Growing demand for consumer electronic devices and telecommunications is expected to drive the quantum cascade laser (QCL) market. The increase in the production rate of QCLs increases the likelihood of production failures and anomalies. The detection of waveguide defects and dirt using automatic optical inspection (AOI) and deep learning (DL) is the main focus of this study. The images samples of QCLs were collected from a laser manufacturing plant in Europe. Due to the lack of sufficient dirt and defect samples, automatic and manual data augmentation approaches were used to increase the number of images. A combination of an improved capsule neural network (WaferCaps) and convolutional neural network (CNN) based on parallel decision fusion is used to classify the samples. The output of these classifiers were combined based on rule-based selection algorithm that chooses the performance of the best classifier according to the class. The proposed approach was compared with the performance of standalone models, different state-of-the-art DL models such as CapsNet, ResNet-50, MobileNet, DenseNet, Xception and Inception-V3 and other machine learning (ML) models such as Support Vector Machine (SVM), decision tree, k-NN and Multi-layer Perceptron (MLP). The proposed approach outperformed them all with a validation accuracy of 98.5%.

A technique for registering and relating events that cause an observable and definable system state is proposed. Discrete events of system-state transfer are expressed by event tracking and clustering in the form of contiguous quanta of data. This approach is capable of describing typical processes in industrial systems in a chain of codes that contain system input/output parameters. The constituent nodes of the Markovian Processes chain form a series akin to genes in the deoxyribonucleic acid, repeatable and predictable. The process genes are the quanta of information that aligns to represent a chain of activities (process). They describe the causal links between occurring events forming a pattern (pathway) that leads to a well-specified output (e.g., a product with a defect or otherwise). The creation of process genomics requires the knowledge of system observed or latent parameters (state) as well as the state change at specified time intervals (discretization). The process genomics theory is tested in an industrial case study for quality assessment and control of glue dispensing in micro-semiconductor manufacturing. The resulting definitions of the system state and interrelationship of control parameters contribute to the development of the process genes. The outcome of the gene alignment is the geometric interpretation of the glue droplet formation. A predicted or observed droplet within the production tolerance leads to a nondefective product. The principle of creating production genomics is to find and rectify the defect-causing genes or to disrupt the sequences that lead to producing defective products, leading to a zero-defect manufacturing process.

An improved method for the real time sensitivity analysis in large scale complex systems is proposed in this paper. The method borrows principles from the event tracking of interrelated causal events and deploys clustering methods to automatically measure the relevance and contribution made by each input event data (ED) on system outputs. The ethos of the proposed event modeling (EM) technique is that the behavior or the state of a system is a function of the knowledge acquired about events occurring in the system and its wider operational environment. As such it builds on the theoretical and the practical foundation for the engineering of knowledge and data in modern and complex systems. The proposed EM platform EventiC filters noncontributory ED sources and has the potential to include information that was initially thought irrelevant or simply not considered at the design stage. The real-time ability to group and rank relevant input-output ED in order of its importance and relevance will not only improve the data quality, but leads to an improved higher level of mathematical formulization in the modern complex systems. The contribution of the approach to systems' modeling is in the automation of data analysis, control, and plant process modeling. EventiC has been validated as the monitoring and the control system for a cement factory. In addition to the previously known parameters, the proposed EventiC identified new influential parameters that were previously unknown. It also filtered 18% of the input data without compromising the data quality or the integrity. The solution has improved the quality of input variable selection and simplify plant control strategies.

A number of data analytics techniques are deployed to measure the influence of various waste water treatment operational parameters against the nitrous oxide ($\text{N}_{\mathbf {2}}$O) emission. N2O is a major threat to the ozone layer and constitutes 80% of total Greenhouse Gas emissions of Waste Water Treatment Plants (WWTPs). The measurement and prediction of N2O emission from WWTP is challenging and costly. Thus, it is important to identify key control parameters that allows for accurately predicting and reducing N2O generation and emission. The current work compares various data driven techniques that identify key parameters and methods of predicting N2O emission. It provides insight to the suitability of each technique for control and optimisation of the target process. The main contribution of this research is introducing two new techniques that applied first time in WWTPs and could cover some current techniques shortcomings in real-time.

Gene expression programming (GEP) is a data driven evolutionary technique that well suits for correlation mining. Parallel GEPs are proposed to speed up the evolution process using a cluster of computers or a computer with multiple CPU cores. However, the generation structure of chromosomes and the size of input data are two issues that tend to be neglected when speeding up GEP in evolution. To fill the research gap, this paper proposes three guiding principles to elaborate the computation nature of GEP in evolution based on an analysis of GEP schema theory. As a result, a novel data engineered GEP is developed which follows closely the generation structure of chromosomes in parallelization and considers the input data size in segmentation. Experimental results on two data sets with complementary features show that the data engineered GEP speeds up the evolution process significantly without loss of accuracy in data correlation mining. Based on the experimental tests, a computation model of the data engineered GEP is further developed to demonstrate its high scalability in dealing with potential big data using a large number of CPU cores.

This paper introduces a platform for online Sensitivity Analysis (SA) that is applicable in large scale real-time data acquisition (DAQ) systems. Here, we use the term real-time in the context of a system that has to respond to externally generated input stimuli within a finite and specified period. Complex industrial systems such as manufacturing, healthcare, transport, and finance require high-quality information on which to base timely responses to events occurring in their volatile environments. The motivation for the proposed EventTracker platform is the assumption that modern industrial systems are able to capture data in real-time and have the necessary technological flexibility to adjust to changing system requirements. The flexibility to adapt can only be assured if data is succinctly interpreted and translated into corrective actions in a timely manner. An important factor that facilitates data interpretation and information modeling is an appreciation of the affect system inputs have on each output at the time of occurrence. Many existing sensitivity analysis methods appear to hamper efficient and timely analysis due to a reliance on historical data, or sluggishness in providing a timely solution that would be of use in real-time applications. This inefficiency is further compounded by computational limitations and the complexity of some existing models. In dealing with real-time event driven systems, the underpinning logic of the proposed method is based on the assumption that in the vast majority of cases changes in input variables will trigger events. Every single or combination of events could subsequently result in a change to the system state. The proposed event tracking sensitivity analysis method describes variables and the system state as a collection of events. The higher the numeric occurrence of an input variable at the trigger level during an event monitoring interval, the greater is its impact on the final analysis of the system state. Experiments were designed to compare the proposed event tracking sensitivity analysis method with a comparable method (that of Entropy). An improvement of 10 percent in computational efficiency without loss in accuracy was observed. The comparison also showed that the time taken to perform the sensitivity analysis was 0.5 percent of that required when using the comparable Entropy-based method.

Ontology alignment facilitates exchange of knowledge among heterogeneous data sources. Many approaches to ontology alignment use multiple similarity measures to map entities between ontologies. However, it remains a key challenge in dealing with uncertain entities for which the employed ontology alignment measures produce conflicting results on similarity of the mapped entities. This paper presents OARS, a rough-set based approach to ontology alignment which achieves a high degree of accuracy in situations where uncertainty arises because of the conflicting results generated by different similarity measures. OARS employs a combinational approach and considers both lexical and structural similarity measures. OARS is extensively evaluated with the benchmark ontologies of the ontology alignment evaluation initiative (OAEI) 2010, and performs best in the aspect of recall in comparison with a number of alignment systems while generating a comparable performance in precision.

This paper proposes the foundation for a flexible data input management system as a vital part of a generic solution for quick-response decision making. Lack of a comprehensive data input layer between data acquisition and processing systems has been realized and thought of. The proposed FDILA is applicable to a wide variety of volatile manufacturing environments. It provides a generic platform that enables systems designers to define any number of data entry points and types regardless of their make and specifications in a standard fashion. This is achieved by providing a variable definition layer immediately on top of the data acquisition layer and before data pre-processing layer. For proof of concept, National Instruments' Labview data acquisition software is used to simulate a typical shop floor data acquisition system. The extracted data can then be fed into a data mining module that builds cost modeling functions involving the plant's Key Performance Factors.