Short-Term Power Load Forecasting Using CNN-LSTM Network Optimized by GTO Algorithm

短期电力负荷预测对电力系统调度至关重要，其预测精度直接影响电网运行的经济性和安全性。针对传统预测方法在处理非线性、非平稳负荷数据时存在的局限性，本文提出一种基于人工大猩猩部队优化算法(GTO)优化的CNN-LSTM混合模型(GTO-CNN-LSTM)，通过结合CNN的空间特征提取能力和LSTM的时序建模优势，并利用GTO算法优化关键超参数，显著提升预测精度。基于中国某地2014~2015年的实际负荷数据实验表明，该模型的RMSE (427.8)较单一LSTM和CNN-LSTM分别降低25.5%和14.5%，MAPE降至4.905%，且R2达0.931，验证了其优越性。该模型为电力负荷预测提供了高精度解决方案，可推广至其他时序预测任务。Short-term power load forecasting is crucial for power system dispatching, as its accuracy directly impacts the economic efficiency and security of grid operations. To address the limitations of traditional forecasting methods in handling nonlinear and non-stationary load data, this paper proposes a hybrid CNN-LSTM model optimized by the artificial Gorilla Troops Optimizer (GTO) algorithm (GTO-CNN-LSTM). By integrating the spatial feature extraction capability of CNN and the temporal modeling advantages of LSTM, and leveraging the GTO algorithm to optimize key hyperparameters, the model significantly improves forecasting accuracy. Experimental results based on actual load data from a region in China during 2014~2015 demonstrate that the proposed model achieves an RMSE of 427.8, representing reductions of 25.5% and 14.5% compared to standalone LSTM and CNN-LSTM models, respectively. Additionally, the MAPE drops to 4.905%, while the R² reaches 0.931, confirming its superiority. This model provides a high-accuracy solution for power load forecasting and can be extended to other time-series forecasting tasks.