Intelligent system for a remote diagnosis of a photovoltaic solar power plant

Abstract

Abstract. Usually small and mid-sized photovoltaic solar power plants are located in rural areas and typically they operate unattended. Some technicians are in charge of the supervision of these plants and, if an alarm is automatically issued, they try to investigate the problem and correct it. Sometimes these anomalies are detected some hours or days after they begin. Also the analysis of the causes once the anomaly is detected can take some additional time. All these factors motivated the development of a methodology able to perform continuous and automatic monitoring of the basic parameters of a photovoltaic solar power plant in order to detect anomalies as soon as possible, to diagnose their causes, and to immediately inform the personnel in charge of the plant. The methodology proposed starts from the study of the most significant failure modes of a photovoltaic plant through a FMEA and using this information, its typical performance is characterized by the creation of its normal behaviour models. They are used to detect the presence of a failure in an incipient or current form. Once an anomaly is detected, an automatic and intelligent diagnosis process is started in order to investigate the possible causes. The paper will describe the main features of a software tool able to detect anomalies and to diagnose them in a photovoltaic solar power plant.

Integrated Multi-Domain Risk Assessment Using Automated Hypothesis Testing

“…Once both cyber and physical anomaly detection analysis is performed, cyber-physical metrics need to be developed to combine the information from both domains and address the tight relations between the power system and the ICT domains. Anomaly detection with regression models has been used for discovering cyber a acks on a SCADA system [20], wind turbine fault detection [22] and photovoltaic plant fault diagnostics [14]. A special case of an a ack against voltage control in distribution power grids has been described in [5].…”

Impact of health indicators on maintenance management and operation of power systems

“…CI can be estimated based on different algorithms and sources of information. Approaches on how to obtain the CI are out of the scope of this work; however, interested readers can refer to several studies [54][55][56][57] for some examples of its estimation. As an example, Mazidi et al [58][59][60] consider collected operation data from several sensors in wind turbines.…”

New applications and control strategies for large-scale CPV plants

“…Based on the available data, the control system detects anomalies. IT expert systems identify failures and suggest countermeasures to the maintenance staff [7]. Automatic soil detection is another important function.…”