Elevator Traffic Pattern Recognition Method Using Particle Swarm K-means Clustering Algorithm

In this article, we introduce an elevator traffic pattern recognition method based on a Particle Swarm Optimization (PSO) enhanced K-means clustering algorithm. The core approach involves performing cluster analysis on raw passenger flow data from the previous week to obtain cluster center coordinates representing different traffic patterns[2]. To handle real-time traffic flow variations, we collect passenger statistics in 5-minute intervals and assign them to corresponding cluster centers using nearest neighbor principles, thereby identifying current traffic modes. The implementation typically includes optimizing initial centroids through PSO to avoid local minima in K-means clustering, followed by calculating Euclidean distances between real-time data points and cluster centers for classification. Additionally, alternative clustering algorithms or machine learning techniques for data prediction can be employed to further optimize our traffic pattern recognition method. Overall, this approach provides valuable insights into elevator traffic patterns and regularities, facilitating more intelligent elevator service operations through algorithmic pattern matching and real-time classification.