A Fuzzy Inference System for Forest Fire Prediction Based on Rechargeable Wireless Sensor Networks

摘要

High incidence and destructiveness of forest fire determine the importance of forest fire prediction or early detection. As the most important factors, local weather observations, environments and human behaviors' characteristics are found highly correlated to forest fire occurrence. Therefore, we introduce a set of fuzzifkation to assess fire risk in the study area and establish a quantitative potential fire risk scheme. These factors converted to triangular fuzzy numbers are calculated by fuzzift'cation and output fire rating level. These results can be utilized as a strategic planning tool to deal with broad-scale fire hazard concerns. At the same time, the wireless sensor network's technology is utilized for collecting 24-hour weather data continuously, which provides a high chance to reflect accurately the status of forest environment. The fuzzy reasoning system is evaluated for Nanjing City region, the capital of Jiangsu Province, China. Depending on the system, we can obtain in which days the high possibility of forest fires danger lies and special attention need to be paid to forest fire prevention for forest guards.