Inter national J our nal of P o wer Electr onics and Dri v e System (IJPEDS) V ol. 16, No. 3, September 2025, pp. 2072 2081 ISSN: 2088-8694, DOI: 10.11591/ijpeds.v16.i3.pp2072-2081 2072 Digital twin-based perf ormance e v aluation of a photo v oltaic system: A r eal-time monitoring and optimization framew ork Mustafa F adel 1,2 , F ajer M. Alelaj 2,3 1 Electrical Engineering Department, Mustansiriyah Uni v ersity , Baghdad, Iraq 2 School of Engineering, Ne wcastle Uni v ersity , Ne wcastle upon T yne, United Kingdom 3 K uw ait Institute for Scientic Research, K uw ait City , K uw ait Article Inf o Article history: Recei v ed May 23, 2025 Re vised Jun 17, 2025 Accepted Jul 23, 2025 K eyw ords: Digital twin Optimization Photo v oltaic system Real-time monitoring Rene w able ener gy system ABSTRA CT The digital t win (DT) technology implementation in photo v oltaic (PV) systems pro vides an inno v ati v e approach to real-time performance monitoring and predicti v e maintenance. In this paper , an end-to-end DT frame w ork for real-time performance analysis, f ault detection, and optimization of a 250 W PV system is propos ed. A ph ysics-based equation and AI-based prediction h ybrid DT model is de v eloped t hrough MA TLAB/Simulink, trained from real data acquired by means of a testbed. The DT simulates the dynamic ph ysical PV system beha vior and adjusts itself using self-correcting algorithms to enhance precision in prediction and forecast po wer out put at high delity . Results indicate that the DT gi v es the true response of the PV system with v ery small dif ferences attrib utable to model approximations and sensor f aults, 95% error minimization after compensation, and a root mean square error (RMSE) of 2.8 W , indicating its applicability for real-time monitoring and predicti v e main-maintenance. The w ork here focuses on the feasibility of applying DTs t o w ards the autonomous optimization of distrib uted rene w able ener gy systems. This is an open access article under the CC BY -SA license . Corresponding A uthor: Mustaf a F adel Electrical Engineering Department, Mustansiriyah Uni v ersity Baghdad, Iraq Email: mustaf a 1988abbas@uomustansiriyah.edu.iq 1. INTR ODUCTION Rene w able ener gy systems based on photo v oltaic (PV) ener gy ha v e become a crucial part of sustainable ener gy portfolios w orldwide, with increasing adoption dri v en by their sustainability and declining cost. Ho we v er , lar ge PV systems are subjected to man y challenges that af fect their performance and ef cienc y , due to their outdoor installations [1], [2]. Challenges such as technical system issues and en vironmental conditions act as obstacles af fecting the system’ s reliability and life span. The technical challenges include v oltage and current uctuations, high DC v oltage ripple, con v ersion ef cienc y , po wer quality issues, thermal management, and de grada tion o v er time [3], [4]. En vironmental challenges, including dust, sno w co v erage, and weather v ariability , also play important roles by impacting the PV system performance [5], [6]. T o maintain high operational ef cienc y under dif ferent en vironmental and technical conditions, sophisticated monitoring and optimization techniques are required. Digital twin (DT) technology emer ges as g ame-changing, by pro viding a virtual match to the real-w orld system in real-time. This mirror model of the ph ysical system can J ournal homepage: http://ijpeds.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 2073 of fer unprecedented potential in autonomous optimisation and predicti v e analytics [7]. DT f acilitates ongoing e v aluation of PV system performance by fus ing sensor -dri v en data collection and computational modelling, enabling proacti v e f ault diagnosis and impro v ed predicti v e maintenance techniques. The recent state of the art highlights the importance of emer ging DT in rene w able ener gy systems, where researchers in [8] found that the DT system could detect dif ferent f aults, such as 20% drift in sensor reading in the PV con v ersion unit, within a short duration. Delussu et al. [9] compare tw o dif ferent DT approaches to predict solar ener gy production and create a h ybrid DT system by combining the tw o studied systems to impro v e predictions. Lee et al. [10] propose a no v el generati v e data-dri v en model based on numerical weather prediction that ef fecti v ely produces en vironmental data to simulate the future beha viour of PV DT systems. Moreo v er , the study in [11] introduces an inno v ati v e DT system that inte grates a no v el circuit-long short-term memory (LSTM) model with a proposed triangle-shading pattern estimation method, eliminating dependencies on direct irradiance sensing and historical data. 2. DIGIT AL TWIN TECHNOLOGY T o understand ho w a DT functions, tw o k e y aspects must be addressed. This section will outline the DT design requirements, including the necessary tools for b uildi ng an ef cient virtual system. Furthermore, it e xplores the de v elopment stages of the DT , focusing on its intended function and operational capabilities. 2.1. Digital twin design r equir ements T o de v elop an ef fecti v e DT , there are se v eral technical, data, softw are, and engineering requirement s that must be met to ensure simulation accurac y and optimize performance. Figure 1 illustrates the DT design system, where each PV DT needs hardw are infrastructures represented as dif f erent types of sensors such as en vironmental sensors represented by Pyranometers that are used to measure the amount of f alling irradiance on PV modules, temperature sensors for PV panels and ambient, humidity , wind speed, and dust sensors to assess the impact of the en vironment on performance, or electrical measurements sensors such as v oltage, current, and po wer measurements from PV modules and in v erters [12], [13]. In addition, each DT requires an edge computing unit that is used to process the initial data before sending it to the cloud to reduce latenc y and impro v e DT responsi v eness. T o ensure data inte grity , a reliable communication netw ork is required, for e xample, dif ferent IoT protocols such as Mess age Queuing T elemetry T ransport (MQTT), Open Platform Communications Unied Architecture (OPC U A), or Modb us is required, especially for applications needing f ast response and in remote locations [4], [5]. Moreo v er , the quality of recei v ed data and its accurac y play a crucial role in DT per formance, where data must be up-to-date, accurate, and uninterrupted to ensure the reliability of the numerical simulation. Additionally , using cloud databases can be ef cient for big data processing g athered from real-time analysis and using machine learning (ML) techniques to analyze big data, predict f ailures, and impro v e performance. The presence of softw are support is essential in b uilding a rugged DT system [14], [15]. Data management and articial intelligence (AI) platforms can predict ener gy consumption, detect f aults, and create interacti v e dashboards to display data and analyze performance. Furthermore, softw are programs lik e MA TLAB/Simulink © and NSYS © that use modelling and simulation engines can play an important role in de v eloping accurate ph ysical numerical models based on PV performance equations and the ef fect of heat on solar panels. Engineering and ph ysical requirements are important to design inte grated models for PV systems, where the DT must be able to simulate all PV system components. Components such as solar panels and the ef fect of gradual corrosion, in v erters’ ef fect on performance, batteries (storage systems) and their life c ycle based on char ging and dischar ging patterns, and inte gration with the elect rical grid and ener gy o w management [16]. Moreo v er , the DT must be designed to be e xpandable to include lar ge solar po wer plants or inte grated smart electrical grids, and the design must be compatible with the e xisting solar infrastructure to f acilitate inte gration without the need to redesign the system from scratch [17]. 2.2. Digital twin de v elopment stages Digital twin creation is a highly adv anced process that mo v es in dif ferent steps to de v elop an accurat e digital model of the ph ysical system’ s real life. As indicated in Figure 2, the process of DT de v elopment can be di vided into a series of stages. Data collection and system inte gration phase, where the DT is based primarily on real data coming from the s olar system pro vided by dif ferent types of sensors deplo yed, and ensuring that data coming from dif ferent de vices and sources is synchronized. Furthermore, pro viding communication Digital twin-based performance e valuation of a photo voltaic system: A r eal-time ... (Mustafa F adel) Evaluation Warning : The document was created with Spire.PDF for Python.
2074 ISSN: 2088-8694 infrastructure to ensure that the netw ork can transmit data in real time wi thout signicant delay and store the g athered data in an appropriate storage mechanism [5]. Model de v elopment and simulation is the ne xt phase of DT de v elopment, where in this stage the DT will be b uilt depending on type select ion whether it is a ph ysics-based twin that relies on ph ysical equations to model the beha vior of the system, data-dri v en twin that relies on AI and data analysis, or h ybrid twin that combines ph ysical modeling and AI to inc rease accurac y [16]. At this stage, model v alidation and calibration are e xamined by comparing DT results with real data to detect discrepancies and correct errors, and tune the model based on actual operating data to impro v e accurac y and reliability . The ne xt stage is system inte gration and control, where the digital twin is link ed to the actual s y s tems, and intelligent control mechanisms are acti v ated. Inte gration with systems t o monitor the performance of the PV system and to sho w information and analysis. Additionally , with the use of predicti v e analytics for f ault detection in adv ance by de v eloping deep learning systems and analyzing patterns for impro v ed ener gy ef cienc y , and enabling automatic control and performance optimization t hrough dynamic adjustment of solar in v erter parameters based on actual operating condit ions [14]. The stage of softw are de v elopment is the nal and continuous observ ation, wherein DT must be e x ecuted and observ ed so that the goals to be realized are achie v ed by e x ecuting and testing the system in real l ife to see ho w the digital twin beha v es for dif ferent operating conditions to conrm its stability , and comparing the predicted results with real data from the real system. Furthermore, continuously updating digital models as ne w operating data becomes a v ailable and tuning softw are and algorithms to impro v e performance based on changes in operating conditions by relying on the DT to predict potential f ailures and tak e proacti v e measures, and run real-time analytics to detect an y decline in system ef cienc y [18]. D i g i t al   d at a   ex ch an g e Rea l - t i me   d at a A ct i o n A g g reg at e Fee d b ack A n a l y s e Co n t ro l Sen s o rs D at a A n a l y t i cs A ct u at o r Ph y s i cal   s y s t em V i rt u al   s y s t em Figure 1. Digital twin system architecture D i g i t al  T w i n   T ech n o l o g y Dat Sen s i n g D at Visu al i s at i o n D i g i t al   T u n i n g Cl o u d   Co mp u t i n g D at In t e g ra t i o n D at Ana l y s i s A ct u al   s y s t e m A rt i fi ci al   In t e l l i g en ce Figure 2. Digital twin de v elopment stages 3. DIGIT AL TWIN APPLICA TIONS IN PV SYSTEM The modern PV systems utilise the DT to predict and optimise their performance. Pro viding rea l-time monitoring to the full PV system can maximise its performance under dif ferent en vironmental conditions. Real-time data are g athered from the PV system to be analysed by the virtual twin, resulting in early prediction of future system issues [6]. Moreo v er , by using AI tools, the DT can predict possible f aults in the PV system, whether at the PV side le v el, such as PV modules ef cie n c y dropping due to en vironmental condition Int J Po w Elec & Dri Syst, V ol. 16, No. 3, September 2025: 2072–2081 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 2075 uctuations, or con v erter le v el, where components f ail (inductor , capacitor , or switches). Spotting these issues can increase the system’ s reliability by reducing system f ault periods and enabling predicti v e maintenance [19], [20]. Furthermore, increasing the PV system ef cienc y can be achie v ed by inte grating such technology and e xperimenting with dif ferent scenarios, such as PV module implementation angles to optimise irradiance e xposure, the shado wing ef fect, and dif ferent en vironmental conditions, pro viding the best settings to increase system ef cienc y [21]. Also, the DT can be used in smart grids to distrib ute the po wer ef ciently by storing it or redirecting it to the main grid when needed. Moreo v er , incorporating DT can help to modify the system design, such as adding ne w units or changing storage technologies before implementing the ph ysical system, impacting the cost and time [22]. 4. CHALLENGES OF DT IN PV SYSTEM In spite of man y benets being link ed with combining DT technology with PV systems, there are se v eral issues related to the use of such technologies. Use of DT entails se v eral technical issues, including the creation of a precise virtual model for the PV system that requires proper understanding of dif ferent beha viour of the components of the PV system, such as PV modules, in v erters, batteries, and en vironmental conditions. Modelling such a system requires adv anced calculations, including the dynamic changes in irradiance, ambient temperature, and normal ageing in PV modules [14]. In a dd i tion, dif ferent designs and PV system performance can v ary from one location to another making it challenging to de v elop a general model that suits all applications. Furthermore, DT relies on data acquired from sensor equipment, ener gy transducers, weather databases, and IoT analytics, with non-unied standardisation for v arious system equipment mak es it dif cult to inte grate information from dif ferent m ak ers [23]. Response time and real-time analysi s are others tough ones, in which the DT must assess data and handle errors in real time, that can be back ed by po werful cloud computing serv ers and rapid connecti vity netw orks [16]. Latenc y in data transmission may af fect the speed of decision-making in critical situations, such as sudden f ailures or lo w ener gy ef cienc y . In addition, DT accurac y can be af fected by uctuations in weather f actors, where DT depends on the a v ailability of accurate climate data, and some areas lack of accurate data or are af fected by une xpected climate changes, such as dust storms or sudden irradiance changes that may af fect the accurac y of DT forecasts [5]. Furthermore, operating in harsh en vironments requires re gular maintenance and continuous replacement, leading to increase operating and maintenance costs. 5. PR OPOSED DT SYSTEM T o understand the DT w ork and test it in a real e n vi ronment, it is essential to e xamine it in real -w orld circumstances with actual data. The full PV system is illustrated in Figure 3(a), which consists of multiple PV units, each connected to the main DC-b us through DC boost con v erter and connected to a central in v erter for microgrid applicat ion. The e xamination will tak e place through b uilding a virtual twin that simulates the performance of the ph ysical system via sensing and analysing real-w orld data, where a data acquisition unit logging me asurements at 1-minute interv als through sensor netw ork for capturing irradiance, temperature, v oltage, and current. The model uses a trained feedforw ard neural netw ork (FNN) de v eloped using MA TLAB’ s Neural Netw ork T oolbox with 10,080 training samples (1-week, 1-minute resolution). It is a tw o-hidden-layer , 10 then 15 neuron model, and uses Le v enber g-Marquardt (trainl m) training algorithm. There is a module for detecting errors, which measures the dif ference between the prediction and actual po wer . When the dif ference e xceeds 5W , a rolling a v erage for 5 interv als is computed and is used to bia s the prediction result adapti v ely . Resulting in a mirrored system uses a feedback loop for error detection and correction, used to o pt imise the o v erall system performance and minimise future maintenance. Figure 3(b) sho ws the circuit diagram of the DC/DC boost con v erter po wered by a PV module with its specications illustrated in T able 1. Building the DT will be based on modelling the basic ph ysical beha viour of the PV system using mathematical equations for the PV cell [ ? ]. F or a single-diode PV module, the current-v oltage relationship for a PV module is gi v en by (1). I pv = I ph I d exp V + I pv R s nV t 1 V + I pv R s R sh (1) Where I pv is the PV output current, V pv is output v oltage, I ph is the photocurrent, I 0 is the diode re v erse saturation current, R s is the series resistance, R sh is the shunt resistance, n diode ideality f actor , and V t is the Digital twin-based performance e valuation of a photo voltaic system: A r eal-time ... (Mustafa F adel) Evaluation Warning : The document was created with Spire.PDF for Python.
2076 ISSN: 2088-8694 thermal v oltage gi v en by V t = k T q . Where k is Boltzmann constant ( 1 . 380649 × 10 23 J/K), T is the PV cell temperature, and q is the electron char ge ( 1 . 602176634 × 10 19 C). G ri d G 1 ,T 1 + _ + _ D C co n v ert er PV  mo d u l e G 2 ,T 2 P V 1 + _ + _ P V 2 + _ + _ P V n G n ,T n D C/ A Co n v er t er D C bus D i g i t al  T w i n Co n t ro l A n a l y s e  &  Feed b ack G ,T I pv ,V pv I,V ( boost ) Ph y s i cal   s y s t em V i rt u al   t w i n (a) C in D b o o s t I D I c in I pv + _ G, T PV PV PV + _ I L I b o o s t L b o o s t C b o o s t I c o S b o o s t (b) Figure 3. The o v erall DT proposed system: (a) PV po wer system and (b) po wer con v ersion unit T able 1. PV panel parameters P arameter Symbol V alue Unit Rated po wer P pv r ated 250 W V oltage at max po wer V mp 30 V Current at max po wer I mp 8.33 A Open-circuit v oltage V oc 37 V Short-circuit current I sc 8.75 A Number of cells N cells 60 - T emperature coef cient of po wer β -0.45 % / C T emperature coef cient of v oltage K V -0.34 V / C T emperature coef cient of current K I 0.05 A / C The photocurrent ( I ph ) released by the PV cell is a function of the solar irradiance ( G ) and temperature ( T ) gi v en as (2) [8]. I ph = ( I ph,S T C + K I ( T T S T C )) G G S T C (2) Where the standard test conditions (STC) are at T = 25 C and G = 1000 W/m 2 . The diode re v erse saturation current is represented as (3) [12]. I 0 = I 0 ,S T C T T S T C 3 exp E g nk 1 T S T C 1 T  (3) Int J Po w Elec & Dri Syst, V ol. 16, No. 3, September 2025: 2072–2081 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 2077 Both open-circuit v oltage ( V oc ) and short-circuit current ( I sc ) are sho wn as (4) [25]. V oc ( T ) = V oc,S T C + K V ( T T S T C ) (4) I sc ( T ) = I sc,S T C + K I ( T T S T C ) (5) Where K I and K V are the temperature coef cients of current ( A/ C) and v oltage ( V / C) respecti v ely . The maximum PV po wer is illustrated as (6). P max ( T ) = P max,S T C (1 + β ( T T S T C )) (6) Where β is the temperature coef cient of po wer . 6. SIMULA TION RESUL TS AND DISCUSSION The proposed system will ha v e a 250 W PV module, where a parallel MA TLAB/Simuli nk © model will simul ate the e xpected PV output using single-di od e mathematical equat ions and utilise ML for po wer forecasting. Using irradiance (p yranometer) and temperature (PT100) sensors to tak e real-time measurements will help to compute PV panel performance under dif ferent en vironmental conditions. The operation of the DT w as v alidated using real and synthetic data under incremental irradiance (200 to 1000 W/m 2 ) and temperature (15 C to 35 C) v ariations. The proposed system uses adv anced visualisation by pro viding comprehensi v e performance plots for system beha viour to v arying en vironmental f actors. Figure 4(a) illustrates both real and digital output po wer response to step-wise irradiance increase, while Figure 4(b), represents the performance under ambient temperature increase, where the output po wer is af fected ne g ati v ely at higher temperature. Furthermore, Figure 4(c), re v eal s the ef fects on PV performance under v arying sunlight and temperature conditions. From the illustrated results, it is noticeable that the DT correlate d well with the measurements, presenting a small error f actor . The AI-enriched DT e xhibited high accurac y and lo w lag in transitions. (a) (b) (c) Figure 4. PV output po wer beha viour under: (a) increasing irradiance, (b) temperature increase, and (c) v arying irradiance and temperature. Digital twin-based performance e valuation of a photo voltaic system: A r eal-time ... (Mustafa F adel) Evaluation Warning : The document was created with Spire.PDF for Python.
2078 ISSN: 2088-8694 T o test the reliability of DT and ho w it can adapt to sensor errors and en vironmental changes without manual interv ention, a self-de gradation and correction analysis tak es place by continuously comparing the predicted po wer output with the actual measured output. The DT detects periods when predictions de viate signicantly from the real system be yond a set threshold v alue of 5 W ( 2%) of the maximum PV po wer (250 W), and this de viation persists for three consecuti v e samples, it ags an anomal y . T o correct t his, the system emplo ys a v e-frame rolling a v erage of past prediction errors and adjusts subsequent predictions by an adapti v e bias. This light-weight correction scheme greatly impro v es real-time alignment without computational o v erhead. As sho wn in Figure 5, 39 errors were correctly detected, with no f alse positi v es. Observing the system’ s adaptability to these errors. The quantitati v e numerical performance of the DT model w as e v aluated quantitati v ely ag ainst three common measures of error: root m ean square error (RMSE), mean absolute error (MAE), and the coef cient of determination (R 2 ). The DT achie v ed an RMSE of 2.8 W , which indicates that, on a v erage, the simulated po wer output de viated from the measurement by 2.8 w atts, with greater sensiti vity to lar ger errors. The MAE w as 2.1 W , representing a lo w and relati v ely stable mean absolute prediction error o v er the dataset. The model also had an R 2 of 0.984, which conrms that 98.4 % of the v ariance i n actual PV po wer output w as well represented by the predictions from the DT . These results conrm the high accurac y and reliability of the DT model in the simulation of the real-time operating beha viours of the photo v oltaic system under v arious en vironmental conditions. Figure 6 displays ho w self-correction mechanisms adjust these de viations as the DT corrections align closely with real system performance, with error reduction after correction reaching 95%. T o ensure rugged performance under v arying conditions, po wer loss trends o v er time are highlighted in Figure 7, where it represents ho w much po wer the DT model underesti mates compared to the ph ysical system o v er time, and the maximum po wer error found is 11.2 W . This is crucial as identifying where and when the losses occur can be ef cient for rening the DT’ s prediction model, resulting in optimizing the o v erall performance. Figure 8 compares the dynamic ef cienc y of the ph ysical PV system and digital twin in a period of 50 seconds. The tw o curv es are virtually identical with a dif ference of less than 0.1%, conrming the reliability of the DT to reproduce the dynamic performance of the system. The agreement pro v es the reliability of the DT to track ef cienc y and its applicability for diagnosti cs and optimization. Figure 9 illustrates a histogram of po wer prediction errors between the actual PV system and digital twin, from 4 W to 11 W . Most errors are concentrated in the ranges 5–6 W and 10–11 W , indicating t hat while DT is generally accurate, lar ger de viations could occur during dynamic transitions or sensor f aults. These v ariations are within the real-time prediction tolerance, and the spread w arrants the use of DT’ s self-correction f acility to adapti v ely reduce such discrepancies. Figure 5. V irtual twin f ault detection and de viation Figure 6. Po wer output with digital twin correction Int J Po w Elec & Dri Syst, V ol. 16, No. 3, September 2025: 2072–2081 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 2079 Figure 7. Mirrored system po wer loss o v er Figure 8. Histogram of digital twin po wer prediction Figure 9. Histogram of digital twin po wer prediction errors 7. CONCLUSION The imple mentation of DT technology on rene w able ener gy systems is a tremendous adv ancem ent in real-time monitoring, f ault diagnosis, performance impro v ement, and predicti v e maintenance. A h ybrid DT model of a 250 W PV system w as proposed in this study by the fusion of ph ysics-based equations and AI-based prediction. The system w as v ery accurate in replicating real-w orld PV beha viour , with an RMSE of 2.8 W and a coef cient of determination (R 2 ) of 0.984. Reliability w as also enhanced by an inherent self-correcting mechanism, with reductions in error de viations of up to 95% across 39 anomalies that were detected. Po wer loss analysis sho wed the 11.2 W peak prediction error in the dynamic en vironment conditions, re v ealing the DT’ s v ariability diagnosis and compensation capability in pract ical operation. System e xperimental v erication w as conducted using r eal-time data, illustrati ng pragmatic usefulness. These ndings depict the potential of DTs in f acilitating smart, autonomous ener gy systems that react dynamically and pa v e the w ay for inte gration with storage, smart grids, and edge computing platforms. A CKNO WLEDGMENTS The rst author w ould lik e to thank the Iraqi go v ernment, represented by the Ministry of Higher Education and Scientic Research, Mustansiriyah Uni v ersity (www .uomustansiriyah.edu.iq), for their support in this w ork. Digital twin-based performance e valuation of a photo voltaic system: A r eal-time ... (Mustafa F adel) Evaluation Warning : The document was created with Spire.PDF for Python.
2080 ISSN: 2088-8694 FUNDING INFORMA TION Authors state no funding in v olv ed. A UTHOR CONTRIB UTIONS ST A TEMENT (MAND A T OR Y) This journal uses the Contrib utor Roles T axonomy (CRediT) to recognize indi vidual author contrib utions, reduce authorship disputes, and f acilitate collaboration. Name of A uthor C M So V a F o I R D O E V i Su P Fu Mustaf a F adel F ajer M. Alelaj C : C onceptualization I : I n v estig ation V i : V i sualization M : M ethodology R : R esources Su : Su pervision So : So ftw are D : D ata Curation P : P roject Administration V a : V a lidation O : Writing - O riginal Draft Fu : Fu nding Acquisition F o : F o rmal Analysis E : Writing - Re vie w & E diting CONFLICT OF INTEREST ST A TEMENT Authors state no conict of interest. D A T A A V AILABILITY Deri v ed data support ing the ndings of this study are a v ailable from the corresponding author , [MF], on request. REFERENCES [1] G. S. Sarma, B. R. Reddy , P . Nir gude, and P . V . Naidu, A long short-term me mory based prediction model for transform er f ault diagnosis using dissolv ed g as analysis with digital twin technology , International J ournal of P ower Electr onics and Drive Systems (IJPEDS) , v ol. 13, no. 2, pp. 1266–1276, Jun. 2022, doi: 10.11591/ijpeds.v13.i2.pp1266-1276. [2] Y . Fu, Y . Huang, F . Hou, and K. Li, A brief re vie w of digital twin in electric po wer industry , in 2022 IEEE 5th International Electrical and Ener gy Confer ence (CIEEC) , IEEE, May 2022, pp. 2314–2318. doi: 10.1109/CIEEC54735.2022.9846081. [3] J . Y uan, Z. T ian, J. Ma, K. L. Man, and B. Li, A digital twin approach for modeling electrical characteristics of bif acial solar panels, in 2022 International Confer ence on Industrial IoT , Big Data and Supply Chain (IIoTBDSC) , IEEE, Sep. 2022, pp. 317–321. doi: 10.1109/IIoTBDSC57192.2022.00065. [4] N . J. Johannesen, M. L. K olhe, and A. D. Jacobsen, “Correlation analysis of potential solar photo v oltaic po wer plant inte gration at wind f arm with grid connection limits, in 2023 8th International Confer ence on Smart and Sustainable T ec hnolo gies (SpliT ec h) , IEEE, Jun. 2023, pp. 1–6. doi: 10.23919/SpliT ech58164.2023.10193149. [5] K. T . Alao, S. I. U. H. Gilani, K. Sopian, and T . O. Alao, A re vie w on digital twin application in photo v oltaic ener gy systems: challenges and opportunities, JMST Advances , v ol. 6, pp. 257–282, Sep. 2024, doi: 10.1007/s42791-024-00083-z. [6] S. Ebrahimi, M. Se yedi, S. M. S. Ullah, and F . Ferdo wsi, “Resilient space operations with digital twin for solar PV and storage, IEEE Open Access J ournal of P ower and Ener gy , v ol. 11, pp. 624–636, 2024, doi: 10.1109/O AJPE.2024.3508576. [7] N. K umari, A. Sharma, B. T ran, N. Chilamkurti, and D. Alahak oon, A comprehensi v e re vie w of digital twin technology for grid-connected microgrid systems: state of the art, potential and chal lenges f aced, Ener gies , v ol. 16, no. 14, p. 5525, Jul. 2023, doi: 10.3390/en16145525. [8] P . Jain, J. Poon, J. P . Singh, C. Spanos, S. R. Sanders, and S. K. P anda, A digital twin approach for f ault diagnosis in distrib uted photo v oltaic systems, IEEE T r ansactions on P ower Electr onics , v ol. 35, no. 1, pp. 940–956, Jan. 2020, doi: 10.1109/TPEL.2019.2911594. [9] F . Delussu, D. Manzione, R. Meo, G. Ottino, and M. Asare, “Experiments and comparison of digital twinning of photo v oltaic panels by machine learning models and a c yber -ph ysical model in Modelica, IEEE T r ansactions on Industrial Informatics , v ol. 18, no. 6, pp. 4018–4028, Jun. 2022, doi: 10.1109/TII.2021.3108688. [10] J. Lee, J. Kang, S. Son, and H.-M. Oh, “Numerical weather data-dri v en sensor data generation for PV digital twins: a h ybrid model approach, IEEE Access , v ol. 13, pp. 5009–5022, 2025, doi: 10.1109/A CCESS.2025.3525659. [11] D. Hong, J. Ma, K. W ang, K. L. Man, H. W en, and P . W ong, “Real-time po wer prediction for bif a cial PV systems in v aried shading conditions: a circuit-LSTM approach within a digital twin frame w ork, IEEE J ournal of Photo vol taics , v ol. 14, no. 4, pp. 652–660, Jul. 2024, doi: 10.1109/JPHO T O V .2024.3393001. [12] M. S. Abdelrahman, H. M. Hussein, I. Kharchouf, S. M. S. H. Ran, and O. A. Mohammed, “Digital twin- based approach for monitoring and e v ent detection in PV systems, in 2024 IEEE International Confer ence on En vir onment and Electrical Engineering and 2024 IEEE Industrial and Commer cial P ower Systems Eur ope (EEEIC / I&CPS Eur ope) , IEEE, Jun. 2024, pp. 1–6. doi: 10.1109/EEEIC/ICPSEurope61470.2024.10751085. Int J Po w Elec & Dri Syst, V ol. 16, No. 3, September 2025: 2072–2081 Evaluation Warning : The document was created with Spire.PDF for Python.
Int J Po w Elec & Dri Syst ISSN: 2088-8694 2081 [13] M. W alters, J. Y once, and G. K. V enayag amoorth y , “Data-dri v en digital t wins for po wer estimations of a solar photo v oltaic plant, in 2023 IEEE Symposium Series on Computational Intellig ence (SSCI) , IEEE, Dec. 2023, pp. 246–251. doi: 10.1109/SSCI52147.2023.10371834. [14] M. K olahi, S. M. Esmailif ar , A. M. M. Sizk ouhi, and M. Aghaei, “Digital-PV : A digital twin-based platform for autonomous aerial monitoring of lar ge-scale photo v oltaic po wer pl ants, Ener gy Con ver sion and Mana g ement , v ol. 321, p. 118963, Dec. 2024, doi: 10.1016/j.enconman.2024.118963. [15] M. P asandideh, S. R. T ito, M. Apperle y , and M. Atkins, “Design and implementation of a simple lo w-cost and real-time solar po wer monitoring system, in 2023 IEEE F ifth International Confer ence on DC Micr o gri ds (ICDCM) , IEEE, No v . 2023, pp. 1–6. doi: 10.1109/ICDCM54452.2023.10433596. [16] X. Zhang, Y . Li, T . Li, Y . Gui, Q. Sun, and D. W . Gao, “Digital twin empo wered PV po wer prediction, J ournal of Modern P ower Systems and Clean Ener gy , v ol. 12, no. 5, pp. 1472–1483, 2024, doi: 10.35833/MPCE.2023.000351. [17] S. Zhang, Z. W ang, J. M. Cheung, E. W . L. Y ip, Z. Xu, and D. Qi, A no v el solar ener gy management syst em based on digital twin technique, in IET Confer ence Pr oceedings , Feb . 2023, pp. 353–357. doi: 10.1049/icp.2023.0126. [18] S. Ro y , A. Sarw at, M. Behnamf ar , A. Debnath, M. T ariq, and P . McClusk e y , An ML-enhanced digital twin mode l of photo v oltaic in v erter for estimating component de gradation, in 2024 IEEE Ener gy Con ver sion Congr ess and Exposition (ECCE) , IEEE, Oct. 2024, pp. 1610–1617. doi: 10.1109/ECCE55643.2024.10861035. [19] P . Jain, J. P . Singh, and S. K. P anda, “F ault remediation for distrib uted photo v oltaic (PV) system, in 2019 IEEE Applied P ower Electr onics Confer ence and Exposition (APEC) , IEEE, Mar . 2019, pp. 1057–1064. doi: 10.1109/APEC.2019.8722145. [20] Z. W enhan et al. , Application of digital twin system in po wer transformer f ault detection, in 2023 IEEE 6th International Electrical and Ener gy Confer ence (CIEEC) , IEEE, May 2023, pp. 2123–2127. doi: 10.1109/CIEEC58067.2023.10167284. [21] A. Driesse, “Identication of de gradation and f aults in PV systems using a time-v arying ef cienc y model, in 2024 IEEE 52nd Photo voltaic Specialist Confer ence (PVSC) , IEEE, Jun. 2024, pp. 0425–0429. doi: 10.1109/PVSC57443.2024.10749357. [22] A. Li v era et al. , “Intelligent cloud-based monitoring and control digital twin for photo v oltaic po wer plants, in 2022 IEEE 49th Photo voltaics Specialists Confer ence (PVSC) , IEEE, Jun. 2022, pp. 0267–0274. doi: 10.1109/PVSC48317.2022.9938505. [23] A. V . V oloso v a, P . F . Y urchik, V . B. Golubk o v a, B. S. Subbotin, and A. V . V asilie v , “Using a digital twin in ultra-lar ge-scale system, in 2023 Intellig ent T ec hnolo gies and Electr onic De vices in V ehicle and Road T r ansport Comple x (TIR VED) , IEEE, No v . 2023, pp. 1–5. doi: 10.1109/TIR VED58506.2023.10332708. [24] P . J. Hueros-Barrios, F . J. R. S ´ anchez, M. T radacete- ´ agreda, P . Mart ´ ın, C. S. P ´ erez, and D. Perez-Saura, A lo w-cost digital twin for real-time monitoring of photo v oltaic panels, in 2023 IEEE 32nd International Symposium on Industrial Ele ctr onics (ISIE) , IEEE, Jun. 2023, pp. 1–6. doi: 10.1109/ISIE51358.2023.10228172. [25] K. K. Saini, C. Y ada v , H. D. Mathur , and A. R. Gautam, “Digital twin simulation and comparati v e study to understand the microgrid operation and control, in 2024 IEEE International Confer ence on P ower and Ener gy (PECon) , IEEE, No v . 2024, pp. 116–121. doi: 10.1109/PECon62060.2024.10827482. BIOGRAPHIES OF A UTHORS Mustafa F adel recei v ed his B.Sc. de gree in Electri cal Engineering from Mustansiriyah Uni v ersity , Baghdad, Iraq, in 2010, and his M.Sc. de gree in Electrical Engineering from the same uni v ersity in 2016. He is currently pursuing a Ph.D. at Ne wca stle Uni v ersity , United Kingdom. From No v ember 2016 to May 2019, he serv ed as an Assistant Lecturer in the Electrical Engineering Department at Mustansiriyah Uni v ersity , where he w as later promoted to Lecturer in October 2019. His r esearch interests focus on the design and control of po wer elect ronic con v erters f or rene w able ener gy syste ms, el ectric v ehicles, battery c har ging systems, and wirele ss po wer transfer applications. Mustaf a w as the recipient of the First Prize P aper A w ard at the IEEE International Symposium on Electrical, Ele ctronics, and Information Engineering, Leicester , United Kingdom, in 2024. He can be contacted at email: mustaf a 1988abbas@uomustansiriyah.edu.iq. F ajer M. Alelaj w as a w arded a Bachelor’ s de gree in Electrical Engineering from K uw ait Uni v ersity in 2014, earning a place on the honor list twice. She obtained her MSc in Electrical Po wer from Ne wcastle Uni v ersity , United Kingdom, in 2020, graduating with First Class Honours. She has recentl y completed her Ph.D. in Hybrid Distrib ution T ransformers at Ne wcastle Uni v ersity . She w as also the recipient of the M.Sc. Prize for the top-rank ed postgraduate taught student in Electrical & Electronic Engineering for the academic year 2019/2020. She can be contacted at email: felaj@kisr .edu.kw . Digital twin-based performance e valuation of a photo voltaic system: A r eal-time ... (Mustafa F adel) Evaluation Warning : The document was created with Spire.PDF for Python.