Reducing congestion and improving transportation systems heavily depend on accurate traffic flow forecasting. However, conventional models frequently overlook extraneous variables such as real-time dynamic weather changes, special events, temperature, and rainfalls that might have a big impact on traffic patterns. Therefore, the proposed study aims to use Artificial Intelligence supported deep learning techniques to investigate how real-time weather data affect the forecast of traffic flow. To improve prediction performance, the incorporation of external variables has also been utilized in the proposed study. Quality of Service (QoS) has also been taken into consideration on account of analyzing and recognizing its pivotal role in predicting the traffic flow. Hence, in this paper, QoS-based Deep Learning scheme (QSTP) has been proposed which is classified into two different levels for enhancing the accuracy of predictions and reducing overhead during the process. This research underscores the significance of integrating external variables and QoS considerations along with a deep learning approach to contribute to accurate and robust traffic flow predictions required for mitigating congestion and optimizing transport systems.