论文部分内容阅读
为了智能化预测森林火情,在分析火情传感器的信息参量的基础上,提出了模糊隶属函数与神经网络相融合的预测方法。利用欧洲标准试验火TF1模型的运算分析表明,该方法能有效降低概率在0.5附近的森林火情误报率;进一步引入干扰信息参量后预测时间虽有短暂的延迟,但仍能比较准确地预测火情并输出其概率特征。该文提出的模糊神经网络研究方法对复杂度较高的森林火情传感网络及其预测系统具有较强的实用价值。
In order to intelligently predict the forest fire, the forecasting method based on fuzzy membership function and neural network is proposed based on the analysis of the information parameters of the fire sensor. The computational analysis of the European standard test fire TF1 model shows that the proposed method can effectively reduce the false alarm rate of forest fire with a probability near 0.5. However, the prediction time of the disturbance information parameter is shortened after further introduction of the disturbance information parameters, but it can still predict more accurately Fire and output its probability characteristics. The proposed fuzzy neural network research method is of great practical value for the complex forest fire sensing network and its prediction system.