1 of 20

PREDICTION OF SOLAR ENERGY �BY MACHINE LEARNING

1806044-Javid Saifullah

1806058-Biswas Rudra Jyoti Arka

1806071-Promit Biswas

​

​

1806078-Sourv Chandra Roy

1806081-Md. Motasim Billah

1806084-K.M. Zulfikkar Sadik

​

Presented by

2 of 20

ABSTRACT

  • Photovoltaic (PV) Electricity Generation in power distribution systems is increasing day by day.
  • PV power generation forecasting can result from weather data and Global Horizontal Irradiation (GHI) or simply the Irradiation data.
  • The weather data can be used to predict the power generation of a single day

3 of 20

Project Overview

  • Forecasting the Power generation from Weather Data
  • Modelling a battery management system.

4 of 20

Primary Idea Behind the Project

  • The GHI Data depends on several parameters such as Air Temperature, relative humidity, wind direction, and wind speed.
  • Using machine learning models to forecast the solar power generation.
  • A small-scale Battery Management System consisting of PV integration, distribution system, and loads is designed to estimate how much of the generated power is required on a particular day.

​

5 of 20

Workflow

Visualization

Preprocessing

Model Training

Error Evaluation

Designing the Battery Management System

Estimation of the Battery Storage Capacity

6 of 20

DATASET

  • 9 parameters
  • Timestamps for 2-minute intervals for three days.
  • Parameters
    • Air Temperature
    • Relative Humidity
    • Wind Speed
    • Wind Direction
    • Solar Radiation

​

​

​

​

    • Resistance Temperature Detector
    • Array Voltage
    • Array Current
    • Power Generated

7 of 20

PREPROCESSING

Standard Scaling from Python Library Scikit-learn (Sklearn) was used to make the data suitable to feed into the models.

The data was split into 80-20 per cent as the training and testing data.

8 of 20

Data Visualization

9 of 20

Data Visualization

10 of 20

Power Generation Forecasting

ANN

  • Artificial Neural Network
  • RMSE Score 0.0073

11 of 20

Power Generation Forecasting

ANN

​

12 of 20

Power Generation Forecasting

ANN

​

13 of 20

Power Generation Forecasting

LSTM

  • Long short-Term Memory
  • Deep learning Model
  • RMSE score: 0.0460

14 of 20

Power Generation Forecasting�

LSTM

15 of 20

Designing the Power Storage System�

16 of 20

BATTERY MANAGEMENT SYSTEM�

Lithium-Ion battery:

Nominal Voltage: 24V

Rated Capacity: 50 Ah

Initial State of Charge: 45%

​

17 of 20

Workflow of Simulink model

Battery specifications

18 of 20

Outputs of the System

19 of 20

Outputs of the System

20 of 20

Result

The outputs of the Simulink model show the workability of the proposed system. Also, changing the load parameters of the system, the battery storage capacity can be modified according to the requirements fulfilled by the power generated by the PV generation.

​