The Research On Dynamic Reliability Allocation And Assessment Of Complex Software System

Under the rapid development of current information society,many software with different functions have emerged.These software not only covers all aspects of people's daily life,but also always promotes the rapid development of the entire society.However,various software brings many conveniences and also causes many large or small hidden dangers,such as system failure and equipment loss,which may cause serious economic losses and even endanger personal safety.Most of the hidden dangers come from unreliable software.Software reliability research has gradually attracted people's attention,among which software reliability allocation and assessment are important parts of software reliability engineering.Reliability allocation can reduce software development costs and provide developers with an improved design basis under the premise of ensuring software reliability goals.Reliability assessment builds reliability assessment models based on various basic assumptions,and provides a reliable basis and guarantee for correct assessment of software reliability.Currently,many scholars at home and abroad have conducted reliability research on complex software systems and achieved certain results,but there are still some problems to be solved,such as the reliability allocation model cannot effectively adapt to the changing system structure and the allocation algorithm is inefficient.Moreover,the single reliability assessment model cannot adapt to multiple situations,and the combined model has problems such as simple selection principle.fixed combination method,and poor prediction performance.Therefore,this paper studies the reliability allocation and assessment methods of complex software systems.The specific contents include:(1)Most of existing complex software system reliability allocation models are based on a certain system structure,which cannot effectively adapt to the uncertainty and incompleteness of complex software modules.Combining the characteristics of hierarchical structure of complex software system,a dynamic reliability allocation model of complex software systems based on D-S evidence theory is proposed and solved by an hybrid optimization algorithm based on particle swarm optimization algorithm and self-adaptive differential evolution algorithm.In this way,the reliability allocation results of complex software systems can be obtained quickly and accurately in the uncertain reliability allocation model to guide software design and improvement.(2)For a single reliability assessment model that is difficult to adapt to complex and changeable situations,combined model prediction has the problems of simple selection principles,fixed model combination cannot adapt to multiple situations,etc.A software reliability assessment method based on Multi-model combination is proposed.On the one hand,the fuzzy comprehensive evaluation method is used to select the optimal model;on the other hand,the chaotic particle swarm optimization algorithm is used to adaptively combine the optimal models to complement the advantages of different types of models.Experimental results show that the reliability assessment method has good assessment accuracy and evaluation efficiency.(3)Based on the above relevant theory and method research,a software reliability allocation and assessment platform for complex software systems was designed and implemented.Taking a UAV control station system as an example,the practicality of the platform was verified,which further showed has a good practical application value.