改进免疫遗传算法的篦冷机二次风温故障诊断

针对遗传算法学习贝叶斯结构时局部寻优能力差的问题, 本文提出一种改进的免疫遗传算法(IIGA)学习贝 叶斯结构. 首先利用最大支撑树与评分函数构建两个初始种群, 然后在种群内部引入改进免疫算子与自动交叉变 异算子, 在种群之间引入改进的联姻策略与师生交流机制, 最后通过迭代搜索到最优贝叶斯结构. 在标准网络中与 遗传算法相比, 提升了遗传算法的局部寻优能力. 利用IIGA算法得到篦冷机水泥熟料换热工艺参数的结构, 并以此 结构为基础进行参数学习与故障推理, 最终得到二次风温的故障诊断模型, 对节约燃煤, 保护环境具有一定实际意 义.

Fault diagnosis of secondary air temperature of grate cooler based on improved immune genetic algorithm

For the problem of poor local optimization ability of genetic algorithm in learning Bayesian structure, an improved immune genetic algorithm (IIGA) is proposed. First, the construction of two initial populations uses maximum support tree and scoring function. Then, IIGA introduces improved immune operator and automatic cross and mutation operator into the population, and imports improved marriage strategy and teacher-student communication mechanism between the populations. Finally, the optimal Bayesian structure is found through iteration. Compared with the original genetic algorithm in the standard network, the local optimization ability of genetic algorithm is improved. IIGA gets the structure of technological parameter that is in terms of heat transfer of cement clinker in grate cooler, and the progress of parameter learning and fault reasoning is executed based on this structure. This research obtains the fault diagnosis model of secondary air temperature, which provides some support for saving coal and protecting environment.