Electrical condition monitoring (CM) normally involves the collection of high-frequency, instantaneous data for feature extraction. This paper presents a novel development of an electrical condition monitoring system for wind turbines. The system is developed based upon a control and data acquisition system, for which hardware modules can be configured for a particular set of signals, thus tailoring the system to a specific range of monitoring tasks. A wavelet-based singularity detection method is proposed, which automatically calculates the Lipschitz exponent, a measure to describe the local transient activities in the measurement signal. The relationship between the Lipschitz exponent and the type and severity of faults occurring on the grid and in the power electronics is explored. The proposed algorithms are tested and validated using simulation data from computer simulations of a doubly-fed induction generator (DFIG) wind turbine with a grid connection. A field-programmable gate array (FPGA) embedded in the system has been utilised, allowing the signal processing tasks to be undertaken in real-time for monitoring purposes. The paper demonstrates that a fault signal of small magnitude generated at the early stage of a fault carries the same Lipschitz exponent as the signal of large magnitude generated at the late stage of the fault, thereby providing an early warning before the fault develops into a detrimental one.