Research and Development of Synthesis Technologies of Transport Enterprise Multi-control Neural Network Algorithms

Currently, the problem of designing automatic control systems of dynamically variable objects is characterized by the transition from adaptive management paradigm to intelligent control paradigm. This is caused by continuous complication of objects and conditions of their operation, the advent of new classes of computing devices (distributed computing), high-performance telecommunication channels, and a sharp increase in the requirements for reliability and efficiency of control processes in a significant priori and posteriori uncertainty. Accounting for these factors is possible only through the transition from «hard» algorithms of parametric and structural adaptation to the anthropomorphic principle of forming control.Given the characteristics of the modern enterprise, when the head and structural units quickly make decisions and monitor its implementation, it comes very clearly understand the need of artificial intelligence as an assistant in the work of transport enterprise. However, existing methods are outdated and not fully perform the role of assistant. The latest trends in this matter are modern methods of creating intelligent systems that can learn in the process, based on neural networks.The paper proposed synthesis technologies of transport enterprise neural network algorithms. Better use of major resources of the enterprise is possible through the use of self-learning neural networks to control transport enterprise. Using a synthesis of known algorithms may be more correct setup of the whole system and increase the speed of information processing and decision of optimal solution