This　article　assesses　the　energy　management　of　reconfigurable　residential　smart　hybrid　AC/DC　microgrids　considering　the　combined　heat　and　power　(CHP)　loads　as　well　as　the　electric　vehicles　charging/discharging　behaviors.　A　holistic　model　is　developed　for　the　proton　exchange　membrane　fuel　cell　to　retrieve　the　unwanted　thermal　energy　generated　at　the　operation　time.　The　proposed　model　makes　use　of　the　unoccupied　capacity　of　the　fuel　cell　for　producing/storing　hydrogen　for　the　later　usage　and　increasing　its　efficiency.　A　stochastic　framework　is　designed　using　point　estimate　method　(PEM)　to　capture　the　uncertainties　of　the　photovoltaic　and　wind　turbine　forecast　error,　power　company　price,　the　operating　temperature　of　the　proton　exchange　membrane　fuel　cell,　the　price　for　natural　gas,　price　for　selling　hydrogen,　and　the　pressure　of　the　H2　and　O2　in　the　fuel　cell　stack.　The　PEM　approach　has　shown　superior　advantages　in　terms　of　accuracy　and　running　time.　Considering　the　complex　and　nonlinear　structure　of　the　proposed　framework,　a　proficient　optimization　technique　based　on　the　teacher　learning　algorithm　(TLA)　is　devised.　A　two-phase　modification　method　is　proposed　to　increase　the　algorithm　variety　and　help　its　convergence　characteristics.　The　performance　of　the　proposed　algorithm　is　compared　with　the　TLA,　particle　swarm　optimization　(PSO)　algorithm　and　genetic　algorithm　(GA).　For　enhancing　the　security　of　the　energy　and　data　transaction　within　the　system,　a　directed　acyclic　graph　(DAG)-based　security　framework　is　introduced　to　guarantee　the　performance　of　the　system　against　the　subversive　accesses.　By　using　this　scheme,　the　essential　data　of　the　units　are　recorded　and　secured　in　the　form　of　public,　private　and　transaction　blockchains.　The　economic　characteristics　of　the　proposed　method　are　assessed　on　a　residential　hybrid　AC-DC　microgrid　test　system.