Hasil Pencarian  ::  Simpan CSV :: Kembali

"The combination between frames and walls is frequently used in the earthquake resistant design of multistorey reinforced concrete buildings Frames are relatively flexible , it will deform according to the shear mode. Meanwhile walls are usually very stift and it will deform with the flexural one. Accordingly between frames and walls will have a conflict of deformation modes. There is no clear guidance regarding the appropriate ratio between the number of walls and frames. In addition, effects of the rocking foundation to the structural response need to be investigated The seismic behavior of the multistorey reinforced concrete building has been investigated Three types of structure i.e 2 Walls +5 Frames or 2W+5F, 2W+7F and 2W+9F have been used for the structural models. The corresponding structural models are 12- storey buildings with 2 symmetrical beams span and _symmetrical buildings plan. The North-South Component of the 1940 El Centro earthquake record has been used for the input motion. The stifness and damping interaction between the soil and the foundation according to the Lumped ParameterMethod are also used The results of investigation show that the smaller the wall-frame ratio , the bigger the base shear coefficient resisted by walls, the smaller the plastic hinge rotation of the column 's bases, the smaller the total plastic hinge rotation and the hysteretic energy dissipation of every frame. In general, the structural response of the rocking structures are smaller than those the _fixed base structures. The appropriate wall frame-ratio can not be defined definitely without any clear requirement of the design criteria."

"Saat ini terjadi kelebihan pasokan listrik di Jawa-Bali seiring beroperasinya PLTU dari program 35.000 Mega Watt. Untuk itu dibutuhkan prediksi kebutuhan energi listrik yang lebih akurat sebagai dasar perencanaan dan pengoperasian sistem tenaga listrik, serta peningkatan efisiensi sistem tenaga listrik. Mengembangkan model yang mencapai presisi peramalan tertinggi dalam konteks tenaga listrik telah menjadi objek studi di beberapa negara. Penelitian ini berusaha menemukan model prediksi konsumsi energi listrik di Indonesia menggunakan LSTM (Long Short Term Memory). LSTM merupakan jenis jaringan syaraf berulang (Recurrent Neural Network) yang sering digunakan untuk mengidentifikasi pola periodik dalam deret waktu (time series) karena merekam hubungan antara nilai yang berurutan. Pelatihan dan pengujian model menggunakan data konsumsi energi listrik dari Januari 2013 hingga Februari 2023 yang diperoleh dari laporan penjualan tenaga listrik PT PLN. Dilakukan percobaan dengan berbagai kombinasi nilai hyperparameter jumlah unit neuron dan jumlah epoch untuk masing-masing model prediksi konsumsi energi listrik Nasional, segmen Rumah Tangga, Bisnis, dan Industri. Evaluasi kinerja model diukur dengan MAPE (rata-rata persentase kesalahan absolut). Model prediksi konsumsi energi listrik nasional dapat memprediksi dengan baik dilihat dari nilai MAPE sebesar 2,212%. Dengan membandingkan akurasi model LSTM yang dibangun dengan prediksi manual yang dilakukan PLN per bulan, maka model LSTM yang dibangun mampu melakukan prediksi lebih akurat.

---

Currently, there is excess power or oversupply of power generation capacity in Jawa-Bali in line with the operation of the PLTU from the 35,000 Mega Watt program. For this reason, a more accurate prediction of electricity demand is needed as a basis for planning and operating the electric power system, as well as increasing the efficiency of the electric power system. Developing a model that achieves the highest forecasting precision in the context of electric power has been the object of study in several countries. This research seeks to find a predictive model of electricity consumption in Indonesia using LSTM (Long Short Term Memory). LSTM is a type of recurrent neural network (RNN) that is often used to identify periodic patterns in a time series because it records the relationship between successive values. Model training and testing uses electricity consumption data from January 2013 to February 2023 obtained from PT PLN's electricity sales report. Experiments were carried out with various combinations of hyperparameter values for the number of neuron units and the number of epochs for each prediction model for the National electricity consumption, the Household, Business and Industrial segments. Model performance evaluation is measured by MAPE (Mean Absolute Percentage Error). The National electricity consumption prediction model can predict well with MAPE value of 2.212%. By comparing the accuracy of the built LSTM model with manual predictions made by PLN per month, LSTM model was able to make more accurate predictions."