Study And Applications On Monitoring Data Integration And Operation Analysis For Smart Hydropower Stations

With the construction of the smart grid and the adjustment of energy structure,the smart hydropower station(SHS)has become the main development direction and new target of hydropower industry.As an important part of SHSs,the condition monitoring system plays an important role in ensuring the safety and stability of the hydropower station(HS).However,due to the independent operation of condition monitoring systems and the lack of interconnect channels and information sharing mechanisms,the monitoring data are stored in various monitoring systems in different formats,leading to a large number of “isolated information islands”.The “isolated information island” which not only seriously affects the sharing and application of the monitoring data,but also is not conducive to the operation analysis and fault diagnosis of strong coupled and complex system of HSs has become a bottleneck problem in the construction of SHSs.How to integrate the discrete data resources and build an integrated data platform,based on which carry out comprehensive data analysis to further improve the fault alarm and early warning capability,the automatic and intelligent level of operation analysis has become an urgent problem to be solved in the construction of SHSs.To solve the practical engineering problems in the construction of SHSs,the study and applications on monitoring data integration and operation analysis for SHSs is conducted under the instruction of the theories and methods of data integration,data reliability,system engineering as well as the knowledge and experience accumulated by field experts after summarizing the current situation of the condition monitoring technology,monitoring data integration,data cleaning and operation analysis of HSs.To solve the problem of "information isolated island" in HSs,the study on data integration approach is conducted combining with the actual engineering requirements in the construction of SHSs.The target of monitoring data integration is expounded,an integrated monitoring data platform(IMDP)for SHSs has been constructed with the reference of achievements obtained by the studying team.The organization approach and association strategies of monitoring data in the IMDP are designed,based on which,the association integration of monitoring data in unit level is realized with the driven of operation condition synchronization and event synchronization,the association integration of monitoring data in station level is realized with the driven of data calling and data transfer.As a result,the unified data warehouse is constructed and the integrated model of "system dispersion and data centralization" is achieved.Further more,the monitoring metadata model is built and the monitoring data sharing service model based on metadata is proposed to solve the standardization problem in the integration and application of mult-isource,heterogeneous and multi-degree data.A large number of erroneous data with serious negative impact on operation analysis and fault diagnosis are mixed into the IMDP because of monitoring equipment defects and signal interference.To solve this problem,the study on data cleaning apporach is conducted combining with multi-source and redundancy of monitoring data in the IMDP and the inherent characteristics of the monitoring data.On the basis of accurate identification of the operating conditions of the hydroelectric generating set,the erroneous data are identified by combining of validity check,multi-source redundancy check,multi-state association check,and time delay clustering.After which,the erroneous data rejected by data identification are restored by multi-source redundant switch,multi-state association analytic calculation,or soft sensor based on operating condition and grey-related fuzzy support vector regression.Finally,the data cleaning experiments for oil system state data of governor,the dissolved gas state data of the transformer,and vibration state data are performed and achieve satisfactory application results.Fault alarm and early warning capability is an important index to characterize the intelligent level of HSs.Aiming to solve the problem of frequent false alarm,frequent missed alarm,lack of early warning,and low alarm coverage in the existing automatic systems,a automatic alarm and early warning approach based on states and performance is proposed according to the system description and relationships model.On the one hand,the abnormalities of states are detected by combining of dynamic threshold analysis,trend analysis with similar influence factors and analogy analysis with similar influence factors,based on which the alarm and early warning based on states are realized.On the other hand,the performance indices of tasks executed by systems are calculated by the macro test models and experts’ experience models,and the performance quality is evaluated,based on which the alarm and early warning based on performance are realized.Finally,the field knowledge of automatic fault alarm and early warning is medeled with the reference of ontology theory,thus to provide standardized and formal knowledge support for high coverage alarm and early warning of the operation analysis system(OAS).Operation analysis is an important means to detect anomalies of equipment and prevent failures of equipment.In order to meet the requirements of operation analysis in HSs,the research and construction of the OAS has been carried out based on the IMDP and the automatic alarm and early warning approach.The software framework with four layers consisting of multi-source and heterogeneous data oriented universal data layer,object-oriented distributed business layer,presentation layer based on MVC framework,self-reconfiguration expert interaction layer is proposed.Based on which,a universal OAS integrating the functions modules of remote real-time monitoring,hierarchical data configuration query,intelligent report,intelligent inspection,special equipment analysis is developed.The practicability of the system,the effectiveness of data integration and the fault alarm and early warning approach are verified by the industrial application of Gezhouba Hydropower Station and Geheyan Hydropower Station.