CN116956203B - Method and system for measuring action characteristics of tapping switch of transformer - Google Patents
Method and system for measuring action characteristics of tapping switch of transformer Download PDFInfo
- Publication number
- CN116956203B CN116956203B CN202311218208.6A CN202311218208A CN116956203B CN 116956203 B CN116956203 B CN 116956203B CN 202311218208 A CN202311218208 A CN 202311218208A CN 116956203 B CN116956203 B CN 116956203B
- Authority
- CN
- China
- Prior art keywords
- data
- transformer
- action
- sequence
- time
- 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.)
- Active
Links
- 230000009471 action Effects 0.000 title claims abstract description 125
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000010079 rubber tapping Methods 0.000 title claims description 9
- 230000008859 change Effects 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 12
- 238000012795 verification Methods 0.000 claims description 11
- 230000001105 regulatory effect Effects 0.000 claims description 5
- 230000005856 abnormality Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000012423 maintenance Methods 0.000 abstract description 6
- 230000009286 beneficial effect Effects 0.000 abstract description 4
- 238000005259 measurement Methods 0.000 description 7
- 230000002159 abnormal effect Effects 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000010276 construction Methods 0.000 description 5
- 238000013500 data storage Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/327—Testing of circuit interrupters, switches or circuit-breakers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application relates to the technical field of transformer detection, in particular to a method and a system for measuring the action characteristics of a tap switch of a transformer. Firstly, constructing a prediction model of the load and the loss of a transformer in an operating state and a shifting state, then respectively constructing the sequence numbers of the action characteristic data of the on-load switch of the transformer in all shifting actions according to historical data to form a sample data set, clustering the sequence data set of the action characteristic data to obtain different load trend clusters, forming the samples of the different load trend clusters into a load trend sample set to construct an action characteristic model, and finally, inputting the samples into the model according to the acquired new action data to verify to judge whether the action characteristic is normal or not. By the technical scheme disclosed by the application, the overlong power failure time caused by transformer maintenance can be reduced, and the beneficial effect of improving the power supply quality is further achieved.
Description
Technical Field
The application relates to the technical field of transformer detection, in particular to a method and a system for measuring the action characteristics of a tap switch of a transformer.
Background
In order to realize uninterrupted power supply and improve the power supply quality, the integration of an on-load tap changer in a transformer capable of voltage regulation is the preferred choice of most transformers. The quality of the transformer on-load tap-changer is directly related to the safe operation of the power supply. Because the on-load tap-changer is an executive component, the abrasion and looseness of equipment cannot be avoided, and in order to ensure the performance of the on-load tap-changer, the transformer and the on-load tap-changer are required to be overhauled and maintained regularly. An important item in the overhaul and maintenance of an on-load tap-changer is the detection of the action characteristics of the on-load tap-changer. And whether the on-load tap-changer is in a higher-performance state or not is judged by detecting the action characteristics, and whether the on-load tap-changer meets the requirement of safe operation or not is judged. Whether based on a direct current power supply or an alternating current power supply, the current detection device is mainly based on detection in a power failure state. Therefore, the power failure is still needed when the test is performed. In order to further improve the power supply quality and reduce the time of power failure, based on the advantages of big data operation of the company, the design of a transformer tap switch operation characteristic measuring method capable of detecting the transformer on-load tap switch operation characteristic in real time on line becomes an urgent requirement.
Disclosure of Invention
The application aims to solve the technical problems that: how to realize the detection of the action characteristics of the on-load tap-changer in the running state of the transformer.
The technical scheme for solving the technical problems is as follows: a method for measuring the action characteristics of a tapping switch of a transformer comprises the following steps:
step 1, constructing a prediction model of the load and the loss of the running state transformer;
step 2, constructing a prediction model of load and loss of the shift state transformer;
step 3, respectively constructing the sequence number of the on-load switch action characteristic data of the transformer under all shift actions according to the historical data, and forming a sample data set:
step 3.1, obtaining n data points in the action time length to construct a data point vector X i (Uh i ,Ih i ,Ul i ,Il i ,ΔP i ,ΔPt i ) Wherein DeltaP i =Uh i Ih i -Ul i Il i ，ΔPt i =ΔP i -Ps, ps calculated from the predictive model in step 1;
step 3.2, calculating error rate
Wherein, px is the actual loss value, pz is the predicted loss value, and the information of the data point is removed when rp is greater than the error rate limit value, and the information is used as the sequence data when rp is less than the error rate limit value, and the final sequence is formed as X, wherein the number of terms of X is less than n;
step 3.3 repeating step 3.1 and step 3.2 until all the motion characteristic data are acquired to form a sample data set;
step 4, clustering the action characteristic data sequence data set to obtain different load trend clusters, and forming a load trend sample set from samples of the different load trend clusters;
step 5, dividing the load trend sample set into data sets to construct an action characteristic model and verifying the model;
step 6, inputting the acquired data of the new gear adjustment into the model constructed in the step 5; if the model does not belong to any model, the abnormality of the voltage regulating action characteristic is determined.
More preferably, in step 3:
dividing the action time Td into m time periods with equal interval time durations, and collecting 3 to 5 data points in each time period;
error rate calculation of data pointsAnd selecting the data of one data point with the minimum error rate in each time period with equal interval duration to represent the data of the time period and forming an action characteristic sequence X.
Preferably, the method further comprises the step of 3.4, obtaining the length of the action characteristic data sequence X, namely X in X i Number n of (2) max I is an integer, and i is more than or equal to 1 and less than or equal to n max When n is max /n<75% of the time, the data of the voltage regulation action is abandoned, namely the obtained sequence X is invalidated.
More preferably, in the step 5:
step 5.1, simplifying vector Xi of sample data;
step 5.2, composing the vector after simplification into new sequence data;
and 5.3, establishing the loss change in the action process into an action characteristic model related to the change of the load.
More preferably, in the step 6:
evenly extracting n data points within the action time Td after the new gear shifting instruction starts;
the error rate of each data point is calculated, and the data of the data points with rp smaller than the error rate limit value are formed into a final real-time action characteristic data sequence X.
Preferably, the validity of the data sequence X is verified,
determining the length n of the real-time motion characteristic data sequence max If not, the value of n is increased until the value is established.
More preferably, in the step 6:
dividing the action time Td after the new gear shifting instruction starts into m time periods;
collecting a plurality of data in each time period and selecting k data with rp smaller than the error rate limit value;
one data among k data is randomly selected in each time period to form a real-time motion characteristic data sequence.
Preferably, a real-time motion characteristic data sequence Xa is randomly selected from k data in each time period, and then a real-time motion characteristic data sequence Xb is randomly selected from k data in each time period; and finally, inputting Xa and Xb into the constructed action characteristic model for verification, and if the Xa and Xb belong to the action characteristic model, indicating that the shift action characteristic of the transformer is normal.
The transformer tapping switch action characteristic measurement system comprises a control system, a data storage system and a real-time acquisition system; the real-time acquisition system comprises an intelligent transformer measurement and control terminal, a high-voltage transformer for acquiring high-voltage side voltage, a high-voltage transformer for acquiring high-voltage side current, a low-voltage transformer for acquiring low-voltage side voltage and a low-voltage current transformer for acquiring low-voltage side current; the control system constructs a transformer tap switch action characteristic model through the data stored in the data storage system, verifies the new real-time shift data acquired by the real-time acquisition system, and judges that the action characteristic is normal if the real-time shift data accords with the action characteristic model.
The beneficial effects of the application are as follows:
a model capable of reflecting the action characteristics of the transformer on-load tap-changer is constructed through analysis operation and modeling of historical data. After the gear shifting operation is carried out, whether the data belong to the constructed model is verified according to the collected data, so that whether the on-load tap-changer of the transformer is in a normal state is judged, and if the on-load tap-changer is in the abnormal state, power failure maintenance is carried out. By the technical scheme disclosed by the application, the overlong power failure time caused by transformer maintenance can be reduced, and the beneficial effect of improving the power supply quality is further achieved.
Drawings
FIG. 1 is a schematic diagram of the system components of an embodiment of the present application.
FIG. 2 is a flow chart of a method of an embodiment of the present application.
In the figure: 310. an intelligent measurement and control terminal of the transformer; 350. a low-voltage current transformer; 340. a low voltage transformer; 330. a high-voltage current transformer; 320. a high voltage transformer; 300. an acquisition system; 200. a data storage system; 100. and a control system.
Detailed Description
In order to make the technical scheme and beneficial effects of the present application clearer, the following further explain the embodiments of the present application in detail.
A method for measuring the action characteristics of a tap switch of a transformer is used for detecting the action characteristics of the tap switch of the transformer. Transformers are commonly single-phase transformers and three-phase transformers, and data of single-phase current and voltage are processed on the single-phase transformers. If the transformer is a three-phase transformer, each phase is processed one by one to judge whether each phase is normal or not. And if any one of the phases is abnormal, an alarm signal is sent out. The specific method for detecting one of the phases comprises the following steps.
Step 1, constructing a prediction model of the load and loss of the running state transformer, and using a function P in the embodiment y =f y (Ul, il) represents a predictive model of load and loss under normal operating conditions, where Ul is the transformer low side voltage and Il is the transformer low side current. The transformer inevitably generates loss of electric energy due to iron loss and copper loss in the running state, and the difference of the loss of the transformer is caused by the different loads of the load carried by the transformer. In order to provide data support for the subsequent steps, a predictive model of the operating state transformer load and loss is constructed primarily to predict the magnitude of transformer loss at a particular load.
In order to reduce the influence of the on-load tap-changer action, only the data of the transformer in the database in the normal running state, namely the data of the on-load tap-changer in the non-action state, are extracted in the process of acquiring the historical data, because the action contact resistance of the execution component also generates loss in the action process of the on-load tap-changer.
In the process of extracting data, step 1.1, extracting a history record of remote adjustment operation from a remote adjustment operation database, and obtaining a voltage regulator gear and a gear shift control instruction at a starting time point and a moment of the starting time point of the remote adjustment operation in the past, wherein the gear shift control instruction is an upshift or a downshift. Step 1.2, the obtained history information is formed into node information, and the node information comprises a starting time point Ts, an action duration Td, a current gear Db, a final gear Da and a control instruction. Step 1.3, when data is extracted from the current voltage database, judging whether the time point of the extracted data is within the action duration Td after the starting time point, and if so, discarding the data. And if the Td after the starting time point is outside, acquiring the data, and looking up the corresponding node information according to the starting time point and marking the gear information of the data. Step 1.4, the acquired data are combined into sample data and a running sample data set Sy is formed. The sample data in the sample data set is three-dimensional vector data (Ul, il, ps), where Ul is the transformer low side voltage, il is the transformer low side current, and Ps is the power loss of the transformer at that power. ps=uhih-Pl, where pl=ulil, pl is the power on the low side of the transformer, uh is the voltage on the high side of the transformer, and Ih is the current on the high side of the transformer.
After data acquisition, a predictive model P of operating state transformer load and loss is constructed using sample data in the operating sample data set Sy y =f y (Ul, il). The regression method in supervised learning can be used for establishing a prediction model, and the neural network-based learning method in the deep learning method can also be used for establishing the prediction model. Linear regression is a basic predictive model construction method and is also a method most suitable for the present application.
Based on the data acquisition mode, classifying the data according to the gear information marked in the acquired data. For example, the data of the transformer operating in the first gear form a first-gear operation sample data set Sy 1 The data run in second gear constitute a second gear run sample data set Sy 2 And by analogy, respectively utilizing each gear operation sample data set to construct a prediction model. E.g. first-gear running sample data set Sy 1 After modeling, a predictive model P of the load and loss of the transformer in the first-gear operating state is finally formed y1 =f y1 (Ul, il), second-gear running sample dataset Sy 2 Finally, a predictive model P of the load and the loss of the transformer in the second-gear running state is formed y2 =f y2 (Ul, il). And further, predicted data can be more accurate in the subsequent data calling process.
Step 2, constructing a prediction model of load and loss of the shift state transformer and using a function P t =f t (Ul, il) represents a predictive model of upshift state load and loss.
The data extracted in step 1 is data other than the action duration Td after the start time point of the shift operation, and in the present step, the extracted data is data within Td after the start time point. In this step, an independent model construction is required for each shift action. Therefore, when data are collected, the data in the gear-shifting state are classified according to the current gear information and the control instruction in the node information. In upshift adjustment from first gear to second gear, the constituent sample data set is a first gear upshift sample data set S t1 The constructed model is P t1 =f t1 (Ul, il) a sample data set consisting of an adjustment from a second gear down to a first gear down is a first gear down sample data set S t-1 The constructed model is P t-1 =f t-1 (Ul, il) the method of constructing a model in this step is the same as in step 1; correspondingly, in the adjustment of the second gear to the third gear up-shift, the composed sample data set is a second gear up-shift sample data set S t2 The constructed model is P t2 =f t2 (Ul, il) a sample data set consisting of a three-gear down to two-gear adjustment is a two-gear down sample data set, and the constructed model is P t-2 =f t-2 (Ul,Il)。
And 3, constructing a transformer on-load tap-changer action characteristic data sequence data set Sd.
In the process of detecting the action characteristics of the on-load tap-changer of the transformer, the whole action process needs to be detected, namely, the whole action process in the action time period Td after the starting time point Ts is detected and recorded, so that data of a plurality of time points in the action time period Td after the starting time point Ts need to be collected to form a data sequence for representing the action characteristic curve fd.
Step 3.1, extracting n data points in the action duration Td time period, wherein the n data points can be uniformly distributed in the Td time period. The data collected on the data points include a high-side voltage Uh, a high-side current Ih, a low-side voltage Ul, and a low-side current Il.
And 3.2, calculating a loss value Px from the data in the step 3.1, wherein the loss value Px is an actual loss value, and the loss value Px is calculated through currents and voltages at a high voltage side and a low voltage side.
The P constructed in step 2 is applied from the low side voltage and the low side current t =f t (Ul, il) to derive a predicted loss value Pz. I.e. according to the predictive model P of the gear ti =f ti (Ul, il) gives a predicted loss value Pz, where i is the positive and negative value of the gear value.
Step 3.3, calculating the absolute value of the difference between Px and Pz as the error loss, and calculating the error rate. And meanwhile, judging whether the data is valid or not according to the set error rate limit value. And when rp is smaller than the error rate limit value, the data is taken as one time point of the sequence data to form a final action characteristic data sequence X.
Wherein each element Xi of the sequence X is a multidimensional vector (Uh i ,Ih i ,Ul i ,Il i ,ΔP i ,ΔPd i ) Wherein DeltaP i As actual loss value, ΔP i =Uh i Ih i -Ul i Il i 。
ΔPd i For the predicted action loss value with the action of the voltage regulating tapping switch according to the prediction model in the step 1, deltaPd i =ΔP i -Ps, ps being derived from model P y =f y (Ul i ,Il i ) Obtained. And 2, verifying the data by utilizing the data in the step 2, and removing unsuitable data or data with larger mutation quantity caused by disturbance so as to form a more accurate dominant sample data set.
And 3.4, verifying the validity of the data again. Acquisition actionThe length of the characteristic data sequence X, i.e., how many data points are contained in the sequence X. Xi is an element in the sequence X, i is an integer, and i is more than or equal to 1 and less than or equal to n max Wherein n is max And < n. After determining the length of sequence X, when n max /n<75% of the time, the data of the voltage regulation action is abandoned, namely the obtained sequence X is invalidated.
If a sequence X (X 1 ，X 2 ，X 3 ，……X i ) Each element in the sequence is a phasor. For a certain voltage regulation action, a plurality of sequence data are collected to form an action characteristic data sequence data set SX. Meanwhile, an action characteristic data sequence data set SX of different gear shifting states is obtained according to the gear information and the control instruction in the data. In upshift adjustment from first gear to second gear, the constituent sample data set is a first gear upshift action characteristic data sequence data set SX 1 The sample data set consisting of the adjustment from the second gear down to the first gear is the first gear down motion characteristic data sequence data set SX -1 Correspondingly, in the adjustment of the second gear to the third gear up-shift, the composed sample data set is a second gear up-shift action characteristic data sequence data set SX 2 The sample data set composed of the adjustment from the third gear down to the second gear is the second gear down shift action characteristic data series data set SX 2 。
Alternatively, in step 3.1, the action duration Td is divided into m time periods of equal interval duration, and 3 to 5 data points are acquired inside each time period. Then using the verification method in step 3.2 to verify the obtained data, verifying 3 to 5 data in the time period of equal interval duration, sequentially calculating error rate, then selecting data of one data point with minimum error rate to represent the data of the time period, and finally forming a sequence X with m elements, wherein the elements are X i 。
And 4, clustering the action characteristic data sequence data set SX to obtain different clusters. The action characteristic data series data set SX including the plurality of data sets of shift actions requires clustering processing for each of the action characteristic data series data sets SX of different shift actions. For SX as i Can be processed byDifferent load trend clusters are obtained, and a corresponding load trend data set SX is obtained according to the different load trend clusters ij Where j is the number of clusters.
And obtaining the power loss change trend under different change trends through clustering. The loss change conditions under different load change trends can be obtained by carrying out clustering processing on a large amount of data. The calculation difficulty under various different load change conditions is reduced by classifying different load change trends. Thereby achieving the purpose of improving the accuracy of the detection of the action characteristic curve.
Step 5, respectively collecting the load trend sample sets SX ij The method comprises the steps of dividing a modeling data set and a verification data set to construct a model and a verification model. The data set may be divided in a random division manner: firstly, randomly scrambling samples in an action characteristic data sequence data set; the samples are then divided into a modeling data set and a validation data set at a ratio (e.g., 70% training set, 30% validation set).
The SVM, support vector machine, is a supervised learning model widely applied to classification and regression problems. It can generate an optimal hyperplane in high-dimensional space, separating two classes of data samples. The SVM is characterized in that it can use different kernel functions to deal with the nonlinearity problem and can maintain a good generalization ability for a small number of support vectors. Neural networks are a model made up of multiple layers of neurons, often used to solve complex classification and regression problems. The weight parameters of the model can be learned through a back propagation algorithm, and the model has strong expression capability. The neural network is suitable for processing large-scale and complex data sets and can mine nonlinear relations in the data. However, the neural network training process is time consuming and requires a large amount of samples and computing resources, so a support vector machine may be employed to construct the predictive model in the early stages of the system operation.
Step 5.1, simplifying the vector Xi of the sample data before processing to construct the prediction model, wherein Xi in the original sample is (Uh i ,Ih i ,Ul i ,Il i ,ΔP i ,ΔPd i ) X after simplification i "is (Ul) i ,Il i ,ΔP i ,ΔPd i ). And 5.2, composing the vector after simplification into new sequence data. Step 5.3 after simplification and modeling, the loss change during the action is built into an action characteristic model associated with the change in load. To facilitate the construction of an action characteristic model for describing action characteristics and load changes, yf is adopted ij An operation characteristic model indicating an operation characteristic and a load change corresponding to the load trend sample set. Where i is + -1, + -2 … … + -7, + -8, similar to the previous labeling relationship, 1 indicates upshift from first gear to second gear, -1 indicates downshift from second gear to first gear, and the other is similar. j represents different cluster types obtained after clustering the data of the shift action.
In the construction of the motion characteristic model Yf ij Then, the verification set is applied to verify the model, and the model is continuously corrected through the verification of the verification set to obtain a more accurate action characteristic model Yf ij 。
Step 6, inputting the action characteristic model Yf of action characteristic and load change constructed in the step 5 according to the collected new gear adjustment data ij And (3) performing verification. As a plurality of action characteristic models for adjusting gears are provided, if new data belong to any one of the models, the action characteristic of gear shifting is normal, and if the new data do not belong to any one of the models, the action characteristic of gear shifting is abnormal.
And constructing a real-time action characteristic data sequence.
1. The down shift instruction is started as a starting time point Ts, and data within the action duration Td are collected at the starting time point. In the process of collecting data, n data points can be extracted in the time of Td, and the n data points can be uniformly distributed in the time of Td. The data collected on the data points include a high-side voltage Uh, a high-side current Ih, a low-side voltage Ul, a low-side current Il, from which the loss value Px is calculated. Then calculating error rate, removing information of data point when rp is greater than error rate limit value, forming sequence data when rp is less than error rate limit value, finally forming data meeting standard into final real-time action specialThe sexual data sequence is X. The validity of the data sequence X is then again verified according to the method in step 3.4. If the data sequence is invalid, after the action characteristic data sequence is abandoned, changing the value of n, further, adjusting the value of n to be larger, further, acquiring the data at different time points, and then verifying again. Up to the length n of the real-time motion characteristic data sequence max When n > 75%, the generated sequence data X is then inputted as sample data into the operation characteristic model for verification. After the shift command is issued, the current shift and the control command can be obtained. And then selecting a corresponding model according to the current gear and the control instruction. All motion characteristic models Yf for a certain shift motion ij If the action characteristic model corresponding to any load trend sample set is met, the shift action is normal, otherwise, the shift action is abnormal, and power failure maintenance is needed.
2. The data may be acquired in such a manner that Td is divided into m time periods, as follows. The action time length is divided into m time periods with equal interval time lengths. Multiple data are collected within each time period and k data with rp less than the error rate limit are selected. Where each time period has k valid data. Selecting one data among k data randomly in each of m time periods to constitute a real-time motion characteristic data sequence Xa, one element of the sequence being a vector (Ul i ,Il i ,ΔP i ,ΔPd i ). Then, one data is randomly selected from k data in each of m time periods to form a real-time action characteristic data sequence Xb. And finally, inputting Xa and Xb into the constructed action characteristic model, and judging whether the voltage regulation action characteristic is normal by judging whether the model accords with the prediction model.
If the two times are normal, the action characteristic of the transformer during gear shifting is judged to be normal, otherwise, the action characteristic is judged to be abnormal. And then, data acquisition and verification are carried out again, and the abnormal action characteristic of the gear shifting of the transformer is determined after the abnormality is determined, so that maintenance is needed.
The application also discloses a transformer tapping switch action characteristic measurement system which comprises a control system 100, a data storage system 200 and a real-time acquisition system 300. Wherein the control system 100 may be a computer or a cloud processor for performing operations and processing of data. The data storage system 200 is used to enable storage of data, including permanent storage and temporary storage of data. The real-time acquisition system 300 comprises a transformer intelligent measurement and control terminal 310, a high-voltage transformer 320 for acquiring high-voltage side voltage, a high-voltage transformer 330 for acquiring high-voltage side current, a low-voltage transformer 340 for acquiring low-voltage side voltage and a low-voltage current transformer 350 for acquiring low-voltage side current, wherein the transformer intelligent measurement and control terminal 310 is provided with four remote functions for converting signals acquired by the high-voltage transformer 320, the high-voltage current transformer 330, the low-voltage transformer 340 and the low-voltage current transformer 350 into digital information, and meanwhile, the system is provided with a remote signaling acquisition module and a remote regulation acquisition module for acquiring action information of an on-load voltage regulating switch of the transformer. The intelligent measurement and control terminal 310 of the transformer can receive data of a transformer substation monitoring system and control the transformer to realize gear shifting, and meanwhile collect various state information of the transformer.
The control system 100 can be arranged in an independent computer or a computer where a monitoring system of a transformer substation is located, and can realize data exchange with the monitoring system of the transformer substation or data exchange with a dispatching center, and if the measuring system disclosed by the application is newly installed, the historical data of the transformer is required to be transferred through the data exchange.
The control system is provided with a data processing program designed according to the method disclosed by the application, and is used for realizing the processing of the collected real-time data and the historical data, constructing a model by collecting and calling the historical data, then verifying whether the real-time data belongs to any one of the models, and if the real-time data belongs to one of the models, indicating that the action characteristic of the tapping switch of the transformer is normal.
After the on-load tap-changer operation model is built, the operation characteristic curve can be built by using the data for building the operation model. One element of the data sequence of the build model is a vector (Ul i ,Il i ,ΔP i ,ΔPd i ) Selecting delta Pdi and corresponding i composition data，ΔPd i As a result, i is a parameter, i.e., in the plane coordinate system, (i, Δpd) i ) For the point coordinates, all the points are constructed into an action characteristic curve which changes along with time in a linear regression mode. And extracting the latest 10 groups of action characteristic data sequences to construct an action characteristic curve, and taking the constructed curve function as an action characteristic curve of the on-load tap-changer.
In summary, the above description is only a preferred embodiment of the present application, and is not intended to limit the scope of the present application, and the related workers can make various changes and modifications without departing from the scope of the technical idea of the present application. The technical scope of the present application is not limited to the description, but includes all equivalent changes and modifications in shape, construction, characteristics and spirit according to the scope of the claims.
Claims (8)
1. The method for measuring the action characteristics of the tapping switch of the transformer is characterized by comprising the following steps of:
step 1, extracting data of a transformer in a normal running state in a database to form a running sample data set, and constructing a running state transformer load and loss prediction model Py=fy (Ul, il) by using the running sample data set; ul is the voltage of the low-voltage side of the transformer, il is the current of the low-voltage side of the transformer, and Py is the loss value in the running state;
step 2, extracting data of the transformer in the shift state in the database to form a sample data set, and constructing a prediction model P of the load and loss of the transformer in the shift state by using the sample data set t =f t (Ul,Il)，P t Is the loss value in the shift state;
step 3, respectively constructing the sequence number of the on-load switch action characteristic data of the transformer under all shift actions according to the historical data, and forming a sample data set:
step 3.1, obtaining n data points in the action time length to construct a data point vector X i (Uh i ,Ih i ,Ul i ,Il i ,ΔP i ,ΔPd i )，
Wherein Uh is i High side voltage, ih, for the ith data point i High side current, ul, for the ith data point i Low side voltage, il, for the ith data point i Low side current for the i-th data point;
wherein ΔPd i Action loss value, delta Pd, of action of voltage regulating tapping switch predicted by prediction model of ith data point i =ΔP i -Ps, wherein ΔP i The actual loss value for the i-th data point,
ΔP i =Uh i Ih i -Ul i Il i the method comprises the steps of carrying out a first treatment on the surface of the Loss of the transformer under load of the ith data point of Ps, wherein Ps is calculated according to the prediction model in the step 1;
step 3.2, calculating error rate
Wherein, px is the actual loss value; pz is a predicted loss value, and is calculated by the prediction model in the step 2; removing the data point information when rp is larger than the error rate limit value, and taking the data point information as sequence data when rp is smaller than the error rate limit value, wherein the final sequence is composed of X, and the number of terms of X is smaller than n;
step 3.3 repeating step 3.1 and step 3.2 until all the motion characteristic data are acquired to form a sample data set;
step 4, clustering the action characteristic data sequence data set to obtain different load trend clusters, and forming a load trend sample set from samples of the different load trend clusters;
step 5, dividing the load trend sample set into data sets to construct an action characteristic model and verifying the model;
step 6, inputting the acquired data of the new gear adjustment into the model constructed in the step 5; if the model does not belong to any model, the abnormality of the voltage regulating action characteristic is determined.
2. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 1, wherein:
in step 3:
dividing the action time Td into m time periods with equal interval time durations, and collecting 3 to 5 data points in each time period;
then calculating error rate rp of data points, selecting data of a data point with the smallest error rate in each time period with equal interval duration to represent the data of the time period, and forming action characteristic sequence X.
3. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 1 or 2, wherein:
also comprises the step 3.4 of obtaining the length of the action characteristic data sequence X, namely X in X i Number n of (2) max I is an integer, and i is more than or equal to 1 and less than or equal to n max When n is max /n<75% of the time, the data of the voltage regulation action is abandoned, namely the obtained sequence X is invalidated.
4. A method for measuring the operating characteristics of a tap changer of a transformer according to claim 3, wherein:
in the step 5:
step 5.1 vector X of sample data i Simplifying the process;
step 5.2, composing the vector after simplification into new sequence data;
and 5.3, establishing the loss change in the action process into an action characteristic model related to the change of the load.
5. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 1, wherein:
in the step 6:
evenly extracting n data points within the action time Td after the new gear shifting instruction starts;
the error rate of each data point is calculated, and the data of the data points with rp smaller than the error rate limit value are formed into a final real-time action characteristic data sequence X.
6. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 5, wherein:
the validity of the data sequence X is verified,
determining the length n of the real-time motion characteristic data sequence max If not, the value of n is increased until the value is established.
7. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 1, wherein:
in the step 6:
dividing the action time Td after the new gear shifting instruction starts into m time periods;
collecting a plurality of data in each time period and selecting k data with rp smaller than the error rate limit value;
one data among k data is randomly selected in each time period to form a real-time motion characteristic data sequence.
8. The method for measuring the operating characteristics of a tap changer of a transformer according to claim 7, wherein:
randomly selecting one data from k data in each time period to form a real-time action characteristic data sequence Xa, and then randomly selecting one data from k data in each time period to form a real-time action characteristic data sequence Xb; and finally, inputting Xa and Xb into the constructed action characteristic model for verification, and if the Xa and Xb belong to the action characteristic model, indicating that the shift action characteristic of the transformer is normal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311218208.6A CN116956203B (en) | 2023-09-21 | 2023-09-21 | Method and system for measuring action characteristics of tapping switch of transformer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311218208.6A CN116956203B (en) | 2023-09-21 | 2023-09-21 | Method and system for measuring action characteristics of tapping switch of transformer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116956203A CN116956203A (en) | 2023-10-27 |
| CN116956203B true CN116956203B (en) | 2023-12-15 |
Family
ID=88458744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311218208.6A Active CN116956203B (en) | 2023-09-21 | 2023-09-21 | Method and system for measuring action characteristics of tapping switch of transformer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116956203B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117435947A (en) * | 2023-12-20 | 2024-01-23 | 山东和兑智能科技有限公司 | Lightning arrester state monitoring system and method |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105759200A (en) * | 2016-02-18 | 2016-07-13 | 中国农业大学 | Method for online monitoring of work state of on-load tap-changer and early warning |
| CN206788325U (en) * | 2017-06-26 | 2017-12-22 | 福建瑞能博尔电力设备有限公司 | Load switch of transformer monitors and diagnostic device |
| CN108593099A (en) * | 2018-04-28 | 2018-09-28 | 国家电网公司 | A kind of on-load regulator transformer tap switch vibration signal extracting method |
| CN110196370A (en) * | 2019-06-26 | 2019-09-03 | 山东电工电气集团智能电气有限公司 | The monitoring method and device of transformer |
| CN110766059A (en) * | 2019-10-14 | 2020-02-07 | 四川西部能源股份有限公司郫县水电厂 | Transformer fault prediction method, device and equipment |
| CN111239646A (en) * | 2020-01-17 | 2020-06-05 | 南方电网科学研究院有限责任公司 | On-load tap-changer interstage short circuit fault early warning method and device and storage medium |
| WO2020125716A1 (en) * | 2018-12-21 | 2020-06-25 | 中兴通讯股份有限公司 | Method for realizing network optimization and related device |
| AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
| WO2021028948A1 (en) * | 2019-08-15 | 2021-02-18 | Easun - Mr Tap Changers (P) Ltd | System for detecting the incomplete operation of diverter in on-load tap-changing transformers |
| CN113325302A (en) * | 2021-04-29 | 2021-08-31 | 国网河南省电力公司平顶山供电公司 | Transformer on-load tap-changer state monitoring device |
| KR102350323B1 (en) * | 2021-04-21 | 2022-01-12 | 케이.엘.이.에스 주식회사 | Pipe load monitoring system and method |
| CN114325361A (en) * | 2021-11-15 | 2022-04-12 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Transformer tap switch state online monitoring method based on vibration principle |
| KR20220052200A (en) * | 2020-10-20 | 2022-04-27 | 한국전력공사 | Machine learning based load prrdiction model cmparative verification system and mthod |
| JP2023063152A (en) * | 2021-10-22 | 2023-05-09 | 中国電力株式会社 | Transformer load prediction system and transformer load prediction method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3576029A1 (en) * | 2018-05-31 | 2019-12-04 | IMEC vzw | Method and device for determining energy system operating scenarios |
-
2023
- 2023-09-21 CN CN202311218208.6A patent/CN116956203B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105759200A (en) * | 2016-02-18 | 2016-07-13 | 中国农业大学 | Method for online monitoring of work state of on-load tap-changer and early warning |
| CN206788325U (en) * | 2017-06-26 | 2017-12-22 | 福建瑞能博尔电力设备有限公司 | Load switch of transformer monitors and diagnostic device |
| CN108593099A (en) * | 2018-04-28 | 2018-09-28 | 国家电网公司 | A kind of on-load regulator transformer tap switch vibration signal extracting method |
| WO2020125716A1 (en) * | 2018-12-21 | 2020-06-25 | 中兴通讯股份有限公司 | Method for realizing network optimization and related device |
| CN110196370A (en) * | 2019-06-26 | 2019-09-03 | 山东电工电气集团智能电气有限公司 | The monitoring method and device of transformer |
| WO2021028948A1 (en) * | 2019-08-15 | 2021-02-18 | Easun - Mr Tap Changers (P) Ltd | System for detecting the incomplete operation of diverter in on-load tap-changing transformers |
| CN110766059A (en) * | 2019-10-14 | 2020-02-07 | 四川西部能源股份有限公司郫县水电厂 | Transformer fault prediction method, device and equipment |
| CN111239646A (en) * | 2020-01-17 | 2020-06-05 | 南方电网科学研究院有限责任公司 | On-load tap-changer interstage short circuit fault early warning method and device and storage medium |
| KR20220052200A (en) * | 2020-10-20 | 2022-04-27 | 한국전력공사 | Machine learning based load prrdiction model cmparative verification system and mthod |
| AU2020104000A4 (en) * | 2020-12-10 | 2021-02-18 | Guangxi University | Short-term Load Forecasting Method Based on TCN and IPSO-LSSVM Combined Model |
| KR102350323B1 (en) * | 2021-04-21 | 2022-01-12 | 케이.엘.이.에스 주식회사 | Pipe load monitoring system and method |
| CN113325302A (en) * | 2021-04-29 | 2021-08-31 | 国网河南省电力公司平顶山供电公司 | Transformer on-load tap-changer state monitoring device |
| JP2023063152A (en) * | 2021-10-22 | 2023-05-09 | 中国電力株式会社 | Transformer load prediction system and transformer load prediction method |
| CN114325361A (en) * | 2021-11-15 | 2022-04-12 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Transformer tap switch state online monitoring method based on vibration principle |
Non-Patent Citations (1)
| Title |
|---|
| 基于变压器特性数据聚类的电力负荷分布式预测;吴琼;段炼;乡立;黄锦增;魏艳霞;孙毅;;广东电力(第08期);正文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116956203A (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102930344B (en) | A kind of ultra-short term bus load Forecasting Methodology based on load trend change | |
| CN116956203B (en) | Method and system for measuring action characteristics of tapping switch of transformer | |
| CN114429153B (en) | Gear box increment fault diagnosis method and system based on life learning | |
| CN105718958B (en) | Circuit breaker failure diagnostic method based on linear discriminant analysis and support vector machines | |
| CN105740975A (en) | Data association relationship-based equipment defect assessment and prediction method | |
| CN109214708A (en) | Study of Risk Evaluation Analysis for Power System method based on cross entropy theoretical optimization support vector machines | |
| CN116031888B (en) | Dynamic load prediction-based power flow optimization method, system and storage medium | |
| CN109818349A (en) | A kind of power grid robust state prediction technique based on multi-dimensional state matrix shiding matching | |
| CN112330165A (en) | Power grid transient stability evaluation method and system based on feature separation type neural network | |
| CN109740727A (en) | A kind of hydraulic turbine shaft state monitoring method neural network based and system | |
| CN110837915A (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
| CN111555363B (en) | AGC real-time control strategy based on deep learning under big data environment | |
| Xue et al. | Adaptive ultra-short-term wind power prediction based on risk assessment | |
| CN115640748A (en) | Dynamic frequency response prediction method for generators after disturbance of power system | |
| CN113627032B (en) | Intelligent decision-making method for equipment design/maintenance scheme based on digital twinning | |
| Amjady | A framework of reliability assessment with consideration effect of transient and voltage stabilities | |
| CN111244937B (en) | Method for screening serious faults of transient voltage stability of power system | |
| Ramirez-Gonzalez et al. | Convolutional neural network based approach for static security assessment of power systems | |
| CN113807027B (en) | Wind turbine generator system health state evaluation model, method and system | |
| CN114692729A (en) | New energy station bad data identification and correction method based on deep learning | |
| Khalyasmaa et al. | Training sample formation for intelligent recognition of circuit breakers states patterns | |
| CN115374957B (en) | Radial flow type small hydropower station multiscale missing measurement data reconstruction method | |
| Bai et al. | Abnormal Detection Scheme of Substation Equipment based on Intelligent Fusion Terminal | |
| CN114219147B (en) | Power distribution station fault prediction method based on federal learning | |
| Zhu et al. | Frequency Stability Control Method of AC/DC Power System Based on Convolutional Neural Network |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |