Research On Liquid Multiparameter Sensing Using Love Wave Device

In many applications of liquid sensing,people are concerned about not the measurement of only a certain characteristic parameter of liquid,but the measurement of all the characteristic parameters of liquid at the same time,such as the parallel measurement of liquid density,viscosity,dielectric constant and conductivity.The Love wave device is suitable for liquid sensing,but the multiple characteristic parameters of the liquid are coupled together,which influence the change of the velocity and attenuation of the Love wave,so it's hard to use the Love wave device for the liquid multiparameter sensing directly.At the same time,it‘s necessary to adopt an effective method to improve the excitation efficiency of Love wave device in liquid environment.To solve the problems mentioned above,the theory model of double piezoelectric layer structure "piezoelectric substrate – piezoelectric film" of the Love wave device,which is for increasing the overall piezoelectric effect of the device,is established.What's more,a two-stage combined neural network method is proposed to realize the liquid multi-parameter sensing of Love wave device,which provides a new idea for the liquid multi-parameter sensing problem.In terms of mechanism research,this article establishes the theory analysis model of Love wave sensor with double piezoelectric structure which including nine different combination of coupling on the basis of the propagation characteristics of acoustic wave in laminated medium.And the model is based on Christoffel equations,boundary conditions and three coupling modes both in piezoelectric substrate and piezoelectric film.The velocity and attenuation of Love wave corresponding to different liquid parameters can be calculated by combining the model and the two-dimensional search method about the viscous liquid.Through the theory analysis model,the influence of piezoelectric film thickness on sensitivity and electromechanical coupling coefficient is simulated and analyzed,and the idea that structure of double piezoelectric can enhance the excitation efficiency of the sensor is verified theoretically.On the basis of the multiple-input-multiple-output characteristic of artificial neural network,large volume of simulation data available in theory model and multiple calibration samples which can be used in actual measurement of liquid,this article presents the way of using two-stage cascading neural network to realize the liquid multi-parameter parallel measurement.According to the property of the metallizing device and the liberalizing device,a two-stage neural network is constructed,the activation functions and the number of hidden neurons are optimized..By using the data from the theory analysis model,the two-stage cascading neural network is tested and the result shows that the relative errors are lower than 7%.Based on the way mentioned above,genetic algorithm is used to optimize the initial weights and threshold,the result of optimization shows the relative errors are lower than 4%.This article not only fabricates the Love wave sensor with the technology of photolithography and magnetron sputtering,but also establishes two test systems based on vector network analyzer and the method of power detection respectively.After testing,it not only experimentally verified the idea of improving the excitation efficiency of the double piezoelectric structure,but also verified the accuracy of the theoretical analysis model.Beyond that,the experimental calibration data is used for two-stage cascading heredity-neural network,and the relative errors of the liquid multi-parameter parallel detection are lower than 4 %,and the accuracy of liquid muti-parameter sensing using Love wave sensor is verified in this article.