Teaching Learning-based Algorithm Research Articles

Deep contextual reinforcement learning algorithm for scalable power scheduling

This article introduces a deep contextual reinforcement learning (DCRL) optimization algorithm to tackle the NP-hard power scheduling problem. The algorithm is based on a multi-agent simulation environment that decomposes the power scheduling problem into sequential Markov decision processes (MDPs). The simulation environment inherently corrects incorrect decisions and adjusts supply capacities so that the MDPs adhere to the optimization constraints. A deep reinforcement learning (DRL) model is trained on these MDPs to provide optimal solutions. Demonstrating its applicability and effectiveness, the proposed method is evaluated on various test systems with different unit numbers, constraints, and production cost functions. The experimental results show that the proposed method has better performance relative to alternative methods, such as binary alternative moth-flame optimization (BAMFO), binary particle swarm optimization (BPSO), teaching learning-based optimization (TLBO), new binary particle swarm optimization (NBPSO), binary learning particle swarm optimization (BLPSO), quasi-oppositional teaching learning-based algorithm (QOTLBO), binary-real-coded genetic algorithm (BRCGA), hybrid genetic-imperialist competitive algorithm (HGICA), three-stage priority list (TSPL), and hybridized evolutionary algorithm and sequential quadratic programming (HEPSQP). The novelties of the proposed method lie in its adaptability to longer planning horizons and linear dimensionality complexity, rendering it suitable for large-scale power systems.

Reinforcement learning-based optimization for power scheduling in a renewable energy connected grid

Power scheduling is an NP-hard optimization problem that demands a delicate equilibrium between economic costs and environmental emissions. In response to the growing concern for climate change, global environmental policies prioritize decarbonizing the electricity sector by integrating renewable energies (REs) into power grids. While this integration brings economic and environmental benefits, the intermittency of REs amplifies the uncertainty and complexity of power scheduling. Existing optimization approaches often grapple with a limited number of units, overlook critical parameters, and disregard the intermittency of REs. To address these limitations, this article introduces a robust and scalable optimization algorithm for renewable integrated power scheduling based on reinforcement learning (RL). In this proposed methodology, the power scheduling problem is decomposed into Markov decision processes (MDPs) within a multi-agent simulation environment. The simulated MDPs are used to train a deep reinforcement learning (DRL) model for solving the optimization. The validity and effectiveness of the proposed method are validated across various test systems, encompassing single-to tri-objective problems with 10–100 generating units. The findings consistently demonstrate the superior performance of the proposed DRL algorithm compared to existing methods, such as multi-agent immune system-based evolutionary priority list (MAI-EPL), binary real-coded genetic algorithm (BRCGA), teaching learning-based optimization (TLBO), quasi-oppositional teaching learning-based algorithm (QOTLBO), hybrid genetic-imperialist competitive algorithm (HGICA), three-stage priority list (TSPL), real-coded grey wolf optimization (RCGWO), multi-objective evolutionary algorithm based on decomposition (MOEAD), and non-dominated sorting algorithms (NSGA-II and NSGA-III). Regarding the experimental results, it is important to highlight the importance of integrating RESs into larger power systems. In a 10-unit system with 2.81 % RE penetration, reductions of 3.42 %, 4.03 %, and 3.10 % were observed in costs, CO2 emissions, and SO2 emissions, respectively. Similarly, in a 100-unit system with a RE penetration rate of only 0.28 %, reductions of 3.75 % in cost, 4.42 % in CO2, and 3.34 % in SO2 were observed. These findings emphasize the effectiveness of RES integration, even at lower penetration rates, in larger-scale power systems.

Optimizing Multimodal Transportation Systems Using the Teaching–Learning-Based Algorithm

Optimizing Multimodal Transportation Systems Using the Teaching–Learning-Based Algorithm

A Machine Learning Perspective to the Investigation of Surface Integrity of Al/SiC/Gr Composite on EDM

Conventional mechanical machining of composite is a challenging task, and thus, electric discharge machining (EDM) was used for the processing of the developed material. The processing of developed composite using different electrodes on EDM generates different surface characteristics. In the current work, the effect of tool material on the surface characteristics, along with other input parameters, is investigated as per the experimental design. The experimental design followed is an RSM-based Box–Behnken design, and the input parameters in the current research are tool material, current, voltage, pulse-off time, and pulse-on time. Three levels of each parameter are selected, and 46 experiments are conducted. The surface roughness (Ra) is investigated for each experimental setting. The machine learning approach is used for the prediction of surface integrity by different techniques, namely Xgboost, random forest, and decision tree. Out of all the techniques, the Xgboost technique shows maximum accuracy as compared to other techniques. The analysis of variance of the predicted solutions is investigated. The empirical model is developed using RSM and is further solved with the help of a teaching learning-based algorithm (TLBO). The SR value predicted after RSM and integrated approach of RSM-ML-TLBO are 2.51 and 2.47 µm corresponding to Ton: 45 µs; Toff: 73 µs; SV:8V; I: 10A; tool: brass and Ton: 47 µs; Toff: 76 µs; SV:8V; I: 10A; tool: brass, respectively. The surface integrity at the optimized setting reveals the presence of microcracks, globules, deposited lumps, and sub-surface formation due to different amounts of discharge energy.

Techno-economic analysis of grid-connected hybrid renewable energy system adapting hybrid demand response program and novel energy management strategy

Over the last two decades, there has been a significant interest in developing solutions to improve the efficiency of power systems, such as optimal energy consumption and management. The demand response programs (DRP) have long been one of the most effective strategies to encourage consumers to change their consumption behavior. In this paper, a novel hybrid DRP has been implemented by combining the time of use (TOU) and incentive-based (IB) DRP methods. To meet the load demand, a PV-WT-BESS hybrid renewable energy system (HRES) has been proposed in this work. Further, to analyze the impact of hybrid DRP on the capacity allocation of different system components, optimal sizing has been performed in two scenarios, one with the actual load and the second with the load after considering the hybrid DRP. The optimization has been carried out by using an improved search space reduction (ISSR) algorithm. To demonstrate the effectiveness of the ISSR algorithm, the obtained results are compared with other algorithms such as particle swarm optimization (PSO), grey wolf optimization (GWO), teaching-learning based algorithm (TLBO), seagull optimization algorithm (SOA), and sine cosine algorithm (SCA). Finally, a novel energy management strategy (EMS) is also proposed in this work on the supply side of the HRES, which intelligently uses the variations in the grid tariff to charge the BESS when the tariff is low. To demonstrate the proposed DRP and EMS-based sizing, a case study has been performed for a grid-connected PV-WT-BESS for the location of Kanyakumari, India. From the results obtained, it has been concluded that the hybrid DRP reduces the peak demand up to 5%, which reduces the levelized cost of energy (LCE) by 22%. Furthermore, the proposed novel EMS reduces the LCE further by 5%. The overall cost reduction of up to 25% has been achieved by using both the proposed hybrid DRP and EMS, while satisfying all necessary constraints.

Comparison of different regulations and metaheuristic algorithms in beam design

In this first study, the rectangular reinforced concrete beam’s costs and cross-section sizes are found by using Harmony Search (HS), Differential Evolution Algorithm (DE), Jaya Algorithm, Teaching- Learning Based Algorithm (TLBO), Hybrid algorithm (Jaya-TLBO) and Flower Pollination Algorithm (FPA) separately by using ACI 318 building code. In addition, in order to better see how successful the algorithms are, the standard deviation of the algorithms used in the project in a certain number of iterations, price changes and in which iteration the minimum cost is compared. As a result of running different algorithms 5, 10, 15 and 20 times, separate values are recorded, and the average number of iterations of the algorithms for each is shown by finding the standard deviation values. Furthermore, Hybrid Algorithm reached the objective function in fewer iterations and their standard deviations reached 0 earlier. In the second study, the beam design is made according to the ACI 318, TS500 and Eurocode 2 regulations under certain loads by using a Hybrid Algorithm with different concrete classes. Optimization of this design is made using the Matlab program, and comparisons are made between regulations. Eurocode and TS500 design costs are roughly the same; however, ACI 318’s design is the cheapest.

Extract nonlinear operating rules of multi-reservoir systems using an efficient optimization method

Hydropower plants are known as major renewable energy sources, usually used to meet energy demand during peak periods. The performance of hydropower reservoir systems is mainly affected by their operating rules, thus, optimizing these rules results in higher and/or more reliable energy production. Due to the complex nonlinear, nonconvex, and multivariable characteristics of the hydropower system equations, deriving the operating rules of these systems remains a challenging issue in multi-reservoir systems optimization. This study develops a self-adaptive teaching learning-based algorithm with differential evolution (SATLDE) to derive reliable and precise operating rules for multi-reservoir hydropower systems. The main novelty of SATLDE is its enhanced teaching and learning mechanism with three significant improvements: (i) a ranking probability mechanism is introduced to select the learner or teacher stage adaptively; (ii) at the teacher stage, the teaching mechanism is redefined based on learners’ performance/level; and (iii) at the learner stage, an effective mutation operator with adaptive control parameters is proposed to boost exploration ability. The proposed SATLDE algorithm is applied to the ten-reservoir benchmark systems and a real-world hydropower system in Iran. The results illustrate that the SATLDE achieves superior precision and reliability to other methods. Moreover, results show that SATLDE can increase the total power generation by up to 23.70% compared to other advanced optimization methods. Therefore, this study develops an efficient tool to extract optimal operating rules for the mentioned systems.

Enhanced Teaching Learning-Based Algorithm for Fuel Costs and Losses Minimization in AC-DC Systems

The Teaching Learning-Based Algorithm (TLBA) is a powerful and effective optimization approach. TLBA mimics the teaching-learning process in a classroom, where TLBA’s iterative computing process is separated into two phases, unlike standard evolutionary algorithms and swarm intelligence algorithms, and each phase conducts an iterative learning operation. Advanced technologies of Voltage Source Converters (VSCs) enable greater active and reactive power regulation in these networks. Various objectives are addressed for optimal energy management, with the goal of attaining economic and technical advantages by decreasing overall production fuel costs and transmission power losses in AC-DC transmission networks. In this paper, the TLBA is applied for various sorts of nonlinear and multimodal functioning of hybrid alternating current (AC) and multi-terminal direct current (DC) power grids. The proposed TLBA is evaluated on modified IEEE 30-bus and IEEE 57-bus AC-DC networks and compared to other published methods in the literature. Numerical results demonstrate that the proposed TLBA has great effectiveness and robustness indices over the others. Economically, the reduction percentages of 13.84 and 21.94% are achieved for the IEEE 30-bus and IEEE 57-bus test systems when the fuel costs are minimized. Technically, significant improvement in the transmission power losses with reduction 28.01% and 69.83%, are found for the IEEE 30-bus and IEEE 57-bus test system compared to the initial case. Nevertheless, TLBA has faster convergence, higher quality for the final optimal solution, and more power for escaping from convergence to local optima compared to other published methods in the literature.

A novel transient search optimization for optimal allocation of multiple distributed generator in the radial electrical distribution network

Abstract Most of the generated electricity is lost in power loss while transmitting and distributing it to the consumer end. The power losses occurring in the distribution network cause deviation in voltage and lower stability due to increased load demand. The integration of multiple Distributed Generation (DG) will enable the existing radial electrical distribution network efficient by minimizing the power losses and improving the voltage profile. Metaheuristic optimization techniques provide a favorable solution for optimal location and sizing of DG in the distribution network. A novel modern metaheuristic Transient Search Optimization (TSO) algorithm, inspired by the electrical network’s transient response of storage components implemented in the proposed work. The TSO formulated optimal DGs allocation to minimize total active power loss, voltage deviation and enhance voltage stability index as minimization optimization problem satisfying various equality and inequality constraints. The installation of multiple DG units at unity, fixed, and optimal power factors were examined. The TSO algorithm’s effectiveness was tested on standard IEEE 33-bus and 69-bus radial distribution networks, including various operating events developed in the form of single and multi-objective fitness functions. The active power loss reduced to 94.29 and 94.71% for IEEE 33 and 69 bus distribution systems. The obtained results trustworthiness is confirmed by comparison with well-known optimization methods like Genetic Algorithm (GA), Particle Swarm Optimization (PSO), combined GA/PSO, Teaching Learning Based Algorithm (TLBO), Swine influenza model-based optimization with quarantine (SIMBO-Q), Multi-Objective Harris Hawks optimizer (MOHHO) and other provided in the literature. The presented numerical studies represent the usefulness and out-performance of the proposed TSO algorithm due to its exploration and exploitation optimization mechanisms for the DG allocation problem meticulously.

A novel method to reduce dross in laser beam cutting of Ti-6Al-4 V alloy sheet

Abstract During the laser cutting of Ti-6Al-4 V sheets, different dross heights are obtained for different thicknesses of the sheet. The maximum value of normalized dross height (= dross height per unit thickness of laser cut sheet), calculated from reported data, was 0.155. In the present investigation, the maximum normalized dross height, obtained by use of an additional nozzle (secondary gas-flow nozzle) and by parametric optimization through Teaching–Learning-based algorithm, was 0.053. The present study shows that the conventional set-up for laser beam cutting of Ti-6Al-4 V alloy, i.e., cutting with co-axial assist gas flow, may not be able to reduce the dross height completely. However, it was observed that the use of a secondary gas-flow nozzle, in tandem with the co-axial nozzle, i.e., a modified set-up, was able to reduce the dross height due to the shearing action of the argon jet flow directed at the dross. The thrust force of the argon jet acting on the dross has been estimated in the present study using analytical equations and was found to be 76.17 gf. When the longitudinal profile of cut surface was mapped using a laser displacement sensor, it was observed that the roughness of the cut surface obtained using the modified set-up was found to be less (i.e., approximately half) as compared to that obtained in case of the conventional set-up. In addition to the dross height, straightness of the cut sample (kerf wall convexity) was also taken into account for optimization.

Parameter extraction of solar photovoltaic models with an either-or teaching learning based algorithm

This paper presents an advanced variant of teaching learning based algorithm (TLBO) called either-or teaching learning based algorithm (EOTLBO) to extract accurate and reliable parameters of solar photovoltaic (PV) models. EOTLBO synergizes three enhanced strategies to accelerate the convergence rate and boost the search efficiency of TLBO. (i) A median learner based teacher phase excluding the mean position used in the original TLBO is designed to avoid infeasible and inefficient learners and to form a rational and advisable moving mechanism around the teacher. (ii) A higher-achieving learner based learner phase using three sorted learners is devised to directionally guide the target learner to jump out of local optima and to move towards a more promising region. (iii) A chaotic map based either-or teaching-learning strategy is developed to give each dimension of each learner a chance to go through either the median learner based teacher phase or the higher-achieving learner based learner phase. EOTLBO is applied to three PV cells/modules including seven cases. Compared with the original TLBO, four non-TLBO variants, and five TLBO variants, experimental results verify the superior performance of EOTLBO in terms of both the quality of final solutions and the convergence speed on all cases. In addition, the current-voltage characteristics yielded by EOTLBO agree well with the measured data independently of different PV models at different operating conditions.

Quasi-oppositional-based Rao algorithms for multi-objective design optimization of selected heat sinks

Abstract In this paper, an endeavor is made to enhance the convergence speed of the recently proposed Rao algorithms. The new upgraded versions of Rao algorithms named as “quasi-oppositional-based Rao algorithms” are proposed in this paper. The quasi-oppositional-based learning is incorporated in the basic Rao algorithms to diversify the searching process of the algorithms. The performance of the proposed algorithms is tested on 51 unconstrained benchmark functions. Also, three multi-objective optimization case studies of different heat sinks such as a single-layered microchannel heat sink (SL-MCHS), a double-layered microchannel heat sink (DL-MCHS), and a plate-fin heat sink (PFHS) are attempted to investigate the effectiveness of the proposed algorithms in solving real-world complex engineering optimization problems. The results obtained using the proposed algorithms are compared with the results obtained using the well-known advanced optimization algorithms such as genetic algorithm (GA), artificial bee colony (ABC), differential evolution (DE), particle swarm optimization (PSO), teaching-learning-based algorithm (TLBO), Jaya algorithm, multi-objective genetic algorithm (MOGA), non-dominated sorting genetic algorithm (NSGA-II), real-coded GA (RCGA), direction-based GA, self-adaptive multi-population (SAMP) Rao algorithms, and basic Rao algorithms. The proposed quasi-oppositional-based Rao algorithms are found superior or competitive to the other optimization algorithms considered.

Effect of Electric Vehicle on Load Frequncy Control by Using SALP Swarm Algorithm Optimized FOPID

Now a days due to increasing participation of renewable sources in the production of energy there are more frequency fluctuations and this frequency instability is not possible to be controlled with load frequency control (LFC) alone. Meanwhile electric vehicles (EV) are gaining popularity to improve the frequency stability. In this paper multi area multi source power system with electrical vehicle by using Salp Swarm Algorithm (SSA) optimized Fractional order PID (FOPID) controller in the presence of EV is considered. The superiority of SSA is expressed by comparing with other algorithms like local unimodal sampling (LUS), differential algorithm (DE), teaching learning based algorithm (TLBO). Likewise FOPID controller is also compared with other controllers like I, PI, PID and the superiority of FOPID is demonstrated. The investigation reveal that SSA tuned FOPID controller provides the best dynamic performance for settling time and peak undershoot in deviation of frequency and deviation in the power of tie-line with EV.

The Optimal Power Flow Solution by Optimal Location of STATCOM Device using AHP Method

Optimum power flow is a useful tool for planning and operating the electrical system and maintains the economy and safety of the modern electrical system. Teaching, Learning-based Algorithm is one of the new Metaheuristic algorithms which can influence both teachers and students by expediting the interaction among them in sharing the necessary knowledge. The proposed TLBO is designed here to solve the problem of an optimum power flow with STATCOM FACTS device. The optimal location of STATCOM FACTS device on the weak bus is obtained by Analytical Hierarchy Process (AHP) method. The main objective of this study is to reduce fuel cost of generation, reduce active and reactive power loss, improve voltage deviation and enhance voltage stability index within the given control variable constraints. The proposed TLBO algorithm with STATCOM device is evaluated on the standard IEEE-57 bus system. From the simulation result, it shows that the Teaching Learning-based algorithm gives the optimal solution as compared to the recent algorithm mentioned in the literature with some IEEE-57 bus test system.

Gene selection for cancer types classification using novel hybrid metaheuristics approach

Abstract With the advancement of microarray technology, gene expression profiling has shown remarkable effort to predict the different types of malignancy and their subtypes. In microarrays, predicting highly discriminative genes is a challenging task and existing hybrid methods fail to deal with efficiently. To mitigate the curse of dimensionality problem and to improve the interpretability of discriminative genes, in this study, we developed a new hybrid wrapper approach which integrates the characteristics of teaching learning-based algorithm (TLBO) and gravitational search algorithm (GSA), called TLBOGSA. A new encoding strategy is also integrated into TLBOGSA to transmute the continuous search space to binary search space and form binary TLBOGSA. In the proposed method, firstly, minimum redundancy maximum relevance (mRMR) feature selection is employed to select relevant genes from the gene expression datasets. Then, wrapper method is applied to select the informative genes from the reduced data produced by mRMR. To improve the search capability during the evolution process, we have incorporated the gravitational search mechanism in the teaching phase. The proposed method uses naive bayes classifier as a fitness function to select the extremely judicious genes which can help to classify cancer accurately. The efficiency of proposed method is tested on ten biological datasets and compared with state-of-art computational intelligence approaches for tumor prediction. Experimental results and statistical analysis demonstrate that proposed method is significantly outperforms existing metaheuristic approaches regarding convergence rate, classification accuracy and optimal number of feature sets. The proposed method reaches above 98% classification accuracy in six datasets and maximum accuracy is achieved as 99.62% in DLBCL dataset.

Teaching learning-based whale optimization algorithm for multi-layer perceptron neural network training.

This paper presents an improved teaching learning-based whale optimization algorithm (TSWOA) used the simplex method. First of all, the combination of WOA algorithm and teaching learning-based algorithm not only achieves a better balance between exploration and exploitation of WOA, but also makes whales have self-learning ability from the biological background, and greatly enriches the theory of the original WOA algorithm. Secondly, the WOA algorithm adds the simplex method to optimize the current worst unit, averting the agents to search at the boundary, and increasing the convergence accuracy and speed of the algorithm. To evaluate the performance of the improved algorithm, the TSWOA algorithm is employed to train the multi-layer perceptron (MLP) neural network. It is a difficult thing to propose a well-pleasing and valid algorithm to optimize the multi-layer perceptron neural network. Fifteen different data sets were selected from the UCI machine learning knowledge and the statistical results were compared with GOA, GSO, SSO, FPA, GA and WOA, severally. The statistical results display that better performance of TSWOA compared to WOA and several well-established algorithms for training multi-layer perceptron neural networks.

Optimization of groundwater resource for balanced cropping pattern

Abstract In this paper, an optimization model is formulated and optimal cropping pattern is suggested. Conjunctive use of water is not feasible for the study area as groundwater is the only source to fulfil irrigation demand. Water resources for the study area are limited. Best utilization of available water resources for increased net benefits is always advisable. Sinnar Taluka of Nasik district of Maharashtra state in India is considered as a study area. An existing cropping pattern is studied extensively and a new cropping pattern is suggested. Teaching–learning-based algorithm (TLBO) and particle swarm optimization (PSO) algorithm are applied to solve the optimization model. TLBO algorithm provides higher net returns as compared to PSO algorithm. TLBO and PSO saves up to 16.52% of water and benefits are increased by 35.81%. The study area is overexploited, this is fact. The new cropping pattern is suggested by considering minimum rainfall, i.e., 400 mm, so that in years when rainfall is above minimum rainfall the groundwater levels will be raised. In the proposed cropping pattern a few crop areas are majorly increased and a few crop areas are majorly reduced. The remaining crop areas are minorly increased or reduced and net benefits are increased and water demand decreased.

Parameter identification of solid oxide fuel cells with ranking teaching-learning based algorithm

The performance of a solid oxide fuel cell (SOFC) is tightly related to relevant parameters associated with the internal multi-physicochemical processes. Accurate identification of these parameters is considerably important for modelling the voltage versus current (V-I) characteristic of SOFCs. In this paper, an improved teaching-learning based algorithm (TLBO) referred to as RTLBO is proposed to identify the exact values for these parameters. The parameter identification of SOFCs is transformed into a minimization optimization problem. The mean square error (MSE) between the measured output voltage and the calculated output voltage is used as the objective function. TLBO has been shown to be competitive with other population-based algorithms. However, its convergence rate is relatively slow especially for complex optimization problems. Inspired by the ranking mechanism in the actual scenarios of teaching-learning process, a ranking based learner selection method is proposed and integrated into both the teacher and learner phases of RTLBO. In RTLBO, poor learners are more likely to be eliminated from the current class in the ranking based teacher phase and good learners are more likely to be chosen to interact with others in the ranking based learner phase, which hence can improve the overall performance of the class quickly. The experimental results on a 5-kW SOFC stack comprehensively demonstrate that RTLBO is able to achieve a better trade-off between the exploration and exploitation compared with twelve advanced TLBO variants and eight popular advanced non-TLBO based methods. In addition, the sensitivity of RTLBO to variations of population size is empirically investigated.

Solution of unit commitment problem using quasi-oppositional teaching learning based algorithm

In this article, quasi-oppositional teaching learning based algorithm (QOTLBO) is proposed to solve thermal unit commitment (UC) problem. Teaching learning based algorithm (TLBO) is a recently developed meta-heuristic algorithm based on the effect of the influence of a teacher on the output of learners in a class. The objective of UC is to economically schedule generating units over a short-term planning horizon subjugating to the forecasted demand and other system operating constraints in order to meet the load demand and spinning reserve for each interval. The proposed method is implemented and tested using MATLAB programming. The tests are carried out using 10-unit system during a scheduling period of 24h for four different cases. Additionally, the QOTLBO algorithm is also carried out for large scale power systems viz. 20, 60, 80 and 100 units to prove the scalability of the algorithm. The results confirm the potential and effectiveness of the proposed algorithm after comparison with various methods such as, simulated annealing (SA), genetic algorithm (GA), evolutionary programming (EP), differential evolution (DE), particle swarm optimization (PSO), improved PSO (IPSO), hybrid PSO (HPSO), binary coded PSO (BCPSO), quantum-inspired evolutionary algorithm (QEA), improved quantum-inspired evolutionary algorithm (IQEA), Muller method, quadratic model (QM), iterative linear algorithm (ILA), binary real coded firefly algorithm (BRCFF) and basic TLBO.