论文部分内容阅读
为避免反传学习(BP)算法易于落入局部极小点,该文提出一种基于新填充函数的小波神经网络全局优化学习算法,用来解决连铸连轧过程的产品质量建模问题.该过程很复杂,影响其产品性能的因素很多,物理模型难以建立.该文以小波神经网络为模型,建立连铸连轧产品质量与其化学成分和轧制参数之间的复杂非线性模型.该模型用来对板材产品的断裂延伸率、屈服强度等质量性能指标进行预测.数值实验表明:所建立的模型拟合与校验命中率较高,能够较好地预测产品的物理性能.
In order to avoid the BP algorithm falling into the local minimum easily, this paper proposes a global optimization learning algorithm based on the new filling function to solve the problem of product quality modeling in the continuous casting and rolling process. The process is very complex, many factors affect the performance of its products, physical model is difficult to establish.In this paper, the wavelet neural network as a model to establish the continuous casting and rolling product quality and its chemical composition and rolling parameters between the complex nonlinear model. The model is used to predict the quality of sheet products, such as elongation at break, yield strength, etc. The numerical experiments show that the established model has a high probability of fitting and checking, and can predict the physical properties of the product well.