Based On Redundant Measurement Of Power Plant Energy Consumption Measurement Data Cleaning Technology And Application

At present, the power plant and electric power-related research department have put into operation like performance monitoring, power monitoring, operation optimization and fault diagnosis systems, and have achieved certain economic benefits. Relying on computer network technology, mass data access technology,high-speed real-time information transmission technology to obtain and store a large number of redundant data from power plant. It is necessary to identify these data that used in work before embarking in work to ensure the accuracy of the data.First, proposing a method to find parameters that have a greater impact on energy consumption from power plant unit measurement data through the application of energy-efficient distribution mechanism matrix equation model, namely that,through data calculation of all temperature and pressure in the main loop thermodynamic system to obtain impact factors of thermodynamic system cycle efficiency, and making the measuring point that have a greater impact on energy consumption as a key point of concern.Coupling relationship exist in thermal system parameters, and all abnormal data will inevitably affect these relations and cause some regularity results, so finding these rules can be helpful for finding the basis of suspected outliers. Based on the principle of curve similarity, the article find that similarities exist in efficiency curve under different load, and finding out the differences between abnormal data efficiency curve and design values efficiency curve by the use of calculation of discrete Fr'echet distance, which can be the basis of finding the suspected outliers,and then make further screening to the focus point to identify suspicious points.Finally, thermal power plant performance remote test systems make line calibration of abnormalities suspicious data that between measured data from field test and running measured data of power plant that from laboratory, in order to achieve online calibration of operational measurement data and reduce system error in operational measurements data. Through calculation of real-time operating parameters of a reheat unit which model is C300 / 227.6-16.7 / 537/537 to find out abnormal data suspicion points, and then make online calibration to the aberrant data suspect points to determine measuring point position that produce the abnormal suspect data.Then replacing the abnormal data with the test data to achieve effect of data cleaning. The results show that the data cleaning method is correct.The data screening method is used in big data platform of power plant energy that under construction of State Grid Corporation of Hebei Electric Power CompanyElectric Power Research Institute and provides accurate data results for platforms.