A METHOD OF PREDICTING FAILURE RATES OF AN UNMANNED AIRCRAFT COMPLEX BASED ON CONTROLLED OPERATION DATA

Abstract

The article provides a description of the method for determining the fault-free work indicators (FWI) of unmanned aerial systems (UAS) during their controlled exploitation (CE). It is noted that’s standard approaches to determining the reliability indicators operate with idealized models of situations in which the number of failures is calculated by various ways, and then statistical models are built on this basis. Further, taking in to account various assumptions of the author, reliability indicators are formulated, and, as their subsets, FWI as well.
 Thus, there is a situation of a contradiction between the “noisy” experimental model data and the ultimate UAS FWI (number of failures, MTBF, failure rate and probability of failure-free operation), which are formulated by analogy with their idealized prototypes.
 It is proposed to overcome this contradiction by developing a method that would calculate the number of UAS failures N* during the standard СE period (conventionally 100 flights), the average integral MTBF T* in the CE period of UAS, the average integrated failure rate λ* in the CE period of UAS, and the average integral probability P* in the CE period of UAS.
 The determination of the devalues is proposed to be carried out on the basis of data obtained during the controlled UAS exploitation, according to the method described in this article here by.
 Monitoring the progress of the controlled UAS exploitation is carried out by documenting the facts concerning failures and updates using the Dynamic Bayesian Trust Network (DBN). By the CE beginning, DBN is created using the Bayes Fusion GeNiE Academic 2.5 simulation environment.
 The tracking process is implemented by making to the DBN records (evidences) about the failure and restoration of the complex in the CE process. These significations are inputting into the model the information about whether at a certain point of time a certain UAS element failed or it was restored after the failure. This information is the result of the UAS observation, and in DBN this causes an immediate recalculation of conditional posterior probabilities of failure-free operation of all whose operability depends on the operability of the nodes got evidenced. This information appears in the actions flow of the practical use of the BPAC and is transmitted to the flow of documenting failures that the operator performs in the DBN. These actions enable the external, relatively DBN, software to calculate the ultimate UAS FWI.