Our nations investment in our nuclear power plants is extensive and valuable. However, high operating costs, relative to alternative technologies, threaten the economic viability our nuclear fleet. Operation and maintenance are labor intensive, accounting for nearly two-thirds of the operating cost of a nuclear plant. Artificial intelligence and machine learning can automate monitoring and planning to help optimize decision making and substantially lower operating costs. The proposed solution creates a virtual expert that will continuously monitor each important item of equipment in the plant. This virtual expert never sleeps and will monitor and detect any abnormal changes in the operation of the equipment. When abnormal changes occur, the virtual expert will determine the most likely cause for the problem, based on the symptoms detected, and will propose the corrective action steps to resolve the problem with the lowest risk and cost. Additionally, the virtual expert will continuously estimate the remaining time available before the problem must be corrected to prevent the equipment from reaching an operating or maintenance limit, shutting down or failing. Phase I demonstrates feasibility of a Virtual Subject Matter Expert for the main steam turbine in a nuclear plant. Ultimately, the solution will entail a comprehensive library of these reusable virtual experts that is tailored and provided to equipment owners in nuclear power and other market segments. Three key innovations are integrated and verified in the project. These are 1) hybrid digital twin models for predicting the expected performance of the equipment while it is operating, 2) diagnostic decision models for evaluating any abnormal behavior and suggesting the best corrective action, and 3) prognostic models for accurately predicting the amount of time available to take a corrective action. These integrated models combine multiple artificial intelligence and machine learning algorithms to replicate the analysis methods used by human subject matter experts. The new innovations are built and demonstrated on top of an existing commercial software platform that is widely used across the power industry today for continuous online monitoring of plant equipment. The primary benefit to the public will be lower electricity rates and lower CO2 emissions. Public utilities will benefit from lower operating costs and avoided capital costs to replace nuclear assets with new generating assets. Commercially, the project innovations have broad applicability across nearly all process industries. As an established provider of online monitoring systems to all types of power plants, oil and gas plants and other manufacturers, the company is well positioned to put these innovations into immediate commercial use as an extension of our existing product lines.
