Radar software reliability test case intelligent generation based on domain knowledges

Due to a lack of expertise in the radar field and intelligent technology, testers typically rely on their personal experience and knowledge to generate reliable test cases. This approach is highly subjective and lacks objectivity, leading to a substantial workload and low efficiency in software reliability testing. Additionally, the absence of a test profile construction method oriented towards task reliability makes it difficult to fully identify weak links that affect the completion of system tasks. To address this problem, this paper proposes a new approach for radar software reliability test profile construction using intelligent test-case generation technology. First, a test profile construction technology oriented towards task reliability is introduced. Next, a radar domain knowledge dictionary is constructed to facilitate text segmentation of test data based on domain knowledge. Then, leveraging the LDA (Latent Dirichlet Allocation) topic model, a deep learning algorithm, and cosine distance, the intelligent reuse and generation of radar software reliability test cases based on semantic similarity is achieved. Finally, case studies of radar data processing software and radar control software are conducted. The results demonstrate that the proposed approach outperforms three classic intelligent algorithms in terms of the reliability test case reuse accuracy rate (P) and recall rate (R), with the accuracy rate (P) increased by at least 10.04% and the recall rate (R) increased by at least 14.94%.