Monitoring on triboelectric nanogenerator and deep learning method

As the basic hydrological parameters, the concentration and type of suspended sediment particles greatly influence the aquatic ecological environment and the water conservancy infrastructure. However, because of the various composition and changing concentration, challenges still remain in low-cost and real-time sediment monitoring. Herein, we report a potential method to realize real-time monitoring of sediment particles parameters by introducing a particles-laden droplet-driven triboelectric nanogenerator (PLDD-TENG) combined with deep learning method. The mechanism of PLDD-TENG was proved to be induced by the liquid-PTFE electrification and particles-electrode electrostatic induction. The output signals of PLDD-TENG were measured under different particles types and mass fractions. The results indicated that the output signals were sensitive to the particles type and mass fraction. A convolutional neural network (CNN) deep learning model was adopted to identify the particles parameters based on the output signals of PLDD-TENG and high identifying accuracy was achieved. Meanwhile, this model can identify particles types in the mixed solution. An intelligent system was finally developed to realize visualization of real-time monitoring for sediment particles. These findings are crucial in both fundamental understanding and application prospect of triboelectric effect in two-phase flow.