APPLICATION OF PROBABILISTIC METHODS FOR EFFECTIVE AND RELIABLE OPERATION OF ELECTRICAL AND ELECTROMECHANICAL SYSTEMS

This PhD thesis presents novel system control methods that can be utilized for effective and reliable operation of electric grids and passenger elevators. First of all, this study introduces a new spinning reserve allocation optimization technique that takes into account load and renewable power generation, inter-zonal conventional power generating capacity and demand response. Using the bivariate Farlie-Gumbel-Morgenstern probability density function, the framework presented in this thesis utilizes a new method to simulate the power generation of wind farms. In addition, the presented framework uses a Bayesian Network (BN) algorithm to fine-tune the spinning reserve allocation based on previous hours' actual unit commitment, as well as the hour and day types. The model proposed in this study has been tested on the IEEE Two-Area Reliability Test System (RTS) to quantify the effect of the bivariate wind prediction model and the Bayesian network-based Reserve Allocation Adjustment Algorithm (RAAA) on reliability and cost-effectiveness of the power grid. The findings show that combining a bivariate wind forecast model with RAAA improves power grid stability by 2.66 percent while lowering overall system running costs by 1.12 percent. Secondly, the present work introduces an algorithm with an objective of optimal dispatching control of passenger lifts. The algorithm utilizes the data received from video cameras and dispatches the elevator cars based on the passenger count. The proposed algorithm utilizes the information on the number of people and dispatches the lifts with an objective to move the maximum number of passengers to the desired building levels within the minimum amount of time. In addition, the algorithm considers each person's size and whether or not they have luggage. To account for uncertainty in image acquisition, the algorithm assigns the probability weights to the number of people who are waiting for a lift and riding the lifts. The main purpose of the algorithm is to minimize the following performance metrics: average travel time (ATT), average journey time (AJT) and average waiting time (AWT). The suggested algorithm works well in situations of limited traffic sizes, according to a test case scenario conducted on a ten-story office building having four elevator cars (less than 200 people). In a scenario with large up-peak high intensity traffic, the proposed algorithm primarily underperforms. The proposed algorithm's best output was seen in situations with random inter-floor passenger movement. In scenarios of changing traffic intensity and size ATT increased by 39.94 percent and 19.53 percent, respectively