Potential of Using Simulated Data in Processing Photoacoustic Measurement Data

Abstract

This paper explores the potential of using simulated data in calibration of photoacoustic measurement system. The database of simulated experimental values is created using software developed on the bases of the theory-mathematical model. Reliability of the data was gained thanks to the expert knowledge. An artificial neural network as a precise prediction tool is trained on the developed database of simulated data to recognize type of the microphone used as a detector in photoacoustic experiment. The result is classification model satisfies the basic requirements of a photoacoustic experiment: accuracy, reliability and real time operations. The paper discusses the optimization of classification model in terms of used computational power, required precision and process rate in relation with defined problem. The obtained results justify the idea of using simulated data in photoacoustic. Presented theory-mathematical model and classification model are part of developed machine learning framework for processing photoacoustic measurement data.