Development of a method for intelligent diagnostics of the technical condition of a synchronous turbine generator based on electrical parameters

Abstract

Relevance: synchronous turbo generators are one of the main electro-technological devices for generating electricity in the power supply system, and their stable and reliable operation is important for ensuring the continuity of power supply. The turbogenerator operates in a mode of active power P, reactive power Q, output voltage U, excitation current I_ex and rotor angular velocity deviation Δω, as well as electrical and Deviations of electromechanical parameters from their nominal values negatively affect the dynamic stability of the turbogenerator, the voltage mode, and the quality of operation of the excitation system. The variation of these electrical parameters in an interlocking way makes it possible to evaluate failure States based on only one parameter complicates. Therefore, the use of calculated diagnostic indicators along with measured parameters, their comprehensive assessment of deviations, and the intellectual determination of the current state of the turbine generator are urgent scientific and technical problems.

Aim: the development of an intellectual diagnostic method that allows you to assess the violations in the basic electrical and electromechanical parameters of a synchronous turbogenerator in a complex way, form diagnostic residual signals between measured and calculated parameters and, on their basis, determine the working condition of the turbogenerator by normal, disruptive and dangerous conditions.

Methods: this article presents an intelligent diagnostic method based on fuzzy logic and artificial intelligence to assess the technical condition of a synchronous turbo generator based on electrical parameters and to detect faults in them.

Results: the research results were obtained based on the electrical parameters of a 110 MW synchronous turbine generator operating at the JSC “Navoi Thermal Power Plant” and based on the disturbances occurring in the network. In the proposed method, I_ex based on changes in the parameters P, Q, and U_ac and w_h accounting magnitudes were formed, and due to their comparison with real measured values, it was possible to identify inconsistencies in the excitation system and electromechanical mode as an additional diagnostic sign. An algorithm for classifying the state of a turbogenerator into normal, abnormal and dangerous modes based on normalized parameter disturbances and residual signals was developed. As a result, it was possible to comprehensively assess the disturbances in the electrical parameters of the turbine generator, early detect dangerous conditions, and form diagnostic information for the subsequent adaptive control system.