结构的边界与材料分布不确定性在结构动力学优化中不可忽略。本文关注考虑结构中不同材料组份之间扩散界面宽度的不确定性对结构动力学性能的影响和及其鲁棒性拓扑优化方法。研究基于EOLE展开和混沌多项式的结构动力学特性的扩散界面不确定性传播分析方法。基于相场模型的扩散界面描述方法（相场方法示意图如图1所示）,给出不含扩散界面的多相材料初始设计,并定义相场演化模型中的扩散界面宽度等参数信息,通过求解包含扩散项的Allen-Cahn方程更新相场函数,得到包含扩散界面的多材料结构分布。建立基于相场模型的结构动力学鲁棒性多目标拓扑优化模型。结合非侵入混沌多项式展开技术,给出确定性灵敏度分析的结构动力学性能的特性期望与方差的灵敏度算法。

Accurate prediction of the performance of tunnel boring machines (TBMs) is important to safe and efficient tunneling. Traditional prediction method of TBM performance is restricted by the lack of sufficient rock mass data and the prediction accuracy is low. A comprehensive management and preprocessing framework of TBM big operational data is proposed in this paper, after which a new prediction method of TBM performance is proposed. The effectiveness of the proposed method is tested on 4 different tunnels. Prediction results show that the big operational data based method can predict TBM performance with high accuracy, and in most cases, the random forest algorithm generates better prediction results than the long-short term memory (LSTM) neural network. The data imbalance phenomenon leads to the emergence of big prediction errors on certain operational segments, highlighting the necessity of adopting new sampling methods to create a more balanced dataset, and incremental learning methods to update the prediction model timely. © 2020 IEEE.

Multi-relational data (e.g., product design knowledge graph) learning has attracted great attention in the various research areas recently. Motivated by the assumption that the entities have appropriate relations with respect to their belonging categories (e.g., two entities in the (process parameter) and the (machine unit) categories respectively likely bear upon the relationship of (belong to)), this paper proposes a tensor factorization approach for multi-relational data in a supervised way from a Bayesian perspective, referred as supervised Bayesian tensor factorization (SetF). The proposed SetF is formulated as a joint optimization framework of probabilistic inference and Ε-insensitive support vector regression. The generative-discriminative nature of the proposed SetF is able to discover the latent representation of multi-relational data under the principle of the max-margin learning. The interplay between entities, relations, and the categories (entities involved) yields more predictive latent representations that are particularly appropriate for discriminant analysis. We conduct a set of experimental comparisons in terms of multi-relational learning tasks over several benchmark datasets, and also over antenna arrays design parameter relation analysis, which demonstrate the effectiveness of the proposed method. © Springer Nature Singapore Pte Ltd. 2018.

Load prediction of tunnel boring machines (TBMs) is crucial for the design and safe operation of these complex engineering systems. However, to date, studies have mostly used only geological data, but the operation of TBMs also has an important effect on the load, especially its dynamic behavior. With the development of measurement techniques, large amounts of operation data are obtained during tunnel excavation. Mining these heterogeneous in-situ data, including geological data and operation data, is expected to improve the prediction accuracy and to realize dynamic predictions of the load. In this paper, a dynamic load prediction approach is proposed based on heterogeneous in-situ data and a data-driven technique. In this approach, the integration of heterogeneous in situ data is conducted as follows: i) the geological data are extended to match the scale of the operation data using an interpolation method; ii) the categorical data and numerical data are fused through a proposed encoding method; and iii) the geological data are combined with the operation data according to the location of each operation datum. A data-driven technique, Random forest, is used to construct the prediction model based on the integrated heterogeneous in-situ data. The approach is applied to a collection of heterogeneous in-situ TBM data from a tunnel in China, and the results indicate that the approach can not only accurately predict the dynamic behaviour of the load but can also precisely estimate the statistical characteristics of the load. This work also highlights the applicability and potential of data-driven techniques in the design and analysis of other complex engineering systems similar to TBMs.

Load prediction of tunnel boring machines (TBMs) is crucial for the design and safe operation of these complex engineering systems. However, to date, studies have mostly used only geological data, but the operation of TBMs also has an important effect on the load, especially its dynamic behavior. With the development of measurement techniques, large amounts of operation data are obtained during tunnel excavation. Mining these heterogeneous in-situ data, including geological data and operation data, is expected to improve the prediction accuracy and to realize dynamic predictions of the load. In this paper, a dynamic load prediction approach is proposed based on heterogeneous in-situ data and a data-driven technique. In this approach, the integration of heterogeneous in situ data is conducted as follows: i) the geological data are extended to match the scale of the operation data using an interpolation method; ii) the categorical data and numerical data are fused through a proposed encoding method; and iii) the geological data are combined with the operation data according to the location of each operation datum. A data-driven technique, Random forest, is used to construct the prediction model based on the integrated heterogeneous in-situ data. The approach is applied to a collection of heterogeneous in-situ TBM data from a tunnel in China, and the results indicate that the approach can not only accurately predict the dynamic behaviour of the load but can also precisely estimate the statistical characteristics of the load. This work also highlights the applicability and potential of data-driven techniques in the design and analysis of other complex engineering systems similar to TBMs.

The hoist ropes play a vital role in excavators and have undergone a considerable research in the last decade. Although much research has been implemented in this two ways to enhance the performance of the excavator, it can not be denied that the influences, of different, modeling, methods on the hoist rope are always ignored. In this study, dynamic simulation analysis based on virtual prototyping has been proposed to WK-55, m3excavator, whose hoist rope model is established with the macro command and the Adams/cable module. The results of dipper handle show that the maximum displacement is 1.53, mm with macro command method, which occurs in 7.8, s. However, the maximum displacement is 0.63, mm which occurs in 7.6, s of the Adams/cable modeling method. the dynamic and static performance of the excavator was reflected more factually by Adams/cable method. © Springer Nature Singapore Pte Ltd. 2018.