Test Case Generation Method Based on Industrial Software Workflow State Machine and Large Language Model

A component-based industrial software system is a novel type of industrial software that supports deep integration and dynamic interaction between physical and informational processes. Due to its inherent complexity and cross-domain integration, testing such systems has become a significant challenge. To address the limitations of traditional manual and partially automated test case generation methods—such as insufficient coverage and low efficiency—this study proposes an automated test case generation method based on industrial software component workflow state machines and large language models (LLMs). Starting from system user input, the method integrates an industrial software knowledge base and employs LLMs to assist in generating workflow state machine models. Under the constraints of these state machine models, preliminary test case data are generated, followed by path verification and filtering to ensure the accuracy and reliability of the testing process. Experimental results demonstrate that this method can generate precise test cases across diverse industrial software scenarios, achieving state coverage exceeding 95% and reducing generation time compared to traditional methods, thereby improving testing efficiency.