Generator differential expansion directional monitoring and regulation method under frequent and rapid deep peak regulation conditions

Abstract

The present application belongs to the technical field of generator operation monitoring and regulation, and discloses a generator differential expansion directional monitoring and regulation method under frequent, rapid and deep peak regulation working conditions. The method comprises operation data acquisition and synchronous processing, working condition identification and temperature difference baseline establishment, differential expansion characteristic calculation, state evaluation and graded early warning, operation parameter linkage control, multi-source abnormality correlation determination, and model correction and parameter updating. By establishing the corresponding relationship between temperature difference and load, current and reactive power, a copper-iron differential expansion characteristic parameter set is constructed, real-time identification and graded response of differential expansion deviation are realized, and based on historical sample correction model and threshold, dynamic optimization of determination parameters is realized. The method can realize continuous monitoring and active regulation of generator thermal stress in peak regulation operation, and improve operation stability and equipment safety.

Description

Technical Field

[0001] This invention belongs to the field of generator operation monitoring and control technology, specifically a method for directional monitoring and control of generator expansion difference under frequent, rapid and deep peak shaving conditions. Background Technology

[0002] When a generator operates under deep peak shaving conditions in the power grid, its output power, excitation current, and cooling conditions change frequently. The stator and rotor conductors and the iron core will experience periodic thermal expansion differences due to temperature fluctuations, i.e., the copper-iron expansion difference phenomenon. The expansion difference not only affects the symmetry of the air gap magnetic field, but also leads to an imbalance in the distribution of electromagnetic forces. If it continues to exceed the limit, it can easily cause rotor eccentricity, stator bar deformation, and local stress concentration in the insulation material.

[0003] Current monitoring systems mostly collect data on single operating parameters such as current, voltage, and temperature, lacking continuous tracking of the coupling relationship between the thermal state and electromagnetic characteristics of the stator and rotor. Especially under frequent and rapid peak-shaving conditions, the unit load changes at a high rate, and traditional methods that rely on periodic inspections or distributed monitoring cannot reflect the dynamic correlation between the copper-iron temperature difference and excitation parameters in real time. Although some technologies use infrared or thermocouple arrays to estimate the temperature difference distribution, they are limited by the arrangement of measuring points, response speed, and data synchronization accuracy, and cannot accurately capture the transient change process of the expansion difference.

[0004] Existing control systems generally use fixed threshold triggering logic. When the ambient temperature, cooling medium flow rate, or hydrogen purity changes, the threshold often lags behind the actual operating conditions, which can easily lead to false alarms or missed alarms. At the same time, the cooling, excitation, and insulation systems have obvious coupling during peak shaving. Anomalies in partial discharge signals, dielectric loss angle, and hydrogen humidity are often related to deviations in copper-iron expansion difference. However, existing technologies lack time correlation analysis and joint judgment mechanisms for multi-source data, resulting in delayed anomaly identification and slow control response. Summary of the Invention

[0005] The purpose of this invention is to provide a method for directional monitoring and control of generator expansion differential under frequent, rapid, and deep peak shaving conditions, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for directional monitoring and control of generator expansion difference under frequent, rapid and deep peak shaving conditions. This method constructs a mechanism for synchronous acquisition of multiple parameters of unit operating parameters, condition identification, feature extraction, hierarchical early warning, linkage control and model self-correction. This enables real-time monitoring and adaptive control of generator expansion difference under peak shaving conditions such as rapid load changes, deep output reduction and frequent start-stop, effectively suppressing stress accumulation caused by the thermal expansion difference of copper and iron, and improving the safety and stability of long-term unit operation. During generator operation, the system synchronously collects parameters such as stator and rotor voltage, current, active power, reactive power, excitation current, excitation voltage, stator bar temperature, core temperature, cooling water inlet and outlet temperatures, cooling flow rate, and hydrogen humidity. Based on the unit's unified clock, it performs time alignment, data integrity verification, and abnormal data removal to obtain calibrated operating parameter time series data, providing reliable input for subsequent feature calculations and operating condition identification. Based on the load power, excitation current and cooling flow parameters in the operating data, the operating conditions of the unit are identified, and the states of full load, partial load, deep peak shaving, hot standby and turning gear are distinguished. Under stable operating conditions, characteristic quantities such as copper-iron temperature difference, stator-rotor temperature rise difference and cooling efficiency are extracted, and the correspondence between temperature difference and load, current and reactive power is established to form a temperature difference baseline model that is adapted to multiple operating conditions. This model can adaptively adjust the fitting parameters under different operating conditions, thereby reducing the impact of load fluctuation on temperature difference judgment. By combining the temperature difference baseline model with the thermal expansion characteristics of materials, the expansion difference signal between the stator and rotor is calculated. Furthermore, by integrating the vibration acceleration signal, excitation current waveform, and cooling water temperature, key parameters such as the vibration main frequency energy ratio, the correlation coefficient between excitation current and reactive power, and the stability of the cooling water temperature difference are extracted to construct a set of characteristic parameters for copper-iron expansion difference. This fusion calculation method can distinguish different types of expansion difference caused by structural deformation and thermal unevenness, thereby improving the accuracy and interpretability of the judgment. The system performs real-time analysis of the expansion difference characteristic parameter set, calculates the deviation and rate of change of the copper-iron expansion difference relative to the temperature difference baseline, and generates an expansion difference warning when the deviation or rate of change exceeds the preset threshold. Based on the degree of deviation and duration, the system classifies the warning signals into three levels: prompt level, warning level, and severe level. This graded judgment mechanism enables operators to take corresponding intervention measures according to different risk levels, realizing full-process monitoring from trend prompts to severe anomalies. When the warning level reaches the warning level or the severe level, the control system executes the linkage control of the operating parameters in priority order. First, the cooling water pump speed is increased to enhance the heat exchange efficiency. If the expansion difference still does not return to the safe temperature difference threshold range after the cooling adjustment, the upper limit of the excitation current is gradually reduced to reduce the rate of change of the copper-iron temperature difference. At the same time, the slip ring temperature and carbon brush current distribution are monitored. If the slip ring temperature exceeds the limit or the current distribution is abnormal, the exit phase operation is immediately executed and the control event is recorded to prevent the rotor circuit from overheating and the carbon brush from burning. This progressive control logic takes into account both response speed and operation stability. During the control process, the system continuously monitors multi-source status parameters such as partial discharge signal, insulation dielectric loss angle and hydrogen humidity change. When any two abnormal signals and the copper-iron expansion difference deviation occur at the same time, and their time overlap ratio exceeds the set threshold, the system determines it as a multi-source correlation anomaly and automatically outputs load reduction or shutdown command. This multi-parameter correlation judgment method can effectively reduce the misjudgment of a single signal and improve the accuracy of anomaly identification under complex working conditions. During long-term operation, the system stores information such as control results, alarm records, and maintenance conclusions in the database, and periodically corrects the temperature difference baseline model and early warning threshold. When the model prediction error exceeds the preset error threshold, the system adjusts the model parameters according to the error direction and magnitude, and regenerates the temperature difference baseline and threshold range for each operating condition, so that the model maintains calculation accuracy in different seasons and operating environments, and realizes adaptive correction of expansion difference judgment and control strategy. This data-driven parameter update mechanism can track equipment performance changes in the long term, avoiding the insensitivity of traditional static models to equipment aging and environmental drift.

[0007] Furthermore, in the process of acquiring and synchronizing operational data, this invention constructs a multi-system parallel acquisition and unified processing mechanism for generator operating parameters. By integrating data from generator online monitoring devices, on-site inspection systems, and production management information systems, it achieves comprehensive acquisition and centralized management of the unit's electrical, thermal, and environmental operating parameters. Among them, the online monitoring device continuously acquires electrical signals such as voltage, current, excitation current, excitation voltage, active power, and reactive power of the stator and rotor; the point inspection system periodically collects stator bar temperature, core temperature, and cooling water inlet and outlet temperatures; the production management information system provides environmental operating parameters such as cooling flow rate and hydrogen humidity. After being collected in parallel by multiple systems, a complete dataset covering the main operating characteristics of the unit can be formed, thereby realizing comprehensive monitoring of the operating status. To ensure consistency in time and accuracy of data collected from different systems, this invention uses the unified clock signal of the unit as a reference for time synchronization; and uses the moving average filtering method to identify and remove abnormal measurement point values ​​caused by sensor drift or transient interference. Subsequently, a cross-system data correspondence is established based on the measurement point number and system identifier to complete the association and calibration of multi-source data, so that data from different sources can be accurately matched under the same time reference. After time synchronization and calibration, the operational data is stored in a structured manner in a time series format. Each record contains a timestamp, measurement point number and corresponding measured value. This storage format facilitates subsequent operating condition identification and temperature difference baseline modeling, significantly improves the accuracy of data correspondence between different systems in the analysis stage, and provides reliable data input for expansion difference feature extraction and model updating. Through the above-described acquisition and synchronization process, this invention realizes the transformation from multi-source heterogeneous sampling to fusion processing under a unified time reference, eliminating the deviation problem of traditional generator operating data in terms of sampling period and time alignment, and providing highly consistent data support for subsequent operating condition identification and temperature difference model calculation.

[0008] Furthermore, in the process of identifying operating conditions and establishing a temperature difference baseline, this invention selects key characteristic quantities such as load power, excitation current, and cooling flow rate as the basis for determining operating conditions based on the calibrated operating parameter time series. The system analyzes the time variation trend of load power and excitation current, and combines it with the scheduling information in the unit operation plan to comprehensively determine the current operating status of the generator, classifying it into operating condition types such as full load, partial load, deep peak shaving, hot standby, and turning gear. This determination method enables the system to accurately divide the operating stage under conditions of frequent load fluctuations, avoiding operating condition identification errors caused by instantaneous power changes. When a stable operating condition is identified where the load fluctuation is within a set threshold, the system calculates the average temperature difference between the copper conductor and the iron core to obtain the copper-iron temperature difference. At the same time, it calculates the temperature rise difference between the stator and the rotor and determines the cooling efficiency based on the temperature difference between the inlet and outlet of the cooling water. These thermal parameters reflect the heat transfer characteristics and cooling capacity of the unit under different load conditions, providing a quantifiable physical basis for the subsequent establishment of a temperature difference model. Based on the acquired thermal parameters, the system pairs the copper-iron temperature difference, stator-rotor temperature rise difference, and cooling efficiency with the load power, excitation current, and reactive power at the corresponding time to generate a temperature difference sample set with time index. Then, the least squares fitting method is used to fit the multidimensional functional relationship between temperature difference and load, current, and reactive power to obtain a temperature difference baseline model. This model reflects the thermal balance state of the generator under different operating conditions and can be used as a quantitative reference for subsequent copper-iron expansion difference calculation and state determination. Through the above data identification and modeling process, this invention transforms the multidimensional thermal parameter relationship under complex operating conditions into an adaptive temperature difference baseline, realizing a dynamic description of the generator's thermal characteristics. This method can maintain the consistency and accuracy of expansion difference judgment under conditions of drastic load changes or deep peak shaving operation, significantly improving the reliability of monitoring the thermal deformation trend of the unit, and overcoming the problem that traditional fixed threshold methods are prone to misjudgment under multiple operating conditions.

[0009] Furthermore, in the calculation and extraction of copper-iron expansion difference characteristics, this invention, based on the aforementioned temperature difference baseline model, retrieves the copper conductor temperature and iron core temperature data for the corresponding time period, and calculates the instantaneous expansion difference between the stator and rotor by combining the linear expansion coefficients of the conductor and iron core materials. By calculating the unit time change amplitude of the expansion difference in a continuous time series, a time-series signal reflecting the dynamic change trend of the expansion difference is obtained. This signal can objectively reflect the transient response of the thermal expansion difference between the stator and rotor under rapid load changes, providing quantifiable input data for subsequent anomaly identification. Under a unified time reference, the system synchronously acquires multi-source operating data such as vibration acceleration signals, excitation current waveforms, and cooling water outlet temperatures. Based on the spectral power distribution of the vibration signals, the proportion of the main frequency energy is extracted to characterize the main vibration response of the structure under dynamic loads. Based on the time series of the excitation current waveform and reactive power, the correlation coefficient is calculated to reflect the degree of time coupling between electromagnetic excitation and output power. At the same time, the cooling stability index is determined based on the variance of the cooling water outlet temperature fluctuation to evaluate the stability of the temperature distribution of the cooling system. The system matches the main frequency energy ratio, the correlation coefficient between excitation current and reactive power, and the cooling stability index with the expansion difference signal according to the time index, and constructs a set of expansion difference characteristic parameters composed of vibration characteristics, electromagnetic characteristics and thermal characteristics. This set comprehensively describes the operating status of the unit under thermal, mechanical and magnetic coupling conditions, and provides data support for subsequent expansion difference status assessment and graded early warning. By introducing a feature-based computational mechanism that integrates multi-source signal fusion, this invention comprehensively considers the interconnected changes in structural vibration, electromagnetic load, and cooling conditions on the basis of traditional temperature difference monitoring, and achieves directional identification of the source of expansion difference. This method can effectively improve the accuracy and sensitivity of expansion difference abnormality judgment, reduce the probability of misjudgment and missed detection, and thus improve the operational safety and reliability of the unit under deep peak shaving operation conditions.

[0010] Furthermore, in the state assessment and graded early warning stage, this invention calculates and determines the operating status and trend of the copper-iron expansion difference of the generator in real time based on the aforementioned set of expansion difference characteristic parameters. The system calculates the deviation of the copper-iron expansion difference from the temperature difference baseline model at each sampling time, and calculates the change difference of the deviation value in adjacent sampling periods to depict the dynamic change trend of the expansion difference over time. When either the deviation value or the change difference exceeds the deviation judgment interval determined by the temperature difference baseline model, the system automatically determines it as an expansion difference abnormal event and generates a corresponding early warning signal to the monitoring system. To facilitate timely monitoring of equipment status by operators, this invention classifies the degree of anomaly based on the magnitude and duration of the expansion deviation, dividing the warning signals into three levels: alert, warning, and severe. The alert level reflects minor expansion fluctuations, the warning level corresponds to continuous and significant deviations, and the severe level indicates that the expansion change may affect the structural safety of critical components of the unit. Warning signals of different levels are transmitted to the alarm signal channel of the monitoring system via a data bus, and the trigger time, threshold, and corresponding expansion parameters are automatically recorded in the system log, providing accurate input for subsequent operational linkage control and model parameter correction. Through the above-mentioned graded assessment and early warning mechanism, this invention establishes a hierarchical response system for abnormal expansion, realizing early identification and layered intervention. When the unit's operating parameters are still within the safe range, the system can issue a trend risk warning, avoiding delayed alarms or misjudgments caused by a single threshold judgment, thereby significantly improving the generator's operational safety and real-time monitoring capabilities under frequent, rapid, and deep peak shaving conditions.

[0011] Furthermore, the linkage control of the operating parameters includes: When the warning level is determined to be a warning level or a severe level, the control system adjusts the unit operating parameters according to a preset control sequence that prioritizes the cooling system and then the excitation system. When making adjustments, first increase the speed of the cooling water pump to increase the cooling water flow. When the speed of the cooling water pump reaches the upper limit and the copper-iron expansion difference deviates from the allowable range specified for unit operation, further reduce the upper limit of the excitation current to slow down the rate of change of the copper-iron temperature difference. During parameter adjustment, the slip ring temperature and carbon brush current distribution status are continuously monitored. When the slip ring temperature exceeds the specified value or the carbon brush current distribution is uneven, the exit phase control command is executed to limit the heat accumulation in the excitation circuit. Each parameter adjustment command, execution time, corresponding warning level, and monitoring result is recorded in the control log as input for subsequent anomaly detection and model correction.

[0012] Furthermore, in the linkage control of operating parameters, when the differential expansion warning level reaches the warning level or the severe level, the present invention automatically triggers the parameter linkage adjustment mechanism of the control system; the system adjusts the key operating parameters of the unit according to the preset control sequence of cooling system priority and excitation system second, so as to achieve dynamic suppression and thermal balance restoration of copper-iron differential expansion. When the system receives an early warning signal, the cooling system takes the lead in responding by increasing the cooling water pump speed to increase the cooling water flow, thereby improving heat exchange efficiency and reducing the temperature difference between the stator and rotor. If the copper-iron expansion difference still does not return to the allowable temperature difference range specified in the unit's operating technical specifications after the cooling water pump speed is increased to the set upper limit, the system further adjusts the excitation circuit parameters and lowers the upper limit of the excitation current to reduce the additional heat accumulation caused by magnetic field changes and suppress the continued expansion of the copper-iron temperature difference. During the parameter adjustment process, the system continuously monitors the slip ring temperature and carbon brush current distribution status. When the slip ring temperature exceeds the safety control threshold set by the sensor, or the difference in current distribution between carbon brushes exceeds the set balance threshold, the system automatically executes the control command to exit the leading phase operation to prevent local overheating or current concentration in the excitation circuit and ensure stable operation of the motor under high load conditions. The control system automatically records the execution time, corresponding warning level, monitoring results and feedback status of each parameter adjustment in the control log, forming a closed-loop control link with parameter adjustment, monitoring feedback and result verification as the core. This data link provides a real and reliable input basis for subsequent abnormal trend analysis and model parameter correction. Through the above-mentioned linkage control strategy, this invention realizes coordinated and adaptive adjustment of the cooling and excitation systems under frequent and rapid peak shaving conditions, transforming differential expansion control from passive response to active correction, effectively preventing differential expansion runaway caused by thermal imbalance or excitation abnormality. This method significantly improves the unit's operational safety, thermal stability, and system adaptability under deep peak shaving conditions.

[0013] Furthermore, in the model correction and parameter update process, this invention constructs a multi-condition sample database covering different loads and operating states based on the control results, early warning records and maintenance conclusions during unit operation. This sample data comes from the real records of the unit's online monitoring system and operation management system, ensuring that the model correction process has traceability and engineering accuracy. The system compares the actual expansion difference values ​​recorded in historical samples with the predicted values ​​of the temperature difference baseline model, calculates the absolute and relative deviation percentages of the two, and quantitatively evaluates the accuracy of the model prediction. When the model prediction error exceeds the preset error threshold range, the system automatically adjusts the correction ratio of relevant parameters in the temperature difference baseline model according to the direction and magnitude of the error change, and simultaneously corrects the warning threshold parameters, so that the temperature difference prediction value output by the model is consistent with the actual operation monitoring results in a statistical sense. The corrected model regenerates a new set of temperature difference baseline parameters and upper and lower thresholds for each operating condition, and stores the updated results in the unit operation database in real time to replace the original parameters for subsequent expansion difference determination and operation control. The model correction process can be triggered automatically according to a preset time period or when the model error continues to exceed the limit, ensuring that the model parameters are dynamically updated as the equipment status and operating environment change. By introducing a model adaptive correction mechanism based on operational feedback and maintenance results, this invention enables the temperature difference baseline model to be gradually optimized according to the random group operating state, thereby effectively improving the long-term stability and prediction accuracy of expansion difference judgment and control strategy, preventing misjudgment problems caused by model aging or parameter drift, and ensuring that the generator always maintains operation within the safe thermal balance range under frequent, rapid and deep peak shaving conditions.

[0014] Furthermore, the present invention provides the following technical solution: an operation data acquisition and synchronization processing module, used to acquire electrical quantities, temperature quantities and cooling system parameters of generator stator and rotor, and to perform time synchronization, anomaly rejection and standardization processing based on the unit's unified clock to form a calibrated sequence of operation parameters; The operating condition identification and temperature difference baseline establishment module is used to identify the unit's operating status based on the load power, excitation current and cooling flow rate in the operating parameter sequence, extract the copper-iron temperature difference and cooling efficiency characteristics under stable operating conditions, establish a correspondence model between temperature difference and load, current and reactive power, so as to generate a temperature difference baseline model. The copper-iron expansion difference characteristic calculation and extraction module is used to calculate the expansion difference between the stator and rotor based on the temperature difference baseline model and the thermal expansion characteristics of the conductor and iron core materials, and to form a set of expansion difference characteristic parameters by combining the vibration signal, excitation current waveform and cooling temperature. The status assessment and graded early warning module is used to calculate the deviation magnitude and rate of change based on the set of expansion difference characteristic parameters, and generate a graded early warning signal when it exceeds a set threshold. The operating parameter linkage control module is used to adjust the cooling flow rate and the upper limit of the excitation current in priority order when the warning level of the graded warning signal reaches the warning level or the severe level, and to perform the exit phase advance operation when abnormal temperature rise or uneven current distribution is detected. The multi-source anomaly correlation judgment and handling module is used to comprehensively analyze partial discharge signals, dielectric loss angle and hydrogen humidity. When multiple anomalies and expansion difference deviations occur simultaneously and the time overlap rate exceeds the threshold, it outputs a load reduction or shutdown signal. The model correction and parameter update module is used to correct the temperature difference baseline and early warning threshold based on the load reduction or shutdown signal and the operation and maintenance data. When the prediction error exceeds the limit, the model parameters are automatically updated to achieve dynamic optimization of the judgment and control threshold.

[0015] Furthermore, the operational data acquisition and synchronization processing module includes: The data acquisition unit is used to collect operating parameter data from the generator online monitoring device, the field inspection system and the production management information system respectively. The data providing unit is used for the online monitoring device to collect stator and rotor voltage, current, excitation current, excitation voltage, active power and reactive power signals; the point inspection system collects stator bar temperature, core temperature and cooling water inlet and outlet temperature; and the production management information system provides cooling flow rate and hydrogen humidity data. The relationship establishment unit is used to align multi-source data with the unit's unified clock, remove outlier values ​​at measurement points through moving average filtering, and establish cross-system data correspondence based on the measurement point identification number. The sequence storage unit is used to store the calibrated data in chronological order as a sequence of operating parameters. The sequence of operating parameters includes timestamps, measurement point numbers, and corresponding measured values, which serve as inputs for subsequent operating condition identification and temperature difference baseline establishment.

[0016] The beneficial effects of this invention are as follows: 1. This invention synchronously acquires multiple parameters of the stator, rotor, and cooling system, realizing the modeling of the correspondence between copper-iron temperature difference, excitation current, and load power. The established temperature difference baseline model can describe the normal expansion difference distribution of the unit under different loads and excitation conditions, making the expansion difference change a quantifiable basis for judgment. The system can continuously track the dynamic changes of copper-iron expansion difference during peak-shaving operation and promptly identify deviation trends caused by sudden load changes or uneven cooling. Compared with the method of relying solely on single-quantity monitoring of temperature or current, this method realizes time synchronization and correlation analysis of multi-source data, improving the accuracy of expansion difference identification and the continuity of monitoring.

[0017] 2. This invention increases the cooling water flow rate, limits the upper limit of the excitation current, and exits the advanced phase operation when necessary, thereby controlling the rate of change of the copper-iron temperature difference. This linkage control method maintains stable output power while suppressing the accumulation of thermal stress and reducing the risk of thermal deformation between the stator and rotor. Compared with the traditional passive protection with a fixed threshold, this method realizes active response control based on multi-parameter judgment, which can maintain the stability of the air gap and thermal balance under frequent and rapid peak-shaving conditions, delay insulation aging, and improve operational reliability.

[0018] 3. This invention periodically analyzes and updates the parameters of the temperature difference baseline model and early warning threshold by statistically analyzing operating data, early warning records, and maintenance conclusions. The model correction is based on the prediction deviation of historical samples and adjusts the parameter weights in combination with the error change trend, so that the model is consistent with the actual thermal characteristics of the unit. The continuously updated threshold parameters can reflect the real operating range under different cooling conditions and load characteristics, ensuring the long-term stability of expansion difference judgment and control logic. This mechanism forms a closed loop of monitoring, regulation and model correction, enabling the system to have continuous adaptability, reducing false alarms and delayed responses caused by environmental changes, and maintaining the safety and thermal balance of the generator under peak-shaving operation. Attached Figure Description

[0019] Figure 1 This is a flowchart of the generator operation data acquisition and synchronization processing of the present invention; Figure 2 This is a flowchart of the calculation and hierarchical early warning system for the copper-iron expansion difference characteristics of the present invention; Figure 3 This is a schematic diagram of the structure of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figures 1 to 2 As shown, this embodiment of the invention provides a method for directional monitoring and control of generator expansion difference under frequent, rapid and deep peak shaving conditions. In the specific implementation process, after the unit enters the peak shaving operation stage, the system first performs operation data acquisition and synchronous processing. Sensors arranged in the stator, rotor and cooling circuit collect operating parameters such as voltage, current, active power, reactive power, excitation current, excitation voltage, stator bar temperature, core temperature, cooling water inlet and outlet temperatures, cooling flow rate and hydrogen humidity in real time. The collected multi-source data are time-aligned with the unit's unified clock as the reference, and the moving average method and median filtering algorithm are used to correct the interference of the measurement points. Outliers are marked and removed. After standardization, a time series dataset of operating parameters with time index is formed, which provides accurate and traceable basic data for subsequent operating condition identification and model building. Through the above synchronization and calibration process, the consistency of data from different monitoring systems in terms of time scale and measurement accuracy is ensured, which improves the accuracy of subsequent expansion difference calculation and the reliability of data correlation analysis. In the stage of operating condition identification and temperature difference baseline establishment, the system extracts key parameters such as load power, excitation current and cooling flow from the calibrated dataset, and judges the current operating condition type by combining the unit dispatch records, distinguishing between full load, partial load, deep peak shaving, hot standby and turning gear status. Within the stable operating range, the system calculates the average temperature difference between the copper conductor and the iron core as the copper-iron temperature difference, calculates the temperature rise difference between the stator and rotor and the cooling water temperature difference efficiency, and establishes the temperature difference correspondence based on the load, current and reactive power at each operating point; the temperature difference baseline model obtained by least squares fitting is used to reflect the thermal balance characteristics of the unit under different loads, and provides a quantitative reference for subsequent expansion difference determination. This step enables the model to maintain its correspondence with the thermal behavior characteristics of the unit over a long period of time, providing a reliable benchmark for anomaly judgment. In the stage of copper-iron expansion difference characteristic calculation and extraction, the system calls the historical parameters in the temperature difference baseline model, combines the linear expansion coefficients of the conductor and the iron core to calculate the instantaneous expansion difference between the stator and rotor, and calculates the rate of change of the expansion difference over a continuous time period to form a time series expansion difference signal. Simultaneously, the monitoring system extracts signals such as vibration acceleration, excitation current waveform, and cooling water temperature. It calculates the vibration dominant frequency energy ratio through spectrum analysis, obtains the correlation coefficient between excitation current and reactive power, and calculates the variance of water temperature fluctuation. It constructs a multi-dimensional expansion feature set composed of thermal parameters, mechanical parameters, and electromagnetic parameters, providing a complete input for expansion status assessment. This feature fusion process realizes the physical coupling characterization of temperature, vibration, and electromagnetic signals, which helps to distinguish between normal thermal deformation and abnormal expansion phenomena. Based on a feature set, the system determines the deviation and rate of change of the copper-iron expansion difference relative to the temperature difference baseline. When any parameter exceeds a set threshold, the system automatically generates an expansion difference anomaly warning, which is divided into three levels according to the deviation magnitude and duration: prompt level, warning level, and severe level. The prompt level corresponds to slight fluctuations, the warning level represents continuous deviation, and the severe level corresponds to the risk of thermal expansion imbalance. The warning signals at each level are reported to the monitoring system in real time via the communication module, and the trigger time and corresponding characteristic parameters are recorded. This warning mechanism can realize graded response to different levels of anomalies, ensuring that monitoring personnel can take timely intervention measures. When the warning level reaches the alarm level or the critical level, the system automatically executes the linkage control strategy of operating parameters: first, the cooling water pump speed is increased to enhance the cooling flow and heat exchange effect; if the copper-iron expansion difference still exceeds the allowable range, the system further lowers the upper limit of the excitation current to suppress the rate of increase in stator bar temperature. During the adjustment process, the slip ring temperature and carbon brush current distribution are monitored in real time. When the slip ring temperature is detected to exceed the rated temperature rise limit of the unit or the current distribution deviation between carbon brushes exceeds the balance tolerance, the system issues a stop operation command to restore the excitation circuit to a safe state; all adjustment commands, feedback signals and threshold states are automatically recorded in the control log for subsequent trend correction and model optimization; this linkage control strategy realizes the coordinated response of thermal management and excitation control, effectively suppresses the expansion trend of expansion difference, and ensures the thermal safety of the unit during peak shaving. During operation and control, the system continuously collects partial discharge signals, insulation dielectric loss angle and hydrogen humidity data. When any two abnormal signals and the copper-iron expansion difference deviation occur simultaneously in time and the time overlap ratio exceeds the set threshold, it is determined to be a multi-source associated abnormal event and triggers a load reduction or shutdown command. This mechanism realizes the timely identification of composite anomalies through the physical coupling of three types of signals: electrical, thermal and environmental, effectively reducing the risk of missed detection by single parameter monitoring. During the model correction and parameter update phase, the system inputs operation control records, early warning logs, and maintenance conclusions into the database and compares the actual expansion difference value with the temperature difference baseline model prediction value. If the prediction error exceeds the preset range, the system corrects the model parameters and threshold settings according to the error direction and magnitude, and generates a new temperature difference baseline parameter set for subsequent operating cycles. This correction process can be executed periodically or automatically triggered when the error continues to exceed the limit, enabling the model to dynamically adapt to changes in unit status. By introducing an adaptive correction mechanism based on operational feedback, the system achieves closed-loop management from monitoring to judgment and from control to model optimization, ensuring that the expansion difference assessment always remains consistent with the actual thermal characteristics of the unit, thereby effectively improving the thermal stability and safety margin of generator operation under frequent, rapid, and deep peak-shaving conditions.

[0022] Among them, the operation data acquisition and synchronization processing module is used to centrally collect, uniformly synchronize and standardize multi-dimensional operation parameters involving electrical, thermal and environmental aspects during the period when the generator is in frequent start-stop and deep peak shaving operation, so as to provide a consistent data foundation for subsequent operating condition identification and temperature difference baseline model establishment. The generator operation monitoring system consists of an online generator monitoring device, a field inspection system, and a production management information system. The online generator monitoring device is installed at key nodes of the stator, rotor, and excitation system to collect real-time stator and rotor voltage, current, active power, reactive power, excitation current, and excitation voltage signals. These signals are collected by current transformers and voltage transformers located at the outgoing terminals and busbar side, and transmitted to the monitoring host after analog-to-digital conversion. The field inspection system uses temperature sensors located in the stator bars, core, and cooling water channels to periodically collect stator bar temperature, core temperature, and cooling water inflow and outflow data. The system collects temperature data with a sampling period of 30 seconds to 1 minute. Temperature sensors can be fiber optic thermometers or thermocouples. The collected signals are pre-processed by an edge acquisition module before being uploaded. The production management information system provides operating parameters for the cooling system, including cooling water flow rate, inlet and outlet water pressure difference, and hydrogen humidity. This system exchanges and synchronizes data with the online monitoring system via an industrial communication bus (such as Modbus or OPC-UA protocol). The multi-source monitoring system, formed by these three components, continuously outputs electromagnetic, thermal, and environmental information under different loads and start-up / shutdown conditions, providing a comprehensive input basis for unit operating status identification and data modeling. During periods of frequent load fluctuations, the sampling clocks of various monitoring systems in the unit may become offset. To maintain time synchronization, a unified time synchronization mechanism is adopted. Each monitoring system receives the standard time signal output by the unit control system through the time synchronization module. Using this signal as a time reference, the sampling data is timestamped and interpolated to ensure that different sampling channels have data correspondence at the same time. For outliers that may appear in the measurement point data, an anomaly identification strategy combining the moving average method and the median filtering method is adopted. For measurement point data that deviates from the moving average by more than the statistically determined anomaly identification multiple threshold, the system performs smoothing correction or rejection operations and performs linear interpolation with the nearest valid data. This process can effectively eliminate measurement abrupt changes caused by electromagnetic interference, sensor loosening, or transient fluctuations, ensuring the continuity and authenticity of the input data. After time synchronization and anomaly removal, the multi-source operational data is stored in chronological order as a structured time-series dataset. Each record contains a timestamp, measurement point number, parameter type, and measured value, and is updated at second or minute intervals. The dataset covers the entire process of generator startup, operation, peak shaving, and shutdown, providing a complete historical sample basis for temperature difference model calculation and expansion difference characteristic analysis.

[0023] Among them, the operating condition identification and temperature difference baseline establishment are used to identify the operating status of the generator under different load conditions based on the calibrated operating parameter time series, extract temperature difference feature data in the stable operating condition range, establish a temperature difference baseline model that reflects the thermal balance law of the generator, and provide a quantitative reference for subsequent expansion difference determination and operation control. The system first reads key parameters such as load power, excitation current, and cooling flow rate from the calibrated operating parameter time series, and constructs a time-load correspondence table by combining the load scheduling information in the unit operation plan. Based on the changing trends of load power and excitation current, the system calculates the rate of change in adjacent sampling periods and takes the moving average of the rate of change for multiple consecutive sampling points. When the average value is lower than the steady-state threshold set according to the fluctuation range of operating data, the unit is determined to be in a stable operating state. Based on the combined range of load power, excitation current, and cooling flow rate, the system classifies the unit's operating status into the following five typical operating conditions: full-load operation; partial-load operation; deep peak shaving operation; hot standby operation; and turning gear operation. The above operating condition classification rules are based on the range of unit operation control parameters and are consistent with the operation plan and scheduling records. They can be automatically indexed and retrieved in the database. By introducing a three-dimensional judgment logic of load power, excitation current change rate, and operation plan, the system can avoid the operating condition identification error caused by using only a single power threshold as a criterion. Once the system identifies a stable operating range, it calls temperature parameters to perform thermal characteristic calculations. First, it calculates the average difference between the copper conductor temperature and the iron core temperature to obtain the copper-iron temperature difference, which reflects the difference in thermal expansion between the stator and the rotor. Second, it calculates the average difference between the stator temperature rise and the rotor temperature rise to obtain the stator-rotor temperature rise difference. Finally, it uses the inlet and outlet temperatures of the cooling water and the cooling flow rate parameters to calculate the heat exchange per unit flow rate and obtain the cooling efficiency index to characterize the heat exchange capacity of the cooling system. The three sets of thermal characteristic data were indexed by the sampling timestamp and matched with the load power, excitation current and reactive power data at the corresponding time to form a multi-dimensional temperature difference sample set containing the time dimension. This temperature difference sample set reflects the thermal response law of the unit under different load and excitation conditions, and provides a statistical basis for the establishment of the temperature difference baseline model. After forming a temperature difference sample set, the system uses a least squares fitting algorithm to establish a multi-parameter correspondence model between temperature difference and load power, excitation current, and reactive power; referred to as the temperature difference baseline model. This model uses load power, current, and reactive power as input variables, and copper-iron temperature difference, stator-rotor temperature rise difference, and cooling efficiency as output variables. The weights of each parameter are obtained by minimizing the fitting residual. During the modeling process, the system performs group fitting according to different operating conditions, generating model subsets under full load, partial load, and deep peak shaving, and records the applicable range and fitting coefficients of each subset to enhance the adaptability of the model under different operating conditions. During system operation, the temperature difference baseline model can be compared with real-time monitoring data to identify thermal characteristic drift and long-term trend changes, providing a benchmark reference for subsequent expansion difference calculation and anomaly early warning; It forms a complete data chain with the operation data acquisition and a logical closed loop with the expansion difference characteristic calculation module. By introducing multi-parameter joint fitting and sub-condition modeling, the model can maintain high fitting accuracy under different power output and cooling conditions. This technical approach can reduce the temperature difference baseline drift problem caused by frequent peak shaving, so that the thermal characteristic assessment no longer depends on the static curve of a single load point, thereby improving the stability and accuracy of subsequent expansion difference determination.

[0024] Among them, the calculation and extraction of copper-iron expansion difference characteristics are used to calculate the instantaneous expansion difference between the stator and rotor after establishing the temperature difference baseline model, combined with the unit operating temperature data and the thermal expansion characteristics of conductor and iron core materials, and simultaneously extract multi-source operating signal characteristics to form a quantifiable expansion difference characteristic parameter set, providing input basis for subsequent status assessment and early warning control. When the generator is in deep peak shaving or rapid load change conditions, the stator and rotor are affected by temperature fluctuations, and their thermal expansion difference changes significantly. The system calls up the copper conductor temperature and iron core temperature data corresponding to the current time period, aligns them according to the temperature difference baseline model, and calculates the thermal expansion difference between the two under a unit temperature rise by combining the linear expansion coefficients of the conductor and iron core materials, thus obtaining the instantaneous expansion difference between the stator and rotor. The system uses the initial structural gap between the stator and rotor as a reference, and obtains the theoretical expansion difference value at the current moment by superimposing the expansion difference caused by temperature. The expansion difference results obtained by continuous sampling constitute time series data, and by calculating the rate of change of adjacent sampling points, a signal sequence reflecting the trend of expansion difference change over time is formed. This signal sequence serves as the thermodynamic input for subsequent multi-source feature synchronization and deviation analysis, realizing the continuous calculation and dynamic tracking of the copper-iron expansion difference inside the generator. To ensure the accuracy of the differential expansion analysis, the system acquires vibration, electromagnetic, and cooling parameter signals under the same time reference through a multi-channel synchronous acquisition device. The vibration acceleration signal is acquired by a triaxial accelerometer arranged at the bearing housing and the end of the machine base, with a sampling frequency of not less than 1kHz; the excitation current waveform signal is provided by the current detection module of the excitation control system; and the cooling water temperature is detected by a thermocouple sensor installed at the outlet of the cooling pipe. All signals are timestamped and the delay is calibrated by the time synchronization module, so that the vibration, electromagnetic, and cooling channels are sampled and aligned in the time dimension. This synchronization mechanism ensures the comparability and consistency of each signal in the subsequent feature calculation. The system performs frequency domain and statistical feature extraction on the synchronously acquired signals: Vibration feature extraction: The power spectral density distribution is obtained by performing a Fast Fourier Transform (FFT) on the vibration signal, and the main frequency energy ratio (the ratio of main frequency energy to total energy) is calculated to reflect the energy concentration of the mechanical system. When rotor eccentricity or air gap changes cause electromagnetic force imbalance, this index will fluctuate significantly. Excitation-power correlation extraction: Pearson correlation analysis is performed on the excitation current waveform and reactive power sequence to obtain the correlation coefficient between excitation current and reactive power, which reflects the electromagnetic coupling characteristics. When the expansion difference between copper and iron causes a shift in magnetic flux distribution, this coefficient decreases or shows an inverse phase change. Cooling stability feature extraction: The variance of the cooling water outlet temperature in adjacent sampling periods is calculated to obtain the temperature difference stability index, which reflects the temperature uniformity of the cooling system under transient thermal disturbances. The feature parameters are matched one-to-one with the expansion difference signal sequence according to the sampling timestamp to form a structured expansion difference feature parameter set, referred to as the feature parameter set. This parameter set contains multi-dimensional vector terms, including the main frequency energy ratio, excitation correlation coefficient, and temperature difference stability index. The feature parameter set is managed through a unified index table and is associated with the temperature difference baseline model and the original operating data. The system generates a complete record for each sampling period, which includes: expansion difference, rate of change and corresponding multi-source feature parameters. All records are stored as a time-continuous data sequence for subsequent state assessment and early warning model input, and can support anomaly identification and trend analysis. Compared with existing generator expansion difference monitoring methods, this invention not only calculates the expansion difference value based on temperature difference, but also integrates vibration, electromagnetic and cooling system signals, and realizes cross-physical domain coupled analysis through time synchronization and feature matching; by establishing the correspondence between the expansion difference change rate and multi-source features, this method realizes the directional identification of mechanical, electromagnetic and cooling anomalies caused by thermal stress, so as to accurately locate the expansion difference deviation in the early stage of the problem.

[0025] Furthermore, all data in this solution originates from the existing monitoring system signals of the generator, without relying on additional sensor devices. It can be directly embedded into the operational data acquisition system, demonstrating strong feasibility and real-time performance.

[0026] Among them, based on the aforementioned expansion difference characteristic parameter set and temperature difference baseline model, the operating status of the generator copper-iron expansion difference is quantitatively evaluated, and graded early warnings are output according to the degree of deviation and duration, providing input for subsequent linkage control of operating parameters and model correction; The system reads the measured expansion difference value from the expansion difference characteristic parameter set at each sampling time, and obtains the baseline expansion difference at the same time according to the temperature difference baseline model. The difference between the two is recorded as the deviation, which is used to reflect the static difference between the current state and the baseline state. To describe the rate of change of deviation, the system calculates the rate of change between two adjacent sampling times. That is, it uses the difference between the current deviation and the previous deviation, and normalizes it according to the sampling time interval to obtain a velocity dimension index. In order to reduce the influence of occasional noise, the deviation and the rate of change are both written into the database after time smoothing and stored together with the corresponding working condition label to form an evaluation parameter set. The system sets two independent thresholds: a deviation threshold for static difference judgment and a change rate threshold for dynamic trend judgment. When either indicator exceeds its corresponding threshold, an expansion abnormality event is triggered. To avoid misjudgment caused by short-term disturbances, the system uses an confirmation period for validity confirmation: the event is only confirmed as a valid abnormality if the triggering condition remains true for a consecutive number of sampling periods (e.g., 3 to 5 consecutive periods). The length of the confirmation period is configured according to the unit's sampling period and historical fluctuation characteristics and can be adjusted during operation and maintenance. For confirmed abnormal events, the system classifies them according to the magnitude and duration of deviation, and supplements this with a check of the rate of change, forming a three-level early warning system: Warning Level: When the deviation is in the first warning range and the duration is shorter than the set upper limit, the system will prompt you to pay attention to the trend and record the event on the monitoring interface; Warning level: When the deviation enters the second warning range, or the rate of change exceeds the corresponding dynamic threshold, the system records a snapshot of the operating parameters and enters the preparatory logic for cooling and excitation control. Severity level: When the deviation enters the highest warning range and the duration reaches or exceeds the corresponding threshold, the system immediately generates an alarm message and sends a protection command suggestion to the control system (the specific action is executed by the linkage control of the operating parameters); the above range threshold and duration threshold are configured based on the unit's historical operating data and typical load fluctuation characteristics, and are stored in the parameter table; during model correction and parameter updates, dynamic correction can be made based on operating feedback; When an early warning is generated, the system writes the warning level, trigger time, threshold range used, current deviation and rate of change, duration of deviation, operating condition label, and corresponding expansion feature vector into the monitoring alarm channel, and sends the early warning data packet to the control system through the communication interface; at the same time, it establishes standardized records in the event database for subsequent linkage control and model correction, realizing a closed loop of monitoring-judgment-control-correction. Among them, based on the warning results of the aforementioned state assessment and graded early warning output, the operating parameters of the generator are adaptively linked and controlled to suppress the abnormal expansion of copper and iron expansion under frequent and rapid peak shaving conditions, reduce the risk of thermal stress accumulation, and ensure the safe and stable operation of the unit. When the warning level is determined to be warning or severe, the system initiates linkage control of operating parameters. The control system adjusts the cooling system first, followed by the excitation system. This order is set based on the characteristics of high response speed of the cooling medium and large thermal inertia of the excitation system. By prioritizing the improvement of cooling capacity, the temperature rise of the stator, rotor and iron core can be reduced in a short time. If the copper-iron expansion difference still exceeds the allowable range specified in the operation after cooling adjustment, the intensity of the internal heat source is further reduced by adjusting the electromagnetic parameters of the excitation system, thereby achieving coordinated control of temperature and power through two paths. The system first detects the real-time speed, flow rate, and temperature difference between the inlet and outlet of the cooling water pump to determine if it is within the rated range. When there is room for improvement in cooling capacity, the control system automatically increases the pump speed and the flow rate of the cooling medium to improve heat exchange efficiency and reduce the temperature difference between copper and iron. When the cooling water pump speed has reached the set upper limit and the expansion difference has not returned to the allowable range, the system determines that the cooling adjustment is insufficient and automatically switches to the excitation system adjustment stage to ensure that physical cooling control is completed while maintaining the stability of the main circuit. After entering the excitation regulation stage, the system gradually reduces the upper limit of the excitation current, thereby reducing the magnetic field strength and reducing the rate of heat generation in the iron core and stator bars. The range of excitation current reduction is automatically calculated based on the real-time operating load and scheduling strategy to ensure that the voltage, power factor and reactive power distribution remain within a safe range. During this process, the system continuously monitors the output power and voltage fluctuations to prevent power quality instability caused by excitation adjustment. Throughout the adjustment process, the system monitors the slip ring temperature and carbon brush current distribution in real time. When the slip ring temperature exceeds the set safety value or the carbon brush current distribution imbalance exceeds the set threshold, the system immediately issues an exit phase-leading control command, causing the excitation system to exit the phase-leading mode and return to normal power factor control, thereby limiting circuit heat accumulation and preventing insulation material aging. This protection mechanism realizes multi-parameter linkage of temperature signal, current distribution and excitation status. After each parameter adjustment is executed, the system records the adjustment command, execution time, corresponding warning level, key operating parameters before and after the adjustment, and monitoring results, and forms a standardized control log. After being organized, the log is entered into the database to provide a basis for subsequent multi-source anomaly correlation judgment and handling, as well as model correction and parameter update. Through the accumulation of log data, the temperature difference baseline model and control parameters can be corrected in long-term operation, so that the control strategy is more in line with the actual thermal characteristics of the unit.

[0027] In the process of executing the linkage control of operating parameters, the multi-source monitoring signals of the generator insulation status and cooling status are synchronously collected and jointly analyzed to identify the correlation anomaly patterns between different monitoring quantities, and to prevent the systemic risks caused by the combined effect of abnormal copper and iron expansion and factors such as insulation aging and cooling performance degradation; this process realizes the synchronous collection, correlation judgment and control of multi-dimensional abnormal data. During generator peak-shaving operation, the system synchronously collects the following signals at a fixed sampling period (e.g., 1-2 seconds): Partial discharge signal: collected by high-frequency current sensors arranged at the ends of the stator winding and the iron core, used to reflect the discharge activity of the insulation system; Insulation loss angle data: measured by an online insulation monitoring device, used to characterize the polarization loss of the insulation material; Cooling system hydrogen humidity data: collected by the humidity sensor of the hydrogen circulation system, used to reflect changes in the water content and purity of the cooling medium; The collected data is synchronized with a unified clock, outlier removal is performed and time series calibration is carried out to form a time-consistent multi-source monitoring dataset for subsequent anomaly determination. The system sets thresholds for each monitored quantity, including: The discharge energy threshold of partial discharge signals; The upper limit of tanδ for the dielectric loss angle; The water content threshold of hydrogen humidity; Abnormal threshold of temperature difference baseline deviation from copper-iron expansion difference; When the values ​​of any two monitored quantities exceed their thresholds within a preset observation window (e.g., 60 seconds), and the deviation of the copper-iron expansion difference also exceeds the abnormal threshold of the temperature difference baseline during this time period, the system initially determines it to be a multi-source abnormal event. Subsequently, the system calculates the ratio of the overlap time of the abnormal signal within the observation window to the total duration of the window, which is defined as the time overlap rate. When the time overlap rate reaches a preset threshold (e.g., 50% to 70%) determined by combining historical operating data and unit design parameters, the system confirms the existence of a multi-source associated anomaly and enters the handling logic. The system automatically executes different handling measures based on the level of abnormality: When the anomaly level is low (two monitored values ​​are slightly out of range and the duration is short), a load reduction command is output to reduce the heat load; When the abnormality level is high (three parameters exceed the limit or the time overlap rate exceeds the upper limit), a shutdown control signal is output, and the main excitation circuit is cut off through the control logic to ensure the safe shutdown of the unit. After the handling is completed, the system records the measured values ​​of each abnormal parameter, the trigger time, the duration, the time overlap rate, the type of action executed, and the execution result, generating a multi-source abnormal event log to provide data support for subsequent model correction and threshold updates; This process maintains real-time linkage with the aforementioned operating parameter linkage control; when the cooling or excitation adjustment action has been triggered but discharge or humidity abnormality still occurs, the system identifies the linkage trend between thermal abnormality and insulation performance degradation through a multi-signal time correlation mechanism, thereby executing protection actions in advance before thermal stress accumulates. The innovative features of this application include a multi-signal joint judgment mechanism: under peak shaving conditions, the deviation of copper-iron expansion difference, partial discharge, dielectric loss angle and hydrogen humidity signals are combined for anomaly identification, forming a comprehensive monitoring framework across thermal, electrical and dielectric dimensions; Time overlap rate quantification standard: Abnormal synchronization is quantified by calculating the signal time overlap ratio, avoiding the false judgment of alarm when both exceed the limit simultaneously using traditional methods; Threshold dynamic correction closed loop: Combined with the feedback from the model correction module, the overlap rate and threshold range are dynamically updated to enable the system to adapt to long-term operation under different unit and operating conditions.

[0028] After completing the multi-source anomaly correlation judgment and handling, the system periodically corrects and updates the temperature difference baseline model and its related threshold parameters based on the control data, early warning records and maintenance conclusions generated during operation; by comparing the deviation between the model prediction results and the actual measurement results, the model parameters are quantitatively corrected so that the expansion difference judgment and control logic continuously conform to the actual operating characteristics of the unit. Construction of historical sample data: During the long-term operation of the generator, the system continuously inputs the following data as historical samples into the database: Control result data: including cooling system speed, excitation current adjustment range and corresponding copper-iron expansion difference response results; Warning record data: including the trigger time, deviation, rate of change, duration and corresponding control actions for each level of warning; Maintenance conclusion data includes: changes in stator-rotor clearance recorded during maintenance, evaluation results of insulation medium performance, and status of cooling system; Each data entry is accompanied by a timestamp, operating condition identifier, and data source identifier, and is archived in the order of collection to form a historical sample dataset with temporal continuity and traceability of source.

[0029] The system calls the historical prediction output of the temperature difference baseline model and compares it with the measured value of copper-iron expansion difference in the same period. It calculates the numerical difference between the two and expresses the model prediction error as an absolute deviation or relative deviation percentage to evaluate the model's adaptability under different working conditions. To ensure the stability of the calculation results, the system first performs a time window-based moving average and outlier removal on the sample data to eliminate the interference of random fluctuations and abnormal measurement points. The processed data is used for error statistics as the basis for subsequent parameter correction. When the prediction error exceeds a preset threshold (e.g., 3%–5%), the system enters the model correction phase. The correction process includes: Parameter scaling adjustment: Based on the direction (overestimation or underestimation) and magnitude of the error, the coefficients corresponding to the input parameters such as load power, excitation current, and cooling flow rate in the temperature difference baseline model are scaled up to gradually make the model output consistent with the measured data. Threshold range synchronous correction: Based on the corrected model output, recalculate the upper and lower limits of the warning threshold under different working conditions to ensure that the new threshold range reflects the normal fluctuation range of the corrected model. Abnormal sample weighting: For samples identified as multi-source abnormal events, the system reduces their participation ratio in subsequent correction calculations to reduce the impact of abnormal samples on model parameter fitting. After the correction is completed, the system generates a new set of temperature difference baseline parameters and stores them in the database in the form of a version number. At the same time, it records the correction time, data range and adjustment range to ensure the traceability of the parameter change process. The corrected model parameters serve as the new basis for judgment in the next operating cycle. When new operating data is collected, the system loads the current version of the model according to the preset calling logic to perform expansion difference calculation and early warning judgment. If the prediction deviation exceeds the threshold again, the system re-executes the model correction process, forming a periodic parameter update closed loop. At the same time, the aforementioned operating parameter linkage control and multi-source anomaly judgment synchronously call the latest threshold parameters to keep the control response consistent with the model state, thereby realizing continuous closed-loop operation of monitoring-control-correction. This mechanism reduces the judgment deviation caused by environmental changes, equipment aging or load fluctuations and maintains the long-term stability of the system's judgment. Compared with existing temperature difference models based on fixed thresholds, the innovations of this implementation method are mainly reflected in the following aspects: A dynamic correction mechanism based on closed-loop feedback: By combining control data, early warning records, and maintenance conclusions, the model parameters are quantitatively corrected to improve model accuracy; a two-dimensional error correction logic: The direction and magnitude of the error are adjusted to make the model more adaptable to changes in operating conditions; collaborative updating of the model and thresholds: By synchronously correcting parameters and alarm thresholds, the judgment criteria are kept consistent with the model status to avoid false alarms; versioned tracking and data verifiability: Each correction generates an independent version and log, forming an auditable data chain to support the long-term safe traceability management of the unit.

[0030] Furthermore, such as Figure 3As shown, the generator expansion difference directional monitoring and control system 10 under frequent, rapid and deep peak shaving conditions includes: an operation data acquisition and synchronization processing module 100, an operation condition identification and temperature difference baseline establishment module 200, a copper-iron expansion difference characteristic calculation and extraction module 300, a status assessment and graded early warning module 400, an operation parameter linkage control module 500, a multi-source anomaly association judgment and handling module 600, and a model correction and parameter update module 700.

[0031] Specifically, the data acquisition and synchronization processing module 100 is used to collect electrical quantities, temperature quantities and cooling system parameters of the generator stator and rotor, and to perform time synchronization, anomaly rejection and standardization based on the unit's unified clock to form a calibrated sequence of operating parameters.

[0032] The operating condition identification and temperature difference baseline establishment module 200 is used to identify the unit's operating status based on the load power, excitation current and cooling flow rate in the operating parameter sequence, extract the copper-iron temperature difference and cooling efficiency characteristics under stable operating conditions, and establish a correspondence model between temperature difference and load, current and reactive power to generate a temperature difference baseline model.

[0033] The copper-iron expansion difference characteristic calculation and extraction module 300 is used to calculate the expansion difference between the stator and rotor based on the temperature difference baseline model and the thermal expansion characteristics of the conductor and iron core materials, and to form a set of expansion difference characteristic parameters by combining the vibration signal, excitation current waveform and cooling temperature.

[0034] The status assessment and graded early warning module 400 is used to calculate the deviation magnitude and rate of change based on the expansion difference characteristic parameter set, and generate a graded early warning signal when it exceeds the set threshold.

[0035] The operating parameter linkage control module 500 is used to adjust the cooling flow rate and the upper limit of the excitation current in priority order when the warning level of the graded warning signal reaches the warning level or the severe level, and to perform the exit phase advance operation when abnormal temperature rise or uneven current distribution is detected.

[0036] The multi-source anomaly correlation judgment and handling module 600 is used to comprehensively analyze partial discharge signals, dielectric loss angle and hydrogen humidity. When multiple anomalies and expansion difference deviations occur simultaneously and the time overlap rate exceeds the threshold, it outputs a load reduction or shutdown signal.

[0037] The model correction and parameter update module 700 is used to correct the temperature difference baseline and early warning threshold based on the load reduction or shutdown signal and the operation and maintenance data. When the prediction error exceeds the limit, the model parameters are automatically updated to achieve dynamic optimization of the judgment and control threshold.

[0038] Furthermore, the data acquisition and synchronization processing module 100 includes: a data acquisition unit, a data provision unit, a relationship establishment unit, and a sequence storage unit.

[0039] The data acquisition unit is used to collect operating parameter data from the generator online monitoring device, the on-site inspection system, and the production management information system.

[0040] The data providing unit is used by the online monitoring device to collect stator and rotor voltage, current, excitation current, excitation voltage, active power and reactive power signals; the point inspection system collects stator bar temperature, core temperature and cooling water inlet and outlet temperature; and the production management information system provides cooling flow rate and hydrogen humidity data.

[0041] The relationship establishment unit is used to align multi-source data with the unit's unified clock, remove outlier values ​​at measurement points through moving average filtering, and establish cross-system data correspondence based on the measurement point identifier.

[0042] The sequence storage unit is used to store the calibrated data in chronological order as a sequence of operating parameters. The sequence of operating parameters includes timestamps, measurement point numbers and corresponding measured values, which serve as inputs for subsequent operating condition identification and temperature difference baseline establishment.

[0043] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for directional monitoring and control of generator expansion difference under frequent, rapid, and deep peak-shaving conditions, characterized in that, include: Operational data acquisition and synchronization processing: Collect electrical quantities, temperature quantities, and cooling system parameters of the generator stator and rotor, and perform time synchronization, anomaly elimination, and standardization based on the unit's unified clock to form a calibrated sequence of operating parameters; Operating condition identification and temperature difference baseline establishment: The unit operating status is identified based on the load power, excitation current and cooling flow rate in the operating parameter sequence. Under stable operating conditions, the copper-iron temperature difference and cooling efficiency characteristics are extracted, and a correspondence model between temperature difference and load, current and reactive power is established to generate a temperature difference baseline model. Calculation and extraction of copper-iron expansion difference characteristics: Based on the temperature difference baseline model and the thermal expansion characteristics of conductor and iron core materials, the expansion difference between the stator and rotor is calculated, and the expansion difference characteristic parameter set is formed by combining the vibration signal, excitation current waveform and cooling temperature; Status assessment and graded early warning: Based on the set of expansion difference characteristic parameters, the deviation magnitude and rate of change are calculated, and a graded early warning signal is generated when the deviation exceeds the set threshold. Operation parameter linkage control: When the warning level of the graded early warning signal reaches the warning level or the severe level, the cooling flow rate and the upper limit of the excitation current are adjusted in priority order, and the exit phase advance operation is performed when abnormal temperature rise or uneven current distribution is detected. Multi-source anomaly correlation judgment and handling: Comprehensive analysis of partial discharge signal, dielectric loss angle and hydrogen humidity. When multiple anomalies and expansion difference deviation occur simultaneously and the time overlap rate exceeds the threshold, output load reduction or shutdown signal. Model correction and parameter update: Based on the load reduction or shutdown signal, the operation and maintenance data are used to correct the temperature difference baseline and early warning threshold. When the prediction error exceeds the limit, the model parameters are automatically updated to achieve dynamic optimization of the judgment and control threshold.

2. The generator expansion differential directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions as described in claim 1, characterized in that, The operational data acquisition and synchronization processing includes: Operating parameter data are collected from the generator online monitoring device, the on-site inspection system, and the production management information system. The online monitoring device collects stator and rotor voltage, current, excitation current, excitation voltage, active power, and reactive power signals; the point inspection system collects stator bar temperature, core temperature, and cooling water inlet and outlet temperatures; and the production management information system provides cooling flow rate and hydrogen humidity data. The multi-source data is time-aligned with the unit's unified clock, outlier values ​​at the measurement points are removed by moving average filtering, and cross-system data correspondence is established based on the measurement point identification number. The calibrated data is stored in chronological order as a sequence of operating parameters, which includes timestamps, measurement point numbers, and corresponding measured values, serving as input for subsequent operating condition identification and temperature difference baseline establishment.

3. The generator expansion difference directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions as described in claim 1, characterized in that, The operating condition identification and temperature difference baseline establishment include: Load power, excitation current and cooling flow parameters are extracted from the calibrated operating parameter sequence. Based on the changing trends of load power and excitation current, and combined with the load scheduling information recorded in the unit operation plan, it is determined whether the unit is currently in full load, partial load, deep peak shaving, hot standby or turning gear state. During the time period determined to be in a stable operating condition, the average temperature difference between the copper conductor and the iron core is calculated as the copper-iron temperature difference. The temperature rise difference between the stator and the rotor is calculated, and the cooling efficiency is calculated based on the temperature difference between the inlet and outlet of the cooling water. The copper-iron temperature difference, the stator-rotor temperature rise difference, and the cooling efficiency are paired with the corresponding load power, excitation current, and reactive power data to form a multi-dimensional temperature difference sample set with a time index. Based on the multi-dimensional temperature difference sample set, a multi-parameter correspondence model between temperature difference and load, current, and reactive power is established using least squares fitting. The temperature difference baseline model is then generated as reference data for subsequent copper-iron expansion difference calculation and determination.

4. The generator expansion differential directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions as described in claim 1, characterized in that, The calculation and extraction of the copper-iron expansion difference characteristics include: Based on the aforementioned temperature difference baseline model, the copper conductor temperature and iron core temperature data for the corresponding time period are called up. The instantaneous expansion difference between the stator and rotor is calculated by combining the linear expansion coefficients of the conductor and iron core materials. The rate of change of the continuously calculated expansion difference is obtained according to the time series to form a signal sequence of expansion difference changing with time. Furthermore, the vibration acceleration signal, excitation current waveform signal, and cooling water temperature data are acquired within the same time period through the synchronized acquisition channel. The main frequency energy ratio is calculated based on the spectral power distribution of the vibration signal. The correlation coefficient between the excitation current waveform and the reactive power data is calculated. The temperature difference stability index is calculated based on the variance of the cooling water temperature fluctuation. The main frequency energy ratio, the correlation coefficient between the excitation current and the reactive power, and the temperature difference stability index are matched with the expansion difference change signal sequence according to the time correspondence to form the expansion difference characteristic parameter set, which serves as the input data for subsequent expansion difference state assessment.

5. The generator expansion differential directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions according to claim 1, characterized in that, The status assessment and graded early warning include: Based on the expansion difference characteristic parameter set, the deviation of the copper-iron expansion difference from the temperature difference baseline model is calculated at each sampling time, and the rate of change of the deviation is obtained in adjacent sampling periods. When either the deviation or the rate of change exceeds a preset threshold, it is determined to be an expansion difference abnormal event, and a corresponding early warning signal is generated and sent to the monitoring system. Based on the numerical range of the deviation and the duration of the deviation, the warning signal is divided into three levels: prompt level, warning level, and severe level. The warning signals of different levels are sent to the alarm channel of the monitoring system, and the trigger time, threshold and corresponding expansion difference parameters are recorded to provide input data for subsequent operation control and model correction.

6. The generator expansion differential directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions according to claim 1, characterized in that, The linkage control of the operating parameters includes: When the warning level is determined to be a warning level or a severe level, the control system adjusts the unit operating parameters according to a preset control sequence that prioritizes the cooling system and then the excitation system. When making adjustments, first increase the speed of the cooling water pump to increase the cooling water flow. When the speed of the cooling water pump reaches the upper limit and the copper-iron expansion difference deviates from the allowable range specified for unit operation, further reduce the upper limit of the excitation current to slow down the rate of change of the copper-iron temperature difference. During parameter adjustment, the slip ring temperature and carbon brush current distribution status are continuously monitored. When the slip ring temperature exceeds the specified value or the carbon brush current distribution is uneven, the exit phase control command is executed to limit the heat accumulation in the excitation circuit. Each parameter adjustment command, execution time, corresponding warning level, and monitoring result is recorded in the control log as input for subsequent anomaly detection and model correction.

7. The generator expansion differential directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions according to claim 1, characterized in that, The multi-source anomaly correlation determination and handling includes: During the execution of the linkage control of operating parameters, the operating data of partial discharge signal, insulation dielectric loss angle and hydrogen humidity of cooling system are continuously collected and monitored at a fixed sampling period. When the values ​​of any two of the partial discharge signal, insulation dielectric loss angle and hydrogen humidity of cooling system exceed their respective set abnormal thresholds within the preset observation window, and the deviation of copper-iron expansion difference exceeds the abnormal judgment threshold set by the temperature difference baseline model at the same time, it is judged as a multi-source abnormal event. Calculate the ratio of the overlap duration of the occurrence time of each abnormal parameter to the duration of the observation window, and use this ratio as the time overlap rate. When the time overlap rate reaches a preset threshold determined by historical operation data, it is confirmed that there is a multi-source correlation anomaly. Output load reduction commands or shutdown control signals to the unit control system, and record the values ​​of each abnormal parameter, the trigger time, and the execution results as input data for subsequent model correction and parameter updates.

8. The generator expansion difference directional monitoring and control method under frequent, rapid, and deep peak-shaving conditions according to claim 1, characterized in that, The model correction and parameter update include: The control results, early warning records, and maintenance conclusions during operation are input into the database as historical sample data; Based on the difference between the actual expansion difference value recorded in the historical sample data and the temperature difference baseline model prediction value, the absolute deviation or percentage deviation between the two is calculated as the model prediction error. When the model prediction error exceeds the preset error threshold, the parameter weights of the temperature difference baseline model are adjusted according to the error change trend and magnitude, and the warning threshold parameters are corrected simultaneously. A new set of temperature difference baseline parameters and upper and lower limits of thresholds under each operating condition are generated using the updated model parameters, and the updated results are stored in the database for subsequent expansion difference determination and control.

9. A generator expansion differential directional monitoring and control device under frequent, rapid, and deep peak-shaving conditions, characterized in that, include: The operation data acquisition and synchronization processing module is used to collect electrical quantities, temperature quantities and cooling system parameters of the generator stator and rotor, and perform time synchronization, anomaly rejection and standardization based on the unit's unified clock to form a calibrated sequence of operating parameters. The operating condition identification and temperature difference baseline establishment module is used to identify the unit's operating status based on the load power, excitation current and cooling flow rate in the operating parameter sequence, extract the copper-iron temperature difference and cooling efficiency characteristics under stable operating conditions, establish a correspondence model between temperature difference and load, current and reactive power, so as to generate a temperature difference baseline model. The copper-iron expansion difference characteristic calculation and extraction module is used to calculate the expansion difference between the stator and rotor based on the temperature difference baseline model and the thermal expansion characteristics of the conductor and iron core materials, and to form a set of expansion difference characteristic parameters by combining the vibration signal, excitation current waveform and cooling temperature. The status assessment and graded early warning module is used to calculate the deviation magnitude and rate of change based on the set of expansion difference characteristic parameters, and generate a graded early warning signal when it exceeds a set threshold. The operating parameter linkage control module is used to adjust the cooling flow rate and the upper limit of the excitation current in priority order when the warning level of the graded warning signal reaches the warning level or the severe level, and to perform the exit phase advance operation when abnormal temperature rise or uneven current distribution is detected. The multi-source anomaly correlation judgment and handling module is used to comprehensively analyze partial discharge signals, dielectric loss angle and hydrogen humidity. When multiple anomalies and expansion difference deviations occur simultaneously and the time overlap rate exceeds the threshold, it outputs a load reduction or shutdown signal. The model correction and parameter update module is used to correct the temperature difference baseline and early warning threshold based on the load reduction or shutdown signal and the operation and maintenance data. When the prediction error exceeds the limit, the model parameters are automatically updated to achieve dynamic optimization of the judgment and control threshold.

10. The generator expansion difference directional monitoring and control device under frequent, rapid, and deep peak-shaving conditions according to claim 9, characterized in that, The operational data acquisition and synchronization processing module includes: The data acquisition unit is used to collect operating parameter data from the generator online monitoring device, the field inspection system and the production management information system respectively. The data providing unit is used for the online monitoring device to collect stator and rotor voltage, current, excitation current, excitation voltage, active power and reactive power signals; the point inspection system collects stator bar temperature, core temperature and cooling water inlet and outlet temperature; and the production management information system provides cooling flow rate and hydrogen humidity data. The relationship establishment unit is used to align multi-source data with the unit's unified clock, remove outlier values ​​at measurement points through moving average filtering, and establish cross-system data correspondence based on the measurement point identification number. The sequence storage unit is used to store the calibrated data in chronological order as a sequence of operating parameters. The sequence of operating parameters includes timestamps, measurement point numbers, and corresponding measured values, which serve as inputs for subsequent operating condition identification and temperature difference baseline establishment.

Images