摘要
针对差分进化算法(Differential Evolution Algorithm,DEA)容易陷入局部最优值且搜索效率不高的缺点,提出一种单纯形-差分进化算法(SM-DEA)。该算法在DEA中引入计算量小、搜索速度快且局部搜索能力很强的单纯形方法(Simplex Method,SM),在迭代过程中交叉使用DEA和SM。对标准函数的测试表明,该算法具有很强的全局搜索能力和很高的搜索效率。将算法应用于乙炔加氢反应器出口乙炔浓度的软测量建模,结果表明该算法能提高模型的精确度,降低预测误差,具有较高的实用价值。
A simplex method-differential evolution algorithm （SM-DEA）which aims at the disadvantage of easily trapped in the local optimization of the original DEA was proposed. It combines DEA with SM. The new algorithm used DEA and SM alternatively in iteration. Several benchmark functions were used to test the algorithm. Results show that the new algorithm has more effective and stronger global searching ability. Finally it is used in optimizing BP＇ s weights and biases to achieve a soft sensor model to predict the acetylene concentration of the acetylene hydrogenation reactor. Results show that the soft sensor model＇ s output precision is increased and predicting error is decreased greatly.
出处
《化工自动化及仪表》
CAS
2008年第6期21-24,共4页
Control and Instruments in Chemical Industry
基金
国家"863"计划项目(2007AA04Z171)
上海市重点学科建设项目(B504)
