基于模糊综合评价法的输电线路山火遥感甄别算法

传统输电线路山火遥感甄别算法通常采用亮温值阈值法直接对火点进行甄别，阈值难以准确设定会导致山火的误甄别或漏甄别问题。为实现输电线路山火遥感的准确甄别，在传统火点甄别算法基础上，提出了热点再甄别的思想，即依据亮温值阈值确定出可能的山火热点后，再对热点进行模糊综合评价以准确甄别山火火点。结合对广东省山火发生的时空规律进行研究，确定了导致山火发生的6个致灾因子，提出采用模糊综合评价法，实现输电线路山火的准确甄别。首先利用火险评判等级确定火险决策集及阈值，随后通过6个致灾因子建立因子集并作为输入，构建隶属度函数并设置权重向量，根据最大隶属度原则输出热点的置信度，并与阈值进行比较，判断火点的真伪。通过广东省2005—2016年12年间的山火数据库和该模型于2017年在广东电网的应用实例作为算例，对该模型进行验证。结果表明该模型较传统算法准确率高，可以满足输电线路山火的防治要求。

Remote Sensing Discrimination Algorithm for Transmission Line Forest Fire Based on Fuzzy Comprehensive Evaluation Method

The traditional remote sensing discrimination algorithm for transmission line forest fires usually uses the brightness temperature threshold method to directly perform the fire point discrimination, which makes it difficult to accurately set the threshold value and frequently lead to false discrimination or omissions. In order to realize the accurate remote sensing discrimination of transmission line forest fires, the idea of second discrimination of hot spots is proposed based on the traditional fire point discrimination algorithm, that is, after determining the possible hotspots according to the threshold value of the brightness temperature, the fuzzy comprehensive evaluation is carried out for the hotspots to determine the exact point of fire. According to the research of the time and spatial distribution of forest fires in Guangdong Province, six hazard factors causing the occurrence of forest fires are determined, and a fuzzy comprehensive evaluation method is proposed to realize the accurate discrimination of transmission line forest fires. Firstly, the fire risk decision-making set and threshold are determined according to the fire risk evaluation levels. Then, the factor set is established as input with the six hazard factors, and the membership function is constructed and the weight vector is set. The confidence of the hotspot is output according to the principle of maximum membership degrees, and compared with the threshold to determine the authenticity of the fire point. The model is verified using the forest fire database in Guangdong Province from 2005 to 2016 and by a case study of Guangdong Power Grid in 2017. The results show that the model has higher accuracy than traditional algorithms and meets the prevention and control requirements for transmission line forest fires.