The bamboo-wood composite container floor (BWCCF) plays an increasingly important role in the transportation area in recent years. However, the conventional mechanical testing methods are conducted in a time-consuming and resource-wasting way. Therefore, this study is aimed to provide a frugal and high-efficiency method to predict the concentrated load of BWCCF, by comparing models with two sets of parameters. First, three artificial neural network (ANN) models were developed by taking the characteristic parameters of the end face extracted by image processing as input and concentrated load as output. Then, the other three ANN models were presented by taking the vertical density profile (VDP) as input. Of the six models, the two ANN models constructed using all characteristic parameters of cross and vertical sections and all VDP parameters had the strongest generalization. The mean absolute percentage errors were determined as 3.393 and 6.196%, respectively, and the absolute percentage errors were all within 10.000%. The result indicates that the designed model has the potential to be a useful, reliable and effective tool for predicting concentrated load.

竹木复合集装箱地板（BWCCF）近年来在交通领域发挥着越来越重要的作用。然而，传统的机械测试方法以耗时且浪费资源的方式进行。因此，本研究旨在通过比较具有两组参数的模型，提供一种节俭且高效的方法来预测 BWCCF 的集中负荷。首先，以图像处理提取的端面特征参数为输入，集中载荷为输出，开发了三种人工神经网络（ANN）模型。然后，通过将垂直密度分布（VDP）作为输入来呈现其他三个 ANN 模型。在六款车型中，使用横截面和垂直截面的所有特征参数以及所有 VDP 参数构建的两个 ANN 模型具有最强的泛化性。平均绝对百分比误差分别确定为 3.393 和 6.196%，绝对百分比误差均在 10.000% 以内。结果表明，所设计的模型有可能成为预测集中荷载的有用、可靠和有效的工具。