REFERENCE ARCHITECTURE FOR THE SOFTWARE PLATFORM FOR VERIFICATION OF MATHEMATICAL MODELS

Abstract

The purpose of this work is to develop software for verification and validation of mathematical models that automatically evaluates similarities and differences between two curves. This program is created to help engineers and analysts perform curve comparisons during the numerical model verification and validation process. The software package enables automatic pre-processing of two input curves to make them comparable.The methodology used in the work includes several options for preprocessing the input curves before calculating the comparison metrics. The data can be filtered and synchronized, or any shifts or drifts can be removed. Various pre-processing options are available through a user-friendly graphical user interface to ensure the most accurate verification possible. Any operation, from entering curves to selecting pre-processing options and final visualization of the results, is available through this interface.The scientific novelty of the work consists in the automation of the process of pre-processing and comparison of curves, which significantly simplifies and speeds up the process of verification and validation of numerical models. The developed program reduces human errors and provides more accurate results, which is important for improving the quality and reliability of models. In addition, the ability to save numerical results in a convenient format of a spreadsheet and graphs in the form of raster images makes the program even more useful for further research.The conclusions drawn on the basis of the conducted studies emphasize the importance of verification and validation as critical stages in the development of computer models to ensure their accuracy and reliability. The work states that achieving the best results requires constant improvement and adherence to best practices in the process of developing and applying models. Simple examples using an analytical form illustrate the characteristics of metrics that demonstrate the effectiveness and accuracy of the developed software.Thus, the presented work makes a significant contribution to the field of verification and validation of numerical models, offering innovative approaches to the automation of curve comparison and improving the accuracy and efficiency of this process. This program will become an indispensable tool for engineers and analysts, helping to improve the quality of mathematical models and their correspondence to real data.