<![CDATA[DESIGN OPTIMIZATION OF MICRO HYDRO TURBINE USING ARTIFICIAL PARTICLE SWARM OPTIMIZATION AND ARTIFICIAL NEURAL NETWORK aLie Jasa, bRatna Ika Putri, cArdyono Priyadi, dMauridhi Hery Purnomo a,b,c,d Instrumentation, Measurement, and Power Systems Identification Laboratory Electrical Engineering Department, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia a Electrical Engineering Department, Udayana University, Bali, Indonesia b Electrical Engineering Department, Politeknik Negeri Malang, Malang, Indonesia. Email: liejasa@unud.ac.id Abstrak Turbin digunakan mengkonversi energy potensial menjadi energy kinetik. Kapasitas Energy yang dihasilkan dipengaruhi oleh sudu-sudu turbin yang dipasang pada tepi. Sudu turbin dirancang seorang ahli dengan sudut kelengkungan tertentu. Efisiensi dari turbin dipengaruhi oleh besarnya sudut, jumlah dan bentuk sudu. Algoritma PSO dapat digunakan untuk komputasi dan optimasi dari design turbin mikro hidro. Penelitian ini dilakukan dengan; Pertama, Formula design turbin dioptimasi dengan PSO. Kedua, Data hasil optimasi PSO diinputkan kedalam jaringan ANN. Ketiga, training dan testing terhadap simulasi jaringan ANN. Dan yang terakhir, Analisa kesalahanr dari jaringan ANN. Data PSO sebanyak 180 record, 144 digunakan untuk training dan sisanya 40 untuk testing. Hasil penelitian ini adalah MAE= 0.4237, MSE=0.3826, dan SSE=165.2654. Error training terendah didapatkan dengan algoritma pembelajaran trainlm. Kondisi ini membuktikan bahwa jaringan ANN mampu menghasilkan desain turbin yang optimal. Kata kunci: Turbin, PSO, ANN, Energi Abstract Turbines are used to convert potential energy into kinetic energy. The blades installed on the turbine edge influence the amount of energy generated. Turbine blades are designed expertly with specific curvature angles. The number, shape, and angle of the blades influence the turbine efficiency. The particle swarm optimization (PSO) algorithm can be used to design and optimize micro-hydro turbines. In this study, we first optimized the formula for turbine using PSO. Second, we input the PSO optimization data into an artificial neural network (ANN). Third, we performed ANN network simulation testing and training. Finally, we conducted ANN network error analysis. From the 180 PSO data records, 144 were used for training, and the remaining 40 were used for testing. The results of this study are as follows: MAE = 0.4237, MSE = 0.3826, and SSE = 165.2654. The lowest training error was achieved when using the trainlm learning algorithm. The results prove that the ANN network can be used for optimizing turbine designs. Keywords: Turbine, PSO, ANN, Energy]]>
Copyrights © 2014
