The approximation of an isolated epidemic process wave using a combination of exponents

The most commonly used methods for the medium- and long-term forecasting of epidemic processes are based on the classical SIR (susceptible – infected – recovered) model and its numerous modifications. In this approach, the dynamics of the epidemic is approximated using the solutions of differential or discrete equations. The forecasting methods based on the approximation of data by functions of a given class are usually focused on obtaining a short-term forecast. They are not used for the long-term forecasts of epidemic processes due to their insufficient efficiency for forecasting nonstationary processes. In this paper, we formulated a hypothesis that the primary waves of the COVID-19 pandemic, which took place in a number of European countries, including the Republic of Belarus, in the spring-summer of 2020 are isolated and therefore can be regarded as processes close to stationary. On the basis of this hypothesis, a method of approximating isolated epidemic process waves by means of generalized logistic functions with an increased number of exponents was proposed. The developed approach was applied to predict the number of infected people in the Republic of Belarus for the period until August 2020 based on data from the beginning of the epidemic until June 12, 2020.

Hybrid method of intellectual diagnosis and forecasting of complex technical systems

Modern control and diagnostic systems (CDS) usually determine only the technical condition (TC) at the current time, ie the CDS answers the question: a complex technical system (CTS) should be considered operational or not, and may provide little information on performance CTS even in the near future. Therefore, the existing scenarios of CDS operation do not provide for the assessment of the possibility of gradual failures, ie there is no forecasting of the technical condition. The processes of parameter degradation and degradation prediction are stochastic processes, the “behavior” of which is influenced by a combination of external and internal factors, so the deg-radation process can be described as a function that depends on changes in the internal parameters of CTS. The hybrid method involves the following steps. The first is to determine the set of initial characteristics that characterize the CTS vehicle. The second is the establishment of precautionary tolerances of degradation values of the characteristics that characterize the pre-failure technical con-dition of the CTS. The third is to determine the rational composition of informative indicators, which maximally determine the "behavior" of the initial characteristics. The fourth — implementa-tion of multiparameter monitoring, fixation of values of the controlled characteristics, formation of an information array of values of characteristics. Fifth — the adoption of a general model of the process of changing the characteristics of the CTS. Sixth — the formation of a real model of the process of changing the characteristics of Y(t) on the basis of an information array of values of char-acteristics obtained by multi-parameter monitoring. Seventh — forecasting the time of possible oc-currence of the pre-failure state of the CTS, which is carried out by extrapolating the obtained real model of the process of changing the characteristics of Y(t). It is proposed to use two types of mod-els: for medium- and long-term forecasting - polynomial models, for short-term forecasting — a lin-ear extrapolation model. At the final stage, forecast errors are determined for all types of models of degradation of pa-rameters and characteristics. Based on the results of the forecast verification, the models are adjust-ed

Long-Term Forecasting and Optimization of Non-Stationary Well Operation Modes Through Neural Networks Simulation

The article discusses the results of long-term forecasting of non-stationary technological modes of production wells using neural network modeling methods. The main difficulty in predicting unsteady modes is to reproduce the response of producing wells to a sharp change in the mode of one of the wells. Such jumps, as a rule, lead to a rapid increase in the forecast error. Training and forecasting of modes was carried out on the data of numerical hydrodynamic modeling. Two fields with significantly different properties, the number of wells and their modes of operation were selected as objects of modeling. Non-stationarity was set by changing the regime on one or several production wells at different points in time. The LSTM recurrent neural network carried out forecasting of production technological parameters. This made it possible to take into account the time-lagging influence of the wells on each other. It is shown that the LSTM neural network allows predicting unsteady technological modes of well operation with an accuracy of up to 5% for a period of 10 years. The solution of the problem of optimization of oil production is considered on the example of one of the models. It is shown that the optimal solution found by the neural network differs from the solution found by hydrodynamic modeling by 5%. At the same time, a significant gain in calculation time was achieved.

Probabilistic Gridded Seasonal Sea Ice Presence Forecasting using Sequence to Sequence Learning

Accurate and timely forecasts of sea ice conditions are crucial for safe shipping operations in the Canadian Arctic and other ice-infested waters. Given the recent observations on the declining trend of Arctic sea ice extent over the past decades due to global warming, machine learning (ML) approaches are deployed to provide accurate short-term to long-term forecasting. This study unlike previous ML approaches in the sea-ice forecasting domain provides a daily spatial map of the probability of ice in the study domain up to 90 days of lead time. The predictions are further used to predict freeze-up/breakup dates and show their capability to capture these events within a valid time period (7 days) at specific locations of interest to communities.

Studying changes in climate normals in long-term forecasting of the Sea of Okhotsk ice cover.

There is a similarity in dynamics and a quantitative difference in the ice cover regime in four consecutive 30-year periods: 1961–1990, 1971–2000, 1981–2010, 1991–2020 are noted. The greatest differences are observed in the regime characteristics of the first and the last periods. The absolute maximum or minimum recorded in at least one of the months from January to May determines the nature of the ice cover of the entire ice season. The sensitivity of the predictive physical-statistical model to the replacement of climatic norms has been investigated. Estimates of the quality of forecasts of the average monthly ice cover are obtained. Keywords: base period, long-term forecast, physical and statistical model, ice cover, climate characteristics, typification, forecast skill scores

Learning Latent Representation of Freeway Traffic Situations from Occupancy Grid Pictures Using Variational Autoencoder

Several problems can be encountered in the design of autonomous vehicles. Their software is organized into three main layers: perception, planning, and actuation. The planning layer deals with the sort and long-term situation prediction, which are crucial for intelligent vehicles. Whatever method is used to make forecasts, vehicles’ dynamic environment must be processed for accurate long-term forecasting. In the present article, a method is proposed to preprocess the dynamic environment in a freeway traffic situation. The method uses the structured data of surrounding vehicles and transforms it to an occupancy grid which a Convolutional Variational Autoencoder (CVAE) processes. The grids (2048 pixels) are compressed to a 64-dimensional latent vector by the encoder and reconstructed by the decoder. The output pixel intensities are interpreted as probabilities of the corresponding field is occupied by a vehicle. This method’s benefit is to preprocess the structured data of the dynamic environment and represent it in a lower-dimensional vector that can be used in any further tasks built on it. This representation is not handmade or heuristic but extracted from the database patterns in an unsupervised way.