Uncertainty and Artificial Intelligence E-Book
114,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Wissenschaft und neue Technologien
- Sprache: Englisch
Today's information technology, along with Artificial Intelligence (AI), is moving towards total communication between all computerized systems. AI is a representation of human intelligence based on the creation and application of algorithms in specific computer environments. Its aim is to enable computers to act like human beings. For it to work, this type of technology requires computer systems, data with management systems and advanced algorithms, used by AI.
In mechanical engineering, AI can offer many possibilities: in mechanical construction, predictive maintenance, plant monitoring, robotics, additive manufacturing, materials, vibration control and agro composites, among many others.
This book is dedicated to Artificial Intelligence uncertainties in mechanical problems. Each chapter clearly sets out used and developed illustrative examples. Aimed at students, Uncertainty and Artificial Intelligence is also a valuable resource for practicing engineers and research lecturers.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 209
Veröffentlichungsjahr: 2023
Ähnliche
Table of Contents
Cover
Table of Contents
Title Page
Copyright Page
Preface
1 New Intelligence Method for Machine Tool Energy Consumption Estimation
1.1. Introduction
1.2. Mathematical model for estimating power consumption by the spindle and table of a mill
1.3. ICA method
1.4. Results and discussion
1.5. Conclusion
1.6. References
2 Uncertainty and Artificial Intelligence: Applications to Maintenance in Additive Manufacturing
2.1. Introduction
2.2. Integration of uncertainty
2.3. Uncertainty in artificial intelligence
2.4. Artificial intelligence in additive manufacturing
2.5. Predictive maintenance in additive manufacturing
2.6. Proposed strategy and applications
2.7. Conclusion
2.8. References
3 Bio-Composite Structural Durability: Using Artificial Intelligence for Cluster Classification
3.1. Introduction
3.2. The state of acoustic emission technology
3.3. The concept behind classification
3.4. Application to the classification of damage mechanisms in agro-composites
3.5. Artificial neural networks for classifying damage
3.6. Conclusion
3.7. References
4 Intelligent Control for Attenuating Vertical Vibrations in Vehicles
4.1. Introduction
4.2. Limits of passive and semi-active control strategies
4.3. Model-free control (MFC)
4.4. Application to the control of vertical vehicular vibrations
4.5. Conclusion
4.6. References
5 Optimization of the Power Inductor of a DC/DC Converter
5.1. Introduction
5.2. Description of the power inductor
5.3. Thermomechanical modeling of the power inductor
5.4. Optimization methods
5.5. Optimization of the power inductor
5.6. Conclusion
5.7. References
6 Study of the Influence of Noise and Speed on the Robustness of Independent Component Analysis in the Presence of Uncertainty
6.1. Introduction
6.2. The model studied
6.3. Construction of the road surface profile
6.4. The principle of the ICA method
6.5. Monte Carlo technique
6.6. Results and discussion
6.7. Conclusion
6.8. References
7 Multi-Objective Optimization Applied to a High Electron Mobility Transistor
7.1. Introduction
7.2. Description of HEMT technology
7.3. Multi-physical modeling of the HEMT
7.4. The multi-objective optimization approach
7.5. Multi-objective optimization applied to HEMT technology
7.6. Conclusion
7.7. References
List of Authors
Index
Other titles from ISTE in Mechanical Engineering and Solid Mechanics
End User License Agreement
List of Tables
Chapter 1
Table 1.1. Relative errors
Chapter 3
Table 3.1. Parameters defining an AE burst
Table 3.2. Maximal stress and strain at rupture of agro-composites
Table 3.3. Characteristics for classification descriptors
Table 3.4. Amplitude and frequency: classification parameters for damage
Table 3.5. Percentage of signals for each class in agro-composite tests
Chapter 5
Table 5.1. Thermal properties of materials
Table 5.2. Numerical simulation and experimental results
Table 5.3. The physical properties of materials
Table 5.4. Results of the optimization by the two methods
Chapter 6
Table 6.1. Parameters for a quarter vehicle
Table 6.2. Classification of road surfaces
Table 6.3. Relative error between the real road surface and the estimated surf...
Table 6.4. Relative error between the real road surface and the estimated surf...
Table 6.5. Relative error between the real road surface and the estimated surf...
Table 6.6. Relative error between the real road surface and the estimated surf...
Table 6.7. Relative error between the real road surface and the estimated surf...
Chapter 7
Table 7.1. Thermal properties of materials
Table 7.2. Physical properties of materials
Table 7.3. Results of the HEMT reliability analysis
Table 7.4. Optimal results for the transistor
List of Illustrations
Chapter 1
Figure 1.1. Cutting system of a machine tool: a mill
Figure 1.2. Distribution of the cutting effort for a circular machining operat...
Figure 1.3. The principal stages of the separation program
Figure 1.4. Finite element model of the spindle.
Figure 1.5. Implementation of the ICA method
Figure 1.6. Comparison of real and estimated tangential efforts (a) and real a...
Figure 1.7. Calculation of the MAC amplitude for tangential Ft(t) and advancem...
Chapter 2
Figure 2.1. Implementation of the artificial intelligence process.
Figure 2.2. Diagnostic and prognostic stages.
Figure 2.3. Successful trial of a candy box and unsuccessful trial (the produc...
Figure 2.4. Successful trial of a column base and unsuccessful trial (the prod...
Figure 2.5. Problem with the raft (the product can be found on 3d-printing-4u....
Figure 2.6. Failure scenario at two levels (the product can be found on 3d-pri...
Figure 2.7. Proposed flow chart for an advanced quality control system.
Figure 2.8. Maintenance block.
Chapter 3
Figure 3.1. Principle of acoustic emission
Figure 3.2. Mechanisms responsible for damage in composites
Figure 3.3. Propagation of a stationary mechanical wave through materials
Figure 3.4. Parameters characterizing an AE burst
Figure 3.5. Diagram of the principle of the k-NN method with a classification ...
Figure 3.6. Diagram of the K-mean method’s principle for a two-class classific...
Figure 3.7. Choosing the optimal number of classes
Figure 3.8. Forming procedure of bio-based composite sheets reinforced by hemp...
Figure 3.9. Tensile tests: stress–strain evolution and damage mechanisms in ag...
Figure 3.10. Stress–strain curve following acoustic activity in a polypropylen...
Figure 3.11. Amplitude distribution of the polypropylene matrix under tensile ...
Figure 3.12. Stress–strain of tensile tests on unidirectional fiber at 0°...
Figure 3.13. Identification of damage as a function of cumulative energy
Figure 3.14. Stress–strain curve for tensile tests at [+/-45°]...
Figure 3.15. Amplitude distribution of characteristic signals for acoustic emi...
Figure 3.16. Constraint/deformation curve following acoustic activity of compo...
Figure 3.17. Identification of damage to composites [+/-67.5°] as a function o...
Figure 3.18. Stress–strain curve for unidirectional composites tested perpendi...
Figure 3.19. Amplitude distribution of characteristic acoustic emission signal...
Figure 3.20. Evolution in the number of AE signal bursts by damage class for c...
Figure 3.21. Basic structure of a network of artificial neurons
Figure 3.22. Supervised and unsupervised learning
Figure 3.23. General architecture of a NN
Figure 3.24. Principle of mini-batch gradient descent
Figure 3.25. Time–frequency–amplitude map of AE signals in agro-composites...
Figure 3.26. Frequency–amplitude before damage classification for agro-composi...
Figure 3.27. Projection of damage classes for agro-composites
Chapter 4
Figure 4.1. Model-free control block diagram.
Figure 4.2. Integral suspension system with seven degrees of freedom.
Figure 4.3. The road surface used
Figure 4.4. Vertical acceleration of the center of gravity.
Figure 4.5. Accelerations of suspended masses without active control and with ...
Figure 4.6. Normalized deformations of tires.
Figure 4.7. Oscillatory acceleration without an active controller and with MFC...
Figure 4.8. Rotational acceleration without an active controller and with MFC.
Figure 4.9. Forces generated by the controllers.
Chapter 5
Figure 5.1. Power inductor structure.
Figure 5.2. Distribution of the temperature in the structure of the power indu...
Figure 5.3. Evolution of the temperature as a function of time.
Figure 5.4. Distribution of the displacement in the structure.
Figure 5.5. Temperature distribution in the structure of the power inductor af...
Figure 5.6. Distribution of von Mises stresses in the structure after optimiza...
Figure 5.7. Distribution of von Mises stresses in the structure after optimiza...
Chapter 6
Figure 6.1. ICA and Monte Carlo coupling
Figure 6.2. Quarter vehicle
Figure 6.3. Simplified concept of ICA (Haddar et al. 2019).
Figure 6.4. The principal stages in the separation process
Figure 6.5. Steps for whitening the vector X(t)
Figure 6.6. Steps for separating sources
Figure 6.7. Comparison between the real surface D and the estimated surface.
Chapter 7
Figure 7.1. HEMT structure.
Figure 7.2. Evolution in temperature relative to dissipated power.
Figure 7.3. Distribution of von Mises constraints in the HEMT structure.
Figure 7.4. Evolution of von Mises constraints relative to dissipated power.
Figure 7.5. Pareto solutions to the HEMT problem.
Guide
Cover Page
Table of Contents
Title Page
Copyright Page
Preface
Begin Reading
List of Authors
Index
Other titles from ISTE in Mechanical Engineering and Solid Mechanics
WILEY END USER LICENSE AGREEMENT
Pages
iii
iv
ix
x
xi
xii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
179
180
181
182
183
184
185
186
187
188
203
Artificial Intelligence in Mechanics Set
coordinated byAbdelkhalak El Hami
Volume 1
Uncertainty and Artificial Intelligence
Additive Manufacturing, Vibratory Control, Agro-composite, Mechatronics
Edited by
Abdelkhalak El Hami
First published 2023 in Great Britain and the United States by ISTE Ltd and John Wiley & Sons, Inc.
Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1988, this publication may only be reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms and licenses issued by the CLA. Enquiries concerning reproduction outside these terms should be sent to the publishers at the undermentioned address:
ISTE Ltd27-37 St George’s RoadLondon SW19 4EUUK
www.iste.co.uk
John Wiley & Sons, Inc.111 River StreetHoboken, NJ 07030USA
www.wiley.com
© ISTE Ltd 2023The rights of Abdelkhalak El Hami to be identified as the author of this work have been asserted by him in accordance with the Copyright, Designs and Patents Act 1988.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s), contributor(s) or editor(s) and do not necessarily reflect the views of ISTE Group.
Library of Congress Control Number: 2023944019
British Library Cataloguing-in-Publication DataA CIP record for this book is available from the British LibraryISBN 978-1-78630-942-6
Preface
Today, computing, along with artificial intelligence (AI), is moving towards complete communication between all computerized systems. AI is a representation of human intelligence that depends on the creation and application of algorithms in a precise computerized environment. Its objective is to allow computers to act like human beings. This type of technology needs computer systems, data with management systems and advanced algorithms capable of being used by AI.
In mechanics, AI offers several possibilities for mechanical construction, predictive maintenance, monitoring of installations, and in the fields of robotics, additive manufacturing, vibratory control and agro-composites.
AI needs a certain amount of data and a high processing capacity. It plays a role in a great number of fields, more specifically in enormous companies capable of using such technology. Its principal purpose is to improve the working conditions of employees in order to diminish risk.
The objective of this book is to explore the uncertainty of AI in mechanical problems. It is made up of seven chapters.
Chapter 1 applies one of the methods of AI called independent component analysis (ICA) to evaluate the power used during a circular operation so that we can estimate the energy used by a milling machine. This method is a technique of blind division of sources and is based on inverse problems. Its robustness has been demonstrated in several fields, for example, its use in approximating road surfaces using the knowledge of vehicular vibration or in the identification of defects related to gear systems. The only thing required is knowledge concerning the dynamic response, for example, accelerations in response estimates. This chapter presents two digital models meant for estimating the power used by the spindle and the table of a mill during a circular milling operation. The method relies only on receivers that capture the vibratory responses such as movement or perhaps accelerations, and then estimates the additional effort made by the machine tool.
Chapter 2 uses several fields to define the best strategies for resolving a specific problem in one of the many applications of AI. The application is related to the use of AI during maintenance in additive manufacturing. However, the specific problem is associated with the existence of uncertainty in the performance of AI for this type of application. Several components are examined: uncertainty, AI, maintenance and additive manufacturing. The concept of uncertainty is first addressed separately in order to provide the reader with a clear explanation of this component. The study is made up of two threads: the first thread represents a proposed strategy, while the second thread is related to a specific application.
The strategy can be summarized as taking uncertainty into account when AI makes decisions. The degree of decisions made by AI can affect the result of the application either directly or indirectly. There are certain intervals for making the right decisions, which can be taught to the computer that is being used to avoid making the wrong decisions.
The principal objective of this chapter is to address questions of uncertainty in order to contribute to the industrialization of technology in additive manufacturing. The industrialization of additive manufacturing must implement several studies to prepare for different kinds of errors and the concept of uncertainty. A high rate of failure raises the total cost, which can be a major obstacle to the industrialization of technology in additive manufacturing.
Chapter 3 looks into the question of the durability of bio-sourced materials for industrial applications. The requirements and expectations of clients are becoming greater and greater with respect to the use of agro-composite components, which undergo mechanical uses and variations in temperature and humidity, repeated throughout their use, leading to a degradation of the mechanical properties and an accelerated aging of the vegetable fibers. The development of agro-composites for industrial applications requires precise information concerning the processes causing the damage in order to better predict their lifespan. Indeed, when there is a process that causes damage to these materials, a transitory wave, resulting from the release of the energy that was stored in it, propagates from the source of the damage towards the surface of the material. This wave can be recorded by receptors attached to the material. In such a context, this chapter looks for ways to make use of artificially intelligent methods to analyze the acoustic emission data and predict the appearance of critical mechanisms in order to identify scenarios where damage is caused. The purpose of our work is to implement methods that essentially allow us to analyze AI data. It depends on the use of recent classification methods (either unsupervised as in k-averages or supervised as with neural networks). The purpose of this multi-parameter statistical analysis is to identify the meaning of the data obtained during the monitoring of acoustic emissions for damage to the polymer matrix composite materials being mechanically worn along one axis.
Chapter 4 presents an intelligence method based on model-free control for the ease of managing an entire active suspension system in a vehicle. In the proposed controller, algebraic estimators are created for approximating the total perturbance in three senses (vertical, oscillation, slope), including the external effect of the road, the nonlinearities of the system and the modeling error, if it exists. A direct method does not require an overall mathematical model with an easy implementation. The intelligence of the controller boils down to its capacity to react quickly and online in fractions of a second to reject all unnecessary types of perturbance without needing to know their model or their frequential characteristics.
Chapter 5 aims to propose a methodology for studying the reliability of mechatronic systems through the reliability of high electron mobility transistors (HEMT) by using AI in the thermal model and the reliability and statistical model. Thermal modeling using the finite element method is presented to observe the thermal behavior of the transistor, and the influence of the working environment, such as the dissipated power and reference temperature on the working temperature. It then develops a thermo-reliability coupling by integrating two models: thermal and statistical, the thermal model being developed with Comsol Multiphysics and the statistical model (reliability) with Matlab. This coupling allows us to estimate the thermal reliability of the HEMT and identify the influence of the working conditions on its reliability and performance.
Chapter 6 is devoted to studying how robust the intelligence method of independent component analysis (ICA) is when estimating a road surface for a quarter vehicle. To do this, the Monte Carlo stochastic technique is used in the presence of the inevitable variables of uncertainty: the mass of the vehicle, the stiffness of the spring and shock absorption. The effect of noise generated by wind is also considered. The convergence of the ICA intelligence method was evaluated when comparing real surfaces as defined by the ISO norm to the surfaces approximated with uncertainties. The results obtained prove the robustness of the ICA in the estimation of different road surfaces.
Chapter 7 focuses on the study of the high electron mobility transistor (HEMT), which is one of the most important components in high-power mechatronic systems. HEMT is the most commonly used technology in complex systems in general and in mechatronic systems in particular. Consequently, the optimization of this technology is a major stake for engineers and researchers in this field. In this chapter, we present a method for multi-objective optimizing applied to HEMT in order to improve its thermal and mechanical performance. The optimization process is based on the coupling of two methods: the finite element model using Comsol Multiphysics software and the coded optimization model on Matlab software. The second model is used to solve the problem of optimization by coupling it with the first model. The optimal values of the conception variable obtained from the application of the AI methods and from the optimization process allow us to optimize the thermomechanical behavior of the HEMT structure, making it more reliable.
Abdelkhalak EL HAMIRouenJuly 2023
1New Intelligence Method for Machine Tool Energy Consumption Estimation
1.1. Introduction
Modern life has led us to consume high levels of electric energy, which is required for our needs. Research has demonstrated that the role of industry in worldwide energy consumption will rise from 3,000 Mtoe in 2010 to 5,000 Mtoe in 2050. This increase is associated with high emissions from the greenhouse effect, which has a negative impact on the environment. This is why eco-production has become increasingly necessary: to limit the high consumption of energy by industry. To do this, researchers have used tools for estimating energy consumed by machine tools. This chapter applies an artificial intelligence method, called independent component analysis (ICA), to evaluate the power consumed during a peripheral milling operation.
Our life is dominated by new technologies that have led to a heightened consumption of electrical energy. For example, in the United States, 31% of all energy is consumed by industry. 90% of this consumption goes to manufacturing, of which 70% is used by machine tools (Zhou et al. 2016). From there, atmospheric CO2 emissions are caused by the energy consumption of the manufacturing industry. According to Herzog (2009), 99% of environmental problems are caused by the consumption of energy in manufacturing processes. The harmful effects are enormous, and the research by Herzog (2009) has indeed shown that manufacturing operations are responsible for 19% of these emissions. This is why the optimization of energy consumption by machine tools is urgent for protecting the environment. In order to minimize this consumption, various studies have relied on modeling the energy consumed by machine tools, especially during cutting operations. Let us cite the model created by Rajemi et al. (2010) as an example, which allows for the quantification of the required energy by turning operations. This model accounts for the lifetime of a tool. We also note the model proposed by Avram and Xirouchakis (2011) for quantifying the power consumed by the spindle and the table of a mill while cutting. They calculate the energy of the spindle by multiplying the force of the cut by the speed of the cut. The energy of the table, meanwhile, is calculated using the product of the cutting torque and the angular speed. The Kara and Li (2011) model was based on the material removal rate (MRR). Another model was established by Calvanese et al. (2013), which presented the energy of an axis feed. It is equal to the product of the effort made in cutting and the cutting speed. To obtain this effort, the average value of the width of the shaving must be determined. Taking this model as a basis, an improvement was presented by Alberteli et al. (2016). They make use of the constant angular position of the cutting tool. Other recent work has been published in the literature by Ben Jdidia et al. (2019b). This work aims to estimate the dynamic cutting energy consumed by the spindle and the table of a mill for resurfacing operations. To estimate the effort involved in cutting, the authors apply the method of finite elements, which is a complex method.
Therefore, this chapter aims to estimate the energy consumed by a mill by applying a new method, namely ICA. This method is a major technique for blindly separating sources. It is based on inverse problems and is a simple method whose robustness has been demonstrated in numerous fields. For example, Ben Hassen et al. (2019) and Chaabane et al. (2019) use this method in estimating road surfaces, using knowledge related to vehicular vibration. Taktak et al. (2012) demonstrate the efficiency of this method in identifying the defects associated with a gear system. It only requires the knowledge of the dynamic response, such as accelerations, for estimating the responses. This is the inverse method.
1.2. Mathematical model for estimating power consumption by the spindle and table of a mill
Figure 1.1 describes the machine tool studied (a mill) during a peripheral cutting operation. Two principal efforts are studied, the tangential component of the cutting effort Ft(t) applied to the spindle and the effort made in advancing Fx(t), applied to the table. These two efforts are considered dynamic. Indeed, one tooth, denoted i, that removes material has a momentary angular position called Φi(t). A shaving whose width varies over time is generated. This width is formed by two components: one static, caused by the rigid body, and one dynamic, caused by the motion of the tool at time t and t-τ:
where Ω is the speed of the spindle rotation in–rpm, ϕp is the angle between two successive teeth and ψ is the cutting angle.
Figure 1.1.Cutting system of a machine tool: a mill
The power consumed by the spindle and the table during the machining operation is given by the following two equations:
where:
– V
c
represents the cutting speed in m/min;
– V
f
represents the feed speed in mm/min.
Figure 1.2 shows the distribution of cutting efforts during a circular machining operation (Romdhane 2017).
Figure 1.2.Distribution of the cutting effort for a circular machining operation
