AAGLMES: an intelligent expert system realization of adaptive autonomy using generalized linear models

摘要

Abstract—We earlier introduced a novel framework forudrealization of Adaptive Autonomy (AA) in human-automationudinteraction (HAI). This study presents an expert system forudrealization of AA, using Support Vector Machine (SVM),udreferred to as Adaptive Autonomy Support Vector MachineudExpert System (AASVMES). The proposed system prescribesudproper Levels of Automation (LOAs) for variousudenvironmental conditions, here modeled as PerformanceudShaping Factors (PSFs), based on the extracted rules from theudexperts’ judgments. SVM is used as an expert system inferenceudengine. The practical list of PSFs and the judgments ofudGTEDC’s (the Greater Tehran Electric DistributionudCompany) experts are used as expert system database. Theudresults of implemented AASVMES in response to GTEDC’sudnetwork are evaluated against the GTEDC experts’ judgment.udEvaluations show that AASVMES has the ability to predict theudproper LOA for GTEDC’s Utility Management Automationud(UMA) system, which changes in relevance to the changes inudPSFs; thus providing an adaptive LOA scheme for UMA.udKeywords-Support Vector Machine (SVM); AdaptiveudAutonomy (AA); Expert System; Human Automation Interactionud(HAI); Experts’ Judgment; Power System; DistributionudAutomation; Smart Grid.