The measurement of particulate matter covers a wide range of aeras and is of great importance for the development of society and the progress of science and technology. The light extinction method is one of the particle measurement methods, which is simple in principle, easy to measure, and can be applied to the measurement of particles in three phases: gas-liquid-solid. The traditional light extinction model is applicable to a single particle system, in which the extinction particle size measurement model is typically constructed based on Mie scattering theory and Lambert-Beer Law. However, the light extinction phenomena of mixed particles are more complex. Factors such as particle size, refractive index, mixing ratio, and particle size distribution affect the light extinction characteristics, making the traditional model inadequate. A mixed extinction model had been developed by weighting the extinction cross-section in terms of the mixing ratio. This model aimed to investigate the extinction characteristics of mixed particles, categorized as monodisperse and polydisperse, while analyzing the effects of particle size, refractive index, mixing ratio, and distribution parameters on the extinction spectra. Furthermore, a measurement system for extinction spectra was established, and a series of experiments were conducted using properly prepared suspensions of silica and polystyrene particles with mixing ratios of 0.25, 0.5, and 0.75. The experimental spectra were then inverted using an improved differential evolutionary algorithm. The results showed that the calculations using the mixed extinction method were in better agreement with the experimental extinction spectra, with root mean square errors within 0.08. Afterwards, simultaneous multi-parameter inversion of the polydisperse mixed particle system under different mixing ratios had been achieved. The absolute values of maximal inversion errors for particle sizes and mixing ratios were 7.44% and 7.48%, respectively, which were smaller than those obtained under the monodisperse hypothesis. The distribution parameters also reflect the characteristics of the experimental samples with narrow particle size distribution.