Decentralized electric heating load based on equivalent thermal parameter model faces difficulties in parameter identification and large simulation errors

which cannot meet the needs of power grid regulation. Therefore

a simulation model for decentralized electric heating load is proposed based on long short term memory（LSTM）network. Firstly

the model parameters are determined according to the heat transfer process between the distributed electric heating load and the building. Then the LSTM network parameters are determined by combining the model input variables and output variables to establish the load model. By dynamically updating the indoor temperature in the input data of the test set

the long-term temperature prediction is realized. In order to measure the accuracy of the model

model error evaluation indexes in both vertical and horizontal dimensions are proposed. Finally

the example analysis results show that compared with the second-order equivalent thermal parameter model of distributed electric heating load

the longitudinal error and lateral error of the prediction results of the distributed electric heating load model based on LSTM network are smaller and the model accuracy is higher.