Radio Frequency (RF) energy harvesting has
been employed to power wireless devices. Nevertheless, RF
energy harvesting encounters restrictions regarding the
quantity of power it can harvest depending on signal
accessibility. As a result, accurately predicting energy levels
becomes crucial for enhancing the performance of energy
harvesting circuits. Most research efforts have concentrated
on enhancing power harvesting policies or theoretically
estimating the energy obtained through RF energy
harvesting. Moreover, the existing literature has primarily
focused on single-band prediction approaches. This paper
presents a multi-band RF energy prediction approach for RF
energy harvesting systems. We collect real-time RF energy
using software-defined radio technology. The proposed
approach leverages Long Short-Term Memory (LSTM)
neural networks to accurately predict the mean RF energy in
different frequency bands for the next 100 samples, which
corresponds to approximately one hour and a half. The
research explores the research gap in modeling the radio
frequency signal and the need for multi-band prediction
techniques. The results demonstrate the effectiveness of the
proposed approach in predicting RF energy across different
frequency bands, with average accuracies above 98%