Study of the possibility of improving the audit process based on its optimization by the Taguti method

The assessment process is a key component of the activity included in many production and non-production processes. The article discusses the plan for the upcoming study of the factors affecting the quality of the assessment process, which makes a significant contribution to the effectiveness of the quality management systems (QMS) audit process. The possibility of optimizing these factors based on the robust1 parametric design (RPD) method developed by Genichi Taguchi [1] is considered. The possibility2 and limitations of the orthogonal planning of RPD experiments in this study are discussed. Based on the definition of the term «assessment quality» adopted in the study, a list of factors influencing the effectiveness of the process of assessing an object for its compliance with requirements (CER) has been compiled. The composition and structure of these factors have been determined. It has been shown that the CER process has properties characteristic of the field of providing services, the quality of which is ensured only by their preliminary validation, i.e. before the interaction with the party interested in the results of the CER process (for example, the consumer, the customer of the assessment, audit, etc.). It has been shown that one of the reasons for failures (failure to fulfill the ideal assessment function - the absence of errors of the 1st and 2nd kind) is the lack of robustness of the ideal CER function when exposed to various types of interference or noise (hereinafter referred to as noise). It is proposed to apply RPP to the CER process as a validation before its interaction with the consumer. It is proposed to use reference audit situations (RAS) with standard properties as input signals for CER. The concept of the study has been developed, including internal and external orthogonal plans of the experiment for designing the OOST process for the subsequent selection of its optimal conditions based on the data on the estimates of the signal-to-noise ratio - S/N (dB) for different combinations of factors affecting the OOST. Examples are given demonstrating the course of data processing that can be obtained within the framework of the experiment for calculating the S/N ratio for different combinations of factors affecting the OOST and, as a consequence, the audit process.