A water supply pre-regulation method and system for pumped storage power station technology

Abstract

This application relates to a pre-regulation method for water supply in pumped storage power stations. This application is applicable to the field of intelligent operation and maintenance technology for pumped storage units. The technical solution is as follows: First, historical operating data of the unit and water supply system are collected, covering operating conditions, temperature of the object being cooled, cooling flow rate, and environmental parameters; then, key variables strongly correlated with temperature are extracted using a feature filtering algorithm, and the optimal delay time is determined using mutual information; next, a temperature prediction model is constructed and trained, with time-series data of key variables as input and delayed temperature data as output; simultaneously, a mapping relationship model between branch cooling flow rate and temperature changes is established. Real-time operating data is collected and input into the model to predict the temperature of the object being cooled; based on the deviation between the predicted temperature and the preset operating range, the branch flow rate adjustment amount and target flow rate are determined in conjunction with the mapping relationship; a pre-regulation command is issued to the actuator, and the temperature and actual flow rate are monitored in real time and dynamically corrected to stabilize the temperature of the object being cooled within the operating range.

Description

Technical Field

[0001] This invention relates to a method for pre-regulating the water supply of a pumped storage power station. It is applicable to the field of intelligent operation and maintenance of pumped storage units. Background Technology

[0002] In recent years, as intermittent new energy sources such as wind power and photovoltaics have gradually become the mainstay of new power systems, pumped storage power stations, as an excellent energy storage resource, have also entered a new era of large-scale construction. At the same time, their safe, stable, and efficient operation has received increasing attention from the industry. The technical water supply system is one of the important auxiliary systems of pumped storage units. It is designed according to the specific operating conditions of each water-using equipment. The main technical water-using equipment and components of the power station include the circulating oil coolers for each bearing, the main shaft seal, the air cooler, the upper / lower labyrinth ring cooling of the runner, and the main transformer load cooling. The unit's technical water supply system draws water from the tailrace channel, pressurizes it through the water supply pump, and then exchanges heat with the medium in the generator motor air cooler, each bearing oil cooler, and the main transformer oil cooler to achieve a cooling effect.

[0003] Traditionally, technical water supply pumps have generally been fixed-frequency pumps, operating primarily according to design conditions. They lack the ability to dynamically adjust water volume, relying instead on valves in branch pipelines for localized flow regulation. However, the valve flow adjustment range is very limited and is mostly manual. Consequently, current technical water supply systems operate at the maximum demand required by users after commissioning, with minimal additional adjustments. This results in excessive redundancy in cooling water volume under most operating conditions.

[0004] Furthermore, the cooling water in the technical water supply system is typically used as a secondary cooling medium, cooling equipment such as thrust bearings and guide bearings by cooling primary cooling media such as lubricating oil and air. However, primary cooling media have unique operating characteristics. For example, the lubricating oil in thrust bearings and guide bearings exhibits different viscosity at lower temperatures, increasing flow resistance and bearing churning resistance, thus reducing unit efficiency. Conversely, higher oil temperatures result in lower viscosity, making it difficult for the lubricating oil to form a sufficiently thick oil film, potentially leading to direct contact between the bearing bush and rotating parts, affecting unit stability and safe operation. Therefore, it is necessary to predict the water demand of each cooling object and pre-adjust based on heat conduction delay. This allows for proactive control of the water flow in each technical water supply branch pipe in response to future temperature changes, thereby reducing the magnitude of future temperature fluctuations and minimizing temperature fluctuations in cooling equipment such as thrust bearings, ensuring the unit operates within its high-efficiency range as much as possible.

[0005] In existing technologies, water supply regulation largely relies on real-time temperature feedback or cooling water volume prediction. For example, the cooling water volume prediction method based on feature optimization and LSTM disclosed in CN117909799A can only achieve accurate water volume prediction and provide a control reference, but it does not involve temperature prediction and advance adjustment logic. CN117666339A achieves intelligent regulation through temperature prediction and flow mapping, but it still relies on the current temperature for real-time response and does not design a pre-adjustment mechanism to address the heat conduction delay problem. Existing technologies generally lack closed-loop control logic of "temperature prediction-advance adjustment," which cannot fundamentally solve the problem of temperature fluctuations caused by changes in operating conditions and makes it difficult to ensure that the unit always operates in the high-efficiency range under all operating conditions.

[0006] Therefore, there is an urgent need for a technical water supply pre-regulation method based on the temperature prediction of the object being cooled, which can minimize temperature fluctuations by intervening in the cooling flow rate in advance to offset the effects of heat conduction delay. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a method for pre-regulating water supply in pumped storage power stations, addressing the aforementioned problems.

[0008] The technical solution adopted in this invention is: a method for pre-regulating water supply in pumped storage power stations, comprising: S1. Collect historical operating data of pumped storage units and technical water supply systems. The historical operating data shall include at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters. By comprehensively collecting multi-dimensional operational data that affects the temperature of the object being cooled, covering all key influencing factors such as unit operation, cooling system, and environment, a complete and reliable data foundation is provided for subsequent feature selection, model training, and pre-adjustment calculations. S2. Based on historical operating data, the key influencing variables that are strongly correlated with the temperature change of the object being cooled are selected using a feature screening algorithm. The optimal delay time between each key influencing variable and the temperature of the object being cooled is determined by mutual information calculation. By using a feature filtering algorithm to remove variables that are irrelevant to temperature changes or are redundant, the complexity of data processing and the amount of model computation are reduced, thereby improving the efficiency of subsequent model training. By using mutual information calculation to accurately match the heat conduction delay between key variables and the temperature of the object being cooled, the problem of asynchronous timing between variables and temperature changes caused by the lag in heat conduction is solved, eliminating control deviations from the source and laying the foundation for accurate prediction and pre-adjustment. S3. Construct a temperature prediction model for the object being cooled, using the time series data of the key influencing variables as the model input and the time series data of the temperature of the object being cooled after the corresponding optimal delay time as the model output, and train and optimize the temperature prediction model. By constructing a temperature prediction model adapted to the heat conduction delay characteristics, and combining the time-series characteristics of key variables with the time delay matching results, we can accurately predict future temperature changes of the cooled object, anticipate temperature deviation risks in advance, and provide a scientific and reliable basis for pre-regulation. S4. Establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions. Establish a quantitative correlation between cooling flow rate and temperature change, transforming abstract temperature requirements into specific flow rate adjustment parameters. This provides a precise mathematical basis for calculating the target cooling flow rate based on temperature deviation, achieving accurate matching between temperature control requirements and flow rate adjustment, and avoiding blind adjustment. S5: Real-time acquisition of current operating data of the unit and technical water supply system, generating time-series data of key influencing variables up to the current moment, inputting it into the trained temperature prediction model, and predicting the time-series temperature data of the cooling object. S6. Based on the deviation between the predicted cooling object temperature time series data and the preset stable operating range, and in conjunction with the mapping relationship model, determine the cooling flow adjustment amount of the corresponding branch pipe, and then obtain the target cooling flow of the corresponding branch pipe. S7. Based on the target cooling flow rate of the branch pipe, generate a pre-adjustment command and send it to the actuator of the technical water supply system to adjust the cooling flow rate of the main pipe and each branch pipe. During the adjustment process, monitor the temperature of the object being cooled and the actual cooling flow rate in real time and dynamically correct the adjustment command to keep the temperature of the object being cooled within the preset stable operating range.

[0009] As a preferred embodiment, the operating parameters include the unit's active power, speed, head, and load change rate; the cooling object temperature data include the oil temperature of each bearing oil groove, the temperature of each bearing bush, the transformer oil temperature, the transformer winding temperature, the stator winding temperature, the rotor winding temperature, and the upper and lower labyrinth ring temperatures of the impeller; the cooling flow rate data includes the cooling water velocity and flow rate of the main pipe and each branch pipe of the technical water supply system; and the environmental parameters include the ambient air temperature and the cooling water inlet temperature.

[0010] As a preferred embodiment, the feature selection algorithm is the maximum correlation minimum redundancy (mRMR) algorithm. The selection process includes: initializing a feature subset, calculating the correlation between each candidate variable and the temperature of the cooling object and the redundancy between variables, and selecting key influencing variables by sorting them according to the evaluation function values.

[0011] The mRMR algorithm efficiently filters out key features that are strongly correlated with temperature changes and have low redundancy among variables. This ensures the effectiveness of the input variables in temperature prediction, minimizes the dimensionality of the model input, reduces the computational load, improves the training efficiency and prediction accuracy of the temperature prediction model, and avoids interference from redundant variables in the prediction results.

[0012] As a preferred embodiment, the mutual information calculation involves calculating the time delay and maximum mutual information coefficient between each key influencing variable and the temperature change of the cooled object, and selecting the time point corresponding to the maximum mutual information coefficient as the optimal delay time.

[0013] By accurately quantifying the time lag characteristics between key variables and temperature changes through mutual information calculation, the optimal delay time is determined, enabling precise alignment between the time series data of key variables and the time series data of temperature changes. This completely solves the control lag problem caused by heat conduction delay and provides a guarantee for the accurate construction of subsequent temperature prediction models.

[0014] As a preferred embodiment, the temperature prediction model for the cooled object is a Long Short-Term Memory Neural Network (LSTM) model optimized based on the Sparrow Search Algorithm (SSA). The SSA is used to optimize the number of hidden layer neurons, training batches, and iterations of the LSTM.

[0015] Leveraging the strength of LSTM neural networks in processing time-series data, this model adapts to the continuous and dynamic changes in unit operation data. By automatically optimizing the key hyperparameters of LSTM using the SSA algorithm, it addresses the issues of low fitting accuracy and poor generalization ability caused by unreasonable manual setting of hyperparameters in traditional LSTM models. This significantly improves the fitting accuracy and generalization ability of the time-series temperature prediction model, enabling high-precision prediction of the future temperature of the cooling object.

[0016] As a preferred embodiment, the process of constructing the mapping relationship model includes: A simulation model of the technical water supply system pipeline was established, and numerical simulations were performed under different cooling flow conditions. The temperature change data of the cooled object was recorded to obtain simulation samples. Combined with the flow-temperature correlation samples in the historical operating data, the simulation samples and the flow-temperature correlation samples were fitted by the stepwise regression method to obtain the quantitative mapping relationship between the cooling flow adjustment and the temperature change under different operating conditions.

[0017] By integrating simulation data with historical measured data to construct a mapping model, the shortcomings of insufficient coverage of single historical data operating conditions can be effectively compensated, and the relationship between flow rate and temperature change under all operating conditions can be covered. A quantitative mapping relationship is obtained by fitting through stepwise regression method, which ensures the accuracy and universality of the mapping model, provides reliable mathematical support for the accurate calculation of target cooling flow rate, and avoids the deviation caused by empirical adjustment.

[0018] As a preferred embodiment, the actuators of the technical water supply system include electromagnetic regulating valves for each branch pipe and a variable frequency water supply pump set; the pre-adjustment commands include the opening adjustment value of the electromagnetic regulating valves and the frequency adjustment value of the water supply pumps.

[0019] The actuator, which combines an electromagnetic regulating valve with a variable frequency water pump set, replaces the traditional fixed frequency pump and manual valve, enabling continuous, precise, and dynamic adjustment of cooling flow. It can quickly respond to pre-adjustment commands and meet the flow adjustment needs under different operating conditions. At the same time, by precisely controlling the valve opening and pump set frequency, it can achieve on-demand water supply, avoid redundant cooling water volume, and reduce system energy consumption.

[0020] A pumped storage power station water supply pre-regulation device includes: The data acquisition module is used to collect historical operating data of the pumped storage unit and the technical water supply system. The historical operating data includes at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters. The data processing module is used to filter out key influencing variables that are strongly correlated with the temperature change of the cooled object based on historical operating data and feature filtering algorithms, and to determine the optimal delay time between each key influencing variable and the temperature of the cooled object through mutual information calculation. The model building module is used to build a temperature prediction model for the object being cooled. It takes the time series data of the key influencing variables as the model input and the time series data of the temperature of the object being cooled after the corresponding optimal delay time as the model output, and trains and optimizes the temperature prediction model. The mapping relationship construction module is used to establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions. The temperature prediction module is used to collect real-time operating data of the unit and technical water supply system, generate time-series data of key influencing variables up to the current moment, input the trained temperature prediction model, and predict the time-series temperature data of the cooled object. The relationship mapping module is used to determine the cooling flow adjustment amount of the corresponding branch pipe based on the deviation between the predicted cooling object temperature time series data and the preset stable operating range, combined with the mapping relationship model, so as to obtain the target cooling flow of the corresponding branch pipe. The regulation and control module is used to generate pre-regulation instructions based on the target cooling flow rate of the branch pipes and send them to the actuators of the technical water supply system to regulate the cooling flow rate of the main pipe and each branch pipe. During the regulation process, the temperature of the object being cooled and the actual cooling flow rate are monitored in real time and the regulation instructions are dynamically corrected to keep the temperature of the object being cooled within the preset stable operating range.

[0021] A storage medium storing a computer program executable by a processor, wherein the computer program, when executed, implements the steps of the pumped storage power station technical water supply pre-regulation method.

[0022] A pumped storage power station water supply pre-regulation device has a memory and a processor. The memory stores a computer program that can be executed by the processor. When the computer program is executed, it implements the steps of the pumped storage power station water supply pre-regulation method.

[0023] The beneficial effects of this invention are as follows: This invention determines the optimal delay time through mutual information calculation, achieves advanced prediction through the SSA-LSTM temperature prediction model, and combines it with active pre-adjustment execution to accurately match the heat conduction delay characteristics, predict the temperature change trend in advance and actively intervene in the flow rate, thereby offsetting the impact of heat conduction delay from the root, upgrading the traditional passive temperature response to active pre-control, significantly reducing the temperature fluctuation of the cooled object, avoiding equipment damage caused by excessive temperature fluctuations, and ensuring the stability of unit operation.

[0024] This invention uses an electromagnetic regulating valve and a variable frequency water pump set as the actuator, combined with flow adjustment calculation and dynamic adjustment, to replace the traditional fixed frequency pump and manual valve, realize continuous, accurate and dynamic adjustment of cooling flow, supply cooling water on demand, completely solve the problem of cooling water redundancy, reduce the energy consumption of the technical water supply system, and improve the economic efficiency of system operation.

[0025] This invention constructs a fully closed-loop control mechanism consisting of "data acquisition - feature processing - temperature prediction - flow calculation - pre-adjustment execution - real-time feedback - dynamic correction", ensuring that the pre-adjustment command can be dynamically optimized according to the real-time operating conditions of the unit, completely solving the limitations of traditional passive adjustment and ensuring the temperature stability of the unit during operating condition switching.

[0026] This invention, by pre-setting a stable operating range and making precise adjustments, combined with a flow-temperature quantization mapping model, strictly controls the temperature of the cooled object within a narrow and efficient range. At the same time, it improves the temperature control accuracy through techniques such as mRMR feature screening and SSA-LSTM optimization prediction, ensuring that the unit always operates under optimal conditions, balancing operating efficiency and equipment safety.

[0027] This invention organically combines data processing, time-series prediction, precise mapping, active pre-adjustment, and closed-loop control technologies to meet the full-condition operation requirements of pumped storage units. It not only solves the core pain points of existing technologies but also realizes intelligent and automated control of the technical water supply system, reduces manual operation and maintenance costs, and improves the overall safety, stability, and economy of pumped storage power stations. Attached Figure Description

[0028] Figure 1 This is a flowchart of the water supply pre-regulation method for pumped storage power station technology in the embodiment. Detailed Implementation

[0029] Example 1: This example is a method for pre-regulating water supply in a pumped storage power station, specifically including the following steps: S1. Collect historical operating data of pumped storage units and technical water supply systems. The historical operating data includes at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters.

[0030] S11. Collect historical operating data of the unit for the past year, with a data collection cycle of 1 minute, covering: Operating parameters: unit active power, speed, head, and load change rate; Temperature data of objects to be cooled: oil temperature in the thrust bearing oil sump, temperature of the thrust bearing bush; Cooling flow data: main pipe flow rate and velocity, thrust bearing cooling branch pipe cooling flow rate and velocity; Environmental parameters: ambient air temperature, cooling water inlet temperature.

[0031] S12. Preprocess the collected data: Linear interpolation is used to fill missing values, outliers are removed using the 3σ criterion, and the data is normalized according to the formula x=(x−min) / (max−min) to obtain a standardized dataset, providing clean and effective data support for subsequent steps. S2. Based on historical operating data, a feature filtering algorithm is used to identify key influencing variables strongly correlated with the temperature changes of the cooled object. Mutual information calculation is used to determine the optimal delay time between each key influencing variable and the temperature of the cooled object.

[0032] This embodiment uses the maximum correlation minimum redundancy (mRMR) algorithm to screen key influencing variables based on preprocessed historical operating data: initialize feature subset S, calculate the correlation between each candidate variable and the temperature of the cooling object and the redundancy between variables, sort by evaluation function value, select variables that are strongly correlated with temperature and have low redundancy as key influencing variables, add them to S, and achieve data dimensionality reduction and redundancy removal.

[0033] In this example, the evaluation function value of the variable is calculated using the following formula: Where I is the mutual information value, the variables with the largest evaluation function values ​​are selected and added to S in turn, and finally the active power of the unit, the cooling flow rate of the thrust bearing cooling branch pipe, the ambient temperature, and the cooling water inlet temperature are determined to be the key influencing variables.

[0034] In this embodiment, the optimal delay time for each key influencing variable is determined by mutual information calculation: the time delay and maximum mutual information coefficient between the key variable and the temperature change of the cooled object within 0-30 minutes are calculated, and the time point corresponding to the maximum mutual information coefficient is selected as the optimal delay time, so as to achieve time synchronization between the variable and the temperature change and solve the problem of time synchronization caused by heat conduction delay.

[0035] In some specific embodiments, the unit's active power is ultimately delayed by 8 minutes, the ambient temperature by 12 minutes, the cooling water inlet temperature by 7 minutes, and the thrust bearing cooling branch pipe cooling flow rate by 6 minutes. S3. Construct a temperature prediction model for the cooling object, using the time series data of the key influencing variables as the model input and the time series data of the cooling object temperature after the corresponding optimal delay time as the model output, and train and optimize the temperature prediction model.

[0036] This embodiment constructs a Long Short-Term Memory Neural Network (LSTM) temperature prediction model based on the Sparrow Search Algorithm (SSA). It takes time-series data of key influencing variables as input and outputs time-series temperature data of the object being cooled after the optimal delay time. The SSA algorithm is used to optimize the number of hidden layer neurons, training batches, and iterations of the LSTM. Historical data is divided into training and testing sets in an 8:2 ratio, and the Adam optimizer is used for training. The root mean square error (RMSE) and mean absolute error (MAE) are used as evaluation metrics. Training stops when the metrics meet preset thresholds, resulting in a high-precision temperature prediction model that accurately predicts future temperature changes of the object being cooled.

[0037] In some embodiments, the SSA parameter settings are: population size 30, number of iterations 50, and optimization variables are the number of hidden layer neurons in the LSTM (20-80), training batches (16-64), and number of iterations (100-500); LSTM model structure: the input layer dimension is 5 (number of key variables) × the length of the delay time series, the hidden layer is set to 45 after SSA optimization, and the output layer is the predicted value of the thrust bearing bush temperature for the next 10 minutes; model training: the preprocessed historical data is divided into training and testing sets in an 8:2 ratio, the Adam optimizer is used, the learning rate is 0.001, and RMSE and MAE are used as evaluation metrics; training results: after the model converges, the RMSE is 0.85℃ and the MAE is 0.62℃, which meets the preset accuracy requirements.

[0038] S4. Establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions.

[0039] This embodiment establishes a mapping model between the cooling flow rate of each branch pipe and the temperature change of the cooled object: First, a simulation model of the technical water supply system pipeline is established using Fluent software, and numerical simulation is performed under different cooling flow rate conditions. Temperature change data is recorded to obtain simulation samples. Combining the flow-temperature correlation samples in historical operating data, the stepwise regression method is used to fit the simulation samples and the flow-temperature correlation samples to obtain a quantitative mapping relationship between the cooling flow rate adjustment and the temperature change under different operating conditions, providing an accurate mathematical basis for subsequent target cooling flow rate calculation.

[0040] In some specific embodiments, a Fluent pipeline simulation model is established: based on the pipeline parameters of the power plant's technical water supply system, a flow domain is constructed in SCDM, an unstructured mesh is generated in ICEM, and boundary conditions are set in Fluent; simulation operating conditions are set: the cooling flow rate is changed (100-250 m³ / h), and the temperature change data of the thrust bearing bush under different flow rates are recorded to obtain 12 sets of simulation samples; data fitting: combined with 80 sets of flow-temperature correlation samples from historical operating data, a stepwise regression method is used to fit the mapping relationship. Where ΔQ is the flow rate adjustment, ΔT is the temperature deviation, P is the unit power, and T is the temperature deviation. env Given the ambient air temperature, k1=15.2, k2=0.03, and k3=2.1. S5: Real-time acquisition of current operating data for the unit and technical water supply system; generation of time-series data for key influencing variables up to the current moment; inputting this data into the trained temperature prediction model to predict the time-series temperature data of the cooled object.

[0041] This embodiment collects real-time operating data of the unit and technical water supply system, generates time-series data of key influencing variables, and inputs them into the trained SSA-LSTM model to predict the time-series temperature data of the cooling object in the next 5-15 minutes. The sliding window method is used to input the latest operating data every 1-3 minutes to update the temperature prediction curve in real time, ensuring the real-time nature of the prediction results and timely adapting to the dynamic changes in the unit's operating conditions.

[0042] In some specific embodiments, real-time data acquisition: Current operating data is collected every minute, including the unit's active power, the cooling flow rate of the thrust bearing cooling branch pipe, the ambient air temperature, and the cooling water inlet temperature; Temperature prediction: The processed real-time data is input into the temperature prediction model to obtain the thrust bearing bush temperature prediction curve for the next 10 minutes as the unit load increases from 100MW to 200MW. The predicted value rises from 56℃ to 72℃, and the predicted temperature will exceed the preset stable operating range of 55℃~65℃, but will still be within the normal operating temperature of 75℃. S6. Based on the deviation between the predicted cooling target temperature time series data and the preset stable operating range, combined with the mapping relationship model, the cooling flow rate adjustment amount of the corresponding branch pipe is determined, thereby obtaining the target cooling flow rate of the corresponding branch pipe.

[0043] In this embodiment, the predicted temperature time series data is compared with the preset stable operating range (e.g., the stable range for bearing oil temperature is set to 35-45℃, and the stable range for bearing bush temperature is set to 55-65℃). The deviation between the predicted temperature and the stable range is calculated. Combined with the flow-temperature mapping model, the adjustment amount of the cooling flow rate of the branch pipe is determined. At the same time, the constraint that the flow rate adjustment amount does not exceed 30% of the current flow rate is strictly followed (the adjustment amount of the flow rate at time T+1 compared to time T does not exceed 30% of the flow rate at time T). Finally, the target cooling flow rate of each branch pipe is obtained to ensure the rationality and safety of the adjustment. S7. Based on the target cooling flow rate of the branch pipe, a pre-adjustment command is generated and sent to the actuator of the technical water supply system to adjust the cooling flow rate of the main pipe and each branch pipe. During the adjustment process, the temperature of the cooled object and the actual cooling flow rate are monitored in real time, and the adjustment command is dynamically corrected to keep the temperature of the cooled object within the preset stable operating range.

[0044] In this embodiment, a pre-adjustment command is generated based on the target cooling flow rate. The command includes the opening adjustment value of the electromagnetic regulating valve and the frequency adjustment value of the water supply pump, and is sent to the actuator. The actuator adjusts the cooling flow rate of the main pipe and each branch pipe through the electromagnetic regulating valve of each branch pipe and the variable frequency water supply pump group to achieve active pre-adjustment, replacing the traditional passive response.

[0045] In this example, the temperature of the object being cooled and the actual cooling flow rate are monitored in real time during the adjustment process. When the actual temperature deviates from the predicted temperature after pre-adjustment by more than 3°C, the adjustment command is dynamically corrected to form a closed-loop control of "prediction-decision-execution-feedback", so that the temperature of the object being cooled is always maintained within the preset stable operating range, ensuring temperature control accuracy.

[0046] Example 2: This example is a pre-regulation device for water supply in a pumped storage power station, comprising: The data acquisition module is used to collect historical operating data of the pumped storage unit and the technical water supply system. The historical operating data includes at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters. The data processing module is used to filter out key influencing variables that are strongly correlated with the temperature change of the cooled object based on historical operating data and feature filtering algorithms, and to determine the optimal delay time between each key influencing variable and the temperature of the cooled object through mutual information calculation. The model building module is used to build a temperature prediction model for the object being cooled. It takes the time series data of the key influencing variables as the model input and the time series data of the temperature of the object being cooled after the corresponding optimal delay time as the model output, and trains and optimizes the temperature prediction model. The mapping relationship construction module is used to establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions. The temperature prediction module is used to collect real-time operating data of the unit and technical water supply system, generate time-series data of key influencing variables up to the current moment, input the trained temperature prediction model, and predict the time-series temperature data of the cooled object. The relationship mapping module is used to determine the cooling flow adjustment amount of the corresponding branch pipe based on the deviation between the predicted cooling object temperature time series data and the preset stable operating range, combined with the mapping relationship model, so as to obtain the target cooling flow of the corresponding branch pipe. The regulation and control module is used to generate pre-regulation instructions based on the target cooling flow rate of the branch pipes and send them to the actuators of the technical water supply system to regulate the cooling flow rate of the main pipe and each branch pipe. During the regulation process, the temperature of the object being cooled and the actual cooling flow rate are monitored in real time and the regulation instructions are dynamically corrected to keep the temperature of the object being cooled within the preset stable operating range.

[0047] Example 3: This example is a storage medium that stores a computer program that can be executed by a processor. When the computer program is executed, it implements the steps of the pumped storage power station water supply pre-regulation method described in Example 1.

[0048] Example 4: This example describes a pumped storage power station water supply pre-regulation device, which has a memory and a processor. The memory stores a computer program that can be executed by the processor. When the computer program is executed, it implements the steps of the pumped storage power station water supply pre-regulation method described in Example 1.

Claims

1. A method for pre-regulating water supply in a pumped storage power station, characterized in that, include: S1. Collect historical operating data of pumped storage units and technical water supply systems. The historical operating data shall include at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters. S2. Based on historical operating data, the key influencing variables that are strongly correlated with the temperature change of the object being cooled are selected using a feature filtering algorithm. The optimal delay time between each key influencing variable and the temperature of the object being cooled is determined by mutual information calculation. S3. Construct a temperature prediction model for the object being cooled, using the time series data of the key influencing variables as the model input and the time series data of the temperature of the object being cooled after the corresponding optimal delay time as the model output, and train and optimize the temperature prediction model. S4. Establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions. S5. Real-time acquisition of current operating data of the unit and technical water supply system, generation of time series data of key influencing variables up to the current moment, input into the trained temperature prediction model, and prediction of the time series data of the cooling object temperature; S6. Based on the deviation between the predicted cooling object temperature time series data and the preset stable operating range, and in conjunction with the mapping relationship model, determine the cooling flow adjustment amount of the corresponding branch pipe, and then obtain the target cooling flow of the corresponding branch pipe. S7. Based on the target cooling flow rate of the branch pipe, generate a pre-adjustment command and send it to the actuator of the technical water supply system to adjust the cooling flow rate of the main pipe and each branch pipe. During the adjustment process, monitor the temperature of the object being cooled and the actual cooling flow rate in real time and dynamically correct the adjustment command to keep the temperature of the object being cooled within the preset stable operating range.

2. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The operating parameters include the unit's active power, speed, head, and load change rate; the cooling object temperature data include the oil temperature of each bearing oil groove, the temperature of each bearing bush, the transformer oil temperature, the transformer winding temperature, the stator winding temperature, the rotor winding temperature, and the upper and lower labyrinth ring temperatures of the impeller; the cooling flow rate data includes the cooling water velocity and flow rate of the main pipe and each branch pipe of the technical water supply system; the environmental parameters include the ambient air temperature and the cooling water inlet temperature.

3. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The feature selection algorithm is the maximum correlation minimum redundancy (mRMR) algorithm. The selection process includes: initializing the feature subset, calculating the correlation between each candidate variable and the temperature of the cooling object and the redundancy between variables, and selecting key influencing variables by sorting them according to the evaluation function values.

4. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The mutual information calculation involves calculating the time delay and maximum mutual information coefficient between each key influencing variable and the temperature change of the cooled object, and selecting the time point corresponding to the maximum mutual information coefficient as the optimal delay time.

5. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The temperature prediction model for the cooled object is a Long Short-Term Memory (LSTM) neural network model optimized based on the Sparrow Search Algorithm (SSA). The SSA is used to optimize the number of hidden layer neurons, training batches, and iterations of the LSTM.

6. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The process of constructing the mapping relationship model includes: A simulation model of the technical water supply system pipeline was established, and numerical simulations were performed under different cooling flow conditions. The temperature change data of the cooled object was recorded to obtain simulation samples. By combining flow-temperature correlation samples from historical operating data, stepwise regression is used to fit the simulation samples and flow-temperature correlation samples to obtain a quantitative mapping relationship between cooling flow adjustment and temperature change under different operating conditions.

7. The method for pre-regulating water supply in a pumped storage power station according to claim 1, characterized in that, The actuators of the technical water supply system include electromagnetic regulating valves for each branch pipe and variable frequency water supply pump sets; the pre-adjustment commands include the opening adjustment value of the electromagnetic regulating valves and the frequency adjustment value of the water supply pumps.

8. A pre-regulation device for water supply in a pumped storage power station, characterized in that, include: The data acquisition module is used to collect historical operating data of the pumped storage unit and the technical water supply system. The historical operating data includes at least operating parameters, cooling object temperature data, cooling flow rate data, and environmental parameters. The data processing module is used to filter out key influencing variables that are strongly correlated with the temperature change of the cooled object based on historical operating data and feature filtering algorithms, and to determine the optimal delay time between each key influencing variable and the temperature of the cooled object through mutual information calculation. The model building module is used to build a temperature prediction model for the object being cooled. It takes the time series data of the key influencing variables as the model input and the time series data of the temperature of the object being cooled after the corresponding optimal delay time as the model output, and trains and optimizes the temperature prediction model. The mapping relationship construction module is used to establish a mapping relationship model between the cooling flow rate of each branch pipe and the temperature change of the corresponding cooling object. The mapping relationship model is constructed based on the correspondence between the cooling flow rate adjustment and the temperature change of the cooling object under different operating conditions. The temperature prediction module is used to collect real-time operating data of the unit and technical water supply system, generate time-series data of key influencing variables up to the current moment, input the trained temperature prediction model, and predict the time-series temperature data of the cooled object. The relationship mapping module is used to determine the cooling flow adjustment amount of the corresponding branch pipe based on the deviation between the predicted cooling object temperature time series data and the preset stable operating range, combined with the mapping relationship model, so as to obtain the target cooling flow of the corresponding branch pipe. The regulation and control module is used to generate pre-regulation instructions based on the target cooling flow rate of the branch pipes and send them to the actuators of the technical water supply system to regulate the cooling flow rate of the main pipe and each branch pipe. During the regulation process, the temperature of the object being cooled and the actual cooling flow rate are monitored in real time and the regulation instructions are dynamically corrected to keep the temperature of the object being cooled within the preset stable operating range.

9. A storage medium having a computer program stored thereon that can be executed by a processor, characterized in that, When the computer program is executed, it implements the steps of the pumped storage power station technical water supply pre-regulation method according to any one of claims 1 to 7.

10. A pre-regulation device for water supply in a pumped storage power station, comprising a memory and a processor, wherein the memory stores a computer program executable by the processor, characterized in that, When the computer program is executed, it implements the steps of the pumped storage power station technical water supply pre-regulation method according to any one of claims 1 to 7.

Images