(Luận văn thạc sĩ) áp dụng thuật toán meta heuristic điều độ tối ưu hệ thống điện

BỘ GIÁO DỤC VÀ ĐÀO TẠO TRƯỜNG ĐẠI HỌC SƯ PHẠM KỸ THUẬT THÀNH PHỐ HỒ CHÍ MINH LUẬN VĂN THẠC SĨ HỒ HỮU THÌN ÁP DỤNG THUẬT TỐN META-HEURISTIC ĐIỀU ĐỘ TỐI ƯU HỆ THỐNG ĐIỆN NGÀNH: KỸ THUẬT ĐIỆN - 60520202 SKC 0 Tp Hồ Chí Minh, tháng 03/2017 BỘ GIÁO DỤC VÀ ĐÀO TẠO TRƢỜNG ĐẠI HỌC SƢ PHẠM KỸ THUẬT THÀNH PHỐ HỒ CHÍ MINH LUẬN VĂN THẠC SĨ HỌ VÀ TÊN HỌC VIÊN HỒ HỮU THÌN TÊN ĐỀ TÀI LUẬN VĂN THẠC SĨ ÁP DỤNG THUẬT TOÁN META-HEURISTIC ĐIỀU ĐỘ TỐI ƢU HỆ THỐNG ĐIỆN NGÀNH: KỸ THUẬT ĐIỆN Hướng dẫn khoa học: PGS.TS VÕ NGỌC ĐIỀU Tp Hồ Chí Minh, tháng năm 2017 LÝ LỊCH KHOA HỌC I LÝ LỊCH SƠ LƢỢC: Họ & tên: HỒ HỮU THÌN Giới tính: Nam Ngày, tháng, năm sinh: 08/11/1989 Nơi sinh: Nghệ an Quê quán: Quỳnh lưu, Nghệ an Dân tộc: Kinh Chỗ riêng địa liên lạc: 127 Chương Dương, Linh Chiểu, Thủ Đức, TP.HCM Điện thoại quan: Điện thoại nhà riêng:0979241145 Fax: E-mail: huuthin08@gmail.com II QUÁ TRÌNH ĐÀO TẠO: Trung học chuyên nghiệp: Hệ đào tạo: Thời gian đào tạo từ ……/…… đến ……/ …… Nơi học (trường, thành phố): Ngành học: i Đại học: Hệ đào tạo: Chính Quy Thời gian đào tạo từ 9/2012 đến 3/ 2015 Nơi học (trường, thành phố): Trường ĐHSPKT TP.HCM Ngành học: Kỹ Thuật Điện Tên đồ án, luận án môn thi tốt nghiệp: Thiết kế nghịch lƣu pha hòa lƣới điều khiển DSP Ngày & nơi bảo vệ đồ án, luận án thi tốt nghiệp: Người hướng dẫn: Th.S Trần Quang Thọ III Q TRÌNH CƠNG TÁC CHUN MƠN KỂ TỪ KHI TỐT NGHIỆP ĐẠI HỌC: Thời gian 3/20159/2015 9/20153/2016 3/20168/2016 Nơi công tác Công việc đảm nhiệm Greytone data system viet nam Kỹ thuật Việt đông nam Giám sát Công ty cổ phần ngoại thương ME 8/2016- Nay Công ty cổ phần BDS Vinhomes i Kỹ thuật LỜI CAM ĐOAN Tơi cam đoan cơng trình nghiên cứu Các số liệu, kết nêu luận văn trung thực chưa công bố cơng trình khác Tp Hồ Chí Minh, ngày tháng năm 2017 (Ký tên ghi rõ họ tên) HỒ HỮU THÌN ii CẢM TẠ Trong thời gian bắt đầu học đến nay, em nhận nhiều quan tâm, giúp đỡ q Thầy Cơ, gia đình, đồng nghiệp bạn bè Với lòng biết ơn chân thành sâu sắc nhất, em xin gửi đến quý Thầy Cô Trường Đại học Sư phạm Kỹ thuật Tp.HCM, đặc biệt quan tâm giúp đỡ Thầy hướng dẫn PGS.TS Võ Ngọc Điều với tri thức tâm huyết để truyền đạt vốn kiến thức quý báu cho em thực đề tài nghiên cứu Em xin cảm ơn thầy ThS.Nguyễn Trung Thắng tận tình giúp đỡ hướng dẫn trình thực Luận văn Em chân thành cảm ơn thầy cô giáo Khoa Điện – Điện Tử cung cấp kiến thức, phòng Đào tạo sau ĐH trường Đại Học Sư Phạm Kỹ Thuật TP HCM tạo điều kiện suốt trình học tập trường, góp ý nhiều ý kiến quý báu để em hoàn thành tốt luận văn Mặc dù thời gian qua cố gắng nỗ lực nghiên cứu kiến thức cịn nhiều hạn chế nên khơng thể tránh thiếu sót Kính mong thầy hội đồng khoa học xem xét góp ý, chỉnh sửa để luận văn hoàn thiện tốt nghiên cứu sau TP HCM, tháng 03 năm 2017 Học viên thực HỒ HỮU THÌN iii TĨM TẮT Vận hành tối ưu tổ máy phát (OOGTU) toán quan trọng hệ thống điện giá nhiên liệu hóa thạch đắt đỏ nguồn lượng cạn kiệt tương lai Nhiệm vụ vận hành tổ máy xác định công suất tổ máy sẵn có để cực tiểu chi phí nhiên liệu phát điện ràng buộc từ tổ máy nhiệt điện hệ thống đảm bảo Trong luận văn này, áp dụng thuật toán Dơi (BA) giải toán vận hành tối ưu tổ máy nhiệt điện với tổn thất công suất xem xét Mặc dù BA áp dụng thành công chất lượng lời giải tỉ lệ thành cơng cịn chưa cao Do đó, việc cải thiện BA điều cần thiết để có thuật tốn Dơi cải tiến (MBA) BA có trở ngại dễ rơi vào vùng cực trị địa phương đạt lời giải không khả thi Ở thuật toán BA, tần số khởi tạo ngẫu nhiên hạn chế việc tìm kiếm khơng mở rộng vùng tìm kiếm giai đoạn đầu co hẹp vùng tìm kiếm giai đoạn cuối Do đó, tần số giảm dần số vòng lặp tăng dần cải biên thứ cho MBA Bên cạnh đó, lần tạo nghiêm thứ hai BA dựa vào bước nhảy ngẫu nhiên mà không dựa sở thực tế kết đạt Từ đo, cải biên thứ hai BA tạo bước nhảy dựa hai nghiệm có MBA BA kiểm tra ba hệ thống khác có xét đến tổn thất công suất hệ thống gồm hệ thống ba, sáu hai mươi tổ máy So sánh kết BA MBA MBA hiệu BA nhiều đạt chất lượng lời giải tốt hơn, tỉ lệ thành công cao độ ổn định lời giải tốt So sánh với phương pháp khác, MBA trội đạt lời giải tốt xấp xỉ thời gian tính tốn nhanh Từ kết luận MBA hiệu cho toán OOTGU iv ABSTRACT Optimal operation of thermal generating units (OOTGU) is an important problem in power system since the price of fossil fuel is expensive and the amount of the energy will be exhausted in the future The task of operation of thermal generating units is to determine the power output of the set of available units so that the objective of minimization of electricity generation of fuel cost is yielded while all constraints from thermal units and power system are completely met In the thesis, we apply Bat algorithm for solving the optimal operation of thermal generating unit problem in which power balance constraint with power losses is considered Although BA is successfully applied for solving the OOTGU problem, the quality of optimal solution and the success rate are still low Consequently, the improvement of BA is really necessary to obtain a new version of modified Bat algorithm (MBA) The disadvantages that BA still suffers are to trap to local optimum and to converge to an in feasible solution In BA, the frequency of each Bat is randomly generated for each Bat at each iteration This way leads to the limitation to open the search space at the beginning of the search and narrow the search space at the end of the search Consequently, the first modification need to be performed for the proposed modified BA (MBA) is that the frequency is reduced when the iteration is increased Furthermore, in the second new solution generation, BA has used a random value to be a step size to update new solution This is not a good change for each new solution and lead to reduce the quality of solution So, the second modification is carried out to determine an appropriate step size The MBA and BA have been tested on three different systems with three units, six units and twenty units considering power loss in transmission line The result comparison between BA and MBA have indicated that MBA is much better than BA since it has obtained better solution, higher success rate and better stabilization of solutions Compared to other methods, MBA is superior since it has had better or approximate solution and faster execution time As a result, it is concluded that MBA is very efficient for the problem of OOTGU iv CƠNG TRÌNH ĐƯỢC HỒN THÀNH TẠI TRƢỜNG ĐẠI HỌC SƢ PHẠM KỸ THUẬT THÀNH PHỐ HỒ CHÍ MINH Cán hướng dẫn khoa học : (Ghi rõ họ, tên, chức danh khoa học, học vị chữ ký) Cán chấm nhận xét 1: (Ghi rõ họ, tên,, chức danh khoa học, học vị chữ ký) Cán chấm nhận xét 2: (Ghi rõ họ, tên, chức danh khoa học, học vị chữ ký) Luận văn thạc sĩ bảo vệ trước HỘI ĐỒNG CHẤM BẢO VỆ LUẬN VĂN THẠC SĨ TRƯỜNG ĐẠI HỌC SƯ PHẠM KỸ THUẬT, Ngày tháng năm iv Hệ số tổn thất hệ thống (x10-3) 56 PHỤC LỤC B Bảng B1 Cơng suất tối ưu tìm cho hệ thống từ BA MBA Hệ TỔ MÁY TỔ MÁY 20 TỔ MÁY thống Unit BA MBA BA MBA BA MBA 436.1847 435.2236 100.4453 100.0001 561.0671 513.4844 299.8118 299.9684 100.0175 100.0008 178.027 166.0462 129.8235 130.6368 50.0119 50.0003 144.9659 126.6985 - - 304.9511 305.7229 117.404 102.4436 - - 134.5506 122.1925 120.7154 114.4381 - - 110.0237 122.0834 76.1194 72.6718 - - - - 97.6916 115.9593 - - - - 128.8088 116.9583 - - - - 90.8568 99.8427 10 - - - - 37.7599 108.0182 11 - - - - 152.3878 151.3077 12 - - - - 280.4063 293.1104 13 - - - - 100.478 120.0397 14 - - - - 54.068 30.4631 15 - - - - 111.1834 116.8039 16 - - - - 36.4114 36.2222 17 - - - - 70.7449 67.0181 18 - - - - 80.8103 87.0738 19 - - - - 110.6709 100.1917 20 - - - - 42.4298 53.3253 57 BÀI BÁO Modified Bat Algorithm for Combined Economic and emission Dispatch Problem Ly Huu Pham1*, Thin Huu Ho2, Thang Trung Nguyen1, and Dieu Ngoc Vo3 Faculty of Electrical and Electronics Engineering, Ton Duc Thang University, Vietnam 19, Nguyen Huu Tho str., 7th dist., HCM City, Vietnam Faculty of Electrical and Electronics Engineering, University of Technical Education HCM City Vo Van Ngan str., Thu Duc dist., HCM City, Vietnam Department of Power Systems, HCM City University of Technology 268 Ly Thuong Kiet str., 10th dist., HCM City, Viet Nam huuthin08@gmail.com@gmail.com, nguyentrungthang@tdt.edu.vn, vndieu@gmail.com *Corresponding author: phamhuuly@tdt.edu.vn Abstract: The paper proposes a new modified Bat algorithm (MBA) for solving combined economic and emission load dispatch (CEED) problems where transmission power losses are considered The MBA is first developed in the paper by modifying the several modifications on the conventional Bat algorithm (BA) in aim to improve the performance of the BA The MBA is tested on two different systems with the transmission power losses The performance of the MBA is evaluated by comparing obtained results with BA and other existing algorithms available in the study As a result, it can be concluded that the MBA outperforms the BA and is very strong for solving the CEED problem Keywords: Modified Bat algorithm, transmission power losses, economic load dispatch, emission dispatch, combined economic and emission dispatch Introduction BÀI BÁO The main task of the CEED problem is to determine the optimal power output of thermal units so that the fuel cost and emission can be minimized significantly while exactly meeting all constraints from current set of units and power system such as limitations on capacity of thermal units, power balance constraints considering power losses in transmission lines The fact that fossil fuels will become exhausted in the near future because people have exploited and used the fuels with a large amount yearly Furthermore, during the process of generating electricity, polluted emissions such as NOx, SO2 and CO2 are released into the air without control Consequently, the purpose of minimization of both fuel cost and emission has a significantly high role in the power systems Due to the importance of the problem, a huge number of researchers have been attracted and published a lot of papers so far such as Improved Hopfield Neural Network Model (IHNN) [1], Tabu Search (TS) [2], fuzzy logic controlled genetic algorithm (FCGA) [3], the Non-dominated Sorting Genetic Algorithm - II (NSGA-II) [4], Differential Evolution (DE) [5], Genetic algorithm (GA) [6], Particle swarm optimization (PSO) [6], biogeography-based optimization (BBO) [7], pareto differential evolution (PDE) [8], nondominated sorting genetic algorithm-II (NSGA-II) [8], strength pareto evolutionary algorithm (SPEA 2) [8], Hybrid Differential evolutionsequential quadratic programming (DE-SQP) [8], Hybrid Particle Swarm optimization- sequential quadratic programming (PSO-SQP) [9], parallel synchronous PSO algorithm (PSPSO) [10], ABC_PSO [11], multi-objective cultural algorithm (MOCA) [12], Basic Cuckoo Search Algorithm (CSA) [13], Lambda method (LM) [14], Hopfield Lagrange Network (HNN) [14], and flower pollination algorithm (FPA) [15] Among these considered methods, IHNN [1], LM [14] and HNN [14] belong to the family of deterministic algorithms where other ones are included in the meta-heuristic algorithms A big difference between the two ones is the way for optimal solution obtained In fact, the comer owns only one solution and the solution tends to be improved gradually when the search process goes to the end whereas the latter consists of a set of solutions, which are improved by evaluating fitness function When the search process of the comer ends, it means that all the BÀI BÁO constraints can be exactly met; however, the manner is not always true for the latter The latter’s stopping criteria is always based on the maximum number of iterations and all constraints can be exactly satisfied although the search process does not terminate On the other hand, it can be concluded that the applicability of the meta-heuristic algorithm is more potential than the deterministic ones because the deterministic ones must stop dealing with problems where nonconvex function as well as non-differential functions are included meanwhile the meta-heuristics can solve easily Bat Algorithm (BA), a new meta-heuristic inspired from the behavior of Bat for searching food, was developed by Xin-She Yang in 2010 [16] The BA has been implemented for solving the CEED problem [17] where three dispatch cases including economic emission dispatch, emission dispatch and combined emission and economic dispatch are considered The performance of method has been validated by comparing with other methods; however, the optimal solution and the execution time still suffer from low quality and slow convergence Consequently, a modified version of the BA (MBA) is developed in the paper for solving the considered problem based on implementing several improvements on the conventional BA The first improvement is to focus on frequency of each Bat In the BA, the values of each Bat is generated randomly ranging from zero to two, leading to an unstable updated step for updating optimal solution The improvement will enable the MBA to narrow the search zone when the process of search goes to the end of the process The second improvement is to focus on the loudness, which is set to the range of values from zero to one in the BA In the MBA, it is fixed at one to guarantee all better solution of the previous generation must be kept The improvement can reduce a huge number of trials for selecting the loudness values The final improvement aim to restrict the limitations of velocity for each bat In BA, there is no limit for the value and therefore, it results in the assignment to either upper or lower value of variables To validate the effectiveness and robustness of the MBA, two systems with three units and power losses in transmission lines are employed BÀI BÁO and the results from the systems are the evidence to be compared The comparisons among the MBA and other ones indicate that the MBA is more efficient than BA and other ones Problem Formulation 2.1 Objective function In the CEED problem, the fuel cost and polluted emission of each generating unit are minimized, thus the objective of the problem is written as: N N i 1 i 1 Min  Fi    w1F1i ( Pi )  w2 PR.F2i ( Pi ) (1) where w1 and w2 are weight factor associate with fuel cost and emission; PR is the price penalty factor [18]; F1i and F1i are the fuel cost function and emission function The fuel cost function and emission function are represented by: (2) Fi ( Pi )   bi Pi  ci Pi N F2i   (a ei Pi  bei Pi  cei ) (3) i 1 where N is the number of generators; Pi is real power output of generator i; Pi,min is minimum power output of unit i; ai, bi, ci and aei, bei, cei are fuel cost and emission coefficients of unit i, respectively 2.2 Constraints Real power balance: The total real power output of generating units satisfies total load demand plus system power losses N P  P i D  PL i 1 where PD is total system load demand; PL is total transmission loss (4) BÀI BÁO Generator capacity limits: The real power output of generating units should be limited between their upper and lower bounds represented by: Pi ,min  Pi  Pi ,max (5) where Pi,min and Pi,max are maximum and minimum power outputs of unit i Modified Bat Algorithm for CEED Problem 3.1 Conventional Bat Algorithm Bat algorithm is mathematically constructed by three idealized rules as below 1) Each Bat d has a position Xd and a velocity Vd and it generates a pulse with frequency fd and loudness Ad and rate rd 2) The loudness Ad that a Bat generates is the largest value at the beginning of the search process and it reduces gradually and reaches the lowest one as approached to the food or prey 3) During the search process, the frequency fd is randomly produced in eq (6) meanwhile pule rate rd and loudness Ad of each pule are automatically tuned as in eqs (7) and (8) At the beginning of the search process, the frequency of each Bat is randomly initialized within its minimum and maximum values: f dG  f  rand *( f max  f ) (6) where fmax and fmin are the maximum and minimum frequencies and set to and 0; The new velocity and the position of each bat are then updated as bellows VdG  VdG 1  f dG  X dG  X * ; d  1, , NB (7) X dG  X dG 1  VdG ; d  1, , NB (8) BÀI BÁO where X* is the global best position of a Bat among the population For local search, each current solution is selected to generate a new solution around the current one X new  X old   AG (9) G where ε is a random num er ranging from to 1; and A is the average loudness The updated loudness and pulse rate are determined by 3.2 AdG   AdG 1 (10) r  r.[1  exp( G)], (11) Modified Bat Algorithm 1) Modification on frequency f dG  f max f max  f G Gmax (12) where G and Gmax are current iteration and maximum number of iterations 2) Modification on the loudness value: The loudness A in eq (10) is set to to guarantee that all better solutions are kept without the value of the loudness A 3) Modification on the velocity: In the MBA, the maximum and minimum velocity are adjusted based on to a range to limit the search zone and avoid the violation of solution Vd max  0.15*( X d max  X d ) Vd  Vd max Implementation of the MBA for the considered problem 4.1 Initialization (13) (14) BÀI BÁO A population of NB bats is represented by Xd (d = 1, , NB) where each solution corresponding to each at’s position given by X  [ P , P , , P ] The power outputs are randomly d 2, d 3, d N ,d initialized satisfying the limitation, Pi,min  Pi,d  Pi,max The fitness function of each bat for the considered problem is calculated as:  N FTd    w1 F1i ( Pi )  w2 PR.F2 i ( Pi )   K s  P1, d  P1lim i 1  (15) where Ks is the penalty factors; and P1,d is the power output of slack thermal unit [14] and obtained by using eq (4) The limit for slack thermal unit in (15) is determined as follows: P1 lim  P1,max    P1,min P  1, d if P1, d  P1,max if P1, d  P1,min (16) otherwise where P1,max and P1,min are the maximum and minimum power outputs of slack thermal unit 1, respectively 4.2 Updating new velocity and new position for each bat The adaptive frequency is determined by using (12) and then the velocity of each bat is updated by using eq (7) If the velocity violates its limitation, it will be set to its limits as below Vdmax if Vd  Vdmax  Vd  Vdmin if Vd  Vdmin ; d  1, , N p V otherwise  d (17) Finally, the position of each bat is then determined by employing eq (8) 4.3 Searching a new solution around the global best solution There is a possibility that each solution Xd is newly generated around its global best solution by using (9) if its random value ranging from to is less than pulse rate rd, which is fixed at a value at the beginning of the search process and increases at the following iteration Clearly, if the rd is predetermined at a high value approximately equaling 1, almost solutions are BÀI BÁO newly generated near the global best solution and the final optimal solution is able to fall in the local optimal one On the contrary, just a few solutions are newly generated nearby the global best one if the rd is set to low value nearly equaling zero Therefore, the value of the pulse rate is chosen in range from to with a step of 0.2 There are no criteria to ensure all the existing solutions satisfying their limits Therefore, they need to be checked and adjusted as the following rules  Pi ,max if Pi ,d  Pi ,max  Pi ,d   Pi ,min if Pi ,d  Pi ,min ; i  2, , N  P otherwise  i ,d 4.4 (18) Selection of new solution using loudness and fitness comparison All the current solutions are evaluated by calculating the fitness function and then the selection of between the current ones and the initialized solutions Better ones with lower fitness function are kept The best one with the lowest fitness is set to Global one 4.5 The termination criterion of the search process The search process will be terminate when the current iteration is equal to the maximum number of iterations Numerical results The proposed MBA is tested on two systems with three units and power losses in transmission lines The proposed algorithm is coded in Matlab platform and run 50 independent trials for each case on a 2.4 GHz PC with GB of RAM Case 1: Three-unit system with load of 850 MW and power losses in transmission lines In this case, three dispatch case including economic dispatch, emission BÀI BÁO dispatch and combined economic and emission dispatch are considered The data of the system are from [4] For implementation of the MBA to the cases, the number of bats and the maximum number of iterations are respectively set to 20 and 30 In addition, the conventional BA is also implemented for validating the superiority of the MBA over the BA Consequently, the two control parameters with the same values are set for the BA The result from BA and MBA compared to those from other for the three cases are reported in the table Table Comparison of results for case Dispatch Method TS [2] NSGA-II [4] BBO [7] CSA [13] BA MBA Cost ($) 8344.6 8344.6 8344.59 8344.59 8345.15 8344.59 Cpu (s) - - - 0.09 0.06 0.06 Em (kg) 0.0958 0.09593 0.09592 0.09592 0.0965 0.09592 Cpu (s) - - - 0.07 0.059 0.061 Cost ($) - 8349.72 8349.722 8349.248 8349.08 Em (kg) - 0.09654 0.09654 0.0966 0.1001 Cpu (s) - - 0.09 0.063 0.062 Eco Em CEED - It is clear from the table that MBA is much superior to BA for the first two cases and there is a trade-off for the remaining case where MBA obtains lower cost but higher emission Time for simulation between the two methods seems to be identical Compared to others, MBA obtains the same quality of solution as them except Tabu search for the emission dispatch case; however, MBA is very fast when compared to CSA In fact, MBA has taken only about 0.06 second for optimal each run meanwhile CSA has spent 0.09 second and there have no times reported for others BÀI BÁO Case 2: Three-unit system with load of 400 MW and power losses in transmission lines In the case, a three-unit system supplying to load demand of 400 MW is considered The data of the system are taken from [6] The control parameters for BA and MBA are set to the same values as in case Comparison from the results obtained by BA, MBA and other ones such as GA [6], PSO [6] and FPA [15] is reported in table As seen from the cost and emission from BA and MBA, MBA obtains less cost than BA, $ 0.0041 but the emission from the two ones are identical Comparison among the MBA and others in [6] and [15] reveals that there is a trade-off between cost and emission but the execution times are significantly different In fact, that time by MBA is only 0.06 second meanwhile that is 0.282 second for GA, 0.235 for PSO [6] and 0.175 for FPA The fact that the time from these methods are very low; however, MBA is faster than these methods from three time to five times In conclusion, it is stated that the MBA is superior to conventional BA and is a promising method for solving the CEED problem Table Comparison of result for case Method GA [6] PSO [6] FPA [15] BA MBA Cost ($) 20840.1 20838.3 20838.1 20848.1 20848.0957 Em.(Kg) 200.256 200.221 200.2238 200.1700 200.17 CPU (s) 0.282 0.235 0.175 0.06 0.06 Conclusion In this paper, the MBA method has been proposed for finding the optimal solutions for CEED problem where both fuel cost, emission and power losses in transmission lines are considered The MBA has been built by performing several 10 BÀI BÁO modifications on the conventional BA in aim to enhance the optimal solution quality and speed up convergence The MBA has been tested on two three-unit systems with different load and data Comparison between the MBA and BA has shown that MBA is better than BA for optimal solutions In addition, the MBA has also been compared to other existing methods The comparisons have indicated that MBA is a very promising for solving the problem and it will be applied for solving the problems with larger scale and more complicated References Kin TD, El-Hawary ME and El-Hawary F (1995) Optimal Environmental Dispatching of Electric Power Systems via an Improved Hopfield Neural Network Model IEEE Trans on Power Systems Vol 10, 1559-1565 Roa-Sepulveda CA, Salazar-Nova ER, Gracia-Caroca E, Knight UG and Coonick A (1996) Environmental Economic Dispatch via Hopfield Neural Network and Taboo Search UPEC’96 Universities Power Engineering Conference, Crete, Greece, 10011004 Song YH, Wang GS, Wang PY and Johns AT (1997) Environmental/econoimic dispatch using fuzzy logic controlled genetic algorithm IEE Proc-Gener Transm Distrib, vol 44, 377-382 Harry CS, Rughooputh and Robert TF, King A (2003) Environmental/Economic Dispatch of Thermal Units using an Elitist Multiobjective Evolutionary Algorithm Industrial Technology, 2003 IEEE International Conference in Maribor, Slovenia 1, 48-53 Mandal KK and Chakraborty N (2008) Effect of Control Parameters on Differential Evolution based Combined Economic Emission Dispatch with Valve-Point Loading and Transmission Loss International Journal of Emerging Electric Power Systems, vol 9, 1-18 Devi AL, Krishna OV (2008) Combined economic and emission dispatch using evolutionary algorithms-a case study ARPN J Eng Appl Sci, vol 3, No 6, 28-35 Roy PK, Ghoshal SP and Thakur SS (2010) Combined economic and emission dispatch problems using biogeography-based optimization Electr Eng, vol.92, 173– 184 Basu M (2011) Economic environmental dispatch using multi-objective differential evolution Applied Soft Computing, vol 11, 2845–2853 11 BÀI BÁO 10 11 12 13 14 15 16 17 Elaiwa AM, Xiab X and Shehata AM (2013) Hybrid DE-SQP and hybrid PSO-SQP methods for solving dynamic economic emission dispatch problem with valve-point effects Electric Power Systems Research, vol 103, 192–200 Hamedi H (2013) solving the combined economic load and emission dispatch problems using new heuristic algorithm Electrical Power and Energy Systems, vol 46, 10–16 Manteaw ED, Odero NA (2012) Combined economic and emission dispatch solution using ABC_PSO hybrid algorithm with valve point loading effect Int J Sci Res Publ, No 12, 1e9 Zhang R, Zhou J, Mo L, Ouyang S, Liao X (2013) Economic environmental dispatch using an enhanced multi-objective cultural algorithm Electr Power Syst Res, vol 99, 18e29 Thao NTP and Thang NT (2014) Environmental Economic Load Dispatch with Quadratic Fuel Cost Function Using Cuckoo Search Algorithm International Journal of u- and e- Service, Science and Technology, Vol.7, No.2, 199-210 Thang NT (2013) Economic emission load dispatch with multiple fuel options using Hopfiled Lagrange Network International Journal of Advanced Science and Technology, vol 57, 9-24 Abdelaziz AY, Ali ES, Abd Elazim SM (2016) Implementation of flower pollination algorithm for solving economic load dispatch and combined economic emission dispatch problems in power systems Energy, vol 101, 506e518 Yang XS (2010) A New Metaheuristic Bat-Inspired Algorithm, in: Nature Inspired Coop-erative Strategies for Optimization Studies in Computational Intelligence Springer Berlin, 284, Springer, 65-74 Nguyen TT, Ho SD (2015) Bat Algorithm for Economic Emission Load Dispatch Problem International Journal of Advanced Science and Technology, vol 86, 51-60 18 Kulkarni PS, Kothari AG, Kothari DP (2000) Combined economic and emission dispatch using improved back-propagation neural network Electric Power Comp Syst, vol 28, 31–44 12 ... ? ?Áp dụng thuật toán Meta- heuristic điều độ tối ưu hệ thống điện? ?? góp phần giải vấn đề Với kết nghiên cứu đề tài có thể:  Ứng dụng cho tổ máy nhiệt điện hệ thống  Sử dụng tài liệu cho toán tối. .. PHẠM KỸ THUẬT THÀNH PHỐ HỒ CHÍ MINH LUẬN VĂN THẠC SĨ HỌ VÀ TÊN HỌC VIÊN HỒ HỮU THÌN TÊN ĐỀ TÀI LUẬN VĂN THẠC SĨ ÁP DỤNG THUẬT TOÁN META- HEURISTIC ĐIỀU ĐỘ TỐI ƢU HỆ THỐNG ĐIỆN NGÀNH: KỸ THUẬT ĐIỆN... thuật tốn BA phân tích nhược điểm thuật toán BA để xây dụng thuật toán MBA Chƣơng 5: Áp dụng thuật toán Dơi cải tiến điều độ tối ƣu tổ máy phát Trình bày cách áp dụng thuật tốn MBA vào toán điều