Short-Term Prediction of Residential Power Energy Consumption via CNN and Multi-Layer Bi-Directional LSTM Networks

Abstract

Excessive power consumption (PC) and demand for power is increasing on a daily basis, due to advancements in technology, the rise in electricity-dependent machinery, and the growth of the human population. It has become necessary to predict PC in order to improve power management and cooperation between the energy used in a building and the power grid. State-of-the-art energy consumption prediction (ECP) methods are limited in terms of predicting the energy effectively, due to various challenges such as weather conditions and the dynamic behaviour of occupants. Thus, to overcome the drawbacks of these methods, we present an intelligent hybrid technique that combines a convolutional neural network (CNN) with a multi-layer bi-directional long-short term memory (M-BLSTM) method using three steps. When applied to short-term power ECP, this approach helps to provide efficient power management i.e. it can assist the supplier to produce the optimum amount of power. The first step in our proposed method integrates the pre-processing and data organisation mechanisms to refine the data and remove abnormalities. The second step employs a deep learning network, where the sequence of refined data is fed into the CNN via the M-BLSTM network to learn the sequence pattern effectively. The third step generates the ECP/PC by comparing actual and predicted data series and evaluates the prediction using error metrics. The proposed method achieves better prediction results than existing techniques, thus demonstrating its effectiveness. Furthermore, it achieved the smallest value of the Mean Square Error (MSE) and Root Mean Square Error (RMSE) for individual household dataset using 10-fold cross validation (CV) and a hold-out (CV) method. INDEX TERMS Artificial intelligence, deep learning, power consumption, CNN, bi-directional LSTM, and short-term energy consumption.

A Novel CNN-GRU-Based Hybrid Approach for Short-Term Residential Load Forecasting

“…A numbers of studies have been conducted in literature for electricity consumption prediction i.e. ARIMA [7], SVR [8], time series modeling [9], linear regression and neuro fuzzy models [10], ANN [11], sequence to sequence learning [12], Deep Recurrent Neural Network (DRNN) [13] and a number of hybrid models [14][15][16][17]. The statistics presented in [18] show different methods' utilization for energy consumption prediction where 47% methods utilize ANN and rest of the researchers employ SVM, decision tree and other models with percentage of 25%, 4% and 25%, respectively.…”

“…In this section, the performance of the proposed CNN-GRU model over IHEPC dataset is compared with baseline models. The results are compared with, linear regression [49] SVM [54], CNN-LSTM [48], autoencoder [49], multilayer bidirectional LSTM (MLBD_LSTM) [14] and deep neural network (DNN) [50] as shown in Figure 6. For instance, linear regression attained 0.…”

“…et al developed CNN-LSTM model for electricity prediction and recorded the lowest error rates as compared to other baseline models, because CNN-LSTM model learned from both spatial and temporal features[39]. Similarly, Ullah et al developed a hybrid model by integrating CNN with multilayer bi-directional LSTM[14]. In hybrid deep learning models, CNNs are used to model spatial features and recurrent models are used to model temporal features but still the error rate is much higher.…”

Domain Fusion CNN-LSTM for Short-Term Power Consumption Forecasting

“…Cascading CNN-LSTM extracts the features from input data by using CNN first, and then CNN extracted hidden features are fed into LSTM to mining the features that have a long-time dependency. e.g., Hu et al applied a cascading structure based on CNN and Bi-LSTM for urban water demand forecasting [25]; Yan et al [26] applied a cascading CNN-LSTM for multi-step power consumption forecasting; Ullah et al [27] combined CNN and multi-layer Bi-directional LSTM for short-term residential power energy consumption prediction with three steps: data refinement, training the model and prediction evaluation. By the contrary, parallel CNN-LSTM extracts the features individually, extracted features from CNN and LSTM are merged as the final features to forecast.…”

From Load to Net Energy Forecasting: Short-Term Residential Forecasting for the Blend of Load and PV Behind the Meter

“…Short-term load forecasting is an essential task to provide reliable operation of modern power systems [4,5]. Due to power system modernization and decentralization, applications of load forecasting have become even more highlighted in today's distribution networks.…”