Two stage robust economic dispatching of microgrid considering uncertainty of wind, solar and electricity load along with carbon emission predicted by neural network model

The fluctuation characteristics of wind, solar and electric load in microgrid are predicted by BP neural network. Based on forecast error, an economic robust optimization model is established considering uncertain variables, flexible load and carbon emission, solving by C&CG algorithm, which adjusted for individual uncertainty degree and overall conservatism of variables. The power of gas turbine is closely related to time-of-use price while the power of energy storage depends on the relationship of electrovalence and operation cost and flexible load is transferred to valley consumption time. As overall conservatism increases, the electricity purchase increases where microgrid refers to buy more electricity to resist risks, leading to 1897.4 yuan and 1007.2 kg of carbon emission added. Robust scheduling has lower costs compared to intra-day scheduling of 14.2 % decreasing when the intra-day electrovalence increases. Considering carbon quota constraint dispatching is more inclined to utilize gas turbine and energy storage while the total cost would rise by 247.6 yuan. With changing the operating cost of gas turbine, a balance point could be obtained in the game between economy and carbon emission. And there would exist an optimal equipment operation strategy when carbon emission is sufficiently enough, which realizes a balance of energy-carbon-economy. •Use neural network prediction errors as uncertainty set for robust day-ahead optimization.•Realize uncertainty degree adjustment for multi-variables and overall conservatism changing.•Explain the operation law along with boundary conditions of energy equipment under time-of-use price.•Obtain a balance point of economy and carbon emission simultaneously together with energy distribution.