We propose a novel model-based surface approximation method for three-dimensional (3-D) surface quality inspection that combines a machine learning approach with multiresolution paradigms. Acceptable surface deviations are modeled by learning a number of 3-D measurements of tolerance samples. At the same time, areas with high surface details are hierarchically refined, allowing an improved spatial localization of the surface model. The method is based on a dual eigenvalue decomposition, which leads to fast computation for large datasets of ordered 3-D point clouds. The proposed algorithm is easy to configure and requires few parameters by automatically determining the areas for local refinement. It yields a better defect detection on deformable parts and parts with high tolerance ranges, especially on critical areas with high surface curvature. Experimental results show the effectiveness compared to model-based approaches without multiresolution as well as nonmodel-based methods. Examples are given for successful defect detection where previous methods have failed.

Smart maintenance offers a promising potential to increase efficiency of the maintenance process, leading to a reduction of machine downtime and thus an overall productivity increase in industrial manufacturing. By applying fault detection and prediction algorithms to machine and sensor data, maintenance measures (i.e., planning of human resources, materials and spare parts) can be better planned and thus machine stoppage can be prevented. While many examples of Predictive Maintenance (PdM) have been proven successful and commercial solutions are offered by machine and part manufacturers, wide-spread implementation of Smart Maintenance solutions and processes in industrial production is still not observed. In this work, we present a case study motivated by a typical maintenance activity in an industrial plant. The paper focuses on the crucial aspects of each phase of the PdM implementation and deployment process, toward the holistic integration of the solution within a company. A concept is derived for the model transfer to a different factory. This is illustrated by practical examples from a lighthouse factory within the BOOST 4.0 project. The quantitative impact of the deployed solutions is described. Based on empirical results, best practices are derived in the domain and data understanding, the implementation, integration and model transfer phases.

Mounting sensors in disk stack separators is often a major challenge due to the operating conditions. However, a process cannot be optimally monitored without sensors. Virtual sensors can be a solution to calculate the sought parameters from measurable values. We measured the vibrations of disk stack separators and applied machine learning (ML) to detect whether the separator contains only water or whether particles are also present. We combined seven ML classification algorithms with three feature engineering strategies and evaluated our model successfully on vibration data of an experimental disk stack separator. Our experimental results demonstrate that random forest in combination with manual feature engineering using domain specific knowledge about suitable features outperforms all other models with an accuracy of 91.27 %.

We propose a novel method for registration of partly overlapping three-dimensional surface measurements for stereo-based optical sensors using fringe projection. Based on two-dimensional texture matching, it allows global registration of surfaces with poor and ambiguous three-dimensional features, which are common to surface inspection applications. No prior information about relative sensor position is necessary, which makes our approach suitable for semi-automatic and manual measurement. The algorithm is robust and works with challenging measurements, including uneven illumination, surfaces with specular reflection as well as sparsely textured surfaces. We show that precisions of 1 mm and below can be achieved along the surfaces, which is necessary for further local 3D registration.

This chapter presents the advanced manufacturing processes and big data-driven algorithms and platforms leveraged by the Boost 4.0 big data lighthouse project that allows improved digital operations within increasingly automated and intelligent shopfloors. The chapter illustrates how three different companies have been able to implement three distinct, open, yet sovereign cross-factory data spaces under a unified framework: the Boost 4.0 big data reference architecture and Digital Factory Alliance (DFA) service development framework.

Manufacturing systems are dynamic and exhibit increasing complexity and uncertainty. Smart manufacturing uses Data Analytics methods to optimize manufacturing processes, systems and products. One approach to structure use cases in production management in smart manufacturing is the Product-Process-Resource (PPR) model, where the resource executes a process on a given product. The PPR model needs to be extended for smart manufacturing, to meet the requirements of prescriptive analytics (but not exclusively). Our contributions are an extended PPR model for prescriptive analytics (P2PR) that involves environmental effects, expert knowledge and adds a process sub-model distinguishing between manufacturing and supervisory processes. We develop prescriptive analytics decision-making categories based on the area of validity and the degree of interconnectivity. The combination results in a systematization scheme for prescriptive analytics use cases in a smart factory environment. It assists entities to find shared characteristics in different prescriptive smart factory use cases within one production ecosystem. A mapping of prescriptive algorithms (as part of a use case) to a category and domain is enabled for future case studies.