中文摘要：

融合Gamma滤波和PI模型的结构,提出压电陶瓷的动态迟滞模型.借鉴PI模型的结构, 对Gamma滤波的各个环节的输出通过RBF神经网络进行静态非线性变换,并在其后与一个动态神经网络串联, 构成了一种新的神经网络迟滞模型.与静态PI 模型不同,由于新模型中含有动态的类迟滞算子与动态神经网络串联.所以, 所提出的迟滞模型是一种动态的迟滞模型.对压电陶瓷实际测量数据逼近和预测的结果表明,所提出的动态迟滞模型精度高,具有较强的泛化能力.

英文摘要：

The dynamic hysteresis model was proposed by combining Gamma filters and structure PI model. Based on the structure of PI model, the output of Gamma filters was used for input to static neural networks series with a dynamic neural networks. In comparison with PI model the proposed hysteresis model is a dynamic hyststeresis model because of dynamic hystreretic cell in the new model. The approximation and prediction results on pizeocamaic actuator show that the proposed hysteresis model has high precision and good generalization capability.