{"id":"W4406709262","doi":"10.1016/j.engappai.2025.110091","title":"A critical review of safe reinforcement learning strategies in power and energy systems","year":2025,"lang":"en","type":"review","venue":"Engineering Applications of Artificial Intelligence","topic":"Smart Grid Security and Resilience","field":"Engineering","cited_by":30,"is_retracted":false,"has_abstract":false,"ca_institutions":"Université Laval","funders":"University of Michigan","keywords":"Computer science; Reinforcement learning; Energy (signal processing); Power (physics); Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002502071,0.0002357725,0.0008852372,0.0003283642,0.00002392441,0.00002327287,0.0002622534,0.0001538887,0.00001274736],"category_scores_gemma":[0.0001972141,0.000238167,0.0001140976,0.0007374161,0.00008173424,0.00007558055,0.00005608248,0.0003078923,0.000003329249],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004573525,"about_ca_system_score_gemma":0.00009084965,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003311575,"about_ca_topic_score_gemma":0.000002730132,"domain_scores_codex":[0.9983724,0.00002893656,0.001046529,0.0002118984,0.0001527329,0.0001875363],"domain_scores_gemma":[0.998981,0.0005052236,0.00009336766,0.0002904746,0.00008279361,0.00004713617],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","study_design_scores_codex":[7.463055e-7,0.00002379385,1.572652e-7,0.2903751,0.0000385347,0.000001177774,0.00006789302,0.2838226,0.000008471954,0.1750873,0.00002189162,0.2505523],"study_design_scores_gemma":[0.00001078538,0.00004884444,4.945701e-7,0.2466794,0.0001901435,0.00001155725,0.0002080222,0.09066106,0.0001428238,0.000224504,0.6612877,0.0005346896],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[3.942471e-7,0.7330719,0.2659713,0.000003248624,0.0001153839,0.000238707,0.000006334775,0.00004288491,0.0005497778],"genre_scores_gemma":[0.002974955,0.9961081,0.0004224507,0.000001701673,0.00003176016,0.000406255,0.00001461208,0.00001891614,0.00002127556],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.6612657,"threshold_uncertainty_score":0.971217,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02016980415017551,"score_gpt":0.3009118143372166,"score_spread":0.2807420101870411,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}