基于BP神经网络的体外脱钙牛股骨松质骨蠕变力学性能的预测

Predicting Model for Tensile Properties of Bovine Femoral Cancellous Bone Decalcified in Vitro Based on BP Neural Network

本文采用体外脱钙方法对牛股骨松质骨进行了不同程度的骨质疏松模拟,并开展了单拉蠕变试验获得了不同脱钙程度牛股骨松质骨的蠕变力学性能.试验结果发现,由于松质骨的高度各向异性、不均匀性及骨小梁分布结构的差异性,Norton幂律模型无法完全描述试样蠕变性能.基于本文试验数据,结合BP神经网络,考虑体外脱钙程度及松质骨的分形维数和孔隙率,对牛股骨松质骨的蠕变性能的进行了预测.结果表明,BP神经网络预测数据与试验数据吻合良好,相关度达到0.9992.

The effect of different degree of decalcification on the tensile properties of bovine femoral cancellous bone was studied by using in vitro decalcification method to simulate osteoporosis.The results show that decalcification could affect the mechanical properties of cancellous bone by affecting the fractal dimension of cancellous bone.Based on BP neural network,the tensile properties of bovine femoral cancellous bone were predicted by considering the degree of decalcification in vitro and the fractal dimension of cancellous bone.The results show that the BP neural network prediction data is in good agreement with the test data,and the correlation is up to 0.9936.