一种基于特征提取方法的智能预测算法

An intelligent prediction algorithm based on feature extraction methodes

针对色织生产调度过程中的一类整经轴数预测问题,提出一种整经轴数智能预测算法.首先基于线性特征提取方法(PCA)和非线性特征提取方法(LLE)对影响整经轴数的多维属性参数进行特征提取;然后采用前向神经网络进行整经轴数预测.数值计算结果表明,所提出的方法能满足实际生产过程整经轴数预测的需要.

An intelligent algorithm is proposed to predict the trim-beam number in the colored weaving scheduling process. Linear and nonlinear feature extraction methods including PCA and LLE are used to extract features from high-dimensional properties which are relative to the trim-beam number. Then, neural networks are applied to predict the trim-beam number. Numerical computational results based on practical production data show that the proposed algorithm can satisfy the trim-beam number prediction requirements in the practical manufacturing process.