognosis of electrical machines...Figure 2.48 Triggering of fault parameter estimation when fault occurs or changes...Figure 2.49 DC Simulation of the field in healthy...Figure 2.50 Simulated current in the stator frame before and after...Figure 2.51 Numerical analysis of stresses and displacements after a broken bar...Figure 2.52 Three-phase rotor resistance with and without broken bars...Figure 2.53 Comparison between simulation and experimental results for...Figure 2.54 Location of Hall sensor installed for FE and experimental ...Figure 2.55 Experimental results for three IPMSM faults, Park et al...Figure 2.56 RFOC strategy with SVM applied to the PMSM-side converterFigure 2.57 Block diagram of the DTC strategy applied to the PMSM..Figure 2.58 Ray ratios and IRP ratios for online ITSCs detection under different...Figure 2.59 Current and voltage ratios and Ray ratios (a) and online...Figure 2.60 Five typical fault-isolation schemes, Zhang et al. (2014).Figure 2.61 Three-phase open circuit