| CPC H02J 3/00 (2013.01) [H02J 2203/20 (2020.01)] | 8 Claims |
|
1. A method for performing analysis and calculation of multi-energy flow for integrated energy system (IES), comprising:
establishing an electrical power flow model, a hydraulic model and a thermal model based on a structure of an IES, to form a preliminary model of the IES; wherein, the structure of the IES is of an electric-thermal interconnection system, comprises a thermal network, an electrical power network connected with a large power grid, and a CHP (combined heat and power) unit connecting the thermal network and the electrical power network together;
constructing a double-hidden-layer (DHL) long short-term memory (LSTM) neural network for nonlinear regression of the electrical power flow model, constructing a single-hidden-layer (SHL) LSTM neural network for nonlinear regression of hydraulic model, and finding optimal parameters of the DHL-LSTM neural network and the SHL-LSTM neural network by using Adam (Adaptive Moment Estimation);
training the electrical power flow model and the hydraulic model by using historical data, obtaining regression error data by using the trained electrical power flow model and hydraulic model; constructing error compensation regression models driven by the LSTM neural networks, for adding error compensations into the electrical power flow model and the hydraulic model in training, respectively;
obtaining a mechanism-driven linear thermal model by simplifying the thermal model;
embedding the mechanism-driven linear thermal model into the error-compensated hydraulic model and the error-compensated electrical power flow model respectively, to form a final model of the IES;
performing a multi-energy flow calculation and analysis by inputting parameter variables of a current load demand of the IES into the final model of the IES, outputting physical parameter values for current energy sources dispatching of the electrical power network and physical parameter values for current energy sources dispatching of the thermal power network; and
providing, by a coupling device of the CHP unit in the electric-thermal interconnection system, thermal power to the thermal network according to the output physical parameter values for current energy sources dispatching of the thermal power network; then, calculating electrical power output by the coupling device according to a thermoelectric ratio of the CHP unit, providing the calculated electrical power to the electrical power network from a coupling node where the coupling device of the CHP unit connecting to the electrical power network, and dispatching electrical power energy from the large-scaled power grid to the electrical power network according to the output physical parameter values for current energy sources dispatching of the electrical power network, to balance energy of the electrical power network;
wherein, a process of adding error compensation comprises: dividing historical data into two parts: an electrical power part and a hydraulic part; constructing training sets and test sets respectively; independently training the electrical power flow model and the hydraulic model constructed by the neural networks, to obtain an electrical power regression model and a hydraulic regression model, respectively;
performing difference between preliminary regression results output by corresponding regression models and true values, to obtain electrical power regression error data and hydraulic regression error data; combining and arranging the electrical power regression error data and the hydraulic regression error data with the historical data into a data set, and dividing the data set in an error training set and an error test set of the electrical power and an error training set and an error test set of the hydraulic; and training error compensation regression models constructed by the neural networks, to obtain an electrical power error compensation regression model and a hydraulic error compensation regression model, respectively.
|