Dry-Type Air-Core Reactor Fault Detection Method, System, Terminal and Storage Medium

Abstract

The present invention relates to the technical field of data processing, and specifically provides a method, system, terminal and storage medium for detecting faults in dry-type air-core reactors, including: performing an attenuated oscillation voltage test on a target dry-type air-core reactor, and obtaining the oscillation voltage waveform and inductance value generated by the target dry-type air-core reactor during the test; obtaining the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor; calculating the peak difference sequence and zero-crossing time difference sequence between the oscillation voltage waveform and the standard voltage waveform; inputting the peak difference sequence and zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features; inputting the difference between the inductance value and the standard inductance value, and the waveform features into a pre-trained random forest model to obtain a fault diagnosis result. The present invention improves the accuracy of fault detection for dry-type air-core reactors.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to a method, a system, a terminal and a storage medium for detecting faults of dry-type air-core reactors. Background Art

[0002] As an inductive high-voltage electrical appliance, the air-core reactor plays a key role in the power system in restricting short-circuit current, realizing reactive power compensation and phase shifting. Its uniqueness lies in that the magnetic flux path forms a loop through air, so it is named the air-core reactor. However, dry-type air-core reactors often encounter challenges during actual operation. In particular, fire incidents caused by insulation damage are not uncommon. These major faults often originate from tiny inter-turn insulation defects, which may even be latent at the initial stage when the reactor leaves the factory. If these insulation hidden dangers cannot be detected in time, the power system may face the risk of shutdown and even more severe accidents, causing significant economic losses to the power operation department. In view of this, it is particularly urgent and important to implement strict factory inspections on the reactor in order to identify potential insulation defects as early as possible.

[0003] Currently, the industry uses the oscillating voltage detection method to diagnose insulation faults. The specific operation is to manually observe the oscillating voltage waveform presented on the oscilloscope. However, this method has limitations and can only make a judgment when the fault signs are obvious, and it is difficult to capture subtle abnormalities. Summary of the Invention

[0004] Aiming at the above deficiencies of the prior art, the present invention provides a method, a system, a terminal and a storage medium for detecting faults of dry-type air-core reactors to solve the above technical problems.

[0005] In the first aspect, the present invention provides a method for detecting faults of dry-type air-core reactors, including:

[0006] Performing a decaying oscillating voltage test on a target dry-type air-core reactor, and obtaining the oscillating voltage waveform and inductance value generated by the target dry-type air-core reactor during the test;

[0007] Obtaining the standard voltage waveform and standard inductance value of the target dry-type air-core reactor stored in advance;

[0008] Calculating the peak difference sequence and zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform;

[0009] Inputting the peak difference sequence and zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features;

[0010] Inputting the difference between the inductance value and the standard inductance value, and the waveform features into a pre-trained random forest model to obtain a fault diagnosis result.

[0011] In an alternative embodiment, obtaining the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor includes:

[0012] According to the model information of the target dry-type air-core reactor, retrieving the corresponding standard voltage waveform and standard inductance value from the database, where the database stores the standard voltage waveforms and standard inductance values of multiple models of dry-type air-core reactors.

[0013] In an alternative embodiment, calculating the peak difference sequence and zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform includes:

[0014] Calculating the peak difference and zero-crossing time difference of the corresponding periods between the oscillating voltage waveform and the standard voltage waveform, and sorting the peak difference and zero-crossing time difference in the order of the periods to obtain the peak difference sequence and the zero-crossing time difference sequence.

[0015] In an alternative embodiment, the method further includes:

[0016] Collecting dry-type air-core reactors with minor defects as defect samples, performing damped oscillating voltage tests on the defect samples to obtain the oscillating voltage waveforms and inductance values of the defect samples under various oscillating voltages;

[0017] Collecting normal dry-type air-core reactors as normal samples, performing damped oscillating voltage tests on the normal samples to obtain the oscillating voltage waveforms and inductance values of the normal samples under various oscillating voltages;

[0018] Marking the oscillating voltage waveforms and inductance values of the defect samples as defect test data and saving them to the data set, and marking the oscillating voltage waveforms and inductance values of the normal samples under various oscillating voltages as normal test data and saving them to the data set;

[0019] Using the data set to train and validate the LSTM model and the random forest model.

[0020] In a second aspect, the present invention provides a dry-type air-core reactor fault detection system, including:

[0021] A data acquisition module for performing a damped oscillating voltage test on the target dry-type air-core reactor and obtaining the oscillating voltage waveform and inductance value generated by the target dry-type air-core reactor during the test;

[0022] A standard acquisition module for obtaining the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor;

[0023] A difference calculation module for calculating the peak difference sequence and zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform;

[0024] A feature extraction module, configured to input the peak difference sequence and the zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features;

[0025] A result diagnosis module, configured to input the difference between the inductance value and the standard inductance value, and the waveform features into a pre-trained random forest model to obtain a fault diagnosis result.

[0026] In an optional embodiment, the standard acquisition module includes:

[0027] A standard acquisition unit, configured to retrieve corresponding standard voltage waveforms and standard inductance values from a database according to the model information of the target dry-type air-core reactor, where the database stores standard voltage waveforms and standard inductance values of multiple models of dry-type air-core reactors.

[0028] In an optional embodiment, the difference calculation module includes:

[0029] A difference calculation unit, configured to calculate the peak difference and the zero-crossing time difference of the corresponding periods between the oscillating voltage waveform and the standard voltage waveform, sort the peak difference and the zero-crossing time difference in the order of the periods, and obtain a peak difference sequence and a zero-crossing time difference sequence.

[0030] In an optional embodiment, the system further includes:

[0031] A first collection module, configured to collect dry-type air-core reactors with minor defects as defect samples, perform damped oscillation voltage tests on the defect samples, and obtain the oscillating voltage waveforms and inductance values of the defect samples under multiple oscillating voltages;

[0032] A second collection module, configured to collect normal dry-type air-core reactors as normal samples, perform damped oscillation voltage tests on the normal samples, and obtain the oscillating voltage waveforms and inductance values of the normal samples under multiple oscillating voltages;

[0033] A data marking module, configured to mark the oscillating voltage waveforms and inductance values of the defect samples as defect test data and save them to a data set, and mark the oscillating voltage waveforms and inductance values of the normal samples under multiple oscillating voltages as normal test data and save them to the data set;

[0034] A training execution module, configured to train and validate the LSTM model and the random forest model using the data set.

[0035] In a third aspect, a terminal is provided, including:

[0036] A processor and a memory, where

[0037] The memory is used to store a computer program,

[0038] The processor is used to call and run the computer program from the memory, so that the terminal executes the method of the terminal described above.

[0039] In a fourth aspect, a computer storage medium is provided. Instructions are stored in the computer-readable storage medium, and when they run on a computer, the computer is made to execute the methods described in the above aspects.

[0040] The beneficial effects of the present invention are as follows. The dry-type air-core reactor fault detection method, system, terminal and storage medium provided by the present invention extract the difference parameters between the oscillating voltage waveform and the standard voltage waveform, and use the LSTM model to extract the difference time series features from the difference parameters. The random forest model is used to accurately capture small anomalies based on the inductance difference and the difference time series features, greatly improving the accuracy of dry-type air-core reactor fault detection.

[0041] In addition, the design principle of the present invention is reliable and the structure is simple, having a very wide application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0043] Figure 1 is a schematic flowchart of the method according to an embodiment of the present invention.

[0044] Figure 2 is a schematic diagram of the principle of the damped oscillating voltage test device according to an embodiment of the present invention.

[0045] Figure 3 is a schematic block diagram of the system according to an embodiment of the present invention.

[0046] Figure 4 is a schematic structural diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of this invention herein are for the purpose of describing specific embodiments only and are not intended to limit the invention.

[0049] The dry-type air-core reactor fault detection method provided by the embodiments of this invention is executed by a computer device. Correspondingly, the dry-type air-core reactor fault detection system runs in the computer device.

[0050] Figure 1 It is a schematic flowchart of the method of an embodiment of this invention. Among them, Figure 1 The execution subject can be a dry-type air-core reactor fault detection system. According to different requirements, the order of the steps in this flowchart can be changed, and some can be omitted.

[0051] As Figure 1 shown, this method includes:

[0052] Step 110, perform a damped oscillation voltage test on the target dry-type air-core reactor, and obtain the oscillation voltage waveform and inductance value generated by the target dry-type air-core reactor during the test;

[0053] Step 120, obtain the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor;

[0054] Step 130, calculate the peak difference sequence and zero-crossing time difference sequence between the oscillation voltage waveform and the standard voltage waveform;

[0055] Step 140, input the peak difference sequence and zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features;

[0056] Step 150, input the difference between the inductance value and the standard inductance value, and the waveform features into a pre-trained random forest model to obtain a fault diagnosis result.

[0057] For the convenience of understanding this invention, the following further describes the dry-type air-core reactor fault detection method provided by this invention in combination with the principle of the dry-type air-core reactor fault detection method of this invention and the process of fault detection of the dry-type air-core reactor in the embodiments.

[0058] Specifically, the dry-type air-core reactor fault detection method includes:

[0059] S1. Perform a damped oscillation voltage test on the target dry-type air-core reactor, and obtain the oscillation voltage waveform and inductance value generated by the target dry-type air-core reactor during the test.

[0060] The damped oscillation voltage test device is asFigure 2 As shown, it includes a first resistor R1, a second resistor R2, a voltage-dividing resistor R H and the voltage-dividing resistor R L , a sphere-gap switch S, a capacitor C C , the capacitor C H , the capacitor C L , the capacitor C, an inductor L and a diode D; the diode D, the first resistor R1 and the sphere-gap switch S are connected in series in turn across both ends of the output side of the target dry-type air-core reactor, and the second resistor R2, the voltage-dividing resistor R H and the voltage-dividing resistor R L are connected in series in turn and are connected in parallel with the sphere-gap switch S; the branch in which the capacitor C C , the capacitor C H , the capacitor C L are connected in series in turn is connected in parallel with the branch in which the voltage-dividing resistor R H and the voltage-dividing resistor R L are connected in series; the inductor L is connected in parallel with the branch in which the capacitor C H , the capacitor C L are connected in series; the capacitor C is connected in parallel with the inductor L.

[0061] The resistor voltage divider is used to measure the DC charging voltage value of the main capacitor C C , and the capacitor voltage divider is used to measure the exponentially decaying oscillating voltage output by the device. The impedance values of both are very large, and their existence is ignored when analyzing the voltage-current relationship of the circuit. The working process of the original circuit is described as follows: The sphere-gap switch is controlled to conduct once per power frequency voltage cycle, that is, the main capacitor discharges to the inductor once per power frequency power supply cycle, just meeting the requirement of 4000 discharges per minute specified by the standard. In the negative half-cycle of the output voltage of the test transformer, the test transformer charges the main capacitor through the protection resistor, the damping resistor, and the inductor coil. Compared with the impedance of the capacitor, the impedances of the protection resistor, the damping resistor, and the inductor coil can be ignored. Therefore, the main capacitor is charged with the peak voltage of the test transformer, the potential of the left plate is the negative peak value, and the potential of the right plate is zero. In the positive half-cycle of the output voltage of the test transformer, the components on the left side of the sphere-gap switch are all open-circuited. At this time, a trigger signal is given, and the sphere-gap switch discharges and conducts. Ignoring the voltage drop on the damping resistor, the potential of the left plate of the main capacitor is forced to be zero. Since the energy stored in the capacitor cannot change suddenly, the potential of the right plate of the main capacitor jumps to the positive peak value, and the main capacitor discharges to the inductor coil through the damping resistor (the capacitor is very small and can be ignored). Then the energy is converted between the electric field energy of the main capacitor and the magnetic field energy of the inductor coil until all of it is absorbed by the damping resistor and the loop resistor, and an exponentially decaying oscillating voltage is obtained on the inductor coil and the model specimen connected in parallel with it.

[0062] It can be seen from the analysis of the test principle that during the negative half - cycle of the output voltage of the test transformer, the sphere - gap switch is disconnected, and the principle circuit works in the charging state. The components involved in the work include a voltage regulator, a test transformer, a protective resistor, a damping resistor, a main capacitor, and an inductance coil. Under direct current, the impedance of the inductance coil can be ignored. Therefore, the amplitude of the voltage charged on the main capacitor depends on the output voltage of the test transformer and the resistance values of the protective resistor and the damping resistor.

[0063] To ensure that after the damping resistor completely absorbs the electric - field energy in the main capacitor (the arc in the sphere - gap switch is extinguished), the recovery time of the air insulation between the sphere - gaps is long enough to ensure the controllability of the sphere - gap switch, the triggering moment of the sphere - gap breakdown is selected at the zero - crossing moment when the output voltage of the test transformer changes from negative to positive.

[0064] The PSCAD software is used to simulate and analyze the above - mentioned circuit principle, and a simulation study on the inter - turn fault discrimination of an inductor with an inductance of 75 mH is carried out. Existing research shows that if there are short - circuited turns in a dry - type air - core reactor, the following changes will occur: short - circuited coils appear in the reactor, reducing the effective number of turns and causing the overall inductance of the reactor to decrease. At the same time, an induced electromotive force will be generated in the short - circuited turns, and the circulating current caused by the induced electromotive force will demagnetize the reactor, resulting in a decrease in the inductance. The circulating current in the short - circuited turns will also increase the overall loss of the reactor.

[0065] The high - frequency pulse oscillation test device charges the main capacitor Cc during the negative half - wave of the power supply and triggers during the positive half - wave of the power supply. The advantage of this is that the capacitor Cc is charged for 1 / 2 of the time in each cycle, ensuring that the charging voltage of the capacitor reaches the predetermined value. When the capacitor Cc is charged to the predetermined value, the ignition controller sends out a trigger signal. The trigger signal is transmitted to the trigger controller through an optical fiber, and the trigger controller ignites, causing the sphere - gap S to break down. At this time, the main capacitor Cc and the equivalent inductance L of the test - sample reactor form a damped - oscillation circuit. The oscillating discharge current gradually decays. When the current decays to a level insufficient to maintain the arc combustion, a damped oscillation is completed, and this damped - oscillation process is less than 5 ms. During the negative half - wave of the next power - frequency cycle, the main capacitor C is charged again. When the voltage reaches the predetermined value, the discharge sphere - gap is triggered at the set moment, and the ignition sphere - gap conducts again, and the RLC circuit forms a damped oscillation again.

[0066] To improve the accuracy of the test voltage applied to the reactor, the detection system does not adopt the discharge method of natural breakdown. The triggering moment of the ignition controller is precisely adjustable, and the triggering of the discharge sphere - gap of the detection system is controlled by the ignition controller. When the detection system is working, it generates a large amount of electromagnetic interference to the surrounding electronic devices. Therefore, the ignition controller does not use any electronic devices and all high - voltage - resistant devices are selected, improving the stability and reliability of the ignition controller.

[0067] Through analysis, it can be known that during the damped oscillation process of the pulse capacitor Cc and the test reactor, if there are inter-turn insulation defects in the reactor, due to the reduction of the number of turns of the reactor coil and the demagnetization effect, the fixed reactance rate of the entire reactor will change, and the oscillation frequency of the entire oscillation circuit will change accordingly; the circulating current in the short-circuited turn will increase the loss of the reactor, and the aluminum wire in the short-circuit ring will heat up severely. The attenuation speed of the voltage and current of the entire oscillation circuit will accelerate. Therefore, by comparing the voltage or current waveform frequencies and the zero-crossing changes at both ends of the test reactor under the system rated voltage and the oscillating wave test voltage, it is possible to determine whether there are inter-turn insulation defects in the reactor coil.

[0068] The pulse capacitor is used to discharge the smoothing reactor by adjusting the sphere gap to form a high-frequency oscillating voltage with an oscillating frequency of about 100 kHz on the UHV dry-type smoothing reactor. The charging voltage polarity of the capacitor is negative, and the high-frequency oscillation test sequence is as follows: 1 time of negative half-voltage; 3 times of negative full-voltage; 1 time of negative half-voltage.

[0069] On the premise of completing the medium-frequency oscillation test sequence required by the standard, finally add 1 time of half-voltage test, observe the half-voltage waveform frequency, and check whether the previous medium-frequency oscillation full-voltage test caused damage to the reactor winding.

[0070] S2. Obtain the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor.

[0071] According to the model information of the target dry-type air-core reactor, retrieve the corresponding standard voltage waveform and standard inductance value from the database, and the database stores the standard voltage waveforms and standard inductance values of various models of dry-type air-core reactors.

[0072] Specifically, according to the specific model information of the target dry-type air-core reactor, we can accurately retrieve the matching standard voltage waveform and standard inductance value from the carefully constructed database. This database is a comprehensive information resource library that widely and detailedly stores the standard voltage waveform data and standard inductance value data of various different models of dry-type air-core reactors. These data not only cover various common models on the market, but also include information on some specially customized or legacy models, ensuring wide applicability and high accuracy.

[0073] During the retrieval process, the system will first parse the input model information of the dry-type air-core reactor, including but not limited to key parameters such as its manufacturer, series number, rated capacity, and rated voltage. Subsequently, using an efficient retrieval algorithm, the database will quickly locate the data records that exactly correspond to this model. Among these records, the standard voltage waveform is presented in high-precision digital or graphical form, depicting in detail the law of voltage change over time when the reactor is in normal operation; while the standard inductance value provides a quantitative index for the reactor's impedance to alternating current and is an important parameter for evaluating its electrical performance.

[0074] In addition, this database also has a dynamic update mechanism, which can be continuously updated as new product models are launched or existing models are technically upgraded, ensuring the timeliness and reliability of the information provided. Users can not only obtain the required data through direct queries but also use the analysis tools provided by the database to further compare, analyze, and optimize the standard voltage waveform and standard inductance value, providing strong support for the design, selection, and maintenance of dry-type air-core reactors.

[0075] S3. Calculate the peak difference sequence and zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform.

[0076] Calculate the peak difference and zero-crossing time difference of the corresponding periods between the oscillating voltage waveform and the standard voltage waveform, sort the peak difference and zero-crossing time difference in the order of the periods, and obtain the peak difference sequence and zero-crossing time difference sequence.

[0077] Divide the oscillating voltage waveform into periods and mark the serial numbers for the waveforms of each period in chronological order. At the same time, divide the standard voltage waveform into periods and mark the serial numbers for the waveforms of each period in chronological order. Compare the waveforms with the same serial numbers, calculate the peak difference and zero-crossing time difference. Arrange the peak difference and zero-crossing time difference according to the serial numbers to obtain the peak difference sequence and zero-crossing time difference sequence.

[0078] For the oscillating voltage waveform, use specialized waveform analysis software or algorithms to identify its periodic characteristics, such as peaks, valleys, or specific phase change points, and based on this, divide the waveform into several complete periods. Subsequently, in chronological order, we assign an increasing serial number to each segmented periodic waveform, such as Period 1, Period 2,... and so on.

[0079] For the standard voltage waveform, we use the same method for period segmentation and serial number marking. Since the standard voltage waveform is usually a known and stable waveform, the process of its period segmentation and serial number marking may be more direct and accurate.

[0080] After completing waveform segmentation and serial number marking, the next step is to compare the oscillation voltage waveforms with the cycles having the same serial numbers in the standard voltage waveforms. The main purpose of the comparison is to find the differences between the two in terms of wave peaks and zero-crossing points.

[0081] For the calculation of the wave peak difference, compare the peak values of the two waveforms within the same cycle and calculate the voltage difference between them. This difference reflects the degree of deviation of the oscillation voltage waveform from the standard voltage waveform in terms of voltage amplitude.

[0082] For the calculation of the zero-crossing time difference, first determine the zero-crossing moments of the two waveforms within the same cycle, and then calculate the time difference between these two moments. This difference reflects the phase shift of the oscillation voltage waveform relative to the standard voltage waveform.

[0083] After calculating the wave peak differences and zero-crossing time differences for all cycles, we sort these difference values according to the serial numbers (i.e., time order). The purpose of doing this is to facilitate the observation and analysis of the variation trend of waveform differences over time.

[0084] After sorting, two sequences are obtained: the wave peak difference sequence and the zero-crossing time difference sequence. The wave peak difference sequence shows the voltage differences between the oscillation voltage waveform and the standard voltage waveform at the wave peaks in each cycle, while the zero-crossing time difference sequence reveals the variation of the phase shift of the waveform over cycles.

[0085] S4. Input the wave peak difference sequence and the zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features.

[0086] The wave peak difference sequence and the zero-crossing time difference sequence have the same length. Perform weighted summation on the corresponding elements of the wave peak difference sequence and the zero-crossing time difference sequence to obtain a comprehensive data sequence. Normalize the comprehensive data sequence to obtain a data sequence.

[0087] The LSTM (Long Short-Term Memory) model is a special type of recurrent neural network (RNN) architecture designed to address the problems of vanishing gradients and exploding gradients in traditional RNNs when dealing with long sequence data. The key of the LSTM model lies in its special cell structure, and each cell contains three gates and a cell state. These three gates are respectively:

[0088] Forget Gate: Controls how much information in the current cell state is retained or forgotten. It decides which information needs to be forgotten based on the hidden state of the previous time step and the input of the current time step.

[0089] Input Gate: Controls how much new information is stored in the cell state. It also decides which new information needs to be added based on the hidden state of the previous time step and the input of the current time step.

[0090] Output Gate: Determines how much information is output from the cell state. It calculates the output based on the current cell state and the hidden state of the previous time step.

[0091] Input the data sequence into a pre-trained LSTM model and extract waveform features from the middle layer of the LSTM model.

[0092] After calculating the period segmentation, peak difference, and zero-crossing time difference, two sequences of the same length are obtained: the peak difference sequence [d1, d2, ..., dn] and the zero-crossing time difference sequence [t1, t2, ..., tn], where n represents the number of periods.

[0093] To combine the information of the peak difference and the zero-crossing time difference, the corresponding elements of these two sequences are weighted and summed. Assume the weight assigned to the peak difference sequence is w1, and the weight assigned to the zero-crossing time difference sequence is w2 (and w1 + w2 = 1 to ensure that the weighted value is still within a reasonable range), then the comprehensive data sequence can be calculated by the following formula:

[0094] Comprehensive data sequence = [w1 * d1 + w2 * t1, w1 * d2 + w2 * t2, ..., w1 * dn + w2 * tn].

[0095] This comprehensive data sequence contains both the voltage information of the peak difference and the phase information of the zero-crossing time difference, providing a more comprehensive description of the waveform features.

[0096] Since the elements in the comprehensive data sequence may have different dimensions and value ranges, to eliminate the impact of this difference on model training, the comprehensive data sequence is normalized.

[0097] Select the min-max normalization method, then the normalized data sequence can be calculated by the following formula:

[0098] Normalized data sequence = [(element - minimum value) / (maximum value - minimum value) for element in comprehensive data sequence].

[0099] Input the normalized data sequence into a pre-trained LSTM model. The LSTM model is a type of recurrent neural network, which is good at processing time series data and can capture long-term dependencies in the data.

[0100] In the LSTM model, the data sequence is sequentially input into the input layer of the model in chronological order and then undergoes feature extraction and transformation through multiple hidden layers. To extract waveform features, an output is set on a certain intermediate layer (such as the last hidden layer) of the LSTM model to obtain the feature vector after being processed by the model.

[0101] This feature vector not only contains the information in the original data but also has undergone the non-linear transformation and feature extraction of the LSTM model, so it has stronger expressive ability and generalization ability. This feature vector is used in subsequent tasks such as fault diagnosis, classification, or regression.

[0102] S5. Input the difference between the inductance value and the standard inductance value, and the waveform feature into a pre-trained random forest model to obtain a fault diagnosis result.

[0103] Input the difference between the inductance value and the standard inductance value, and the waveform feature into a pre-trained random forest model to obtain a fault diagnosis result.

[0104] Calculation of the inductance value difference: Measure the actual inductance value of the measuring device, denoted as L_actual. Obtain the standard inductance value of this device, denoted as L_standard. Calculate the difference between the inductance value and the standard inductance value, that is, ΔL = L_actual - L_standard.

[0105] Denote the waveform feature extracted from the intermediate layer of the LSTM model as F_waveform.

[0106] Combine the inductance value difference ΔL and the waveform feature vector F_waveform into a new input vector X_input, that is:

[0107] X_input = [ΔL, F_waveform];

[0108] This input vector contains the inductance value change information and waveform feature information of the device, providing comprehensive data support for subsequent fault diagnosis.

[0109] In practical applications, input the new input vector X_input into a pre-trained random forest model. The model will traverse and vote on decision trees based on the features of the input vector and finally give a fault diagnosis result. This result may include information such as the type of fault, the severity of the fault, and the possible causes of the fault, providing strong support for the maintenance and management of the device.

[0110] In addition, before using the model, the model needs to be trained. The training methods include:

[0111] Collect dry-type air-core reactors with minor defects as defect samples, conduct damped oscillation voltage tests on the defect samples, and obtain the oscillation voltage waveforms and inductance values of the defect samples under various oscillation voltages; collect normal dry-type air-core reactors as normal samples, conduct damped oscillation voltage tests on the normal samples, and obtain the oscillation voltage waveforms and inductance values of the normal samples under various oscillation voltages; mark the oscillation voltage waveforms and inductance values of the defect samples as defect test data and save them to the dataset, and mark the oscillation voltage waveforms and inductance values of the normal samples under various oscillation voltages as normal test data and save them to the dataset.

[0112] Defect sample collection: Select dry-type air-core reactors with minor defects from the production line or those already in use as defect samples. These defects may include loose windings, aging of insulating materials, partial discharge, etc. Keep a detailed record of each defect sample, including information such as defect type, defect location, and defect degree.

[0113] Normal sample collection: Select several dry-type air-core reactors without obvious defects and operating normally from the same batch or production line as normal samples. Ensure that the normal samples are comparable to the defect samples in terms of electrical performance, structural characteristics, etc.

[0114] Test equipment preparation: Prepare a test equipment capable of generating damped oscillation voltage, and ensure that the equipment accuracy meets the test requirements. Set the parameters of the test equipment, such as voltage amplitude, frequency, decay rate, etc., to simulate the voltage fluctuations in actual operation.

[0115] Test process: Conduct damped oscillation voltage tests on each defect sample and normal sample. Record the oscillation voltage waveforms and inductance values of each sample under various oscillation voltages (such as different amplitudes, different frequencies, etc.). Ensure that the samples are in a stable state during the test to avoid the influence of external interference on the test results.

[0116] Data marking: Mark the oscillation voltage waveforms and inductance values of the defect samples as defect test data. Mark the oscillation voltage waveforms and inductance values of the normal samples as normal test data.

[0117] Data arrangement: Arrange the test data to ensure that the data format is unified, clear, and easy to understand. Number each data point for convenient subsequent data analysis and processing.

[0118] Data saving: Save the defect test data and normal test data to dedicated datasets respectively. The datasets should contain information such as the original records, processing procedures, and analysis results of the data for subsequent research and applications.

[0119] Data quality control: Perform quality control on the dataset to ensure the accuracy, integrity, and consistency of the data. Process abnormal data, such as deletion, correction, or retesting, etc.

[0120] Use the dataset to train and validate the LSTM model and the random forest model.

[0121] Use the training set to train the LSTM model. Minimize the loss function by iteratively updating the weights and biases of the model. During the training process, the validation set can be used to monitor the performance of the model, so as to implement strategies such as early stopping or adjusting the learning rate to prevent overfitting.

[0122] In the model training stage, we use a large amount of historical fault data and normal data to train the random forest model. These data include the difference in inductance values, waveform feature vectors, and corresponding fault diagnosis results. Through training, the random forest model can learn the mapping relationship between the difference in inductance values, waveform features, and fault diagnosis results.

[0123] In some embodiments, the dry-type air-core reactor fault detection system may include multiple functional modules composed of computer program segments. The computer programs of each program segment in the dry-type air-core reactor fault detection system can be stored in the memory of the computer device and executed by at least one processor to perform (see details in Figure 1 description) the functions of dry-type air-core reactor fault detection.

[0124] In this embodiment, according to the functions it performs, the dry-type air-core reactor fault detection system can be divided into multiple functional modules, such as Figure 3 shown. The functional modules of system 400 may include: a data acquisition module 410, a standard acquisition module 420, a difference calculation module 430, a feature extraction module 340, and a result diagnosis module 350. The module referred to in the present invention means a series of computer program segments that can be executed by at least one processor and can complete fixed functions, and are stored in the memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0125] The data acquisition module is used to perform a damped oscillation voltage test on the target dry-type air-core reactor and obtain the oscillation voltage waveform and inductance value generated by the target dry-type air-core reactor during the test;

[0126] The standard acquisition module is used to obtain the pre-stored standard voltage waveform and standard inductance value of the target dry-type air-core reactor;

[0127] The difference calculation module is used to calculate the peak difference sequence and zero-crossing time difference sequence between the oscillation voltage waveform and the standard voltage waveform;

[0128] A feature extraction module, configured to input the peak difference sequence and the zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features;

[0129] A result diagnosis module, configured to input the difference between the inductance value and the standard inductance value, and the waveform features into a pre-trained random forest model to obtain a fault diagnosis result.

[0130] Optionally, as an embodiment of the present invention, the standard acquisition module includes:

[0131] A standard acquisition unit, configured to retrieve corresponding standard voltage waveforms and standard inductance values from a database according to the model information of the target dry-type air-core reactor, where the database stores standard voltage waveforms and standard inductance values of multiple models of dry-type air-core reactors.

[0132] Optionally, as an embodiment of the present invention, the difference calculation module includes:

[0133] A difference calculation unit, configured to calculate the peak difference and the zero-crossing time difference of the corresponding periods between the oscillating voltage waveform and the standard voltage waveform, sort the peak difference and the zero-crossing time difference in the order of the periods, and obtain a peak difference sequence and a zero-crossing time difference sequence.

[0134] Optionally, as an embodiment of the present invention, the system further includes:

[0135] A first collection module, configured to collect dry-type air-core reactors with minor defects as defect samples, perform damped oscillating voltage tests on the defect samples, and obtain the oscillating voltage waveforms and inductance values of the defect samples under multiple oscillating voltages;

[0136] A second collection module, configured to collect normal dry-type air-core reactors as normal samples, perform damped oscillating voltage tests on the normal samples, and obtain the oscillating voltage waveforms and inductance values of the normal samples under multiple oscillating voltages;

[0137] A data marking module, configured to mark the oscillating voltage waveforms and inductance values of the defect samples as defect test data and save them to a data set, and mark the oscillating voltage waveforms and inductance values of the normal samples under multiple oscillating voltages as normal test data and save them to the data set;

[0138] A training execution module, configured to train and verify the LSTM model and the random forest model using the data set.

[0139] Figure 4 It is a schematic structural diagram of a terminal 400 provided by an embodiment of the present invention, and the terminal 400 can be used to execute the dry-type air-core reactor fault detection method provided by the embodiment of the present invention.

[0140] Among them, the terminal 400 may include: a processor 410, a memory 420, and a communication unit 430. These components communicate through one or more buses. Those skilled in the art can understand that the structure of the server shown in the figure does not constitute a limitation on the present invention. It can be a bus structure, a star structure, and may also include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0141] Among them, the memory 420 can be used to store the execution instructions of the processor 410. The memory 420 can be implemented by any type of volatile or non-volatile storage terminal or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc. When the execution instructions in the memory 420 are executed by the processor 410, the terminal 400 is enabled to execute some or all of the steps in the above method embodiments.

[0142] The processor 410 is the control center of the storage terminal, connecting various parts of the entire electronic terminal through various interfaces and lines. By running or executing the software programs and / or modules stored in the memory 420, and calling the data stored in the memory, it executes various functions of the electronic terminal and / or processes data. The processor may be composed of an integrated circuit (IC). For example, it may be composed of a single packaged IC, or may be composed of connecting multiple packaged ICs with the same or different functions. For example, the processor 410 may only include a central processing unit (CPU). In the embodiment of the present invention, the CPU may be a single operation core or may include multiple operation cores.

[0143] The communication unit 430 is used to establish a communication channel so that the storage terminal can communicate with other terminals. It receives user data sent by other terminals or sends user data to other terminals.

[0144] The present invention also provides a computer storage medium. Among them, the computer storage medium can store a program, and when the program is executed, it may include some or all of the steps in the various embodiments provided by the present invention. The storage medium may be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0145] Therefore, in the present invention, by extracting the difference parameters between the oscillating voltage waveform and the standard voltage waveform, and using the LSTM model to extract the differential timing features from the difference parameters, and using the random forest model to accurately capture the minute anomalies based on the inductance difference and the differential timing features, the accuracy of the dry-type air-core reactor fault detection is greatly improved. The technical effects achievable in this embodiment can be referred to the descriptions above and will not be elaborated here.

[0146] Those skilled in the art can clearly understand that the technologies in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solutions in the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disc, etc., which can store program codes, including several instructions for causing a computer terminal (which can be a personal computer, a server, or a second terminal, a network terminal, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0147] For the same or similar parts among the various embodiments in this specification, reference can be made to each other. In particular, for the terminal embodiments, since they are basically similar to the method embodiments, the descriptions are relatively simple, and for the relevant parts, reference can be made to the descriptions in the method embodiments.

[0148] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the systems or modules can be in electrical, mechanical or other forms.

[0149] The modules described as separate components may or may not be physically separated. The components displayed as modules may or may not be physical modules, that is, they can be located in one place, or they can be distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0150] In addition, in each embodiment of the present invention, each functional module may be integrated into one processing module, may exist separately as individual physical modules, or two or more modules may be integrated into one module.

[0151] Although the present invention has been described in detail by referring to the accompanying drawings and in conjunction with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, those of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and all such modifications or substitutions should fall within the scope of the present invention. / Any person skilled in the art within the technical scope disclosed by the present invention can easily conceive of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A dry-type air-core reactor fault detection method, characterized in that: include: Performing a damped oscillation voltage test on the target dry-type air-core reactor, and obtaining an oscillation voltage waveform and an inductance value generated by the target dry-type air-core reactor during the test; Obtaining a pre-stored target dry-type air-core reactor standard voltage waveform and standard inductance value; Calculating a peak difference sequence and a zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform; The peak difference sequence and the zero-crossing time difference sequence are input into a pre-trained LSTM model to obtain waveform features; in the LSTM model, the data sequence is input into the input layer of the model in chronological order, and then feature extraction and conversion are performed through multiple hidden layers. In order to extract the waveform features, an output is set on a certain intermediate layer of the LSTM model, so as to obtain a feature vector processed by the model; The difference between the inductance value and the standard inductance value and the waveform feature are input into a pre-trained random forest model to obtain a fault diagnosis result.

2. The method according to claim 1, characterized in that Obtain the pre-stored target dry-type air-core reactor standard voltage waveform and standard inductance value, including: According to the model information of the target dry-type air-core reactor, the corresponding standard voltage waveform and standard inductance value are retrieved from a database, wherein the database stores the standard voltage waveforms and standard inductance values ​​of various models of dry-type air-core reactors.

3. The method according to claim 1, characterized in that Calculating a peak difference sequence and a zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform, including: The peak difference and zero-crossing time difference of the corresponding periods of the oscillating voltage waveform and the standard voltage waveform are calculated, and the peak difference and zero-crossing time difference are sorted in order of period to obtain the peak difference sequence and the zero-crossing time difference sequence.

4. The method according to claim 1, characterized in that The method further comprises: Dry-type air-core reactors with tiny defects are collected as defective samples, and attenuated oscillation voltage tests are performed on the defective samples to obtain oscillation voltage waveforms and inductance values ​​of the defective samples under various oscillation voltages; Normal dry-type air-core reactors are collected as normal samples, and attenuated oscillation voltage tests are performed on the normal samples to obtain oscillation voltage waveforms and inductance values ​​of the normal samples under various oscillation voltages; Mark the oscillation voltage waveform and inductance value of the defective sample as defective test data and save them in the data set, and mark the oscillation voltage waveform and inductance value of the normal sample under various oscillation voltages as normal test data and save them in the data set; The dataset is used to train and validate the LSTM model and random forest model.

5. A dry-type air-core reactor fault detection system, characterized in that: include: A data acquisition module is used to perform an attenuated oscillation voltage test on the target dry-type air-core reactor and obtain an oscillation voltage waveform and an inductance value generated by the target dry-type air-core reactor during the test; A standard acquisition module is used to acquire a pre-stored target dry-type air-core reactor standard voltage waveform and standard inductance value; A difference calculation module, used for calculating a peak difference sequence and a zero-crossing time difference sequence between the oscillating voltage waveform and the standard voltage waveform; A feature extraction module is used to input the peak difference sequence and the zero-crossing time difference sequence into a pre-trained LSTM model to obtain waveform features; in the LSTM model, the data sequence is input into the input layer of the model in chronological order, and then feature extraction and conversion are performed through multiple hidden layers. In order to extract waveform features, an output is set on a certain intermediate layer of the LSTM model, thereby obtaining a feature vector processed by the model; The result diagnosis module is used to input the difference between the inductance value and the standard inductance value and the waveform characteristics into a pre-trained random forest model to obtain a fault diagnosis result.

6. The system according to claim 5, characterized in that The standard acquisition module includes: The standard acquisition unit is used to retrieve the corresponding standard voltage waveform and standard inductance value from a database according to the model information of the target dry-type air-core reactor, wherein the database stores the standard voltage waveforms and standard inductance values ​​of various models of dry-type air-core reactors.

7. The system according to claim 5, characterized in that The difference calculation module comprises: The difference calculation unit is used to calculate the peak difference and zero-crossing time difference of the corresponding cycles of the oscillating voltage waveform and the standard voltage waveform, and sort the peak difference and zero-crossing time difference in order of cycle to obtain the peak difference sequence and the zero-crossing time difference sequence.

8. The system according to claim 5, characterized in that The system further comprises: The first collection module is used to collect dry-type air-core reactors with tiny defects as defective samples, perform attenuated oscillation voltage tests on the defective samples, and obtain oscillation voltage waveforms and inductance values ​​of the defective samples under various oscillation voltages; The second collection module is used to collect normal dry-type air-core reactors as normal samples, perform attenuated oscillation voltage tests on the normal samples, and obtain oscillation voltage waveforms and inductance values ​​of the normal samples under various oscillation voltages; A data marking module, used to mark the oscillation voltage waveform and inductance value of the defective sample as defective test data and save them in a data set, and mark the oscillation voltage waveform and inductance value of the normal sample under various oscillation voltages as normal test data and save them in a data set; The training execution module is used to train and verify the LSTM model and random forest model using the dataset.

9. A terminal, characterized in that: include: A memory, used for storing a dry-type air-core reactor fault detection program; A processor is used to implement the steps of the dry-type air-core reactor fault detection method as described in any one of claims 1 to 4 when executing the dry-type air-core reactor fault detection program.

10. A computer-readable storage medium storing a computer program, characterized in that: The readable storage medium stores a dry-type air-core reactor fault detection program, which, when executed by a processor, implements the steps of the dry-type air-core reactor fault detection method according to any one of claims 1 to 4.

Images