Scientists from Northeastern University in China have created a new method that could detect a big electrical problem in microgrids. Microgrids are pods of power generation that look like islands, with bridges that connect to the main power grid. If power from the main grid fails or stops, the microgrid can disconnect from the main power and continue to supply power to local areas.
The new inverter takes the current that is being supplied by the main grid and converts it into an alternating current. This is something that household electronics can use every day. Power flows through the circuit from a source, like a coffeemaker. When the appliance is no longer in use, there is a switch that interrupts the circuit and reroutes the electrical current to wherever it is needed next.
The problem with this system is that sometimes the switch sticks and the current continues to flow to an appliance that doesn’t need it anymore.
"[An open switch fault] often affects the normal operation of the entire drive system and causes many serious influences," said Zhanshan Wang, Senior Member of IEEE. "For example, [it can cause]... overcurrent stress to other power switches or electronic components... low efficiency; [and] high repair costs."
The switch can be flipped back, which corrects the problem, but you have to know if there is a fault and where the fault is. A switch fault can be dangerous as they often cause electrical fires. Switch faults may not be obvious and are very easy to miss. Many are switched through the microgrid system, which makes it nearly impossible to figure out which one needs to be fixed.
The researchers at Northeastern have been working to solve this problem. They developed an algorithm that accurately identifies multiple signals at multiple levels in a circuit, which can determine if there is a switch fault. The algorithm locates the faulty signals through an artificial neural network, made up of a series of connected computers that have learned to process information based on the information itself.
The algorithm and neural network help detect and identify which of the switches have a fault, according to the scientists. Detection and identification occur simultaneously, so scientists say that this method can improve reliability, efficiency and cost of a microgrid.
The paper on this research was published in IEEE/CAA Journal of Automatica Sinica (JAS) and can be accessed here.