Decision support at industrial workstations based on a desktop AI agent

Abstract

This paper addresses the problem of decision support at industrial workstations within a closed-loop control environment. The relevance of the study is determined by the distribution of operational data across heterogeneous digital systems, which increases the time required to retrieve relevant information and complicates operational decision-making. A decision support approach based on a desktop AI agent is proposed; the agent interacts with corporate and local digital systems and performs contextual analysis of the current production situation. An algorithm comprising six sequential stages is developed: registration of a production event, collection of data from digital systems, contextual analysis of the obtained information, generation of recommendations, execution of the selected action, and re-evaluation of the state of the production facility. An architecture of interaction between the desktop AI agent, the specialist workstation, corporate information systems, and a computation and analytical processing server is proposed. The approach is demonstrated on a typical production scenario related to a decline in equipment performance. The proposed approach is aimed at reducing decision-making time, improving the completeness of available information used during analysis, and decreasing the probability of errors in the operation of industrial facilities.