固体燃料冲压发动机燃速的人工神经网络辨识

由于固体燃料冲压发动机 (SFRJ)结构的特殊性 ,其燃速的预测比较困难。为探讨燃速对于不同飞行工况的依赖性 ,对模拟试验的结果采用 BP人工神经网络进行了辨识 ,并将辨识结果与用最小二乘法辨识的结果进行比较。从辨识的结果来看 ,这种辨识方法具有精度高、处理实验数据迅速、能进行在线辨识等特点 ,能够较好地满足工程应用的要求。此外 ,在利用人工神经网络对类似的大样本系统进行辨识时 ,如果能采用一些数学上的处理技巧 ,可大大加快训练速度

Artificial Neural Network Identification of Burning Rate in Solid Fuel Ramjet

In order to search the dependency of the burning rate on the flying conditions of SFRJ,we identify the results of the simulated experiments by means of BP artificial neural network,and compare the results with those obtained with the help of Least Square Method.It is shown that the presented method of identification features high precision,fast experimental data handling and on-line process identification,and is superior in engineering application.Furthermore,it will greatly enhance the speed of training if some proper mathematic techniques are adopted in the application of artificial neural network to identifying similar big sample system.