I.I. Ulshin, R.E. Pervezentsev

In the article improving meteorological support of the Air Force under stage decision of important problems under time and limited spare material resource deficiency conditions is considered. The new order making management decisions to maximize the probability of the problem fulfillment and to minimize the average loss of the consumer of the meteorological information is proposed. Optimization problem is solved for the aircraft organizing-technical system depending on the meteorological conditions. To increase the adequacy of the model not to considered an only one solution but the optimal phase-planning system for a long enough period of time. For this purpose the mathematical tools of the dynamic programming is used. The essence of this approach lies in the fact that at each stage the decision on a start is made on the basis of the ratio of the probability of the problem fulfillment and the threshold probability. The probability of the problem fulfillment is calculated with the help of a special recurrence relation. The threshold probability is found from the maximum likelihood of the problem fulfillment. By use of the recurrence relation is provided consistent optimal taking the management decisions at each stage of solving the problem and provides a framework for solving dynamic problems (Bellman equation). The disadvantage of this approach is the subjective nature of some raw data and, in particular, specialized aviation weather forecasts. One of the ways to address this shortcoming is to develop at each stage of solving the problem is not the weather forecast , but forecast the possibility of the problem fulfillment. It is suggested to use classification trees (decision trees). The use of decision trees can not only exclude the subjective factor, but also automating the creation of the weather forecasts. To ensure the statistical significance of the initial data should be corrected any forward-looking statements or increase the amount of archival material, or reduce the number of considered atmospheric parameters. Can be obtained also an interval estimates using the formulas presented in this paper. The practical test of the proposed approach has shown its advantage over existing practice of aviation meteorological services. The probability of the problem fulfillment at the recommended approach is higher than the traditional, 5-15%, depending on the level of the preparation of flight crews. The effectiveness of the proposed approach is greatest when headroom is limited and equal to the number of remaining steps to complete the task, and therefore the price of each mistake is very high