Research On Short-term Load Forecasting Of Power System Based On MP-RLST

Electricity is an important energy source that is related to the national economy and people’s livelihood.Because it is difficult to make large-scale storage on it at present,the power of the power system must maintain a dynamic balance.The short-term load prediction of high-precision power systems is not only an important basis for ensuring the safety and reliable operation of the power system,but also directly affects the economy and stability of the power system.Therefore,short-term load predictions have become more and more focus on domestic and foreign scholars’ research.Power load data is not only about the sequence of time,but it has many influencing factors.With the development of the power system,traditional prediction methods have been difficult to meet the requirements of the grid in terms of accuracy.The artificial neural network studied,and improved the neural network based on the minimum peeping Kong long short-term memory(MP-LSTM)neural network,and proposed a variety of prediction models.First of all,a short-term load prediction model that uses the PARTICLLE SWARM Optimization(PSO)algorithm optimization of the MP-LSTM neural network.The MP-LSTM model not only overcomes the problems such as the disappearance of gradient disappearance and gradient explosion of the recurring neural network(RNN).Compared with the conventional long-term memory(LSTM)model,MP-LSTM only retains.A gate-control unit,the only door,the model includes a Sigmoid network layer and a TANH network layer,and adds the state of memory cell status value to the previous moment,reduce model parameters,optimize the model structure,improve the model’s input information of the input information Utilization rate.The improvement of the PSO algorithm to optimize the MP-LSTM neural network,which improves the global excellence of predictive models.The simulation results show that the short-term load prediction model of the MP-LSTM neural network proposed in this article is high Prediction accuracy.In order to improve the processing ability of the model to uncertain data,this article proposes a short-term load prediction model based on the vague-minimum-minor-minimum peep of the minor ridgelet Long Short-Term Memory(MP-RLSTM)neural network.First of all,the three temperature rally functions are blurred for temperature data.Because the spine function has a strong direction selectivity,it can effectively reflect the dynamic and directional characteristics in the load.Therefore Calculation.Through the comparison of simulation experiments,the results show that the fuzzy MP-RLSTM neural network model has a better predictive effect.The Bagging algorithm can increase the difference in neural network integration,improve the generalization capacity of the model,and in order to further improve the performance of the short-term load prediction model,the article proposes a short-term load prediction model based on the integration of the Bagging algorithm and the fuzzy MP-RLSTM neural network.First,use the gray association projection method to screen the original dataset,and then train the base learning device composed of a fuzzy MP-RLSTM neural network of the Bagging algorithm by the collection of similar dates.The combination obtains the final prediction result.Example simulation shows that the models mentioned in this article have higher predictive accuracy and stability on loads in different seasons.