Abstract

This study describes a non-contact low-cost X-band sensor system for determining the soluble solid content (SSC) of a sugar solution. The system adopts a transmission signal technique with two frequency pairs (10.2 GHz paired with 10.4 GHz and 10.2 GHz paired with 10.6 GHz) from three transceiver modules. Each module has a microstrip patch antenna, mixer circuit, and dielectric resonator oscillator. To simplify the transmission power frequency of each frequency pair, the frequency is down-converted to an intermediate frequency (IF) signal using a frequency mixer. The IF signals are then compared using a gain and phase detector to find their magnitude ratio and phase difference. The measured SSC-level data are randomly divided into three datasets and input to an artificial neural network (ANN) for training. The training output is the SSC level in Brix degree. The proposed ANN structure comprises four input nodes, eight hidden nodes, and four output nodes, affording low complexity and resource savings while providing 92.98% accuracy. Therefore, the proposed low-cost sensor system can achieve precise decision-making and real-time measurement.