MODELLING THE QUALITY ASSURANCE OF AI-BASED INTELLIGENT ENERGY MANAGEMENT SOFTWARE

DOI: 10.31673/2409-7292.2025.030192

Abstract

This research is conducted within the Department of Software Engineering for Power Industry, NTUU KPI and
Foreign Expert Studio for Demand Response at the Shandong-Uzbekistan Technological Innovation Research Institute
collaboration under the Project H20240943 Quality Assurance Project for Intelligent Energy Management Software
Based on AI Methods and the Development and Industrialization of Intelligent Grid Demand Response Technology
Project. Intelligent Energy Management Software (IEMS) must operate reliably across heterogeneous sites where data
distributions, sensor suites, code bases, and operating policies evolve over time. This paper presents a unified framework
for cross-domain adaptation and trusted quality assurance (QA) that combines supervised transfer learning, domainadversarial alignment, and federated aggregation with release gates for calibration, robustness, and explainability. The
framework is validated on benchmarks spanning software engineering and energy analytics: NASA MDP and PROMISE
defect datasets for classification, the Numenta Anomaly Benchmark (NAB) for time-series anomaly detection, and the
UCI energy dataset for reliability assessment. Strong baselines (Random Forest, SVM, CNN, GRU) are tuned under
identical protocols to ensure fair comparison. The proposed method consistently improves predictive performance,
yielding absolute F1-score gains of 5–10 points on defect prediction and an 8-point increase on NAB anomaly detection
(from 0.70 to 0.78). Trustworthiness also increases: the Expected Calibration Error (ECE) is reduced to 0.032 (a 22–42%
reduction relative to Bayesian/CNN baselines), the Negative Log-Likelihood (NLL) falls to 0.18, and the Brier score
improves, indicating better probabilistic accuracy. Ablation studies show that adversarial alignment drives the most crossdomain generalization gains, whereas temperature scaling and entropy regularization deliver the largest calibration
improvements. Stress tests with injected noise and gradual drift confirm stable precision–recall trade-offs and bounded
error propagation under distributional shift. In privacy-constrained settings, federated aggregation maintains these
benefits without exchanging raw data, while lightweight explainability checks (e.g., SHAP/LIME) flag low-confidence
predictions for human review, enabling actionable QA. Together, these results demonstrate that coupling adaptive transfer
with formal QA checks provides a principled and practical route to reliable IEMS deployment across residential,
commercial, and industrial environments.
Keywords: Intelligent Energy Management Software (IEMS); Cross-Domain Adaptation; Transfer Learning;
Domain-Adversarial Training; Federated Learning; Software Quality Assurance; Calibration; Explainability.

References
1. ISO/IEC 25010:2011. Systems and software engineering—Systems and software Quality Requirements and
Evaluation (SQuaRE), System and software quality models. International Organization for Standardization, Geneva,
2011.
2. Felderer, M., & Ramler, R. (2021). Quality assurance for AI-based systems: Overview and challenges. arXiv
preprint arXiv:2102.05351. https://doi.org/10.48550/arXiv.2102.05351.
3. Ali, M. A., Yap, N. K., Ghani, A. A. A., Zulzalil, H., Admodisastro, N. I., & Najafabadi, A. A. (2022). A
systematic mapping of quality models for AI systems, software and components. Applied Sciences, 12(17), 8700.
https://doi.org/10.3390/app12178700.
4. Lakshminarayanan, B., Pritzel, A., & Blundell, C. (2017). Simple and scalable predictive uncertainty
estimation using deep ensembles. In Advances in Neural Information Processing Systems (NeurIPS 30) (pp. 6402–6413).
Curran Associates, Inc. https://doi.org/10.48550/arXiv.1612.01474
5. Angelopoulos, A. N., & Bates, S. (2021). A gentle introduction to conformal prediction and distribution-free
uncertainty quantification. arXiv preprint arXiv:2107.07511. https://doi.org/10.48550/arXiv.2107.07511
6. Higham, N. J. (2002). Accuracy and Stability of Numerical Algorithms (2nd ed.). Philadelphia, PA: SIAM.
https://doi.org/10.1137/1.9780898718027
7. Goodfellow, I., Shlens, J., & Szegedy, C. (2015). Explaining and harnessing adversarial examples. In
Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015). San Diego, CA. Retrieved
from https://arxiv.org/abs/1412.6572
8. Carlini, N., & Wagner, D. (2017). Towards evaluating the robustness of neural networks. In 2017 IEEE
Symposium on Security and Privacy (SP) (pp. 39–57). IEEE. https://doi.org/10.1109/SP.2017.49
9. Hendrycks, D., & Dietterich, T. G. (2019). Benchmarking neural network robustness to common corruptions
and perturbations. In Proceedings of the 7th International Conference on Learning Representations (ICLR 2019). New
Orleans, LA. Retrieved from https://arxiv.org/abs/1903.12261
10. Mounir, N., Ouadi, H., & Jrhilifa, I. (2023). Short-term electric load forecasting using an EMD–BiLSTM
approach for smart grid energy management system. Energy and Buildings, 288, 113022. https://doi.org/10.1016
/j.enbuild.2023.113022
11. Abumohsen, M., AlQahtani, M., Alsanad, A., Alqahtani, A., & Alkahtani, H. (2023). Electrical load forecasting
using LSTM, GRU, and RNN. Energies, 16(5), 2283. https://doi.org/10.3390/en16052283
12. Bayram, F., Aupke, P., Ahmed, B. S., Kassler, A., Theocharis, A., & Forsman, J. (2023). DA-LSTM: A
dynamic drift-adaptive learning framework for interval load forecasting with LSTM networks. Engineering Applications
of Artificial Intelligence, 123, 106480. https://doi.org/10.1016/j.engappai.2023.106480
13. Mischos, S., Iakovidis, D. K., & Katsikas, A. K. (2023). Intelligent energy management systems: A review.
Artificial Intelligence Review. https://doi.org/10.1007/s10462-023-10506-5