Determination of Elements Reliability for Power Plants Based on Internal Combustion Engines by Lowest Residual Entropy Method

The selection technique of diagnostic parameters for the creation of fault detection system of autonomous electric power sources based on gasoline and diesel engines is given in the paper. An analysis of the design features for autonomous electric power sources based on internal combustion engines, which are the most common on the Ukrainian market, was carried out. Thanks to this, a logical model of the research object, which establishes the relation between the main structural elements of the system and determines the possible states of the system, was developed. The effect of fault state initiation for each element on the other system elements was analyzed. An informative criterion – Shannon information entropy is proposed to determine the finite number of diagnostic parameters among the infinite number of possible combinations for physical parameters that characterize the system. The equal-probable cases of exit from operational state of each system elements are considered. The residual entropies of the system at the fault state for one of the autonomous power sources assembly are determined, having applied the concept of Shannon information entropy. The residual entropy value is the informative criterion. The application of this criterion allowed to determine the system elements that most effectively reduce the system uncertainty degree. Based on the residual entropy values, the system assemblies, the state of which should be primarily monitored by diagnostic system, are selected. The diagnostic parameters are determined for such elements, and the ways to obtain them are given