دانلود رایگان مقاله مدیریت بهره برداری از شبکه تامین-انرژی مسکونی بر اساس روش بهینه سازی

Abstract

An operation management system for residential energy-supplying networks using multiple cogeneration units was developed by hierarchically integrating energy demand prediction, operational planning, and operational control, using optimization approaches. The energy demand for multiple dwellings was predicted by support vector regression with information on occupant behavior as well as forecasted weather and energy demand history. Mixed-integer linear programming was employed for the operational planning of the cogeneration units to the predicted energy demand. The energy demand prediction and operational planning were updated using a variable frequency receding horizon approach. This was done to limit the unnecessary shutdown and start-up of the cogeneration units and to reduce the influences of prediction errors for energy demand. Regarding the operational control, the actual on–off schedule of the cogeneration units conformed to the operational planning result. Additionally, the power and heat outputs of the cogeneration units and the heat supply from the storage tanks were modulated in response to the actual energy demand, based on predefined rules. The developed operation management system was applied to annual operation simulation of a residential energy-supplying network consisting of four cogeneration units using fuel cells in a housing complex. For comparative analysis, history-based approaches for energy demand prediction and separate operation of each cogeneration unit were also considered. The results revealed the effectiveness of the developed operation management system as well as the high energy-saving performance of the residential energy-supplying network.

8. Conclusions

An operation management system for residential energysupplying networks (R-ESN) using multiple CGUs and storage tanks was developed that hierarchically integrates energy demand prediction, operational planning, and operational control. This was achieved by employing an MPC approach. The energy demand for multiple dwellings in the prediction horizon was predicted by SVR, using occupant behavior information as well as forecasted weather and energy demand history. The MILP-based operational planning was conducted to determine the on–off schedule of the CGUs in the control horizon, by using the predicted energy demand, the current on–off status of the CGUs, and the current amount of stored heat. The energy demand prediction and operational planning were updated by using a variable frequency receding horizon approach, in which the prediction and control horizon recedes after a lapse of multiple sampling times. This was done not only to limit the unnecessary shutdown and start-up of CGUs and the influence of prediction errors for the energy demand but also to lessen the computational load in the operation management system. In the operational control, the actual on–off schedule of the CGUs complied with the operational planning results. Further, the power and heat outputs of the CGUs and the heat output from the storage tanks were modulated in response to the actual energy demand, based on predefined rules.