A method for intelligent protection of water pump without water empty pumping in a water collecting well of a hydropower station

Abstract

The application relates to the technical field of automatic control of hydropower stations, and discloses a water-pump water-emptying intelligent protection method for a catch basin of a hydropower station, which comprises the following steps: collecting operation data of sensors arranged in a water pump of the catch basin and a connecting pipeline system of the water pump and an electric control system of the water pump; pre-processing the collected operation data to obtain pre-processed operation data; calculating characteristic deviation amounts based on the pre-processed operation data and a normal operation condition model; calculating a risk value according to the characteristic deviation amounts; and judging the risk of the water pump of the catch basin and the connecting pipeline system of the water pump according to the risk value. The application can accurately identify the water-emptying state.

Description

Technical Field

[0001] This application relates to the field of hydropower station automation control technology, and in particular to an intelligent protection method for water pumps in hydropower station sump pumps that are running dry. Background Technology

[0002] Hydropower station sump wells are used to collect seepage water from various levels of the powerhouse, dam, and generator maintenance drainage. The water from the sump wells must be promptly pumped out of the powerhouse to ensure its safety and dryness. As a core drainage device, the reliability of the pumps directly affects the safety of the power station. In existing technologies, sump well pumps mostly employ simple start-stop control based on a float, i.e., starting at high liquid levels, stopping at low liquid levels, and forced stopping via a low-level switch. However, in actual operation, problems often arise such as float jamming, malfunctions of the level sensor and switch, abnormal signal transmission, or abnormal water flow, causing the pumps to continue running dry even when the sump well is actually empty, the liquid level is extremely low, or the pumps are manually operated by the operator. Dry running of water pumps poses significant hazards: On the one hand, prolonged dry running without water can cause the mechanical seal to burn out due to loss of cooling and lubrication, and the impeller to generate severe friction with the pump casing, leading to overheating and deformation of the pump body, damage to the bearings, and in severe cases, overload and burnout of the motor; on the other hand, the vibration and cavitation caused by dry running can accelerate the fatigue damage of components, significantly shorten the equipment life, increase maintenance costs, and even lead to the paralysis of the drainage system, endangering the safety of the plant.

[0003] Currently, level switches and overcurrent protection devices are commonly used as the basis for dry-running protection for maintenance and leakage pumps in hydropower stations. Traditional protection methods have the following main problems: 1. Single-parameter judgment is prone to malfunction; 2. Fixed thresholds cannot adapt to complex operating conditions; 3. Lack of trend prediction capability; 4. Inability to distinguish between transient disturbances and actual dry-running; 5. Lack of self-learning and automatic recovery capabilities. Existing technologies mostly use simple relay control or single-parameter joint judgment methods, which have poor adaptability to complex hydropower station operating conditions. Summary of the Invention

[0004] In view of this, this application provides an intelligent protection method for water pumps in hydropower station sump pumps that are running dry.

[0005] This application discloses an intelligent protection method for water pumps in hydropower station sump pumps that are depleted due to lack of water, which includes: Step 1: Collect operational data from sensors and the pump's electrical control system deployed in the water collection well pump and its connecting pipeline system; Step 2: Preprocess the collected operational data to obtain preprocessed operational data; Step 3: Calculate the deviation of each feature based on the preprocessed operating data and the normal operating condition model; Step 4: Calculate the risk value based on the deviation of each feature; Step 5: Based on the risk value, assess the risk of the water collection well pump and its connecting pipeline system.

[0006] Furthermore, prior to step 1, the procedure also includes: The system is initialized by establishing a normal operating condition model corresponding to the water collection well pump and its connecting pipeline system. The normal operating condition model includes a reference current parameter I0, a reference pressure parameter P0, a reference flow parameter V0, a ​​reference temperature parameter T0, and a reference vibration parameter S0, and a corresponding historical operating database is established.

[0007] Furthermore, the operational data includes: current data I, pressure data P, flow data V, temperature data T, and vibration data S, with the sampling period set according to monitoring requirements.

[0008] Further, step 2 includes: The collected operational data are sequentially denoised and validated to obtain preprocessed operational data. The denoising process employs any one or a combination of the following methods: moving average filtering, Kalman filtering, wavelet denoising, and median filtering. Data validation involves cleaning outliers and missing values ​​from the denoised operational data, identifying invalid data, and then correcting the data using interpolation compensation, nearest neighbor substitution, and historical trend estimation. Preprocessing includes denoising and data validation.

[0009] Furthermore, the characteristic deviations include current deviation ΔI=|I-I0|, pressure deviation ΔP=|P-P0|, flow deviation ΔV=|V-V0|, temperature deviation ΔT=|T-T0|, and vibration deviation ΔS=|S-S0|.

[0010] Further, step 4 includes: The risk value K is calculated using the following formula:

[0011] Among them, w1 to w5 are the corresponding weights, and the sum of w1 to w5 is 1. The weights can be obtained by training based on historical running data.

[0012] Further, step 5 includes: Level 1 Risk Assessment: When K≤K1, the water collection well pump and its connecting pipeline system are determined to be in normal operating condition, and monitoring of the water collection well pump and its connecting pipeline system continues; K1 is the Level 1 threshold. Level 2 Risk Assessment: When K1 < K ≤ K2, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of slight water shortage risk, triggering a Level 1 warning and issuing an audible and visual alarm; K2 is the Level 2 threshold. Level 3 Risk Assessment: When K2 < K ≤ K3, the water collection well pump and its connecting pipeline system are determined to be in a state of moderate water shortage risk, triggering a Level 2 alarm, limiting the speed through the frequency converter, and sending maintenance suggestions to the operators; K3 is the Level 3 threshold. Level 4 Risk Assessment: When K > K3, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of severe water shortage risk, triggering a Level 3 alarm, forcing a shutdown, and initiating the emergency response procedure.

[0013] Furthermore, prior to step 5, the procedure also includes: The first-level threshold, second-level threshold, and third-level threshold are dynamically adjusted using the following formulas:

[0014] Where E represents the historical false alarm rate, D represents the degree of operating condition drift, and α and β are adjustment coefficients. The first-level threshold, second-level threshold, or third-level threshold corresponding to time n-1.

[0015] Furthermore, after step 5, the method further includes: First, the normal operating condition model is updated using newly added historical data in the historical operating database. Next, alarms and records are generated, including: alarm time, risk level, risk value, deviation of each characteristic, equipment operating status, environmental operating conditions information, and saved to the historical operation database. Finally, real-time trend analysis is performed. If data anomalies are found and early warnings can be issued, return to step 1 for online monitoring and risk assessment. Data anomalies include those caused by sensor or electrical quantity distortion or sudden changes, such as current data I, pressure data P, flow data V, temperature data T, and vibration data S. Data measurement and judgment need to be performed again.

[0016] Furthermore, the step of updating the model using newly added historical data in the historical operation database includes: Parameter weight update: The weight coefficients in the normal operating condition model are readjusted based on the verification results of the dry pumping event of the water collection well pump and its connecting pipeline system. Threshold update: Automatically optimize the first-level, second-level, and third-level thresholds based on historical false alarm rates; Model update: Retrain the normal operating condition model corresponding to the water collection well pump and its connecting pipeline system by adding new historical data in the historical operation database, improve the accuracy of subsequent risk value K identification, and reduce the time required for the normal operating condition model to make risk judgments.

[0017] Due to the adoption of the above technical solution, this application has the following advantages: 1. Multi-parameter fusion for high recognition accuracy: It integrates five types of parameters, including current and pressure, to accurately identify the evacuation status.

[0018] 2. Dynamic threshold adapts to complex working conditions, with real-time threshold correction to adapt to complex working conditions such as water level fluctuations and sediment.

[0019] 3. The self-learning algorithm can be continuously optimized, automatically updating weights and models based on historical data, becoming more accurate the more it is used.

[0020] 4. It has strong trend prediction capabilities and can predict the risk of short selling 5 to 15 seconds in advance, thus achieving early warning.

[0021] 5. It has automatic recovery capability, automatically resets and monitors after the fault is eliminated, and does not require manual intervention to restart.

[0022] 6. It can effectively reduce the false alarm rate, and the multi-parameter comprehensive judgment criteria can greatly reduce false alarms caused by single-parameter faults.

[0023] 7. It has strong engineering adaptability and is suitable for drainage systems of various hydropower stations. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments recorded in the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0025] Figure 1 This is a flowchart illustrating an intelligent protection method for water pumps in a hydropower station's sump pumping without water, according to an embodiment of this application. Figure 2 This is a schematic diagram of the multi-parameter fusion analysis process in an embodiment of this application. Detailed Implementation

[0026] The present application will be further described in conjunction with the accompanying drawings and embodiments. The described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of the present application.

[0027] See Figure 1 and Figure 2 This application provides an embodiment of an intelligent protection method for water pumps in a hydropower station's sump pumping without water, comprising: Step 1: Collect operational data from sensors and the pump's electrical control system deployed in the water collection well pump and its connecting pipeline system; Step 2: Preprocess the collected operational data to obtain preprocessed operational data; Step 3: Calculate the deviation of each feature based on the preprocessed operating data and the normal operating condition model; Step 4: Calculate the risk value based on the deviation of each feature; Step 5: Based on the risk value, assess the risk of the water collection well pump and its connecting pipeline system.

[0028] Optionally, before step 1, the method further includes: The system is initialized by establishing a normal operating condition model corresponding to the water collection well pump and its connecting pipeline system. The normal operating condition model includes a reference current parameter I0, a reference pressure parameter P0, a reference flow parameter V0, a ​​reference temperature parameter T0, and a reference vibration parameter S0, and a corresponding historical operating database is established.

[0029] Optionally, the operating data includes: current data I, pressure data P, flow data V, temperature data T, and vibration data S, with the sampling period set according to monitoring requirements.

[0030] Optionally, step 2 includes: The collected operational data are sequentially denoised and validated to obtain preprocessed operational data. The denoising process employs any one or a combination of the following methods: moving average filtering, Kalman filtering, wavelet denoising, and median filtering. Data validation involves cleaning outliers and missing values ​​from the denoised operational data, identifying invalid data, and then correcting the data using interpolation compensation, nearest neighbor substitution, and historical trend estimation. Preprocessing includes denoising and data validation.

[0031] Optionally, the characteristic deviations include current deviation ΔI=|I-I0|, pressure deviation ΔP=|P-P0|, flow deviation ΔV=|V-V0|, temperature deviation ΔT=|T-T0|, and vibration deviation ΔS=|S-S0|.

[0032] Optionally, step 4 includes: The risk value K is calculated using the following formula:

[0033] Among them, w1 to w5 are the corresponding weights, and the sum of w1 to w5 is 1. The weights can be obtained by training based on historical running data.

[0034] Optionally, step 5 includes: Level 1 Risk Assessment: When K≤K1, the water collection well pump and its connecting pipeline system are determined to be in normal operating condition, and monitoring of the water collection well pump and its connecting pipeline system continues; K1 is the Level 1 threshold. Level 2 Risk Assessment: When K1 < K ≤ K2, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of slight water shortage risk, triggering a Level 1 warning and issuing an audible and visual alarm; K2 is the Level 2 threshold. Level 3 Risk Assessment: When K2 < K ≤ K3, the water collection well pump and its connecting pipeline system are determined to be in a state of moderate water shortage risk, triggering a Level 2 alarm, limiting the speed through the frequency converter, and sending maintenance suggestions to the operators; K3 is the Level 3 threshold. Level 4 Risk Assessment: When K > K3, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of severe water shortage risk, triggering a Level 3 alarm, forcing a shutdown, and initiating the emergency response procedure.

[0035] Optionally, before step 5, the method further includes: The first-level threshold, second-level threshold, and third-level threshold are dynamically adjusted using the following formulas:

[0036] Where E represents the historical false alarm rate, D represents the degree of operating condition drift, and α and β are adjustment coefficients. The first-level threshold, second-level threshold, or third-level threshold corresponding to time n-1.

[0037] Optionally, after step 5, the method further includes: First, the normal operating condition model is updated using newly added historical data in the historical operating database. Next, alarms and records are generated, including: alarm time, risk level, risk value, deviation of each characteristic, equipment operating status, environmental operating conditions information, and saved to the historical operation database. Finally, real-time trend analysis is performed. If data anomalies are found and early warnings can be issued, return to step 1 for online monitoring and risk assessment. Data anomalies include those caused by sensor or electrical quantity distortion or sudden changes, such as current data I, pressure data P, flow data V, temperature data T, and vibration data S. Data measurement and judgment need to be performed again.

[0038] Optionally, updating the model using newly added historical data in the historical operation database includes: Parameter weight update: The weight coefficients in the normal operating condition model are readjusted based on the verification results of the dry pumping event of the water collection well pump and its connecting pipeline system. Threshold update: Automatically optimize the first-level, second-level, and third-level thresholds based on historical false alarm rates; Model update: Retrain the normal operating condition model corresponding to the water collection well pump and its connecting pipeline system by adding new historical data in the historical operation database, improve the accuracy of subsequent risk value K identification, and reduce the time required for the normal operating condition model to make risk judgments.

[0039] For ease of understanding, this application provides a more specific embodiment: A multi-parameter fusion criterion for venting risk assessment integrates five parameters, including current and pressure, to calculate a weighted risk value K. A dynamic threshold correction mechanism adjusts the three-level threshold in real time based on historical false alarms and operating condition drift. A self-learning weight update algorithm optimizes weights based on historical events, improving the accuracy of the criterion. Trend prediction logic fits the slope of K value changes, providing a 5-15 second advance warning of venting risk. Staged protection control responds in stages according to K value ranges: alarm → speed reduction → shutdown and interlocking. An automatic recovery and trial operation mechanism automatically resets and performs a trial run after fault resolution, only resuming operation after confirming normal operation.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of the claims of this application.

Claims

1. A method for intelligent protection against dry pumping of a water collection well pump in a hydropower station, characterized in that, include: Step 1: Collect operational data from sensors and the pump's electrical control system deployed in the water collection well pump and its connecting pipeline system; Step 2: Preprocess the collected operational data to obtain preprocessed operational data; Step 3: Calculate the deviation of each feature based on the preprocessed operating data and the normal operating condition model; Step 4: Calculate the risk value based on the deviation of each feature; Step 5: Based on the risk value, assess the risk of the water collection well pump and its connecting pipeline system.

2. The method according to claim 1, characterized in that, Before step 1, the following are also included: The system is initialized by establishing a normal operating condition model corresponding to the water collection well pump and its connecting pipeline system. The normal operating condition model includes a reference current parameter I0, a reference pressure parameter P0, a reference flow parameter V0, a ​​reference temperature parameter T0, and a reference vibration parameter S0, and a corresponding historical operating database is established.

3. The method according to claim 2, characterized in that, The operational data includes: current data I, pressure data P, flow data V, temperature data T, and vibration data S. The sampling period is set according to the monitoring requirements.

4. The method according to claim 1, characterized in that, Step 2 includes: The collected operational data are sequentially denoised and validated to obtain preprocessed operational data. The denoising process employs any one or a combination of the following methods: moving average filtering, Kalman filtering, wavelet denoising, and median filtering. Data validation involves cleaning outliers and missing values ​​from the denoised operational data, identifying invalid data, and then correcting the data using interpolation compensation, nearest neighbor substitution, and historical trend estimation. Preprocessing includes denoising and data validation.

5. The method according to claim 1, characterized in that, The characteristic deviations include current deviation ΔI=|I-I0|, pressure deviation ΔP=|P-P0|, flow deviation ΔV=|V-V0|, temperature deviation ΔT=|T-T0|, and vibration deviation ΔS=|S-S0|.

6. The method according to claim 1, characterized in that, Step 4 includes: The risk value K is calculated using the following formula: Among them, w1 to w5 are the corresponding weights, and the sum of w1 to w5 is 1. The weights can be obtained by training based on historical running data.

7. The method according to claim 1, characterized in that, Step 5 includes: Level 1 Risk Assessment: When K≤K1, the water collection well pump and its connecting pipeline system are determined to be in normal operating condition, and monitoring of the water collection well pump and its connecting pipeline system continues; K1 is the Level 1 threshold. Level 2 Risk Assessment: When K1 < K ≤ K2, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of slight water shortage risk, triggering a Level 1 warning and issuing an audible and visual alarm; K2 is the Level 2 threshold. Level 3 Risk Assessment: When K2 < K ≤ K3, the water collection well pump and its connecting pipeline system are determined to be in a state of moderate water shortage risk, triggering a Level 2 alarm, limiting the speed through the frequency converter, and sending maintenance suggestions to the operators; K3 is the Level 3 threshold. Level 4 Risk Assessment: When K > K3, the water pump and its connecting pipeline system in the water collection well are determined to be in a state of severe water shortage risk, triggering a Level 3 alarm, forcing a shutdown, and initiating the emergency response procedure.

8. The method according to claim 1, characterized in that, Before step 5, the following is also included: The first-level threshold, second-level threshold, and third-level threshold are dynamically adjusted using the following formulas: Where E represents the historical false alarm rate, D represents the degree of operating condition drift, and α and β are adjustment coefficients. The first-level threshold, second-level threshold, or third-level threshold corresponding to time n-1.

9. The method according to claim 1, characterized in that, After step 5, the following is also included: First, the normal operating condition model is updated using newly added historical data in the historical operating database. Next, alarms and records are generated, including: alarm time, risk level, risk value, deviation of each characteristic, equipment operating status, environmental operating conditions information, and saved to the historical operation database. Finally, real-time trend analysis is performed. If data anomalies are found and early warnings can be issued, return to step 1 for online monitoring and risk assessment. Data anomalies include those caused by sensor or electrical quantity distortion or sudden changes, such as current data I, pressure data P, flow data V, temperature data T, and vibration data S. Data measurement and judgment need to be performed again.

10. The method according to claim 9, characterized in that, The method of updating the model using newly added historical data in the historical operation database includes: Parameter weight update: The weight coefficients in the normal operating condition model are readjusted based on the verification results of the dry pumping event of the water collection well pump and its connecting pipeline system. Threshold update: Automatically optimize the first-level, second-level, and third-level thresholds based on historical false alarm rates; Model update: Retrain the normal operating condition model corresponding to the water collection well pump and its connecting pipeline system by adding new historical data in the historical operation database, improve the accuracy of subsequent risk value K identification, and reduce the time required for the normal operating condition model to make risk judgments.

Images