Dynamic Optimization of Automated Test Cases for Industrial HMI Based on a Time-Weighted Bayesian Model

In traditional industrial human-machine interface (HMI) testing, test case selection is predominantly driven by engineer’s subjective experience, leading to static, non-adaptive test strategies that lack quantitative decision support. This often results in inefficient resource allocation, redundant testing of low-risk scenarios, and potential omission of critical functional paths. To address these limitations, this paper proposes a dynamic optimization algorithm for industrial automated testing based on a time-weighted Bayesian model (TWBM). The TWBM framework incorporates a time decay factor into historical test outcomes to construct weighted observational data, enabling the model to dynamically compute the posterior probability of each test case detecting defects. By replacing manual, binary decisions with data-driven probabilistic evaluation, the proposed method intelligently selects high-value test cases for execution. A predefined probability threshold is used to determine inclusion in the final test suite, ensuring adaptability to evolving system states and update patterns. The optimized test set is then executed within existing automation frameworks, and test outcome data are updated for future iterations. Experimental evaluation on over 6000 HMI systems demonstrates that the TWBM-based method significantly improves testing efficiency by reducing execution time, while achieving higher fault detection effectiveness compared to conventional methods. The results validate that the proposed algorithm enables scientific, sustainable, and dynamic optimization of test case selection in industrial HMI testing.