A novel deep learning prediction model for concrete dam displacements using interpretable mixed attention mechanism

Dam displacements can effectively reflect its operational status, and thus establishing a reliable displacement prediction model is important for dam health monitoring. The majority of the existing data-driven models, however, focus on static regression relationships, which cannot capture the long-term temporal dependencies and adaptively select the most relevant influencing factors to perform predictions. Moreover, the emerging modeling tools such as machine learning (ML) and deep learning (DL) are mostly black-box models, which makes their physical interpretation challenging and greatly limits their practical engineering applications. To address these issues, this paper proposes an interpretable mixed attention mechanism long short-term memory (MAM-LSTM) model based on an encoder-decoder architecture, which is formulated in two stages. In the encoder stage, a factor attention mechanism is developed to adaptively select the highly influential factors at each time step by referring to the previous hidden state. In the decoder stage, a temporal attention mechanism is introduced to properly extract the key time segments by identifying the relevant hidden states across all the time steps. For interpretation purpose, our emphasis is placed on the quantification and visualization of factor and temporal attention weights. Finally, the effectiveness of the proposed model is verified using monitoring data collected from a real-world dam, where its accuracy is compared to a classical statistical model, conventional ML models, and homogeneous DL models. The comparison demonstrates that the MAM-LSTM model outperforms the other models in most cases. Furthermore, the interpretation of global attention weights confirms the physical rationality of our attention-based model. This work addresses the research gap in interpretable artificial intelligence for dam displacement prediction and delivers a model with both high-accuracy and interpretability.     

Excavator 3D pose estimation using deep learning and hybrid datasets

Earthwork operations are crucial parts of most construction projects. Heavy construction equipment and workers are often required to work in limited workspaces simultaneously. Struck-by accidents resulting from poor worker and equipment interactions account for a large proportion of accidents and fatalities on construction sites. The emerging technologies based on computer vision and artificial intelligence offer an opportunity to enhance construction safety through advanced monitoring utilizing site cameras. A crucial pre-requisite to the development of safety monitoring applications is the ability to identify accurately and localize the position of the equipment and its critical components in 3D space. This study proposes a workflow for excavator 3D pose estimation based on deep learning using RGB images. In the proposed workflow, an articulated 3D digital twin of an excavator is used to generate the necessary data for training a 3D pose estimation model. In addition, a method for generating hybrid datasets (simulation and laboratory) for adapting the 3D pose estimation model for various scenarios with different camera parameters is proposed. Evaluations prove the capability of the workflow in estimating the 3D pose of excavators. The study concludes by discussing the limitations and future research opportunities.     

Model transformation and formal verification for Semantic Web Services composition

Web services composition is becoming more and more important in today’s service oriented business environment. We need to compose services from different providers together to fulfill the business goals that cannot be satisfied by any single service. However, different services often have semantic inconsistencies which may lead to the failure of the services composition. In order to verify the correctness of the Semantic Web Services composition, this paper proposed a composition model of Coloured Petri Nets which is transformed from OWL-S model. This model can express the logical relations among the sub-processes of the services composition explicitly, and verify the correctness of the services composition using formalized methods of Coloured Petri Nets. This paper presented the verification algorithms for the reachability, boundness and semantic consistency of composed services. Furthermore, an example of collaborating design process was given to simulate and execute the model.     

Prediction of rank deficiency in partition of unity-based methods with plane triangular or quadrilateral meshes

In the partition of unity (PU)-based methods, the global approximation is built by multiplying a partition of unity by local approximations. Within this framework, high-order approximations are achieved by directly adopting high-order polynomials as local approximations, and therefore nodes along sides or inside elements, which are usually adopted in the conventional finite element methods, are no more required. However, the PU-based approximation constructed in this way may suffer from rank deficiency due to the linear dependence of the global degrees of freedom. In this paper, the origin of the rank deficiency in the PU-based approximation space is first dissected at an element level, and then an approach to predict the rank deficiency for a mesh is proposed together with the principle of the increase of rank deficiency. Finally, examples are investigated to validate the present approach. The current work indicates such a fact that the rank deficiency is an unrelated issue to the nullity of the global matrix. It can be resolved in its own manner.     

A new boundary condition for energy radiation in covered reservoirs using BEM

In this paper, a new boundary condition accounting for energy radiation at the far end of covered reservoirs is proposed. Using boundary element modelling (BEM), the boundary condition is investigated through analysis of the hydrodynamic pressure within a two-dimensional ice-covered reservoir impounded by a gravity dam. The proposed boundary condition accounts for reservoir bottom absorption effects and the presence of an ice cover at infinity. Seismic excitation is introduced by subjecting the dam and the reservoir to a horizontal harmonic ground motion. The effectiveness of the proposed method and the accuracy of the boundary condition are examined through a parametric study. The boundary condition is shown to be accurate even when placed near the dam upstream face, and the results obtained are in excellent agreement when compared to those from a mathematical model developed by the authors in a previous work. Some fundamental aspects of hydrodynamic pressure within ice-covered reservoirs are also discussed.     

Mercury-promotion of platinum catalysis of dehydrogenation and dehydrochlorination

Mercury in very low concentrations (i.e. when there is up to about one Hg atom in one hundred Pt atoms) promotes the dehydrogenation activity of supported Pt towards cyclohexane and to a lesser extent cyclohexene. However, dehydrochlorination activity towards monochlorocyclohexane is relatively unaffected by Hg additions. Interestingly, cyclohexene is a significant intermediate from dehydrochlorination at high Hg concentrations; consideration is given to reaction mechanisms. The low concentrations at which Hg promotes catalytic activity suggests that it may disrupt the lattices of the Pt crystallites producing surface sites of low coordination and high activity.     

Optimum design,redundancy and reliability of structural systems

The need for structural safety under a variety of loading and accident conditions has focused attention on redundancy, ductility and reliability of structural systems. The concepts of component reserve strength and system residual strength, system reliability and system residual reliability and their application are described. Several different structural configuration examples are illustrated in which component sizes are optimized. Design models for extreme loading and accident conditions for both brittle and ductile models are developed. System design methods are recommended.     

The effect of Zr on the microstructure and tensile properties of hot-extruded Al–Mg2Si composite

This work was carried out to investigate the effect of different amounts of Zr on the microstructure and tensile properties of homogenized and hot extruded Al-15% Mg₂Si composite using optical microscopy and scanning electron microscopy (SEM). The results showed that Zr addition has no significant effect on the morphology of both primary and eutectic Mg₂Si phase in as-cast condition. But, applying homogenizing and extrusion processes changed the morphology of Mg₂Si phases from irregular to a more spherical shape. Further results demonstrated that the average size of primary Mg₂Si decreases with the addition of Zr up to 0.1% from 56 μm to 24 μm in hot-extruded condition. As the mount of Zr increased up to 0.1 wt.%, ultimate tensile strength (UTS) and elongation values were also increased from 160 MPa and 3.2% to 292 MPa and 9.5%, respectively. Fracture surface examinations revealed a transition from brittle fracture mode in as-cast composite to ductile fracture in hot-extruded Zr-modified specimens. This can be attributed to the changes in size and morphology of Mg₂Si intermetallic and porosity content.     

Electrokinetically deposited coating for increasing the service life of partially deteriorated concrete sewers

Microbiological induced corrosion (MIC) is a leading deterioration mechanism in concrete wastewater conveyance systems. The work reported herein focuses on breaking the MIC cycle by preventing colonization of the bacteria responsible for converting hydrogen sulfide to sulfuric acid. Electrokinetics was used to drive an antimicrobial agent (cuprous oxide) into the porous wall surface of a pre-cast concrete pipe. An electric potential difference, applied between the steel reinforcement embedded in the concrete and a copper electrode placed in the coating solution, provides the driving force to the copper ions. Atomic absorption tests performed on the coated specimens were used to determine the percentage of cuprous oxide that penetrated the porous surface and migrated into the hardened concrete matrix. A pilot study conducted on three mock pipe specimens demonstrated that the process is effective on partially corroded and non-corroded pipes. Test data also revealed that the amount of copper that migrated into the concrete matrix is time dependent. Based on preliminary test results a treatment procedure was designed and implemented on a 380 mm diameter commercial pre-cast concrete pipe. The data suggest that the method could be deployed to “immune” new and partially deteriorated steel reinforced concrete pipes against MIC.     

Seq2Seq模型的短期水位预测

为有效预测未来一定时间内的连续水位,提出了基于序列到序列(Seq2Seq) 的短期水位预测模型,并使用一个长短期记忆神经网络(LSTM)作为编码层,将历史水位序列编码为一个上下文向量,使用另一个LSTM 作为解码层,将上下文向量解码来预测目标水位序列。以流溪河为研究对象,针对不同预测长度分别建立水位预测模型,并与LSTM 模型和人工神经网络(ANN)模型进行了对比。结果表明:Seq2Seq 模型对连续6 h、12 h 和24 h 水位预测的纳什效率系数最高分别为0.93、0.90和0.85;当预测长度为6 h 时,LSTM 和Seq2Seq 模型预测结果相似,ANN 模型精度较低;当预测长度为12 h 和24 h 时,Seq2Seq 模型相比LSTM 模型和ANN 模型预测效果更好,收敛速度更快。