Processing Sensor Signal Under Low Values of Signal to Noise Ratio
Zinovii Liubun, Bohdan Bryk, Vasyl Mandziy, Oleksandr Karpin, Bogdana Kalivoshka, Serhiy Velhosh · 2023
Creating effective means for the implementation of the filtering algorithms is an important task. The method significantly reduces the cost of the implementation due to the unnecessity of involving high-level experts and creating adaptive means, whereas it ensures the maximum speed with minimum computing resources. Such processing algorithms are expected to provide the possibility of easy implementation on microcontrollers and microprocessors. The usage of the neural network (NN) approach provides the possibility to quickly and with insignificant costs implement the compact NNs for filtering noise. This article considers three structures of NNs for filtering noisy signals. Herein, you can learn the following about NNs: •the possibility of their fast adaptation - tuning or training •the proof of their effectiveness for the emittance of signal under low value of signal to noise ratio •the results, which confirm the possibility to obtain simple NNs for signal filtering under high noise levels.