基于灰色神经网络模型的强力旋压连杆衬套屈服强度预测

Prediction of yield strength of power spinning connecting rod bushing based on grey neural network model

强力旋压连杆衬套的力学性能与旋压参数之间的关系复杂,难以用较为系统的数学公式进行表达。为科学准确预测连杆衬套的屈服强度,采用灰色系统理论,选取旋压工艺参数作为系统相关因素,建立预测连杆衬套屈服强度的GM（0,3）模型,经过后验差检验,模型精度二级,平均相对误差为3.25%。为提高模型精度,利用RBF神经网络对预测残差进行修正后,将模型精度提高为一级。在将外来数据回代检验时,GM（0,3）＋RBF模型预测结果相对误差在1%左右。研究结果表明：利用灰色理论建立的GM（0,3）＋RBF模型能够较精确的预测连杆衬套屈服强度,且预测能力较强,建模简单快速。

The relationship between mechanical properties and spinning parameters of power spinning connecting rod bushing is complicate,and it is difficult to express in systematic mathematical formula. To predict the yield strength of connecting rod bushing scientifically and accurately,GM（0,3） model was established to predict yield strength of connecting rod bushing with the grey system theory and selecting spinning process parameters as the system relevant factors. The posterior difference test shows that the model precision is in level two and the average relative error is 3. 25%. In order to improve the accuracy of the model,the precision of the model was improved to level one after correcting the predicting residual error by RBF neural network. The relative error of GM（0,3） ＋ RBF model is about 1%when the external data is returned to the test. The research results show that GM（0,3） ＋ RBF model established by the grey theory can predict yield strength of connecting rod bushing more accurately. Besides,its predictive ability is stronger and the model-builder process is simple and fast.