Fig. 4: Performance comparison of the AI model leveraging predictions from all sensors with various input features and frequency ranges.
From: Real-time concrete strength monitoring using piezoelectric sensors and deep learning
a Benchmark model performance with the optimized frequency range (10–200 kHz). b Performance without the baseline mechanism. c Comparison of model performance with and without temperature as an input feature. d Contour plot of the frequency spectra for free sensors, showing two main peak clusters (10–200 kHz and 200–500 kHz) and variations between individual sensor signals, underscoring the importance of the baseline mechanism. e Violin plots showing sensor temperature data during slab measurements.
