Evolutionary approach for achieving structural coverage in testing IEC 61499 function block systems

Abstract

The topic of this thesis is automated test generation for control software represented in a specific standard, the IEC 61499. This standard, which is largely based on the concept of function block, establishes a way to design distributed control systems in a visually clear way. The goal of the thesis was to design a test generation approach or a number of such approaches that would produce input test data with high coverage of the implementation of systems under test. Coverage is a measure which expresses the fraction of the system that was exercised at least ones when all tests in a test suite were run on this system. To reach the stated goal, evolutionary computation, a general optimization methodology, was employed. In this methodology, possible solutions of the problem (in our case, test suites) are developed during a simulated evolution process which involves mutating solutions (that is, altering them insignificantly) and combining them into new ones.

Two methods of test suite generation were designed based on the mentioned approaches. The experimental evaluation showed that one of them produces test suites with high coverage but is time consuming, and another one is more flexible and fast, but produces test suites with lower coverage. It was also shown that the proposed methods are capable of identifying faults in control software under test, which are mainly connected with unreachable system segments.