Fault sample expansion method for excitation system of pumped storage power station

Abstract

The invention discloses a fault sample expansion method for an excitation system of a pumped storage power station, and the method comprises the steps: carrying out the fixed-point-crossing quadratic fitting of the characteristic quantity data of a small-power motor test group and the characteristic quantity data of a large-power motor under the normal operation condition of the excitation system of the pumped storage power station, and obtaining a mapping matrix; small motor test sample data under specific fault conditions are obtained, a fault sample matrix is formed, a calculation method of correlation judgment factors and monotonicity judgment factors is put forward, characteristic quantities in different fault types are divided into irrelevant characteristic quantities, stable characteristic quantities and follow-up characteristic quantities, and the characteristic quantities in different fault types are calculated. Different mapping coefficient calculation methods are defined for different types of characteristic quantities to form a fault mapping matrix, matrix operation is performed on a fault sample matrix by using a mapping function and the fault mapping matrix to realize fault sample expansion, and the problems of small number and low quality of fault samples of the excitation system of the pumped storage power station are effectively solved.

Description

Technical Field

[0001] The invention belongs to the technical field of fault monitoring of hydropower stations, and in particular relates to a method for expanding fault samples of an excitation system of a pumped storage power station. Background Art

[0002] The excitation system of a pumped storage power station is a key part of the normal operation of the power station. It provides the necessary magnetic field for the generator to maintain its stable operation. When there are power supply voltage fluctuations, rectifier component damage, rotating parts failure and control loop problems, the excitation system may fail. If the type of excitation system failure cannot be discovered and repaired in time, it will affect the normal operation of the pumped storage power station.

[0003] The current method for distinguishing fault types is mainly to adopt threshold analysis. By collecting and monitoring data such as stator voltage, stator current, power, rotor voltage, rotor current and speed of the excitation system of the pumped-storage power station, the threshold of each characteristic quantity is determined, and the normal and abnormal states of the system are distinguished. Combined with artificial intelligence, state analysis is performed to determine the fault type of the equipment and make timely repairs.

[0004] The sensitivity of the threshold value that needs to be set by threshold analysis should be within an appropriate range. If the sensitivity is too low, the fault cannot be detected. If the sensitivity is too high, too many false alarms of fault types will be generated. This leads to the need for a large number of samples to determine the threshold value and the heavy reliance on the quality of the sample data. When the sample data is inaccurate or incomplete, the result of the threshold analysis may be biased, affecting the identification of the fault type. For large units, the number of samples of various fault types in the excitation system is very small. In this case, the threshold analysis method is difficult to detect the fault type in a timely and accurate manner. Summary of the invention

[0005] The main purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art and to provide a method, device, electronic device and storable medium for expanding fault samples of the excitation system of a pumped storage power station. By collecting operating data of motors of different powers under normal and fault conditions, the fault sample data of the excitation system is expanded using the test sample data of small power motors.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: The present invention provides a method for expanding fault samples of an excitation system of a pumped storage power station, which is applied to the excitation system of a pumped storage power station. The excitation system of the pumped storage power station includes: an excitation power supply and an excitation winding for providing a magnetic field for a motor, an armature power supply for providing a constant DC voltage for the motor, and a motor carried by the excitation system includes a stator and a rotor and a measuring device. The stator measures the stator voltage, stator current and reactive power by means of an external voltmeter, an ammeter and a reactive power meter. The rotor measures the rotor voltage and rotor current by means of a voltage sensor and a current sensor. The measuring device includes a force sensor for measuring torque and a tachometer for measuring rotational speed. The method for expanding fault samples includes the following steps: S1. Obtain the characteristic quantity data of the small power motor test group and the characteristic quantity data of the high power motor under the normal operation of the excitation system: Through the voltmeter, ammeter, voltage sensor, current sensor, tachometer and reactive meter, the characteristic quantities of the motor carried by the excitation system are obtained, including: stator voltage , stator current , rotor voltage , rotor current and speed , reactive power ; Motors with rated powers of 5kW, 10kW, 5nkW, ..., 95kW are selected as the low-power motor test group, where n is the corresponding number of the small-power motor sample, n=1, 2, ..., 19; When the excitation system is operating normally, when the excitation system of the pumped storage power station is equipped with a 5kW motor, the motor operation data is recorded, that is, the stator voltage time series data of the excitation system under the operation of the 5kW motor is obtained. , stator current timing data , rotor voltage timing data , rotor current timing data , speed timing data and reactive power time series data , forming a sample timing array of the small power motor numbered 1 under normal operating conditions of the excitation system ; The same process is performed on other motors in the low-power motor test group, and the second sample timing array corresponding to the motor with a rated power of 10kW is obtained in turn. ; The motor with a rated power of 5nkW corresponds to the Sample time series array ; Equipped with the 19th sample timing array corresponding to the rated power of 95kW ; The above 19 sample time series arrays are characteristic quantity time series arrays of the small power motor test group under the condition of normal operation of the excitation system; Under the condition of normal operation of the pumped storage excitation system, the high-power motors actually put into use in the pumped storage power station are monitored and recorded. If the rated power of the high-power motors actually put into use in the pumped storage power station is , and its corresponding feature array number is , get the timing array corresponding to the high-power motor actually put into use in the pumped storage power station ; S2. Calculate the standard values ​​of the characteristic quantities of the small power motor test group and the high power motor: Select a time period without missing values ​​in each feature time series array , is the time corresponding to the starting data point of the selected time segment, To select the moment corresponding to the last data point of a time segment, the average of each characteristic value of all low-power motor test samples in the segment is taken as the standard value. The specific steps are as follows: The sample time series array numbered 1 For example, the stator voltage in the time period The stator voltage is obtained by taking the average value

[0007] (1) Similarly, find the mean values ​​of the stator voltage, stator current, reactive power, rotor voltage, rotor current, and speed of the sample time series array numbered 1 , , , , , The specific calculation formula is: (2) The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of them, get the characteristic value standard value of the small power motor test sample numbered 1 when the excitation system is operating normally , The subscript 0 indicates that it is the standard value; Similarly, for the sample time series array numbered n

[0008] Do the same (3) The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the above can obtain the standard value of the characteristic quantity of the nth small power motor test sample when the excitation system is operating normally. , The subscript 0 indicates that it is the standard value; The above processing was performed on 19 small power motor test samples and high power motor samples to obtain the standard value of the characteristic quantity of the small power motor test group when the excitation system is normal. ,in , and the standard value of the characteristic quantity of high-power motors when the excitation system is operating normally .

[0009] In the above step S2, the standard values ​​of the characteristic quantities of the 19 small-power motor test samples when the excitation system is operating normally are used for fitting each function in the subsequent step S3, and the standard values ​​of the characteristic quantities of the high-power motor when the excitation system is operating normally are used as fixed points of the fitting function to play a fitting and constraint role.

[0010] S3. Fitting to obtain the mapping matrix: Obtain the stator voltage standard value of the small power motor test group from the characteristic quantity standard value in step S2 And the standard value of stator voltage for high-power motors , define equivalent numbers is a positive real number less than or equal to 20. The stator voltage standard value is calculated by the least squares binomial over-determined value fitting method. About Equivalent Numbers The stator voltage function , with equivalent numbers as The standard value of the stator voltage of a high-power motor is As constraints, the function Perform fitting and take the stator voltage function , the fitting formula is as follows: (4) in are the coefficients of the quadratic term and the linear term to be fitted respectively. are constants, is the equivalent number corresponding to the small power motor sample, , Corresponding number for high-power motors, Obtained by step S13; Use the equivalent number x as the sample number n, that is, x=1,2,…,19 and its corresponding stator voltage standard value The coefficients of the quadratic term and the linear term are fitted and solved. The specific steps of fitting and solving are as follows: Calculate the average value of the difference between the sample number and the corresponding number of the high-power motor, and use the result to express,

[0011] Calculate the average of the squares of the differences between the sample numbers and the corresponding numbers of the high-power motors. The result is represented by b. Calculate the average value of the cube of the difference between the sample number and the corresponding number of the high-power motor, and the result is represented by c.

[0012] Calculate the average value of the fourth power of the difference between the sample number and the corresponding number of the high-power motor, and the result is represented by d.

[0013] Calculate the average value of the stator voltage standard value corresponding to n=1,2,…,19, and the result is represented by e.

[0014] Calculate the average value of the difference between the sample number and the corresponding number of the high-power motor and the product of the corresponding stator voltage standard value. The result is represented by f.

[0015] Calculate the average value of the product of the square of the difference between the sample number and the corresponding number of the high-power motor and the corresponding stator voltage standard value. The result is expressed in g.

[0016] Then the coefficient of the quadratic term is

[0017] The coefficient of the first-order term is

[0018] Substitute the calculated quadratic and linear coefficients into the stator voltage function ; The stator current, reactive power, rotor voltage, rotor current and speed are fitted with binomial over-fixed value using the same method to obtain the stator current function. , rotor voltage function , rotor current function , speed function and reactive function , define a sixth-order diagonal matrix as the mapping matrix , write the above function into the mapping matrix , diag represents the operation of constructing the elements in the brackets into a diagonal matrix; Furthermore, by substituting any positive real number x less than or equal to 20 into the mapping matrix, the rated power is obtained as The measured mean value of each characteristic quantity of the motor in the normal operation state of the excitation system , The rated power is The measured mean value of the stator voltage of the motor in the normal operation state of the excitation system, The rated power is The measured mean value of the stator current of the motor in the normal operation state of the excitation system, The rated power is The measured mean value of the motor rotor voltage in the normal operation state of the excitation system, The rated power is The measured mean value of the motor rotor current in the normal operation state of the excitation system, The rated power is The measured average value of the motor speed in the normal operation state of the excitation system, The rated power is The measured mean value of the motor's reactive power under normal operating conditions of the excitation system.

[0019] S4, obtaining small motor test sample data under specific fault conditions to form a fault sample matrix; The excitation system under different fault conditions has different equipment problems, and the detection value of the characteristic quantity will also change differently. For specific fault conditions, by changing the motor carried, the changes in the characteristic quantity data when the motors with different powers are working are studied, and the regularity at the data level is obtained. The specific steps are as follows: Since the failure rate of the excitation system of the large motor put into operation is low, the abnormal data is difficult to obtain, and the cost of the fault test is high, the small power motor test group in step S1 is selected to perform a specific fault type test, and the fault time period is recorded as , n is the corresponding number of the small power motor sample, n=1,2,…,19, the values ​​of each characteristic quantity of the small power motor numbered n during the excitation system failure are monitored to obtain the characteristic quantity of the small power motor numbered n The time series of fault data , The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of ; Obtain the characteristic quantity of the low-power motor numbered n in step S2 The mean As standard value; Select time period Characteristic quantity of small power motor with internal number n Failure data Average and standard value The one with the smallest absolute value of the difference is recorded as ,Right now satisfy (5) The fault time period is , the number of fault data in this time period is m, calculate the mean value of fault data in this time period (6) Then the abnormal value of the characteristic quantity obtained from the fault experiment is selected as (7); Take is the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed , the stator voltage of the small power motor numbered n is calculated , stator current 、Reactive , rotor voltage , rotor current , speed Outliers under specific fault conditions , , , , , , the subscript b indicates that it is an outlier; Define a sixth-order diagonal matrix is the fault sample matrix, and the abnormal values ​​of the characteristic quantities under the above specific fault conditions are written into the matrix to obtain the fault sample matrix ; Repeat the above operations for 19 low-power motors in turn to obtain the fault sample matrix of the 19 low-power motors under specific fault conditions; In this step S4, the fault sample matrix The six diagonal elements of represent the six characteristic quantity abnormal values ​​of the small power test motor numbered n under specific fault conditions. The fault sample matrix is ​​obtained by the fault test of the small power test motor group, reflecting the abnormal values ​​of the six characteristic quantities of the small power motor test group under specific fault conditions.

[0020] S5. Calculate the correlation judgment factor to distinguish the relevant feature quantities from the irrelevant feature quantities: The abnormal values ​​of each characteristic quantity in 19 low-power motors under a specific fault type are obtained from the fault sample matrix of step S4. , , , , , ; Define feature quantity Correlation judgment factor : (8)

[0021] in, is the abnormal value of the characteristic quantity in 19 low-power motors, is the characteristic quantity of the small power motor numbered n under normal operating conditions of the excitation system The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of; If the correlation factor , it is considered that the characteristic quantity is not a representation of the fault type and is defined as an irrelevant characteristic quantity of the fault type; if , it is considered that the characteristic quantity is the characterization of this fault type and can reflect the fault state, and is defined as the relevant characteristic quantity of this fault type.

[0022] S6. Calculate the monotonicity judgment factor to distinguish between stable feature quantities and follow-up feature quantities: Defining the monotonicity factor , used to classify relevant feature quantities, where the relevant feature quantities are those that meet the correlation judgment factor in step S5 The characteristic quantity of the monotonicity judgment factor is calculated as follows: (9) in, Indicates the characteristic quantity of the small power motor numbered n The outliers in the fault test, is the average abnormal value of the characteristic quantities of 19 small power motors in the fault test, and the calculation formula is: (10) like , it is considered that the relevant characteristic quantity is a stable characteristic quantity, that is, the relevant characteristic quantity does not change with the change of motor power. If , it is considered that the relevant characteristic quantity is a follow-up characteristic quantity, that is, the relevant characteristic quantity changes with the change of motor power, that is, (10).

[0023] S7. Calculate the fault mapping matrix: Defining Fault Mapping Coefficients , represents the characteristic quantity of the small power motor test group The mapping relationship between the fault data obtained in the fault test and the characteristic value of the high-power motor actually put into operation under the corresponding fault conditions; Corresponding feature quantity The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the following defines the fault mapping coefficient The value is obtained as follows: For irrelevant feature quantities, the fault mapping coefficient The value is 0; For stable features among relevant features, the fault mapping coefficient The value is 1; For the follow-up feature in the relevant feature, assume that the fault mapping coefficient The change of conforms to the linear law, and the least square method is used to calculate its mapping coefficient, that is, (11) Define a sixth-order diagonal matrix is the fault mapping matrix, The values ​​are selected from the stator voltage , stator current , rotor voltage , rotor current , speed 、Reactive , 6 characteristic quantities are calculated , , , , , The mapping coefficient , , , , , , the fault mapping matrix is ​​composed of the above mapping coefficients , the expression is as follows ; In this step S7, the fault mapping matrix It represents the mapping relationship between the fault sample matrix of the low-power motor in step S4 and the fault sample matrix of the high-power motor put into actual operation under the corresponding fault condition.

[0024] S8. Expand the fault samples using the mapping function and the fault mapping matrix: Using the fault sample matrix obtained in step S4 , and the mapping matrix obtained in step S3 and the fault mapping matrix obtained in step S7 Perform matrix operations to obtain the expanded sample matrix , the above fault sample matrix and the mapping matrix is the function matrix, the fault mapping matrix is a constant coefficient matrix, the specific process is: Select a small power motor for the excitation system specific fault test, where the small power motor is any motor with a power less than 100kW. Obtain the fault sample matrix according to the method in step S4. , the rated power of the motor used is , then the fault sample matrix subscript number is ; Take the equivalent number x as the number , substitute into the mapping matrix get , take the equivalent number x as the rated power of the pumped storage power station actually in operation The corresponding number of high-power motor , substitute into the mapping matrix get , perform the following matrix operations (12) The non-zero elements in are the relevant characteristic quantities under this fault condition, and their values ​​can be used as the actual power put into operation in the pumped storage system. The fault test data of high-power motors is used to expand the fault samples.

[0025] By performing multiple fault tests on low-power motors, a large number of high-quality fault samples can be obtained, effectively improving the accuracy and sensitivity of threshold analysis.

[0026] Compared with the prior art, the present invention has the following advantages and beneficial effects: 1. The present invention proposes a correlation judgment factor, which calculates the degree of change of the characteristic quantity under the fault condition of the excitation system and the normal operation condition of the excitation system, and sets a certain threshold to achieve that the correlation judgment factor is greater than 1 when the degree of change of the characteristic quantity is large, and the correlation judgment factor is less than or equal to 1 when the degree of change of the characteristic quantity is small. It is used to distinguish the relevant characteristic quantities and irrelevant characteristic quantities in different fault types. The irrelevant characteristic quantities cannot reflect the operation status of the excitation system. The correlation judgment factor can effectively screen out the characteristic quantities that can reflect the specific fault state.

[0027] 2. The present invention proposes a monotonicity judgment factor, which calculates the degree of change of relevant characteristic quantities when equipped with motors of different powers in the case of a fault in the excitation system, and sets a certain threshold to ensure that the monotonicity judgment factor is greater than 1 when the degree of change is large, and the monotonicity judgment factor is less than or equal to 1 when the degree of change is small, thereby screening out stable characteristic quantities that do not change significantly when the motor power changes and follow-up characteristic quantities that change significantly with the change of the motor power, and being able to identify the change characteristics of the relevant characteristic quantities under specific fault conditions.

[0028] 3. The present invention uses correlation judgment factors and monotonicity judgment factors to calculate the fault mapping matrix, takes the fault mapping coefficient corresponding to the irrelevant feature quantity as 0, takes the fault mapping coefficient corresponding to the stable feature quantity as 1, and fits the fault mapping coefficient corresponding to the follow-up feature quantity according to its change law, so that when the fault mapping matrix and the fault sample matrix are operated, the feature quantities with different change characteristics obtained from the small power motor fault test are transformed as required to obtain the fault samples corresponding to the required power motor, wherein the fault mapping matrix can effectively screen out the feature quantities related to the specific fault type, which is simple and efficient. At the same time, the characteristics of each feature quantity of the fault type can be directly seen from the fault mapping matrix, and the feature quantities with monitoring significance can be screened out through the fault mapping matrix, simplifying the fault monitoring procedure.

[0029] 4. The present invention adopts a matrix operation method, which can quickly and effectively transform test samples of various fault types and various low-power motors into fault samples corresponding to the required power motor, thereby achieving the purpose of expanding the sample.

[0030] 5. The present invention utilizes the changing law of various characteristic quantities when the power of the motor carried by the excitation system changes from small to large, adopts a small-power motor to conduct tests, and processes the test results to obtain fault samples of the high-power motor, which can save test costs while expanding effective samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0032] Figure 1 It is a structural block diagram of the components of the excitation system of the pumped storage power station in Embodiment 1 of the present invention; Figure 2 It is a flow chart of a method for expanding fault samples of an excitation system of a pumped storage power station disclosed in the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.

[0034] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments.

[0035] Example 1 like Figure 2 As shown, this embodiment discloses a method for expanding fault samples of an excitation system of a pumped storage power station, and the steps are as follows: S1. Obtain the characteristic quantity data of the small power motor test group and the characteristic quantity data of the high power motor under the normal operation of the excitation system: Assuming that the power of the high-power motor actually put into operation in the pumped-storage power station is 1 MW, in this embodiment, the expansion of the high-power motor fault sample includes the following steps: S1. Obtain the characteristic quantity data of the small power motor test group and the characteristic quantity data of the high power motor under the normal operation of the excitation system: S11. Obtain characteristic quantities of the motor carried by the excitation system through a voltmeter, an ammeter, a voltage sensor, a current sensor, a tachometer and a reactive meter, including: stator voltage , stator current , rotor voltage , rotor current and speed , reactive power ; S12, select motors with rated power of 5kW, 10kW, 5nkW, ..., 95kW in descending order as the low-power motor test group, where n is the corresponding number of the small-power motor sample, n=1, 2, ..., 19; Under normal operation of the excitation system, when the excitation system of the pumped storage power station is equipped with a 5nkW motor, the motor operation data is recorded to obtain the stator voltage time series data of the excitation system under the operation of the 5nkW motor. , stator current timing data , rotor voltage timing data , rotor current timing data , speed timing data and reactive power time series data , forming a sample timing array of a small power motor numbered n under normal operating conditions of the excitation system ; S13. Under the condition of normal operation of the pumped storage excitation system, the high-power motors actually put into use in the pumped storage power station are monitored and recorded. The characteristic quantity array number corresponding to the high-power motors actually put into use in the pumped storage power station is , get the timing array corresponding to the high-power motor actually put into use in the pumped storage power station .

[0036] S2. Calculate the standard values ​​of characteristic quantities of the low-power motor test group and the high-power motor; Select a time period without missing values ​​in each feature time series array , is the time corresponding to the starting data point of the selected time segment, To select the moment corresponding to the last data point of the time segment, the average of each characteristic value of all low-power motor test samples in the segment is taken as the standard value. The specific steps are as follows: The sample time series array numbered n For example, the stator voltage in the time period The stator voltage mean is obtained by taking the internal mean value :

[0037] Similarly, find the mean values ​​of stator voltage, stator current, reactive power, rotor voltage, rotor current, and speed of the sample time series array numbered n. , , , , , The specific calculation formula is:

[0038] The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of them is used to obtain the standard value of the characteristic quantity of the small power motor test sample numbered n when the excitation system is operating normally. , The subscript 0 indicates that it is the standard value; The above processing was performed on 19 small power motor test samples and high power motor samples to obtain the standard value of the characteristic quantity of the small power motor test group when the excitation system is normal. ,in , and the standard value of the characteristic quantity of a high-power motor with a rated power of 1MW during normal operation of the excitation system ,The simulation results are shown in Table 1.

[0039] Table 1. Standard values ​​of characteristic quantities for small power motor test group and high power motor

[0040] S3, fitting to obtain the mapping matrix; Obtain the stator voltage standard value of the small power motor test group from the characteristic quantity standard value in S2 And the standard value of stator voltage for high-power motors , define equivalent numbers is a positive real number less than or equal to 20. The stator voltage standard value is calculated by the least squares binomial over-determined value fitting method. About Equivalent Numbers The stator voltage function , with equivalent numbers as The standard value of high power motor stator voltage is As constraints, the function Perform fitting and take the stator voltage function , the fitting formula is as follows:

[0041] in are the coefficients of the quadratic term and the linear term to be fitted, respectively. are constants, is the equivalent number corresponding to the small power motor sample, , Corresponding numbers for high-power motors; Use the equivalent number x as the sample number n, that is, x=1,2,…,19 and its corresponding stator voltage standard value Fit and solve the coefficients of the quadratic term and the linear term. Substitute the calculated quadratic and linear coefficients into the stator voltage function ,get ; The stator current, reactive power, rotor voltage, rotor current and speed are fitted with binomial over-fixed value using the same method to obtain the stator current function. ; Rotor voltage function ; Rotor current function ; Speed ​​function ; Reactive Function , define a sixth-order diagonal matrix as the mapping matrix , write the above function into the mapping matrix , diag represents the operation of constructing the elements in the brackets into a diagonal matrix; S4, obtaining test sample data of a small-power motor under specific fault conditions to form a fault sample matrix; Select the small power motor test group in step S1 to perform a specific fault type test, and record the fault time period as , n is the corresponding number of the small power motor sample, n=1,2,…,19, the values ​​of each characteristic quantity of the small power motor numbered n during the excitation system failure are monitored to obtain the characteristic quantity of the small power motor numbered n Time series of failure data , The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the ; Obtain the characteristic quantity of the low-power motor numbered n in step S2 The mean As standard value; Select time period Characteristic quantity of small power motor with internal number n Failure data Average and standard value The one with the smallest absolute value of the difference is recorded as ,Right now satisfy

[0042] Record fault time period The number of fault data in the time period is m, and the mean value of the fault data in this time period is calculated:

[0043] Then the abnormal value of the characteristic quantity obtained from the fault experiment is selected as

[0044] Take is the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed , the stator voltage of the small power motor numbered n is calculated , stator current 、Reactive , rotor voltage , rotor current , speed Outliers under specific fault conditions , , , , , , the subscript b indicates that the value is an outlier; For example, it is assumed that the abnormal values ​​obtained in a certain specific fault test are as shown in Table 2.

[0045] Table 2. Abnormal values ​​obtained in a specific fault test

[0046] Define a sixth-order diagonal matrix is the fault sample matrix, and the abnormal values ​​of the characteristic quantities under the above specific fault conditions are written into the matrix to obtain the fault sample matrix ; The above operations are repeated for 19 low-power motors in turn to obtain the fault sample matrix of the 19 low-power motors under specific fault conditions.

[0047] S5. Calculate the correlation judgment factor to distinguish the relevant feature quantities from the irrelevant feature quantities; From the fault sample matrix Obtain the abnormal values ​​of 6 characteristic quantities in 19 low-power motors under specific fault types , , , , , ; Define feature quantity Correlation judgment factor :

[0048] in, is the characteristic quantity Outliers in 19 small power motors, is the characteristic quantity of the small power motor numbered n under normal operating conditions of the excitation system The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of; If the correlation factor , it is considered that the characteristic quantity is not a representation of the fault type and is defined as an irrelevant characteristic quantity of the fault type; if , it is considered that the characteristic quantity is the characterization of this fault type and can reflect the fault state, and is defined as the relevant characteristic quantity of this fault type.

[0049] S6. Calculate the monotonicity judgment factor to distinguish between stable feature quantities and follow-up feature quantities; Defining the monotonicity factor , used to classify relevant feature quantities, where the relevant feature quantities are those that meet the correlation judgment factor in step S5 The characteristic quantity of the monotonicity judgment factor is calculated as follows:

[0050] in, Indicates the characteristic quantity of the small power motor numbered n The outliers in the fault test, is the average abnormal value of the characteristic quantities of 19 small power motors in the fault test, and the calculation formula is:

[0051] like , it is considered that the relevant characteristic quantity is a stable characteristic quantity, that is, the relevant characteristic quantity does not change with the change of motor power. If , it is considered that the relevant characteristic quantity is a follow-up characteristic quantity, that is, the relevant characteristic quantity changes with the change of motor power.

[0052] S7, calculating the fault mapping matrix; Defining Fault Mapping Coefficients , represents the characteristic quantity of the small power motor test group The mapping relationship between the fault data obtained from the fault test and the characteristic value of the corresponding fault condition of the high-power motor actually put into operation; Corresponding feature quantity The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the following defines the fault mapping coefficient The value is obtained as follows: For irrelevant feature quantities, the fault mapping coefficient The value is 0; For stable features among relevant features, the fault mapping coefficient The value is 1; For the follow-up feature in the relevant feature, assume that the fault mapping coefficient The change of conforms to the linear law, and the least square method is used to calculate its mapping coefficient, that is,

[0053] Define a sixth-order diagonal matrix is the fault mapping matrix, The values ​​are selected from the stator voltage , stator current , rotor voltage , rotor current , speed 、Reactive , and 6 characteristic quantities are calculated respectively The mapping coefficient , , , , , , the fault mapping matrix is ​​composed of the above mapping coefficients , the expression is as follows .

[0054] Calculated .

[0055] S8. Expand the fault samples using the mapping matrix and the fault mapping matrix.

[0056] Using the fault sample matrix , mapping matrix and the fault mapping matrix Perform matrix operations to obtain the expanded sample matrix , the above fault sample matrix and the mapping matrix is the function matrix, the fault mapping matrix is a constant coefficient matrix, specifically: Select any low-power motor to perform a specific fault test on the excitation system, where the low-power motor is a motor with a power less than 100 kW. Obtain the fault sample matrix according to the method in step S4. , the rated power of the motor used is , then the fault sample matrix subscript number is ; For example, Corresponding fault sample matrix Expand the fault sample, take the equivalent number x as number 1, and substitute it into the mapping matrix get , take the equivalent number x as 200 and substitute it into the mapping matrix get , perform the following matrix operations: , The non-zero elements in are the relevant characteristic quantities under this fault condition, and the actual power put into operation in the pumped storage system is The fault test data of high-power motors is used to expand the fault samples.

[0057] Calculated

[0058] From the results, it can be seen that the non-zero elements of the matrix, stator voltage, rotor voltage, rotor current, and reactive power, are relevant characteristic quantities of this type of fault and have monitoring significance. In this fault case, the stator voltage fault abnormal value is 380V, the rotor voltage fault abnormal value is 300V, the rotor current fault abnormal value is 300A, and the reactive power fault abnormal value is 1090.4kVar. The expanded fault sample information is shown in Table 3.

[0059] Table 3. Expanded fault sample information

[0060] The above abnormal values ​​can be used to formulate early warning thresholds in the monitoring process. Performing the above operation multiple times can obtain fault samples of multiple high-power motors.

[0061] Taking reactive power as an example, the extended fault samples obtained by applying the above method to the fault data in Table 2 are shown in Table 4.

[0062] Table 4. Reactive power abnormal values ​​of expanded fault samples

[0063] It can be seen from Table 4 that the reactive fault abnormality values ​​obtained by performing specific fault tests with different small power motors are different. It is difficult to directly obtain useful information from the reactive fault abnormality values ​​of small power motors. The reactive fault abnormality values ​​of small power motors are converted by the method of this patent to obtain the reactive abnormality values ​​of high power motors actually put into operation in pumped storage power stations. The values ​​tend to be stable and are closer to the reactive values ​​of high power motors actually put into operation in specific fault tests, which is more accurate and intuitive.

[0064] To sum up, the fault sample expansion method disclosed in the present invention can identify the type of feature quantity, screen out feature quantities with detection significance, and calculate the corresponding feature quantity abnormal values ​​of the high-power motor in the fault test according to the feature quantity abnormal values ​​obtained in the fault test of different small-power motors. Through multiple small-motor specific fault tests, the fault samples of the high-power motor can be expanded without destroying the high-power motor, thereby reducing the fault test cost and simplifying the detection process while retaining the detection accuracy.

[0065] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0066] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for expanding fault samples of an excitation system of a pumped storage power station, characterized in that: The fault sample expansion method comprises the following steps: S1. Acquire characteristic quantity data of a small power motor test group and characteristic quantity data of a large power motor under normal operation of an excitation system of a pumped storage power station; S2. Calculate the standard values ​​of characteristic quantities of the low-power motor test group and the high-power motor; S3, fitting to obtain the mapping matrix; S4, obtaining test sample data of a small-power motor under specific fault conditions to form a fault sample matrix; S5. Calculate the correlation judgment factor to distinguish the relevant feature quantities from the irrelevant feature quantities; S6. Calculate the monotonicity judgment factor to distinguish between stable feature quantities and follow-up feature quantities; S7, calculating the fault mapping matrix; S8. Expand the fault samples using the mapping matrix and the fault mapping matrix.

2. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 1, characterized in that: The process of step S1 is as follows: S11. Obtain characteristic quantities of the motor carried by the excitation system through a voltmeter, an ammeter, a voltage sensor, a current sensor, a tachometer and a reactive meter, including: stator voltage , stator current , rotor voltage , rotor current and speed , reactive power ; S12, select motors with rated power of 5kW, 10kW, 5nkW, ..., 95kW in descending order as the low-power motor test group, where n is the corresponding number of the small-power motor sample, n=1, 2, ..., 19; Under normal operation of the excitation system, when the excitation system of the pumped storage power station is equipped with a 5nkW motor, the motor operation data is recorded to obtain the stator voltage time series data of the excitation system under the operation of the 5nkW motor. , stator current timing data , rotor voltage timing data , rotor current timing data , speed timing data and reactive power time series data , forming a sample timing array of a small power motor numbered n under normal operating conditions of the excitation system ; S13. Under the condition of normal operation of the pumped storage excitation system, the high-power motors actually put into use in the pumped storage power station are monitored and recorded. If the rated power of the high-power motors actually put into use in the pumped storage power station is , and its corresponding feature array number is , get the timing array corresponding to the high-power motor actually put into use in the pumped storage power station .

3. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 2, characterized in that: The process of step S2 is as follows: Select a time period without missing values ​​in each feature time series array , is the time corresponding to the starting data point of the selected time segment, To select the moment corresponding to the last data point of the time segment, the average of each characteristic value of all low-power motor test samples in the segment is taken as the standard value. The specific steps are as follows: The sample time series array numbered n For example, the stator voltage in the time period The stator voltage mean is obtained by taking the internal mean value Similarly, find the mean values ​​of stator voltage, stator current, reactive power, rotor voltage, rotor current, and speed of the sample time series array numbered n. , , , , , The specific calculation formula is: The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of them is used to obtain the standard value of the characteristic quantity of the small power motor test sample numbered n when the excitation system is operating normally. , The subscript 0 indicates that it is the standard value; The above processing was performed on 19 small power motor test samples and high power motor samples to obtain the standard value of the characteristic quantity of the small power motor test group when the excitation system is normal. ,in , and the standard value of the characteristic quantity of high-power motors when the excitation system is operating normally .

4. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 3, characterized in that: The process of step S3 is as follows: Obtain the stator voltage standard value of the small power motor test group from the characteristic quantity standard value obtained in step S2 And the standard value of stator voltage for high-power motors , define equivalent numbers is a positive real number less than or equal to 20. The stator voltage standard value is calculated by the least squares binomial over-determined value fitting method. About Equivalent Numbers The stator voltage function , with equivalent numbers as The standard value of high power motor stator voltage is As constraints, the function Perform fitting and take the stator voltage function , the fitting formula is as follows: in are the coefficients of the quadratic term and the linear term to be fitted, respectively. are constants, is the equivalent number corresponding to the small power motor sample, , Corresponding numbers for high-power motors; Use the equivalent number x as the sample number n, that is, x=1,2,…,19 and its corresponding stator voltage standard value Fit and solve the quadratic term coefficient and the linear term coefficient, and substitute the calculated quadratic term coefficient and the linear term coefficient into the stator voltage function ; The stator current, reactive power, rotor voltage, rotor current and speed are fitted with binomial over-fixed value using the same method to obtain the stator current function. , rotor voltage function , rotor current function , speed function and reactive function , define a sixth-order diagonal matrix as the mapping matrix , write the above function into the mapping matrix , diag represents the operation of constructing the elements in the brackets into a diagonal matrix; Substituting any positive real number x less than or equal to 20 into the mapping matrix, the rated power can be obtained as The measured mean value matrix of each characteristic quantity of the motor in the normal operation state of the excitation system , The rated power is The measured mean value of the stator voltage of the motor in the normal operation state of the excitation system, The rated power is The measured mean value of the stator current of the motor in the normal operation state of the excitation system, The rated power is The measured mean value of the motor rotor voltage in the normal operation state of the excitation system, The rated power is The measured mean value of the motor rotor current in the normal operation state of the excitation system, The rated power is The measured average value of the motor speed in the normal operation state of the excitation system, The rated power is The measured mean value of the motor's reactive power under normal operating conditions of the excitation system.

5. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 4, characterized in that: The process of step S4 is as follows: Select the small power motor test group in step S1 to perform a specific fault type test, and record the fault time period as , n is the corresponding number of the small power motor sample, n=1,2,…,19, the values ​​of each characteristic quantity of the small power motor numbered n during the excitation system failure are monitored to obtain the characteristic quantity of the small power motor numbered n Time series of failure data , The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the ; Obtain the characteristic quantity of the low-power motor numbered n in step S2 The mean As standard value; Select time period Characteristic quantity of small power motor with internal number n Failure data Average and standard value The one with the smallest absolute value of the difference is recorded as ,Right now satisfy Record fault time period The number of fault data in the time period is m, and the mean value of the fault data in this time period is calculated: Then the abnormal value of the characteristic quantity obtained from the fault experiment is selected as Take is the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed , the stator voltage of the small power motor numbered n is calculated , stator current 、Reactive , rotor voltage , rotor current , speed Outliers under specific fault conditions , , , , , , the subscript b indicates that the value is an outlier; Define a sixth-order diagonal matrix is the fault sample matrix, and the abnormal values ​​of the characteristic quantities under the above specific fault conditions are written into the matrix to obtain the fault sample matrix ; The above operations are repeated for 19 low-power motors in turn to obtain the fault sample matrix of the 19 low-power motors under specific fault conditions.

6. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 5, characterized in that: The process of step S5 is as follows: From the fault sample matrix Obtain the abnormal values ​​of 6 characteristic quantities in 19 low-power motors under specific fault types , , , , , ; Define feature quantity Correlation judgment factor : in, is the characteristic quantity Outliers in 19 small power motors, is the characteristic quantity of the small power motor numbered n under normal operating conditions of the excitation system The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of; If the correlation factor , it is considered that the characteristic quantity is not a representation of the fault type and is defined as an irrelevant characteristic quantity of the fault type; if , it is considered that the characteristic quantity is the characterization of this fault type and can reflect the fault state, and is defined as the relevant characteristic quantity of this fault type.

7. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 6, characterized in that: The process of step S6 is as follows: Defining the monotonicity factor , used to classify relevant feature quantities, where the relevant feature quantities are those that meet the correlation judgment factor in step S5 The characteristic quantity of the monotonicity judgment factor is calculated as follows: in, Indicates the characteristic quantity of the small power motor numbered n Outliers in the fault test, is the average abnormal value of the characteristic quantities of 19 small power motors in the fault test, and the calculation formula is: ; like , it is considered that the relevant characteristic quantity is a stable characteristic quantity, that is, the relevant characteristic quantity does not change with the rated power of the motor. If , it is considered that the relevant characteristic quantity is a follow-up characteristic quantity, that is, the relevant characteristic quantity changes with the rated power of the motor.

8. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 7, characterized in that: The process of step S7 is as follows: Defining Fault Mapping Coefficients , represents the characteristic quantity of the small power motor test group The mapping relationship between the fault data obtained from the fault test and the characteristic value of the corresponding fault condition of the high-power motor actually put into operation; Corresponding feature quantity The standard value of The value is selected from the stator voltage , stator current 、Reactive , rotor voltage , rotor current , speed One of the following defines the fault mapping coefficient The value is obtained as follows: For irrelevant feature quantities, the fault mapping coefficient The value is 0; For stable features among relevant features, the fault mapping coefficient The value is 1; For the follow-up feature in the relevant feature, assume that the fault mapping coefficient The change of conforms to the linear law, and the least square method is used to calculate its mapping coefficient, that is, , define a sixth-order diagonal matrix is the fault mapping matrix, The values ​​are selected from the stator voltage , stator current , rotor voltage , rotor current , speed 、Reactive , and 6 characteristic quantities are calculated respectively The mapping coefficient , , , , , , the fault mapping matrix is ​​composed of the above mapping coefficients , the expression is as follows .

9. A method for expanding fault samples of an excitation system of a pumped storage power station according to claim 8, characterized in that: The process of step S8 is as follows: Using the fault sample matrix , mapping matrix and the fault mapping matrix Perform matrix operations to obtain the expanded sample matrix , the above fault sample matrix and the mapping matrix is the function matrix, the fault mapping matrix is a constant coefficient matrix, specifically: Select a small power motor for the excitation system specific fault test, where the small power motor is a motor with a power less than 100 kW. Obtain the fault sample matrix according to the method in step S4. , the rated power of the motor used is , then the fault sample matrix subscript number is ; Take the equivalent number x as the number , substitute into the mapping matrix get , take the equivalent number x as the rated power of the pumped storage power station actually in operation The corresponding number of high-power motor , substitute into the mapping matrix get , perform the following matrix operations: , The non-zero elements in are the relevant characteristic quantities under this fault condition, and the actual power put into operation in the pumped storage system is The fault test data of high-power motors is used to expand the fault samples.

Images