RLDEAO optimized of air quality data clustering analysis(RLDEAO优化的空气质量数据聚类分析)

Abstract

Aiming at the problems of high randomness, low clustering accuracy and multiple central points in clustering the same air quality data by traditional clustering methods due to different choice of initial points, a K-means complementary iterative air quality data clustering analysis method optimized by RLDEAO is proposed. Although aquila optimizer (AO) has a strong exploration ability, and its performance is not easily affected by different choice of initial points, it is easy to fall into local optimization. In view of this deficiency, we propose the adaptive dimension-by-dimension keyhole imaging reverse learning strategy, Levy flight combined with stagnation perturbation strategy and mutation evolution of the survival of the fittest to improve the search performance of the algorithm, thus avoiding local optimization; Secondly, a weighted maximum minimum distance product (WMMP) is designed to calculate the cluster center point, which can reflect the importance of each feature in the data and play a good role to improve the clustering results; Finally, RLDEAO and WMMP are combined to optimize K-means complementary iteration. Through clustering tests on multiple data sets, the convergence curves and evaluation indicators of the observed results show that the convergence accuracy and clustering effect of RLDEAO-KMC are better than AO-KMC, FCM, KMC and KMC++. It can be seen that RLDEAO-KMC can cluster information more efficiently and make better prediction and response.(对空气质量数据进行聚类，传统聚类方法因受初始点的影响，存在随机性高、聚类精度低以及多个中心点出现在同一簇中的问题，为此提出了一种反向学习差分进化天鹰优化器（RLDEAO）优化的K-means互补迭代空气质量数据聚类方法。天鹰优化器（aquila optimizer，AO）算法具有很强的探索能力，不易受初始点的影响且更易实现，但易陷入局部最优。基于自适应逐维小孔成像反向学习策略、停滞扰动结合莱维飞行策略以及生物进化策略等改进思想，对AO算法进行了改进，有效提高了搜索性能，避免了局部最优；在求取聚类中心点时，设计了一种加权最大最小距离积法（weighted maximum minimum distance product，WMMP），能反映各特征的重要性，对改进聚类结果作用良好；将RLDEAO与WMMP相结合优化K-means互补迭代，提高了搜索速率和搜索精度。通过在多个数据集上的聚类测试，发现RLDEAO-KMC算法的收敛精度和聚类效果较AO-KMC、FCM、KMC、KMC++算法更优。可知，RLDEAO-KMC算法可以更高效地对空气质量数据进行聚类分析，有针对性地做出预测和应对。)

Keywords

• k-means clustering algorithm(k-means聚类算法)
• aquila optimizer (ao)(天鹰优化器（ao）)
• weighted maximum minimum distance method(加权最大最小距离积法)