Finite test sets development method for test execution of safety critical software

The V and V method has been utilized for this safety critical software, while SRGM has difficulties because of lack of failure occurrence data on developing phase. For the safety critical software, however, failure data cannot be gathered after installation in real plant when we consider the severe consequence. Therefore, to complement the V and V method, the test-based method need to be developed. Some studies on test-based reliability quantification method for safety critical software have been conducted in nuclear field. These studies provide useful guidance on generating test sets. An important concept of the guidance is that the test sets represent 'trajectories' (a series of successive values for the input variables of a program that occur during the operation of the software over time) in the space of inputs to the software.. Actually, the inputs to the software depends on the state of plant at that time, and these inputs form a new internal state of the software by changing values of some variables. In other words, internal state of the software at specific timing depends on the history of past inputs. Here the internal state of the software which can be changed by past inputs is named as Context of Software (CoS). In a certain CoS, a software failure occurs when a fault is triggered by some inputs. To cover the failure occurrence mechanism of a software, preceding researches insist that the inputs should be a trajectory form. However, in this approach, there are two critical problems. One is the length of the trajectory input. Input trajectory should long enough to cover failure mechanism, but the enough length is not clear. What is worse, to cover some accident scenario, one set of input should represent dozen hours of successive values. The other problem is number of tests needed. To satisfy a target reliability with reasonable confidence level, very large number of test sets are required. Development of this number of test sets is a herculean task. Therefore, another approach to cover software context without trajectory form of input is needed. This study proposed a test set generation method for PLD based safety critical software without trajectory form of input sets. This method also considered the test coverage which was hard to deal with trajectory form based approach. To substitute trajectory form of input sets, possible ranges of variables for each situation should be identified. For this purpose, assigned range of each variable, logical relations between variables, plant dynamics under certain situation, and characteristics of obtaining information of digital device are considered