本文提出一种新的神经网络模型，它是由ＲＢＦ（ＲａｄｉａｌＢａｓｉｓＦｕｎｃｔｉｏｎ）函数构成的一种两层网络模型。利用此模型不仅可以进行一般的多类模式的分类识别，还可以进行多目标的旋转不变分类识别，本文对轰炸机（ＰＢ）、战斗机（ＰＦ）和客机（ＰＬ）三种飞机的旋转不变分类进行了计算机模拟，模拟表明这种模型对于多目标的旋转不变分类识别是非常有效的。本文还提出一种实现此模型的光电混合系统。 This paper presents a new neural network model, which is a two-layer network model composed of Radial Basis Function (RBF) functions. This model not only can be used for general multi-class classification and recognition, but also for multi-objective rotation invariant classification and recognition. In this paper, the rotation of three kinds of aircraft, PB, Fighter and Passenger aircraft (PL) The classification is carried out by computer simulation. Simulation shows that the model is very effective for multi-objective rotation invariant classification and recognition. This paper also presents a hybrid optoelectronic system that implements this model.