{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33306"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33306","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"Machine Learning-Augmented Energy Management of Electric Vehicle Powertrains with Hybrid Energy Storage","abstract":"Battery electric vehicles depend on a single battery pack to handle steady-state cruising, high-power acceleration transients, and regenerative braking events across ambient temperatures from sub-zero winter mornings to hot summer highway runs. High current loads accelerate capacity fade, and degradation is strongly temperature dependent. At 40 °C, the same current that produces negligible wear at −20 °C can result in a substantial increase in capacity loss. Pairing the battery with an ultracapacitor (UC) relieves this stress, but only if the power split is managed intelligently across the operating envelope. This thesis develops a physics-informed neural networks (PINNs) energy management strategy for a battery-UC hybrid energy storage system in a battery electric vehicle. Dynamic programming (DP) is solved offline across three drive cycles and six ambient temperatures from −20 °C to 40 °C, using an Arrhenius capacity-fade model fitted to Samsung INR21700-30T cell data. The PINN embeds three physics-informed terms in its training loss: a data fidelity term anchored to the DP-optimal policy, a DC-bus feasibility penalty, and an Arrhenius aging residual that penalises battery-intensive operation at elevated temperatures. A temperature-compensated, rate-limited fuzzy logic (FL) controller serves as a credible real-time baseline. On a fully held-out WLTP cycle, the surrogate achieves an RMSE of 0.0383 with zero DC-bus violations. In closed-loop operation, the PINN recovers 90-99% of the DP aging benchmark across the evaluated drive cycles and reduces cumulative capacity loss by 8-29% relative to the fuzzy baseline at 40 °C, with the largest improvement observed under urban driving conditions. Inference requires approximately 16,640 floating-point operations per step and fewer than 5,000 parameters, well within automotive microcontroller constraints. The framework generalises to hybrid energy storage systems with differentiable degradation models, providing a pathway for translating offline optimal solutions into deployable real-time controllers.","abstract_html":"Battery electric vehicles depend on a single battery pack to handle steady-state cruising, high-power acceleration transients, and regenerative braking events across ambient temperatures from sub-zero winter mornings to hot summer highway runs. High current loads accelerate capacity fade, and degradation is strongly temperature dependent. At 40 °C, the same current that produces negligible wear at −20 °C can result in a substantial increase in capacity loss. Pairing the battery with an ultracapacitor (UC) relieves this stress, but only if the power split is managed intelligently across the operating envelope. This thesis develops a physics-informed neural networks (PINNs) energy management strategy for a battery-UC hybrid energy storage system in a battery electric vehicle. Dynamic programming (DP) is solved offline across three drive cycles and six ambient temperatures from −20 °C to 40 °C, using an Arrhenius capacity-fade model fitted to Samsung INR21700-30T cell data. The PINN embeds three physics-informed terms in its training loss: a data fidelity term anchored to the DP-optimal policy, a DC-bus feasibility penalty, and an Arrhenius aging residual that penalises battery-intensive operation at elevated temperatures. A temperature-compensated, rate-limited fuzzy logic (FL) controller serves as a credible real-time baseline. On a fully held-out WLTP cycle, the surrogate achieves an RMSE of 0.0383 with zero DC-bus violations. In closed-loop operation, the PINN recovers 90-99% of the DP aging benchmark across the evaluated drive cycles and reduces cumulative capacity loss by 8-29% relative to the fuzzy baseline at 40 °C, with the largest improvement observed under urban driving conditions. Inference requires approximately 16,640 floating-point operations per step and fewer than 5,000 parameters, well within automotive microcontroller constraints. The framework generalises to hybrid energy storage systems with differentiable degradation models, providing a pathway for translating offline optimal solutions into deployable real-time controllers.","abstract_has_math":false,"creators":["Otoo, Esi Afariwa"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Mechanical Engineering","school":null,"contributors":[],"advisors":["Emadi, Ali"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:37:15Z","subjects":["Electric Vehicles","Hybrid Energy Storage Systems","Energy Management Systems","Physics-Informed Neural Networks","Dynamic Programming","Battery Degradation","Ultracapacitors","Machine Learning","Battery Electric Vehicles","Real-Time Control"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/168"],"render_values":[{"text":"https://doi.org/10.71548/168","href":"https://doi.org/10.71548/168","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33306","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33306","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33306","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Emadi, Ali"]},{"key":"dc:contributor.department","label":"Department","values":["Mechanical Engineering"]},{"key":"dc:creator","label":"Author","values":["Otoo, Esi Afariwa"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-06-29T14:49:12Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Electric Vehicles","Hybrid Energy Storage Systems","Energy Management Systems","Physics-Informed Neural Networks","Dynamic Programming","Battery Degradation","Ultracapacitors","Machine Learning","Battery Electric Vehicles","Real-Time Control"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33306","https://doi.org/10.71548/168"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Battery electric vehicles depend on a single battery pack to handle steady-state cruising, high-power acceleration transients, and regenerative braking events across ambient temperatures from sub-zero winter mornings to hot summer highway runs. High current loads accelerate capacity fade, and degradation is strongly temperature dependent. At 40 °C, the same current that produces negligible wear at −20 °C can result in a substantial increase in capacity loss. Pairing the battery with an ultracapacitor (UC) relieves this stress, but only if the power split is managed intelligently across the operating envelope. This thesis develops a physics-informed neural networks (PINNs) energy management strategy for a battery-UC hybrid energy storage system in a battery electric vehicle. Dynamic programming (DP) is solved offline across three drive cycles and six ambient temperatures from −20 °C to 40 °C, using an Arrhenius capacity-fade model fitted to Samsung INR21700-30T cell data. The PINN embeds three physics-informed terms in its training loss: a data fidelity term anchored to the DP-optimal policy, a DC-bus feasibility penalty, and an Arrhenius aging residual that penalises battery-intensive operation at elevated temperatures. A temperature-compensated, rate-limited fuzzy logic (FL) controller serves as a credible real-time baseline. On a fully held-out WLTP cycle, the surrogate achieves an RMSE of 0.0383 with zero DC-bus violations. In closed-loop operation, the PINN recovers 90-99% of the DP aging benchmark across the evaluated drive cycles and reduces cumulative capacity loss by 8-29% relative to the fuzzy baseline at 40 °C, with the largest improvement observed under urban driving conditions. Inference requires approximately 16,640 floating-point operations per step and fewer than 5,000 parameters, well within automotive microcontroller constraints. The framework generalises to hybrid energy storage systems with differentiable degradation models, providing a pathway for translating offline optimal solutions into deployable real-time controllers."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Applied Science (MASc)"]},{"key":"dc:title","label":"Title","values":["Machine Learning-Augmented Energy Management of Electric Vehicle Powertrains with Hybrid Energy Storage"]},{"key":"others:access_status","label":"Others Access Status","values":["embargo"]}]}],"canonical_facts":{"dc:contributor.advisor":["Emadi, Ali"],"dc:contributor.department":["Mechanical Engineering"],"dc:creator":["Otoo, Esi Afariwa"],"dc:date.accessioned":["2026-06-29T14:49:12Z"],"dc:date.issued":["2026"],"dc:description.abstract":["Battery electric vehicles depend on a single battery pack to handle steady-state cruising, high-power acceleration transients, and regenerative braking events across ambient temperatures from sub-zero winter mornings to hot summer highway runs. High current loads accelerate capacity fade, and degradation is strongly temperature dependent. At 40 °C, the same current that produces negligible wear at −20 °C can result in a substantial increase in capacity loss. Pairing the battery with an ultracapacitor (UC) relieves this stress, but only if the power split is managed intelligently across the operating envelope. This thesis develops a physics-informed neural networks (PINNs) energy management strategy for a battery-UC hybrid energy storage system in a battery electric vehicle. Dynamic programming (DP) is solved offline across three drive cycles and six ambient temperatures from −20 °C to 40 °C, using an Arrhenius capacity-fade model fitted to Samsung INR21700-30T cell data. The PINN embeds three physics-informed terms in its training loss: a data fidelity term anchored to the DP-optimal policy, a DC-bus feasibility penalty, and an Arrhenius aging residual that penalises battery-intensive operation at elevated temperatures. A temperature-compensated, rate-limited fuzzy logic (FL) controller serves as a credible real-time baseline. On a fully held-out WLTP cycle, the surrogate achieves an RMSE of 0.0383 with zero DC-bus violations. In closed-loop operation, the PINN recovers 90-99% of the DP aging benchmark across the evaluated drive cycles and reduces cumulative capacity loss by 8-29% relative to the fuzzy baseline at 40 °C, with the largest improvement observed under urban driving conditions. Inference requires approximately 16,640 floating-point operations per step and fewer than 5,000 parameters, well within automotive microcontroller constraints. The framework generalises to hybrid energy storage systems with differentiable degradation models, providing a pathway for translating offline optimal solutions into deployable real-time controllers."],"dc:description.degree":["Master of Applied Science (MASc)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33306","https://doi.org/10.71548/168"],"dc:language.iso":["en"],"dc:subject":["Electric Vehicles","Hybrid Energy Storage Systems","Energy Management Systems","Physics-Informed Neural Networks","Dynamic Programming","Battery Degradation","Ultracapacitors","Machine Learning","Battery Electric Vehicles","Real-Time Control"],"dc:title":["Machine Learning-Augmented Energy Management of Electric Vehicle Powertrains with Hybrid Energy Storage"],"dc:type":["Thesis"],"others:access_status":["embargo"]},"updated_at":"2026-11-18T19:37:15Z"}