论文部分内容阅读
基于锅炉燃烧模型的非线性寻优和基于历史运行工况的数据挖掘是两种常见的锅炉燃烧优化技术,且各有利弊。前者可得到全局最优解,但算法复杂度较高;后者计算较为简易,但只能实现局部最优。结合两种方案的优点,提出基于离线非线性寻优所得最优知识库,采用模糊关联规则挖掘算法,建立最优操作变量(manipulated variables,MVs)决策模型,实现高效、稳定的锅炉燃烧优化。关联规则挖掘中,提出基于k-均值聚类的语言变量非均等模糊分割,以提高所得规则库的可信度;并基于改进的支持度和置信度概念实现规则库的精简。仿真结果表明,基于该文最优MVs决策模型的锅炉燃烧优化结果与全局寻优结果接近,且算法复杂度低、稳定性高,适合于在线实时优化与自适应更新。
Nonlinear optimization based on boiler combustion model and data mining based on historical operating conditions are two common boiler combustion optimization techniques, each with its own advantages and disadvantages. The former can get the global optimal solution, but the algorithm is more complex; the latter is easier to calculate, but can only achieve the local optimum. Combining the advantages of the two schemes, an optimal knowledge base based on off-line nonlinear optimization is proposed. The fuzzy association rule mining algorithm is used to establish the optimal manipulated variables (MVs) decision model to achieve efficient and stable boiler combustion optimization. In association rule mining, a non-uniform fuzzy segmentation of linguistic variables based on k-means clustering is proposed to improve the credibility of the resulting rule base. And the rule base is simplified based on the improved concept of support and confidence. The simulation results show that the boiler combustion optimization results based on the optimal MVs decision model are close to the global optimization results, and the algorithm has low complexity and high stability, which is suitable for online real-time optimization and adaptive updating.