Hydraulic turbine governor control method fusing big data self-learning algorithm and PID algorithm

Abstract

The application discloses a water turbine governor control method fusing a big data self-learning algorithm and a PID algorithm, which records the result of each time of governor adjustment, including a current water head h, a current power P and corresponding guide vane opening Y, and inputs the historical database to form a data point set stored in a PLC, and introduces a water turbine model characteristic curve as a reference item to perform point error determination; the water turbine governor adjustment is guided by the historical big data, the adjustment speed and effect of the water turbine governor can be quickly improved, the adjustment record of the water turbine governor can be continuously enriched, and the big data abundance can be continuously improved; and the problems of low adjustment efficiency and long time consumption of the prior art PID adjustment mode are solved.

Description

Technical Field

[0001] This invention belongs to the technical field of turbine governors, specifically relating to a turbine governor control method that integrates big data self-learning algorithm and PID algorithm. Background Technology

[0002] The main function of a hydroelectric power station is to raise the water head through a dam, converting the potential energy of water into electrical energy, thereby providing clean energy to society. The turbine governor is a key piece of equipment in a hydroelectric power station. It controls the flow rate through the turbine by controlling the guide vane opening, thus controlling the output of the turbine generator unit. The main purpose of the turbine governor is to regulate power; the power P is a function of the water head h and the guide vane opening Y. The propeller turbine's blade opening φ will follow the guide vane opening Y.

[0003] Currently, the commonly used control method for hydro turbines is the Proportional Integral-Derivative (PID) regulation algorithm. This algorithm compares the difference between the target value and the actual value, the previous regulation speeds and effects, and uses the length of the pulses to increase or decrease the guide vane opening. This regulation method cannot achieve a one-time adjustment and will form multiple regulation cycles, oscillating around the target value to gradually correct the error. This results in a slow regulation speed, and each regulation requires a few seconds to ensure that the mechanical equipment has responded correctly before the next cycle can begin. Generally, load regulation requires several or even more than a dozen cycles, resulting in a very long regulation time. Therefore, it is necessary to design a hydro turbine governor control method that integrates big data self-learning algorithms and PID algorithms to solve the above problems. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a turbine governor control method that integrates big data self-learning algorithm and PID algorithm. This method records the results of each governor adjustment, including the current head h, current power P and corresponding guide vane opening Y, forming a data point set stored in the PLC. By using historical big data to guide the turbine governor adjustment, the adjustment speed and effect of the turbine governor can be rapidly improved. At the same time, the adjustment records of the turbine governor can be continuously enriched, and the richness of big data can be continuously improved. This solves the problems of low adjustment efficiency and long time consumption of the existing PID control method.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a turbine governor control method integrating big data self-learning algorithm and PID algorithm, comprising the following steps:

[0006] S1 records the data for each adjustment of the turbine governor to a stable operating condition, including head h, power P, and guide vane opening Y. The specific method is as follows:

[0007] Each time the turbine governor adjusts to a stable operating condition, it records a set of data on the head h, power P, and guide vane opening Y, and stores this data in the computer monitoring system as a historical database. After recording n sets of data, a structure A = {(h1, P1, Y1) ... (h...} is formed. m P m Y m )...(h n P n Y n The data point set A is};

[0008] S2. Based on the characteristic curve of the turbine model, calculate the head error value ΔY for each set of data points in data point set A. The specific method is as follows:

[0009] S201, read each group of data in data point set A, reading when the water head is h. m Power is P m， The water head is Y m At that time, the guide vane opening Y corresponding to the characteristic curve of the turbine model m ';

[0010] S202, let △Y=|Y m -Y m '|, and import the calculated n sets of △Y corresponding to the historical database to form a new data point set R={ (h1, P1, Y1, △Y1), (h2, P2, Y2, △Y2)...(h n P n Y n , △Y n )};

[0011] S3, when the computer monitoring system issues the target power command P0, it matches the current head data h0 and the power command P0 in the data point set R to find the closest set of historical reference data (h0, h0, h0). x P x Y x , △Y x ), of which (h x P x Y x , △Y x )∈R, the specific method is as follows:

[0012] S301, search within the data point set R in the historical database; when the water head h is found... x = h0 and power P x =A set of data for P0 (h) x P x Y x , △Y xWhen selecting this set of data as historical reference data, Y... x As the target value of the guide vane opening, directly jump to step S5 to adjust the speed controller;

[0013] S302, when no data set with the same head and power is found, calculate the values ​​for multiple data sets in the data point set R, and select the data set with the smallest distance difference ΔU as the historical reference data. The calculation method for ΔU is as follows:

[0014] △U=(|P x - P0| / P 额定 +|h x - h0| / h 额定 ) / 2;

[0015] Among them, P x h represents the power of any set of data points in the data point set R; x Let P be the head of any set of data in the data point set R; 额定 The rated power of the unit, h 额定 The rated head of the unit;

[0016] S4, regarding the historical reference data (h) selected in S302 x P x Y x , △Y x To determine the reference point with the smallest error relative to the true value, the following method is used:

[0017] S401, Obtain the current target power command P0 and the current head h. 0， Substitute the power P0 and head h0 into the turbine model characteristic curve in step S2, and read the corresponding model guide vane opening Y0'.

[0018] S402, historical reference data (h x P x Y x , △Y x The power P in ) x and water head h x Substitute the turbine model characteristic curve from step S2 and read the corresponding model guide vane opening Y. x ';

[0019] S403, Calculate the point selection error △E=|Y x '- Y0'|;

[0020] S404, compare △E and △Y x Size:

[0021] When △E is less than or equal to △Y xWhen the error is small, it indicates that the selected historical reference point is close to the current command point. In this case, Y is selected. x As the target value for the guide vane opening;

[0022] When △E is greater than △Y x If the selected historical reference point is far from the current command point, it indicates that the error in point selection is large, and the selected historical reference point is far from the current command point, resulting in a large error. In this case, Y0' in the turbine model characteristic curve is selected as the target value for the guide vane opening.

[0023] S5, the target value of the guide vane given opening obtained in S3 or S4 is assigned to the speed controller control system, and the speed controller directly and quickly drives the guide vane to reach the target value.

[0024] S6, after the operating conditions stabilize, reads the current power data and corrects the error through PID fine-tuning of the power closed loop, thus completing the entire speed controller control process;

[0025] S7. After completing the error correction, record and store the head, power and guide vane opening data after the operating conditions have stabilized into the historical database of the computer monitoring system, and update the data point set A.

[0026] Preferably, the turbine model characteristic curve in step S2 is the model turbine characteristic curve that comes with the turbine unit at the factory.

[0027] Preferably, in steps S202 and S302, each group of data in the database is read and calculated using a traversal algorithm.

[0028] The beneficial effects of the turbine governor control method integrating big data self-learning algorithm and PID algorithm provided by this invention are as follows:

[0029] 1. This method uses the historical big data of the governor's power command actions as a database to establish a three-dimensional data point set based on head, power and guide vane opening. This allows for the creation of a self-updating data model for subsequent governor actions, avoiding the problem of long adjustment cycles caused by simply using PID control.

[0030] 2. In this method, two adjustment modes are set up based on historical big data and turbine model characteristic curves. When the power command changes, the PLC automatically reads historical data from the database for search and matching. When a data point with the same power and head is matched, the PID calculation process is skipped directly, and the guide vane opening value of the historical data is assigned to the guide vane opening target value for governor action. When no matching data point is found, the closest historical data point is automatically selected by introducing the data point distance difference △U and the point selection error △E as reference items, and the availability of the selected historical data reference point is determined. This process of selecting the closest historical data point allows the governor to skip the PID calculation first, so that the guide vane opening is directly adjusted to a position close to the actual value, saving multiple PID calculation loops. Finally, the traditional PID calculation is used to correct the remaining small part of the error, which greatly improves the governor's action efficiency.

[0031] 3. This method does not require mechanical modification of existing equipment, making it easy to implement; and the data after each adjustment will update the historical database, so that the efficiency and accuracy of the next adjustment will continue to improve. Attached Figure Description

[0032] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0033] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0034] like Figure 1 A method for controlling a turbine governor that integrates big data self-learning algorithms and PID algorithms includes the following steps:

[0035] S1 records the data for each adjustment of the turbine governor to a stable operating condition, including head h, power P, and guide vane opening Y. The specific method is as follows:

[0036] Each time the turbine governor adjusts to a stable operating condition, it records a set of data on the head h, power P, and guide vane opening Y, and stores this data in the computer monitoring system as a historical database. After recording n sets of data, a structure A = {(h1, P1, Y1) ... (h...} is formed. m P m Y m )...(h n P n Y n The data point set A is};

[0037] S2. Based on the characteristic curve of the turbine model, calculate the head error value ΔY for each set of data points in data point set A. The specific method is as follows:

[0038] S201, read each group of data in data point set A, reading when the water head is h. m Power is P m， The water head is Y m At that time, the guide vane opening Y corresponding to the characteristic curve of the turbine model m ';

[0039] S202, let △Y=|Y m -Y m '|, and import the calculated n sets of △Y corresponding to the historical database to form a new data point set R={ (h1, P1, Y1, △Y1), (h2, P2, Y2, △Y2)...(h n P n Y n , △Y n )};

[0040] S3, when the computer monitoring system issues the target power command P0, it matches the current head data h0 and the power command P0 in the data point set R to find the closest set of historical reference data (h0, h0, h0). x P x Y x , △Y x ), of which (h x P x Y x , △Y x )∈R, the specific method is as follows:

[0041] S301, search within the data point set R in the historical database; when the water head h is found... x = h0 and power P x =A set of data for P0 (h) x P x Y x , △Y x When selecting this set of data as historical reference data, Y... x As the target value of the guide vane opening, directly jump to step S5 to adjust the speed controller;

[0042] S302, when no data set with the same head and power is found, calculate the values ​​for multiple data sets in the data point set R, and select the data set with the smallest distance difference ΔU as the historical reference data. The calculation method for ΔU is as follows:

[0043] △U=(|P x - P0| / P 额定 +|h x - h0| / h 额定 ) / 2;

[0044] Among them, P x h represents the power of any set of data points in the data point set R; x Let P be the head of any set of data in the data point set R; 额定 The rated power of the unit, h 额定 The rated head of the unit;

[0045] S4, regarding the historical reference data (h) selected in S302 x P x Y x , △Y x To determine the reference point with the smallest error relative to the true value, the following method is used:

[0046] S401, Obtain the current target power command P0 and the current head h. 0， Substitute the power P0 and head h0 into the turbine model characteristic curve in step S2, and read the corresponding model guide vane opening Y0'.

[0047] S402, historical reference data (h x P x Y x , △Y x The power P in ) x and water head h x Substitute the turbine model characteristic curve from step S2 and read the corresponding model guide vane opening Y. x ';

[0048] S403, Calculate the point selection error △E=|Y x '- Y0'|;

[0049] S404, compare △E and △Y x Size:

[0050] When △E is less than or equal to △Y x When the error is small, it indicates that the selected historical reference point is close to the current command point. In this case, Y is selected. x As the target value for the guide vane opening;

[0051] When △E is greater than △Y x If the selected historical reference point is far from the current command point, it indicates that the error in point selection is large, and the selected historical reference point is far from the current command point, resulting in a large error. In this case, Y0' in the turbine model characteristic curve is selected as the target value for the guide vane opening.

[0052] S5, the target value of the guide vane given opening obtained in S3 or S4 is assigned to the speed controller control system, and the speed controller directly and quickly drives the guide vane to reach the target value.

[0053] S6, after the operating conditions stabilize, reads the current power data and corrects the error through PID fine-tuning of the power closed loop, thus completing the entire speed controller control process;

[0054] S7. After completing the error correction, record and store the head, power and guide vane opening data after the operating conditions have stabilized into the historical database of the computer monitoring system, and update the data point set A.

[0055] Preferably, the turbine model characteristic curve in step S2 is the model turbine characteristic curve that comes with the turbine unit at the factory.

[0056] Preferably, in steps S202 and S302, each group of data in the database is read and calculated using a traversal algorithm.

[0057] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The embodiments and features described in these embodiments can be arbitrarily combined without conflict. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for controlling a turbine governor that integrates a big data self-learning algorithm and a PID algorithm, characterized in that, Includes the following steps: S1 records the data for each adjustment of the turbine governor to a stable operating condition, including head h, power P, and guide vane opening Y. The specific method is as follows: Each time the turbine governor adjusts to a stable operating condition, it records a set of data on the head h, power P, and guide vane opening Y, and stores this data in the computer monitoring system as a historical database. After recording n sets of data, a structure A = {(h1, P1, Y1) ... (h...} is formed. m P m Y m )...(h n P n Y n The data point set A is}; S2. Based on the characteristic curve of the turbine model, calculate the head error value ΔY for each set of data points in data point set A. The specific method is as follows: S201, read each group of data in data point set A, reading when the water head is h. m Power is P m， The water head is Y m At that time, the guide vane opening Y corresponding to the characteristic curve of the turbine model m '; S202, let △Y=|Y m -Y m '|, and import the calculated n sets of △Y corresponding to the historical database to form a new data point set R={ (h1, P1, Y1, △Y1), (h2, P2, Y2, △Y2)...(h n P n Y n , △Y n )}; S3, when the computer monitoring system issues the target power command P0, it matches the current head data h0 and the power command P0 in the data point set R to find the closest set of historical reference data (h0, h0, h0). x P x Y x , △Y x ), of which (h x P x Y x , △Y x )∈R, the specific method is as follows: S301, search within the data point set R in the historical database; when the water head h is found... x = h0 and power P x = A set of data for P0 (h) x P x Y x , △Y x When selecting this set of data as historical reference data, Y... x As the target value of the guide vane opening, directly jump to step S5 to adjust the speed controller; S302, when no data set with the same head and power is found, calculate the values ​​for multiple data sets in the data point set R, and select the data set with the smallest ΔU as the historical reference data. The calculation method for ΔU is as follows: △U=(|P x - P0| / P 额定 +|h x - h0| / h 额定 ) / 2; Among them, P x h represents the power of any set of data points in the data point set R; x Let P be the head of any set of data in the data point set R; 额定 The rated power of the unit, h 额定 The rated head of the unit; S4, regarding the historical reference data (h) selected in S302 x P x Y x , △Y x To determine the reference point with the smallest error relative to the true value, the following method is used: S401, Obtain the current target power command P0 and the current head h. 0， Substitute the power P0 and head h0 into the turbine model characteristic curve in step S2, and read the corresponding model guide vane opening Y0'. S402, historical reference data (h x P x Y x , △Y x The power P in ) x and water head h x Substitute the turbine model characteristic curve from step S2 and read the corresponding model guide vane opening Y. x '; S403, Calculate the point selection error △E=|Y x '- Y0'|; S404, compare △E and △Y x Size: When △E is less than or equal to △Y x When the error is small, it indicates that the selected historical reference point is close to the current command point. In this case, Y is selected. x As the target value for the guide vane opening; When △E is greater than △Y x If the historical reference point is far from the current command point, it indicates that the selection error is large and the selected historical reference point is far from the current command point, resulting in a large error. In this case, Y0' is selected as the target value for the guide vane opening. S5, the target value of the guide vane given opening obtained in S3 or S4 is assigned to the speed controller control system, and the speed controller directly and quickly drives the guide vane to reach the target value. S6, after the operating conditions stabilize, reads the current power data and corrects the error through PID fine-tuning of the power closed loop, thus completing the entire speed controller control process; S7. After completing the error correction, record and store the head, power and guide vane opening data after the operating conditions have stabilized into the historical database of the computer monitoring system, and update the data point set A.

2. The turbine governor control method integrating big data self-learning algorithm and PID algorithm according to claim 1, characterized in that: The turbine model characteristic curve in step S2 is the model turbine characteristic curve that comes with the turbine unit from the factory.

Images