The paper presents a combined approach to forecasting of the manufacturing system target indicators, depending on the key factors determining the functioning of this system and business environmental factors. The approach includes: (i) identification of key factors and weak signals influencing the target indicators by analyzing the system functioning model in the external environment represented by a cognitive map of the situation; (ii) analysis of dynamics of the identified key factors and manufacturing system target indicators and the construction of a forecast model using time series analysis methods; (iii) monitoring of the heterogeneous information space in order to detect changes in trends and the composition of parameters of the forecast model; (iv) correction of the forecast model and cognitive map of the situation according to the results of the detected changes. The approach is focused on the formation of long-term forecasts of the target indicators and increasing the accuracy of these forecasts by identifying significant influencing events whose influence does not have time to reflect in time series.