Clustering of thermal loads for participating in demand response programs

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Clustering of thermal loads for participating in demand response programs

ارائه دهنده: Provider: morteza matin nezhad

اساتید راهنما: Supervisors: Dr mohamad hasan moradi-Dr alireza hatami

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr saleh razini-Dr mohsen hasan babae nozadian

زمان و تاریخ ارائه: Time and date of presentation: 2024

مکان ارائه: Place of presentation: سالن آمفی تئاتر

چکیده: Abstract: Today's energy supply systems are in conflict with power quality and customer satisfaction. Reducing the rate of load interruption or reducing the rate of unsold energy of the network has a direct dependence on the separation of network loads and the speed of timely action of the network control center. The sustainable supply of energy demanded by the network includes macro and long-term policies of power systems and distribution of electrical energy and depends on the high costs of building power plants and designing rotating storage systems. Therefore, the rapid growth of energy consumption of subscribers often exceeds the sustainable policies of building energy production units, and in this way, network operators turn to load demand management programs to meet average load demand. In this study, due to the importance of thermal load clustering with the help of genetic artificial intelligence algorithm, in order to participate in load response planning, it is targeted. In this way, three steps were planned to solve the problem. In the first step, the clustering of all loads of the 17-basin test network with uniform load distribution was done with the help of genetic artificial intelligence algorithm. In the second step, based on matching the impedance of the clusters, the smallest cluster with the highest thermal load rate was separated. In the third and final step, in the clusters with more than one busbar, LMP was calculated to determine the right of way to disconnect the busbar. Finally, in the 8 scenarios defined in this study, it was observed that firstly, the higher the participation coefficient of chromosomes in the production of the new generation, it has no effect on improving the performance of clustering many times, and on the contrary, the speed of solving the problem increases greatly. . Also, with the increase of power grids, the number of chromosomes at the beginning of the genetic algorithm should be less than 50, and in the next stages, the participation of chromosomes in the production of the new generation should be considered less than 0.85. Secondly, it was observed that the number of repetitions of the genetic algorithm should not be large to solve this problem. Because it has no effect on improving the convergence of the algorithm. In scenario 3, the number of genetic algorithm iterations was 50 iterations for dividing the network into 3 clusters, and the problem solving time was recorded as 61.52 seconds. While in scenario 4, even though the clustering algorithm was ordered to divide the network into 4 clusters, the number of iterations of the genetic algorithm was considered to be 5, which solved the problem after 3.54 seconds. This result helps us to solve wider networks with more busbars with an acceptable speed. Thirdly, it was observed that the context of the problem is such that in order to achieve the desired answer, the genetic mutation should be small in the amount of 0.001, and by choosing the genetic mutation to the value of 0.3 genes, the algorithm completely diverges. Fourthly, it was finally observed that in scenarios 6, 5 and 8, the genetic algorithm introduces clusters with only one busbar to us, to interrupt at the peak of energy consumption. This result makes it possible to suggest to electric companies to cut off thermal and cheap loads at the peak of network consumption, the best practice is to install remote control switches on specific busbars of the distribution network and cut off only a few busbars at different points of the network. The place of disconnection is a feeder or a cluster with several busbars

فایل: ّFile: Download فایل