为确认燃烧调整过程中影响燃气轮机运行状态的主导因素，首先将影响运行状态的天然气压力、天然气温度、压气机排气温度、压气机进口温度等14个参数作为输入变量，将表征燃气轮机运行状态的功率、燃烧室、加速度、NOx质量浓度作为输出变量，建立粒子群算法优化的Elman神经网络模型，得到隐含层与输入层、输出层之间的连接权值；然后利用Olden方法处理神经网络的连接权值，获得各因素对燃气轮机运行状态影响显著性的量化值表达式，建立了燃烧调整过程中燃气轮机运行状态影响因素显著性分析的方法；最后结合燃气轮机运行数据进行计算分析。结果表明：燃气轮机的运行状态主要受排气温度、预混气压力及流量、值班气流量、压气机进口差压以及进气导流叶片开度5个因素的影响，并且燃烧调整过程中需要统筹调整输入参数，以保证燃气轮机燃烧始终处于稳定、低NOx排放区域。

To confirm the leading factors affecting the operation status of gas turbine during combustion adjustment, firstly, 14 parameters, such as natural gas pressure, natural gas temperature, compressor exhaust temperature and compressor inlet temperature and so on, which affect the operating state, are taken as input variables, and the power, combustion chamber acceleration and NOx mass concentration that characterize the operating state of the gas turbine are taken as output variables to establish an Elman neural network model optimized by particle swarm optimization algorithm and obtain the connection weights between the hidden layer, the input layer and the output layer; then, Olden method is used to deal with the connection weights of the neural network to obtain the quantitative value expression of the significance of each factor on the operation status of the gas turbine, and a method for the significance analysis of the factors affecting the operation status of the gas turbine in the combustion adjustment process is established; finally, combined with the operation data of the gas turbine, the calculation and analysis are carried out. The results show that the operation status of the gas turbine is mainly affected by five factors, such as exhaust temperature, premixed gas pressure and flow, duty gas flow, compressor inlet differential pressure and inlet guide vane opening. In addition, the input parameters need to be adjusted as a whole in the combustion adjustment process to ensure that the gas turbine combustion is always in a stable and low NOx emission area.

TK14