Continued increase in system load leading to a reduction in operating margins, as well as the tendency to move towards a deregulated grid with renewable energy sources has increased the vulnerability of the grid to blackouts. Advanced intelligent techniques are therefore required to design new monitoring schemes that enable smart grid operation in a secure and robust manner. As the grid is highly interconnected, monitoring of transmission and distribution systems is increasingly relying on digital communication. Conventional security assessment techniques are slow, hampering real-time decision making. Hence, there is a need to develop fast and accurate security monitoring techniques. Intelligent techniques that are capable of processing large amounts of captured data are finding increasing scope as essential enablers for the smart grid. The research work presented in this thesis has evolved from the need for enhanced monitoring in transmission and distribution grids. The potential of intelligent techniques for enhanced system monitoring has been demonstrated for disturbed scenarios in an integrated power system. In transmission grids, one of the challenging problems is network partitioning, also known as network area-decomposition. In this thesis, an approach based on relative electrical distance (RED) has been devised to construct zonal dynamic equivalents such that the dynamic characteristics of the original system are retained in the equivalent system within the desired accuracy. Identification of coherent generators is another key aspect in power system dynamics. In this thesis, a support vector clustering-based coherency identification technique is proposed for large interconnected multi-machine power systems. The clustering technique is based on coherency measure which is formulated using the generator rotor measurements. These rotor measurements can be obtained with the help of Phasor Measurement Units (PMUs). In distribution grids, accurate and fast fault identification of faults is a key challenge. Hence, an automated fault diagnosis technique based on multi class support vector machines (SVMs) has been developed in this thesis. The proposed fault location scheme is capable of accurately identify the fault type, location of faulted line section and the fault impedance in the distributed generation (DG) systems. The proposed approach is based on the three phase voltage and current measurements available at all the sources i.e. substation and at the connection points of DGs. An approach for voltage instability monitoring in 3-phase distribution systems has also been proposed in this thesis. The conventional single phase L-index measure has been extended to a 3-phase system to incorporate information pertaining to unbalance in the distribution system. All the approaches proposed in this thesis have been validated using standard IEEE test systems and also on practical Indian systems.