多种群遗传算法在湿蒸气参量测量中的应用

Application of multi-population genetic algorithm in wet steam parameter measurement

为了利用激光散射法快速准确地测量出汽轮机中湿蒸气的参量值,实时监测汽轮机末级蒸汽湿度,指导汽轮机安全运行,采用多种群遗传算法,用三次样条插值的方法代替散射理论公式,以理论值与测量值的最小二乘值为目标函数,设置合适的控制参量,对加入高斯白噪声的仿真数据和模拟汽缸的实验数据进行搜索寻优。结果表明,多种群遗传算法能够克制过早收敛,准确地搜索到全局最优解,在遗传代数内反演出湿蒸气的各个参量。该研究对激光散射法测量湿蒸气参量或其它微粒粒度分布的反演是有帮助的。

In order to measure the wet steam parameters with laser scattering method,monitor the steam humidity at the final stage and guide the safe operation of turbine,the multi-population genetic algorithm was adopted,cubic spline interpolation was used in stead of scattering theory,the least squares of theoretical value and measured value were adopted as the objective functions and the appropriate control parameters were set up for searching the optimization among the simulation data added Gaussian white noise and the experimental data of analog cylinder.The experimental results show that multi-population genetic algorithm can effectively restraint premature convergence,locate the global optimal solution accurately and inverse every parameter of wet steam prominently among some hereditary algebras.The study is helpful for the measurement of wet steam parameters with laser scattering method and the inversion of particle size distribution.