本文選用鐘型函數做為輸入函數,將影響降雨量之因素,最高輻射、風向、風速、大氣壓力、溫度、溼度作為六個輸入變數,以建立模糊隸屬函數,來探討苗栗地區降雨量。其建構流程為先將輸入變數値做模糊化的轉換,利用MATLAB中之ANFIS軟體做模糊類神經網路演算建立模糊資料庫,建立資料庫後,再將演算結果反模糊化轉換輸出,以推估其測站之降雨量,與測站之實際降雨量作檢驗,分析評估其學習模擬效果。 降雨量之模組建構以西元2005年至2007年苗栗縣公館國小氣象站和國立苑裡高中氣象站之基本降雨資料,藉由降雨資料來做模糊演算和分析,計算結果可以看出本文所述影響降雨量之六個因素,對其模擬學習階段所推估之降雨量與實際降雨量之平均相對誤差,在苗栗公館國小站約為30.3%,在苑裡高中站約為19.2%。而預測階段所推估之降雨量與實際降雨量之平均相對誤差,在苗栗公館國小站約為74.5%,在苑裡高中站約為59.8%,精度上比模擬學習階段雖較為大,此結果希望可提供未來降雨量預測上參考。
In this research, the six factors affecting rainfall are selected as the input variables. They are the maximum radiation, the wind direction, the wind velocity, atmospheric pressure, temperature and humidity. These variables were considered to have bell-shaped function distribution to set up the membership of fuzzy function to discuss the rainfall of Miaoli. Fuzzification transfers the input data and creates the fuzzy database from the method of the neural nets and its computation program MATLAB/ANFIS. Then the resulting rainfalls were obtained. The estimated rainfall was compared with the observation rainfall of the station .From the comparison, the learning effects were analyzed. Using the typhoon data from the weather stations of Gongguan Elementary School in Miaoli County and National Yuanli Senior High School, a fuzzy calculation and analysis by using rainfall data can be done. For the results in the learning stage, the accuracies of the estimate rainfalls for the weather stations were good and the relative error was about 30.3% in Gongguan Elementary School in Miaoli County, and that was 19.2% in National Yuanli Senior High School. For the results in predicting stage, the accuracies of rainfalls of relative error for weather stations was about 74.5% in Gongguan Elementary School in Miaoli County, and that was 59.8% in National Yuanli Senior High School. The relative error were larger than that of the learning stage. The results of this study can be used as a reference for the rainfall prediction in the future.