Short-Term Load Forecasting Based on Deep Learning Bidirectional LSTM Neural Network

Abstract

Accurate load forecasting guarantees the stable and economic operation of power systems. With the increasing integration of distributed generations and electrical vehicles, the variability and randomness characteristics of individual loads and the distributed generation has increased the complexity of power loads in power systems. Hence, accurate and robust load forecasting results are becoming increasingly important in modern power systems. The paper presents a multi-layer stacked bidirectional long short-term memory (LSTM)-based short-term load forecasting framework; the method includes neural network architecture, model training, and bootstrapping. In the proposed method, reverse computing is combined with forward computing, and a feedback calculation mechanism is designed to solve the coupling of before and after time-series information of the power load. In order to improve the convergence of the algorithm, deep learning training is introduced to mine the correlation between historical loads, and the multi-layer stacked style of the network is established to manage the power load information. Finally, actual data are applied to test the proposed method, and a comparison of the results of the proposed method with different methods shows that the proposed method can extract dynamic features from the data as well as make accurate predictions, and the availability of the proposed method is verified with real operational data.

Dimensionality Reduction and Prediction of Impedance Data of Biointerface

“…Today, LSTM applications are mainly concerned with predicting upcoming events or signals [25,30,31]. In our research, we used the LSTM network, and its main use was as a forecasting tool of a signal [30,31] to study the ability to estimate upcoming changes in impedance behavior using the normalized DF signal which briefly describes the impedance changes.…”

“…The LSTM network in this research was used as a forecasting method [25] for a values sequence or signal by training the neural network on DF values and checking the difference between the predicted behavior and the actual one. The principle is easy: at each time step of the input sequence, the LSTM network learns to predict the value of the next time step.…”

Early Warning Method for Coal and Gas Outburst Prediction Based on Indexes of Deep Learning Model and Statistical Model

“…During the training of the network model, problems such as the insufficient training data sets, the sample uniformity, the excessive noise interference in the training data, or the model over complexity may arise, resulting in over-fitting. Aiming at the over-fitting problem of the bidirectional LSTM model in predicting with waveform data, the dropout mechanism is introduced to restrict the sparsity in the random region of the model (Cai et al, 2021). The dropout mechanism randomly discards a certain fraction of neural units from the network temporarily.…”

Short-Term Electrical Load Forecasting Based on VMD and GRU-TCN Hybrid Network

“…Furthermore, load forecasting methods based on combined algorithms emerge continuously, which have achieved results with a faster calculation speed and higher precision. A deep learning bidirectional LSTM neural network is proposed in [17]; the method gives a deep learning structure of bidirectional LSTM to avoid the gradient explosion problem. A hybrid short-term load forecasting is proposed in [18], which uses the combination of correlation analysis and appropriate inputs to the individual Bayesian neural network, but the parallel training and its weight of Bayesian neural network introduce an extra computational burden.…”