Cross-flow fans are widely used in heating, wind-curtains, and air-conditionings, as well as other ventilation systems. A single or double arc is generally used as the camber line of cross-flow fans, but this design leads to constraints in the geometry of the blade profiles. In this study, the camber line of a cross-flow fan blade was parameterized by five parameters based on the fourth-order Bezier curve. A two-dimensional computational fluid dynamics (CFD) simulation was conducted to predict the aerodynamic characteristics and the internal flow field. It is necessary in multi-condition optimization, to evaluate the relative importance of the performance parameters under different working conditions and determine their weight factors. Here, a novel maximum entropy method (MEM) was proposed to quantify of volume flow rate, because the method avoids the subjectivity in the selection of the weights. Subsequently, a multi-island genetic algorithm (MIGA), combined with numerical simulation, was used to search the global optimum in the given design space. The results indicated that the optimum combination of the structural parameters reduced the blade channel vortex in a particular location of the impeller and changed the position and size of the eccentric vortex. The volume flow rate of the optimized impeller was 4.28% higher at the minimum rotation speeds and 12.87% higher at the maximum rotation speeds.

横流风扇广泛用于供暖，风幕，空调和其他通风系统。通常使用单弧或双弧作为横流风扇的外倾线，但是这种设计会导致叶片轮廓的几何形状受到限制。在这项研究中，基于四阶贝塞尔曲线，通过五个参数对横流风扇叶片的外倾线进行了参数化。进行了二维计算流体动力学（CFD）仿真，以预测空气动力学特性和内部流场。在多条件优化中，有必要评估不同工作条件下性能参数的相对重要性并确定其权重因子。在这里，提出了一种新颖的最大熵方法（MEM）来量化体积流量，因为该方法避免了权重选择的主观性。随后，使用多岛遗传算法（MIGA）与数值模拟相结合，在给定的设计空间中搜索全局最优值。结果表明，结构参数的最佳组合减少了叶轮在特定位置的叶片通道涡流，并改变了偏心涡流的位置和大小。优化的叶轮的体积流量在最小转速下提高4.28％，在最大转速下提高12.87％。结果表明，结构参数的最佳组合减少了叶轮在特定位置的叶片通道涡流，并改变了偏心涡流的位置和大小。优化的叶轮的体积流量在最小转速下提高4.28％，在最大转速下提高12.87％。结果表明，结构参数的最佳组合降低了叶轮特定位置处的叶片通道涡流，并改变了偏心涡流的位置和大小。优化的叶轮的体积流量在最小转速下提高4.28％，在最大转速下提高12.87％。