Faults are natural for any practical system that needs to be diagnosed accordingly to maximize the efficiency of the system. However, faults can be rectified only if they are diagnosed correctly. Therefore, a credible diagnosis is required for the maintenance of the system. Furthermore, communication of diagnosis data from remote sites to the control center is also critical for a reliable diagnosis.

In this project, solar photo-voltaic (PV) systems are considered that harvest the electrical energy from sunlight and are usually deployed at remote sites. First, a solar PV fault diagnosis technique is developed to detect and localize various faults occurring in the solar PV system. Then, an independent component analysis (ICA)-based multiple-input multiple-output (MIMO) system is proposed to transmit data through wireless channels from remote sites to the control center. ICA is an unsupervised machine learning technique that enables us to communicate without having any prior information about the signals and the wireless channel.

The ICA-based MIMO system transmits independent copies of the source signals of each string obtained from the fault diagnosis algorithm. The receiver receives linear mixtures of the transmitted source signals. The received mixed signals are separated through the ICA algorithm to get back the original source signals for further investigation. Furthermore, the ICA-based MIMO system did not require training bits to estimate the wireless channel, and also multiple signals can be transmitted through a single channel. These advantages make this technique a good choice to transmit data wirelessly in a solar PV fault diagnosis system.