The basic principle of power system relay design and operation has not changed for more than half a century, even after the introduction of microprocessor-based relays in the early 1980s to replace its analogue counterparts. All power system protection relays, as the basis of their operating principle, measure attributes of signals from sensors connected to the power system. The major attributes are those of the fundamental frequency and harmonics of the power system. To detect these attributes mathematical integral transforms, such as the Fourier Transform and Wavelet analysis, are commonly applied. The mathematically intensive integral transform based protection relay algorithms have difficulty in satisfying the demands of ultra-high-speed protection where fast relay reactions to the transient features of fault voltage and current are crucial. It is also unable to extract the fundamental waveform and the exponentially decaying DC components of transient post-fault signals accurately, when using a very small sampling window. For most of the protection relays, complex calculations of integral transform coefficients, relevant to the attributes of fault signals and fault conditions, have to be undertaken during a sampling interval. These calculations introduce many problems for the relay performance, such as accuracy, fast responses, noise, disturbance rejections and reliability. Existing protection relay devices in power substations and plants are physically large and constructed from a large number of electronics components and boards, which easily suffer from electromagnetic disturbances. In contrast to the many applications of VLSI and embedded system technologies, the protection relay devices seem to be behind the times.One of the main investigations of the project is to develop a generic approach of applying Mathematical Morphology (MM) techniques in the design of protection relays. MM has been designated as a new branch of mathematics, which is totally different from the integral transform based methods, in basic principles, algorithmic operations and approach. This project will focus, in depth, on the investigation of a new methodology for extracting components of transient and non-periodic signals by a further study of AMOs. Based on the studies of AMOs, the project will focus on making a prototype of a new generation of power system protection relays - MPRs. This work will contain two aspects. The first part is concerned with investigations on developments of ultra-high-speed protection relays, and the second will focus on extracting fundamental frequency components of post-fault signals and identifying the shape of distorted waveforms, which are commonly required in many protection relays.The MPRs will require much less computation than the conventional integral transform based protection relays, as the computation only involves a few mathematical operations and time delay shifts. Reliability is a crucial issue in protection relay design. The current problem of protection relays' reliability has mainly arisen from the complex calculations for sampling, filtering, integral transform and fault identification, etc., which requires an assembly of many integrated circuits (ICs) to implement. The reliability of current protection relays is greatly affected by the complex algorithms and their associated hardware implementation. Therefore, the MPRs will be much more reliable than the conventional relays if implemented in a single IC using SOC techniques. This implementation will not only significantly enhance the reliability and reduce the size and cost of the whole protection relay system, but potentially create a step change in the methodology for the design and implementation of many protection systems.The MPRs will be comprehensively evaluated in simulation and on a VATECH Reyrolle ACP Ltd's test bed. The project is challenging and adventurous and may lead to a revolution in power system automation.