MODELLING AND SIMULATION OF MAXIMUM POWER POINT TRACKER WITH DEEP NEURAL NETWORK METHOD FOR SOLAR POWER PLANT PERFORMANCE IMPROVEMENT
Maximum power point tracker is a control device that accomodates functions such as conversion, regulation, and filtering both voltage and current output from a solar panel in solar power plant system to overcome the problem of power generation. MPPT can be combined with both converter and inverter f...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/68359 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Maximum power point tracker is a control device that accomodates functions such as conversion, regulation, and filtering both voltage and current output from a solar panel in solar power plant system to overcome the problem of power generation. MPPT can be combined with both converter and inverter for assisting in tracking and extracting function of the maximum power point, that can be generated by the solar panel. MPPT carry out it features with calculated duty ratio value for a certain condition of solar irradiance and ambient temperature and then transmit that processed signal to DC DC converter. MPPT controller works with a certain algorithm such as incremental conductance, perturb and observe, fuzzy logic control, as well as more advance algorithm such as machine learning.
In this research, with the CRISP-DM framework, MPPT controller will be modelled and designed with deep neural network algorithm trained with real weather data in the form of 1.228 x 3 matrix data. The DNN-based MPPT implemented on a solar power plant system that consists of PV array, DC DC converter and control algorithm as MPPT, and an adjusted resistive load. In this research, the MPPT will modelled with three types of DC-DC converter topologies such as buck, boost, and buck-boost. The system model that has been successfully designed is tested through simulation with several test conditions such as standard testing conditions, step test, and real environmental conditions test with determined evaluation parameters.
Based on the test results, the DNN algorithm shows mean square error for buck, boost, and buck-boost MPPT is 2,20×10?6, 2,27×10?5, and 4,80×10?6consecutively and improves efficiency up to 7,8% on STC. Solar power plant that used the DNN-based MPPT controller simulated under real environmetal conditions and the simulation results of the efficiency of the MPPT for buck, boost, and buck-boost is 95,47%, 90,97%, and 79,34% consecutively. |
|---|