CN116702630A - Error iterative evaluation method for capacitive voltage transformer CVT - Google Patents
Error iterative evaluation method for capacitive voltage transformer CVT Download PDFInfo
- Publication number
- CN116702630A CN116702630A CN202310980854.XA CN202310980854A CN116702630A CN 116702630 A CN116702630 A CN 116702630A CN 202310980854 A CN202310980854 A CN 202310980854A CN 116702630 A CN116702630 A CN 116702630A
- Authority
- CN
- China
- Prior art keywords
- error
- prediction
- cvt
- prediction result
- cndot
- 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.)
- Granted
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 claims abstract description 80
- 238000004088 simulation Methods 0.000 claims abstract description 10
- 238000012549 training Methods 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 26
- 230000001932 seasonal effect Effects 0.000 claims description 17
- 238000000354 decomposition reaction Methods 0.000 claims description 15
- 230000000737 periodic effect Effects 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 125000003275 alpha amino acid group Chemical group 0.000 claims description 6
- 150000001875 compounds Chemical class 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 4
- 230000009191 jumping Effects 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000013526 transfer learning Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 abstract 1
- 239000011159 matrix material Substances 0.000 description 8
- 230000006870 function Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R35/00—Testing or calibrating of apparatus covered by the other groups of this subclass
- G01R35/02—Testing or calibrating of apparatus covered by the other groups of this subclass of auxiliary devices, e.g. of instrument transformers according to prescribed transformation ratio, phase angle, or wattage rating
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/096—Transfer learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Power Engineering (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an error iterative evaluation method of a capacitive voltage transformer CVT, which comprises the following steps: build laboratory simulation platform, construct CVT offline data D L The method comprises the steps of carrying out a first treatment on the surface of the Predicting by using the first to sixth prediction models to obtain first to sixth prediction results respectively; migrating the sixth prediction model to the CVT real-time working condition to obtain a seventh prediction model; decomposing the online CVT error data into an error period term, an error trend term and an error remainder; obtaining seventh to ninth prediction results by using a seventh prediction model; utilizing a seventh prediction result, an eighth prediction result, and a ninth prediction junctionTraining fruits to obtain an eighth prediction model; and predicting the error of the CVT to be detected by using an eighth prediction model. The error iterative evaluation method of the capacitive voltage transformer CVT has the advantages that the real error state of the voltage transformer CVT can be obtained under the condition of fewer data samples, the accuracy of the error prediction model of the voltage transformer CVT is improved, and the like.
Description
Technical Field
The invention relates to an online monitoring method for electric power metering, in particular to an error iteration evaluation method for a capacitive voltage transformer CVT.
Background
A voltage transformer (Voltage Transformer, VT for short) is an electric appliance indispensable to a power transmission and supply system of a power plant, a substation, or the like. The precision voltage transformer is an instrument used for enlarging the limit of measurement and measuring voltage, power and electric energy in an electric measuring laboratory. Voltage transformers and transformers are very similar and are used to transform the voltage on a line.
The voltage transformer is used as key equipment in the power system, realizes accurate measurement of primary voltage on the premise of ensuring the safety of secondary equipment and electricity consumption, and provides reliable basis for electric energy metering, state monitoring, relay protection and the like. The voltage transformer is ensured to be in a stable running state, the accuracy of the power system in metering and measuring can be ensured, the reliability of the automatic device and relay protection actions is improved, and the safe, stable and economic running of the power system is facilitated.
The capacitive voltage transformer (Capacitance voltage transformer, CVT for short) is a voltage transformer which is divided by a series capacitor, is reduced in voltage and isolated by an electromagnetic transformer, and is used as a meter, relay protection and the like. The capacitive voltage transformer may also couple carrier frequencies to the power line for long distance communications, remote measurement, selective line high frequency protection, remote control, teletyping, etc.
In the actual operation of the power system, due to the complex and changeable operation conditions of the power grid and the increase of the service life of the voltage transformer CVT, the operation state of the voltage transformer CVT and the accuracy of the voltage transformer CVT are changed. The reduction of the CVT accuracy of the voltage transformer presents a potential hazard for safe, stable and economical operation of the power system. Long-term operation experience shows that the voltage transformer CVT runs for several years with a certain risk of out-of-tolerance due to its longer life. The continuous operation of the out-of-tolerance voltage transformer CVT brings huge loss to the gateway metering trade settlement of the power supply and use parties, and even affects the stable operation of the power system.
Therefore, the metering error state of the voltage transformer CVT in operation needs to be effectively evaluated, the out-of-tolerance problem of the voltage transformer CVT is timely found, and a reliable basis is provided for formulating corresponding maintenance and overhaul strategies; the risk early warning is carried out on the voltage transformer CVT with high out-of-tolerance risk, the voltage transformer with serious degradation trend is found in time, and the timeliness of overhauling the voltage transformer CVT is ensured; meanwhile, necessary maintenance can be carried out on the voltage transformer CVT with requirements, the conventional blind overhaul and maintenance on part of the voltage transformer CVT are avoided, the workload is reduced, and the labor efficiency is improved.
The existing online evaluation method is to analyze and process signals collected by all devices in an electric power system based on a data driving principle, so as to evaluate the error state of the voltage transformer CVT, namely, the error state of the voltage transformer CVT is real-time represented by constructing an approximate model by means of historical data, real-time data and relational data and relying on a large amount of data and calculation. However, the method has the problem that once the data samples are fewer, the real error state of the voltage transformer CVT is difficult to reflect, and the accuracy of the real error state of the voltage transformer CVT is affected.
Disclosure of Invention
The invention provides an error iterative evaluation method of a capacitive voltage transformer CVT to avoid the defects in the prior art, so that the actual error state of the voltage transformer CVT can be obtained under the condition of fewer data samples.
The invention adopts the following technical scheme for solving the technical problems.
The invention discloses an error iterative evaluation method of a capacitive voltage transformer CVT, which specifically comprises the following steps:
step S100: build laboratory simulation platform, construct CVT offline data D L The method comprises the steps of carrying out a first treatment on the surface of the The CVT offline data D L Including error truth data D L1 Fix error data D L2 Cycle error data D L3 Gradual error data D L4 Random error data D L5 ；
Step S200: using a first predictive model f 1 (. Cndot.) second prediction model f 2 (. Cndot.) third prediction model f 3 (. Cndot.) fourth prediction model f 4 (. Cndot.) fifth predictive model f 5 (. Cndot.) and sixth predictive model f 6 (. Cndot.) are predicted to obtain first prediction results y respectively 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 Fifth prediction result y 5 And a sixth prediction result y 6 ；
Using a first predictive model f 1 (. Cndot.) for error truth data D L1 Predicting to obtain a first prediction result y 1 ；
Using a second predictive model f 2 (. Cndot.) for fixed error data D L2 Predicting with the first prediction result to obtain a second prediction result y 2 ；
Using a third predictive model f 3 (. Cndot.) for periodic error data D L3 Predicting with the second prediction result to obtain a third prediction result y 3 ；
Using a fourth predictive model f 4 (. Cndot.) for gradation error data D L4 Predicting with the third predicted result to obtain a fourth predicted result y 4 ；
Using a fifth predictive model f 5 (. Cndot.) for random error data D L5 Predicting with the fourth predicted result to obtain a fifth predicted result y 5 ；
Using the first prediction result y 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 And a fifth prediction result y 5 Training to obtain a sixth predictive model f 6 （·）；
y 1 =f 1 （D L1 ） （6）；
y 2 =f 2 （y 1 ，D L2 ） （7）；
y 3 =f 3 （y 2 ，D L3 ） （8）；
y 4 =f 4 （y 3 ，D L4 ） （9）；
y 5 =f 5 （y 4 ，D L5 ） （10）；
y 5 =f 6 （y 1 ，y 2 ，y 3 ，y 4 ，y 5 ） （11）；
Step S300: the sixth predictive model f in step S200 is calculated by the transfer learning algorithm 6 (. Cndot.) migration to CVT real-time conditions to obtain a seventh predictive model f 7 （·）；
Step S400: acquiring online CVT error data, and decomposing the online CVT error data into an error period term, an error trend term and an error remainder by using an STL decomposition method;
step S500: utilizing a seventh predictionModel f 7 (. Cndot.) obtaining the seventh prediction result y 7 Eighth prediction result y 8 And a ninth prediction result y 9 ；
Step S600: using the seventh predictor y 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 （·）；
Step S700: using an eighth predictive model f 8 (. Cndot.) the error of the CVT to be measured is predicted.
The error iterative evaluation method of the capacitive voltage transformer CVT is also characterized by comprising the following steps of:
further, in the step 300, the pre-training process is as follows: seventh prediction model f 7 The pretrained model parameters of (-) are set to be the same as the sixth predictive model f 6 Parameters of (-) are consistent.
Further, in the step 300, a seventh prediction model f 7 The training process of (-) is that the online error data at the time t-1 is used as input, the online error data at the time t is used as output, and a sixth prediction model f is obtained 6 Training to obtain a seventh prediction model f 7 （·）。
Further, in step 400, online CVT error dataCan be decomposed into STL:
(12)
in the formula (12), the amino acid sequence of the compound,、/>and->And respectively representing an error period term, an error trend term and an error remainder, wherein t is the sampling moment.
Further, in the step 400, the STL decomposition method includes an inner loop iteration process and an outer loop iteration process.
Further, the (i+1) th iteration of the STL decomposition method includes the following steps:
step S1: historical error data for CVTCarrying out trending item removal treatment;
step S2: a sub-sequence smoothing step;
step S3: calculating a trend item component of the new time sequence and a seasonal component of the new time sequence;
step S4: obtaining a seasonal component and a trend component of the original sequence;
step S5: judging whether the seasonal component and the trend component of the original sequence are converged or not; and (4) jumping out of the inner loop after convergence to obtain the remainder component of the time sequence.
Further, in the step 500, a seventh prediction model f is utilized 7 (. Cndot.) time-domain multi-head self-attention prediction is carried out on the error period term to obtain a seventh prediction result y 7 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) frequency domain multi-head self-attention prediction is carried out on the error remainder to obtain an eighth prediction result y 8 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) predicting the trend term to obtain the ninth predicted result y 9 。
Further, in the step 600, a seventh prediction result y is used 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 The specific process of the (-) is as follows: the input is as follows: seventh prediction result y at time t-1 7 Eighth prediction result y 8 And a ninth prediction result y 9 The method comprises the steps of carrying out a first treatment on the surface of the The pre-training model is as follows: seventh prediction model f 7 (. Cndot.); the output is: the CVT error value is online at time t.
The invention also discloses an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of iterative error assessment of the capacitive voltage transformer CVT.
The invention also discloses a computer readable storage medium storing a computer program. The computer program when executed by the processor implements the method for iterative error assessment of the capacitive voltage transformer CVT.
Compared with the prior art, the invention has the beneficial effects that:
the invention discloses an error iterative evaluation method of a capacitive voltage transformer CVT, which comprises the following steps: build laboratory simulation platform, construct CVT offline data D L The method comprises the steps of carrying out a first treatment on the surface of the Predicting by using the first to sixth prediction models to obtain first to sixth prediction results respectively; migrating the sixth prediction model to the CVT real-time working condition to obtain a seventh prediction model; decomposing the online CVT error data into an error period term, an error trend term and an error remainder; obtaining seventh to ninth prediction results by using a seventh prediction model; training by using the seventh prediction result, the eighth prediction result and the ninth prediction result to obtain an eighth prediction model; and predicting the error of the CVT to be detected by using an eighth prediction model.
The invention discloses an error iterative evaluation method of a capacitive voltage transformer CVT, which has the following technical advantages.
1. By adopting an off-line training-on-line modeling method, the problem of few data samples in actual working conditions is solved;
2. the accuracy of the prediction model is improved by a method of adding difference iteration on normal data;
3. according to the characteristics of CVT time sequence data, a method of combining time domain self-attention prediction and frequency domain self-attention prediction is adopted, so that the accuracy of a prediction model is improved.
The error iterative evaluation method of the capacitive voltage transformer CVT has the advantages that the real error state of the voltage transformer CVT can be obtained under the condition of fewer data samples, the accuracy of the error prediction model of the voltage transformer CVT is improved, and the like.
Drawings
Fig. 1 is a block diagram of an error iterative evaluation method of a capacitive voltage transformer CVT according to the present invention.
Fig. 2 is a flowchart of an error iterative estimation method of the capacitive voltage transformer CVT according to the present invention.
Fig. 3 is a flowchart of a time-domain multi-head self-attention prediction process performed by an error period term of an error iterative evaluation method of a capacitive voltage transformer CVT according to the present invention.
Fig. 4 is a flowchart of a frequency domain multi-head self-attention prediction process by using the error remainder of the error iterative estimation method of the capacitive voltage transformer CVT according to the present invention.
The invention is further described below by means of specific embodiments in connection with the accompanying drawings.
Detailed Description
Referring to fig. 1 to 4, the error iterative evaluation method of the capacitive voltage transformer CVT according to the present invention specifically includes the following steps:
step S100: build laboratory simulation platform, construct CVT offline data D L The method comprises the steps of carrying out a first treatment on the surface of the The CVT offline data D L Including error truth data D L1 Fix error data D L2 Cycle error data D L3 Gradual error data D L4 Random error data D L5 ；
Step S200: using a first predictive model f 1 (. Cndot.) second prediction model f 2 (. Cndot.) third prediction model f 3 (. Cndot.) fourth prediction model f 4 (. Cndot.) fifth predictive model f 5 (. Cndot.) and sixth predictive model f 6 (. Cndot.) are predicted to obtain first prediction results y respectively 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 Fifth prediction result y 5 And a sixth prediction result y 6 ；
Using a first predictive model f 1 (. Cndot.) for error truth data D L1 Predicting to obtain the firstA prediction result y 1 ；
Using a second predictive model f 2 (. Cndot.) for fixed error data D L2 Predicting with the first prediction result to obtain a second prediction result y 2 ；
Using a third predictive model f 3 (. Cndot.) for periodic error data D L3 Predicting with the second prediction result to obtain a third prediction result y 3 ；
Using a fourth predictive model f 4 (. Cndot.) for gradation error data D L4 Predicting with the third predicted result to obtain a fourth predicted result y 4 ；
Using a fifth predictive model f 5 (. Cndot.) for random error data D L5 Predicting with the fourth predicted result to obtain a fifth predicted result y 5 ；
Using the first prediction result y 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 And a fifth prediction result y 5 Training to obtain a sixth predictive model f 6 （·）；
y 1 =f 1 （D L1 ） （6）；
y 2 =f 2 （y 1 ，D L2 ） （7）；
y 3 =f 3 （y 2 ，D L3 ） （8）；
y 4 =f 4 （y 3 ，D L4 ） （9）；
y 5 =f 5 （y 4 ，D L5 ） （10）；
y 5 =f 6 （y 1 ，y 2 ，y 3 ，y 4 ，y 5 ） （11）；
Wherein the first predictive model f 1 (. Cndot.) second prediction model f 2 (. Cndot.) third prediction model f 3 (. Cndot.) fourth prediction model f 4 (. Cndot.) fifth predictive model f 5 (. Cndot.) and sixth predictive model f 6 The process of formation of (-) is:
the error true value data at the time t-1 is used as input, the error true value data at the time t is used as output, and the LSTM algorithm model is trained to obtain a first prediction model f 1 （·）；
A first prediction model f at time t-1 using fixed error data at time t-1 1 The output of (-) is taken as input, error true value data at the time t and fixed error data at the time t are taken as output, and the first prediction model f is obtained 1 Training to obtain a second prediction model f 2 （·）；
A second prediction model f at time t-1 using the cycle error data at time t-1 2 The output of (-) is taken as input, error true value data at the time t, fixed error data at the time t and periodic error data at the time t are taken as output, and the second prediction model f is obtained 2 Training to obtain a third prediction model f 3 （·）；
The third prediction model f is outputted by using the gradation error data at time t-1 and the output of the third prediction model at time t-1 as inputs, and the error true value data at time t, the fixed error data at time t, the period error data at time t, and the gradation error data at time t as outputs 3 Training to obtain a fourth prediction model f 4 （·）；
The random error data at the time t-1 and the output of the fourth prediction model at the time t-1 are used as input, and the error true value data at the time t, the fixed error data at the time t, the periodic error data at the time t, the gradual error data at the time t and the random error data at the time t are used as output, so that the fourth prediction model f 4 Training to obtain a fifth prediction model f 5 （·）；
Using the first prediction result y at time t-1 1 Second prediction result y at time t-1 2 Third prediction result y at time t-1 3 Fourth prediction result y at time t-1 4、 Fifth prediction result y at time t-1 5 As input, error truth data at time t, fixed at time tError data, periodic error data at time t, gradual error data at time t, random error data at time t are output to a fifth predictive model f 5 Training to obtain a sixth prediction model f 6 （·）；
Step S300: the sixth predictive model f in step S200 is calculated by the transfer learning algorithm 6 (. Cndot.) migration to CVT real-time conditions to obtain a seventh predictive model f 7 （·）；
Step S400: acquiring online CVT error data, and decomposing the online CVT error data into an error period term, an error trend term and an error remainder by using an STL decomposition method;
step S500: using a seventh predictive model f 7 (. Cndot.) obtaining the seventh prediction result y 7 Eighth prediction result y 8 And a ninth prediction result y 9 ；
Step S600: using the seventh predictor y 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 （·）；
Step S700: using an eighth predictive model f 8 (. Cndot.) the error of the CVT to be measured is predicted.
The time series decomposition algorithm STL (Seasonal and Trend decomposition using Loess) is a very general and robust and widely used method of decomposing time series, where Loess is a method of estimating nonlinear relationships. Time series refers to data that varies over time, such as stock prices, air temperatures, sales, etc. Time series decomposition is the decomposition of a time series into different components to better understand its characteristics and regularity. The basic idea of the STL algorithm is to decompose the time sequence into three parts: seasonal, trending, and randomness. Seasonal refers to a pattern of cyclic occurrences in a time series, such as the four seasons of the year or weekdays and weekends of the week, etc. Trending refers to a long-term trend of the time series, such as rising or falling year by year. Randomness refers to irregular fluctuations in time series.
D L =（D L1 ，D L2 ，D L3 ，D L4 ，D L5 ） （1）；
In the formula (1), D L1 Simulating the output of a normal CVT for an experimental simulation platform; d (D) L2 The method is characterized in that a fixed error generated when a switching phenomenon occurs to a secondary load of the CVT is simulated in an experimental simulation platform; d (D) L3 The method is used for simulating a periodic error generated under the influence of temperature and humidity in an experimental simulation platform; d (D) L4 The method comprises the steps of simulating gradual change errors generated by gradual change of parameters of a CVT capacitive voltage division unit in an experimental simulation platform; d (D) L5 The random error generated by simulating the power grid frequency in the experimental simulation platform is generated.
D L2 =C （2）；
D L3 =F t （3）；
D L4 =G t （4）；
D L5 ～N（0，σ 2 ） （5）；
Wherein C is a constant, F t As a periodic variation function, in one embodiment, one can assume that（/>、/>、/>Constant), G t As a gradient function, in one of the embodiments they can assume +.>（/>、/>Constant), σ is variance; f (f) 1 （·）~f 8 The base models of (-) can be LSTM algorithms, with only differences in parameter settings.
In specific implementation, in the step 300, the pre-training process is as follows: seventh prediction model f 7 The pretrained model parameters of (-) are set to be the same as the sixth predictive model f 6 Parameters of (-) are consistent.
In particular, in the step 300, a seventh prediction model f 7 The training process of (-) is that the online error data at the time t-1 is used as input, the online error data at the time t is used as output, and a sixth prediction model f is obtained 6 Training to obtain a seventh prediction model f 7 （·）。
In particular, in step 400, online CVT error dataCan be decomposed into STL:
(12)
in the formula (12), the amino acid sequence of the compound,、/>and->And respectively representing an error period term, an error trend term and an error remainder, wherein t is the sampling moment.
In specific implementation, in the step 400, the STL decomposition method includes an inner loop iteration process and an outer loop iteration process.
In specific implementation, the (i+1) th iteration process of the STL decomposition method comprises the following steps:
step S1: historical error data for CVTCarrying out trending item removal treatment;
step S2: a sub-sequence smoothing step;
step S3: calculating a trend item component of the new time sequence and a seasonal component of the new time sequence;
step S4: obtaining a seasonal component and a trend component of the original sequence;
step S5: judging whether the seasonal component and the trend component of the original sequence are converged or not; and (4) jumping out of the inner loop after convergence to obtain the remainder component of the time sequence.
The specific decomposition process of STL can be divided into two steps: internal and external circulation. At the time of internal loop iteration, the smooth period update is mainly utilizedAnd->Removing abnormal disturbance of each component; during the iteration of the outer loop, the method is mainly used for calculating +.>。
In the STL decomposition method of the invention, the process of the (i+1) th iteration is as follows:
step S1: historical error data for CVTAnd (5) carrying out trending item removal treatment:
（13）
in the formula (13), the amino acid sequence of the compound,representing the result after removing the trend item; />Representing the result of the ith iteration of the trend component.
Step S2: the subsequence is smoothed. Carrying out local weighted regression (Lowess) smoothing treatment on each subsequence, and extending 1 time point before and after each subsequence, and combining to obtain a new subsequence;
step S3: calculating trend term components of the new time series using low pass filtering, removing periodic differences, and then calculating seasonal components of the new time series from the additivity of the time seriesThe method comprises the following steps:
（14）
in the formula (14), the amino acid sequence of the compound,a seasonal component representing the new time series; />Representing a new time sequence;a trend term component representing a time series.
Step S4: removing seasonal components from the sub-sequences to obtain seasonal and trend components of the original sequence, i.e
（15）
（16）
In the formulas (15) and (16),representing trend components of the (i+1) th iteration, lowess represents locally weighted regression, +.>Representing the seasonal component of the i+1st iteration.
Step S5: judging whether the convergence is carried out, if the convergence is not carried out, repeating the processes from the step S1 to the step S4; if the convergence is carried out, the internal circulation is jumped out, and the remainder component of the time sequence is obtained, namely:
（17）
in the formula (17), the amino acid sequence of the compound,and the time sequence remainder component obtained by the end of the inner loop is represented.
In specific implementation, in the step 500, a seventh prediction model f is used 7 (. Cndot.) time-domain multi-head self-attention prediction is carried out on the error period term to obtain a seventh prediction result y 7 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) frequency domain multi-head self-attention prediction is carried out on the error remainder to obtain an eighth prediction result y 8 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) predicting the trend term to obtain the ninth predicted result y 9 。
The principle of the multi-head self-attention structure is as follows:
（18）
（19）
（20）
（21）
（22）
（23）
in the formulas (18) to (23), I is an input matrix; q is a Query matrix (Query); k is a Key matrix (Key); v is a Value matrix (Value); w (W) Q 、W K 、W V Respectively corresponding conversion matrixes; attention (&) is a scaled dot product Attention function that first computes an Attention score matrixMultiplying the obtained product with matrix V to obtain corresponding output d K A dimension of K; head part i For the i-th group self-attention output, after obtaining h-th group of different self-attention outputs, it is spliced (connected) and passed through a conversion matrix W O And converting the spliced matrix into an output vector with a specified length, namely a multi-head self-attention weight vector. Wherein W is Q 、W K 、W V And W is O Are all parameters to be learned.
In the invention, the process of performing time-domain multi-head self-attention prediction by the error period term is shown in fig. 3. The process of performing frequency domain multi-head self-attention prediction on the error residuals is shown in fig. 4.
（24）
（25）
（26）
In the formulas (24) to (26),αweights are weighted for the temporal multi-headed self-attention,βfor the time domain multi-headed self-attention weighting, F (·) is a fourier transform function, specifically:
（27）
wherein,,for frequency +.>As a complex function, t is time.
In particular, in step 600, the seventh prediction result y is used 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 The specific process of the (-) is as follows: the input is as follows: seventh prediction result y at time t-1 7 Eighth prediction result y 8 And a ninth prediction result y 9 The method comprises the steps of carrying out a first treatment on the surface of the The pre-training model is as follows: seventh prediction model f 7 (. Cndot.); the output is: the CVT error value is online at time t.
The invention also discloses an electronic device, which comprises at least one processor and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of iterative error assessment of the capacitive voltage transformer CVT.
The invention also discloses a computer readable storage medium storing a computer program. The computer program when executed by the processor implements the method for iterative error assessment of the capacitive voltage transformer CVT.
According to the error iterative evaluation method of the capacitive voltage transformer CVT, the error of the CVT to be tested is subjected to iterative evaluation, the abnormal state of the CVT is evaluated in real time, and therefore the stability and the safety performance of the operation of a power grid are guaranteed.
The error iterative evaluation method of the capacitive voltage transformer CVT has the following technical advantages.
1. By adopting an off-line training-on-line modeling method, the problem of few data samples in actual working conditions is solved;
2. the accuracy of the prediction model is improved by a method of adding difference iteration on normal data;
3. according to the characteristics of CVT time sequence data, a method of combining time domain self-attention prediction and frequency domain self-attention prediction is adopted, so that the accuracy of a prediction model is improved.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.
Claims (10)
1. The error iterative evaluation method of the capacitive voltage transformer CVT is characterized by comprising the following steps of:
step S100: build laboratory simulation platform, construct CVT offline data D L The method comprises the steps of carrying out a first treatment on the surface of the The CVT offline data D L Including error truth data D L1 Fix error data D L2 Cycle error data D L3 Gradual error data D L4 Random error data D L5 ；
Step S200: using a first predictive model f 1 (. Cndot.) second prediction model f 2 (. Cndot.) third prediction model f 3 (. Cndot.) fourth prediction model f 4 (. Cndot.) fifth predictive model f 5 (. Cndot.) and (b)Six predictive model f 6 (. Cndot.) are predicted to obtain first prediction results y respectively 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 Fifth prediction result y 5 And a sixth prediction result y 6 ；
Using a first predictive model f 1 (. Cndot.) for error truth data D L1 Predicting to obtain a first prediction result y 1 ；
Using a second predictive model f 2 (. Cndot.) for fixed error data D L2 Predicting with the first prediction result to obtain a second prediction result y 2 ；
Using a third predictive model f 3 (. Cndot.) for periodic error data D L3 Predicting with the second prediction result to obtain a third prediction result y 3 ；
Using a fourth predictive model f 4 (. Cndot.) for gradation error data D L4 Predicting with the third predicted result to obtain a fourth predicted result y 4 ；
Using a fifth predictive model f 5 (. Cndot.) for random error data D L5 Predicting with the fourth predicted result to obtain a fifth predicted result y 5 ；
Using the first prediction result y 1 Second prediction result y 2 Third prediction result y 3 Fourth prediction result y 4 And a fifth prediction result y 5 Training to obtain a sixth predictive model f 6 （·）；
y 1 =f 1 （D L1 ） （6）；
y 2 =f 2 （y 1 ，D L2 ） （7）；
y 3 =f 3 （y 2 ，D L3 ） （8）；
y 4 =f 4 （y 3 ，D L4 ） （9）；
y 5 =f 5 （y 4 ，D L5 ） （10）；
y 5 =f 6 （y 1 ，y 2 ，y 3 ，y 4 ，y 5 ） （11）；
Step S300: the sixth predictive model f in step S200 is calculated by the transfer learning algorithm 6 (. Cndot.) migration to CVT real-time conditions to obtain a seventh predictive model f 7 （·）；
Step S400: acquiring online CVT error data, and decomposing the online CVT error data into an error period term, an error trend term and an error remainder by using an STL decomposition method;
step S500: using a seventh predictive model f 7 (. Cndot.) obtaining the seventh prediction result y 7 Eighth prediction result y 8 And a ninth prediction result y 9 ；
Step S600: using the seventh predictor y 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 （·）；
Step S700: using an eighth predictive model f 8 (. Cndot.) the error of the CVT to be measured is predicted.
2. The method for iterative error assessment of a capacitive voltage transformer CVT according to claim 1, characterized in that in step 300, the pre-training process is: seventh prediction model f 7 The pretrained model parameters of (-) are set to be the same as the sixth predictive model f 6 Parameters of (-) are consistent.
3. The method for iterative error assessment of a capacitive voltage transformer CVT according to claim 2, characterized in that in step 300, a seventh prediction model f 7 The training process of (-) is that the online error data at the time t-1 is used as input, the online error data at the time t is used as output, and a sixth prediction model f is obtained 6 Training to obtain a seventh prediction model f 7 （·）。
4. According toThe method for iterative error assessment of a capacitive voltage transformer CVT according to claim 1, characterized in that in step 400, the CVT error is onlineThe data may be decomposed into STLs:
(12)
in the formula (12), the amino acid sequence of the compound,、/>and->And respectively representing an error period term, an error trend term and an error remainder, wherein t is the sampling moment.
5. The method for iterative error assessment of a capacitive voltage transformer CVT according to claim 4, wherein in step 400, the STL decomposition method includes an inner loop iteration process and an outer loop iteration process.
6. The iterative error assessment method for a capacitive voltage transformer CVT of claim 5, characterized in that the process of the (i+1) th iteration of the STL decomposition method comprises the steps of:
step S1: historical error data for CVTCarrying out trending item removal treatment;
step S2: a sub-sequence smoothing step;
step S3: calculating a trend item component of the new time sequence and a seasonal component of the new time sequence;
step S4: obtaining a seasonal component and a trend component of the original sequence;
step S5: judging whether the seasonal component and the trend component of the original sequence are converged or not; and (4) jumping out of the inner loop after convergence to obtain the remainder component of the time sequence.
7. The method for iterative error assessment of a capacitive voltage transformer CVT according to claim 1, characterized in that in step 500, a seventh prediction model f is used 7 (. Cndot.) time-domain multi-head self-attention prediction is carried out on the error period term to obtain a seventh prediction result y 7 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) frequency domain multi-head self-attention prediction is carried out on the error remainder to obtain an eighth prediction result y 8 The method comprises the steps of carrying out a first treatment on the surface of the Using a seventh predictive model f 7 (. Cndot.) predicting the trend term to obtain the ninth predicted result y 9 。
8. The method of claim 1, wherein in the step 600, a seventh prediction result y is used 7 Eighth prediction result y 8 And a ninth prediction result y 9 Training to obtain an eighth prediction model f 8 The specific process of the (-) is as follows: the input is as follows: seventh prediction result y at time t-1 7 Eighth prediction result y 8 And a ninth prediction result y 9 The method comprises the steps of carrying out a first treatment on the surface of the The pre-training model is as follows: seventh prediction model f 7 (. Cndot.); the output is: the CVT error value is online at time t.
9. An electronic device comprising at least one processor and a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of iterative error assessment of a capacitive voltage transformer CVT of any one of claims 1 to 8.
10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the method of iterative error assessment of a capacitive voltage transformer CVT according to any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310980854.XA CN116702630B (en) | 2023-08-07 | 2023-08-07 | Error iterative evaluation method for capacitive voltage transformer CVT |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310980854.XA CN116702630B (en) | 2023-08-07 | 2023-08-07 | Error iterative evaluation method for capacitive voltage transformer CVT |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116702630A true CN116702630A (en) | 2023-09-05 |
| CN116702630B CN116702630B (en) | 2023-11-10 |
Family
ID=87832580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310980854.XA Active CN116702630B (en) | 2023-08-07 | 2023-08-07 | Error iterative evaluation method for capacitive voltage transformer CVT |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116702630B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117972537A (en) * | 2024-04-02 | 2024-05-03 | 国网山东省电力公司营销服务中心(计量中心) | Voltage transformer metering state evaluation method and system based on wide area measurement |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107544046A (en) * | 2016-06-24 | 2018-01-05 | 中国电力科学研究院 | A kind of online error measuring system of voltage transformer and method |
| CN113447783A (en) * | 2021-08-30 | 2021-09-28 | 武汉格蓝若智能技术有限公司 | Voltage transformer insulation fault identification model construction method and device |
| CN113567904A (en) * | 2021-07-02 | 2021-10-29 | 中国电力科学研究院有限公司 | Method and system suitable for metering error of capacitive mutual inductor |
| US20220037879A1 (en) * | 2018-09-28 | 2022-02-03 | Abb Power Grids Switzerland Ag | Method and device for controlling at least one circuit breaker of a power system |
| CN116049629A (en) * | 2023-03-29 | 2023-05-02 | 国网福建省电力有限公司 | Voltage transformer error state prediction method, system, equipment and medium |
-
2023
- 2023-08-07 CN CN202310980854.XA patent/CN116702630B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107544046A (en) * | 2016-06-24 | 2018-01-05 | 中国电力科学研究院 | A kind of online error measuring system of voltage transformer and method |
| US20220037879A1 (en) * | 2018-09-28 | 2022-02-03 | Abb Power Grids Switzerland Ag | Method and device for controlling at least one circuit breaker of a power system |
| CN113567904A (en) * | 2021-07-02 | 2021-10-29 | 中国电力科学研究院有限公司 | Method and system suitable for metering error of capacitive mutual inductor |
| CN113447783A (en) * | 2021-08-30 | 2021-09-28 | 武汉格蓝若智能技术有限公司 | Voltage transformer insulation fault identification model construction method and device |
| CN116049629A (en) * | 2023-03-29 | 2023-05-02 | 国网福建省电力有限公司 | Voltage transformer error state prediction method, system, equipment and medium |
Non-Patent Citations (2)
| Title |
|---|
| 卞志刚: "电流互感器误差测量结果的不确定度评定", 《第十一届全国电工数学学术年会》 * |
| 张竹: "电容式电压互感器计量误差状态评估和预测方法研究", 《中国优秀博士论文全文数据库》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117972537A (en) * | 2024-04-02 | 2024-05-03 | 国网山东省电力公司营销服务中心(计量中心) | Voltage transformer metering state evaluation method and system based on wide area measurement |
| CN117972537B (en) * | 2024-04-02 | 2024-07-19 | 国网山东省电力公司营销服务中心(计量中心) | Voltage transformer metering state evaluation method and system based on wide area measurement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116702630B (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tan et al. | Prediction of water quality time series data based on least squares support vector machine | |
| CN111191191B (en) | Construction method of combined model for accurately predicting deformation effect of concrete dam | |
| Fan et al. | A novel online estimation scheme for static voltage stability margin based on relationships exploration in a large data set | |
| CN109948833A (en) | A kind of Hydropower Unit degradation trend prediction technique based on shot and long term memory network | |
| CN116702630B (en) | Error iterative evaluation method for capacitive voltage transformer CVT | |
| CN103996084B (en) | Wind power probability Forecasting Methodology based on longitudinal moment Markov chain model | |
| CN111582551A (en) | Method and system for predicting short-term wind speed of wind power plant and electronic equipment | |
| CN112100711B (en) | ARIMA and PSO-ELM based concrete dam deformation combined forecasting model construction method | |
| Liu et al. | Combined forecasting method of dissolved gases concentration and its application in condition-based maintenance | |
| CN110991725B (en) | RBF ultra-short-term wind power prediction method based on wind speed frequency division and weight matching | |
| CN109840691A (en) | Non-intrusion type subitem electricity estimation method based on deep neural network | |
| CN103778341B (en) | Method for estimating probabilistic load flow of harmonic pollution user | |
| Xiyun et al. | Wind power probability interval prediction based on bootstrap quantile regression method | |
| CN103226737B (en) | Based on the chemical industry unusual service condition trend forecasting method of Kalman filtering and gray prediction | |
| CN107134813A (en) | A kind of power distribution network photovoltaic exports equilibrium index Forecasting Methodology with energy storage active power | |
| Xu et al. | Short-term wind speed prediction based on GRU | |
| CN116840767A (en) | Electric energy metering device overall error assessment method and device, storage medium and terminal | |
| Xiao et al. | Remaining useful life prediction of lithium-ion battery based on unscented Kalman filter and back propagation neural network | |
| CN114971024A (en) | Fan state prediction method and device | |
| Zheng et al. | Short-term load forecasting of BP network based on EMD | |
| CN115932704A (en) | CVT error measurement method and system based on model and data dual drive | |
| CN112581311B (en) | Method and system for predicting long-term output fluctuation characteristics of aggregated multiple wind power plants | |
| CN114487975A (en) | Remote resonance monitoring equipment, method and equipment for voltage transformer and storage medium | |
| Zhang et al. | Health status characterization and RUL prediction method of escalator bearing based on VMD-GRU | |
| CN113722977B (en) | Gas turbine rotor fault early warning method based on hybrid prediction |
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 |