Multilayer polymer composite structures have been playing important roles in various fields, but the voids inside are not allowed in most scenarios. Ultrasonic technology has been widely used to inspect voids in concrete and metal structures. However, the application of ultrasonic imaging in polymer composite structures is severely blocked by the coating and lamination structures and unstable manufacturing induced sound speed variations. In this paper, a method to autofocus imaging of internal voids in multilayer polymer composite structures with ultrasonic phased array is firstly proposed. The method processes the full matrix capture (FMC) and focuses all voids in the multilayer structure automatically without the prior information of the speed of sound (SOS). The method utilizes the focus criterions to evaluate the focusing quality and then estimates the SOS with differential evolution layer by layer from surface to deep, which improves the robustness and computational efficiency. The method was examined with simulation data from three multilayer structures and well-focused all voids with position error less than 0.6 mm and SOS error less than 6 %. Moreover, the method was verified with the experimental data and focused voids with position error less than 1 mm.

多层聚合物复合结构在各个领域都发挥着重要作用，但在大多数情况下，内部的空隙是不允许的。超声波技术已广泛用于检查混凝土和金属结构中的空隙。然而，超声成像在聚合物复合结构中的应用受到涂层和层压结构以及不稳定的制造引起的声速变化的严重阻碍。本文首次提出了一种利用超声相控阵对多层聚合物复合结构内部空隙进行自动聚焦成像的方法。该方法处理全矩阵捕获（FMC）并自动聚焦多层结构中的所有空隙，而无需声速（SOS）的先验信息。该方法利用聚焦准则评估聚焦质量，然后从表层到深层逐层差分进化估计SOS，提高了鲁棒性和计算效率。使用来自三个多层结构的模拟数据对该方法进行了检查，并很好地聚焦了位置误差小于 0.6 mm 和 SOS 误差小于 6% 的所有空洞。此外，该方法通过实验数据和位置误差小于 1 mm 的聚焦空洞进行了验证。