Reliability analysis is an essential part of design process of complex-structured systems, especially systems with high demands of reliability. Analysis of system reliability is used iteratively many times in different stages of system design. The analysis results help to define system structure with appropriable reliability characteristics, comparison of various structures of the system, prediction of future system properties. For some classes of systems components with redundancy are used. Reliability analysis allows to determine components with higher influence on overall system reliability and estimate effect of redundancy application. All of the above is true for automotive control systems. These systems become one of the important subsystem in car safety. The control system in a contemporary car contains up to several thousands components. There usually are complex interconnections between components and complex rules of their operability. Manual calculation of reliability characteristics of real-world automotive control systems is too time-consumption. So, it is necessary to use automation of reliability analysis. The result of reliability analysis is reliability characteristics. Calculation of these characteristics are based on system reliability function. System reliability function can be calculated from system operational function or from system fault tree.

There are several methods for Reliability Function construction which are based on operational function or fault tree. All of these methods have in general case exponential complexity of system size. We found out that in practice, using Cutting algorithm with some heuristics has good scalability. So, we use this algorithm in our approach.

The main problem of reliability analysis is high time-consumption of manual Fault tree or Operational function building. Because of this, Fault tree-based methods can not be directly applied for reliability analysis of automotive control systems. There are some methods for automation of fault tree construction. Using of Reliability block diagrams before Fault tree construction reduces model size, but it is efficient only in some simple cases. The other method is Fault tolerant data flow. The main limitation is that this method does not support of cyclic dependencies between system components. We propose a novel approach for automation of operational function construction and the following reliability characteristics calculation.

Our approach is based on a meta-model, that allows to analyze various classes of systems. The meta-model consists of components types, connection line types, failure types and rules for their propagation and high-level description rules of system operability. The meta-model is adjusted for a class of systems with help of patterns. For automotive control systems analysis we have developed a special pattern. In the result we have got a model, which allows to represent systems of the class. So, user makes high-level description of a system in terms of the model. Reliability analysis based on the description is performed fully automatic. For this propose we have developed a tool which automatically constructs operational function from the given description, reliability function and calculate reliability characteristics.

In the pattern for automotive control systems we define four component types: ECU, gateway, battery and fuse. The system topology is defined via two graphs, one for the data paths and one for the power lines. The automotive pattern contains the following failure types. Some failure types can propagate through the system components. The failure propagation rules are predefined in the pattern. We set up conditions of component operable state and general rules of operability function description.

Automotive pattern

System description is created by user, who defines set of system components, choose inner structure and failure rate of each component. Further, user adds power and data lines, and set up lines failure rates. At the end, user describes system operability using the model rules.

We have developed Reliability Analyzer tool which has convenient GUI. Reliability Analyzer allows calculation reliability characteristics.

RA