偏最小二乘和ANFIS在飞机燃油油量测量中的应用

分析了飞机燃油箱剩余油量测量系统具有变量多、非线性及变量间存在多重相关性的特点,指出单独采用自适应神经模糊推理系统(Adaptive network-based fuzzy inference system．ANFIS)和偏最小二乘(Partial least- squares,PLS)解决飞机燃油箱剩余油量软测量问题均有各自的局限性。结合上述两种方法的优点,提出了一种基于PLS回归与减法聚类的ANFIS软测量模型。将其应用于预测飞机油箱燃油剩余的建模过程中,得到了比采用PLS回归方法更好的精度,从而证明了该建模方法的有效性。

Partial Least-Squares and ANFIS for Measuring Aircraft Fuel Volume

The measurement system of the aircraft fuel volume has characteristics of high dimensions of inputs,the nonlinearity and the strong correlation among inputs.Using adaptive network-based fuzzy inference system(ANFIS)or partial least-squares(PLS)to construct a model has some limitations. The soft sensor model combined by PLS,substractive clustering and ANFIS is used to measure the air- craft fuel volume during flight.The precision of two types of models using PLS and PLS-ANFIS is com- pared.Results show that the PLS-ANFIS modeling method has better performance.