In the vicinity of weaving areas, freeway congestion is nearly unavoidable due to their negative effects on the continuous freeway mainline flow. The adverse impacts include increased collision risks, extended travel time, and excessive emissions and fuel consumption. Dynamic Speed Harmonization (DSH), which is also known as Variable Speed Limit (VSL), has the potential to dampen traffic oscillation during congestion. However, the effectiveness of this strategy is typically limited by the low compliance rates of drivers and potential delays in information transmission and dissemination, and that control strategies can only affect a small area. Fortunately, new opportunities are emerging with the development of Connected and Automated Vehicles (CAVs) that can completely comply with the control system. CAVs can greatly help complement the intelligent transportation systems to enhance a variety of Measures of Effectiveness (MOEs), such as safety, mobility, and environmental sustainability. Optimizing mixed flow involving Human-Driven Vehicles (HDVs) and CAVs is expected to exist as a challenging issue for a long time prior to the full adoption of CAVs.The objective of this dissertation is to investigate the effects of DSH in mixed traffic flow involving HDVs and CAVs on the freeway. A safety-oriented DSH strategy based on Deep Reinforcement Learning (DRL) is developed to better understand how CAVs can improve operational performance. A holistic performance evaluation is conducted to quantify the impacts under different Market Penetration Rates (MPRs) of CAVs in multiple simulated scenarios. The mixed traffic flow integrated with DSH highlights the synergies and trade-offs across different metrics. The results reveal that for the recurrent congestion, the proposed method can enhance freeway mobility and achieve co-benefits with safety, and environmental sustainability could be improved under higher MPRs. Spatiotemporal features of bottleneck speed demonstrate that DSH powered by CAVs can smooth the speed variations for partial areas. Sensitivity analysis of headways indicates that high-level CAVs can further improve performance. For the nonrecurrent congestion, the implementation of DSH can further improve safety and enhance mobility with increasing CAV penetration rates. While special events may exacerbate congestion, their impact can be mitigated to some extent through DSH. Spatiotemporal patterns of speed variations at the bottleneck demonstrate that the DRL controller has the capability to dampen oscillations. A series of numerical experiments also indicate the adaptability of the agent under adverse weather scenarios, and the differences of surrogate safety measurements in response to various parametric thresholds. Moreover, a lane-based Multi-Agent Dynamic Speed Harmonization (MADSH) system prevents the proposed strategy getting stuck in local optimization. This study provides essential insights to foster a deeper understanding of the transformative potential of the CAV-powered DSH technique in promoting intelligent transportation systems.