预测神经网络控制器及其在禽蛋孵化过程中的应用

Neural Network Predictable Control in the Prosess of Hatching Poultry egg

本文针对常规神经网络控制通常需要辨识网络和控制网络两个网，提出将神经网络用于提前学习系统动态特性，借助于运动方程来获得控制信号的预测控制算法，极大地提高了算法的运算速度。将此算法用于非线性多变量耦合的禽蛋孵化过程，取得了最佳的动、静态特性。

Usually neural network control needs identification network and control network.This prper proposes application of neural network to the prelearning of system's dynamic characteristics. This method acquires predictable control algorithm by using motion equation,thus computing speed is greatly increased.The algorithm is used in the process of hatching poultry egg with non linear and multiple varation .The dynamic and static characteristics are optimum.