Nowadays the Modular Multilevel Cascaded Converter (MMCC) is a family of emerging high-voltage multilevel converters that are configured with a cascaded connection of identical submodules with low-voltage ratings by distinct topological structures. The MMCC system is featured with a high quantity of coupled system variables (converter currents and floating submodule voltages) and abundant discrete control inputs (submodule switching states). To guarantee a stable and optimal system operation, it is a fundamental challenge to fully model and control these variables. This thesis addresses two frameworks for the control-oriented MMCC modeling as well as the hierarchical analysis. The first framework in Part I presents a comprehensive classification of MMCC topologies, analyzes them by replacing converter branches with continuous controllable voltage sources and develops a unified modeling procedure for current and branch energy, aiming for a general understand of MMCC in the context of continuous system theory. The second framework aims to develop an explicit relation between submodule switching states and MMCC system variables, which preserves the characteristics of discrete switched system. Two practical direct control methods, e.g., fast reduced control set and event-based method, are proposed, which achieves comparable harmonic performance and obviously improved submodule voltage balancing under the premise of the same switching frequency as the conventional submodule-voltage-sorting method.