基于热力模型的燃气轮机气路故障预测诊断研究综述

Review of Gas Turbine Gas-path Fault Diagnosis and Prognosis Based on Thermodynamic Model

为避免失修和过修,提高设备的可靠性和可用性,降低燃气轮机机组运维成本,用户宜采用预知维修策略。气路诊断是一种对正在演变或即将发生的恶化情况发布早期预警信息的有效技术。经过多年发展,气路诊断已经取得了许多基于燃气轮机(准)稳态的诊断算法理论成果,但尚未形成一个完整的科学体系。现今燃气轮机越来越需在电网支持模式下更灵活地运行(包括频繁变工况及瞬态加减载)。在此操作条件下,其使用寿命会比在基本负荷(准)稳态运行时消耗更快,因此,迫切需要考虑瞬态变工况下气路故障诊断与预测的新理论新方法。针对瞬态变工况下气路故障预测诊断的基础问题,提出了瞬态变工况下燃气轮机自适应气路故障预测诊断方法研究路线,包括研究可变几何部件及瞬态变工况运行模式的自适应动态热力建模方法;利用机组可测气路参数,研究基于部件特性线形状内在非线性自适应的气路诊断方法,实时诊断得到各主要部件健康参数;最后提出一种融合各主要部件健康参数的多维度时序预测方法,为实现复杂强非线性热力系统故障诊断与预测提出新方法。

To avoid disrepair and over-repair, improve equipment reliability and availability while maximizing service life and reduce O&M costs, users should adopt predictive maintenance strategies. Gas-path diagnosis is an effective technical means of disseminating early warning information for evolving or impending deterioration. After years of development, many diagnosis methods based on gas turbine steady state/quasi-steady state were obtained, but no complete scientific system was formed yet. Today’s gas turbines are increasingly required to operate more flexibly in grid support modes(including frequent change conditions and transient loading/unloading operating modes). Under such operating conditions, the service life of the gas turbine will be consumed faster than that during base load steady-state/quasi-steady-state operation. Therefore, it is urgent to consider new theories and new methods of gas-path diagnosis under transient conditions. This project studied the fundamental problems in the gas-path fault predictive diagnosis under transient conditions. First, an adaptive dynamic performance modeling method with variable geometry component was proposed. Using the measurable gas-path parameters of the unit, the gas-path diagnosis method based on the intrinsic nonlinear adaptation of the shape of the component characteristic maps was studied, to obtain the main components health parameters real-time. Finally, a multi-dimensional time series prediction method was proposed based on the fusion of the diagnostic information of the main components.