基于模糊神经网络的升温超塑性成形工艺参数优化

为了解决恒温超塑性成形工艺效率低的问题，本文提出了一种升温超塑性成形的新工艺方法，并用模糊神经网络对升温超塑性胀形成形的工艺参数进行优化，得出了优化后的TC4钛合金升温超塑性胀形成形工艺的加载曲线，利用优化后的成形工艺参数进行了某航空器用TC4钛合金薄板盒形件的升温超塑性胀形成形试验。试验结果表明：与恒温超塑性成形相比，此工艺可以显著地缩短成形时间，在本文试验条件下每个零件可以缩短成形时间10min，并可改善材料的成形性，获得厚度分布均匀（厚度分布不均匀率〈8％）的成形零件。基于模糊神经网络方法进行成形工艺参数优化的升温超塑性成形方法可在实际生产中应用。

Optimization of Technological Parameters in Calescent Superplastic Forming Based on Fuzzy Neural Network

To improve the productivity of the superplastic forming technology, a calescent superplastic forming(CSPF)technological method is proposed. Its technological parameters by using fuzzy neural network (FNN) are optimized and the optimized loading curve of TC4 sheet is obtained. Meanwhile, the calescent superplastic bulge forming of TC4 sheet is tested by using optimized technological parameters for parts of aerostat. Compared with constant temperature superplastic forming (SPF),the CSPF may save processing time (forming time be shorten 10 min), which can also improve the material formability so that the nonuniformity of wall thickness is less than 8%. Moreover, the CSPF may obtain the ideal microstructure of parts. Result show that the CSPF based on FNN optimized technological parameters is suitable in manufacture.