Modelling and Validation of Industrial Measurement Systems - Aspects of Quality and Human Factors

Abstract: Although there is a fairly large number of books and articles concerning the validation of technical measurement systems, there are only a few which take into account the whole measurement system consisting of both technical and non-technical aspects. A measurement system comprises both the measurement process and the human being who conducts the measurement. In this way, the human being may be a part of the initiating process (e.g. sampling), and/or part of the ongoing process (such as manual handling of the sample). The human being, however, is always a part of the final process, where the results are interpreted. The purpose of this thesis is to validate industrial measurement systems, including both technical and non-technical aspects of these systems, in order to understand them better and to increase the ability to estimate the quality of the product. In order to do so, a general method that estimates a quality index capable of handling both qualitative and quantitative factors is proposed. The suggested method uses a fuzzy neural network as a tool since such a tool has a learning function that allows the integration of the human judgement of quality into a quantitative index. This method thus constitutes a promising starting point for the future development of a representative quality index based on both quantitative and qualitative factors. Such an index may be able to allow better comparison of different products or services. Part I of this thesis gives a brief overview of the research methods used, the research areas investigated, and the theory behind this field. This thesis consists of eight papers, presented in Part II. The papers discuss, among other things, the role of the human being in automated measurement systems, problems that occur when implementing advanced measurement concepts to modern industries, automated control of sensors and quality assessments.