摘要
为了提高小样本的预测精度,提出了一种基于量子行为粒子群优化与数据融合算法的灰色融合预测模型。首先从两个方面改进了GM(1,1)模型,对原始序列进行了幂函数变换,并采用量子行为粒子群优化算法实现了参数的优化。然后提出了多次建模的策略,利用原始序列的不同分量分别建立GM(1,1)改进模型进行预测,将各次预测值进行融合得到最终结果。最后用该模型进行软件故障预测,结果表明其相对误差在3%以内,适用于平滑性较差和高增长的序列预测。
To improve the prediction accuracy of small sample, a grey fusion prediction model based on QPSO and data fusion algorithm is presented. First the GM （1,1） is improved from two aspects, the original sequence is transformed by power function, and the parameter is optimized by QPSO. Then the multiple modeling strategy is presented, using different components of the original sequence to modeling improved GM （ 1,1 ） and make a prediction, the predicted values are fused to get the final result. Finally, it makes a forecast of software faults using the model, the result shows the relative error is within 3%, applying to the forecast of smoothness and high growth sequence.
出处
《信息技术》
2015年第7期161-164,169,共5页
Information Technology
