中速磨直吹式制粉系統是火力發電廠鍋爐運行穩定的一個重要環節，其磨機內存煤量能否獲得可靠監測對發電廠的自動控制、經濟運行和安全生產均有十分重要的意義。通過基于工藝機理知識，分析了中速磨直吹式制粉系統筒內存煤量的各相關因素，提
出了利用人工神經網絡進行存煤量軟測量的方法，給出了基于BP 改進算法的回歸神經網絡與延時神經網絡綜合模型，為中速磨存煤量的檢測提供了新方法。離線訓練與計算機仿真結果證實了軟測量方法的可行性。
關鍵詞：中速磨煤機 存煤量 軟測量 改進算法
Abstract：Powdered system of straight blow medium speed mills is one of the important sections in fossil-fired power plants. Reliable monitoring for the quantity of reserving coal in the mills is significant to automation control，economical operation and safe production. Based on the knowledge of technique mechanism，various related factors to the quantity of reserving coal are analyzed. The method of soft-sensing by using artificial neural network is proposed and the integrated model of regressive neural network and time delay neural network based on BP improved algorithm are given. This provided a new method for detecting the reserving coal for medium speed mills. Through off-line training and computerized simulation，the results prove that the soft-sensing method is feasible.
Keywords：Medium speed mill Quantity of reserving coal Soft-sensing Improved algorithm