Detection of sylvine concentration in water based on permittivity

In order to find out a fast detection method for pollutants in water, through analysis on the relationship between concentration of different ion or cation in sylvine solutions and dielectric properties, the rapid and accurate detection method for sylvine concentration in water based on permittivity was studied. Using the new permittivity detection sensor designed for liquid dielectric materials, frequency-response data of sylvine solutions with different concentrations over a wide frequency range were measured by experiments. With PLS regression, the concentration prediction model for various ions or cation in sylvine solutions was established. By principal component analysis, the characteristic frequencies of potassium cation have been identified. With the characteristic frequencies, the new prediction model for concentration of potassium cation was established. The results showed that each ion or cation solution has its unique frequency-response pattern; frequency-response properties of different kinds of sylvine solution are significant difference at the low frequency region. For mixing solutions of low concentration, the coefficient of determination (R2) was high up to 0.98865 while RMSE was 0.37598 mg/L for training PLS regression model of potassium cation concentration when only using gain data. For validation of PLS regression model, R2 was high up to 0.98861 while RMSE was 0.41031 mg/L. Principal component analysis and identification of the characteristic frequencies can reduce the amount of data significantly, and also have a high accuracy.