Hybrid teaching–learning-based optimization and neural network algorithm for engineering design optimization problems

Zhang, Yiying; Jin, Zhigang; Chen, Ye

doi:10.1016/j.knosys.2019.07.007

Cited by 104 publications

(34 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other variants are the ant colony algorithm [31] and the teaching-learning algorithm [32]. Among these algorithms, the neural network algorithm (NNA) is a new effective and adaptive optimization technique [33][34][35]. It is demonstrated to have a global search feature based on the principles of artificial neural networks.…”

Section: Introductionmentioning

confidence: 99%

Effective Nonlinear Model Predictive Control Scheme Tuned by Improved NN for Robotic Manipulators

et al. 2021

View full text Add to dashboard Cite

The nonlinearities of the robotic manipulators and the uncertainties of their parameters represent big challenges against the controller design. Moreover, the tracking of regular and irregular trajectories with fewer overshoots, short settling time, and small steady-state error is the main target for the robotic response. The model predictive control (MPC) is an efficient controller to handle the performance requirements. However, the conventional MPC requires the linearization of the system model. The linearization of the model does not cover all dynamics of the robotic system. Thus, this paper introduces the nonlinear MPC (NLMPC) as a proper control method for the nonlinear systems instead of the conventional MPC. Specifically, this work proposes the use of NLMPC for controlling robotic manipulators. However, the NLMPC gains need proper tuning to attain good performance rather than the conventional methods. The neural network algorithm (NNA) considers a sufficient adaptive intelligent technique that can be utilized for this purpose. The restriction in a local optimum reveals the main issue versus artificial intelligence techniques. This paper suggests a new improvement to reinforce the exploration behavior of the NNA to overcome the local restriction issue. This modification is carried out by utilizing the polynomial mutation as an effective method to promise the exploration manner of the intelligence techniques. The proposed system can estimate all states from only the output to reduce the cost of the required sensors to measure all states. The results confirm the superiority of the proposed systems with the estimator with negligible change in the output response. The proposed modified NNA (MNNA) is evaluated with the main NNA, genetic algorithm-based PID control scheme, besides the cuckoo search algorithm-based PID control scheme from other works. The results confirm the robustness and effectiveness of the suggested MNNA-based NLMPC to track regular and irregular trajectories compared with other techniques.

show abstract

Section: Introductionmentioning

confidence: 99%

Effective Nonlinear Model Predictive Control Scheme Tuned by Improved NN for Robotic Manipulators

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Whilst gradient-based optimization methods have been utilized for many years for the purpose, they are known to have numerous deficiencies which led to the birth and pursue of metaheuristic optimization algorithms. The latter algorithms involve an iterative procedure Besides, some of the other challenges in optimization of engineering design problems can be mentioned as the epsilon constraint based HTS algorithm for optimization of multi-objective engineering design problems [35], Layout optimization of wind farms with an improved version of TLBO [36], design optimization of engineering problems by a hybrid approach of TLBO and the Neural Network Algorithm (NNA) [37], Symbiotic Organisms Search (SOS) algorithm for optimum design of multi-objective constrained engineering problems [38], Bayesian optimization (BO) for optimum design of engineering design problems, optimum design of real-world problems by Seagull Optimization Algorithm (SOA) [39] and the Black Widow Optimization (BWO) algorithm for optimization purposes in engineering applications [40].…”

Section: Introductionmentioning

confidence: 99%

Optimization of Engineering Design Problems Using Atomic Orbital Search Algorithm

2021

View full text Add to dashboard Cite

In this paper, optimum design of engineering problems is considered by means of the Atomic Orbital Search (AOS), a recently proposed metaheuristic optimization algorithm. The algorithm is inspired by the quantum-based atomic model. In particular, the mathematical development of the algorithm is based on principles of quantum mechanics focusing on the act of electrons around the nucleus of an atom. For numerical investigation, 20 of well-known constrained design problems in different engineering fields are considered; some of which have been benchmarked by the 2020 Competitions on Evolutionary Computation (CEC 2020) for realworld optimization purposes. Statistical results including the best, mean, worst and standard deviation of multiple optimization runs are reported for the AOS algorithm. These results are compared to similar data from previous algorithms found in the literature to establish the efficiency and usefulness of the AOS algorithm. It is concluded that the AOS has acceptable behavior in dealing with all the considered constrained optimization problems while the maximum difference of about 40% between the best optimum values of the AOS and other approaches is noted for the robot gripper benchmark problem.

show abstract

“…Heuristic optimization methods are carried on optimizing the structure of the NN. Some of them are particle swam optimization, genetic algorithm, teaching learning optimization, firefly optimization [31][32][33][34].…”

Section: Introductionmentioning

confidence: 99%

Robot Manipulator Calibration Using a Model Based Identification Technique and a Neural Network With the Teaching Learning-Based Optimization

Kang

2020

IEEE Access

View full text Add to dashboard Cite

This paper proposes a new calibration method for enhancing robot positional accuracy of the industrial manipulators. By combining the joint deflection model with the conventional kinematic model of a manipulator, the geometric errors and joint deflection errors can be considered together to increase its positional accuracy. Then, a neural network is designed to additionally compensate the unmodeled errors, specially, non-geometric errors. The teaching-learning-based optimization method is employed to optimize weights and bias of the neural network. In order to demonstrate the effectiveness of the proposed method, real experimental studies are carried out on HH 800 manipulator. The enhanced position accuracy of the manipulator after the calibration confirms the feasibility and more positional accuracy over the other calibration methods.

show abstract

Hybrid teaching–learning-based optimization and neural network algorithm for engineering design optimization problems

Cited by 104 publications

References 61 publications

Effective Nonlinear Model Predictive Control Scheme Tuned by Improved NN for Robotic Manipulators

Effective Nonlinear Model Predictive Control Scheme Tuned by Improved NN for Robotic Manipulators

Optimization of Engineering Design Problems Using Atomic Orbital Search Algorithm

Robot Manipulator Calibration Using a Model Based Identification Technique and a Neural Network With the Teaching Learning-Based Optimization

Contact Info

Product

Resources

About