针对偏转板伺服阀射流盘组件两腔恢复压力不对称和一致性差的问题，建立了考虑射流盘尺寸和形位误差时的流场仿真模型，采用多元线性回归分析方法研究了射流盘的形状因素与压力特性之间的关系; 通过神经网络算法实现了不同尺寸和形位误差组合下的射流盘组件两腔恢复压力的预测，并研究了导致两腔压差超差的形状因素分布情况。结果表明: 恢复压力的主要影响因素包括劈尖宽度、射流口宽度、射流盘厚度、接收腔圆角以及劈尖对称度; 两腔压差的主要影响因素为劈尖对称度、射流口垂直度、接收腔圆角以及内角对称度; 射流盘的两腔压差大小呈正态分布规律，压差超差主要是多个因素组合引起，但即使所有形状因素符合设计要求，也可能出现压差超差。实验结果与理论结果相符。

For the asymmetry and poor consistency of recovery pressures in the two receiving cavities of deflector jet servo valve，flow field simulation model considering the size and shape error of the jet-pan was established and the relationship between the shape factors of the jet-pan and the pressure characteristics was studied by using multi-ple linear regression analysis． The neural network algorithm was used to predict the recovery pressure of jet-pan components under different size and shape error combinations，and the shape factors distribution which cause the pressure difference excess was studied as well． Ｒesearch shows that the main factors affecting the recovery pressure include the width of wedge，the width of jet orifice，the thickness of jet-pan，the round corners of receiving cavities and the symmetry of wedge． And the main factors of the pressure difference between the two cavities are the symme-try of the wedge，the verticality of the jet orifice，the symmetry of round corners and the inner angle of the receiving cavities． The pressure difference between the two cavities is in normal distribution． The pressure difference excess is caused by the combination of several factors，but even if all the shape factors meet the design requirements，the pressure difference excess may occur． Experimental results agree with the theoretical results．