Measurement system for honey moisture content based on microwave pseudo-waveguide and deep neural network

The moisture content of honey is a critical quality factor that directly affects its viscosity, shelf life and antibacterial properties. Traditional moisture content measurement methods for honey generally have limitations such as destructive sampling, long testing cycles or complicated sample pretreatment. To address this issue, this article proposes a moisture content measurement system based on a microwave pseudo-waveguide and a probe antenna. It realizes rapid and nondestructive moisture content measurement for honey.

The system was firstly modeled and simulated in CST Microwave Studio, and the simulation results verified the feasibility of the proposed scheme. Experimental tests were then carried out: scattering parameters were obtained through multifrequency scanning to characterize interaction between microwave electromagnetic field and honey samples. Frequency selection was performed on the collected scattering parameters to retain dominant frequency bands, and a deep neural network was adopted to construct a honey moisture prediction model. The prediction results were compared with measurements from the standard gravimetric method. The established model delivers excellent performance with a coefficient of determination (R²) of 0.9969, root mean square error of 0.0057 and mean absolute error of 0.0044, demonstrating high prediction accuracy.

The proposed microwave measurement method features the advantages of nondestructiveness, rapid measurement and environmental friendliness. The measurement scheme integrating microwave pseudo-waveguide technology with artificial intelligence algorithms holds broad application prospects in the agricultural and food industries, and provides a new technical approach for rapid quality evaluation of honey. © 2026 Society of Chemical Industry.