Cluster analysis as a method for the planning of production systems

Abstract

Today's situation in manufacturing enterprises is characterized by unpredictability, high-frequency market changes and a turbulent environment. Changing order situation leads to new requirements with regard to output capacity. To some extent, companies have to meet contrary targets in order to stay competitive. These major targets are quality, cost and time. To achieve these targets in a balanced way, manufacturing companies need high flexibility as well as a high productivity at the same time. This is a paradigm shift especially for companies that produce standardized products. This paradigm shift also requires new planning methods. In the paper, the state of the art of methods for the designing of industrial facilities and their relevance to manufacturing enterprises is discussed. By studying those methods, it becomes apparent that planners in companies do not pay enough attention to the analysis of the current situation in production. For this stage of planning the authors improved the method of cluster analysis. The authors describe how to implement this method in industrial practice.

Deep learning‐based clustering of processes and their visual exploration: An industry 4.0 use case for small, medium‐sized enterprises

“…Especially, in the case of OKP planning, there is often no historic manufacturing information available for a given part category or the information is not sufficient to enable an effective estimation. Instead, a multivariate analysis is required that takes into account various aspects such as size, shape, and position in order to identify similar fixture parts so that historical information on these parts can be leveraged for the planning process (Nachtwey et al, 2009). While several methodological attempts at part structuring were undertaken, the clustering approach was defined as one of the most viable options in this regard (Crama & Oosten, 1996; Goswami et al, 2008).…”

“…The heterogeneity of customer orders in terms of unique product structures, different processing pathways, varying process execution durations, due dates, and other factors adds to the complexity of realizing OKP (Huang et al, 2021). A major challenge in this context poses the identification of groups of similar product parts to enable a more fine‐grained statistical exploration as a means of production planning (Nachtwey et al, 2009). We aim to demonstrate that SMEs can tackle such a complicated planning situation using relevant process event data‐driven machine learning solutions on top of the data delivered by affordable information systems.…”

A Design of Experiments Comparative Study on Clustering Methods

“…However, this approximation decreases as the corresponding neuron is further from the winner neuron i. For every neuron within the topological neighborhood of the winner i, the weight vectors are modified according to (10).…”

“…Clustering applications examples found in the literature also demonstrate this diversity. There are papers about the use of clustering algorithms to identify characteristics of people with attempted suicide [6]; to facilitate the diagnosis and treatment of cancer [7]; to identify residential and social patterns of homeless adults [8]; and also in applications in the field of production engineering, e.g., clustering method for production planning [9], [10], and for analyzing product portfolios [11].…”