Context. The article focuses on the question of automated decision-making analysis made by the operator in ergatic systems of critical infrastructures on the example of marine transport control in difficult navigation conditions. It is evident enough that the main criterion for an adequate perception of input information done by an operator is highly likely to predict the choice of behavioral decision-making strategies in discrete time conditions. However, the difficulty of modeling the operator’s actions is found to be lying in non-linear pattern of taking definite decisions in emergency situations and deviations from the Codes and Rules.Objective. The research purpose strategy of conducted investigation can be defined as the development of the mathematical platform for a decision support system (DSS) module with an aim to identify the class-forming set of atomic elements. In particular this issue determines the fact of distortion of the perception of information about navigation risks predicting the operator’s behavior pattern while having vessel running process. This is possible to have it depicted through formal analysis.Method. To capture the analysis of danger perception by the operator the paper introduces a mathematical model of data collection which identifies the fact of perception distortion in the form of attribute space of metadata obtained by the method of converting information from navigation devices. Besides, the factor of disorientation of the operator can be considered to be a shift on a displaced bridge which significantly affects on the analysis of information for adequate decision making. In addition, taking into account the failure of navigation equipment such as: RADAR, ARPA, AIS, ECDIS, especially while doing exit from the automatic control mode, a dangerous precedent can possibly be created for the operator not ready to perceive the complexity of the situation. To make it work a formal analysis was carried out using the extending risks possibility level tasks during the transition under these conditions. In addition to this item, a probabilistic model of perceiving the situation under the conditions of the error set is reported to have been constructed. So, as the result, the modeling process turned out to show the definite evidence of getting no way possibility to have the degree of criticality of the navigation situation determined without a clear identification of factors affecting the distortion of perception of the operator. Nevertheless, generalized statistical data are sure to be not enough and there is a special need of taking into account an individual information model of each operator for the effective work of DSS as this process faces real challenges. It must be significantly noticed that in order to analyze the perception of information by the operator a special test for defining preferences when choosing a strategy of control actions in the form of maneuvering under difficult navigation conditions purpose was created. Regarding the test results, as well as data on the passage of locations, certain attention is advised to be drawn to the classification analysis of 15 parameters using artificial neural networks having been carried out by our team and, as a consequence, the boundaries of deviations in the perception of navigational danger were found out and clarified. Additional superior item to be spoken about is certainly the introduction of rules and algorithms having been welcomed into the DSS core including the following: interaction field, RADAR and NIS synchronization tools; actual navigational hazard in a given cartographic area; ships trajectories and, as a result, simulations of probable deviations in the information perception of the operator.Results. In order to meet beneficial agreement between the effectiveness of the developed DSS with the proposed formalanalytical approaches an experiment was assumed to be appropriate to be conducted using the Navi Trainer 5000 navigation simulator (NTPRO 5000). Based on the foregoing, due to comprehensive results in experiment metadata for the 2.5 years of operation of navigation simulators and DSS software tools the identification of the deviation probabilities in the information perception of dangers was achieved and export the predicted data to new locations for the operator and cartographic areas was performed. Undoubtedly, the experimental investigation confirmed the hypothesis of the study and reflected completely the feasibility of using this DSS to make predictions of possible risks when control the vessel by analyzing the information model of the operator.Conclusions. Formal-analytical approaches presented in the study combined with the developed DSS software tools and the information itself made it possible to classify the decision-making strategies of the operator when control the vessel and to predict the probability of catastrophic consequences. The feasibility of the proposed models and methods was successfully revealed by carried out experiments.
Read full abstract