基于约束主导混合粒子群算法的风力机叶片优化方法研究

Optimization Method for Wind Turbine Blade Based on Dominanted-constraint Hybrid Particle Swarm

为提高叶片在额定风况和低速风况下的功率系数,研究叶片各叶素处的气动外形参数分布。针对风力机通常运行在低风速风况下,而叶片的优化模型很少考虑该因素的影响,建立基于叶素动量理论和Wilson理论的带低风速功率系数的非线性约束优化模型。由于在处理约束条件的惩罚函数法中罚因子难以确定,而导致算法过早陷入局部解的早熟现象,提出一种结合可行性约束主导处理方法的混合粒子群算法。该算法基于粒子群优化和模拟退火理论,采用可行性约束主导在退火概率突跳下对不可行约束解进行随机生存选择,使种群保持多样性,从而朝更优方向进化,解决了非线性约束条件难以处理和种群易陷入局部解的问题。以1.5 MW风力机叶片为研究对象,建立非线性约束优化模型,对该算法进行了验证。研究成果表明该方法可以有效地处理优化模型的非线性约束,避免优化过程陷入早熟,提高了叶片在额定风速和低风速区域的功率系数。为非线性约束处理方法的研究提供了一种很好的理论分析途径。

To increase the power coefficient of blade at both rated and low wind conditions, the distribution of aerodynamic shape parameter at each blade element is studied. The wind turbine is typically operated under low wind conditions, however the influence has rarely been considered in blade optimization model. Hence, a nonlinear constrained optimization model with power coefficient under low wind conditions is introduced based on the blade element momentum theory and Wilson theory. Since the penalty factor is difficult to be determined in penalty function method when dealing with constraints, which may lead to prematurity phenomenon that the algorithm falls into local solution, a hybrid particle swarm algorithm combined with feasible dominated-constraint method is brought up. Based on particle swarm optimization theory and simulated annealing theory, the algorithm applies feasible dominated-constraint method to perform random survival selection under drifting annealing probability, which keeps the population diverse and can be evolving in more optimized direction, thus solves the problem that nonlinear constraint is difficult to handle and the population＇s tendency to fall into local solution. So as to verify the algorithm, a nonlinear constrained optimization model for the 1.SMW wind turbine blade is established. The results indicate that the method can effectively handle nonlinear constraints, avoid prematurity of the process and increase the power coefficient of blade under rated and low wind conditions. It provides an excellent way of theoretical analysis to handle nonlinear constraints.