Capacitance estimation algorithm based on DC-link voltage harmonics using artificial neural network in three-phase motor drive systems

Reliability of dc-link capacitors in modern design of power electronic converters is an important aspect that needs to be considered. The requirement of applying condition monitoring for health status estimation in many reliability-critical applications have been a focused demand. The existing capacitor condition monitoring methodologies are suffering from shortcomings such as, low estimation accuracy, extra hardware, and increased cost, and thereby, they are rarely adopted by industry. Therefore, development of new methods that are based on advanced software algorithms and data processing techniques requiring no extra hardware will be more attractive to industry. In this paper, a condition monitoring methodology is proposed and applied on the dc-link capacitor in a three phase Front-End diode bridge motor drive. The proposed condition monitoring methodology estimates the capacitance value of the dc-link capacitor based on Artificial Neural Network (ANN) algorithm. Two ANNs (ANN1 and ANN2) are proposed, trained and evaluated based on time-domain and frequency-domain parameters. Experiments are conducted to validate the proposed methodology and the effectiveness of the proposed method is examined through an error analysis.