CN115455358A - Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model - Google Patents
Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model Download PDFInfo
- Publication number
- CN115455358A CN115455358A CN202210846263.9A CN202210846263A CN115455358A CN 115455358 A CN115455358 A CN 115455358A CN 202210846263 A CN202210846263 A CN 202210846263A CN 115455358 A CN115455358 A CN 115455358A
- Authority
- CN
- China
- Prior art keywords
- active power
- data
- model
- early warning
- water head
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000003745 diagnosis Methods 0.000 title claims abstract description 39
- 238000004458 analytical method Methods 0.000 claims abstract description 25
- 230000008569 process Effects 0.000 claims abstract description 22
- 230000002159 abnormal effect Effects 0.000 claims abstract description 12
- 238000000611 regression analysis Methods 0.000 claims abstract description 11
- 230000007704 transition Effects 0.000 claims abstract description 8
- 238000012216 screening Methods 0.000 claims abstract description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 44
- 238000010219 correlation analysis Methods 0.000 claims description 15
- 238000004519 manufacturing process Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 8
- 238000007619 statistical method Methods 0.000 claims description 7
- 230000001419 dependent effect Effects 0.000 claims description 4
- 238000012821 model calculation Methods 0.000 claims description 4
- 230000000875 corresponding effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000005856 abnormality Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000035772 mutation Effects 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000013024 troubleshooting Methods 0.000 description 1
- 238000010977 unit operation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Economics (AREA)
- Mathematical Physics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to an electrical parameter trend early warning and fault diagnosis method based on a nonlinear regression model, which is used for trend early warning and fault diagnosis analysis of active power, stator voltage, stator current, rotor voltage and rotor current of a hydropower station and comprises the following steps: acquiring strong correlation parameters of an analysis object in the running process of a unit; acquiring actual data in the actual operation process of the unit; the data volume covers the minimum and maximum ranges of the model related parameters in the procedure, and the same group of data is at the same time; screening the collected historical operating data, and removing abnormal data of a transition state in the data to obtain typical steady-state operating condition data of the unit; performing data modeling based on the data obtained in the step 3, and finally determining an optimal theoretical model by comparing multiple regression analysis equation results; setting a threshold value; alarm judgment; and (5) fault diagnosis. The invention can realize the trend early warning and fault diagnosis of active power, stator voltage, stator current, rotor voltage and rotor current of the hydropower station.
Description
Technical Field
The invention relates to the technical field of hydroelectric power generation, in particular to an electric parameter trend early warning and fault diagnosis method based on a nonlinear regression model.
Background
Under different working conditions of the hydraulic power plant, the operation characteristics and parameters are different, the number of associated parameters is large, and the change trend of the current operation working condition and the normal working condition cannot be judged by manpower basically; and the traditional monitoring system of the hydraulic power plant only has a limit value alarm and no trend analysis function.
Disclosure of Invention
The invention aims to provide an electric parameter trend early warning and fault diagnosis method based on a nonlinear regression model.
The invention provides an electrical parameter trend early warning and fault diagnosis method based on a nonlinear regression model, which is used for trend early warning and fault diagnosis analysis of active power, stator voltage, stator current, rotor voltage and rotor current of a hydropower station and comprises the following steps:
step 1, acquiring strong correlation parameters of an analysis object in the running process of a unit;
step 2, collecting actual data in the actual operation process of the unit; the data volume of the data covers the minimum and maximum ranges of the model related parameters in the regulation, and the same group of data is at the same time;
step 3, screening the collected historical operation data, removing abnormal data of a transition state in the data, and obtaining typical steady-state operation condition data of the unit;
step 4, performing data modeling based on the data obtained in the step 3, and finally determining an optimal theoretical model by comparing multiple regression analysis equation results;
step 5, calculating historical data theoretical error rate e = (n) Theory of the invention -n Real time )/n Real-time Carrying out statistical analysis on the error rate e to obtain a threshold value;
step 6, calculating the predicted error rate e = (n) at the same moment in real time Theory of the invention -n Real-time )/n Real-time Carrying out trend early warning on the parameter model according to a preset alarm threshold value;
and 7, acquiring the relevant parameter condition based on the expert experience diagnosis model, transversely comparing and analyzing the state of the multi-parameter fault when the multi-parameter fault occurs, and judging the fault direction and type.
Further, the hydropower station active power trend early warning and fault diagnosis analysis steps are as follows:
1) Obtaining parameters with strong active power correlation in the operation process of the unit, wherein the parameters comprise active power, guide vane opening and working water head, carrying out correlation analysis on three factors of the active power, the guide vane opening and the water head, and verifying the reliability of the correlation parameters;
2) Acquiring actual data in the actual operation process of the unit, wherein the acquired data quantity covers the rule or the minimum and maximum ranges of active power, water head and guide vane opening in the actual operation, and the water head and guide vane opening data corresponding to each group of active power are at the same moment;
3) Screening and processing abnormal data in a transition state in the acquired data, and deleting numerical values in a non-steady operation state to obtain range data of normal and steady operation states of the active power, the water head and the opening degree of the guide vane of the unit;
4) An active power characteristic model is established by adopting a binary regression analysis mode, and the expression of the active power characteristic model of the model is determined as follows according to the regression coefficient estimation value:
P active power ＝-247.7+7.579*N Opening degree -4.046*H Water head -0.063*N Opening degree ^2+0.09031*N Opening degree *H Water head +0.03347*H Water head ^2; wherein the dependent variable is the active power P Active power The independent variable being the opening N of the guide vane Opening degree And head H Water head ；
5) N to be collected Opening degree And H Water head Respectively importing the data into the established active power characteristic model formula to calculate the theoretical value P of the active power Theory of the invention The active power value P is compared with the active power value P at the same moment Real-time Comparing and calculating the error rate e = (P) Theory of the invention -P Real-time )/P Real time Carrying out statistical analysis on the error rate e to obtain a proper threshold, setting the threshold according to 98% of percentile of the model statistical error rate, rounding to 4%, and carrying out primary early warning threshold L First stage Set to 4%, a secondary early warning threshold L Second stage Set to 4% by 2;
6) N at the same time to be generated in real time Opening degree And H Water head The data is brought into an active power characteristic model to generate an active power theoretical value P Theory of the invention And calculating an error rate e = (P) Theory of the invention -P Real-time )/P Real-time According to a preset alarm threshold value, performing automatic machine early warning on the active power characteristic model;
7) After the active power measuring point is early-warned, whether the corresponding parameters of the direction of the fault are normal or not is automatically judged according to the expert experience diagnosis model, if one parameter is abnormal, the fault is judged to be abnormal, and the fault diagnosis process is displayed to production personnel.
Further, step 6) comprises continuous alarm setting, namely setting the error rate of each active power model calculation as e 1 ,e 2 ,e 3 ,e 4 ,e 5 ,……e n E.g. satisfying e "L for multiple successive error rates First stage Performing active power primary alarm; e.g. any one of which error rates e<L First level Then, the number of times of recounting from this point is 0 until the next occurrence of e "L First stage Starting counting; IIThe level early warning setting mode is consistent with the level early warning, and the threshold value is adjusted to be 4% x 2.
By means of the scheme, the electrical parameter trend early warning and fault diagnosis method based on the nonlinear regression model has the following technical effects:
1) The method can accurately and quickly carry out early warning on the deviation of the electrical parameters of the hydraulic power plant such as active power and the like, and judge the fault direction. The early warning frequency can be adjusted at will according to the requirements of the user.
2) The collected historical data are not directly subjected to modeling analysis, but are scientifically screened to remove unstable-state data, so that the modeled data model can more accurately reflect the current operation condition.
3) The continuous counting setting of the alarm logic effectively avoids data misinformation caused by actual data mutation in production, and improves the early warning accuracy.
4) The threshold setting and selecting mode is more practical, the defect that the early warning is insensitive or misinformed due to too high threshold setting or too low threshold setting is effectively avoided, and the misinformation rate of an early warning model is reduced.
5) The fault diagnosis depends on the early warning result, and an expert fault analysis model is integrated, so that the state of the associated parameters can be quickly positioned, manual query and analysis are not needed, a large amount of human resources and fault troubleshooting time are saved, and the method has high economic benefit.
6) The model analysis mode is suitable for different units and different types of units, customized early warning models are generated according to the operation conditions of the different units, and the early warning result deviation is not large due to the fact that only one formula is applied.
The foregoing description is only an overview of the technical solutions of the present invention, and in order to make the technical solutions of the present invention more clearly understood and to implement them in accordance with the contents of the description, the following detailed description is given with reference to the preferred embodiments of the present invention and the accompanying drawings.
Drawings
FIG. 1 is a plot of active power, vane opening, head scatter plot for an embodiment of the present invention;
FIG. 2 is a scatter plot of active power, vane opening in an embodiment of the present invention;
FIG. 3 is a plot of active power, head scatter, in accordance with an embodiment of the present invention;
FIG. 4 is a three-dimensional mechanism model of active power in an embodiment of the invention;
FIG. 5 is a normal distribution diagram of the active power error rate according to an embodiment of the present invention;
FIG. 6 is a logic diagram for active power fault diagnosis in one embodiment of the present invention;
FIG. 7 is a stator voltage/current fault diagnostic logic diagram in accordance with an embodiment of the present invention;
fig. 8 is rotor voltage/current fault diagnostic logic in an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
The embodiment provides an electrical parameter trend early warning and fault diagnosis method based on a nonlinear regression model, the analysis mode is suitable for trend early warning and fault diagnosis analysis of electrical parameters such as active power, stator voltage, stator current, rotor voltage, rotor current and the like of a hydropower station, and the whole trend early warning and fault diagnosis process is roughly divided into the following steps:
step 1 is correlation analysis; and searching for strong correlation parameters of the analysis object in the running process of the unit according to the working experience.
Step 2, data collection; the method has the advantages that the actual data in the actual operation process of the unit is collected, the data size needs to be as much as possible, the minimum and maximum ranges of the model related parameters in the rules need to be covered, the same group of data needs to be at the same time, the data size needed by modeling is guaranteed to be accurate enough, and the coverage is more comprehensive.
Step 3, data processing; and screening abnormal data of a transition state in the collected historical operation data, and ensuring that the modeling data is a typical steady-state operation condition of the unit.
Step 4, modeling data; and finally determining the optimal theoretical model by comparing the results of the multiple regression analysis equations.
Step 5, setting a threshold value; to select a suitable threshold, the theoretical error rate e = (n) of historical data is calculated Theory of the invention -n Real-time )/n Real time And then carrying out statistical analysis on the error rate e to obtain a proper threshold value.
Step 6, alarm judgment; real-time calculation of the same-time prediction error rate e = (n) Theory of the invention -n Real time )/n Real time . And then realizing the trend early warning function of the parameter model according to the preset alarm threshold value.
Step 7, fault diagnosis; the expert experience diagnosis model is integrated in each parameter early warning, the relevant parameter condition can be quickly checked, the state of multi-parameter faults when occurring is transversely contrasted and analyzed, the fault direction and type are quickly judged, and the quick positioning of the fault direction by production personnel is assisted.
3. Early warning process
Taking the 'active power' parameter of a 350MW mixed-flow type No. 1 unit of a certain hydropower station as an example, the specific analysis process is as follows:
(ii) correlation analysis
Through working experience, in the known unit operation process, parameters strongly correlated with active power are 'guide vane opening degree' and 'working head', and correlation analysis and verification are performed on the parameters. And carrying out correlation analysis on three factors of active power, guide vane opening and water head, and verifying the reliability of the correlation parameters, as shown in table 1.
TABLE 1 Power characteristics correlation analysis
In table 1, the correlations between the active power and the opening degree of the guide vane and the water head are respectively 0.389 and 0.314, which shows that the power and the opening degree of the guide vane and the water head are positively correlated, the significance is less than 0.05, and the statistical significance is satisfied.
And analyzing the active power, the opening degree of the guide vane and the water head by a discrete graph, as shown in figure 1.
Through the analysis of the graph 1, the active power, the opening degree of the guide vane and the water head have a certain linear relation and basically accord with correlation analysis data.
(II) data acquisition
In order to ensure the integrity and correctness of the modeling data, the data volume needs to cover as much as possible the minimum and maximum ranges of active power, head and guide vane opening in the specification or actual operation. The method is characterized in that the water head and the guide vane opening data corresponding to each group of active power need to be ensured at the same time (the minimum steady active power of the hydraulic generator in the actual operation is 170MW, the maximum steady active power is 350MW, the minimum actual operation of the guide vane opening is about 50% and the maximum actual operation is about 95%, the minimum specified by a water head regulation is 52m and the maximum specified by a water head regulation is 81 m). The model selects a plurality of groups of data of different time periods, different water heads, guide vane openness and active power of a unit No. 1 of a certain hydropower station in one year.
(III) data processing
The premise of data modeling is that data in a stable running state is needed, so abnormal data in a transition state in the acquired data needs to be screened and processed, and the numerical value in a non-stable running state is deleted. According to a power generator with the model SF350-70/15880 of a penning hydropower station, the rated power is 350MW, the normal stable state basically runs above 170MW and below 350MW, and therefore, the numerical value that the active power is larger than 350MW and smaller than 170MW in the group of data is deleted. The data ranges of the active power, the water head and the guide vane opening degree after final processing are shown in table 2, and the data ranges basically cover the normal and stable running state ranges of the active power, the water head and the guide vane opening degree of a No. 1 unit of a certain hydropower station.
TABLE 2 Power model analysis sampling parameters
| Number of cases | Minimum value of | Maximum value | |
| Active power | 73551 | 170.00 | 350.00 |
| Water head | 73551 | 54.68 | 78.29 |
| Opening degree of guide vane | 73551 | 49.83 | 95.33 |
Mathematical modeling
The modeling method adopts a binary regression analysis mode. Dependent variable being active power P Active power The independent variable being the opening N of the guide vane Opening degree And head H Water head . The trend changes of the dependent variable and the independent variable are observed through a scatter diagram.
As can be seen from fig. 2 and 3, the linear relationship between the guide vanes and the water head and the active power is not strong, and the linear relationship is also needed to be analyzed.
And (5) comparing the test results of different nonlinear regression analysis formulas by using matlab. Finally determining a quadratic polynomial nonlinear regression analysis mechanism model:
P active power ＝p00+p10*N Opening degree +p01*H Water head +p20*N Opening degree ^2+p11*N Opening degree *H Water head +p02*H Water head ^2
And then calculating the regression coefficient of the formula by using a least square method. The calculation results and the test results are shown in tables 3 and 4:
TABLE 3 results of regression analysis
| Regression coefficient | Regression coefficient estimation | Regression coefficient confidence interval |
| p00 | -247.7 | (-255.3,-240.1) |
| p10 | 7.579 | (7.492,7.665) |
| p01 | -4.046 | (-4.191,-3.902) |
| p20 | -0.063 | (-0.06328,-0.06272) |
| p11 | 0.09031 | (0.08957,0.09106) |
| p02 | 0.03347 | (0.03272,0.03421) |
Table 4 active power model test results
| R-square (coefficient of determination) | SSE (sum of squared error) | RMSE (root mean square error) | |
| Test results | 0.9951 | 4.619e+05 | 2.506 |
From Table 4, it can be seen that the R-square (coefficient of identity) is 0.9951, which has very small SSE and RMSE values, indicating that the model fits very well. And finally, determining the expression of the active power characteristic model of the model according to the regression coefficient estimation value as follows:
P active power ＝-247.7+7.579*N Opening degree -4.046*H Water head -0.063*N Opening degree ^2+0.09031*N Opening degree *H Water head +0.03347*H Water head ^2。
The active power three-dimensional mechanism model diagram is shown in fig. 4.
(V) threshold setting
The threshold setting is very critical, and if the threshold is set to be larger, the early warning sensitivity is reduced, and the purpose of trend early warning cannot be achieved. If the threshold value is set to be smaller, the sensitivity is improved, but false alarm can be frequently carried out, and the normal monitoring of production personnel is influenced.
To select the appropriate threshold, the previous sampling is performedSet 73551 group N Opening degree And H Water head Respectively importing the data into the active power characteristic model formula which is just established, and calculating the theoretical value P of the active power Theory of the invention At the same time as the active power value P Real-time Comparing and calculating the error rate e = (P) Theory of the invention -P Real time )/P Real time . Statistical analysis was performed on the error rate e, the results of which are shown in table 5 and fig. 5:
TABLE 5 active Power error Rate statistics
| Mean value of | Median number | Minimum value | | Percentile | 95% | Percentile 98% | |
| Active power error rate (unit:%) | -0.916 | 0.0186 | -20.43 | 29.47 | 2.3997 | 3.5577 |
As can be seen from table 5 and fig. 5, the active power error rate e is normally distributed, and the average error rate and the median of the error rate are low. In order from small to large, the numbers ranked at the 98 th% of the percentile are only 3.5577%, and are all within the error acceptance range. To avoid outliers in the error rate e affecting the threshold setting, the model threshold is set as 98% of the percentile of the statistical error rate of the model, rounded to 4%. Meanwhile, in order to distinguish the alarm grade, a primary early warning threshold value L First stage Set to 4%, a secondary early warning threshold L Second stage Set to 4% 2.
(VI) alarm judgment
The overall alarm logic is N at the same time to be generated in real time Opening degree And H Water head The data is brought into an active power characteristic model to generate an active power theoretical value P Theory of the invention And calculating an error rate e = (P) Theory of the invention -P Real-time )/P Real time . And then, according to the alarm threshold value set in the past, carrying out automatic machine early warning on the active power model.
However, when the actual production runs, the data can be in a transition stage, and the possibility of data mutation exists. Therefore, continuous alarm setting is added into the alarm logic, and the setting can effectively reduce the false alarm rate caused by data mutation. The alarm logic is as follows:
setting the error rate of each active power model calculation as e 1 ,e 2 ,e 3 ,e 4 ,e 5 ,……e n If e is provided 1 ,e 2 ,e 3 ,e 4 ,e 5 The error rates of 5 times in succession satisfy e [ ] L First stage The system will present a primary alarm of active power; e.g. any one of which error rates e<L First stage Then, the number of times is counted again from that point to 0 until the next occurrence of e "L First level The counting is started. The setting mode of the secondary early warning is consistent with that of the primary early warning, only the threshold value is adjusted to be 4% by 2,
the active power characteristic model has the advantage of flexibility, namely the model calculation frequency and the alarm frequency are set randomly according to requirements. The operation characteristics of the unit and the early warning necessity of the active power are considered, the calculation frequency is set to be 1 time in 1 minute, and early warning of the active power in the minute level is achieved.
(VII) Fault diagnosis
The active power abnormality is only a fault phenomenon, but the fault reason needs to be searched through field inspection and parameter judgment. The traditional mode is that production personnel inquire associated parameter data one by one according to experience and manually judge the fault direction. This approach is both time consuming and can affect the failure analysis results due to the level of individual skill.
The fault diagnosis method integrates an expert experience diagnosis model, can quickly check the condition of the associated parameters, contrasts and analyzes the state of the multi-parameter fault when the multi-parameter fault occurs, quickly judges the fault direction and type, and assists production personnel to quickly position the fault direction. The decision logic and specific decision parameters are shown in fig. 6.
Taking the "active power" fault diagnosis shown in fig. 6 as an example, the active power abnormality may be several fault directions, such as barrier blockage, stator and rotor circuit abnormality, governor fault, and measurement point fault. After the active power measuring point is early-warned, the system can automatically judge whether the corresponding parameters of the direction to be failed are normal or not according to the set expert experience diagnosis model, if one parameter is abnormal, the failure abnormality can be diagnosed, the failure diagnosis process is shown to production personnel, and the production personnel can be assisted to quickly locate the failure direction.
4. Other parameter analysis procedures
Stator voltage U ab Stator current I a Rotor voltage U Rotor Rotor current I Rotor (three-phase voltage and current of the generator stator are the same, so only one phase is taken) parameter analysis steps and methods are similar to active power. The specific analysis process is as follows:
(ii) correlation analysis
TABLE 6 stator Voltage U ab Correlation analysis
TABLE 7 stator Current Ia correlation analysis
TABLE 8 rotor Voltage U Rotor Correlation analysis
TABLE 9 rotor Current I Rotor Correlation analysis
The 'active power' part of the correlation analysis process is analyzed in detail, only the correlation analysis result is shown here, and as can be seen from tables 6 to 9, the assumed correlation parameters of the four parameters, namely active P and reactive q, and the Pearson correlation coefficient meet the requirements, and the significance is less than 0.05, so that the statistical significance is met.
(II) data acquisition and processing
The collecting and processing processes are similar to the active power and are not described in detail, and finally, the results after collecting and processing are shown in table 10:
TABLE 10 model analysis of sample parameters
Mathematical modeling
By utilizing nonlinear regression analysis and comparing and analyzing different analysis models, an optimal mathematical model is obtained, and a model equation is as follows:
Uab＝17.44+0.001009*p+0.007273*q-0.000003416*p^2+0.0000003191*p*q+0.000004803*q^2
Ia＝260.3+30.95*p+1.1*q+0.003796*p^2-0.01856*p*q+0.05265*q^2
U rotor ＝140.2+0.187*p+0.7136*q+0.0001514*p^2-0.0005464*p*q+0.001025*q^2
I Rotor ＝1132+0.8595*p+4.805*q+0.001944*p^2-0.004087*p*q+0.003895*q^2。
TABLE 11 examination of model parameters
As can be seen from Table 11, the R-square (coefficient of identity) for each parametric model was greater than 0.99, and the SSE and RMSE values were small, indicating that the fitness of each parametric model was very high.
(IV) threshold setting
The threshold setting is similar to the active power, data needs to be respectively imported into the optimal mathematical model formula which is just established, then the corresponding parameter error rate e is calculated, statistical analysis is carried out on the error rate e, and the result is shown in table 12:
TABLE 12 statistical results of errors of the parameters
As can be seen from Table 12, the mean error rate, the median, and the 98% percentile of the parameters are all low and within the error acceptance range, and meet the specified values of the operating regulations. To avoid outliers in the error rate e affecting the threshold settings, the model threshold is set as 98% of the percentile of the statistical error rate of the model, and the parameter thresholds are shown in table 13:
TABLE 13 early warning thresholds for each parameter
| Name of model | First-level early warning threshold | Second-level early warning threshold |
| Uab | 1％ | 2％ |
| Ia | 5％ | 10％ |
| U Rotor | 5％ | 10％ |
| I Rotor | 3％ | 6％ |
(V) alarm logic
The alarm logic is consistent with the continuous alarm logic of the active power setting and will not be described repeatedly.
(VI) Fault diagnosis
Stator voltage U ab Stator current I a Rotor voltage U Rotor Rotor current I Rotor The whole diagnosis method is similar to the active power, but the fault direction judged by each model is different, the criterion is different, the stator voltage and the stator current are the same judgment logic, the rotor voltage and the rotor current are the same judgment logic, and the specific diagnosis logic is shown in fig. 7 and 8.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, it should be noted that, for those skilled in the art, many modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (3)
1. A method for trend early warning and fault diagnosis of electrical parameters based on a nonlinear regression model is characterized by being used for trend early warning and fault diagnosis analysis of active power, stator voltage, stator current, rotor voltage and rotor current of a hydropower station, and comprising the following steps of:
step 1, acquiring strong correlation parameters of an analysis object in the running process of a unit;
step 2, collecting actual data in the actual operation process of the unit; the data volume of the data covers the minimum and maximum ranges of the model related parameters in the regulation, and the same group of data is at the same time;
step 3, screening the collected historical operation data, removing abnormal data of a transition state in the data, and obtaining typical steady-state operation condition data of the unit;
step 4, carrying out data modeling based on the data obtained in the step 3, and finally determining an optimal theoretical model by comparing multiple regression analysis equation results;
step 5, calculating historical data theoretical error rate e = (n) Theory of the invention -n Real time )/n Real-time Carrying out statistical analysis on the error rate e to obtain a threshold value;
step 6, real-time calculating the predicted error rate e = (n) at the same moment Theory of the invention -n Real time )/n Real-time Carrying out trend early warning on the parameter model according to a preset alarm threshold value;
and 7, acquiring the relevant parameter condition based on the expert experience diagnosis model, transversely comparing and analyzing the state of the multi-parameter fault when the multi-parameter fault occurs, and judging the fault direction and type.
2. The nonlinear regression model-based electrical parameter trend early warning and fault diagnosis method as claimed in claim 1, wherein the hydropower station active power trend early warning and fault diagnosis analysis steps are as follows:
1) Obtaining parameters with strong active power correlation in the operation process of the unit, wherein the parameters comprise active power, guide vane opening and working water head, carrying out correlation analysis on three factors of the active power, the guide vane opening and the water head, and verifying the reliability of the correlation parameters;
2) Acquiring actual data in the actual operation process of the unit, wherein the acquired data quantity covers the rule or the minimum and maximum ranges of active power, water head and guide vane opening in the actual operation, and the water head and guide vane opening data corresponding to each group of active power are at the same moment;
3) Screening and processing abnormal data in a transition state in the acquired data, and deleting numerical values in a non-steady operation state to obtain range data of normal and steady operation states of the active power, the water head and the opening degree of the guide vane of the unit;
4) An active power characteristic model is established by adopting a binary regression analysis mode, and the expression of the active power characteristic model of the model is determined as follows according to the regression coefficient estimation value:
P active power ＝-247.7+7.579*N Opening degree -4.046*H Water head -0.063*N Opening degree ^2+0.09031*N Opening degree *H Water head +0.03347*H Water head 2, a group B; wherein the dependent variable is the active power P Active power The independent variable being the opening N of the guide vane Opening degree And head H Water head ；
5) N to be collected Opening degree And H Water head Respectively importing the data into the established active power characteristic model formula to calculate the theoretical value P of the active power Theory of the invention The active power value P is compared with the active power value P at the same moment Real time Comparing and calculating the error rate e = (P) Theory of the invention -P Real time )/P Real time Carrying out statistical analysis on the error rate e to obtain a proper threshold, setting the threshold according to 98% of percentile of the model statistical error rate, rounding to 4%, and carrying out primary early warning threshold L First stage Set to 4%, a secondary early warning threshold L Second stage Set to 4% x 2;
6) N at the same time to be generated in real time Opening degree And H Water head The data is brought into an active power characteristic model to generate an active power theoretical value P Theory of the invention And calculating an error rate e = (P) Theory of the invention -P Real time )/P Real time According to a preset alarm threshold value, the active power characteristic model is subjected to automatic machine pre-predictionAlarming;
7) After the active power measuring point is early-warned, whether the corresponding parameters of the direction of the fault are normal or not is automatically judged according to the expert experience diagnosis model, if one parameter is abnormal, the fault is judged to be abnormal, and the fault diagnosis process is displayed to production personnel.
3. The nonlinear regression model-based electrical parameter trend pre-warning and fault diagnosis method as claimed in claim 2, wherein the step 6) comprises a continuous alarm setting, that is, setting the error rate of each active power model calculation as e 1 ,e 2 ,e 3 ,e 4 ,e 5 ,……e n E.g. satisfying e "L for multiple successive error rates First stage Performing active power primary alarm; e.g. any one of which error rates e<L First level Then, the number of times is counted again from that point to 0 until the next occurrence of e "L First level Starting to count; the secondary early warning setting mode is consistent with the primary early warning, and the threshold value is adjusted to be 4% x 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210846263.9A CN115455358A (en) | 2022-07-19 | 2022-07-19 | Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210846263.9A CN115455358A (en) | 2022-07-19 | 2022-07-19 | Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115455358A true CN115455358A (en) | 2022-12-09 |
Family
ID=84296350
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210846263.9A Pending CN115455358A (en) | 2022-07-19 | 2022-07-19 | Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115455358A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116662761A (en) * | 2023-06-28 | 2023-08-29 | 广州发展南沙电力有限公司 | Fuel gas power station important parameter early warning method and system based on data analysis |
| CN116990465A (en) * | 2023-09-25 | 2023-11-03 | 北京金水永利科技有限公司 | Air quality data abnormity early warning method and system thereof |
| CN117688834A (en) * | 2023-12-11 | 2024-03-12 | 北京京能能源技术研究有限责任公司 | Method for early warning insulation overheat fault of coil of water-hydrogen turbine generator |
-
2022
- 2022-07-19 CN CN202210846263.9A patent/CN115455358A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116662761A (en) * | 2023-06-28 | 2023-08-29 | 广州发展南沙电力有限公司 | Fuel gas power station important parameter early warning method and system based on data analysis |
| CN116662761B (en) * | 2023-06-28 | 2024-05-14 | 广州发展南沙电力有限公司 | Fuel gas power station important parameter early warning method and system based on data analysis |
| CN116990465A (en) * | 2023-09-25 | 2023-11-03 | 北京金水永利科技有限公司 | Air quality data abnormity early warning method and system thereof |
| CN116990465B (en) * | 2023-09-25 | 2023-12-19 | 北京金水永利科技有限公司 | Air quality data abnormity early warning method and system thereof |
| CN117688834A (en) * | 2023-12-11 | 2024-03-12 | 北京京能能源技术研究有限责任公司 | Method for early warning insulation overheat fault of coil of water-hydrogen turbine generator |
| CN117688834B (en) * | 2023-12-11 | 2024-05-24 | 北京京能能源技术研究有限责任公司 | Method for early warning insulation overheat fault of coil of water-hydrogen turbine generator |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115455358A (en) | Electrical parameter trend early warning and fault diagnosis method based on nonlinear regression model | |
| CN108375476B (en) | Hydroelectric generating set health assessment method | |
| CN112505549B (en) | New energy automobile battery abnormity detection method based on isolated forest algorithm | |
| CN110008565A (en) | A kind of industrial process unusual service condition prediction technique based on operating parameter association analysis | |
| CN113032454A (en) | Interactive user power consumption abnormity monitoring and early warning management cloud platform based on cloud computing | |
| CN117060409B (en) | Automatic detection and analysis method and system for power line running state | |
| CN113239132B (en) | Online out-of-tolerance identification method for voltage transformer | |
| GB2476246A (en) | Diagnosing an operation mode of a machine | |
| CN109061387A (en) | A kind of power grid abnormality method of discrimination based on annulus theorem | |
| CN114781476B (en) | System and method for analyzing faults of measuring equipment | |
| CN111126759A (en) | Electric energy meter state evaluation method based on abnormal event fault correlation degree | |
| CN111929579B (en) | Generator online fault diagnosis method and device and computer device | |
| CN115685045B (en) | Online evaluation method for voltage transformer | |
| CN112926656A (en) | Method, system and equipment for predicting state of circulating water pump of nuclear power plant | |
| CN115614292A (en) | Vibration monitoring device and method for vertical water pump unit | |
| CN113987817A (en) | Degradation analysis method for power generation equipment of hydraulic power plant | |
| US11339763B2 (en) | Method for windmill farm monitoring | |
| CN115133659A (en) | On-line monitoring method and system for low-voltage line loss of low-voltage distribution network | |
| CN115809805A (en) | Power grid multi-source data processing method based on edge calculation | |
| CN115034094A (en) | Prediction method and system for operation state of metal processing machine tool | |
| CN114970311A (en) | Method for establishing remote module life prediction model and life prediction method | |
| CN115326393A (en) | Wind turbine generator bearing pair fault diagnosis method based on temperature information | |
| CN114442543A (en) | Computer monitoring method suitable for early warning of hydropower station fault | |
| CN112380487A (en) | Method for judging running state of small hydraulic turbine set by using neural network self-coding | |
| CN111986469A (en) | Intelligent diagnosis method for field terminal fault |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |