Adaptive Constraint Handling in Optimization of Complex Structures by Using Machine Learning

2021 ◽  
Author(s):  
Yuecheng Cai ◽  
Jasmin Jelovica
Keyword(s):  
Machine Learning ◽  
Pareto Front ◽  
Search Algorithms ◽  
Constraint Handling ◽  
Computational Time ◽  
Complex Structures ◽  
Computationally Efficient ◽  
Nsga Ii ◽  
Slowing Down ◽  
Handling Methods

Abstract Optimization of complex systems requires robust and computationally efficient global search algorithms. Constraints make this a very difficult task, significantly slowing down an algorithm, and can even prevent finding the true Pareto front. This study continues the development of a recently proposed repair approach that exploits infeasible designs to increase computational efficiency of a prominent genetic algorithm, and to find a wider spread of the Pareto front. This paper proposes adaptive and automatized discovery of sensitivity of constraints to variables, i.e. the link, which needed direct designer’s input in the previous version of the repair approach. This is achieved by using machine learning in the form of artificial neural networks (ANN). A surrogate model is afterwards utilized in optimization based on ANN. The proposed approach is used for the recently proposed constraint handling implemented into NSGA-II optimization algorithm. The proposed framework is compared with two other constraint handling methods. The performance is analyzed on a structural optimization of a 178 m long chemical tanker which needs to fulfil class society’s criteria for strength. The results show that the proposed framework is competitive in terms of convergence and spread of the front. This is achieved while discovering the link automatically using ANN, without an input from a user. In addition, computational time is reduced by 60%.

Multi-Objective Geometric Optimization of Elliptical-Toothed Gerotor Pumps for Kinematics and Wear by Genetic Algorithm

2018 ◽  
Author(s):  
Andrew J. Robison ◽  
Andrea Vacca
Keyword(s):  
Genetic Algorithm ◽  
Pareto Front ◽  
Adhesive Wear ◽  
Geometric Optimization ◽  
Objective Functions ◽  
Computationally Efficient ◽  
Nsga Ii ◽  
Generation Algorithm ◽  
Flow Ripple ◽  
Gerotor Pumps

A computationally efficient gerotor gear generation algorithm has been developed that creates elliptical-toothed gerotor gear profiles, identifies conditions to guarantee a feasible geometry, evaluates several performance objectives, and is suitable to use for geometric optimization. Five objective functions are used in the optimization: minimize pump size, flow ripple, adhesive wear, subsurface fatigue (pitting), and tooth tip leakage. The gear generation algorithm is paired with the NSGA-II optimization algorithm to minimize each of the objective functions subject to the constraints to define a feasible geometry. The genetic algorithm is run with a population size of 1000 for a total of 500 generations, after which a clear Pareto front is established and displayed. A design has been selected from the Pareto front which is a good compromise between each of the design objectives and can be scaled to any desired displacement. The results of the optimization are also compared to two profile geometries found in literature. Two alternative geometries are proposed that offer much lower adhesive wear while respecting the size constraints of the published profiles and are thought to be an improvement in design.

Accurate Detection and Quantization of Leaf- Diseases through Soft Computing

2018 ◽  
Vol 1 (1) ◽  
pp. 236-247
Author(s):  
Divya Srivastava ◽  
Rajitha B. ◽  
Suneeta Agarwal
Keyword(s):  
Machine Learning ◽  
Image Processing ◽  
Agricultural Production ◽  
Bacterial Blight ◽  
Early Stage ◽  
Second Phase ◽  
Computationally Efficient ◽  
Stage Of Disease ◽  
Accurate Detection ◽  
Two Phases

Diseases in leaves can cause the significant reduction in both quality and quantity of agricultural production. If early and accurate detection of disease/diseases in leaves can be automated, then the proper remedy can be taken timely. A simple and computationally efficient approach is presented in this paper for disease/diseases detection on leaves. Only detecting the disease is not beneficial without knowing the stage of disease thus the paper also determine the stage of disease/diseases by quantizing the affected of the leaves by using digital image processing and machine learning. Though there exists a variety of diseases on leaves, but the bacterial and fungal spots (Early Scorch, Late Scorch, and Leaf Spot) are the most prominent diseases found on leaves. Keeping this in mind the paper deals with the detection of Bacterial Blight and Fungal Spot both at an early stage (Early Scorch) and late stage (Late Scorch) on the variety of leaves. The proposed approach is divided into two phases, in the first phase, it identifies one or more disease/diseases existing on leaves. In the second phase, amount of area affected by the disease/diseases is calculated. The experimental results obtained showed 97% accuracy using the proposed approach.

scholarly journals A computationally efficient hybrid 2D–3D subwoofer model

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Ahmad H. Bokhari ◽  
Martin Berggren ◽  
Daniel Noreland ◽  
Eddie Wadbro
Keyword(s):  
Hybrid Model ◽  
3D Model ◽  
Element Model ◽  
Acoustic Properties ◽  
Computational Time ◽  
Average Response ◽  
Frequency Range ◽  
Computationally Efficient ◽  
Lumped Element ◽  
Lumped Element Model

AbstractA subwoofer generates the lowest frequency range in loudspeaker systems. Subwoofers are used in audio systems for live concerts, movie theatres, home theatres, gaming consoles, cars, etc. During the last decades, numerical simulations have emerged as a cost- and time-efficient complement to traditional experiments in the design process of different products. The aim of this study is to reduce the computational time of simulating the average response for a given subwoofer design. To this end, we propose a hybrid 2D–3D model that reduces the computational time significantly compared to a full 3D model. The hybrid model describes the interaction between different subwoofer components as interacting modules whose acoustic properties can partly be pre-computed. This allows us to efficiently compute the performance of different subwoofer design layouts. The results of the hybrid model are validated against both a lumped element model and a full 3D model over a frequency band of interest. The hybrid model is found to be both accurate and computationally efficient.

Multiobjective optimization of hard coating parameters designing for damping of the bladed disk

2017 ◽  
Vol 232 (9) ◽  
pp. 1609-1619 ◽  
Author(s):  
Yugang Chen ◽  
Jingyu Zhai ◽  
Qingkai Han
Keyword(s):  
Pareto Front ◽  
Engineering Application ◽  
Front Surface ◽  
Model Material ◽  
Superior Performance ◽  
Hard Coating ◽  
Damping Capacity ◽  
Nsga Ii ◽  
Bladed Disk ◽  
Design Variables

In this paper, the damping capacity and the structural influence of the hard coating on the given bladed disk are optimized by the non-dominated sorting genetic algorithm (NSGA-II) coupled with the Kriging surrogate model. Material and geometric parameters of the hard coating are taken as the design variables, and the loss factors, frequency variations and weight gain are considered as the objective functions. Results of the bi-objective optimization are obtained as curved line of Pareto front, and results of the triple-objective optimization are obtained as Pareto front surface with an obvious frontier. The results can give guidance to the designer, which can help to achieve more superior performance of hard coating in engineering application.

scholarly journals A machine-learning-based alloy design platform that enables both forward and inverse predictions for thermo-mechanically controlled processed (TMCP) steel alloys

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Jin-Woong Lee ◽  
Chaewon Park ◽  
Byung Do Lee ◽  
Joonseo Park ◽  
Nam Hoon Goo ◽  
...  
Keyword(s):  
Machine Learning ◽  
Feature Space ◽  
Research Strategy ◽  
Nsga Ii ◽  
Steel Alloys ◽  
Non Linear ◽  
Data Driven Approach ◽  
Processing Information ◽  
Set Up ◽  
World Industry

AbstractPredicting mechanical properties such as yield strength (YS) and ultimate tensile strength (UTS) is an intricate undertaking in practice, notwithstanding a plethora of well-established theoretical and empirical models. A data-driven approach should be a fundamental exercise when making YS/UTS predictions. For this study, we collected 16 descriptors (attributes) that implicate the compositional and processing information and the corresponding YS/UTS values for 5473 thermo-mechanically controlled processed (TMCP) steel alloys. We set up an integrated machine-learning (ML) platform consisting of 16 ML algorithms to predict the YS/UTS based on the descriptors. The integrated ML platform involved regularization-based linear regression algorithms, ensemble ML algorithms, and some non-linear ML algorithms. Despite the dirty nature of most real-world industry data, we obtained acceptable holdout dataset test results such as R2 > 0.6 and MSE < 0.01 for seven non-linear ML algorithms. The seven fully trained non-linear ML models were used for the ensuing ‘inverse design (prediction)’ based on an elitist-reinforced, non-dominated sorting genetic algorithm (NSGA-II). The NSGA-II enabled us to predict solutions that exhibit desirable YS/UTS values for each ML algorithm. In addition, the NSGA-II-driven solutions in the 16-dimensional input feature space were visualized using holographic research strategy (HRS) in order to systematically compare and analyze the inverse-predicted solutions for each ML algorithm.

scholarly journals Improved Visible Light-Based Indoor Positioning System Using Machine Learning Classification and Regression

2019 ◽  
Vol 9 (6) ◽  
pp. 1048 ◽  
Author(s):  
Huy Tran ◽  
Cheolkeun Ha
Keyword(s):  
Machine Learning ◽  
Visible Light ◽  
Noise Reduction ◽  
Indoor Positioning ◽  
Computational Time ◽  
Dual Function ◽  
Positioning System ◽  
Positioning Accuracy ◽  
Positioning Systems ◽  
Machine Learning Classification

Recently, indoor positioning systems have attracted a great deal of research attention, as they have a variety of applications in the fields of science and industry. In this study, we propose an innovative and easily implemented solution for indoor positioning. The solution is based on an indoor visible light positioning system and dual-function machine learning (ML) algorithms. Our solution increases positioning accuracy under the negative effect of multipath reflections and decreases the computational time for ML algorithms. Initially, we perform a noise reduction process to eliminate low-intensity reflective signals and minimize noise. Then, we divide the floor of the room into two separate areas using the ML classification function. This significantly reduces the computational time and partially improves the positioning accuracy of our system. Finally, the regression function of those ML algorithms is applied to predict the location of the optical receiver. By using extensive computer simulations, we have demonstrated that the execution time required by certain dual-function algorithms to determine indoor positioning is decreased after area division and noise reduction have been applied. In the best case, the proposed solution took 78.26% less time and provided a 52.55% improvement in positioning accuracy.

scholarly journals Forecasting System of Computational Time of DFT/TDDFT Calculations under the Multiverse Ansatz via Machine Learning and Cheminformatics

ACS Omega ◽  
2021 ◽  
Vol 6 (3) ◽  
pp. 2001-2024
Author(s):  
Shuo Ma ◽  
Yingjin Ma ◽  
Baohua Zhang ◽  
Yingqi Tian ◽  
Zhong Jin
Keyword(s):  
Machine Learning ◽  
Computational Time ◽  
Tddft Calculations ◽  
Forecasting System

scholarly journals A Multidimensional and Multiscale Model for Pressure Analysis in a Reservoir-Pipe-Valve System

2018 ◽  
Author(s):  
Feng Jie Zheng ◽  
Fu Zheng Qu ◽  
Xue Guan Song
Keyword(s):  
Pressure Fluctuation ◽  
Large Scale ◽  
Three Dimensional ◽  
Industrial Process ◽  
Multiscale Model ◽  
Computational Time ◽  
Small Scale ◽  
Cfd Model ◽  
Computationally Efficient ◽  
Proposed Model

Reservoir-pipe-valve (RPV) systems are widely used in many industrial process. The pressure in an RPV system plays an important role in the safe operation of the system, especially during the sudden operation such as rapid valve opening/closing. To investigate the pressure especially the pressure fluctuation in an RPV system, a multidimensional and multiscale model combining the method of characteristics (MOC) and computational fluid dynamics (CFD) method is proposed. In the model, the reservoir is modeled by a zero-dimensional virtual point, the pipe is modeled by a one-dimensional MOC, and the valve is modeled by a three-dimensional CFD model. An interface model is used to connect the multidimensional and multiscale model. Based on the model, a transient simulation of the turbulent flow in an RPV system is conducted, in which not only the pressure fluctuation in the pipe but also the detailed pressure distribution in the valve are obtained. The results show that the proposed model is in good agreement with the full CFD model in both large-scale and small-scale spaces. Moreover, the proposed model is more computationally efficient than the CFD model, which provides a feasibility in the analysis of complex RPV system within an affordable computational time.

scholarly journals A Graph-based Evolutionary Algorithm for Automated Machine Learning

2020 ◽  
Author(s):  
Fei Qi ◽  
Zhaohui Xia ◽  
Gaoyang Tang ◽  
Hang Yang ◽  
Yu Song ◽  
...  
Keyword(s):  
Machine Learning ◽  
Evolutionary Algorithm ◽  
Parameter Optimization ◽  
State Of The Art ◽  
The State ◽  
Complex Structures ◽  
Architecture Evolution ◽  
Automated Machine Learning ◽  
Art Performance

As an emerging field, Automated Machine Learning (AutoML) aims to reduce or eliminate manual operations that require expertise in machine learning. In this paper, a graph-based architecture is employed to represent flexible combinations of ML models, which provides a large searching space compared to tree-based and stacking-based architectures. Based on this, an evolutionary algorithm is proposed to search for the best architecture, where the mutation and heredity operators are the key for architecture evolution. With Bayesian hyper-parameter optimization, the proposed approach can automate the workflow of machine learning. On the PMLB dataset, the proposed approach shows the state-of-the-art performance compared with TPOT, Autostacker, and auto-sklearn. Some of the optimized models are with complex structures which are difficult to obtain in manual design.

Sign in / Sign up

Export Citation Format

Share Document