| Description |
1 online resource (205 p.) |
| Series |
Transportation Series |
|
Transportation Series
|
| Contents |
Intro -- Title -- Copyright -- Contents -- List of figures -- List of tables -- Nomenclature -- About the authors -- Abbreviations -- 1 Introduction -- 1.1 Structure of electric powertrain in EVs -- 1.2 Review of faults in electric powertrain in EVs -- 1.3 Review of monitoring techniques for electric powertrain -- 1.4 Errors in the electric powertrain -- 1.4.1 The three-phase power inverter -- 1.4.2 Analysis of dead-time effects -- 1.4.3 Electric machine (motor) -- 1.5 Analysis of EV powertrain errors -- 1.5.1 Contribution of errors from the PMSM |
|
1.5.2 Contribution of errors from the power inverter -- 1.5.3 Contribution of errors from sensors -- 1.5.4 Contribution of errors from ADC/microcontroller -- 2 Feature extraction engineering (FEE) for EV electric powertrain fault diagnosis -- 2.1 Fast Fourier transform -- 2.1.1 The calculation amount of DFT -- 2.1.2 Operation characteristics of DFT -- 2.1.3 FFT decimated by time -- 2.1.4 FFT by frequency extraction -- 2.2 Hilbert transform -- 2.3 Short-time Fourier transform -- 2.4 Discrete wavelet transform -- 2.4.1 Wavelet function scale discretization methods |
|
2.5 Empirical mode decomposition -- 2.5.1 The sampling rate, interpolation methods, and edge effects -- 2.5.2 Stopping criteria for screening -- 2.5.3 Local EMD -- 2.5.4 Online EMD -- 2.5.5 Performance basis -- 2.5.6 Components and sampling rates -- 2.6 Principal-component analysis -- 2.6.1 Advantages of PCA -- 2.7 Singular-value decomposition -- 3 AI-based electric motor fault diagnosis for electric powertrain in EVs -- 3.1 Support vector machine (SVM)-based approach -- 3.1.1 Generalized linear discriminant function -- 3.1.2 The optimal classification surface |
|
3.1.3 Selection of kernel functions -- 3.2 MLP/KNN/RF-based approach -- 3.2.1 Model of a single-layered perceptron -- 3.2.2 Limitations of the single-layered perceptron -- 3.2.3 Nearest neighbor (k-nearest neighbor (KNN)) classification algorithm -- 3.2.3.1 Advantages and disadvantages of KNN -- 3.2.4 Some improvement strategies for KNN -- 3.2.5 Random forest -- 3.2.6 Awareness of the CART algorithm -- 3.2.7 Random forest generation -- 3.2.8 Random forest generation rules -- 3.2.9 Out-of-bag error rate (OOB error) -- 3.2.10 Overfitting |
|
3.2.11 Why is a random forest effective in avoiding overfitting? -- 3.3 Convolutional neural network -- 3.3.1 Sparse connectivity -- 3.3.2 Pooling -- 3.3.3 Fully connected layers -- 3.4 Recurrent neural network -- 3.5 Generative adversarial network -- 3.6 Deep belief network -- 3.7 Autoencoder -- 3.7.1 Probabilistic interpretation of reconstruction errors -- 3.7.2 Sparse autoencoder -- 3.8 Other AI-based approaches -- 3.8.1 Adaptive neuro-fuzzy system (ANFIS)-based approach -- 3.8.2 Fuzzy control combined with neural networks |
| Summary |
This book comprehensively covers the recently-developed AI techniques for solving condition monitoring and fault detection issues in EV electrical conversion systems. Chapters systematically address condition monitoring and fault detection in EV motors and inverters, with illustrative case studies |
| Notes |
Description based upon print version of record |
|
3.8.3 Fuzzy neural system-neural network fuzzification, which is still essentially an ANN |
| Form |
Electronic book
|
| Author |
Zhang, Xiaotian
|
|
Lang, Wangjie
|
| ISBN |
9781839537639 |
|
1839537639 |
|
