Control charts pattern recognition based on linear vector quantization neural networks: Application on the business produced ready mixed concrete

The objective in this study is to detect the errors that occur or may occur in the future during the process in which the company's quality objectives are fulfilled and to show the applicability of the Artificial Neural Networks (ANN) which is one of the Artificial Intelligence (AI) techniques. Thus, it will be able to contribute to the main purposes which make quality control necessary such as to raise the level of quality, reduce operating costs, time savings, raising employees' motivation and reducing customer complaints. In the first part, in general, quality, quality control and statistical quality control techniques are described. AI techniques, which are the key concepts of this study, are introduced; the definition, structure, model and application of ANN are explained; the ANN models and their results used in the quality control studies are presented. Additionally, ANN model used in the present study is also discussed in detail in this section. In the second part comprising the application, a control charts pattern recognition (CCPR) application has been carried out using ANN which is one of the techniques of AI and this application has been transformed to a visual program using the obtained results. For this purpose, average compressive strength, one of the most important quality indicators, of a company that produces ready-mixed concrete has been used. A LVQ (Linear Vector Quantisation) type ANN model has been established by using the quality characteristics observation values related to control charts and the parameters related to control charts, and when these two models are compared, it has been found out that the model whose quality characteristics have been constructed using the observation values result in more successful results than that constructed with the model's control charts. The visual program which is suitable for LVQ algorithm using weight values about the raw data model having the best performance has been tested and it has been concluded that the CCPR application can also be used to control the average of concrete compressive strength efficiently. Keywords: Quality Control, Pattern Recognition in Control Charts, Neural Networks, LVQ, Concrete Quality