Test Automation for Industrial Programmable Logic Controller Software: From Requirements to Executable Tests

Programmable Logic Controllers (PLCs), which are specialized computers used to control machines and processes in industrial settings, play a vital role in the automation industry. Testing these systems to identify faults during the development phase, especially for safety-critical applications such as crane control in ports or train control systems, is crucial to ensuring the system's safety in operation. Current PLC testing practices are mostly manual, and it has been a challenge for both practitioners and researchers to develop robust, applicable test automation techniques and tools that are relevant in practice. This thesis proposes test automation solutions for PLC programs, which can improve current testing practices by augmenting manual testing. We first propose a method for selecting an appropriate test automation framework for PLCs, derived from a grey literature review and a set of empirically validated evaluation criteria. To achieve practical integration of test automation solutions, we design a method and tool, PyLC+, for automated program-to-program translation from PLC to Python. Our tool automatically translates a PLC program into executable Python code, and then verifies the latter against given assertions, using an automated Python program verifier called Nagini. PyLC+ then uses metaheuristic algorithms for automated test generation. In addition, we introduce a domain-specific LLM, PyLC-M1, which serves as an interactive test-automation assistant within PyLC+. It assists the generation of verification harnesses for formal PLC analysis while operating entirely in secure, offline environments. Investigating the applicability of a well-known syntax notation pattern for semi-formal requirements specification, called Easy Approach to Requirements Syntax, for specification-based testing through a controlled experiment is another contribution of this thesis. The applicability and efficiency of the proposed testing and verification solutions in this thesis have been evaluated on real-world industrial use cases.