Advanced data analytics for additive manufacturing energy consumption modelling, prediction, and management under Industry 4.0

The topic of ‘Industry 4.0’ has become increasingly important in both industry and academia since it was first published. Under this trending topic, many related capabilities required by current manufacturing systems have been pointed out in both academia and industry, such as automation, sustainability, and intelligence. Additive manufacturing (AM) is one of the most popular manufacturing systems in the era of Industry 4.0. Although the AM system tends to become increasingly automated and flexible, the issue of energy consumption still attracts attention. It is related to many attributes in different components of an AM system, which are represented as multiple source data, such as process operation data, working environment data, design-relevant data, and material condition data. How to integrate and analyse the multi-source data for AM energy modelling, prediction, and management has become a crucial research question. This research was structured according to four themes. Firstly, a categorical classification is proposed based on the research gaps between current manufacturing systems and Industry 4.0 requirement. Nine varied applications are generated relying on their classification to provide a roadmap to raise the intelligence level of manufacturing systems to achieve Industry 4.0 requirement. Inspired by this classification, a framework was designed for leading the research of AM energy consumption modelling, prediction, and management. The framework includes four layers, data sensing and collection layer, data pre-process and integration layer, data analytics layer, and knowledge and application layer. This four-layered framework covers the entire knowledge discovery process from data generation to performance presentation. Secondly, due to multi-source data of the AM systems usually involving nested hierarchies, a hybrid approach is proposed to tackle the issue. This hybrid approach incorporates clustering and deep learning technologies to integrate the multi-source data which is collected by the Internet of Things (IoT), to model energy consumption iv for AM systems. Multi-source data is analysed and collected. The data collection methods are introduced within the validation of a selective laser sintering (SLS) system. Results derived using this hybrid approach reveal that it outperforms pre-existing approaches. Thirdly, while existing studies reveal that AM energy consumption modelling largely depends on the design-relevant features in practice, it has not been given sufficient attention. Therefore, in this research, design-relevant features are examined with respect to energy consumption prediction based on the study of AM energy modelling. By reviewing the literature of Design for AM and analysing some representative design models, AM design patterns are obtained and listed. Two types of design-relevant features are found, part-design features and process-planning features. The AM energy consumption knowledge, hidden in the design-relevant features, is exploited for prediction through a design-relevant data analytics approach. Finally, methods enabling energy consumption management are provided in this research, which includes framework, modelling, prediction and the optimisation. The energy consumption optimisation method is based on particle swarm optimisation (PSO), and driven by deep learning technology, named as deep learning driven particle swarm optimisation (DLD-PSO). The proposed optimisation method aims to reduce the energy utility by optimising the design-relevant features. Deep learning was introduced to address several issues, in terms of increasing the search speed and enhancing the global best of PSO. The approaches proposed in this research were validated with the data collected from the target AM system, and the results reveal their merits. The expected main achievement of this research is to pave the way for AM energy consumption modelling, prediction, and management through the advanced data analytics, which provides a feasibility study for achieving Industry 4.0. As such, it offers great potential as a route to achieve a more practical and generalised implementation of digital and intelligent manufacturing