Prediction Accuracy Of Stress Research Articles

Deep Learning-Based Prediction of Stress and Strain Maps in Arterial Walls for Improved Cardiovascular Risk Assessment

Conducting computational stress-strain analysis using finite element methods (FEM) is a common approach when dealing with the complex geometries of atherosclerosis, which is a leading cause of global mortality and complex cardiovascular disease. The considerable expense linked to FEM analysis encourages the substitution of FEM with a considerably faster data-driven machine learning (ML) approach. This study investigated the potential of end-to-end deep learning tools as a more effective substitute for FEM in predicting stress-strain fields within 2D cross sections of arterial walls. We first proposed a U-Net-based fully convolutional neural network (CNN) to predict the von Mises stress and strain distribution based on the spatial arrangement of calcification within arterial wall cross-sections. Further, we developed a conditional generative adversarial network (cGAN) to enhance, particularly from the perceptual perspective, the prediction accuracy of stress and strain field maps for arterial walls with various calcification quantities and spatial configurations. On top of U-Net and cGAN, we also proposed their ensemble approaches to improve the prediction accuracy of field maps further. Our dataset, consisting of input and output images, was generated by implementing boundary conditions and extracting stress-strain field maps. The trained U-Net models can accurately predict von Mises stress and strain fields, with structural similarity index scores (SSIM) of 0.854 and 0.830 and mean squared errors of 0.017 and 0.018 for stress and strain, respectively, on a reserved test set. Meanwhile, the cGAN models in a combination of ensemble and transfer learning techniques demonstrate high accuracy in predicting von Mises stress and strain fields, as evidenced by SSIM scores of 0.890 for stress and 0.803 for strain. Additionally, mean squared errors of 0.008 for stress and 0.017 for strain further support the model’s performance on a designated test set. Overall, this study developed a surrogate model for finite element analysis, which can accurately and efficiently predict stress-strain fields of arterial walls regardless of complex geometries and boundary conditions.

Medically-oriented design for explainable AI for stress prediction from physiological measurements

BackgroundIn the last decade, a lot of attention has been given to develop artificial intelligence (AI) solutions for mental health using machine learning. To build trust in AI applications, it is crucial for AI systems to provide for practitioners and patients the reasons behind the AI decisions. This is referred to as Explainable AI. While there has been significant progress in developing stress prediction models, little work has been done to develop explainable AI for mental health.MethodsIn this work, we address this gap by designing an explanatory AI report for stress prediction from wearable sensors. Because medical practitioners and patients are likely to be familiar with blood test reports, we modeled the look and feel of the explanatory AI on those of a standard blood test report. The report includes stress prediction and the physiological signals related to stressful episodes. In addition to the new design for explaining AI in mental health, the work includes the following contributions: Methods to automatically generate different components of the report, an approach for evaluating and validating the accuracies of the explanations, and a collection of ground truth of relationships between physiological measurements and stress prediction.ResultsTest results showed that the explanations were consistent with ground truth. The reference intervals for stress versus non-stress were quite distinctive with little variation. In addition to the quantitative evaluations, a qualitative survey, conducted by three expert psychiatrists confirmed the usefulness of the explanation report in understanding the different aspects of the AI system.ConclusionIn this work, we have provided a new design for explainable AI used in stress prediction based on physiological measurements. Based on the report, users and medical practitioners can determine what biological features have the most impact on the prediction of stress in addition to any health-related abnormalities. The effectiveness of the explainable AI report was evaluated using a quantitative and a qualitative assessment. The stress prediction accuracy was shown to be comparable to state-of-the-art. The contributions of each physiological signal to the stress prediction was shown to correlate with ground truth. In addition to these quantitative evaluations, a qualitative survey with psychiatrists confirmed the confidence and effectiveness of the explanation report in the stress made by the AI system. Future work includes the addition of more explanatory features related to other emotional states of the patient, such as sadness, relaxation, anxiousness, or happiness.

A mixed inverse differential quadrature method for static analysis of constant- and variable-stiffness laminated beams based on Hellinger-Reissner mixed variational formulation

Increasing applications of laminated composite structures necessitate the development of equivalent single layer (ESL) models that can achieve similar accuracy but are more computationally efficient than 3D or layer-wise models. Most ESL displacement-based models do not guarantee interfacial continuity of shear stresses within laminates. A possible remedy is the enforcement of interlaminar equilibrium in variational formulations, for example, in the framework of the Hellinger-Reissner variational principle, leading to a mixed force/displacement model. In this paper, the governing equations for bending and stretching of laminated beams, comprising only seven stress resultants and two displacement functionals, are obtained using global fifth-order and a local linear zigzag kinematics. As a strong-form solution technique, the differential quadrature method (DQM) is an efficient tool which can provide excellent convergence with relatively few number of grid points. However, in dealing with high-order differential equations, the conventional DQM can incur considerable errors due to the nature of numerical differentiation. Therefore, a mixed inverse differential quadrature method (iDQM) is proposed herein to solve the governing fourth-order differential equations for bending and stretching of laminated beams. This approach involves approximating the first derivatives of functional unknowns, thereby reducing the order of differentiation being performed. Using a non-uniform Chebychev-Gauss-Lobatto grid point profile, numerical results show that the accuracy of stress predictions is improved by using iDQM compared to DQM. In addition, the Cauchy’s equilibrium condition is satisfied more accurately by iDQM, especially in the vicinity of boundaries.

Finite element simulation of welding residual stresses in martensitic steel pipes

Finite element (FE) simulations of the welding of two high grade steel pipes are described. The first is a P91 steel pipe welded with a similar P91 weld consumable, and the second is a P92 steel pipe welded with dissimilar nickel–chromium based weld consumables. Both welds are multipass circumferential butt welds, having 73 weld beads in the P91 pipe and 36 beads in the P92 pipe. Since the pipes and welds are symmetric around their axes, the FE simulations are axisymmetric, allowing high FE mesh refinement and residual stress prediction accuracy. The FE simulations of the welding of the P91 and P92 pipes comprise thermal and sequentially coupled structural analyses. The thermal analyses model the heat evolution produced by the welding arc, determining the temperature history throughout the FE models. Structural analyses use the computed temperature history as input data to predict the residual stress fields throughout the models. Post-weld heat treatment (PWHT) of both pipes has also been numerically simulated by assuming that the FE models obey the Norton creep law during the hold time period at 760°C. The residual stresses presented here have all been validated by corresponding experimental measurements. Before PWHT, it has been found that, at certain locations in the weld region and heat affected zone (HAZ) in the pipes, tensile hoop and axial residual stresses approach the tensile strength of the material, presenting a high risk of failure. It has also been found that PWHT substantially reduces the magnitude of residual stresses by varying degrees depending on the material.

Development of New Hybrid Shell Element Takinging the Normal Stress into Account

The calculation of the stress distribution in the forming stage of sheet metal forming is critical to the accuracy of springback simulation. In order to improve the accuracy of stress prediction, a new mixed shell element is presented. Two pseudo nodes are added in the thickness direction so that the normal strain can be calculated using the displacements of the nodes. Mixed formulation is used instead of that of pure displacement. As a result, the accuracy of stress prediction is improved. Examples are given to show the improvement.

EVALUATION AND MINIMIZATION OF GEOMETRIC RECONSTRUCTION ERRORS IN FEM MODELS GENERATED FROM CT-SCAN IMAGES

CAD reconstruction of anatomical regions from computerized tomography (CT) scans is a very common approach in orthopaedic biomechanics. The CAD model is discretized into finite volume sub-domains and finite element (FE) analyses are performed in order to predict the distribution of stresses generated by applied loads. However, quality and reliability of numerical results depend on the level of accuracy reached in the meshing process. This paper analyzes some critical parameters that may affect the overall efficiency of the CT–FEM transformation process: scan threshold range, object size, and complexity. An optimization procedure for minimizing geometric errors on size and shape of reconstructed objects is presented. Finally, accuracy of stress predictions is evaluated for FE models that include known amounts of geometric errors. Compression and bending loads are considered. Results show that geometric and stress errors rapidly decrease as the objects to be reconstructed become larger in size. Optimal threshold ranges can be identified clearly for both an epoxy-resin benchmark model and a real bone specimen cut from a human lumbar vertebra. This allows geometric errors to be reduced significantly.

Improving the efficiency of finite element formulations in laminated composites

AbstractThis communication extends the principles of an advanced smeared laminate plate theory towards the development of corresponding FE models and codes. The present FE numerical results are compared with those based on exact elasticity solutions, as well as those of corresponding FE models based on three conventional laminate plate theories. These comparisons show that, compared to those conventional FE codes, the proposed FE formulation that uses also a small and fixed number of nodal degrees of freedom improves substantially the accuracy of stress predictions. They also show that the present numerical results are particularly accurate even for very thick laminates. Copyright © 2002 John Wiley & Sons, Ltd.

Contribution to Improved Accuracy in the Calculation and Measurement of Torsional Vibration Stresses in Marine Propeller Shafting

Methods of calculating torsional vibration stresses in marine propeller shafting have differed mainly in the apportionment of damping energy between engine and propeller, and for this purpose a propeller damping constant of average value has commonly been assumed. The author has long considered this simplifying approximation unjustified, since no account is taken of variations in geometrical design factors as affecting propeller damping properties. As the propeller is usually by far the major damping influence in the single-node mode, the accuracy of stress prediction must often suffer correspondingly. The paper details a convenient method of deriving damping coefficients for modern three- and four-bladed bronze propellers. Curves are plotted, for a series of pitch ratios, against a non-dimensional power coefficient involving only quantities usually known in the design stage, namely, horse-power, propeller revolutions and diameter. Allowance for disk area ratio is by interpolation between limiting values. Attention is also drawn to possible serious errors in interpreting torsiograph records taken from the crankshaft end of marine heavy-oil engine systems, arising from the phase difference between engine and propeller vibration at resonance—a consequence of heavy propeller damping. Here again improved knowledge of propeller damping characteristics is desirable if maximum accuracy is to be achieved. The paper concludes with a worked example from practice, indicating the order of magnitude of errors involved, with details of a simple vector method of allowing for the phase effect in any given instance.