Particle size is an essential parameter for industrial processes. The ultrasound method can acquire the particle size distribution (PSD) with multifrequency attenuation characteristics. However, the acoustic path in the traditional measurement mode with a pair of transducers cannot cover the whole measurement zone, and the heterogeneous distribution affects the measurement results. Hence, an ultrasound array strategy with multiple acoustic paths is proposed to eliminate the disadvantage. A hybrid ultrasound attenuation model combining McClements and Bouguer–Lambert–Beer Law (BLBL) is adopted to describe the different attenuation mechanisms. The PSD is obtained with an improved particle swarm optimization (PSO) algorithm, and the efficiency of the proposed PSD inversion algorithm is verified by the simulations. A liquid–solid two-phase measurement device is built for the experimental investigation. The PSD measurements show agreement with the optical measurement results, and the measurement accuracy using the ultrasound array method is improved compared to the traditional one-pair transducer measurement. Combining the PSD measurements of different acoustic paths, the particle distribution of the measurement zone is characterized, and the flow behaviors of the mixing process can be further captured.