This paper presents a hybrid model called VMD-Bi-GRU for forecasting household electricity consumption. The model combines variational mode decomposition (VMD) with bidirectional gated recurrent units (Bi-GRU) to predict energy consumption for the next 24 hours with a 15-minute time granularity. VMD decomposes the power consumption time series into intrinsic mode functions (IMFs), and Bi-GRU is used to predict each IMF separately. The final prediction is obtained by summing the predictions of each model and rebuilding them. The model achieves a mean squared error of 0.0038 KW, a mean absolute error of 0.046 KW, a mean absolute percentage error of 0.11%, and a high R² score of 0.98, indicating its high accuracy. Smart grids have become more dynamic and interconnected due to advancements in technologies such as AMI, IoT, and AI. However, managing the complexity and unpredictability of power consumption in these grids is challenging. Forecasting energy consumption is crucial for harmonizing generation and load, enabling efficient load control and optimal participation of distributed energy resources (DERs). Due to the stochastic nature of power consumption, which is a random non-stationary signal, efficient forecasting is essential. Traditional methods such as ARIMA, SARIMA, SVR, and random forest have been used for power consumption forecasting, but deep learning models like LSTM, GRU, and CNN have shown better performance. However, these models struggle with high-frequency and dynamic power consumption signals. To improve accuracy, researchers have used wavelet decomposition, empirical mode decomposition (EMD), and ensemble empirical mode decomposition (EEMD) to decompose power signals. VMD is a recent method that decomposes the load consumption signal into IMFs, each with its own unique center frequency. The VMD-Bi-GRU model uses VMD to decompose the power consumption signal into seven IMFs, then uses Bi-GRU to predict each IMF. The predicted IMFs are aggregated to generate the forecasted daily energy usage. The model has been applied in various studies, including improving the accuracy of short-term passenger flow forecasts. The proposed model is compared with other models such as Bi-GRU, GRU, VMD-GRU, linear regression, and extreme learning machine to demonstrate its accuracy.This paper presents a hybrid model called VMD-Bi-GRU for forecasting household electricity consumption. The model combines variational mode decomposition (VMD) with bidirectional gated recurrent units (Bi-GRU) to predict energy consumption for the next 24 hours with a 15-minute time granularity. VMD decomposes the power consumption time series into intrinsic mode functions (IMFs), and Bi-GRU is used to predict each IMF separately. The final prediction is obtained by summing the predictions of each model and rebuilding them. The model achieves a mean squared error of 0.0038 KW, a mean absolute error of 0.046 KW, a mean absolute percentage error of 0.11%, and a high R² score of 0.98, indicating its high accuracy. Smart grids have become more dynamic and interconnected due to advancements in technologies such as AMI, IoT, and AI. However, managing the complexity and unpredictability of power consumption in these grids is challenging. Forecasting energy consumption is crucial for harmonizing generation and load, enabling efficient load control and optimal participation of distributed energy resources (DERs). Due to the stochastic nature of power consumption, which is a random non-stationary signal, efficient forecasting is essential. Traditional methods such as ARIMA, SARIMA, SVR, and random forest have been used for power consumption forecasting, but deep learning models like LSTM, GRU, and CNN have shown better performance. However, these models struggle with high-frequency and dynamic power consumption signals. To improve accuracy, researchers have used wavelet decomposition, empirical mode decomposition (EMD), and ensemble empirical mode decomposition (EEMD) to decompose power signals. VMD is a recent method that decomposes the load consumption signal into IMFs, each with its own unique center frequency. The VMD-Bi-GRU model uses VMD to decompose the power consumption signal into seven IMFs, then uses Bi-GRU to predict each IMF. The predicted IMFs are aggregated to generate the forecasted daily energy usage. The model has been applied in various studies, including improving the accuracy of short-term passenger flow forecasts. The proposed model is compared with other models such as Bi-GRU, GRU, VMD-GRU, linear regression, and extreme learning machine to demonstrate its accuracy.