【摘要】 根据灰色系统理论处理贫信息系统的优势,以及神经网络学习和自适应的优点,将灰色神经网络组合算法应用于混凝土结构的徐变预测中。利用GM(1,1)模型和BP人工神经网络,建立灰色新陈代谢短期组合预测模型和长期组合预测模型。该组合模型既克服了原始数据少,数据波动性大对预测精度的影响,也增强了预测的自适应性。通过自密实预应力混凝土梁长期变形试验结果的算例分析,表明短期和长期组合模型的预测结果均与试验结果吻合良好,该模型可以作为混凝土结构徐变预测的有效工具。更多还原

【Abstract】 Based on the predominance of grey system models to deal with the problems of the systems that characterized by poor information,and the advantage of learn and self-adaptability of neural networks,the hybrid gray-neural network algorithms was applied to the creep prediction of concrete structures.Using GM(1,1) model and BP artificial neural networks,a short-term and a long-term combination prediction model was established.These models overcome the influences by little raw data and high data fluctuation precision of prediction,and also enhance the self-adaptability.Through analysis results of long-term deformation self-compacting prestreesed concrete beams,both two models show good agreement with experimental data.The models are effective tools for creep prediction in concrete structures.更多还原