The project aims to experimentally analyse the possibility of implementing artificial intelligence and machine learning (ML) in an assisted selection of measurement parameters using contact and non-contact devices for surface irregularities measurements. The implementation will be concerned with the complex impact of the type and degree of data filtering on the measured metrological parameters when using various measurement techniques. Depending on the measuring object and the parameters of surface irregularities, measurement technique, resolution, measurement uncertainty, and other initial information required for its assessment, the algorithm developed within the specified range will relieve the operator before measuring by selecting the measurement technique from the database of available metrological devices, measurement speed, sampling density, type and degree of data filtering used. The project is based on the vast amount of data obtained from various surfaces, collected using contact and non-contact techniques, and implemented with various measurement parameters and results filtering procedures. The measurement results will be used as a large set of input data for the further preparation and implementation of machine learning algorithms using a pre-trained neural network. Measurement uncertainties for surface irregularities will also be estimated and used as input parameters for the algorithm. The data will be prepared suitable for implementing supervised machine learning algorithms depending on their type (numbers, images, filter types, etc.).
Funding: A grant supported this work: A project entitled: "Analysis of the possibility of using artificial intelligence in surface irregularities measurements", financed by the Ministry of Education and Science of the program: Polish Metrology Nr PM/SP/0021/2021 of 29.06.202