teaching–learning-based optimization

Today, due to the complexity of engineering problems and at the same time the advancement of computer science, the use of machine learning (ML) methods and soft computing methods in solving engineering problems has been considered by many researchers. These methods can be used to find accurate estimates for problems in various scientific fields. This paper investigates the effectiveness of the Adaptive Network-Based Fuzzy Inference System (ANFIS) hybridized with Teaching Learning Based Optimization Algorithm (TLBO), to predict the ultimate strength of columns with square and rectangular cross-sections, confide with various fiber-reinforced polymer (FRP) sheets. In previous studies by many researchers, several experiments have been conducted on concrete columns confined by FRP sheets. The results indicate that FRP sheets effectively increase the compressive strength of concrete columns. Comparing the results of ANFIS-TLBO with the experimental findings, which were agreeably consistent, demonstrated the ability of ANFIS-TLBO to estimate the compressive strength of concrete confined by FRP. Also, the comparison of RMSE, SD, and R2 for ANFIS-TLBO and the studies of different researchers show that the ANFIS-TLBO approach has a good performance in estimating compressive strength. For example, the value of R2 in the proposed method was 0.92, while this parameter was 0.87 at best among the previous studies. Also, the obtained error in the prediction of the proposed model is much lower than the obtained error in the previous studies. Hence, the proposed model is more efficient and works better than other techniques.

As a novel strategy, Quantum-behaved particles use uncertainty law and a distinct formulation obtained from solving the time-independent Schrodinger differential equation in the delta-potential-well function to update the solution candidates’ positions. In this case, the local attractors as potential solutions between the best solution and the others are introduced to explore the solution space. Also,  the difference between the average and another solution is established as a new step size. In the present paper, the quantum teacher phase is introduced to improve the performance of the current version of the teacher phase of the Teaching-Learning-Based Optimization algorithm (TLBO) by using the formulation obtained from solving the time-independent Schrodinger equation predicting the probable positions of optimal solutions. The results show that QTLBO, an acronym for the Quantum Teaching- Learning- Based Optimization, improves the stability and robustness of the TLBO by defining the quantum teacher phase. The two circulant space trusses with multiple frequency constraints are chosen to verify the quality and performance of QTLBO. Comparing the results obtained from the proposed algorithm with those of the standard version of the TLBO algorithm and other literature methods shows that QTLBO increases the chance of finding a better solution besides improving the statistical criteria compared to the current TLBO.

The optimal design of dome structures is a challenging task and therefore the computational performance of the currently available techniques needs improvement. This paper presents a combined algorithm, that is supported by the mixture of Charged System Search (CSS) and Teaching-Learning-based optimization (TLBO). Since the CSS algorithm features a strong exploration and may explore all unknown locations within the search space, it is an appropriate complement to enhance the optimization process by solving the weaknesses with using another optimization algorithm’s strong points. To enhance the exploitation ability of this algorithm, by adding two parts of Teachers phase and Student phase of TLBO algorithm to CSS, a method is obtained that is more efficient and faster than standard versions of these algorithms. In this paper, standard optimization methods and new hybrid method are tested on three kinds of dome structures, and the results show that the new algorithm is more efficient in comparison to their standard versions.

In this study, to enhance the optimization process, especially in the structural engineering field two well-known algorithms are merged together in order to achieve an improved hybrid algorithm. These two algorithms are Teaching-Learning Based Optimization (TLBO) and Harmony Search (HS) which have been used by most researchers in varied fields of science. The hybridized algorithm is called A Discrete Hybrid Teaching-Learning Based Optimization (DHTLBO) that is applied to optimization of truss structures with discrete variables. This new method is consisted of two parts: in the first part the TLBO algorithm applied as conventional TLBO for local optimization, in the second stage the HS algorithm is applied to global optimization and exploring all the unknown places in the search space. The new hybrid algorithm is employed to minimize the total weight of structures. Therefore, the objective function consists of member’s weight, which is depends on the form of stress and deflection limits. To demonstrate the efficiency and robustness of this new algorithm several truss structures which are optimized by most researchers are presented and then their results are compared to other meta-heuristic algorithm and TLBO and HS standard algorithms.

Imperialist Competitive Algorithm, ICA is a meta-heuristic which simulates collapse of weak empires by more powerful ones that take possession of their colonies. In order to enhance performance, ICA is hybridized with proper features of Teaching-Learning-Based Optimization, TLBO. In addition, ICA walks are modified with an extra term to intensify looking for the global best solution. The number of control parameters and consequent tuning effort has been reduced in the proposed Imperialist Competitive Learner-Based Optimization, ICLBO with respect to ICA and several other methods. Efficiency and effectiveness of ICLBO is further evaluated treating a number of test functions in addition to continuous and discrete engineering problems. It is discussed and traced that balancing between exploration and exploitation is enhanced due to the proposed hybridization. Numerical results exhibit superior performance of ICLBO vs. ICA and a variety of other well-known meta-heuristics.