u-Рation of the main equipment and expert experience, a new fault knowledge base was established, includingРhaving online fault diagnosis realized for the condensation and feed-water system, and the deep belief neuralРnetwork (DBNN) applied to established a fault diagnosis model. The results show that, the accuracy of thisРdeep belief neural network-based fault diagnosis model can stay at 98%.РKey words fault diagnosis, condensation and feed-water system, fault knowledge base, DBNNРР■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■РР 严 正 声 明РР 接各大数据库通知,我刊已发表的个别稿件重复率超标,而我刊接收稿件时重复率已控制在 15%以下,故Р 此敬告作者:若因作者一稿多投,或因未能妥善保管自己的稿件造成他人在他刊先发,导致在我刊出刊前重复Р 率超标的,本刊将对此稿件做退稿处理,费用概不返还。