1 Introduction
Global challenge | Contribute to sustainable manufacturing solutions | |
---|---|---|
Short description of levels and objectives | Abstract level: Develop sustainable, agile, and resilient high-quality production systems, e.g., by closely integrating complementary strengths of humans users and cyber-physical-production systems in human-cyber-physical-production systems (HCPPS) | Concrete level: Research, develop, and produce high-quality surfaces for key industrial applications produced with energy- and resource-efficient thermal spray coating technologies utilizing innovative, non-toxic materials that are abundantly available |
(1) Classify and bound challenge extent | Large-scale challenge | Small-scale challenge |
(2) Specify setting | To be addressed by broader transdisciplinary expert teams of researchers and practitioners with participation of affected human users (e.g., operators) | Mainly to be addressed by thermal spraying experts (supported by experts from artificial intelligence modeling and human sciences) and industrial partners from relevant application areas |
(3) Specify challenge by generation of alternative options based on (fundamental) values and desired output criteria | Define and specify set of evaluation criteria that reflect fundamental values such as “sustainability” or “agility” in complex industrial systems Develop scenarios for various levels of automation, flexibility and human-cyber-interconnectedness etc. with value trade-offs | Define set of evaluation criteria such as “resource efficiency” or “coating properties” for concrete thermal spray technologies and applications Specify how the evaluation criteria can be met (e.g., by varying control parameters or using different spraying technologies) and weigh options |
(4) Build and evaluate various (system) models | Simulate scenarios (e.g., large-scale models) and anticipate dynamic consequences at different levels and for affected people | Use suitable modeling methods (e.g., from artificial intelligence or process simulations) to predict outcomes and evaluate empirically |
(5) Validate alternatives | Validate and confirm alternatives based on global criteria such as simulation performance, costs, robustness, and stability of procedures and results | Validate and confirm alternatives based on experimental observations as well as expert judgments especially when multiple (and sometime conflicting) responses are specified |
(6) Decide which alternative to take to meet identified challenge | Decision to be based on validation of alternatives, desired fundamental values, and anticipated outcomes that can be achieved while simultaneously restricting undesired consequences |
1.1 Abstract level: human-cyber-physical-production systems for sustainable manufacturing
1.2 Concrete level: thermal spray coatings for sustainable engineering applications
1.3 Paper objectives
2 Human–machine teaming in industrial decision-making
Human performance level | Human consciousness level/algorithm transparency level | Equivalent AI algorithm class |
---|---|---|
Skill-based | Low | Black-box models (e.g., artificial neural networks, support vector machines, random forest models, Bayesian belief networks) |
Rule-based | Medium | Gray-box models (e.g., decision trees, fuzzy cognitive maps, (fuzzy) rule-based models, ACT-R) |
Knowledge-based | High | Gray/glass/white-box models (explainable artificial intelligence methods, e.g., fuzzy pattern classification) |
-
Human-centeredness by integrating human cognition and performance explicitly using transparent AI algorithms. Still, transparency of knowledge-based engineering approaches [33] and enrichment of semantics of knowledge models are known to be research challenges [34]. As the human cognitive processes [18, 29] work together seamlessly within the human mind, combinations of different XAI methods need to be developed in the long run to achieve effective HMT.
-
Efficient AI performance for big and small/smart data problems. Today, AI algorithms succeed mainly due to the availability of large databases and computational power—both fundamental prerequisites for nowadays machine learning methodology. At the same time, further deployment in resource-restricted application areas may be hindered by the need for large data sets and high energy demand for computation (e.g., extensive training necessity for deep learning models) [35]. This holds true for many industrial manufacturing applications as well because experimentation and data acquisition are often highly expensive.
-
A combined expert knowledge- and data-driven AI modeling approach is worth aspiring. HMT will rely heavily on information from the CPPS in order to monitor and control complex technical processes (e.g., precise technical measurements such as process gas flow rate or voltage characteristics in APS). Although CPPS performance is superior to human cognition in certain characteristics (e.g., data mining, processing capacity, and speed), human knowledge and abilities superior to CPPS performance (e.g., rapid contextualization, uncertainty management, and holistic, linguistic evaluations of process characteristics) need to be integrated as well within one consistent modeling approach [14] to establish complementarity.
-
Robust AI algorithms to manage uncertainty and vagueness. This is of importance because industrial engineering issues involve dealing with different kinds of imprecisions. AI methods based on fuzzy sets [30, 36‐39] are especially suitable for this purpose and some are able to combine the above-mentioned requirements like the FPC approach we chose here for illustration (see below).
3 Formalizing expert knowledge for quality control in thermal spraying
3.1 Expert knowledge for hybrid decision-making in atmospheric plasma spraying
3.2 Approaches for multiple criteria decision-making
3.2.1 Formal model building approaches to MCDM problems
3.2.2 MCDM in APS
Structural MCDM element | APS example |
---|---|
General objective | Resource-efficient production of high-quality ceramic surfaces using a ternary spraying powder blend (Al2O3-Cr2O3-TiO2) for application-specific multi-functional properties combining insulation, corrosion resistance, and wear resistance |
Decision alternatives | APS input variable values (e.g., varied plasma gas flow rates at five discrete levels P1–P5) |
Goal criteria (GC) and subgoal criteria (SGC) with preferred performance | GC 1: microstructure: desirable/undesirable SGC 1.1: proportion Al2O3: maximum SGC 1.2: proportion TiO2: maximum GC 2: coating properties: adequate/inadequate SGC 2.1: electric resistivity: maximum SGC 2.2: hardness: maximum GC 3: resource efficiency: acceptable/unacceptable SGC 3.1: thickness/pass: maximum |
Example-weighting of subgoal criteria (first rank stands for most important) | (1) Electrical resistivity, hardness (2) Thickness/pass (3) Proportion Al2O3, proportion TiO2 |
Subgoal criteria (SGC) in goal criteria space | Measurement methods | Measurement unit |
---|---|---|
SGC 1.1: proportion of Al2O3 and SGC 1.2: proportion of TiO2 | X-ray diffraction | % labeled as linguistic term |
SGC 2.1: electric resistivity | Electrical impedance spectroscopy (EIS) | Ω·m |
SGC 2.2: hardness | Indentation method according to Vickers | HV0.3 |
SGC 3.1: thickness/pass | Images of cross sections taken using an optical microscope to measure coating thickness, later normalized with number of passes | µm/pass |
4 Method
4.1 Fuzzy pattern classification
4.2 Fuzzy pattern model building for decision-making in atmospheric plasma spraying
4.2.1 Decision alternative and goal criteria spaces
Decision alternatives: varied total plasma gas flow rates (P) in l/min | Classes of combined microstructural proportions | Microstructure | |||
---|---|---|---|---|---|
Proportion Al2O3 | Proportion TiO2 | ||||
LT | % | LT | % | ||
P1: 46 | 1 (low–low) | Low | 32 | Low | 26 |
P2: 51 | 2 (medium–very low) | Medium | 44 | Very low | 18 |
P3: 61 | 1 (low–low) | Low | 21 | Low | 37 |
P4: 71 | 3 (very low–medium) | Very low | 19 | Medium | 41 |
P5: 81 | 3 (very low–medium) | Very low | 8 | Medium | 56 |
4.2.2 Fuzzy transformation procedure between decision alternative and goal criteria spaces
5 Results and discussion
-
Scenario I: develop a coating that has a high electrical resistivity to serve as dielectric and also combats wear.
-
Scenario II: develop a coating that has a high electrical resistivity to serve as dielectric; degradation or failure on account of wear is, however, not a consideration.
Goal criteria (GC) | Subgoal criteria (SGC) | Weight | Scenario I | Scenario II |
---|---|---|---|---|
GC 1: microstructure | SGC 1.1: proportion Al2O3 | 1 | Max | Max |
SGC 1.2: proportion TiO2 | 1 | Max | Max | |
GC 2: coating properties | SGC 2.1: electric resistivity | 3 | Max | Max |
SGC 2.2: hardness | 3 | Max | – | |
GC 3: resource efficiency | SGC 3.1: thickness/pass | 2 | Max | Max |
-
Scenario I v = 154 m/s and T = 3245 °C