Home Research Projects Current Neural Network Method for Improving of The Accuracy of Information-Measurement Systems of Ultraviolet Radiation
Neural Network Method for Improving of The Accuracy of Information-Measurement Systems of Ultraviolet Radiation

Principal investigator and project executor: Prof. Anatoly Sachenko

Project is executed within interuniversity network Erasmus Mundus together with partners from Alaxender Ioan Kuza University, Iassi, Romania.

Duration: 01.01.2013-31.12.2014

Goal: development of new neural network method for improving of the accuracy of information measuring systems for measurement of ultraviolet radiation.

Research purpose: neural network methods and means of accuracy improving of the information-measuring systems for measurement of ultraviolet radiation.

Research methods: structural and functional analysis (error analysis of measuring systems for measuring of UV radiation level and UV sensors); methods of neural networks theory, the method of gradient ascent in the space of weight coefficients and neurons thresholds of (for NN training); simulation methods (for experimental research of developed methods); technique of primary transformer investigation.

Current project results:
-    The methods of signal processing of multiparameter sensors was proposed. Simulations was conducted in MathLab.
-    The software for modeling of the real multiparameter sensors behavior was developed. The software allows to type include into the model random and systematic errors as well, and identify the limits of the proposed methods.
-    Application for Ukrainian patent and for useful model was made.

Team:
-    Anatoly Sachenko
-    Olexiy Roshchupkin
-    Volodymyr Kochan