Machine learning based false data injection in smart grid

Smart grids have two-way power flow, two-way communication system, automated and distributed Energy Network. Communication is the main feature that makes a grid smart but that is the feature, which makes it vulnerable to cyber-attacks. Smart meters are installed to measure the real time data and after measurement this data is sent to control room. In the control room, all the control decisions are based on this received data. In communication lines, this data can be tampered or attacked to mislead the decision-making done in the control room. Load shading, power theft, and delay or blocking of data can be the purpose of an attack. State estimation, support vector machine, and observation of previous patterns are the techniques that can be used to detect the false data injected into the power system. In an effort to devise robust strategies against communication line. we put forth a novel attack strategy, which has not been dealt in the literature earlier. We inject false data into the power system by using Linear regression. We also show that none of the existing defence technique are able to detect the false data.