بهبود تولید بیودیزل با کمک اولتراسونیک حاصل از ضایعات صنعت گوشت (چربی خوک) با استفاده از نانوکاتالیزور اکسید مس سبز: مقایسه سطح پاسخ و مدل سازی شبکه عصبی
سال انتشار: 2021 - تعداد صفحات فایل pdf انگلیسی: 11 - تعداد صفحات فایل doc فارسی: 25
سوخت زیستی سبز ، تمیز و پایدار تنها گزینه به منظور کاهش کابرد سوخت های فسیلی ، پاسخگویی به تقاضای زیاد انرژی و کاهش آلودگی هوا است. تولید بیودیزل زمانی ارزان می شود که از یک پیش ماده ارزان ، کاتالیزور سازگار با محیط زیست و فرآیند مناسب استفاده کنیم. پیه خوک از صنعت گوشت حاوی اسید چرب بالا است و به عنوان یک پیش ماده موثر برای تهیه بیودیزل کاربرد دارد. این مطالعه بیودیزل را از روغن پیه خوک از طریق فرآیند استری سازی دو مرحله ای با کمک اولتراسونیک و کاتالیزور تولید می کند. عصاره Cinnamomum tamala (C. tamala) برای تهیه نانوذرات CuO مورد استفاده قرار گرفت و با استفاده از طیف مادون قرمز ، پراش اشعه ایکس ، توزیع اندازه ذرات ، میکروسکوپ الکترونی روبشی و انتقال مشخص شد. تولید بیودیزل با استفاده از طرح Box-Behnken (BBD) و شبکه عصبی مصنوعی (ANN) ، در محدوده متغیرهای زمان اولتراسونیک (us )(20-40 min)، بارگیری نانوکاتالیزور 1-3) CuO درصد وزنی( ، و متانول به قبل از نسبت مولی PTO (10:1e30:1) مدلسازی شد. آنالیز آماری ثابت کرد که مدل سازی شبکه عصبی بهتر از BBD است. عملکرد بهینه 97.82٪ با استفاده از الگوریتم ژنتیک (GA) در زمان US: 35.36 دقیقه ، بار کاتالیزور CuO: 2.07 درصد وزنی و نسبت مولی: 29.87: 1 به دست آمد. مقایسه با مطالعات قبلی ثابت کرد که اولتراسونیک به میزان قابل توجهی موجب کاهش بار نانوکاتالیزور CuO می شود ، و نسبت مولی را افزایش می دهد و این فرایند را بهبود می بخشد.
کلمات کلیدی: چربی خوک | التراسونیک | اکسید مس | سنتز سبز | شبکه عصبی | سطح پاسخ
|مقاله ترجمه شده|
A self-learning genetic algorithm based on reinforcement learning for flexible job-shop scheduling problem
یک الگوریتم ژنتیک خودآموز مبتنی بر یادگیری تقویتی برای مسئله زمان بندی انعطاف پذیر مشاغل فروشگاهی -2020
As an important branch of production scheduling, flexible job-shop scheduling problem (FJSP) is difficult to solve and is proven to be NP-hard. Many intelligent algorithms have been proposed to solve FJSP, but their key parameters cannot be dynamically adjusted effectively during the calculation process, which causes the solution efficiency and quality not being able to meet the production requirements. Therefore, a self-learning genetic algorithm (SLGA) is proposed in this paper, in which genetic algorithm (GA) is adopted as the basic optimization method and its key parameters are intelligently adjusted based on reinforcement learning (RL). Firstly, the selflearning model is analyzed and constructed in SLGA, SARSA algorithm and Q-Learning algorithm are applied as the learning methods at initial and later stages of optimization, respectively, and the conversion condition is designed. Secondly, the state determination method and reward method are designed for RL in GA environment. Finally, the learning effect and performance of SLGA in solving FJSP are compared with other algorithms using two groups of benchmark data instances with different scales. Experiment results show that the proposed SLGA significantly outperforms its competitors in solving FJSP.
Keywords: Flexible job-shop scheduling problem (FJSP) | Self-learning genetic algorithm (SLGA) | Genetic algorithm (GA) | Reinforcement learning (RL)
Optimization strategies for Microgrid energy management systems by Genetic Algorithms
استراتژی های بهینه سازی برای سیستم های مدیریت انرژی میکرو گرید توسط الگوریتم های ژنتیک-2020
Grid-connected Microgrids (MGs) have a key role for bottom-up modernization of the electric distribution network forward next generation Smart Grids, allowing the application of Demand Response (DR) services, as well as the active participation of prosumers into the energy market. To this aim, MGs must be equipped with suitable Energy Management Systems (EMSs) in charge to efficiently manage in real time internal energy flows and the connection with the grid. Several decision making EMSs are proposed in literature mainly based on soft computing techniques and stochastic models. The adoption of Fuzzy Inference Systems (FISs) has proved to be very successful due to their ease of implementation, low computational run time cost, and the high level of interpretability with respect to more conventional models. In this work we investigate different strategies for the synthesis of a FIS (i.e. rule based) EMS by means of a hierarchical Genetic Algorithm (GA) with the aim to maximize the profit generated by the energy exchange with the grid, assuming a Time Of Use (TOU) energy price policy, and at the same time to reduce the EMS rule base system complexity. Results show that the performances are just 10% below to the ideal (optimal) reference solution, even when the rule base system is reduced to less than 30 rules.
Keywords: Microgrids | Genetic algorithms | Fuzzy systems | Energy management systems
XCS with opponent modelling for concurrent reinforcement learners
XCS با مدل سازی حریف برای یادگیرنده تقویتی همزمان-2020
Reinforcement learning (RL) of optimal policies against an opponent agent also with learning capabil- ity is still challenging in Markov games. A variety of algorithms have been proposed for solving this problem such as the traditional Q-learning-based RL (QbRL) algorithms as well as the state-of-the-art neural-network-based RL (NNbRL) algorithms. However, the QbRL approaches have poor generalization capability for complex problems with non-stationary opponents, while the learned policies by NNbRL al- gorithms are lack of explainability and transparency. In this paper, we propose an algorithm X-OMQ( λ) that integrates eXtended Classifier System (XCS) with opponent modelling for concurrent reinforcement learners in zero-sum Markov Games. The algorithm can learn general, accurate, and interpretable action selection rules and allow policy optimization using the genetic algorithm (GA). Besides, the X-OMQ( λ) agent optimizes the established opponent’s model while simultaneously learning to select actions in a goal-directed manner. In addition, we use the eligibility trace mechanism to further speed up the learn- ing process. In the reinforcement component, not only the classifiers in the action set are updated, but other relevant classifiers are also updated in a certain proportion. We demonstrate the performance of the proposed algorithm in the hunter prey problem and two adversarial soccer scenarios where the op- ponent is allowed to learn with several benchmark QbRL and NNbRL algorithms. The results show that our method has similar learning performance with the NNbRL algorithms while our method requires no prior knowledge of the opponent or the environment. Moreover, the learned action selection rules are also interpretable while having generalization capability.
Keywords: Opponent modelling | XCS | Markov games | Reinforcement learning
Mean-conditional value at risk model for the stochastic project scheduling problem
ارزش میانگین شرطی در مدل ریسک برای مسئله برنامه ریزی تصادفی پروژه-2020
Every project faces different opportunities and risks during its lifecycle. Risks are the factors that can disrupt the successful implementation of projects and cause failure in achieving project goals. Advancing the project while considering its risks is one of the most essential aspects of project management. Planning and scheduling can be applied in a way that reduces the risks in the management of projects. In this paper, a new scenario-based meanconditional value-at-risk (CVaR) model is developed to minimize the risk of the project’s net present value (NPV). Moreover, the trade-off between expected NPV and the risk of NPV is considered in this study. Start time policies are also used to specify the start times of project activities. Two multi-objective optimization algorithms including Non-dominated Sorting Genetic Algorithms (NSGA-II), and Multi-Objective Vibration Damping Optimization (MOVDO) are applied to identify the Pareto optimal solution. The efficiency of the algorithms is assessed based on some performance criteria. The results of the computational experiments show that at identical run time MOVDO functions better in terms of hypervolume indicator, while NSGA- II better results in other performance metrics
Keywords: Stochastic project scheduling | Conditional-value-at risk | Scheduling policy
Optimal scheduling of a renewable based microgrid considering photovoltaic system and battery energy storage under uncertainty
برنامه ریزی بهینه از یک میکروگرید مبتنی بر قابل تجدید با توجه به سیستم فتوولتائیک و ذخیره انرژی باتری در عدم قطعیت-2020
This paper suggests a new energy management system for a grid-connected microgrid with various renewable energy resources including a photovoltaic (PV), wind turbine (WT), fuel cell (FC), micro turbine (MT) and battery energy storage system (BESS). For the PV system operating in the microgrid, an innovative mathematical modelling is presented. In this model, the effect of various irradiances in different days and seasons on day-ahead scheduling of the microgrid is evaluated. Moreover, the uncertainties in the output power of the PV system and WT, load demand forecasting error and grid bid changes for the optimal energy management of microgrid are modelled via a scenario-based technique. To cope with the optimal energy management of the grid-connected microgrid with a high degree of uncertainties, a modified bat algorithm (MBA) is employed. The proposed algorithm leads to a faster computation of the best location and more accurate result in comparison with the genetic algorithm (GA) and particle swarm optimization (PSO) algorithm. The simulation results demonstrate that the use of practical PV model in a real environment improve the accuracy of the energy management system and decreases the total operational cost of the grid-connected microgrid.
Keywords: Photovoltaic | Energy management | Battery energy storage system | Uncertainty | Optimization | Microgrid
Complete coverage path planning using reinforcement learning for Tetromino based cleaning and maintenance robot
برنامه ریزی کامل مسیر پوشش با استفاده از یادگیری تقویتی برای تمیز کاری و نگهداری ربات مبتنی بر Tetromino-2020
Tiling robotics have been deployed in autonomous complete area coverage tasks such as floor cleaning, building inspection, and maintenance, surface painting. One class of tiling robotics, polyomino-based reconfigurable robots, overcome the limitation of fixed-form robots in achieving high-efficiency area coverage by adopting different morphologies to suit the needs of the current environment. Since the reconfigurable actions of these robots are produced by real-time intelligent decisions during operations, an optimal path planning algorithm is paramount to maximize the area coverage while minimizing the energy consumed by these robots. This paper proposes a complete coverage path planning (CCPP) model trained using deep blackreinforcement learning (RL) for the tetromino based reconfigurable robot platform called hTetro to simultaneously generate the optimal set of shapes for any pretrained arbitrary environment shape with a trajectory that has the least overall cost. To this end, a Convolutional Neural Network (CNN) with Long Short Term Memory (LSTM) layers is trained using Actor Critic Experience Replay (ACER) reinforcement learning algorithm. The results are compared with existing approaches which are based on the traditional tiling theory model, including zigzag, spiral, and greedy search schemes. The model is also compared with the Travelling salesman problem (TSP) based Genetic Algorithm (GA) and Ant Colony Optimization (ACO) schemes. The proposed scheme generates a path with lower cost while also requiring lesser time to generate it. The model is also highly robust and can generate a path in any pretrained arbitrary environments.
Keywords: Tiling robotics | Cleaning and maintenance | Inspection | Path planing | Reinforcement learning
Application of optimized Artificial and Radial Basis neural networks by using modified Genetic Algorithm on discharge coefficient prediction of modified labyrinth side weir with two and four cycles
استفاده از شبکه های عصبی بهینه سازی شده مصنوعی و شعاعی با استفاده از الگوریتم ژنتیک اصلاح شده بر پیش بینی ضریب تخلیه ریزگرد سمت اصلاح شده با دو و چهار چرخه-2020
Determining the discharge coefficient is one of the most important processes in designing side weirs. In this study, the structure of Artificial Neural Network (ANN) and Radial Basis Neural Network (RBNN) methods are optimized by a modified Genetic Algorithm (GA). So two new hybrid methods of Genetic Algorithm Artificial neural network (GAA) and Genetic Algorithm Radial Basis neural network (GARB), were introduced and compared with each other. The modified GA was used to find the neuron number in the hidden layers of the ANN and to find the spread value and the neuron number of the RBNN method, as well. GAA and GARB were tested for predicting the discharge coefficient of a modified labyrinth side weir he GARB method could successfully predict the accurate discharge coefficient even in cases where there is a limited number of train datasets available.
Keywords: Artificial neural network | Discharge coefficient | Hybrid model | Labyrinth side weir | Modified | Genetic algorithm | Radial basis neural network
Research on BP network for retrieving extinction coefficient from Mie scattering signal of lidar
تحقیقات بر روی شبکه BP برای بازیابی ضریب خاموشی از سیگنال پراکندگی میای LIDAR-2020
Mie lidar is a powerful tool for detecting the optical properties of atmospheric aerosols. However, there are two unknown parameters in the Mie lidar equation: the extinction coefficient and the backscattering coefficient. In the common methods for solving the equation, it is necessary to make assumptions about the relationship between the two unknown parameters. These assumptions will reduce the detection precision of extinction coefficient. In view of this, the back propagation (BP) neural network is used to retrieve extinction coefficient from the Mie scattering signal of lidar. Firstly, the structure and main parameters of the BP network are designed according to the practical application. In order to improve the convergence speed and prevent falling into local minima, the initial weights and thresholds of BP network are optimized by genetic algorithm (GA). Then the GA-BP network is trained with Mie scattering signal and the extinction coefficient retrieved by Raman method. Thus the mathematical relationship between Mie scattering signal and the extinction coefficient is stored in the BP network. The trained GA-BP network is then used to retrieve the extinction coefficient from Mie scattering signal in different conditions and the applicability of the GA-BP network is researched. The research will promote the development of Mie lidar retrieving algorithm.
Keywords: Aerosol | Mie scattering | Lidar | Extinction coefficient | BP network | Genetic algorithm
AI-based optimization of PEM fuel cell catalyst layers for maximum power density via data-driven surrogate modeling
بهینه سازی مبتنی بر هوش مصنوعی لایه های کاتالیزور سلول سوختی PEM برای حداکثر چگالی توان از طریق مدل سازی جایگزین داده محور-2020
Catalyst layer (CL) is the core electrochemical reaction region of proton exchange membrane fuel cells (PEMFCs). Its composition directly determines PEMFC output performance. Existing experimental or modeling methods are still insufficient on the deep optimization of CL composition. This work develops a novel artificial intelligence (AI) framework combining a data-driven surrogate model and a stochastic optimization algorithm to achieve multi-variables global optimization for improving the maximum power density of PEMFCs. Simulation results of a three-dimensional computational fluid dynamics (CFD) PEMFC model coupled with the CL agglomerate model constitutes the database, which is then used to train the data-driven surrogate model based on Support Vector Machine (SVM), a typical AI algorithm. Prediction performance shows that the squared correlation coefficient (R-square) and mean percentage error in the test set are 0.9908 and 3.3375%, respectively. The surrogate model has demonstrated comparable accuracy to the physical model, but with much greater computation- resource efficiency: the calculation of one polarization curve will be within one second by the surrogate model, while it may cost hundreds of processor-hours by the physical CFD model. The surrogate model is then fed into a Genetic Algorithm (GA) to obtain the optimal solution of CL composition. For verification, the optimal CL composition is returned to the physical model, and the percentage error between the surrogate model predicted and physical model simulated maximum power densities under the optimal CL composition is only 1.3950%. The results indicate that the proposed framework can guide the multi-variables optimization of complex systems.
Keywords: Proton exchange membrane fuel cell | Catalyst layer composition | Agglomerate model | Data-driven surrogate model | Stochastic optimization algorithm