Framework for generation scheduling and equivalent dynamic modeling in generation-mix scenarios

With active distribution networks (ADNs) having renewable energy sources (RES), it has become apparent to operate it economically with improved stability. Scheduling of energy resources (ERs) units for generation mix scenarios needs to address stability issues as well. This study precisely proposes a framework using probability density function (PDF) to decide about scheduling, while taking into account voltage stability. The analysis is given on PDF of voltage and active power signals, influenced by dynamic changes in power flow conditions on account of generation mix scenarios. The study presents admittance modeling using recurrent neural network based long short-term memory (LSTM) to forecast the dq components for microgrid formed by combinations of ERs, loads and solar irradiations available on day of test. The same course of PDF analysis and modeling is also performed for an IEEE-123 bus integrated with 15 %, 50 % solar PV penetration plus 50 % battery storage integration. The accuracy in modeling is summarised using a performance indices. This admittance model featured as LSTM network represented as two-input and one-output, can form a basis for signal input to dq component based control, which in turn can serve for scheduling of generation units towards unit commitment objective with voltage stability in mind.