Defense of the dissertation of Taganova Guldana for the degree of Doctor of Philosophy (PhD) in the specialty «8D06103 - Information systems»

L.N. Gumilyov Eurasian National University, a dissertation defense for the degree of Doctor of Philosophy (PhD) by Taganova Guldana on the topic «Development of an information system for load forecasting and optimization of electric power systems» to the educational program «8D06103 – Information systems».

The dissertation was carried out at the «Information Systems education department» of L.N. Gumilyov Eurasian National University.

The language of defense is kazakh

Official reviewers:

Zhanar Orymbayevna Oralbekova – Doctor of Philosophy (PhD), Associate Professor, Acting Professor at the «School of Artificial Intelligence and Data Science», Astana IT University LLP (Astana, Republic of Kazakhstan).

Gulshat Amanzholovna Amirkhanova – Doctor of Philosophy (PhD), Assistant Professor of the Department of Artificial Intelligence and Big Data at Al-Farabi Kazakh National University (Almaty, Republic of Kazakhstan).

Temporary members of the Dissertation Committee:

Vladimir Barakhnin – Doctor of Technical Sciences, Associate Professor, Leading Researcher at the Federal Research Center for Information and Computing Technologies (FRC ICT) (Novosibirsk, Russia);

Orken Zhumazhanovich Mamyrbayev – Doctor of Philosophy (PhD), Professor, Deputy Director for Science at the RSE «Institute of Information and Computational Technologies" of the Science Committee of the Ministry of Science and Higher Education of the Republic of Kazakhstan (Almaty, Republic of Kazakhstan);

Ormanbekova Ainur Alibekovna – Doctor of Philosophy (PhD), Assistant Professor of the Department of Automation and Robotics at the Almaty University of Technology (Almaty, Republic of Kazakhstan).

Scientific advisors:

Tussupov Jamalbek Aliaskarovich – Doctor of Physico-Mathematical Sciences, Professor of the Department of Information Systems of the L. N. Gumilyov Eurasian National University (Astana, Republic of Kazakhstan);

Waldemar Wójcik – Doctor of Technical Sciences, Professor at the Department of Information Technology of Electronics at the Lublin University of Technology (Lublin, Poland).

The defense will take place on July 03, 2026, at 12:00 PM in the Dissertation Council for the training direction «8D061 – Information and communication technologies» in the specialty «8D06103 – Information systems» of L.N. Gumilyov Eurasian National University. The defense meeting is planned to be held online.

Link: https://teams.microsoft.com/meet/4391040662843?p=xVW6Uwdlc4QkkjxISR 

Address: г. Астана, ул. Пушкина, 11, Учебный корпус № 2, аудитория № 222.

Abstract (English): ABSTRACT of the dissertation work by Guldana Taganova on the topic «Development of an information system for load forecasting and optimization of electric power systems» submitted for the degree of Doctor of Philosophy (PhD) under the educational program «8D06103 - Information Systems» Relevance of the study. At the present stage, the operating modes of electric power systems are changing due to the increasing share of renewable energy sources, the growing complexity of consumption patterns, and the expansion of digital data flows. The development of solar energy, in particular, complicates the task of anticipatory assessment of the balance between electricity generation and consumption. Since solar generation depends on weather conditions, time of day, seasonal variation, cloudiness, temperature, and radiation level, it cannot be accurately described by traditional methods alone. In the context of Kazakhstan, this issue has additional significance. The country has a vast territory, diverse climatic zones, and uneven regional distributions of solar resources and consumption structure. While the southern regions have high solar generation potential, the northern and central regions are characterized by different load behaviour and seasonal consumption patterns. Under such conditions, an information system is required that considers load and solar generation not separately, but in their mutual relationship. The relevance of the dissertation research is determined by its treatment of load and solar generation forecasting results not merely as numerical predictions, but as a decision-support tool for identifying energy deficit, surplus energy, the amount of energy imported from the external grid, net load, and peak-load reduction indicators. From this perspective, the study is closely related to the digitalization of the electric power sector, the efficient integration of renewable energy sources, and the management of power-system operating modes. In current scientific research, load forecasting, solar generation forecasting, detection of extreme values, and implementation within an information system are often considered separately. Although a number of studies analyse model accuracy, they do not always fully disclose how the obtained forecasts are applied to practical energy-balance and operational optimization tasks. This dissertation is aimed at reducing this gap: forecasting, detection of extremes, calculation of net load, and visualization through an information system are integrated into a single research logic. The aim of the dissertation research. The aim of the dissertation is to develop an information system that, using modern digital technologies, data analysis methods, and machine-learning models, integrates solar generation and electric-load forecasting, net-load calculation, energy-balance analysis, and support for operational decision-making in electric power systems. To achieve this aim, the following tasks were defined: 1. To analyse contemporary scientific research in the fields of electric-load forecasting, solar-energy generation modelling, and optimization of the generation–consumption balance in electric power systems. 2. To develop a hybrid deep-learning model based on Transformer, CNN, and RNN architectures for forecasting electricity generation by solar panels. 3. To evaluate the effectiveness of the developed model on the basis of experimental analysis and comparative study with existing machine-learning and deep-learning methods. 4. To develop an electric-load forecasting module based on open data and integrate the load-forecasting results with the results of solar-power generation forecasting. 5. To determine energy deficit, surplus energy, and the amount of energy imported from the external electric grid by calculating the difference between forecast electricity consumption and forecast solar generation. 6. To develop a computational module aimed at improving the operating mode of an electric power system. The module is intended to address the task of reducing peak-load values through the efficient use of solar energy. 7. To develop an information system that supports forecasting, analysis, and decision-making for the optimization of operating modes in an electric power system. The main results submitted for defense are as follows: 1. A comprehensive processing method for the joint calculation of electric load and solar generation, taking into account the accumulation of surplus energy under clear-sky conditions and the subsequent dependence on the external grid during evening hours. 2. Hybrid models for the preliminary estimation of solar generation, designed to reduce forecasting error under abrupt weather changes and to enable earlier identification of abnormal operating modes. 3. An information platform that jointly analyses the energy produced by solar power plants and the consumption load within a unified environment, enabling preliminary assessment of energy deficit and surplus power. The scientific novelty of the research is as follows: 1. A multilevel predictive-analytical approach is proposed that links solar generation and electric load within a single information system. 2. A method is implemented for calculating the net-load indicator and using it in optimization by integrating the forecast of solar-panel energy production with the forecast of electric consumption load. 3. A software solution is developed that makes it possible to analyse, within a single interface, the indicators of energy deficit, surplus energy, energy imported from the grid, and peak-load reduction in an electric power system. 4. The developed system is based on a modular architecture that is not limited to a single object and can be adapted to various electric grids and organizations when new smart-meter data, PV installation parameters, and local meteorological data are supplied. The object of the research is the processes of electricity generation by a solar power plant and changes in electric load, as well as temporal, meteorological, and technical data that characterize their influence on the operating mode of an electric power system. The subject of the research is machine-learning, deep-learning, hybrid-modelling, and computational methods for forecasting solar-energy production and electric load, calculating net load, detecting extreme operating modes, and optimizing the energy balance. Research methodology. The study applies methods of time-series analysis, machine learning, deep learning, multitask learning, hybrid neural architectures, regression forecasting, classification, cross-validation, ablation study, and mathematical modelling based on the energy balance. The author’s HST-MB-CREH model was used for solar generation forecasting. In estimating electric load, a load profile reconstructed on the basis of smart-meter data was applied. Research methods. Meteorological, temporal, solar-astronomical, and cyclic features were used as input data. The stages of initial data cleaning, missing-value processing, logarithmic transformation, Min–Max normalization, cyclic time encoding, and time-window formation were carried out. Model effectiveness was evaluated using MAE, RMSE, MAPE, R², EVS, and AUC_ext. To preserve the nature of time series, the TimeSeriesSplit strategy was applied. Software implementation. Python, machine-learning and deep-learning libraries, Flutter Web, Firebase Authentication, Firestore Database, and Google Cloud Run technologies were used to implement the research results. The SUN ENERGY platform integrates forecasting models, the load module, energy-balance calculations, and optimization tasks into a single web interface. The theoretical significance of the research is characterized by the development of methods for solar generation forecasting, detection of extreme operating modes, and analysis of the energy balance associated with the load profile. In the dissertation, solar generation is considered not only as a time series, but also as a multifactorial process linked to meteorological, solar-astronomical, and cyclic factors. The proposed HST-MB-CREH model provides a basis for analysing local patterns, temporal dependencies, long-range relationships, and extreme operating modes within a single computational scheme. The practical significance of the research is determined by the development of the SUN ENERGY information system. The system is designed to present forecasting results in an understandable form to operators of solar power plants, energy organizations, and dispatching services. The user can see in advance when solar generation will decrease, during which period surplus energy will arise, at what intervals energy will need to be imported from the external grid, and when peak load will intensify. This supports the justification of dispatching decisions and contributes to the efficient integration of renewable energy sources into the grid. The doctoral candidate’s personal contribution. The research presented in the dissertation was carried out independently by the applicant. The author directly participated in the stages of data collection, preprocessing, formation of the multifactorial feature space, development of the hybrid model, computational experiments, model comparison, ablation study, and design of the information system. In co-authored publications, the author’s contribution consists of developing models and algorithms, conducting experiments, analysing results, and preparing scientific articles. Approbation of the dissertation results. The main results of the dissertation were discussed at an extended meeting of the Department of Information Systems of L. N. Gumilyov Eurasian National University. Implementation of the results. The results of the dissertation research were implemented in the activities of UMAY R&D LLP. The study was carried out within the framework of the grant-funded project No. AP26105051 of the Ministry of Science and Higher Education of the Republic of Kazakhstan, entitled "Optimization of the Reliability of Distribution Networks through the Implementation of an Integrated Intelligent Control System". Publication of the main dissertation results. Five scientific articles have been published on the research topic. These include two articles in journals indexed in Scopus and Web of Science and three articles in scientific journals recommended by the Committee for Quality Assurance in Science and Higher Education of the Ministry of Science and Higher Education of the Republic of Kazakhstan. In addition, based on the results of the dissertation work, a certificate of entry of information into the State Register of Rights to Copyright-Protected Objects was obtained. Scope and structure of the dissertation. The dissertation is written in Kazakh and consists of an introduction, three interrelated chapters divided into subsections, a conclusion, and a list of references. The dissertation comprises 128 pages, includes 31 figures and 15 tables, and the list of references contains 70 sources. The introduction presents the justification of the research topic, the aim and objectives, the object and subject of the research, the scientific novelty, theoretical and practical significance, the main results submitted for defense, and the research methods. The first chapter analyses methods for load and solar generation forecasting in electric power systems and for optimizing power-system operating modes. It considers global and Kazakhstani trends in the development of renewable energy sources, the main factors affecting the formation of load and solar generation, and traditional and modern forecasting methods. Based on the literature analysis, the need to consider load, solar generation, extreme operating modes, and decision support at the information-system level within a unified research logic is substantiated. The second chapter develops a model for forecasting solar electricity generation and detecting extrema. It describes data preparation, the formation of a multifactorial space consisting of 21 features, the construction of a hybrid model based on CNN, LSTM, GRU, and Transformer, the detection of extreme values, and the mathematical model of the load forecasting and optimization module. The HST-MB-CREH model was compared with Random Forest, XGBoost, LSTM, GRU, and Transformer models, and its effectiveness was experimentally confirmed. The third chapter develops the SUN ENERGY information system for solar electricity forecasting and optimization of power-system operating modes based on the proposed hybrid model. It presents the conceptual model, architecture, functional structure, business processes, data flow, integration with an external meteorological service, visualization, and testing results of the system. The system was formed as a platform that presents solar generation forecasting, load analysis, detection of extreme operating modes, energy-balance calculation, and optimization decisions in a single digital environment. The conclusion presents the main scientific conclusions of the dissertation research, the theoretical and practical significance of the obtained results, and directions for further development of the research.

Dissertation

List of Scientific Works

Consultant's Review

Foreign Consultant's Review

Conclusion of the Research Ethics Committee