Experimental study demonstrates real-time microplastic detection in seawater using a portable microwave sensor, highlighting a practical method for remote marine pollution tracking.
Key Points
To develop a portable, real-time microwave resonant sensor system integrated with microfluidics and machine learning for microplastic detection in seawater.
Fabricated a microwave resonant sensor integrated with a microfluidic channel, an embedded system, and a Bluetooth module for wireless transmission.
Tested detection performance using polyvinyl chloride powder (particle size: 6.5 ± 1 μm) in deionized water, artificial seawater, and natural seawater samples.
Trained a convolutional neural network algorithm on microwave test outputs to predict microplastic concentrations.
Achieved a recognition rate exceeding 96% with a concentration tolerance of ±0.01 mg/mL.
Demonstrated a minimum detection limit of 48.72 ng/mL while mitigating interference from the seawater matrix.
Observed high consistency between convolutional neural network predictions and actual microplastic concentrations.