ENERGY PREDICTION OF PV P ANELS FOR DEMAND AND RESPONSE SYSTEM USING ANN D EEPLEARNING Rohaib Bhatti

2025-05-06 0 0 744.62KB 6 页 10玖币

侵权投诉

ENERGY PREDICTION OF PV PANELS FOR DEMAND AND

RESPONSE SYSTEM USING ANN (DEEP LEARNING)

Rohaib Bhatti

National University Of Sciences and Technology

Islamabad

rbhatti.ee41ceme@student.nust.edu.pk

Ali John Naqvi

National University of Sciences and Technology

Islamabad

salijohn.ce41ceme@student.nust.edu.pk

Abdullah Tauqeer

National University Of Sciences and Technology

Islamabad

atauqeer.ee41ceme@student.nust.edu.pk

ABSTRACT

Renewable sources of energy are the future due to the environmental problems caused by non-

renewable sources to produce energy. The biggest issue with renewable energy sources is that the

power produced by devices such as PV solar panels depend on many uncertain factors. These factors

include Solar irradiation, wind speed, temperature, hours of sunlight per day, and surface temperature

of solar panels. Industries and authorities can use this predicted power through ML to control power

consumption. Power forecast has multiple applications that promote the usage of green energy

in the future. This paper will also help to determine the dependence of PV power production on

various weather/environmental factors. For this paper, we have used regressions and ANN models to

predict power. In the end, results of power prediction using regression as well as the ANN model are

compared with the actual power output. Overall, ANN performs excellently compared to the other

machine learning models because of its advanced feature selection techniques.

Keywords Power, Deep learning

1 Introduction

The motivation to write this research comes from the aim of contributing towards a sustainable world. A sustainable

world is not possible without producing green energy. This research aims to solve the problem of uncertainty of the

power being produced through solar PV panels. Providing accurate predictions of solar power generation will help

industries and power authorities maintain an uninterruptible power supply by demand and response system. This study

will result in more power being produced by renewable sources than non-renewable, thus saving the environment from

global warming and climate change.

In the future, the use of solar energy will grow in enormous numbers because it is environmentally friendly, especially

in regions like Pakistan, where there is great potential for producing PV power. In Pakistan, most areas have extended

hours of sunlight availability throughout the year, especially in the country’s southern region [3]. It is not wrong to

say that solar energy is the most abundant energy source that is available to our planet, and it is something that will

not vanish soon. 100,000 TW of electrical energy can be produced through sunlight every hour, which is such a huge

amount that it can give power to the entire population of humans on earth for a year [2]. This statistic shows that

humans should take advantage of that and produce green energy that can meet our requirements and not damage the

earth’s environment.

Solar panels are easy to install because no complex setup or massive construction is needed. Overall, these cells are

also easy to manufacture for industries, which shows that there is great potential for developing nations like Pakistan to

arXiv:2210.11559v1 [eess.SP] 20 Oct 2022

fulﬁll their energy needs using an economical energy source such as sunlight [5]. The most signiﬁcant disadvantage of

using solar power is that it requires many areas to install PV panels to produce a megawatt of electricity. Currently,

Pakistan’s energy needs are 92 GWs, and Pakistan has the potential to produce 2.9 TW of solar electricity, which shows

that solar alone can meet all of the country’s energy requirements, and still, much energy will be left to export [2].

The system is also clearly suited to off-grid generating and consequently to places with little infrastructure. The public

accepts and generally approves solar technology; it is subject to less geopolitical, environmental, and aesthetic concerns

than nuclear, wind, or hydro. However, substantial desert installations may arouse objections [6].

Photovoltaics are challenging to integrate into power systems since solar energy is mainly reliant on location and

weather, shifting wildly. This uncertainty results in system disturbances, voltage spikes, conﬁguration issues, ineffective

utility management, and economic loss. Forecast models can be helpful, but time stamps, forecast horizons, input

correlation analysis, data pre-and post-processing, weather categorization, capacity planning, risk assessment, and

performance assessments must all be taken into account [1].

One of the signiﬁcant issues the industry is presently facing is incorporating these unique types of power plants into the

electrical system. Because renewable energy is becoming a more signiﬁcant part of the power market, machine learning

algorithms must forecast foreseeable electricity production to a desired level of accuracy. These projections also apply

to power plant owners, the energy trading market, and grid operators. Information on future energy production growth

in the grid decreases all market players’ technical and economic risks [4].

The fundamental objective of this work is to evaluate the use of ML and deep learning techniques to predict solar

irradiance. The paper will also go through the intricacies of the forecasting methodologies with illustrations. Each

approach’s merits and ﬂaws are also discussed. The predicted data and actual data comparisons based on our simulations

are explored. Furthermore, intriguing future research paths are suggested.

2 Method

The methodology used in this research will be using an ML model, a regression model, and a deep learning model,

ANN. Using the data from the world bank about environmental factors of Pakistan, we trained a regression model, as

well as training of the artiﬁcial neural network was also done. Four solar panels were installed in Islamabad to take

real-time data and compare it with forecasted data to get an idea of the accuracy of our predicted data.

Figure 1: PV Setup used to predict power

These four plates were installed in Islamabad, and the world bank’s data was also from Islamabad. Four of these PV

solar panels were each 150w. The nameplate of the solar cell is attached in ﬁgure 2.

The 4 PV Panels of solar are of EverExceed, and their module type is EX150-36P. The maximum power Pmax they

can generate is 150W. These panels can operate at temperatures 47

2. The dimensions of the solar panels are

1480*680*30mm. The max power voltage it can produce is 18.4V, and the max power current that can be produced is

8.16 A.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ENERGYPREDICTIONOFPVPANELSFORDEMANDANDRESPONSESYSTEMUSINGANN(DEEPLEARNING)RohaibBhattiNationalUniversityOfSciencesandTechnologyIslamabadrbhatti.ee41ceme@student.nust.edu.pkAliJohnNaqviNationalUniversityofSciencesandTechnologyIslamabadsalijohn.ce41ceme@student.nust.edu.pkAbdullahTauqeerNationalUniver...

展开>> 收起<<

ENERGY PREDICTION OF PV P ANELS FOR DEMAND AND RESPONSE SYSTEM USING ANN D EEPLEARNING Rohaib Bhatti.pdf

共6页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

ENERGY PREDICTION OF PV P ANELS FOR DEMAND AND RESPONSE SYSTEM USING ANN D EEPLEARNING Rohaib Bhatti

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: