Renewable energy sources, particularly solar energy systems, have an adequate power supply (PS). But still, PS is widely dispersed and heavily weather-dependent. Hence, forecasting solar output offers a robust answer to such issues. Also, it reduces the workload of delivering electricity through transmission and distribution. In this study, deep learning algorithms are used for precise solar power forecasting (SPF), utilizing a short-term forecasting method. In particular, a long short-term memory (LSTM) approach is used for SPF in the time-series dataset. This paper aims to prevent over-fitting issues that arise in time-series datasets. The real-time solar power time-series dataset is collected from Thiagarajar College of Engineering, Madurai. The SPF are used to evaluate the accuracy using different error types, including mean absolute error(MAE), mean squared error(MSE), and root mean square error(RMSE). Additionally, other algorithms like AdaBoost, linear regression, artificial neural networks, neural networks, space vector machines, and recurrent neural networks are compared to the proposed LSTM algorithm