CN-122026529-A - Multi-source data fusion hydroelectric generating set operation control method and system

CN122026529ACN 122026529 ACN122026529 ACN 122026529ACN-122026529-A

Abstract

The invention provides a multi-source data fusion hydroelectric generating set operation control method and system, and relates to the technical field of hydroelectric generating set optimal control. The method comprises the steps of obtaining historical control parameter sequence data and historical control state record data of a target hydroelectric generating set, dividing the historical control parameter sequence data and the historical control state record data to generate a plurality of generating set control data sets, constructing control load parameter vectors and control state vectors, fusing the control load parameter vectors and the control state vectors to obtain control effect files of each generating set control data set, collecting real-time operation monitoring data of the target hydroelectric generating set, determining the target control effect files, constructing a neighborhood sample set according to the target control effect files to conduct abnormal dominance detection on the real-time operation monitoring data, determining abnormal control load parameter types, and conducting control optimization on the target hydroelectric generating set. The invention realizes the improvement of abnormal response sensitivity and control optimization pertinence of the hydroelectric generating set under complex working conditions, and improves the intelligent level of the running management of the hydroelectric generating set.

Inventors

ZHU BINLIN
HONG JUSHUI
WANG XING
WU WEI
WU CHANGQIAN
XIAO PENG

Assignees

国家能源集团江西电力有限公司
国家能源集团江西电力有限公司万安水力发电厂

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The operation control method of the hydroelectric generating set with multi-source data fusion is characterized by comprising the following steps of: Acquiring historical control parameter sequence data and historical control state record data of a target hydroelectric generating set, dividing the historical control parameter sequence data according to the historical control state record data, and generating a plurality of generating set control data sets; Constructing a plurality of groups of control execution signals and control state signals of each unit control data set, extracting a plurality of control load parameters of each group of control execution signals and constructing a control load parameter vector; carrying out state recognition processing on each group of control state signals to construct a control state vector, fusing a plurality of control load parameter vectors and the control state vector, and constructing a control effect file of each unit control data set; Collecting real-time operation monitoring data of a target hydroelectric generating set, identifying control effects of the real-time operation monitoring data, and determining target control effect files to which the real-time operation monitoring data belong; And constructing a real-time control load parameter vector and a real-time control state parameter of the real-time operation monitoring data, constructing a neighborhood sample set according to the target control effect file, detecting abnormal dominance of the real-time operation monitoring data based on the neighborhood sample set, determining an abnormal control load parameter type and performing control optimization on the target hydroelectric generating set.
2. The method for controlling operation of a multi-source data fusion hydroelectric generating set according to claim 1, wherein the construction of the control load parameter vector and the control state vector comprises: Calculating a plurality of variation parameters of each group of control execution signals, wherein the variation parameters are control parameter variation accumulation values of a plurality of sampling point pairs in the control execution signals, constructing a variation load parameter vector of the control execution signals, counting reverse parameters of the control execution signals in each sliding window, constructing a reverse load parameter vector, performing high-frequency filtering processing on the control execution signals, calculating a plurality of energy parameters, constructing an energy load parameter vector, acquiring a signal operation range of the control execution signals, performing overrun detection on the control execution signals, determining a saturation value of each sampling point, calculating to obtain a plurality of limiting trigger parameters, and constructing a limiting load parameter vector of the control execution signals; The control state signal is subjected to low-frequency filtering processing to generate a target state vector, and a plurality of state stability parameters of the target state vector are calculated to construct the control state vector.
3. The method for controlling operation of a multi-source data fusion hydroelectric generating set according to claim 2, wherein the detecting of abnormal dominance of real-time operation monitoring data based on a neighborhood sample set comprises: Calculating a control effect difference index of the real-time operation monitoring data and each neighborhood sample in the neighborhood sample set, wherein the neighborhood samples comprise a state stability parameter and a plurality of control load parameter vectors; Based on the real-time control load parameter vector and the real-time control state parameter of the real-time operation monitoring data, identifying sample dominant states of a plurality of neighborhood samples relative to the real-time operation monitoring data, and generating a sample dominant label of each neighborhood sample; Performing unit operation anomaly detection on the real-time operation monitoring data according to the sample dominant labels of the neighborhood samples, generating an abnormal operation detection result of the target hydroelectric unit, determining an abnormal control load parameter type related to the target hydroelectric unit according to the abnormal operation detection result, and realizing optimal control on the target hydroelectric unit based on the abnormal control load parameter type.
4. A multi-source data fusion hydroelectric generating set operation control method according to claim 3, wherein identifying sample dominant status of a plurality of neighborhood samples with respect to real-time operation monitoring data, generating a sample dominant label for each neighborhood sample comprises: Calculating control effort values of the real-time operation monitoring data and each neighborhood sample according to the real-time control load parameter vector of the real-time operation monitoring data and the control load parameter vector of the neighborhood sample, calculating a control effect optimization value of each neighborhood sample relative to the real-time operation monitoring data according to the real-time control state parameter of the real-time operation monitoring data and the state stability parameter of the neighborhood sample, determining a sample dominant state of each neighborhood sample relative to the real-time operation monitoring data according to the control effect optimization value and the control effort value, and determining a sample dominant label of the neighborhood sample according to the sample dominant state, wherein the sample dominant label comprises a control dominant sample and a control degradation sample.
5. The method of claim 4, wherein determining the type of abnormal control load parameters for the target hydroelectric generating set based on the abnormal operation detection result comprises: for an abnormal operation detection result, determining dominant intensity parameters and degradation intensity parameters of a neighborhood sample set according to sample dominant labels of neighborhood samples, calculating control abnormal parameters of real-time operation monitoring data according to the dominant intensity parameters and the degradation intensity parameters, and generating an abnormal operation detection result; If the abnormal operation detection result is that the real-time operation monitoring data has an operation abnormal state, determining an abnormal control load parameter type of the real-time operation monitoring data according to the real-time operation monitoring data and the control effort value of each neighborhood sample, wherein the abnormal control load parameter type of the real-time operation monitoring data is obtained by calculating the control load contribution degree of a plurality of control parameters according to the control effort value and carrying out abnormal marking on the plurality of control parameters according to the control load threshold value.
6. The method of claim 5, wherein constructing a limiting load parameter vector of the control execution signal comprises: and determining an overrun value of each sampling point in the control execution signal according to the signal operation range, calculating saturation values of a plurality of sampling points based on the overrun value, fusing the saturation values of the plurality of sampling points to obtain a plurality of limiting trigger parameters, and forming a limiting load parameter vector of the control execution signal according to the plurality of limiting trigger parameters.
7. The method of claim 2, wherein calculating a plurality of state stability parameters of the target state vector to construct the control state vector comprises: And traversing the target state vector through the observation windows, calculating a variance value of each observation window as a state stability parameter, and constructing a control state vector based on a plurality of state stability parameters.
8. The method according to claim 5, wherein the ratio of the dominant strength parameter to the degraded strength parameter is calculated as the abnormal control parameter of the real-time operation monitoring data.
9. A multi-source data fusion hydroelectric generating set operation control system, characterized in that the system is used for realizing a multi-source data fusion hydroelectric generating set operation control method as claimed in any one of claims 1-8, comprising: the operation control data acquisition module is used for acquiring historical control parameter sequence data and historical control state record data of the target hydroelectric generating set, dividing the historical control parameter sequence data according to the historical control state record data and generating a plurality of set control data sets; The control load analysis module is used for constructing a plurality of groups of control execution signals and control state signals of each unit control data set, extracting a plurality of control load parameters of each group of control execution signals and constructing a control load parameter vector; The control effect file construction module is used for carrying out state identification processing on each group of control state signals to construct a control state vector, fusing a plurality of control load parameter vectors and the control state vector and constructing a control effect file of each unit control data set; the real-time operation monitoring module is used for collecting real-time operation monitoring data of the target hydroelectric generating set, identifying control effects of the real-time operation monitoring data and determining target control effect files to which the real-time operation monitoring data belong; The control analysis optimization module is used for constructing a real-time control load parameter vector and a real-time control state parameter of the real-time operation monitoring data, constructing a neighborhood sample set according to the target control effect file, detecting abnormal dominance of the real-time operation monitoring data based on the neighborhood sample set, determining an abnormal control load parameter type and performing control optimization on the target hydroelectric generating set.
10. The multi-source data fusion hydroelectric generating set operation control system of claim 9, wherein for the control load analysis module, constructing the control load parameter vector comprises: Calculating a plurality of variation parameters of each group of control execution signals, wherein the variation parameters are control parameter variation accumulation values of a plurality of sampling point pairs in the control execution signals, constructing a variation load parameter vector of the control execution signals, counting reverse parameters of the control execution signals in each sliding window, constructing a reverse load parameter vector, performing high-frequency filtering processing on the control execution signals, calculating a plurality of energy parameters, constructing an energy load parameter vector, acquiring a signal operation range of the control execution signals, performing overrun detection on the control execution signals, determining a saturation value of each sampling point, calculating to obtain a plurality of limiting trigger parameters, and constructing a limiting load parameter vector of the control execution signals.

Description

Multi-source data fusion hydroelectric generating set operation control method and system Technical Field The invention relates to the technical field of hydroelectric generating set optimal control, in particular to a hydroelectric generating set operation control method and system based on multi-source data fusion. Background The hydroelectric generating set is used as core equipment of a hydroelectric generating system, and the operation control performance of the hydroelectric generating set directly influences the energy conversion efficiency, the operation stability and the service life of the hydroelectric generating set. Along with the improvement of the dispatching complexity of the power grid, the hydroelectric generating set is frequently started, stopped and switched in load under different working conditions, the operation control process of the hydroelectric generating set is more and more complex, and higher requirements are put forward on the self-adaptability and the fine management of a control system. At present, some hydropower unit operation control technologies realize the evaluation and optimization of the unit operation state through the monitoring and analysis of signals such as operation parameters, power output, vibration, temperature and the like. The method plays an important role in guaranteeing the safety and stability of the unit. However, the existing research has focused on static or resultant evaluation of the achieved control effect, less from the standpoint of the degree of effort the control system makes in maintaining a high quality control effect. During long-term operation, the crew control system may have implicitly increased the adjustment amplitude or frequency to counteract the potential degradation effects while maintaining normal output. Such a change in control load, if not captured in time, may delay early identification of a decrease in operating efficiency or degradation in performance. Disclosure of Invention In order to solve the technical problems, the invention provides a multi-source data fusion hydroelectric generating set operation control method and system, wherein a comprehensive evaluation thought of control effort degree is adopted from multi-source operation data so as to more accurately reflect the internal working state of a control system, provide an analysis view angle for early abnormality recognition under the angles of the control efficiency and stability of the generating set, realize operation optimization and improve the intelligent level of hydroelectric generating set operation management. The first aspect of the invention provides a hydroelectric generating set operation control method based on multi-source data fusion, which comprises the following steps: Acquiring historical control parameter sequence data and historical control state record data of a target hydroelectric generating set, dividing the historical control parameter sequence data according to the historical control state record data, and generating a plurality of generating set control data sets; Constructing a plurality of groups of control execution signals and control state signals of each unit control data set, extracting a plurality of control load parameters of each group of control execution signals and constructing a control load parameter vector; carrying out state recognition processing on each group of control state signals to construct a control state vector, fusing a plurality of control load parameter vectors and the control state vector, and constructing a control effect file of each unit control data set; Collecting real-time operation monitoring data of a target hydroelectric generating set, identifying control effects of the real-time operation monitoring data, and determining target control effect files to which the real-time operation monitoring data belong; And constructing a real-time control load parameter vector and a real-time control state parameter of the real-time operation monitoring data, constructing a neighborhood sample set according to the target control effect file, detecting abnormal dominance of the real-time operation monitoring data based on the neighborhood sample set, determining an abnormal control load parameter type and performing control optimization on the target hydroelectric generating set. Preferably, the construction of the control load parameter vector and the control state vector includes: Calculating a plurality of variation parameters of each group of control execution signals, wherein the variation parameters are control parameter variation accumulation values of a plurality of sampling point pairs in the control execution signals, constructing a variation load parameter vector of the control execution signals, counting reverse parameters of the control execution signals in each sliding window, constructing a reverse load parameter vector, performing high-frequency filtering processing on the control execution sig