# A Regional Day-Ahead Rooftop Photovoltaic Generation Forecasting Model Considering Unauthorized Photovoltaic Installation

^{*}

## Abstract

**:**

## 1. Introduction

#### 1.1. Background and Motivation

#### 1.2. Literature Review

#### 1.3. Contributions

#### 1.4. Structure of This Study

## 2. Problem Formulation of Unauthorized Photovoltaic (PV) Installation and Regional Rooftop PV Forecasting

#### 2.1. Problem Statement

_{PV}(d, t)

_{PV}(d, t) = 0 ∀ d ∈ D, ∀ t ∈ T. For rooftop PV, the PV output power is BTM except at representative solar sites, implying that most rooftop PV power is not measured and collected. The home classification according to rooftop PV installation, rooftop PV authorization, and sub meter of PV power installation is shown in Figure 1 and Table 2.

#### 2.2. Framework of the Proposed Approach

## 3. Proposed Methodology of Regional Rooftop PV Generation Forecasting

#### 3.1. Unauthorized PV Detection Model

#### 3.1.1. Four Weather Groups Clustering

#### 3.1.2. Generation Real and Virtual Typical Net Load Pattern and Minimum Net Load Pattern

Algorithm 1. Generation virtual TNLP |

Input: ${TNLP}_{Sam}^{A}\left(t\right)$, ${TNLP}_{Sam}^{D}\left(t\right)$, ${TPP}_{Sam}^{A}\left(t\right),{TPP}_{Sam}^{D}\left(t\right),N,\omega PV,Sam\in \left[1,{N}_{H1}\right]$Output: if $\omega PV==1$then${TNLP}_{n}^{A}$(t), ${TNLP}_{n}^{D}$(t) else${TLP}_{n}^{A}$(t), ${TLP}_{n}^{D}$(t) Endfor$n=1:N$do${TLP}_{Sam}\left(t\right)={TNLP}_{Sam}\left(t\right)+{TPP}_{Sam}\left(t\right);$ ${TLP}_{Norm,Sam}\left(t\right)=\frac{{TLP}_{Sam}\left(t\right)}{\mathrm{max}\left({TLP}_{Sam}\left(t\right)\right)}$; for t = 0:23 dorandomly select r $\in \left[1,{N}_{H1}\right]$ ${TLP}_{n,Norm}^{A}$(t) = ${TLP}_{r,Norm}^{A}$(t); ${TLP}_{n,Norm}^{D}$(t) = ${TLP}_{r,Norm}^{D}$(t); End${TLP}_{n}^{A}\left(t\right)={TLP}_{n,Norm}^{A}\left(t\right)\times \mathrm{max}\left({TLP}_{Sam}\left(t\right)\right);$ ${TLP}_{n}^{D}\left(t\right)={TLP}_{n,Norm}^{D}\left(t\right)\times \mathrm{max}\left({TLP}_{Sam}\left(t\right)\right);$ if $\omega PV==1$ then |

${TPP}_{Norm,Sam}^{A}\left(t\right)=\frac{{TPP}_{Norm,Sam}^{A}\left(t\right)}{{C}_{Sam}};$ ${TPP}_{Norm,Sam}^{D}\left(t\right)=\frac{{TPP}_{Norm,Sam}^{D}\left(t\right)}{{C}_{Sam}};$ randomly select PV capacity ${C}_{rand}\in \left[1,10\right]$ ${TPP}_{Vir}^{A}\left(t\right)={TPP}_{Norm,Sam}^{A}\left(t\right)\times {C}_{rand};$ ${TPP}_{Vir}^{D}\left(t\right)={TPP}_{Norm,Sam}^{D}\left(t\right)\times {C}_{rand};$ ${TNLP}_{n}^{A}\left(t\right)={TLP}_{n}^{A}\left(t\right)-{TPP}_{Vir}^{A}\left(t\right);$ ${TNLP}_{n}^{D}\left(t\right)={TLP}_{n}^{D}\left(t\right)-{TPP}_{Vir}^{D}\left(t\right);$ Elsebreak; EndEnd |

#### 3.1.3. Feature Extraction Based on TNLP and MNLP

#### 3.1.4. Training and Test of Unauthorized PV Detection Model

#### 3.2. Unauthorized PV Capacity Estimation Model

#### 3.2.1. Generation Virtual Net Load

Algorithm 2. Generation virtual NL |

Input: ${\mathrm{C}}_{\mathrm{Au}}$ = [${\mathrm{C}}_{1},{\mathrm{C}}_{2},{\mathrm{C}}_{3},\dots ,{C}_{{N}_{Au}}$], ${NL}_{Sam}^{WG}\left(d,t\right)$, ${P}_{PV,Sam}^{WG}\left(d,t\right),Au\in \left[1,{N}_{Au}\right],Sam\in \left[1,{N}_{Sam}\right]$Output: $N{L}^{WG}\left(d,t\right)$Initialize ${\mathrm{Dis}}_{\mathrm{cap}}=0;cap\in \left[1,10\times int\left(Max\left({C}_{Au}\right)\right)\right]$ ${\mathrm{GL}}_{\mathrm{Sam}}^{\mathrm{WG}}\left(d,t\right)={\mathrm{NL}}_{\mathrm{Sam}}^{\mathrm{WG}}\left(d,t\right)+{\mathrm{P}}_{\mathrm{PV},\mathrm{Sam}}^{\mathrm{WG}}\left(d,t\right);$ for $n=1:{N}_{Au}$ dofor i = 1:$int\left(Max\left({C}_{Au}\right)\right)$ dofor j = 1:10 doif $i+0.1\left(j-1\right)\le {C}_{n}<i+0.1j$ then${Dis}_{int\left(Max\left({C}_{Au}\right)\right)+i+j}+=1$ elseContinue; endendend${Count}_{cap}=Max\left({Dis}_{cap}\right)-{Dis}_{cap};$ for $WG=A:D$ do${P}_{PV,mean}^{WG}\left(t\right)=\frac{1}{{N}_{d}^{WG}}{\displaystyle {\displaystyle \sum}_{d=1}^{N\_d^WG}}{P}_{PV,Sam}^{WG}\left(d,t\right);$ ${P}_{PV,Norm}^{WG}\left(t\right)=\frac{1}{{C}_{PV,Sam}}{P}_{PV,mean}^{\mathrm{WG}}\left(t\right);$ endfor $l=1:10\times int\left(Max\left({C}_{Au}\right)\right)$ dofor $c=1:{Count}_{l}$ doSelect randomly PV capacity ${C}_{rand}\in $ [1+(l−1)/10,1+l/10] Select randomly $\mathrm{r}\in \left[1,{\mathrm{N}}_{\mathrm{Sam}}\right]$ for WG = A:D do${NL}_{r}^{WG}\left(d,t\right)={GL}_{r}^{WG\left(d,t\right)}-{P}_{PV,Norm}^{WG}\left(t\right)\times {C}_{rand}$ endendend |

#### 3.2.2. Extracting Minimum Net Load Pattern (MNLP) for Four Weather Classes

_{A}(t) and MNLP

_{D}(t) each denote MNLP in A and D of WG. They are shown in Equations (11) and (12). D

_{A}and D

_{D}in Equations (11) and (12) denote the set of days when the WG is A and the set of days when the WG is D, where d is the day index.

#### 3.2.3. Extracting Features from MNLP

^{A}(t). It varies by PV output and GL values. If PV output is maximum or GL is minimum, it has negative values of significantly larger magnitude. It is available to estimate PV capacity using ${\mathrm{F}}_{1}^{\mathrm{E}}.$ The second feature ${\mathrm{F}}_{2}^{\mathrm{E}}$ denotes difference of MNLP

^{A}(t) and MNLP

^{D}(t) during a day. So, ${\mathrm{F}}_{3}^{\mathrm{E}}$ is originally the sum of the difference of MNLP

^{A}(t) and MNLP

^{D}(t) for intermediate start time and end time of PV generation. However, it is difficult to recognize PV capacity when the PV capacity is small. Therefore, the sum of difference between MNLP

^{A}(t) and MNLP

^{D}(t) for intermediate start time and end time of PV generation is chosen as the third feature. In the testing process, the MNLP of a test home is used to extract three features and unauthorized PV capacity is estimated through three features.

#### 3.2.4. Training and Test PV Capacity Estimation Model

#### 3.3. Regional PV Forecasting Model

#### 3.3.1. Clustering and Sampling of Rooftop PV

Algorithm 3. Clustering of rooftop PV |

Input: $\mathrm{a}\mathrm{given}\mathrm{data}\mathrm{X}=\left\{\left({lat}_{1},{lon}_{1}\right),\left({lat}_{2},{lon}_{2}\right),\dots .,\left({lat}_{n},{lon}_{n}\right)\right\}$The number of cluster k Maximum number of iteration I Output:$\mathrm{clustering}\mathrm{result}$${o}_{nk}$ for all n PV k center of cluster CRandomly initialize C = $\left\{\left({lat}_{1}^{C},{lon}_{1}^{C}\right),\left({lat}_{2}^{C},{lon}_{2}^{C}\right),\dots .,\left({lat}_{k}^{C},{lon}_{k}^{C}\right)\right\}$ for $i=1:I$ do// Assignment step for n = 1:N do${o}_{nk}=\left\{\begin{array}{l}1,\mathrm{if}\mathrm{k}={\mathrm{argmin}}_{\mathrm{i}}{\left|\left({\mathrm{lat}}_{\mathrm{n}}-{\mathrm{lat}}_{\mathrm{k}}^{\mathrm{C}}\right)\right|}^{2}+{\left|\left({\mathrm{lon}}_{\mathrm{n}}-{\mathrm{lon}}_{\mathrm{k}}^{\mathrm{C}}\right)\right|}^{2}\\ 0,\mathrm{otherwise}\end{array}\right.$ end// Update step for $k=1:K$ do${lat}_{k}^{C}=\frac{1}{{{\displaystyle \sum}}_{n=1}^{N}{o}_{nk}}{\displaystyle {\displaystyle \sum}_{n=1}^{N}}{o}_{nk}{lat}_{n};$ ${lon}_{k}^{C}=\frac{1}{{{\displaystyle \sum}}_{n=1}^{N}{o}_{nk}}{\displaystyle {\displaystyle \sum}_{n=1}^{N}}{o}_{nk}{lon}_{n};$ endend |

#### 3.3.2. Individual Rooftop PV Generation Forecasting

_{actual}, F

_{Max}, F

_{Min}and F

_{Norm}denote original feature, maximum of feature, minimum of feature, and normalized feature. Next, the individual rooftop PV generation model is constructed in Equation (17).

_{PVi}

_{nd}(t) is the predicted PV generation at time t; time resolution is one hour as described in Section 2.1. ${\mathrm{F}}_{1}^{\mathrm{Norm}},{\mathrm{F}}_{2}^{\mathrm{Norm}},{\mathrm{F}}_{3}^{\mathrm{Norm}},{\mathrm{F}}_{4}^{\mathrm{Norm}},{\mathrm{F}}_{5}^{\mathrm{Norm}}$ denote normalized one day ahead PV generation, normalized solar irradiance, normalized cloud cover, normalized precipitation, and normalized temperature. Finally, the test of individual PV generation forecasting is performed through the trained PV generation forecasting model.

#### 3.3.3. Upscaling Sample Rooftop PV Generation by Cluster

_{c,ind}(t) denotes predicted individual generation ind th for the representative solar site at time t in cluster c. N

_{rep}denotes the number of representative solar sites. uf

_{c}denotes upscaling factor of cluster c that corresponds to the scaled up coefficient of individual power generation. The PV generation of a cluster (or sub region) is made by multiplying each individual PV generation by the upscaling factor and aggregating them. Equation (19) shows how the upscaling factor is calculated. It is defined as the ratio of total rooftop PV capacity in the cluster to the sum of representative solar sites capacity.

#### 3.3.4. Aggregating PV Generation of a Cluster

## 4. Case Study

#### 4.1. Experimental Data Description

_{PV}, the number of given home with rooftop PV was 300, all home in the region were 1500. ${\mathrm{r}}_{\mathrm{Au}}$ is 0.5 (i.e., 50%), which means the ratio of home installed unauthorized rooftop PV of all rooftop PV. In other words, 150 homes are authorized and the other 150 homes are unauthorized in the case study. Finally, an important assumption in our work is that the number of systems is constant over 2 years. Because it is difficult to find this by complete enumeration, there are few papers on this. According to [29], identified unauthorized rooftop PV installation rate is about 50% in Cape Town, South Africa. Therefore, unauthorized rooftop PV installation rate is assumed 0.5 (i.e., 50%) based on [29]. ${\mathrm{r}}_{\mathrm{Sam}}$, the ratio of the number of home PV generation data is measured and among the home installed authorized PV, is assumed 0.08 (8%). ts and te are assumed to be 9 and 16. This is because the period that PV generation mainly occurs is from 9 to 16.

#### 4.2. Performance Metric

#### 4.2.1. Unauthorized PV Detection Performance Metric

#### 4.2.2. Unauthorized PV Capacity Estimation Performance Metric

#### 4.2.3. Regional PV Generation Forecasting Performance Metric

#### 4.3. Simulation Results

#### 4.3.1. Unauthorized Rooftop PV Detection Results

#### 4.3.2. Unauthorized Rooftop PV Capacity Estimation Results

#### 4.3.3. Regional Rooftop PV Generation Forecasting Results

## 5. Discussion

#### 5.1. Upscaling Factor Analysis

#### 5.2. Feature Correlation Analysis

## 6. Conclusions

- Investigating the impact of NL home-owned energy storage and electric vehicles on unauthorized PV detection performance.
- Exploring rooftop PV capacity uncertainty in addition to unauthorized PV installation. For example, there are rooftop PV faults and real-time rooftop PV penetration.

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Acknowledgments

## Conflicts of Interest

## References

- Fossil Fuel. Available online: https://ourworldindata.org/fossil-fuels (accessed on 15 January 2020).
- Greenhouse Gas Emissions. Available online: https://www.epa.gov/ghgemissions/global-greenhouse-gas-emissions-data (accessed on 9 February 2019).
- The Sources and Solutions. Fossil Fuels. Available online: https://www.epa.gov/nutrientpollution/sources-and-solutions-fossil-fuels (accessed on 9 February 2020).
- Renewable Energy Statistics 2021. Available online: https://www.irena.org/publications/2021/March/Renewable-Capacity-Statistics-2021 (accessed on 2 November 2019).
- Improving Load Forecast with BTM Solar Forecast. Available online: https://www.pjm.com/-/media/committees-groups/committees/oc/20190514/20190514-item-20-improving-load-forecast-with-btm-solar-forecast.ashx (accessed on 9 June 2019).
- World Energy Issue Insight. Available online: http://www.keei.re.kr/keei/download/WEIS1702.pdf (accessed on 6 July 2020).
- PV Status Report. Available online: https://ec.europa.eu/jrc/sites/jrcsh/files/kjna29938enn_1.pdf (accessed on 9 February 2020).
- Katsanevakis, M.; Stewart, R.A.; Lu, J. Energy storage system utilisation to increase photovoltaic penetration in low voltage distribution feeders. J. Energy Storage
**2017**, 14, 329–347. [Google Scholar] [CrossRef] - 7 Reasons Why You Should Use Solar Power. Available online: https://www.greenmatch.co.uk/blog/2014/07/7-reasons-why-you-should-use-solar-power (accessed on 12 April 2020).
- Niccolai, A.; Dolara, A.; Ogliari, E. Hybrid PV Power Forecasting Methods: A Comparison of Different Approaches. Energies
**2021**, 14, 451. [Google Scholar] [CrossRef] - Antonanzas, J.; Osorio, N.; Escobar, R.; Urraca, R.; de Pison, F.M.; Antonanzas-Torres, F. Review of photovoltaic power forecasting. Sol. Energy
**2016**, 136, 78–111. [Google Scholar] [CrossRef] - Lorenz, E.; Scheidsteger, T.; Hurka, J.; Heinemann, D.; Kurz, C. Regional PV power prediction for improved grid integration. Prog. Photovolt. Res. Appl.
**2011**, 19, 757–771. [Google Scholar] [CrossRef] - Lorenz, E.; Heinemann, D.; Kurz, C. Local and regional photovoltaic power prediction for large scale grid integration: Assessment of a new algorithm for snow detection. Prog. Photovolt. Res. Appl.
**2012**, 20, 760–769. [Google Scholar] [CrossRef] - Junior, J.G.d.S.F.; Oozeki, T.; Ohtake, H.; Shimose, K.I.; Takashima, T.; Ogimoto, K. Regional forecasts and smoothing effect of photovoltaic power generation in Japan: An approach with principal component analysis. Renew. Energy
**2014**, 68, 403–413. [Google Scholar] [CrossRef] - Junior, J.; Oozeki, T.; Ohtake, H.; Shimose, K.; Takashima, T.; Ogimoto, K. Forecasting regional photovoltaic power generation-a comparison of strategies to obtain one-day-ahead data. Energy Procedia
**2014**, 57, 1337–1345. [Google Scholar] [CrossRef][Green Version] - Fonseca Junior, J.G.d.S.; Oozeki, T.; Ohtake, H.; Takashima, T.; Ogimoto, K. Regional forecasts of photovoltaic power generation according to different data availability scenarios: A study of four methods. Prog. Photovolt. Res. Appl.
**2015**, 23, 1203–1218. [Google Scholar] [CrossRef] - Zamo, M.; Mestre, O.; Arbogast, P.; Pannekoucke, O. A benchmark of statistical regression methods for short-term forecasting of photovoltaic electricity production, part I: Deterministic forecast of hourly production. Sol. Energy
**2014**, 105, 792–803. [Google Scholar] [CrossRef] - Davò, F.; Alessandrini, S.; Sperati, S.; Delle Monache, L.; Airoldi, D.; Vespucci, M.T. Post-processing techniques and principal component analysis for regional wind power and solar irradiance forecasting. Sol. Energy
**2016**, 134, 327–338. [Google Scholar] [CrossRef][Green Version] - Li, Y.; Sun, Q.; Lehman, B.; Lu, S.; Hamann, H.F.; Simmons, J.; Black, J. A machine-learning approach for regional photovoltaic power forecasting. In Proceedings of the 2016 IEEE Power and Energy Society General Meeting (PESGM), Boston, MA, USA, 17–21 July 2016; pp. 1–5. [Google Scholar]
- Saint-Drenan, Y.M.; Good, G.H.; Braun, M.; Freisinger, T. Analysis of the uncertainty in the estimates of regional PV power generation evaluated with the upscaling method. Sol. Energy
**2016**, 135, 536–550. [Google Scholar] [CrossRef][Green Version] - Pierro, M.; De Felice, M.; Maggioni, E.; Moser, D.; Perotto, A.; Spada, F.; Cornaro, C. Data-driven upscaling methods for regional photovoltaic power estimation and forecast using satellite and numerical weather prediction data. Sol. Energy
**2017**, 158, 1026–1038. [Google Scholar] [CrossRef] - Pierro, M.; De Felice, M.; Maggioni, E.; Moser, D.; Perotto, A.; Spada, F.; Cornaro, C. A new approach for regional photovoltaic power estimation and forecast. In Proceedings of the 33rd European Photovoltaic Solar Energy Conference and Exhibition (EUPVSEC), Amsterdam, The Netherlands, 25–29 September 2017; pp. 25–29. [Google Scholar]
- Fu, L.; Yang, Y.; Yao, X.; Jiao, X.; Zhu, T. A Regional Photovoltaic Output Prediction Method Based on Hierarchical Clustering and the mRMR Criterion. Energies
**2019**, 12, 3817. [Google Scholar] [CrossRef][Green Version] - Koster, D.; Minette, F.; Braun, C.; O’Nagy, O. Short-term and regionalized photovoltaic power forecasting, enhanced by reference systems, on the example of Luxembourg. Renew. Energy
**2019**, 132, 455–470. [Google Scholar] [CrossRef] - Saint-Drenan, Y.M.; Vogt, S.; Killinger, S.; Bright, J.M.; Fritz, R.; Potthast, R. Bayesian parameterisation of a regional photovoltaic model—Application to forecasting. Sol. Energy
**2019**, 188, 760–774. [Google Scholar] [CrossRef][Green Version] - Zhang, X.; Li, Y.; Lu, S.; Hamann, H.F.; Hodge, B.M.; Lehman, B. A solar time-based analog ensemble method for regional solar power forecasting. IEEE Trans. Sustain. Energy
**2018**, 10, 268–279. [Google Scholar] [CrossRef] - Zhang, X.; Grijalva, S. A data-driven approach for detection and estimation of residential PV installations. IEEE Trans. Smart Grid
**2016**, 7, 2477–2485. [Google Scholar] [CrossRef] - Shaker, H.; Zareipour, H.; Wood, D. A data-driven approach for estimating the power generation of invisible solar sites. IEEE Trans. Smart Grid
**2015**, 7, 2466–2476. [Google Scholar] [CrossRef] - Dealing with Unregisterd and Unauthorized Systems. Available online: https://solar-support.org/files/299/dump (accessed on 27 January 2020).
- EPIC 1.21—Pilot Methods for Automatic Identification of Distributed Energy Resources (such as Solar PV) as they Interconnect to the Grid to Improve Safety & Reliability. Available online: https://www.pge.com/pge_global/common/pdfs/about-pge/environment/what-we-are-doing/electric-program-investment-charge/PGE-EPIC-Project-1.21.pdf (accessed on 3 March 2020).
- HECO Customers asked to Disconnect Unauthorized PV Systems. Available online: http://khon2.com/2014/09/05/heco-customers-asked-to-disconnect-unauthorizedpv-systems/ (accessed on 13 June 2021).
- Transient Over-Voltage Mitigation and Its Prevention in Secondary Distribution Networks with High PV-to-Load Ratio, Hawaii Electric Companies. Available online: https://www.hnei.hawaii.edu/wp-content/uploads/Transient-Over-Voltage-Mitigation-and-Prevention-In-Secondary-Distribution-Networks.pdf (accessed on 15 June 2021).
- California Solar Permitting Guidebook, Solar Permitting Work Group, The Governor’s Office of Planning and Research. Available online: https://www.buildingincalifornia.com/wp-content/uploads/2014/02/CaliforniaSolarPermittingGuidebook-2012.pdf (accessed on 15 June 2021).
- Shaker, H.; Zareipour, H.; Wood, D. Estimating power generation of invisible solar sites using publicly available data. IEEE Trans. Smart Grid
**2016**, 7, 2456–2465. [Google Scholar] [CrossRef] - Wang, Y.; Zhang, N.; Chen, Q.; Kirschen, D.S.; Li, P.; Xia, Q. Data-driven probabilistic net load forecasting with high penetration of behind-the-meter PV. IEEE Trans. Power Syst.
**2017**, 33, 3255–3264. [Google Scholar] [CrossRef] - Kapourchali, M.H.; Sepehry, M.; Aravinthan, V. Multivariate Spatio-temporal Solar Generation Forecasting: A Unified Approach to Deal with Communication Failure and Invisible Sites. IEEE Syst. J.
**2018**, 13, 1804–1812. [Google Scholar] [CrossRef] - Shaker, H.; Manfre, D.; Zareipour, H. Forecasting the aggregated output of a large fleet of small behind-the-meter solar photovoltaic sites. Renew. Energy
**2020**, 147, 1861–1869. [Google Scholar] [CrossRef] - He, X.; Qiu, R.C.; Chu, L.; Ai, Q.; Ling, Z.; Zhang, J. Invisible Units Detection and Estimation Based on Random Matrix Theory. IEEE Trans. Power Syst.
**2019**, 35, 1846–1855. [Google Scholar] [CrossRef][Green Version] - Wang, F.; Li, K.; Wang, X.; Jiang, L.; Ren, J.; Mi, Z.; Shafie-khah, M.; Catalão, J.P. A distributed PV system capacity estimation approach based on support vector machine with customer net load curve features. Energies
**2018**, 11, 1750. [Google Scholar] [CrossRef][Green Version] - Li, K.; Wang, F.; Mi, Z.; Fotuhi-Firuzabad, M.; Duic´, N.; Wang, T. Capacity and output power estimation approach of individual behind-the-meter distributed photovoltaic system for demand response baseline estimation. Appl. Energy
**2019**, 253, 113595. [Google Scholar] [CrossRef] - Solar Home Electricity Data. Available online: https://www.ausgrid.com.au/Industry/Our-Research/Data-to-share/Solar-home-electricity-data (accessed on 21 October 2019).
- Yuan, C.; Yang, H. Research on K-value selection method of K-means clustering algorithm. J. Multidiscip. Sci. J.
**2019**, 2, 226–235. [Google Scholar] [CrossRef][Green Version] - PVlib, Python Solar Open Source Library. Available online: https://pvlib-python-dacoex.readthedocs.io/en/latest/index.html (accessed on 8 July 2019).
- Obtain Historical Weather Forecast Data in CSV Format Using Python. Available online: https://www.freecodecamp.org/news/obtain-historical-weather-forecast-data-in-csv-format-using-python/ (accessed on 16 September 2019).
- Home Solar Power in New South Wales. Available online: https://www.solarquotes.com.au/australia/solar-power-nsw/ (accessed on 2 February 2020).

**Figure 1.**Home classification according to photovoltaic (PV) installation, authorization, and sub meter.

**Figure 5.**PV capacity distribution histogram in [41].

**Figure 10.**Regional PV output forecasting distributions; (

**a1**) nRMSE of Case 1 (

**a2**) nMAE of Case1 (

**b1**) nRMSE of Case 2 (

**b2**) nMAE of Case2 (

**c1**) nRMSE of Case 3 (

**c2**) nMAE of Case 3.

Literature Group | Research Subject | Reference |
---|---|---|

1 | Regional utility scale photovoltaic (PV) forecasting | [12,13,14,15,16,17,18,19,20,21,22,23,24,25,26] |

2 | Regional behind the meter (BTM) PV forecasting | [28,34,35,36,37] |

3 | Unauthorized PV detection and PV capacity estimation | [27,38,39,40] |

Question | H1 | H2 | H3 | H4 |
---|---|---|---|---|

Is there a rooftop PV sub meter at home? | Yes | No | No | No |

Is a rooftop PV at home authorized? | Yes | Yes | No | No |

Is a rooftop PV installed at home? | Yes | Yes | Yes | No |

Parameters | Value | Parameters | Value |
---|---|---|---|

N_{home} | 1500 | r_{S}_{a}_{m} | 0.08 |

N_{PV} | 300 | t_{s} | 9 |

r_{Au} | 0.5 | t_{e}, t_{f} | 16, 19 |

**Table 4.**Unauthorized PV detection result by method in [39].

Metric | Best | Average | Worst |
---|---|---|---|

PA | 96.00 | 90.69 | 77.33 |

NPA | 99.67 | 96.58 | 89.67 |

OA | 98.00 | 95.93 | 90.15 |

Metric | Best | Average | Worst |
---|---|---|---|

PA | 100 | 99.81 | 96.67 |

NPA | 98.33 | 97.02 | 95.67 |

OA | 98.52 | 97.33 | 96.07 |

Study | Best (%) | Average (%) | Worst (%) |
---|---|---|---|

[35] | 66.00 | 92.79 | 113.00 |

Proposed method | 34.00 | 44.21 | 63.00 |

Error Metric (%) | Case 1 | Case 2 | Case 3 |
---|---|---|---|

Normalized Root Mean Square Error (nRMSE) | 11.29 | 6.41 | 5.41 |

Normalized Mean Absolute Error (nMAE) | 6.01 | 3.52 | 2.95 |

Feature | ${\mathbf{F}}_{1}^{\mathbf{D}}$ | ${\mathbf{F}}_{2}^{\mathbf{D}}$ | ${\mathbf{F}}_{3}^{\mathbf{D}}$ | ${\mathbf{F}}_{4}^{\mathbf{D}}$ | ${\mathbf{F}}_{5}^{\mathbf{D}}$ | ${\mathbf{F}}_{6}^{\mathbf{D}}$ |
---|---|---|---|---|---|---|

MIC value | 0.445 | 0.135 | 0.217 | 0.207 | 0.600 | 0.722 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Kim, T.; Kim, J. A Regional Day-Ahead Rooftop Photovoltaic Generation Forecasting Model Considering Unauthorized Photovoltaic Installation. *Energies* **2021**, *14*, 4256.
https://doi.org/10.3390/en14144256

**AMA Style**

Kim T, Kim J. A Regional Day-Ahead Rooftop Photovoltaic Generation Forecasting Model Considering Unauthorized Photovoltaic Installation. *Energies*. 2021; 14(14):4256.
https://doi.org/10.3390/en14144256

**Chicago/Turabian Style**

Kim, Taeyoung, and Jinho Kim. 2021. "A Regional Day-Ahead Rooftop Photovoltaic Generation Forecasting Model Considering Unauthorized Photovoltaic Installation" *Energies* 14, no. 14: 4256.
https://doi.org/10.3390/en14144256