This paper discusses the development of wearable devices for monitoring vibrations induced on the hand-arm system, and the challenges involved in modifying measurement procedures defined by the ISO 5349 standard to be employable on-field. The paper proposes the use of embedded Artificial Intelligence (AI) by implementing Machine Learning (ML) algorithms on low-complexity devices, such as microcontrollers and FPGAs, to perform inferences based on data acquired by on-board sensors. The paper presents an ML-based approach for the classification of vibrations applied on the hand-arm system, distinguishing between harmful and harmless accelerations. The paper describes the dataset used for the NN training, validation, and testing, and the input signal preprocessing performed via FFT. The paper also discusses the advantages of using FPGAs for implementing the NN model and input signal processing and presents a lighter NN model exploiting sigmoidal activation functions for equivalent performance.
Addabbo, T., Landi, E., Moretti, R., Parri, L., Peruzzi, G., Pozzebon, A., et al. (2023). A Low-Complexity FPGA-Based Neural Network for Hand-Arm Vibrations Classification. In 2023 IEEE International Workshop on Metrology for Industry 4.0 & IoT (MetroInd4.0&IoT) (pp.19-23). New York : Institute of Electrical and Electronics Engineers Inc. [10.1109/MetroInd4.0IoT57462.2023.10180160].
A Low-Complexity FPGA-Based Neural Network for Hand-Arm Vibrations Classification
Addabbo T.;Landi E.;Moretti R.;Parri L.;Spinelli F.
2023-01-01
Abstract
This paper discusses the development of wearable devices for monitoring vibrations induced on the hand-arm system, and the challenges involved in modifying measurement procedures defined by the ISO 5349 standard to be employable on-field. The paper proposes the use of embedded Artificial Intelligence (AI) by implementing Machine Learning (ML) algorithms on low-complexity devices, such as microcontrollers and FPGAs, to perform inferences based on data acquired by on-board sensors. The paper presents an ML-based approach for the classification of vibrations applied on the hand-arm system, distinguishing between harmful and harmless accelerations. The paper describes the dataset used for the NN training, validation, and testing, and the input signal preprocessing performed via FFT. The paper also discusses the advantages of using FPGAs for implementing the NN model and input signal processing and presents a lighter NN model exploiting sigmoidal activation functions for equivalent performance.File | Dimensione | Formato | |
---|---|---|---|
A_Low-Complexity_FPGA-Based_Neural_Network_for_Hand-Arm_Vibrations_Classification.pdf
non disponibili
Tipologia:
PDF editoriale
Licenza:
NON PUBBLICO - Accesso privato/ristretto
Dimensione
658.43 kB
Formato
Adobe PDF
|
658.43 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11365/1242434