In order to reduce difficulty to obtain accurate in-situ sampling data and enhance correctness of accident diagnosis, a new method of establishing fuzzy Petri net model by transferring a fuzzy fault tree into a fuzzy Petri net model is put forward based on that a fuzzy fault tree and a fuzzy Petri net model have the same characteristics of cause-and-effect relationship which describes occurrence of events, a fuzzy analysis is made to a fault tree based on its corresponding Petri net model. The method can not only estimate all kinds of characteristic values of root event, but also make accident diagnosis by fuzzy inference algorithm of its Petri net model, therefore a tight link between a Petri net model and a fault tree is formed so that accident diagnosis is more suitable for in-site circumstances. The fuzzy set theory is introduced into analytical methods of fault tree where fuzzy numbers represents firing possibilities of events so that difficulty to obtain accurate firing possibilities of bottom events is reduced, errors from statistic data of accidental possibilities are corrected to some extent. The method can put in-site sampling data and experiences of experts together so as to keep analytical results to approximate to engineering fact. In order to reduce difficulty to obtain accurate in-situ sampling data and enhance correctness of accident diagnosis, a new method of establishing fuzzy Petri net model by transferring a fuzzy fault tree into a fuzzy Petri net model is put forward based on that a fuzzy fault tree and a fuzzy Petri net model have the same characteristics of cause-and-effect relationship which describes occurrence of events, a fuzzy analysis is made to a fault tree based on its corresponding Petri net model. The method can not only estimate all kinds of of characteristic values ​​of root event, but also make accident diagnosis by fuzzy inference algorithm of its Petri net model, therefore a tight link between a Petri net model and a fault tree is formed so that accident diagnosis is more suitable for in-site circumstances. fuzzy set theory is introduced into analytical methods of fault tree where fuzzy numbers represents firing possibilities of events so that difficulty to obtain accurate firing possibiliations ties of bottom events is reduced, errors from statistic data of accidental possibilities are are to some extent. The method can put in-site sampling data and experiences of experts together so as to keep analytical results to approximate to engineering fact.