Prediction of physical realizations of Coordinated Universal Time with hybrid quantum neural network

In this paper, we present for the first time the use of a Hybrid Quantum Neural Network (HQNN) to predict the time differences UTCr – UTC(k). We adopt a multiple-input single-output (MISO) prediction strategy, allowing the next UTCr – UTC(k) value to be predicted using p past values. The model has been evaluated with data from three different UTC(k) timescales, and its performance is compared against a classical multi-layer perceptron (MLP). The results indicate that HQNN achieves strong predictive performance for UTC(k) timescales. These findings suggest that HQNN-based predictions could help improve UTC(k) stability, providing more frequent and accurate estimations of its deviations from UTC.