The Information infrastructure of smart grid plays the role of neural system, which carries various types of signal to the brain (control center) to react to the rapidly changing environment. The leakage of confidential message flowing through the network not only compromises the information security but also threatens the operation of the physical system. In Dec. 2015, a coordinated cyberattack on the Ukrainian power grid left more than 200,000 people within blackout for 30 minutes. This cyberattack immediately caught enormous concerns for the future integration of information technology of smart grid. As a large geographically distributed system, the operation of power system is highly nontrivial in the history, where all generators and consumers are connected through alternating coupling line, thus introducing the nonlinear nature to the whole system. To address the desire of a more reliable and efficient system, more real time automations are required to provide full control and extensive visibility of the entire system. All these urge for the new information and communication technologies to integrate all subsystems together. However, in the deregulated power systems since the 1970s, monopoly is eliminated in the power market, and the authority is dispersed to hundreds of companies. The privatization of power system allows power producers to pursue profits in the power market, while they share the same responsibility to maintain the interconnected system. It is well known that a noncooperative control approach for a large distributed system, where each party only considers its own profit, may lead to catastrophe. In such scenarios, information sharing among all subsystems is inevitable.
However, with the merge of competitive generation and transmission market, sharing confidential information comprises the privacy, reveals individual strategies and endangers the future profits. To handle the problem, independent third parties, such as independent system operators (ISOs) and regional transmission organizations (RTOs), will play a critical role in coordinating, controlling and monitoring the operation of the entire system. For the current situation, a contract is signed to authorize the trusted third parties to gather these confidential information with advanced communication network. In spite of that, malicious attackers could access or eavesdrop the information via the vulnerable information network. Therefore, the data must be protected with disguise. In general, there are several approaches to retain privacy. One is the traditional encryption method, which encodes the message in a manner such that only authorized participants can decode it. Another interesting approach, which caught attention recently, is to add dedicated random noise to perturb the true values. Here we consider a new approach to design an approximated control strategy with barely noticeable sacrifice of the performance loss. The basic idea used in this paper is based on the decomposition of functions and the corresponding dimension reduction. Consider two parties (e.g., two regional power providers) that intend to compute the control policy, which requires the local information of the two parties. We can consider the control policy as the computation of a function f(x,y), where x and y are the local parameters (e.g., the parameters of generators) of the two parties. However, they do not want to leak the full information of x and y to each other. Then, we intend to decompose the function f, namely to find functions g, h, s such that f(x,y)= g(h(x),s(y)). In this project, the PI has leveraged the framework of nomography for the function decomposition, such that only the necessary information for the computing is sent to the center, such that the privacy of each sub-system (e.g., different utility companies) is kept.

Last Modified: 12/26/2019
Modified by: Husheng Li