The problem of assessing the state of complex objects in monitoring systems, where the process of analysing the characteristics of the object to determine its current state, detect anomalies and predict possible changes is implemented, is discussed. The main attention is paid to the classification task, when states are considered as images of some classes to which it is necessary to refer the evaluated object. A popular solution for working with input data in monitoring systems is the use of neural network methods. However, their application is limited in the case of small training samples, when it is difficult or impossible to collect data for model training. In this case, instead of artificial neural networks it is proposed to use Haken’s synergetic model designed for pattern recognition. The possibility of its application for estimating the state of objects with small training samples is investigated. Based on the development of Haken’s model, a generalised algorithm of state classification is developed, the main provisions of the method and its structural representation in the form of a three-level sequence of the object state estimation process are considered. The results of the study are presented.