Tiutiunnyk M. Parallel algorithms for complex evaluation of the state and operation quality of complex systems functioning

The necessity of using parallel computations is substantiated for the realization of the technique of complex evaluation of the state and operation quality of complex systems. Parallel-pipeline algorithm of solving the problem of digital filtering is developed for adaptive smoothing of large arrays of input data in real time. The optimality of this al-gorithm for speed and memory use is proved in the class of algorithms that are equi¬valent to the information graph. The hardware architecture for the implementation of developed digital filtering algorithm is proposed. A parallel-sequential approach is proposed for realization of the procedure of local evaluation of the state and operation quality of complex systems. Algorithmic construc-tions are developed and analyzed for separate computational fragments. These construc-tions are oriented to the implementation by parallel computing devices with common me¬mory. Effective algorithmic constructions are proposed for paralell computation of aggre¬gated evalutions. For the first time, a general approach to optimizing the methodology for complex evaluation of complex systems was developed. The proposed approach is based on large-scale block parallelized computations. The software modeling for the parallel-pipeline algorithm of digital filtering was performed. The software implementation of proposed algorithmic computations was carried out on a multi-core computer. The constructed proposed parallel algorithms for the complex assessment of the state and operation quality of functioning of complex systems were used to analyze the current state of railway transport facilities; forecasting, planning and managing traffic flows in the Ukrzaliznytsia and in the educational process.