A microgrid power smoothing control system incorporating artificial intelligence load forecasting

Abstract

A microgrid power smoothing control system combining artificial intelligence load forecasting is proposed. The system first receives and captures real-time load data of the microgrid, renewable energy generation data, state of charge of the energy storage system, and environmental and temporal characteristics. Then, it imports a long short-term memory network model with attention mechanism to perform high-precision prediction calculations to accurately calculate the predicted load curve. Based on the prediction results and the rate of change constraint mechanism, the system generates an optimized target grid-connected power curve. Finally, through the dynamic weighting factor formula of the state of charge, the system calculates the charging and discharging commands of the energy storage system with protection mechanism. In this way, the invention effectively upgrades from the traditional "passive compensation" to "active prediction". While ensuring that the grid-connected power complies with grid specifications, it significantly reduces unnecessary micro-charge and discharge cycles of the energy storage system, maximizes battery life, and achieves high-efficiency intelligent management of the microgrid.

Claims

1. A microgrid power smoothing control system incorporating artificial intelligence load prediction, comprising: a data acquisition module including a processing unit, a grid monitoring unit, an environmental sensing unit, a power calculation unit, and a power sensing unit; the processing unit is responsible for fusing and acquiring real-time data from multiple sources and is electrically connected to the grid monitoring unit, the environmental sensing unit, the power calculation unit, and the power sensing unit; the grid monitoring unit is used to monitor real-time load data of the microgrid; the environmental sensing unit collects environmental characteristics; the power calculation unit monitors power generation data of renewable energy and collects time characteristics; and the power sensing unit monitors the state of charge of an energy storage system (ESS); and an intelligent prediction module connected to the data acquisition module, the intelligent prediction module being used to deploy a trained Long Short-Term Memory (LSTM) network model incorporating an attention mechanism, and based on the real-time load data and the extracted environmental characteristics... Based on the time characteristics, the system outputs a predicted load curve and confidence interval for a future time interval; a target curve generation module, connected to the intelligent prediction module, calculates an optimized target grid-connected power curve based on the predicted load curve and renewable energy generation data, using a rate of change constraint mechanism; a dynamic compensation calculation module, connected to the target curve generation module, calculates the power difference between the target grid-connected power curve and the real-time net load, and combines the dynamic weighting factor of the state of charge to output the final charge and discharge command; a closed-loop execution control module, connected to the dynamic compensation calculation module and the intelligent prediction module, drives the energy storage system (ESS) to execute the charge and discharge command, and compares the deviation between the predicted value and the actual value in real time, feeding back the error to the intelligent prediction module for online learning and correction of the model.

2. The microgrid power smoothing control system incorporating artificial intelligence load prediction as described in claim 1, wherein, The processing unit is integrated with a historical database that stores historical load data.

3. The microgrid power smoothing control system incorporating artificial intelligence load prediction as described in claim 1, wherein, The environmental characteristics include ambient temperature, solar radiation intensity, and wind speed, while the temporal characteristics include weekday / holiday labels, peak / off-peak period labels, and seasonal classification codes.

Images