电动汽车充电站火灾多目标实时检测与预警方法

电动汽车在充电过程中易发生火灾事故，为了提高电动汽车充电站火灾检测实时性，提出一种基于YOLOv4-Tiny-CBAM的电动汽车充电站火焰烟雾多目标实时检测与预警算法：选用YOLOv4-Tiny轻量级网络模型实现在低算力平台流畅运行；引入Kmeans-GA算法重新计算锚框值；引入卷积注意力机制模块（CBAM）以加强网络对火焰烟雾特征提取能力以提升检测精度；将电动汽车充电站监控视频作为模型检测输入源，实现就地端实时检测。实验结果表明：该改进算法模型参数量为6.143M，视频检测FPS值为43，mAP值为86.76%，具有较好的目标连续跟踪能力，满足实时检测的需求，对无人化电动汽车充电站安全运行以及火灾应急处置具有重要意义。

Multi-objective real-time detection and early warning method for electric vehicle charging station fires

To improve the real-time detection of charging station fires, we present a YOLOv4-Tiny-CBAM-based multi-objective real-time flame and smoke detection and warning algorithm for electric vehicle charging stations: The YOLOv4-Tiny lightweight network model is chosen to operate efficiently on a low-power computer platform, and the Kmeans-GA algorithm is introduced. The anchor frame value is recalculated; the Convolutional Attention Mechanism Module (CBAM) is introduced to enhance the network''s ability to extract the flame smoke features to improve the accuracy of detection; and the electric vehicle charging station monitoring video is used as the model detection input source to achieve real-time detection at the local end. The experimental results demonstrate that the improved algorithm has a model parameter number of 6.143M, a video detection FPS value of 43, and a mAP value of 86.76 percent, which has good continuous target tracking capability and satisfies the demand for real-time detection, which is crucial for the safe operation of unmanned electric vehicle charging stations and fire emergency disposal.