An Artificial Neural Network Architecture to Classify Workers’ Operations in Manual Production Processes

This chapter delves into the application of artificial neural networks to classify workers' operations in manual production processes, a crucial aspect of modern manufacturing under the Industry 4.0 paradigm. It begins by highlighting the shift from traditional supervisory methods to data-driven decision-making, facilitated by Industrial Internet of Things (IIoT) sensors. The core proposal is an integrated IIoT architecture that leverages Ultrawide-Band (UWB) based indoor positioning systems (IPS) to gather workers' trajectories and feeds them into a Long Short-Term Memory (LSTM) based artificial neural network. This architecture demonstrates impressive performance, achieving an F1-score of 0.93 in classifying picking/deposit activities. The chapter also includes a detailed case study validating the architecture in a real manufacturing job shop, showcasing its potential to enhance process monitoring and improve working conditions. By combining advanced positioning technology with powerful neural networks, this chapter offers a compelling solution for digitizing and optimizing manual production processes.