Radial basis function neural network model based prediction of weld plate distortion due to pulsed metal inert gas welding

Abstract

Welding shrinkage and distortion affect the shape, dimensional accuracy and strength of the finished product. This work concerns the prediction of welding distortion in a pulsed metal inert gas welding (PMIGW) process. Six different types of radial basis function network (RBFN) models have been developed to predict the distortion of welded plates. Six process parameters, namely, pulse voltage, background voltage, pulse duty factor, pulse frequency, wire feed rate and the welding speed, along with the root mean square (RMS) values of two sensor signals, namely, the welding current and the voltage signals, are used as input variables of these models. The angular distortion and the transverse shrinkage of the welded plate are considered as the output variables. Inclusion of sensor signals in the models, as developed in this work, results in better output prediction.     

Possibility of grain size prediction in AA5754 aluminium ingots using neural networks

Abstract

The approach to the grain size prediction in AA5754 Al alloy ingots based on artificial neural networks (ANN) has been used in the present study. The ANN has been trained on data that was measured in the real industrial conditions during the process of direct chill Al ingots casting. A very complex relation between the numerous casting parameters and the microstructure of the ingots justifies the application of neural networks, which are known for mapping complex and non-linear systems. A feed forward ANN model with the resilient back-propagation learning algorithm and weight decay regularisation has been developed to relate the grain size to casting rate, meniscus level, casting temperature, water flow for the metal mould cooling and speed of wire for master alloy addition. The results obtained from the ANN are found to be consistent with the theoretical researches and experience from the foundry.     

Stochastic Optimisation Methods for Biomolecular Structure Prediction

We discuss the use of stochastic optimization methods for biomolecular structure prediction, particularly applied to protein structure prediction and receptor ligand docking. After a brief introduction we give an overview of the dominating physical effects that are important for protein structure prediction and outline our strategy to address this problem. We discuss the strengths and weaknesses of several possible optimization methods, including the stochastic tunneling method. Finally we give examples of applications of this methodology both for protein structure prediction and receptor ligand docking.     

SLAMMOT-SP: Simultaneous SLAMMOT and Scene Prediction

In recent years, SLAMMOT (simultaneous localization, mapping and moving object tracking) has attracted widespread attention in the mobile robot field. This paper proposes a new approach, SLAMMOT-SP, which combines SLAMMOT and scene prediction (SP). It extends the SLAMMOT problem to simultaneous map prediction and moving object trajectory prediction. The robot not only passively collects the data and executes SLAMMOT, but actively predicts the scene. The recursive Bayesian formulation of SLAMMOT-SP is derived for real-time operations. A generalized framework for tracking and predicting the moving objects is also proposed. Simulations and experiments show that the proposed SLAMMOT-SP is effective and can be performed in real-time.     

Capturing a Qualitative Model of Network Performance and Predicting Behavior

This paper describes a method for constructingbehavior models of communication networks. The methodutilizes archived quantitative performance data createdby a network management platform to create a Quantitative/Qualitative (Q²)Dynamic System representation. The Q²representation captures the predominant qualitative(symbolic) states of the network, qualitative inputevents and transitions among the states resulting from these events. Thissymbolic model allows the network manager to understandthe current system behavior, and predict future possiblebehaviors. We evaluated the method on two sets of archive data. The method shows promise foruse in network management, including network monitoring,fault detection, prognostication andavoidance.     

Single Lap Joints with Rounded Adherend Corners: Experimental Results and Strength Prediction

The objective of the present study was to better understand the effect of the change in the geometry of the adherend corners on the stress distribution in single lap joints and, therefore, on the joint strength. Various degrees of rounding were studied and two different types of adhesives were used: one very brittle and another which had a large plastic deformation. Experimental results on the strength of joints with different degrees of rounding are presented. For joints bonded with brittle adhesives, the effect of the rounded adherend corners is larger than that with ductile adhesives. The strength of joints with brittle adhesives with a large radius adherend corner increases by about 40% compared to that with a sharp adherend corner. It is shown that for joints bonded with brittle adhesives, crack propagation occurs for a short period before it grows into catastrophic failure. However, for ductile adhesives, there is large adhesive yielding and small crack propagation before final failure. Another important feature of joints bonded with ductile adhesives is that there may be more than one crack in the adhesive layer before failure. This makes strength predictions more difficult. The second part of the paper presents an approximate method for predicting the strength of joints bonded with brittle and ductile adhesives, with and without adherend corner rounding. The predictions, based on an average value around the singularity, compare well with the experimental results, especially for joints bonded with ductile adhesives.     

Strength Improvement of Adhesively-Bonded Joints Using a Reverse-Bent Geometry

Adhesive bonding of components has become more efficient in recent years due to the developments in adhesive technology, which has resulted in higher peel and shear strengths, and also in allowable ductility up to failure. As a result, fastening and riveting methods are being progressively replaced by adhesive bonding, allowing a big step towards stronger and lighter unions. However, single-lap bonded joints still generate substantial peel and shear stress concentrations at the overlap edges that can be harmful to the structure, especially when using brittle adhesives that do not allow plasticization in these regions. In this work, a numerical and experimental study is performed to evaluate the feasibility of bending the adherends at the ends of the overlap for the strength improvement of single-lap aluminium joints bonded with a brittle and a ductile adhesive. Different combinations of joint eccentricity were tested, including absence of eccentricity, allowing the optimization of the joint. A Finite Element stress and failure analysis in ABAQUS^® was also carried out to provide a better understanding of the bent configuration. Results showed a major advantage of using the proposed modification for the brittle adhesive, but the joints with the ductile adhesive were not much affected by the bending technique.     

Hardness prediction models based on HAZ simulation for in-service welded pipeline steels

Abstract

Welding onto pressurised pipelines that contain flammable fluid to facilitate repairs or branch connections, is a critical procedure with considerable risk to personnel and infrastructure. Limiting the heat input is obviously an important consideration to prevent 'burn-through', but the potential for rapid cooling of the weldment increases its susceptibility to hydrogen assisted cold cracking (HACC). Therefore, one of the most important factors for in-service welding procedure development relates to the increased risk of formation of hard, brittle microstructures in the grain coarsened heat affected zone (GCHAZ) of the weld, microstructures that increase the risk of HACC. The present work has been successful in utilising heat treatment simulations to derive two new hardness prediction models that more accurately predict hardness in the GCHAZ for typical in-service welding applications than a commercially adopted and widely used hardness prediction algorithm. Although it is acknowledged that further work is required to validate the models for a wider range of in-service welding conditions, the performance of the models demonstrates their potential for developing improved in-service welding procedures.