Application of Machine Learning in manufacturing: advantages and challenges

Author: Vivek Srivastava

The manufacturing industry today is experiencing a never seen increase in complexity, dynamic and chaotic behaviours; in fact this has been the way of working for quite some time. In order to being able to operate in an efficient manner and satisfy the demand for high-quality products, it is essential to utilize available large amounts of data, which the manufacturing environment possess.

Key challenges faced by manufacturing are:

1.   Adoption of advanced manufacturing technologies

2.   Growing importance of manufacturing in high value-added products

3.   Agile and flexible enterprise capabilities and supply chains

4.   Innovation in products, services, and processes

5.   Close collaboration between industry and research to adopt new technologies

To overcome some of today’s major challenges of complex manufacturing systems, we can utilise the Machine Learning (ML) techniques. These data-driven approaches are able to find highly complex and non-linear patterns in data of different types and sources and transform raw data to features spaces, so-called models, which are then applied for prediction, detection, classification, regression, or forecasting.

ML techniques has the ability to handle high-dimensional, multi-variate data, and the ability to extract implicit relationships within large data-sets in a complex and dynamic, often even chaotic environment. Since most engineering and manufacturing problems are data-rich but knowledge-sparse, ML provides a tool to increase the understanding of the domain.

ML allows to reduce cycle time and scrap, and improve resource utilization in certain hard manufacturing problems. Furthermore, ML provides powerful tools for continuous quality improvement in a large and complex process such as semiconductor manufacturing.

Emerging applications of ML in manufacturing includes:

1.   Predictive maintenance or condition monitoring

2.   Warranty cost estimation

3.   Propensity to buy

4.   Demand forecasting

5.   Throughput optimization

Applying ML in manufacturing may result in deriving pattern from existing data-sets, which can provide a basis for the development of approximations about future behaviour of the system. This new information (knowledge) may support process owners in their decision-making or used to automatically improve the system directly. In the end, the goal of certain ML techniques is to detect certain patterns or regularities that describe relations.

An advantage of ML algorithms is the ability to handle high dimensional problems and data. ML algorithms is to discover formerly unknown (implicit) knowledge and to identify implicit relationships in data-sets. ML algorithms provide the opportunity to learn from the dynamic system and adapt to the changing environment automatically to a certain extent. The adaptation is, depending on the ML algorithm, reasonably fast and in almost all cases faster than traditional methods.

The goal of ML techniques is to detect certain patterns or regularities that describe relations based on application of existing data. Data to be fed to ML algorithm after the available data are secured, the data often have to be pre-processed depending on the requirements of the algorithm of choice. Pre-processing of data has a critical impact on the results. However, there are many standardized tools available which support the most common pre-processing processes like normalizing and filtering the data. Also it has to be checked whether the training data are unbalanced. This can present a challenge for the training of certain algorithms. In manufacturing practice, it is a common problem that values of certain attributes are not available or missing in the data-set. These so-called missing values present a challenge for the application of ML algorithms.

Some manufacturing setup i.e. steel manufacturing- kiln operation in particular with several months- and sometimes a few years-long campaign run, have too little data to be statistically meaningful when put under an analyst’s lens. The challenge for senior leaders at these setup will be taking a long-term focus and investing in systems and practices to collect more data. They can invest incrementally—for instance, gathering information about one particularly important or particularly complex process step within the larger chain of activities, and then applying sophisticated analysis to that part of the process. Multiple options are available to quickly capture high volumes of data generated at different stages of the manufacturing process i.e. Oracle offers a range of products including Oracle NoSQL Database and Oracle Database 11g.

ML, steering manufacturing in a new era of man–machine collaboration, it will require the biggest change in the way organisations work. While the machine identifies patterns, the human decoder’s responsibility will be to construe them for different fragments and to indorse a course of action. Behavioural change will be decisive, and one of top management’s key roles will be to guidance and inspire it. And enable frontline managers, armed with insights from increasingly meaningful insights and output from intelligent machine algorithm, must learn to make more decisions on their own, with top management setting the overall direction and intervening to manage the exception.

With fast paced developments in the area of algorithms and increasing availability of data (e.g. due to low cost sensors and the shift toward smart manufacturing) and computing power, the applications for machine learning especially in manufacturing will increase further at a rapid pace. As of today, supervised algorithms have the upper hand in most application in the manufacturing domain. However, with the fast increase in available data, thanks to more and better sensor technologies and increased awareness, unsupervised methods may increase in importance in the future.

One area, which saw fast pace developments in terms of not only promising results but also usability, is machine learning. Promising an answer to many of the old and new challenges of manufacturing, machine learning is widely discussed by researchers and practitioners. However, the field is very broad and even confusing which presents a challenge and a barrier hindering wide application.