System-Optimal Dynamic Traffic Assignment (SO-DTA) problem optimizes the time-dependent traffic flow of a transportation network, with varied objectives and complicated constraints. Traffic management applications of Intelligent Transports Systems (ITS), such as emergency evacuation, network optimization, and route planning, rely on SO-DTA models to obtain sophisticated solutions to optimize network performance. In this paper, an improved route-based linear programming (IRLP) optimization model is proposed, specially tailored for the emergency evacuation problem of pedestrians in ITS. The proposed model utilizes the Cell Transmission Model (CTM) and linear programming to model the pedestrian evacuation and introduces penalty labels in the IRLP formulations to solve the holding-back flow issue. Besides, when designing the model, the categories of transmission cells are simplified to make the model more practical, compared to some previous research work. Finally, a series of simulation experiments, including the comparison to the real-world shortest-path evacuation strategies, are conducted to showcase the effectiveness in saving valuable evacuation time.