Research on Software Reliability Prediction Methods for Intelligent Measurement Equipment
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Aiming at the reliability prediction challenges of intelligent measurement equipment software in dynamic update environments, this study proposes a prediction method integrating multi-model entropy weight optimization and improved particle filtering. By constructing an evaluation framework encompassing five categories (12 models including NHPP and Markov classes), the entropy weight method screens the top-five model combination, reducing prediction errors by 36.0% (RMSE) and 47.5% (MAPE) under dynamic conditions compared to single models. A local Parzen window state space decomposition and adaptive resampling strategy are designed, reducing RMSE to 9.3 hours and MAPE to 6.2%, with a single prediction time of 28ms. A full-life-cycle multi-metric monitoring framework is proposed, improving detection sensitivity by 7.69 percentage points and reducing false alarm rates by 15.53 percentage points in dynamic adaptability scenarios with a 200% fault rate surge. Validated on 220 fault data from smart meter software frequent updates, the method achieves 92.4% prediction accuracy. The achievements have been applied to the management platform of S Electric Power Company, supporting the establishment of a unified testing standard system.
Fields of Study & Subjects
Software Reliability and Analysis Research
Reliability and Maintenance Optimization
Power System Reliability and Maintenance
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 362–365
- Publication date
- 28 January 2026
- Language
- EN