Abstract: Due to the complexity of foundation pit supporting structure, frequent occurance of collapse accidents and applicating limitation of traditional fault tree analysis(FTA) methodinduced by probability and accuracy logical relationship, a method for eveluating the foundation pit collapse probability based on T-S fuzzy fault tree is proposed. This method uses fuzzy numbers to describe the probability and fault state of each bottom event, which sloves the problems suffered by traditional fault tree that relies too much on accurate probability and fails to take the intermediate fault state of events into conisderation. At the same time, T-S model is used to replace the traditional logic and/or relationship, which reflects the fuzziness of fault mechanism and event connection, reducing the difficulty of building fault tree. This method can use the actual failure state or the fuzzy probability of the bottom event two different ways to calculate the collapse possibility of foundation pit, which can also guide the risk control work according to the analysis result of bottom event importance. Case analysis shows that this method is more suitable for engineering practice than traditional fault tree. It can scientifically and accurately evaluate the possibility of collapse and determine the key risk factors which can be an effective tool for evaluating the possibility of foundation pit collapse.

Key words: T-S fuzzy fault tree, foundation pit, collapse possibility, fuzzy number, T-S model