针对传统均值标准差阈值法不能用于变工况故障预警这一问题，以三轴燃气轮机的性能仿真模型为基础，提出了基于Bootstrap方法的气路参数波动比阈值计算模型，对三轴燃气轮机0.8以及1.0工况下的低压涡轮出口温度T6进行故障预警研究，并与传统均值标准差阈值法进行预警结果对比。结果表明：该计算模型用于预警阈值计算的样本量更丰富、具有更高的可信度；其得到的预警阈值单值图，在燃气轮机与样本数据运行情况较为一致时可作为燃气轮机的固有属性，对于变工况故障预警及运行情况较为相近的燃气轮机故障预警同样适用；预警阈值偏差仅为-0.015%，相较于传统的动态阈值法减少了计算量。

In view of the probelm that the traditional mean standard deviation threshold method cannot be applied to fault warning under variable working conditions, based on the performance simulation model of threeshaft gas turbine, a calculation model of gas path parameter fluctuation ratio threshold value by the Bootstrap method was proposed. With this calculation model, failure warning research was conducted on the low pressure turbine outlet temperature T6 of three shaft gas turbine under 0.8 and 1.0 working conditions, and the warning results were compared with the traditional mean standard deviation threshold method. The results show that the calculation model has the following three advantages: the sample size for early warning threshold calculation is more abundant and has higher reliability; the single value diagram of early warning threshold obtained by the calculation model can be used as the inherent attribute of the gas turbine when the operation condition of the gas turbine is relatively consistent with the sample data. It is also applicable to the fault warning under variable working conditions and the fault early warning of gas turbines with similar operation conditions; the deviation of early warning threshold is only -0.015%, compared with the traditional dynamic threshold method, the calculation amount is reduced.

TK14