A Cyber-Security Framework For Boiler of Steam Power Plant Using Simulation Data to Train Machine Learning Model

The paper proposes an architecture of cyber security framework to utilize artificial intelligence (AI) in optimizing thermal power plant operations. Thus, we choose a steam power plant as a case study which consists of several critical subsystems such as; the boiler, condenser, turbine, feed-water, and other auxiliaries. We study a boiler subsystem of an actual power plant and simulated it at an abstract level using Simulink/MATLAB tool. The aim is to generate a data-set from this simulation model in order to be used in an AI approach. We compare our data-set to the real power plant data-set and based on that we do the calibration for our simulation model. The reason behind selecting the simulation data-set is that we can control the input parameters in a simple way without involving and affecting the actual power plant. Therefore, comparing the reality of generated data-set with the real power plant in terms of the outputs. Using IS027001 which is an international standard for handling information security, and its framework and its standards include Annex A for controlling information security. Here, we will focus on Annex A.14 and A.17 which are system acquisition, development and maintenance, and information security aspects of business continuity management, respectively. This paper discusses the proposed cyber security framework and preliminary results for the collected data-set of the initial model which will then move to further process using neural designer software for training and testing using linear regression.