查詢結果分析
來源資料
頁籤選單縮合
題名 | 類神經網路應用於分級泌乳牛乳房炎感染等級之研究=Study of Mastitis Grading of Dairy Cows in Quarter Milk using Neural Network |
---|---|
作者 | 連振昌; 萬一怒; | 書刊名 | 中華農學會報 |
卷期 | 5:1 2004.02[民93.02] |
頁次 | 頁53-64 |
分類號 | 437.37 |
關鍵詞 | 類神經網路; 乳房炎; 體細胞數; 導電度; Neural network; Mastitis; Somatic cell count; Electrical conductivity; |
語文 | 中文(Chinese) |
中文摘要 | 泌乳牛乳房炎的早期診斷對乳房炎的防治是相當重要,本研究目的為利用三層的倒傳遞類神經網路對泌乳牛乳房炎的感染程度作分級判定,採集泌乳牛的分房前乳樣品,量測其體細胞數及導電度,依據牛乳體細胞數多寡將泌乳牛乳房炎感染等級分類為健康分房、輕微感染分房及嚴重感染分房三個等級,除了量測分房乳導電度外另定義三個與其相關之參數:EC、 ECR、 ED及EDR;類神經網路的輸入層以 EC、ECR、ED及EDR作為輸入,隱藏層使用 6 個神經元,而輸出層使用 3 個神經元,其輸出值對應健康分房、輕微感染分房及嚴重感染分房三個等級,類神經網路先以訓練組樣品進行訓練,再以驗證組樣品作測試,由分析結果可得知;對訓練組樣品進行乳房炎感染等級的分級檢測時,其健康分房的分級正確率為 99.4 ﹪,輕微感染分房的分級正確率為 66.7 ﹪,嚴重感染分房的分級正確率為 63.6 ﹪,整體分級的正確率為 94.9﹪;分析結果將有助於分房乳線上導電度檢測泌乳牛乳房炎系統的設計參考。 |
英文摘要 | Mastitis of dairy cows being identified at early stage is very important for the prevention and treatment of mastitis. The purpose of this study is to detect and to grade the dairy cows mastitis in quarter milk using a backpropagation neural network which consisits of input layer, hidden layer and output layer. The somatic cell count (SCC) and electrical conductivity (EC) of the collected quarter foremilk of dairy cows were measured. Based on the scale of SCC in foremilk, the mastitis levels of quarters were classified into three classes: healthy, lightly infected, and seriously infected levels. In addition to the electrical conductivity of quarter milk, three relative parameters - the electrical conductivity ratio (ECR), electrical conductivity difference (ED) and electrical conductivity difference ratio (EDR), were defined. These four parameters were used as the input pattern of the neural network. The output of neural network corresponded to the three mastitis levels. The neural network was trained with training sample set and followed by validation test with validation sample set. The analysis results for training sample set revealed a 99.4% of correct rate in the healthy quarter, 66.7% of correct rate in the slightly infected quarter, and 63.6% of correct rate in the seriously infected quarter. The total correct of quarters is 94.9%. In conclusion, this study will provide a good suggestion for developing an online EC measurement system for dairy cow mastitis inspection. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。