Machine Learning-Based Health Index Evaluation of Power Transformers Using Novel Parameterization for Predictive Maintenance: Data-Driven Research on Pakistan’s National Grid Regarding Maintenance Cost Optimization
Aug 2026· Energies· Vol 19, pp. 3653· 0 citations· 42 references
TL;DR
A sophisticated method for determining the overall health condition of power transformers is presented and it is verified that the omission of four diagnostic tests can reduce the economic burden by 46.99%, yielding a cost saving of 259,000 PKR per transformer unit.
Abstract
Power transformers are an integral part of electrical power system infrastructure and play a vital role in the efficient and reliable transmission of bulk electrical power to distribution networks. The careful use of these assets enables the optimization of transmission voltages in power networks, a crucial step for reducing electrical energy losses, enhancing grid reliability, and ensuring uninterrupted electricity supply to end-users in interconnected power networks. Operational reliability of power transformers is a critical aspect in ensuring a continuous power supply. The health index (HI) is a crucial diagnostic tool to determine their real condition. Historically, HI assessments were based on scoring and weighting. Lately, however, there has been a significant change in the attitude towards the use of artificial intelligence (AI) and machine learning (ML) to predict the health of high-voltage power transformers. Although developments are taking place, the existing studies on ML-based HI prediction models for power transformers largely rely on an incomplete dataset containing improper parameters. Moreover, dependency on traditional ML models is a significant limitation when it comes to achieving a higher degree of predictive accuracy. This article presents a sophisticated method for determining the overall health condition of power transformers. A total of twenty of the most appropriate and highly relevant input parameters were selected to effectively evaluate the transformer condition. The dataset for these parameters was collected from real-time testing in accordance with international industry standards (i.e., IEC, IEEE, and ASTM), conducted at 220 kV and 500 kV grid stations in the Multan and Lahore regions, operated by the National Grid Company (NGC) in Pakistan. This comprehensive dataset was fed to five state-of-the-art ML models. The Categorical Boosting Regression (CatBoost Regressor) model demonstrated superior performance, achieving the highest accuracy (R2 Score) of 97.2% and the lowest mean absolute error (MAE) of 1.73. The best-performing model was then employed to predict the health index of the power transformers at the 500 kV grid station, Rahim Yar Khan, and the 500 kV grid station, Multan, as a practical case study. To demonstrate the economic importance of the proposed framework, an economic analysis was conducted via an iterative, parameter-skipping imputation strategy for maintenance cost optimization of the electrical power grid. The results verify that the omission of four diagnostic tests (i.e., Dissipation Factor, Capacitance, Insulation Resistance, and Transformer Turn Ratio) can reduce the economic burden by 46.99%, yielding a cost saving of 259,000 PKR per transformer unit. The implementation of this data-driven framework in the national grid can significantly reduce maintenance costs and facilitate an operational shift from traditional preventive maintenance to advanced predictive maintenance.
A machine learning-based approach for predicting the Transformer Health Index (HI) and the Remaining Life (RL) by means of the diagnostic parameters of transformer oil is described, allowing for timely planning of maintenance, reducing the occurrence of unexpected transformer failures, extending the service life of the assets and improving the overall efficiency of asset management for transformers in modern power systems.
S. V. Satyanarayana, P. S. Kumar· International journal of com...· 0 citations
Nigeria’s electricity system continues to experience reliability challenges arising from transmission and distribution faults, protection coordination difficulties, voltage and frequency deviations, ageing assets, and uneven network observability. Machine learning (ML) is increasingly proposed as a means of identifying abnormal operating states earlier and supporting faster, more selective intervention. This critical narrative review evaluates the extent to which the international fault-analytics literature can support genuine fault prediction and reliability improvement in the Nigerian smart-grid context. Literature published from 1 January 2010 to 22 June 2026 was examined, while older foundational Nigerian evidence was retained where it remained necessary for context. The evidence shows a pronounced terminological and translational gap. Most studies labelled as predictive apply ML after fault inception to detection, classification or localisation, commonly on simulated or synthetically augmented data. Convolutional, recurrent, ensemble, support-vector and hybrid methods often achieve very high test performance in controlled settings, but these results do not establish prospective prediction, field robustness or reduction in outage burden. Nigeria-specific evidence demonstrates technical feasibility for high-impedance-fault diagnosis and voltage-stability classification, yet field-validated prospective models linked to reliability outcomes remain scarce. The strongest near-term opportunity is therefore not autonomous fault anticipation in the abstract, but staged deployment of ML for condition-based risk scoring, abnormal-event recognition, protection support and maintenance prioritisation using utility data streams. Successful translation will require dependable sensing, labelled event histories, topology-aware modelling, uncertainty estimation, cybersecurity, model-drift governance and prospective evaluation against operational endpoints such as outage frequency, interruption duration, energy not supplied and protection misoperations. ML can contribute meaningfully to Nigerian grid reliability, but only when treated as one component of an engineering, data and governance architecture rather than as a substitute for protection design, asset renewal or network reinforcement.
A. B. Ogundare, J. Ogunyemi, S. Omolola et al.· Journal of Engineering Resea...· 0 citations
High-voltage assets are central to the reliability of modern power systems, yet their maintenance remains difficult because failures are rare, degradation is heterogeneous, and operational data are distributed across transformers, gas-insulated switchgear, circuit breakers, protection systems, and inspection platforms. This review examines how artificial intelligence can support predictive maintenance for high-voltage equipment in Saudi power systems. A structured review of literature published from 2020 to 2025 was conducted in line with PRISMA 2020 reporting principles. The evidence base comprises twenty-eight peer-reviewed studies and two official Saudi energy-sector reports covering dissolved gas analysis, partial-discharge recognition, thermal imaging, mechanical-state prediction, health indexing, remaining-life estimation, digital twins, anomaly detection, and deployment governance. The synthesis finds that artificial intelligence creates the greatest operational value when it combines multi-source condition data with asset context, uncertainty estimates, and maintenance decision rules rather than operating as an isolated fault classifier. Transformer applications are the most mature, particularly for dissolved gas analysis and health-index estimation, while gas-insulated switchgear and circuit-breaker applications show strong diagnostic accuracy but weaker evidence on long-term field generalisation. Saudi deployment conditions add specific requirements: extreme heat and dust, rapidly expanding transmission infrastructure, renewable-energy integration, large geographical service territories, and a strategic shift toward digital operations. The paper proposes a five-layer framework linking sensing, trusted data engineering, hybrid analytics, risk-based decision support, and continuous learning. It also identifies practical priorities for utilities, including asset-specific baselines, physics-informed models, explainability, cybersecurity, model-drift monitoring, and phased field validation. The review concludes that AI-driven predictive maintenance can strengthen reliability and lifecycle value in Saudi high-voltage networks, but its success depends more on data governance, engineering integration, and accountable maintenance workflows than on algorithmic accuracy alone.
M. Javeed· Global academic journal of e...· 0 citations
Condition assessment of a transformer provides information about the overall health status of the insulation. Accurate health index (HI) prediction at regular intervals prevents catastrophic failures. In this study, a novel multi-hierarchically weighted neural network is proposed to estimate the HI of power transformers. It is designed with 12 distinct features of an oil- and paper-insulation system. Initially, collected attributes are normalized using multi-criterion analysis (MCA). The correlation and appropriate weight prioritization of each attribute are determined using the analytical hierarchy process (AHP). The combined MCA-AHP facilitates the calculation of weighted scores and converts the 12 attributes into 3 quality grades. These grades are used as inputs to the hybrid artificial neural network (ANN), and the output is the overall HI. The model is trained and tested using 350 data samples collected from the Himachal Pradesh State Electricity Board, India. The performance of the proposed hybrid model is validated by root mean square error, mean absolute error, mean relative error, and correlation coefficient. Furthermore, a comprehensive comparison is conducted using 300 data samples with pre-known health conditions (HCs), and other expert models in the literature achieved 97% of accuracy. The proposed hybrid model effectively addresses the general issues raised by intelligent models, such as reliance on expert rule-based approaches in Fuzzy models, greater computational demand in ANN and ANFIS models, and increased complexity due to diagnostic attributes. Based on the predicted HCs, preventive maintenance actions are proposed to ensure effective maintenance of the asset.
M. Gopi, C. Ranga, K. Jagtap· Engineering Research Express· 0 citations
This article explores machine learning techniques (MLTs) as a modern alternative to enhance the interpretation of DGA data for early-stage fault detection in service transformers, and demonstrates that random forest and gradient boosting outperform others, achieving up to 98% accuracy.
Rupali Balabantaraya, A. Chatterjee, A. Sahoo et al.· Electrica· 0 citations
Accurate line-loss diagnosis and loss-reduction control are essential for low-voltage distribution transformer areas under high penetration of distributed energy resources (DERs). With electric vehicles (EVs), rooftop photovoltaic (PV) generation, and energy storage systems (ESS) connected at the customer side, traditional line-loss patterns are reshaped by bidirectional power flow, evening-concentrated charging behavior, voltage rise, and local flexibility. To address both abnormal-loss identification and operational mitigation, this paper proposes an integrated framework for transformer-area line-loss governance. First, the physical and operational factors affecting technical and non-technical line losses are analyzed, and simulation-supported sensitivity intervals are linked to DER penetration, power factor, load rate, and feeder impedance. Second, a two-stage anomaly diagnosis scheme is constructed, in which Isolation Forest screens suspicious samples from a 20-dimensional feature matrix and XGBoost classifies the detected abnormal samples into electricity theft, equipment failure, metering error, and topology error. Third, a collaborative loss-reduction strategy coordinates ESS dispatch, ordered EV charging, and PV hosting-capacity screening through time-series MATPOWER AC power-flow calculation. Case studies on a 500-household transformer area and the IEEE 69-bus radial feeder implemented in MATPOWER show that the classification stage achieves 99.02% accuracy and a macro-F1 score of 0.9917. The coordinated strategy selected by enumerative search reduces daily line loss by 25.3%-28.6%, with an average reduction of 27.1%, while reducing the maximum voltage deviation from 6.8% to 2.9%. The results demonstrate that combining data-driven diagnosis with DER-aware operational optimization can provide practical support for high-loss station governance in new power systems.
Cheng-Ming Liu, Jia-Xi Li, Ming Wen et al.· 2026 5th International Confe...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
