A CVT metering error prediction method, device and storage medium based on S-transformation

Abstract

The present invention discloses a CVT metering error prediction method, device and storage medium based on s-transform. The method comprises the steps of: real-time acquisition of primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested; performing s-transformation processing on the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data to a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum; extracting first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively to obtain a first feature data set; performing PCA dimensionality reduction processing on the first feature data set to obtain a second feature data set; inputting the second feature data set into a CVT metering error prediction model based on a DFF neural network, and outputting a CVT metering error prediction result. The present invention improves the efficiency of CVT metering error prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of CVT metering error prediction, and in particular to a CVT metering error prediction method, device and storage medium based on s-transformation. Background Art

[0002] The existing technology uses the RBF neural network model in deep learning to realize the prediction of the error state of the mutual inductor, but its input data, namely the various physical signals sensed by the CVT, is a non-stationary random signal. The simple time series data analysis method cannot distinguish the gradual measurement error of the CVT from the inherent fluctuation of its sensor signal under the specified measurement accuracy requirement, resulting in a decrease in accuracy.

[0003] At present, when technicians check the error status of transformers during power outages, there are disadvantages of high manpower and material resources and low efficiency. At the same time, if the transformer error is not discovered in time, it will damage the fairness of power settlement and even the safety of the power grid. Summary of the invention

[0004] The present invention provides a CVT metering error prediction method, device and storage medium based on s-transformation, which improves the efficiency of CVT metering error prediction.

[0005] An embodiment of the present invention provides a CVT metering error prediction method based on s-transform, comprising the following steps:

[0006] Real-time collection of primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer under test;

[0007] Performing S transformation on the primary voltage frequency data, the ambient temperature data, the leakage current data and the secondary load data to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum respectively;

[0008] Extracting first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum respectively to obtain a first feature data set;

[0009] Performing PCA dimensionality reduction processing on the first feature data set to obtain a second feature data set, and recording the reduced features as second features;

[0010] The second feature data set is input into a CVT metering error prediction model based on a DFF neural network, and a CVT metering error prediction result is output.

[0011] Furthermore, the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are represented in time series.

[0012] Further, first feature data is extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum, respectively, specifically:

[0013] Peak index data, waveform index data, kurtosis index data, Tamura texture data, first-order color moment data, second-order color moment data and third-order color moment data are extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively.

[0014] Furthermore, the raw data for training the CVT metering error prediction model based on the DFF neural network is collected according to the following formula:

[0015]

[0016] Where N represents the number of time points that can be collected, n 1 Indicates the number of time points that do not exceed the tolerance, n 2 The number of time points that represent deviations.

[0017] Further, S transformation processing is performed according to the following formula:

[0018]

[0019] Among them, τ=xT, f=y / nT, τ is the translation factor used to control the position of the Gaussian window on the time axis t, f is the characteristic frequency of the signal, x is the time sampling point number for sampling at a certain time interval, y is the frequency sampling point number for sampling at a certain frequency interval, T is the sampling period, n is the number of sampling points, and sig represents the signal to be measured.

[0020] Furthermore, the CVT metrology error prediction model based on the DFF neural network includes k input units, 2 hidden layers, a plurality of hidden neurons and 1 output unit, wherein k is equal to the number of the second features.

[0021] Another embodiment of the present invention provides a CVT metering error prediction device based on s-transformation, including an S-transformation processing module, a PCA dimension reduction processing module and a metering error prediction module;

[0022] The method is used to collect primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested in real time, and the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are respectively subjected to S transformation processing to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum, and first feature data are respectively extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum to obtain a first feature data set;

[0023] The PCA dimension reduction processing module is used to perform PCA dimension reduction processing on the first feature data set to obtain a second feature data set, and record the reduced features as second features;

[0024] The metering error prediction module is used to input the second feature data set into the CVT metering error prediction model based on the DFF neural network, and output the CVT metering error prediction result.

[0025] Another embodiment of the present invention provides a readable storage medium, which includes a stored computer program. When the computer program is executed, the device where the readable storage medium is located is controlled to execute the CVT metering error prediction method based on s-transformation described in any method embodiment of the present invention.

[0026] The embodiments of the present invention have the following beneficial effects:

[0027] The present invention provides a CVT metering error prediction method, device and storage medium based on s-transformation. The method processes the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested in real time through s-transformation processing, thereby solving the problem of data instability. At the same time, by respectively extracting the first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum, the deep features are mined to enable deeper mining of data information. Then, PCA dimensionality reduction processing is used to further streamline the data input into the CVT metering error prediction model based on the DFF neural network, which not only improves the prediction efficiency of the CVT metering error prediction model based on the DFF neural network, but also improves the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of a CVT metering error prediction method based on s-transformation provided by an embodiment of the present invention;

[0029] Figure 2 It is a structural schematic diagram of a CVT metering error prediction device based on s-transformation provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0030] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] like Figure 1As shown, a CVT metering error prediction method based on s-transform provided by an embodiment of the present invention includes the following steps:

[0032] Step S101: real-time acquisition of primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested, wherein the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are represented in time series.

[0033] Step S102: performing S transformation on the primary voltage frequency data, the ambient temperature data, the leakage current data and the secondary load data to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum respectively.

[0034] As one embodiment, the S transform process is performed according to the following formula:

[0035]

[0036] Among them, τ=xT, f=y / nT, τ is the translation factor used to control the position of the Gaussian window on the time axis t, f is the characteristic frequency of the signal, x is the time sampling point number for sampling at a certain time interval, y is the frequency sampling point number for sampling at a certain frequency interval, T is the sampling period, n is the number of sampling points, sig represents the signal to be measured, and d is the differential sign.

[0037] Step S103: extracting first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively to obtain a first feature data set.

[0038] As one embodiment, first feature data is extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum, respectively, specifically:

[0039] Peak index data, waveform index data, kurtosis index data, Tamura texture data, first-order color moment data, second-order color moment data and third-order color moment data are extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively.

[0040] Step S104: Perform PCA dimensionality reduction processing on the first feature data set to obtain a second feature data set, and record the reduced features as second features.

[0041] Step S105: input the second feature data set into the CVT metering error prediction model based on the DFF neural network, and output a CVT metering error prediction result, which is specifically a CVT error prediction value.

[0042] As one embodiment, a training set for the CVT metering error prediction model based on the DFF neural network is obtained according to the following steps:

[0043] Step S11: Collecting the real error data under power outage detection of the site and the primary voltage frequency data, ambient temperature data, leakage current data, and secondary load data of the corresponding error data, wherein the error data includes ratio difference and angle difference.

[0044] Step S12: Assume that the number of time points (times) that can be collected is N, and the time point n that does not exceed the tolerance is 1 The time point of the extreme difference n 2 Then the amount of data collected is N*:

[0045]

[0046] That is, when the out-of-tolerance time point is greater than 9 times the non-out-of-tolerance time point, all valid data are used; when the out-of-tolerance time point is not greater than 9 times the non-out-of-tolerance time point, all out-of-tolerance time points and the nearest 9 times the non-out-of-tolerance time points are used. This method achieves full coverage of data collection and ensures the comprehensiveness of the error state to which the data belongs. The collected primary voltage frequency data, ambient temperature data, leakage current data, and secondary load data are represented in time series.

[0047] Step S13: Draw a corresponding time series diagram for the primary voltage frequency data, ambient temperature data, leakage current data, and secondary load data of each error point 10 days (i.e., 240 hours) before the error point, and use the s-transform time-frequency analysis method to obtain a corresponding time-frequency spectrum based on the time series diagram; the time-frequency spectrum includes a first time-frequency spectrum of the primary voltage frequency data, a second time-frequency spectrum of the ambient temperature data, a third time-frequency spectrum of the leakage current data, and a fourth time-frequency spectrum of the secondary load data.

[0048] Step S14: Extract the first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively to obtain the first feature data set. The first feature data extracted from the time-frequency spectrum is used as a new prediction variable instead of the original time series data as a prediction variable, which solves the problem of data non-stationarity and deeply mines the data feature information. Among them, the horizontal axis of the time-frequency spectrum corresponds to time information, the vertical axis corresponds to frequency information, and n groups of data correspond to n time-frequency spectra.

[0049] Specifically, h image features are extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively, and a total of 4h features are obtained, that is, the first feature data includes feature data corresponding to the h image features, and the first feature data set includes feature data corresponding to the 4h features.

[0050] The h image features include: peak index data, waveform index data, kurtosis index data, Tamura texture data, first-order color moment data, second-order color moment data and third-order color moment data.

[0051] Peak index: used to describe the peak height and peak surface of the spectrum;

[0052] Waveform index: reflects the difference and distortion between the actual waveform and the standard sine wave, and is used to diagnose imbalance and misalignment in the low-frequency field;

[0053] Kurtosis index: an index that describes the peak degree of the time-frequency spectrum of the CVT error influencing parameter;

[0054] Tamura texture: includes contrast and directionality, where contrast is obtained by statistically analyzing the distribution of pixel intensity in the entire image or a certain area; directionality is obtained by calculating the direction angle at each pixel point.

[0055] Color moment: Any color distribution in an image can be represented by its moment. Since color distribution information is mainly concentrated in low-order moments, the first-order moment, second-order moment, and third-order moment of color are sufficient to express the color distribution of an image. Therefore, the first feature data set includes feature data corresponding to 4h features, specifically h=8, and the first feature data set includes 32 features.

[0056] Step S15: Perform PCA dimensionality reduction processing on the first feature data set to obtain a second feature data set. The image features of the time-frequency spectrum corresponding to the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are as high as 32 items. Therefore, it is difficult to sort out the relationship between the various features. It is often easy to have two features that are strongly correlated but are both used for training. If there are a large number of such redundant features, it will greatly affect the efficiency of machine learning and extremely waste running memory and running time. Therefore, the most important new features are optimized through the PCA dimensionality reduction method, and the reduced features are recorded as the second features, corresponding to the second feature data training set.

[0057] Step S16: The CVT metrology error prediction model based on the DFF neural network is trained using the second feature data set until convergence. The CVT metrology error prediction model based on the DFF neural network includes k input units, 2 hidden layers, a plurality of hidden neurons and 1 output unit, wherein k is equal to the number of the second features.

[0058] Compared with the evaluation model based on neural network in the prior art that directly uses input data for training, the present invention processes the initial feature data through S transformation processing, solves the problem of data instability, and at the same time, excavates deep features to more deeply mine data information, and then uses PCA dimensionality reduction processing to further streamline the data input into the CVT metering error prediction model based on DFF neural network, which not only improves the prediction efficiency of the CVT metering error prediction model based on DFF neural network, but also improves the accuracy of the prediction results. The present invention does not require power outage evaluation to reduce the impact on the power grid, has a small workload, is time-consuming, and is not subject to conditions, which meets current needs.

[0059] Based on the above-mentioned embodiments of the invention, the present invention provides corresponding device embodiments, such as Figure 2 As shown;

[0060] Another embodiment of the present invention provides a CVT metering error prediction device based on s-transformation, comprising an S-transformation processing module 101, a PCA dimension reduction processing module 102 and a metering error prediction module 103;

[0061] The method is used to collect primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested in real time, and the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are respectively subjected to S transformation processing to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum, and first feature data are respectively extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum to obtain a first feature data set;

[0062] The PCA dimension reduction processing module is used to perform PCA dimension reduction processing on the first feature data set to obtain a second feature data set, and record the reduced features as second features;

[0063] The metering error prediction module is used to input the second feature data set into the CVT metering error prediction model based on the DFF neural network, and output the CVT metering error prediction result.

[0064] For the convenience and brevity of description, the device item embodiment of the present invention includes all the implementation methods in the above-mentioned CVT metering error prediction method embodiment based on s-transformation, which will not be repeated here.

[0065] On the basis of the above-mentioned invention item embodiments, the present invention provides a corresponding readable storage medium item embodiment; another embodiment of the present invention provides a readable storage medium, the readable storage medium includes a stored computer program, and when the computer program is executed, the device where the readable storage medium is located is controlled to execute the CVT metering error prediction method based on s-transform as described in any method item embodiment of the present invention.

[0066] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the terminal device.

[0067] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0068] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0069] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the terminal device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0070] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium (i.e., the above-mentioned readable storage medium). Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0071] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0072] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

[0073] Those skilled in the art can understand that all or part of the processes in the above embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above embodiments. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

Claims

1. A CVT metering error prediction method based on s-transform, characterized in that: The following steps are involved: Real-time collection of primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer under test; Performing S transformation on the primary voltage frequency data, the ambient temperature data, the leakage current data and the secondary load data to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum; Extracting first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum respectively to obtain a first feature data set; Performing PCA dimensionality reduction processing on the first feature data set to obtain a second feature data set, and recording the features after dimensionality reduction as second features; Inputting the second feature data set into a CVT metering error prediction model based on a DFF neural network, and outputting a CVT metering error prediction result; The first feature data is extracted from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum respectively, specifically: Extracting peak index data, waveform index data, kurtosis index data, Tamura texture data, first-order color moment data, second-order color moment data, and third-order color moment data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum, respectively; The raw data for training the CVT metering error prediction model based on the DFF neural network is collected according to the following formula: Where N represents the number of time points that can be collected. Indicates the number of time points that do not exceed the tolerance. The number of time points that represent deviations.

2. The CVT metering error prediction method based on s-transform according to claim 1 is characterized in that: The primary voltage frequency data, ambient temperature data, leakage current data and secondary load data are represented in time series.

3. The CVT metering error prediction method based on s-transform according to claim 1 is characterized in that: Perform S transform processing according to the following formula: in, = x T, f = y / nT , is the translation factor, which is used to control the position of the Gaussian window on the time axis t. f is the signal characteristic frequency, x The time sampling point number for sampling at a certain time interval, y is the frequency sampling point number for sampling at a certain frequency long interval, T is the sampling period, n is the number of sampling points, and sig represents the signal to be measured.

4. The CVT metering error prediction method based on s-transform according to any one of claims 1 to 3, characterized in that: The CVT metrology error prediction model based on the DFF neural network includes k input units, 2 hidden layers, a plurality of hidden neurons and 1 output unit, wherein k is equal to the number of the second features.

5. A CVT metering error prediction device based on s-transform, characterized in that: It includes S-transform processing module, PCA dimension reduction processing module and measurement error prediction module; The S-transformation processing module is used to collect primary voltage frequency data, ambient temperature data, leakage current data and secondary load data of the transformer to be tested in real time, and perform S-transformation processing on the primary voltage frequency data, ambient temperature data, leakage current data and secondary load data to obtain a first time-frequency spectrum, a second time-frequency spectrum, a third time-frequency spectrum and a fourth time-frequency spectrum, and extract first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum and the fourth time-frequency spectrum to obtain a first feature data set; The PCA dimension reduction processing module is used to perform PCA dimension reduction processing on the first feature data set to obtain a second feature data set, and record the features after dimension reduction as second features; The metering error prediction module is used to input the second feature data set into the CVT metering error prediction model based on the DFF neural network, and output the CVT metering error prediction result; The S-transform processing module extracts first feature data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum, respectively, specifically: The S-transform processing module extracts peak index data, waveform index data, kurtosis index data, Tamura texture data, first-order color moment data, second-order color moment data, and third-order color moment data from the first time-frequency spectrum, the second time-frequency spectrum, the third time-frequency spectrum, and the fourth time-frequency spectrum, respectively; The metering error prediction module is also used to collect raw data for training the CVT metering error prediction model based on the DFF neural network according to the following formula: Where N represents the number of time points that can be collected. Indicates the number of time points that do not exceed the tolerance. The number of time points that represent deviations.

6. A readable storage medium, characterized in that: The readable storage medium includes a stored computer program. When the computer program is executed, the device where the readable storage medium is located is controlled to execute the CVT metering error prediction method based on s-transformation as described in any one of claims 1 to 4.

Images