The operators of power distribution systems strive to lower their operational costs and improve the quality of the power service provided to their customers. Furthermore, they are faced with the challenge of accommodating large numbers of Distributed Energy Resources (DERs) into their grids. It is expected that these problems will be tackled with a large-scale deployment of automation technology, which will enable the real-time monitoring and control of power distribution systems (i.e., similar to power transmission systems). For this purpose, real-time situation awareness w.r.t. the state and the stability of the system is needed. In view of the deployment of such automation functions into power distribution grids, there are two binding requirements. Firstly, the system models have to account for the inherent unbalances of power distribution systems (i.e., w.r.t. the components of the grid and the loads). Secondly, the analysis methods have to be real-time capable when deployed into low-cost embedded systems platforms, which are the cornerstones of automation. In other words, the analysis methods need to be computationally efficient. This thesis focuses on the modeling of unbalanced polyphase power systems, as well as the development, validation, and deployment of real-time methods for State Estimation (SE) and Voltage Stability Assessment (VSA) of such systems. More precisely, the following theoretical and practical contributions are made to the field of power system engineering. 1. Fundamental properties of the compound admittance matrix of polyphase power grids are identified. Specifically, theorems w.r.t. the rank of the compound admittance matrix, the feasibility of Kron Reduction (KR), and the existence of compound hybrid matrices are stated and formally proven. These theorems hold for generic polyphase power grids (i.e., which may be unbalanced, and have an arbitrary number of phases). 2. A Voltage Stability Index (VSI) for real-time VSA of polyphase power systems is proposed. The proposed VSI is a generalization of the well-known L-index, which is achieved by integrating more generic models of the power system components. More precisely, the grid is represented by a compound hybrid matrix, slack nodes by Thévenin equivalents, and resource nodes by polynomial load models. In this regard, the theorems mentioned under item 1 substantiate the applicability of the proposed VSI. 3. A Field-Programmable Gate Array (FPGA) implementation for real-time SE of polyphase power systems is presented. This state estimator is based on a Sequential Kalman Filter (SKF), which - in contrast to the standard Kalman Filter (KF) - is suitable for implementation in such dedicated hardware. In this respect, it is formally proven that the SKF and the standard KF are equivalent if the measurement noise variables are uncorrelated. To achieve high computational performance, the grid model is reduced through KR, and the SKF calculations on the FPGA are parallelized and pipelined. 4. The methods stated under items 1-3 are deployed into an industrial real-time controller, which is used to control a real-scale microgrid. This microgrid is equipped with a metering system composed of Phasor Measurement Units (PMUs) coupled with a Phasor Data Concentrator (PDC). The real-time capability of the developed methods is validated experimentally by measuring the latencies of the PDC-SE-VSA processing chain w.r.t. the PMU timestamps.