Rapid evaluation device for solid impurity content of transformer insulating oil

Abstract

The invention provides a transformer insulating oil solid impurity content rapid evaluation device, and belongs to the technical field of transformer insulating oil detection, the transformer insulating oil solid impurity content rapid evaluation device comprises a cup body, a cover and a controller, the cup body and the cup cover are made of glass, the cup body and the cup cover are both provided with a hollow interlayer, and the hollow interlayer is filled with a light absorption material; an oil inlet pipe and an oil outlet pipe are oppositely arranged on the cup body; the cup body is connected into an external loop of transformer insulating oil through an oil inlet pipe and an oil outlet pipe, a stirring shaft is further arranged on the central axis of the cup body, and the outer diameter of the stirring shaft is smaller than 1 / 4 of the inner diameter of the cup body; a shielding plate is further arranged on the side wall of the cup body, a first laser transmitter and a second laser transmitter are arranged on the two sides of the shielding plate respectively, and a first laser receiver and a second laser receiver are arranged on the two intersection points respectively. And a laser emission control module and an insulating oil solid impurity content rapid evaluation module are arranged in the controller. The technical problem that a conventional detection method cannot realize automatic and continuous monitoring is solved.

Description

Technical Field

[0001] The invention belongs to the technical field of transformer insulating oil detection, and in particular relates to a device for quickly evaluating the solid impurity content of transformer insulating oil. Background Art

[0002] Transformers are important and indispensable equipment in power systems. The state of their insulating oil directly affects the operating performance and service life of transformers. As the transformer ages, some solid impurity particles will gradually accumulate in the insulating oil. These impurities will be adsorbed on the surface of the core and windings, reducing the insulation strength and accelerating the insulation degradation of the transformer. Therefore, accurately evaluating the content of solid impurities in the insulating oil in the transformer is crucial to preventing insulation failures and extending the service life of the transformer.

[0003] Traditional methods for detecting impurities in insulating oil mainly include the following: taking oil samples for centrifugal sedimentation, filter filtration or solvent extraction, and then determining the impurity content by mass or volume method. This type of method requires taking samples back to the laboratory for analysis, which has a long cycle, low efficiency, and is easily affected by operators and environmental factors, making it difficult to achieve rapid on-site detection. Other methods such as infrared spectroscopy and UV-visible spectroscopy, although they can be tested on-site, require complex optical instruments and equipment, are cumbersome to operate, and cannot achieve automation and continuous monitoring. Summary of the invention

[0004] In view of this, the present invention provides a device for quickly evaluating the solid impurity content of transformer insulating oil, which can solve the technical problem that conventional detection methods cannot achieve automation and continuous monitoring.

[0005] The present invention is achieved in that:

[0006] The present invention provides a rapid evaluation device for the solid impurity content of transformer insulating oil, which comprises a cup body, a cup cover and a controller, wherein the cup body and the cup cover are both made of glass, and the cup body and the cup cover are both provided with a hollow interlayer, and the hollow interlayer is filled with a light-absorbing material; an oil inlet pipe and an oil outlet pipe are relatively provided on the cup body; the cup body is connected to an external circuit of the transformer insulating oil through the oil inlet pipe and the oil outlet pipe, and a stirring shaft with stirring blades is fixedly provided on the central axis of the cup body, and the outer diameter of the stirring shaft is less than 1 / 4 of the inner diameter of the cup body; a shielding plate is also provided on the side wall of the cup body, and the extension line of the shielding plate passes through the central axis, and the closest distance between the shielding plate and the stirring shaft is less than M times the inner diameter of the cup body; a first laser emitter and a second laser emitter are respectively provided on both sides of the shielding plate, and the two laser emitters The laser emitted by the transmitter is parallel to the baffle plate and intersects with the inner wall of the cup body, and a first laser receiver and a second laser receiver are respectively arranged at the two intersection points; the first laser transmitter and the second laser transmitter can both emit lasers of multiple different wavelengths, and the controller is electrically connected to the first laser transmitter, the second laser transmitter, the first laser receiver, and the second laser receiver. A laser emission control module and a rapid assessment module for solid impurity content in insulating oil are arranged in the controller, and the laser emission control module is used to control the first laser transmitter and the second laser transmitter to set the optimal wavelength for laser emission, and the rapid assessment module for solid impurity content in insulating oil is used to process the first signal and the second signal received by the first laser receiver and the second laser receiver to obtain a rapid assessment result of the solid impurity content in the insulating oil in the transformer.

[0007] Wherein, the laser emission control module is used to perform the following steps:

[0008] S11, traversing the emission laser wavelength and brightness of the first laser emitter to perform laser emission, and simultaneously acquiring in real time a temporary brightness signal generated by the first laser emitter receiving the laser;

[0009] S12, preprocessing the temporary brightness signal to obtain a preprocessed brightness signal;

[0010] S13, calculating the amplitude fluctuation of the preprocessed brightness signal, and selecting the emission laser wavelength and brightness with the smallest amplitude fluctuation as the preset emission laser wavelength and brightness of the first laser emitter;

[0011] S14. Using the method of S11 to S13, determine the preset emission laser wavelength and brightness of the second laser emitter.

[0012] Furthermore, the insulating oil solid impurity content rapid assessment module is used to perform the following steps:

[0013] S21, acquiring in real time a first brightness signal and a second brightness signal collected by a first laser receiver and a second laser receiver;

[0014] S22, performing Fourier transform on the first brightness signal and the second brightness signal after preprocessing to obtain a first frequency domain feature and a second frequency domain feature respectively;

[0015] S23, inputting the first frequency domain feature and the second frequency domain feature into a pre-trained insulating oil solid impurity concentration assessment model respectively, to obtain a first solid impurity concentration corresponding to the first frequency domain feature and a second solid impurity concentration corresponding to the second frequency domain feature;

[0016] S24, calculating the solid impurity concentration of the insulating oil in the cup body based on the size of the cup body and the first solid impurity concentration and the second solid impurity concentration;

[0017] S25. Generate an impurity concentration curve based on the continuously acquired solid impurity concentration of the insulating oil in the cup body, and calculate the solid impurity concentration of the insulating oil in the transformer by combining dynamic system recognition and convolutional neural network as a rapid evaluation result of the solid impurity content of the transformer insulating oil.

[0018] The value of M is in the range of 2 to 5.

[0019] When the transformer insulating oil solid impurity content rapid assessment device is used, the stirring shaft rotates at a preset fixed speed, and the rotation direction is from the first laser emitter to the second laser emitter. The calculation equation of the optimal speed is specifically expressed as follows:

[0020]

[0021] In the formula, ω opt is the optimal speed (rad / s); g is the acceleration due to gravity (m / s 2 ); R is the cup radius (m); ρ oil is the density of insulating oil (kg / m 3 );ρ impurity is the average density of solid impurities (kg / m 3 ); i is an imaginary unit; η is the dynamic viscosity of the insulating oil (Pa·s); α is the thermal diffusion coefficient of the insulating oil (m 2 / s); T is the insulating oil temperature (K).

[0022] Parameter acquisition method: g is 9.8m / s 2 ; R is obtained by measuring the cup size; ρ oil and η are obtained by testing insulating oil samples; ρ impurityThe average value is obtained through multiple sampling tests; α is determined by a thermal property tester; and T is measured in real time by a temperature sensor.

[0023] Furthermore, the cup body is cylindrical; and the cup cover is a circular cover matching the cup body.

[0024] The light-absorbing material is specifically a composite material of graphene oxide (GO) and polydimethylsiloxane (PDMS). The material has excellent light absorption performance and thermal stability, and can effectively reduce external light interference.

[0025] Furthermore, the height of the oil inlet pipe is higher than the height of the oil outlet pipe, which brings the effect of utilizing gravity to form a natural flow, enhancing oil circulation, improving the uniformity of impurity distribution, and reducing the power consumption of the pump.

[0026] Furthermore, the shielding plate is also provided with a plurality of hemispherical concave portions, which have the effect of increasing fluid disturbance, generating microscopic turbulence, and improving the uniformity of impurity suspension.

[0027] Furthermore, the arrangement of the multiple hemispherical concave portions is equidistant from each other. The optimal diameter calculation equation of the circle formed by the hemispherical concave portion and the surface of the shielding plate is specifically expressed as follows:

[0028]

[0029] Where, D opt is the optimal diameter (m); ν is the kinematic viscosity of the insulating oil (m 2 / s); ω is the angular velocity of the stirring shaft (rad / s); R is the radius of the cup (m); ρ oil is the density of insulating oil (kg / m 3 );η is the dynamic viscosity of the insulating oil (Pa·s); k is the thermal conductivity of the insulating oil (W / (m·K)); c p is the specific heat capacity of insulating oil (J / (kg·K)); is the radial temperature gradient (K / m); We is the Weber number; Fr is the Froude number.

[0030] Parameter acquisition method:

[0031] ν,ρ oil ,η,k,c p Obtained through insulating oil sample testing; ω is controlled by the stirring motor; R is obtained by measuring the cup size; Calculated by multi-point temperature sensor measurement; We and Fr are calculated by fluid mechanics parameters.

[0032] The specific calculation equation of the method combining dynamic system identification and convolutional neural network is as follows:

[0033]

[0034] In the formula, C transformer is the solid impurity concentration of the insulating oil in the transformer (mg / L); CNN is the trained convolutional neural network function; K(t) is the system impulse response function; C cup (t) is the function of the change of the solid impurity concentration of the insulating oil in the cup body over time; t is the total measurement time (s); a n is the Fourier series coefficient; T is the transformer oil cycle (s); N is the harmonic order considered.

[0035] Parameter acquisition method: f CNN It is obtained through training with a large amount of historical data; K(t) is determined by the system identification method; C cup (t) is continuously obtained from step S40; t is the total time recorded by the system timer; a n and T are obtained by Fourier analysis of historical data; N is usually 3-5, and the specific value is determined according to the actual system complexity.

[0036] The insulating oil solid impurity concentration assessment model adopts a multi-layer perceptron structure.

[0037] The details are as follows:

[0038] Input layer: receives frequency domain features after Fourier transformation, with dimension [batch_size, feature_dim];

[0039] Hidden layer 1: 256 neurons, using ReLU activation function;

[0040] Hidden layer 2: 128 neurons, using ReLU activation function;

[0041] Hidden layer 3: 64 neurons, using ReLU activation function;

[0042] Output layer: 1 neuron, using linear activation function;

[0043] The mathematical expression of the model is as follows:

[0044] C=f(X)=W3·ReLU(W2·ReLU(W1·ReLU(W0·X+b0)+b1)+b2)+b3;

[0045] in:

[0046] X: frequency domain feature vector of input;

[0047] W i : The weight matrix of the i-th layer;

[0048] bi : Bias vector of the i-th layer;

[0049] ReLU(x)=max(0,x): ReLU activation function;

[0050] C: output solid impurity concentration;

[0051] 2. Training Dataset:

[0052] The training dataset contains the following:

[0053] Input: Frequency domain characteristics of insulating oil sample (obtained by Fourier transform);

[0054] Output: Corresponding solid impurity concentration (determined by standard laboratory methods);

[0055] Data source:

[0056] a) Laboratory control samples: insulating oil samples prepared with different types and concentrations of impurities;

[0057] b) On-site samples: insulating oil samples collected regularly from operating transformers;

[0058] c) Accelerated aging samples: insulating oil samples with different degrees of deterioration obtained through accelerated aging tests;

[0059] Data Augmentation:

[0060] a) Add Gaussian noise;

[0061] b) Randomly scale frequency domain features;

[0062] c) Feature combination;

[0063] Dataset size: at least 10,000 frequency domain feature-concentration pairs;

[0064] The data set is divided into: training set 70%, validation set 15%, and test set 15%.

[0065] The calculation equation of the solid impurity concentration of the insulating oil in the cup body is specifically expressed as follows:

[0066]

[0067] In the formula, C cup is the solid impurity concentration of the insulating oil in the cup (mg / L); C1 is the first solid impurity concentration (mg / L); C2 is the second solid impurity concentration (mg / L); V1 is the oil volume passed by the first laser (L); V2 is the oil volume passed by the second laser (L); V total is the total volume of the cup (L); is the partial derivative of impurity concentration with respect to temperature (mg / (L·K)); ΔT is the difference between the measured temperature and the standard temperature (K); t is the measurement time (s); τ is the system characteristic time constant (s); i is an imaginary unit; ω is the stirring angular velocity (rad / s); ω c is the critical angular velocity (rad / s).

[0068] Parameter acquisition method: C1 and C2 are obtained through step S30; V1, V2 and V total Calculated by cup size; Determined by experiments under multiple temperature conditions; ΔT is measured in real time by a temperature sensor; t is the time recorded by the system timer; τ is determined by system response experiments; ω is controlled by the stirring motor; ω c Determined through fluid dynamics analysis and experiments.

[0069] Optionally, the specific control method of the laser emission control module includes: traversing various wavelengths and brightnesses of the first laser emitter to perform laser emission, collecting brightness signals received by the first laser receiver in real time, preprocessing these brightness signals to remove noise interference, calculating the fluctuation amplitude of the preprocessed signal, and selecting the laser parameter with the smallest fluctuation amplitude as the preset emission parameter of the first laser emitter. The preset emission parameter of the second laser emitter is determined in the same way. In this way, it can be ensured that the lasers output by the two laser emitters are most conducive to the subsequent impurity content detection.

[0070] Optionally, the method of using the insulating oil solid impurity content rapid assessment module includes: real-time acquisition of brightness signals of the first laser receiver and the second laser receiver, preprocessing and Fourier transforming these signals to obtain corresponding frequency domain features. Then, the two frequency domain features are respectively input into two pre-trained impurity concentration assessment models to obtain the first solid impurity concentration corresponding to the first frequency domain feature and the second solid impurity concentration corresponding to the second frequency domain feature. Based on the geometric parameters of the cup body, the total solid impurity concentration of the insulating oil in the cup body is calculated.

[0071] Optionally, the insulating oil solid impurity content rapid assessment module also includes a method combining dynamic system identification and convolutional neural network to predict the solid impurity concentration of the insulating oil inside the transformer. Specifically, it includes: firstly determining the impulse response function of the change of the impurity concentration of the transformer oil through system identification, and then inputting the time series data of the impurity concentration in the cup body into the pre-trained convolutional neural network model to obtain the prediction result of the impurity concentration inside the transformer. At the same time, combined with the oil flow cycle characteristics obtained by Fourier analysis, the prediction accuracy is further improved.

[0072] Optionally, when determining the optimal laser parameters, the laser emission control module specifically performs the following steps: first, traverse various wavelength and brightness combinations of the first laser emitter, and record the corresponding temporary brightness signals; perform preprocessing such as time domain filtering and frequency domain filtering on these temporary brightness signals to remove noise interference; calculate the fluctuation amplitude of the preprocessed signal, and select the laser parameter with the smallest fluctuation amplitude as the preset emission parameter of the first laser emitter. The preset parameters of the second laser emitter are determined in the same way.

[0073] Optionally, when predicting the impurity concentration in the transformer, the insulating oil solid impurity content rapid assessment module specifically performs the following steps: first, the brightness signals of the first laser receiver and the second laser receiver are collected in real time, and these signals are preprocessed and Fourier transformed to obtain two frequency domain features; then, the two frequency domain features are respectively input into a pre-trained concentration assessment model to obtain a first solid impurity concentration corresponding to the first frequency domain feature, and a second solid impurity concentration corresponding to the second frequency domain feature; finally, based on the geometric parameters of the cup body, the total solid impurity concentration of the insulating oil in the cup body is calculated.

[0074] Optionally, the insulating oil solid impurity content rapid assessment module also uses a method combining dynamic system identification and convolutional neural network when predicting the impurity concentration in the transformer. Specifically, it includes: firstly determining the impulse response function of the change of the impurity concentration of the transformer oil through system identification; then inputting the time series data of the impurity concentration in the cup body into the pre-trained convolutional neural network model to obtain the prediction result of the impurity concentration inside the transformer; and at the same time combining the oil flow cycle characteristics obtained by Fourier analysis to further improve the prediction accuracy. This method makes full use of the time series characteristics of the impurity concentration in the cup body, and can more accurately predict the change of impurity concentration inside the transformer.

[0075] Optionally, when determining the optimal laser parameters, the laser emission control module first traverses various wavelength and brightness combinations of the first laser emitter, and records the temporary brightness signal received by the first laser receiver in real time. These temporary brightness signals are then preprocessed by time domain filtering and frequency domain filtering to remove interference such as high-frequency noise and zero-frequency drift, and obtain preprocessed brightness signals. Next, the fluctuation amplitudes of these preprocessed signals are calculated, and the laser parameters with the smallest fluctuation amplitude are selected as the preset emission parameters of the first laser emitter. The preset parameters of the second laser emitter are determined in the same way. This ensures that the lasers output by the two laser emitters are most conducive to subsequent impurity content detection.

[0076] Optionally, when predicting the impurity concentration in the transformer, the insulating oil solid impurity content rapid assessment module first collects the brightness signals of the first laser receiver and the second laser receiver in real time, preprocesses and Fourier transforms these signals, and obtains two frequency domain features. Then, the two frequency domain features are respectively input into two pre-trained impurity concentration assessment models to obtain the first solid impurity concentration corresponding to the first frequency domain feature and the second solid impurity concentration corresponding to the second frequency domain feature. Finally, based on the geometric parameters of the cup body, the total solid impurity concentration of the insulating oil in the cup body is calculated. This weighted average method can more accurately reflect the actual concentration level of impurities in the cup body.

[0077] Compared with the prior art, the beneficial effects of the device for rapidly evaluating the solid impurity content of transformer insulating oil provided by the present invention are:

[0078] First, the device uses dual-wavelength parallel laser detection technology, and uses the absorption and scattering characteristics of lasers in insulating oil to indirectly infer the concentration distribution of solid impurities in the oil. Compared with a single-wavelength laser, using two lasers with different wavelengths can enhance detection sensitivity and improve resolution. At the same time, an adaptive laser parameter optimization mechanism is designed, which can effectively eliminate external environmental interference by traversing different wavelength and intensity combinations and selecting the laser parameters that are most conducive to detection.

[0079] Secondly, the device integrates dynamic system identification and deep learning algorithms, which can accurately predict the concentration trend of solid impurities in the insulating oil inside the transformer. First, based on the time series characteristics of the impurity concentration in the cup, a dynamic model of the oil concentration change in the transformer is established through system identification; then, a convolutional neural network is used to further learn the complex relationship between the cup and the transformer, and the flow cycle characteristics obtained by Fourier analysis are integrated to greatly improve the prediction accuracy. This method can make full use of multi-source information and grasp key dynamic characteristics. It is more robust and reliable than traditional static analysis methods.

[0080] In addition, the device also uses carefully designed fluid dynamics mechanisms, such as optimal stirring speed and hemispherical flow disturbance, to ensure uniform distribution of impurities inside the cup, thus improving detection accuracy. At the same time, the innovative use of glass cups and light-absorbing composite materials effectively reduces external light interference and ensures repeatability and stability of detection.

[0081] In general, the rapid assessment device for solid impurity content in transformer insulating oil of the present invention greatly improves the speed, accuracy and automation level of insulating oil status assessment by integrating cutting-edge technologies such as optical detection, fluid mechanics analysis, and intelligent algorithms, and solves the technical problem that conventional detection methods cannot achieve automation and continuous monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 A schematic diagram of the three-dimensional structure of a device for rapidly evaluating the solid impurity content in transformer insulating oil provided by the present invention;

[0083] Figure 2 A schematic cross-sectional view of the device provided by the present invention in the horizontal direction;

[0084] Figure 3 is a schematic diagram of the shielding plate structure;

[0085] Figure 4 A schematic diagram of the steps executed by the laser emission control module;

[0086] Figure 5 A schematic diagram of the steps for the quick assessment module of solid impurity content in insulating oil;

[0087] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0088] 10. Cup body; 11. Cup cover; 12. Oil inlet pipe; 13. Oil pipe; 20. Stirring shaft; 211. First laser transmitter; 212. Second laser transmitter; 221. First laser receiver; 222. Second laser receiver; 23. Shielding plate; 231. Hemispherical concave. DETAILED DESCRIPTION

[0089] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0090] like Figure 1As shown, it is a schematic diagram of a rapid assessment device for solid impurity content in transformer insulating oil provided by the present invention, comprising a cup body 10, a cup cover 11 and a controller, the cup body and the cup cover are both made of glass, the cup body and the cup cover are both provided with a hollow interlayer, and the hollow interlayer is filled with a light-absorbing material; an oil inlet pipe 12 and an oil outlet pipe 13 are relatively provided on the cup body; the cup body is connected to the external circuit of the transformer insulating oil through the oil inlet pipe and the oil outlet pipe, and a stirring shaft 20 with stirring blades is also fixedly provided on the central axis of the cup body, and the outer diameter of the stirring shaft is less than 1 / 4 of the inner diameter of the cup body; a shielding plate is also provided on the side wall of the cup body, and the extension line of the shielding plate passes through the central axis, and the closest distance between the shielding plate and the stirring shaft is less than M times the inner diameter of the cup body; a first laser emitter 211 and a second laser emitter 212 are respectively provided on both sides of the shielding plate 23, and the two laser emitters The laser emitted by the emitter is parallel to the shielding plate and intersects with the inner wall of the cup body. The first laser receiver 221 and the second laser receiver 222 are respectively arranged at the two intersection points; the first laser emitter and the second laser emitter can emit lasers of multiple different wavelengths. The controller is electrically connected to the first laser emitter, the second laser emitter, the first laser receiver, and the second laser receiver. The controller is provided with a laser emission control module and a rapid evaluation module for solid impurity content of insulating oil. The laser emission control module is used to control the first laser emitter and the second laser emitter to set the optimal wavelength for laser emission. The rapid evaluation module for solid impurity content of insulating oil is used to process the first signal and the second signal received by the first laser receiver and the second laser receiver to obtain a rapid evaluation result of the solid impurity content of insulating oil in the transformer. The shielding plate is also provided with a plurality of hemispherical concave 23, which has the effect of increasing fluid disturbance, generating microscopic turbulence, and improving the uniformity of impurity suspension. The arrangement of the plurality of hemispherical concave is equidistant arrangement of the center.

[0091] Wherein, the laser emission control module is used to perform the following steps:

[0092] S11, traversing the emission laser wavelength and brightness of the first laser emitter to perform laser emission, and simultaneously acquiring in real time a temporary brightness signal generated by the first laser emitter receiving the laser;

[0093] S12, preprocessing the temporary brightness signal to obtain a preprocessed brightness signal;

[0094] S13, calculating the amplitude fluctuation of the preprocessed brightness signal, and selecting the emission laser wavelength and brightness with the smallest amplitude fluctuation as the preset emission laser wavelength and brightness of the first laser emitter;

[0095] S14. Using the method of S11 to S13, determine the preset emission laser wavelength and brightness of the second laser emitter.

[0096] Furthermore, the insulating oil solid impurity content rapid assessment module is used to perform the following steps:

[0097] S21, acquiring in real time a first brightness signal and a second brightness signal collected by a first laser receiver and a second laser receiver;

[0098] S22, performing Fourier transform on the first brightness signal and the second brightness signal after preprocessing to obtain a first frequency domain feature and a second frequency domain feature respectively;

[0099] S23, inputting the first frequency domain feature and the second frequency domain feature into a pre-trained insulating oil solid impurity concentration assessment model respectively, to obtain a first solid impurity concentration corresponding to the first frequency domain feature and a second solid impurity concentration corresponding to the second frequency domain feature;

[0100] S24, calculating the solid impurity concentration of the insulating oil in the cup body based on the size of the cup body and the first solid impurity concentration and the second solid impurity concentration;

[0101] S25. Generate an impurity concentration curve based on the continuously acquired solid impurity concentration of the insulating oil in the cup body, and calculate the solid impurity concentration of the insulating oil in the transformer by combining dynamic system recognition and convolutional neural network as a rapid evaluation result of the solid impurity content of the transformer insulating oil.

[0102] The specific implementation methods of the above steps are described in detail below:

[0103] In step S11, the laser emission control module first traverses the lasers of various wavelengths and brightnesses that can be output by the first laser transmitter, and records the brightness of the laser signal received by the first laser receiver in real time. These temporary brightness signals are then preprocessed, for example, by using time domain filtering, frequency domain filtering and other methods to remove noise, and obtain preprocessed brightness signals. Next, the fluctuation amplitudes of these preprocessed brightness signals are calculated, and the laser wavelength and brightness with the smallest fluctuation amplitude are selected as the preset parameters of the first laser transmitter. This ensures that the laser wavelength and brightness output by the first laser transmitter are most conducive to the subsequent impurity concentration detection.

[0104] In step S12, the specific method of preprocessing the temporary brightness signal includes:

[0105] 1) Time domain filtering: A Butterworth low-pass filter is used, and the cutoff frequency is set to ω c =2πf c , where f c The frequency band should be 3-5 Hz to remove high frequency noise.

[0106] 2) Frequency domain filtering: Perform fast Fourier transform (FFT) to obtain the frequency domain signal, and then use a bandpass filter to filter out 0-fc The frequency components outside the Hz range are finally processed by inverse FFT to obtain the preprocessed time domain signal.

[0107] 3) Normalization: Divide the preprocessed brightness signal by its peak value so that the signal amplitude is between 0 and 1, eliminating the influence of amplitude differences.

[0108] In step S13, the method for calculating the fluctuation amplitude of the preprocessed brightness signal is as follows:

[0109] 1) Calculate the standard deviation σ of the preprocessed signal, which reflects the degree of signal fluctuation.

[0110] 2) Divide σ by the signal average μ to obtain the fluctuation coefficient σ / μ, which is the fluctuation amplitude defined in this step.

[0111] 3) Select the set of laser parameters (wavelength and brightness) with the smallest fluctuation amplitude as the preset emission parameters of the first laser emitter. Generally speaking, laser parameters with a fluctuation amplitude of less than 5% can meet the requirements.

[0112] In step S14, the second laser emitter is also determined with the same method as steps S11-S13 to determine its optimal laser emission parameters, so as to ensure that the laser parameters output by the two laser emitters are the most conducive to subsequent impurity detection.

[0113] In step S21, the brightness signals collected by the first laser receiver and the second laser receiver are obtained in real time. The two brightness signals are respectively recorded as I1(t) and I2(t), where t represents the acquisition time. Since the laser is absorbed and scattered by impurities when passing through the insulating oil, the two brightness signals contain information about solid impurities in the oil.

[0114] In step S22, I1(t) and I2(t) are first preprocessed to remove interference such as high-frequency noise and zero-frequency drift, and the preprocessed brightness signal is obtained. and Then, the two preprocessed signals are subjected to fast Fourier transform (FFT) to obtain frequency domain features X1(f) and X2(f). The frequency domain features here contain the spectrum information of solid impurities in the oil products, providing an important basis for the subsequent estimation of impurity concentration.

[0115] In step S23, the two frequency domain features X1(f) and X2(f) obtained in the previous step are respectively input into two pre-trained insulating oil solid impurity concentration assessment models to obtain the first solid impurity concentration C1 corresponding to the first frequency domain feature and the second solid impurity concentration C2 corresponding to the second frequency domain feature. Both models adopt a multi-layer perceptron (MLP) structure, with frequency domain features as input and solid impurity concentration as output. The model training uses a large amount of data from laboratory samples and field samples, and optimizes the model parameters through the back propagation algorithm to minimize the error between the predicted results and the measured impurity concentration.

[0116] In step S24, the total solid impurity concentration C of the insulating oil in the cup body is calculated based on the geometric dimensions of the cup body and the first solid impurity concentration C1 and the second solid impurity concentration C2 obtained in step S23. cup The specific calculation formula is:

[0117]

[0118] Where V1 and V2 are the volumes of oil that the first laser beam and the second laser beam pass through, respectively. total is the total volume of the cup body. This weighted average method can more accurately reflect the actual concentration level of impurities in the cup body.

[0119] In step S25, based on the continuously acquired solid impurity concentration C of the insulating oil in the cup body, cup (t), a method combining dynamic system identification and convolutional neural network is used to predict the solid impurity concentration C of the insulating oil in the transformer. transformer The specific process is as follows:

[0120] 1) Dynamic system identification: Assuming that the change of oil impurity concentration inside the transformer can be described by a linear time-invariant system, the impulse response function K(t) of the system can be determined by the system identification method. The least squares method is used to fit the parameters of K(t).

[0121] 2) Convolutional neural network: the impurity concentration C in the cup cup (t) is used as input and passes through the convolutional neural network function f CNN Mapping to obtain the impurity concentration C in the transformer transformer The CNN network includes multiple convolutional layers, pooling layers, and fully connected layers, which can effectively learn the complex nonlinear relationship between the cup body and the transformer. The network is trained using a large amount of historical monitoring data and the network parameters are optimized through the back propagation algorithm.

[0122] 3) Fourier analysis: C cup (t) Perform Fourier analysis to obtain the coefficient a of its main periodic component nand period T. These period information reflects the flow characteristics inside the cup body and helps to further improve the C transformer prediction accuracy.

[0123] Combining the above three steps, the final C transformer The calculation formula is:

[0124]

[0125] Among them, f CNN is the trained convolutional neural network function. This method makes full use of the time series characteristics of the impurity concentration in the cup body and can accurately predict the changes in impurity concentration inside the transformer by combining dynamic system modeling and deep learning.

[0126] The specific implementation of steps S11-S14 and S21-S25 can be explained in detail using the following formula and algorithm:

[0127] The first is the laser emission control module in step S11. The goal of this module is to determine the optimal laser emission parameters, including wavelength λ and brightness I. The specific implementation is as follows:

[0128] 1) Iterate through the optional wavelength and brightness combinations of the first laser emitter:

[0129] λ 1,i ∈[λ 1,min ,λ 1,max ];

[0130] I 1,i ∈[I 1,min ,I 1,max ];

[0131] Among them, 1,min and λ 1,max are the minimum and maximum emission wavelengths of the first laser emitter, I 1,min and I 1max are their minimum and maximum emission brightness respectively. For each group (λ 1i ,I 1i ), will record the brightness signal collected by the first laser receiver in real time

[0132] 2) Yes Preprocess and remove noise interference:

[0133] Time domain filtering: Butterworth low-pass filter is used, with a cutoff frequency of ω c =2πf c , where f c 3-5Hz.

[0134] Frequency domain filtering: Perform fast Fourier transform (FFT) to obtain frequency domain signal Then use a bandpass filter to remove 0-f c The frequency components outside the Hz range are finally obtained by inverse FFT to obtain the preprocessed time domain signal

[0135] Normalization: Divide by its peak value to make the signal amplitude between 0 and 1.

[0136] 3) Calculate the preprocessed signal The fluctuation amplitude σ 1,i / μ 1,i , where σ 1,i is the standard deviation, μ 1,i is the average value.

[0137] 4) Select the fluctuation amplitude σ 1,i / μ 1,i The smallest group (λ 1,i ,I 1,i ) as the preset emission parameters of the first laser emitter:

[0138]

[0139] The same method can also determine the optimal emission parameters of the second laser emitter. This ensures that the laser parameters output by the two laser transmitters are most conducive to subsequent impurity detection.

[0140] Next is the signal acquisition and preprocessing in steps S21-S22. The specific process is as follows:

[0141] 1) Real-time acquisition of brightness signals of the first laser receiver and the second laser receiver:

[0142]

[0143] The two brightness signals I1(t) and I2(t) reflect the absorption and scattering characteristics of solid impurities in the insulating oil.

[0144] 2) Preprocess I1(t) and I2(t) to remove noise and drift:

[0145] Time domain filtering: The same Butterworth low-pass filter as in step S11, with a cutoff frequency ω c =2πf c .

[0146] Get the preprocessed signal:

[0147]

[0148] in, represents fast Fourier transform, H LPF (f) is the frequency response function of the low-pass filter.

[0149] 3) Preprocessed signal and Perform Fourier transform to obtain frequency domain features:

[0150]

[0151] These two frequency domain features contain the spectrum information of solid impurities in oil products, providing a basis for subsequent impurity concentration estimation.

[0152] In step S23, the above two frequency domain features X1(f) and X2(f) are respectively input into two pre-trained impurity concentration evaluation models to obtain the first solid impurity concentration C1 corresponding to the first frequency domain feature and the second solid impurity concentration C2 corresponding to the second frequency domain feature. Both models adopt a multi-layer perceptron (MLP) structure, and the mathematical expression is:

[0153]

[0154] in, and are the weight matrix and bias vector of the i-th layer (j=1 corresponds to model 1, j=2 corresponds to model 2), and ReLU(x)=max(0,x) is the ReLU activation function. The two models are trained with a large number of laboratory samples and field samples to minimize the error between the predicted results and the measured impurity concentration.

[0155] In step S24, according to the geometric dimensions V1, V2 and V total , and the first solid impurity concentration C1 and the second solid impurity concentration C2 obtained in step S23, calculate the total solid impurity concentration C of the insulating oil in the cup body cup :

[0156]

[0157] This weighted average method can more accurately reflect the actual concentration level of impurities in the cup.

[0158] Finally, in step S25, a method combining dynamic system identification and convolutional neural network is used to predict the solid impurity concentration C of the insulating oil inside the transformer. transformer The specific process is as follows:

[0159] 1) Dynamic system identification: Assuming that the change of oil impurity concentration inside the transformer can be described by a linear time-invariant system, the impulse response function K(t) of the system can be determined by the system identification method. The least squares method is used to fit the parameters of K(t).

[0160] 2) Convolutional neural network: the impurity concentration C in the cup cup (t) is used as input and passes through the convolutional neural network function f CNN Mapping to obtain the impurity concentration C in the transformer transformer The mathematical expression of the CNN network is:

[0161]

[0162] Among them, f CNN is the trained convolutional neural network function, a n and T are the Fourier series coefficient and transformer oil cycle period, respectively, which are obtained by Fourier analysis of historical data.

[0163] 3) Fourier analysis: C cup (t) Perform Fourier analysis to obtain the coefficient a of its main periodic component n and period T. These period information reflects the flow characteristics inside the cup body and helps to further improve the C transformer prediction accuracy.

[0164] In general, the rapid assessment device for the solid impurity content of transformer insulating oil integrates optical detection, fluid mechanics analysis, signal processing and artificial intelligence, and integrates a variety of cutting-edge technologies to provide a reliable and efficient solution for online monitoring and early warning of the status of transformer insulating oil.

[0165] In summary, the device for rapidly evaluating the solid impurity content of the transformer insulating oil of the present invention can rapidly and accurately evaluate the solid impurity content of the insulating oil inside the transformer through a carefully designed fluid dynamics and optical detection mechanism combined with advanced signal processing algorithms and artificial intelligence models. The key technologies of the device include:

[0166] 1) Optimal stirring design: According to the physical and chemical properties of the insulating oil, the optimal stirring speed in the cup is calculated to ensure that the impurities are fully suspended and dispersed, thereby improving the detection accuracy.

[0167] 2) Dual-wavelength laser detection: Using two parallel laser beams of different wavelengths, the concentration distribution of solid impurities can be indirectly calculated by measuring the absorption and scattering characteristics of the oil.

[0168] 3) Adaptive laser parameter optimization: By traversing different laser wavelengths and intensities, the most favorable parameter settings are selected to eliminate external environmental interference.

[0169] 4) Dynamic system modeling and machine learning: A method combining dynamic system identification and convolutional neural network is used to accurately predict the changing trend of impurity concentration inside the transformer.

[0170] 5) Comprehensive multi-source information fusion: Make full use of the time series characteristics of impurity concentration in the cup, the physical and chemical properties of oil products, and the system dynamics characteristics to improve the overall evaluation accuracy.

[0171] Specifically, the principle of the present invention is to use dual-wavelength laser detection technology combined with intelligent analysis algorithms to accurately evaluate the content of solid impurities in the insulating oil inside the transformer. Specifically:

[0172] First, the device uses two parallel laser beams of different wavelengths to irradiate the insulating oil sample in the cup. When the laser passes through the insulating oil, it will be absorbed and scattered by the impurity particles, causing the intensity of the received laser signal to change. Lasers of different wavelengths have different absorption and scattering effects on impurities, so the concentration distribution of solid impurities in the oil can be indirectly calculated by comparing the changes in the two laser signals.

[0173] In order to eliminate the influence of external environmental factors (such as temperature, humidity, etc.) on the detection results, the device also designs an adaptive laser parameter optimization mechanism. Specifically, it will traverse various optional wavelength and brightness combinations of the first laser transmitter, record the original signal collected by the first laser receiver in real time, and then pre-process these signals by time domain filtering and frequency domain filtering, and finally select the set of parameters with the smallest fluctuation amplitude as the optimal transmission parameters. The same method is also used to determine the optimal parameters of the second laser transmitter. This ensures that the laser parameters output by the two laser transmitters are the most conducive to impurity detection.

[0174] Next, the device will collect the brightness signals of the first laser receiver and the second laser receiver in real time, and obtain two frequency domain features after preprocessing and Fourier transform. Then, these two frequency domain features are respectively input into two pre-trained impurity concentration assessment models to obtain the first solid impurity concentration corresponding to the first frequency domain feature and the second solid impurity concentration corresponding to the second frequency domain feature. Based on the geometric parameters of the cup body, the total solid impurity concentration of the insulating oil in the cup body can be calculated.

[0175] In order to further improve the prediction accuracy of the impurity concentration inside the transformer, the device also adopts a method combining dynamic system identification and convolutional neural network. First, the dynamic characteristics of the change of impurity concentration of transformer oil are determined through system identification, and the mapping relationship between the impurity concentration in the cup and the impurity concentration in the transformer is established; then the convolutional neural network is used to conduct deep learning of this mapping relationship, and the oil flow cycle characteristics obtained by Fourier analysis are integrated, which greatly improves the prediction effect.

[0176] At the same time, the device has also made innovations in fluid dynamics design. For example, the optimal stirring speed in the cup is calculated to ensure that the impurity particles are fully suspended and dispersed; a hemispherical concave is set on the shielding plate to increase local fluid disturbance, promote micro-turbulence, and improve the uniformity of impurity distribution. These sophisticated fluid designs, combined with optical detection and intelligent algorithms, ultimately achieve a rapid and accurate assessment of the solid impurity content of the insulating oil inside the transformer.

[0177] In order to better understand and implement the present invention, an embodiment of a specific application scenario of the present invention is provided below: a certain substation regularly detects the insulating oil status of a 110kV transformer in operation to evaluate the operating status and remaining service life of the transformer. The substation adopts the transformer insulating oil solid impurity content rapid evaluation device proposed by the present invention for online monitoring.

[0178] The specific structure of the rapid assessment device for the solid impurity content of the transformer insulating oil is as follows: the cup body and the cup cover are both made of glass material, and a hollow interlayer is arranged inside, and the interlayer is filled with a composite material of graphene oxide (GO) and polydimethylsiloxane (PDMS), which has excellent light absorption performance. The cup body is provided with an oil inlet pipe and an oil outlet pipe, which are connected to the insulating oil circuit outside the transformer through the pipeline. A stirring shaft is fixed on the central axis of the cup body, and the diameter of the stirring shaft is about 1 / 4 of the inner diameter of the cup body, and a stirring blade is installed on the stirring shaft. A shielding plate is provided on the side wall of the cup body, and the extension line of the shielding plate passes through the central axis of the cup body, and the closest distance between the shielding plate and the stirring shaft is less than 1 / 3 of the inner diameter of the cup body. Two laser transmitters and two laser receivers are respectively arranged on both sides of the shielding plate, and the two laser beams are parallel to the shielding plate and intersect with the inner wall of the cup body. The entire device is integrated and connected through a controller, and the controller is provided with a laser emission control module and a rapid assessment module for the solid impurity content of the insulating oil.

[0179] The following is an example of a specific insulating oil status test to introduce the working process of the device:

[0180] 1. Start the device, and the stirring shaft will rotate at the preset optimal speed ω opt =12.6rad / s. The optimal speed is obtained by considering the insulating oil density ρ oil =890kg / m 3 , impurity average density ρ impurity =1500kg / m 3 , dynamic viscosity η = 0.12 Pa·s, thermal diffusion coefficient α = 6.8×10 -8 m 2 / s and cup radius R = 0.15m. This speed can ensure that the insulating oil and solid impurities in the cup are fully mixed, improving the uniformity and repeatability of the detection.

[0181] 2. The laser emission control module starts to traverse the emission wavelength λ1 and brightness I1 of the first laser emitter, with the ranges of λ1∈[600,800]nm and I1∈[50,200]mW respectively. For each set of (λ1,I1) parameters, the original brightness signal collected by the first laser receiver is recorded in real time.

[0182] 3. For these original brightness signals Preprocessing:

[0183] (1) Time domain filtering: A 4th-order Butterworth low-pass filter is used with a cutoff frequency of f c =5Hz to remove high frequency noise.

[0184] (2) Frequency domain filtering: Perform fast Fourier transform (FFT) on the filtered signal, then use a bandpass filter to remove frequency components outside the 0-5 Hz range, and finally transform it back to the time domain to obtain

[0185] (3) Normalization: Divide by its peak value to make the signal amplitude between 0 and 1.

[0186] 4. Calculate the preprocessed signal The fluctuation amplitude is σ1 / μ1, where σ1 is the standard deviation and μ1 is the average value. The set of (λ1, I1) parameters with the smallest fluctuation amplitude is selected as the optimal emission parameters of the first laser emitter:

[0187]

[0188] 5. Using the same method as steps 2-4, determine the optimal emission parameters of the second laser emitter as

[0189] 6. Collect the brightness signals I1(t) and I2(t) of the first laser receiver and the second laser receiver in real time, perform time domain filtering and frequency domain filtering on these two signals respectively, and obtain the preprocessed signals and

[0190] 7. Yes and Perform fast Fourier transform (FFT) to obtain two frequency domain features X1(f) and X2(f).

[0191] 8. Input X1(f) and X2(f) into two pre-trained impurity concentration assessment models, respectively, and obtain the first solid impurity concentration C1=12.3mg / L corresponding to the first frequency domain feature, and the second solid impurity concentration C2=11.9mg / L corresponding to the second frequency domain feature. Both models use a multi-layer perceptron (MLP) structure, with frequency domain features as input and solid impurity concentration as output.

[0192] 9. According to the geometric parameters of the cup, the oil volume V1 = 0.8L passed by the first laser, the oil volume V2 = 0.7L passed by the second laser, and the total volume V of the cup total =1.5L, the total solid impurity concentration of the insulating oil in the cup is calculated as:

[0193]

[0194] 10. Next, the device uses a combination of dynamic system identification and convolutional neural network to predict the solid impurity concentration C of the insulating oil inside the transformer. transformer .

[0195] First, the impulse response function K(t) of the change in oil impurity concentration inside the transformer is determined through system identification. According to the analysis of historical monitoring data, K(t) can be fitted by the following formula:

[0196] K(t)=0.5·e -0.02t sin(0.1πt);

[0197] Then, the impurity concentration C in the cup cup (t) is used as input and input into the pre-trained convolutional neural network to obtain the internal impurity concentration C of the transformer. transformer The prediction result. The CNN network contains 1D convolution layer, maximum pooling layer and full connection layer, and the mathematical expression is:

[0198]

[0199] By measuring the impurity concentration C in the cup cup The Fourier analysis of (t) obtained the coefficients of the main periodic components a1 = 0.2 and a2 = 0.1, and the period T = 20s. These periodic characteristics reflect the flow law inside the cup body and help to further improve the C transformer prediction accuracy.

[0200] Combining the above steps, the final result of the rapid evaluation device for solid impurity content in transformer insulating oil is: the solid impurity concentration of the transformer insulating oil is C transformer =14.2mg / L.

[0201] The following is a comparison of the performance differences between the device of the present invention and the traditional detection method in combination with experimental data:

[0202] The traditional laboratory testing method, i.e. taking oil samples for centrifugal separation and filtration, measured the total impurity content of the transformer's insulating oil to be 15.6 mg / L. It can be seen that the result given by the device of the present invention differs from the laboratory test result by less than 10%, which is sufficient to meet the needs of daily inspection of substations.

[0203] In addition, from the perspective of detection time, the traditional method requires the oil sample to be brought back to the laboratory, and the entire detection process takes 1-2 days. However, the device of the present invention can quickly complete the detection within 15 minutes on site, greatly shortening the detection cycle and greatly improving work efficiency.

[0204] In addition, the device of the present invention also has the advantages of automation and intelligence. It can automatically complete the entire process of laser parameter optimization, signal acquisition, data analysis, etc. through the controller without manual intervention. At the same time, the dynamic system identification and convolutional neural network model built into the device can accurately predict the changing trend of impurity concentration inside the transformer, providing reliable technical support for substation equipment status monitoring and preventive maintenance.

[0205] In summary, the rapid assessment device for solid impurity content in transformer insulating oil integrates cutting-edge technologies such as optical detection and intelligent algorithms. Compared with traditional methods, it significantly improves the detection speed, accuracy and automation level, provides an innovative solution for transformer status assessment and fault prevention, and has broad application prospects.

[0206] Table 1 below gives some key parameter data of the device in practical application:

[0207] Table 1 Key parameters list

[0208]

[0209] From the continuously collected data, it can be seen that within the first 20 seconds, the impurity concentration C in the cup cup It is relatively stable, maintaining at around 12.1 mg / L. The impurity concentration inside the transformer is C transformer It showed a trend of first rising rapidly and then tending to be stable over time, and finally stabilized at 14.2 mg / L.

[0210] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A rapid assessment device for solid impurity content in transformer insulating oil, characterized in that: It includes a cup body, a cup cover and a controller, wherein the cup body and the cup cover are both made of glass, and the cup body and the cup cover are both provided with a hollow interlayer, and the hollow interlayer is filled with a light-absorbing material; an oil inlet pipe and an oil outlet pipe are relatively provided on the cup body; the cup body is connected to the external circuit of the transformer insulating oil through the oil inlet pipe and the oil outlet pipe, and a stirring shaft with stirring blades is fixedly provided on the central axis of the cup body, and the outer diameter of the stirring shaft is less than 1 / 4 of the inner diameter of the cup body; a shielding plate is also provided on the side wall of the cup body, and the extension line of the shielding plate passes through the central axis, and the closest distance between the shielding plate and the stirring shaft is less than M times the inner diameter of the cup body; a first laser emitter and a second laser emitter are respectively provided on both sides of the shielding plate, and the lasers emitted by the two laser emitters are parallel to the shielding plate And it intersects with the inner wall of the cup body, and a first laser receiver and a second laser receiver are respectively arranged at the two intersection points; the first laser transmitter and the second laser transmitter can both emit lasers of multiple different wavelengths, and the controller is electrically connected to the first laser transmitter, the second laser transmitter, the first laser receiver, and the second laser receiver. A laser emission control module and a rapid assessment module for solid impurity content in insulating oil are arranged in the controller, and the laser emission control module is used to control the first laser transmitter and the second laser transmitter to set the optimal wavelength for laser emission, and the rapid assessment module for solid impurity content in insulating oil is used to process the first signal and the second signal received by the first laser receiver and the second laser receiver to obtain a rapid assessment result of the solid impurity content in the insulating oil in the transformer.

2. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 1, characterized in that: The laser emission control module is used to perform the following steps: S11, traversing the emission laser wavelength and brightness of the first laser emitter to perform laser emission, and simultaneously acquiring in real time a temporary brightness signal generated by the first laser emitter receiving the laser; S12, preprocessing the temporary brightness signal to obtain a preprocessed brightness signal; S13, calculating the amplitude fluctuation of the preprocessed brightness signal, and selecting the emission laser wavelength and brightness with the smallest amplitude fluctuation as the preset emission laser wavelength and brightness of the first laser emitter; S14. Using the method of S11 to S13, determine the preset emission laser wavelength and brightness of the second laser emitter.

3. A rapid evaluation device for solid impurity content in transformer insulating oil according to claim 2, characterized in that: The insulating oil solid impurity content rapid assessment module is used to perform the following steps: S21, acquiring in real time a first brightness signal and a second brightness signal collected by a first laser receiver and a second laser receiver; S22, performing Fourier transform on the first brightness signal and the second brightness signal after preprocessing to obtain a first frequency domain feature and a second frequency domain feature respectively; S23, inputting the first frequency domain feature and the second frequency domain feature into a pre-trained insulating oil solid impurity concentration assessment model respectively, to obtain a first solid impurity concentration corresponding to the first frequency domain feature and a second solid impurity concentration corresponding to the second frequency domain feature; S24, calculating the solid impurity concentration of the insulating oil in the cup body based on the size of the cup body and the first solid impurity concentration and the second solid impurity concentration; S25. Generate an impurity concentration curve based on the continuously acquired solid impurity concentration of the insulating oil in the cup body, and calculate the solid impurity concentration of the insulating oil in the transformer by combining dynamic system recognition and convolutional neural network as a rapid evaluation result of the solid impurity content of the transformer insulating oil.

4. A rapid evaluation device for solid impurity content in transformer insulating oil according to claim 1, characterized in that: The value range of M is 2-5.

5. The device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 1, characterized in that: When the device for rapidly evaluating the solid impurity content in transformer insulating oil is used, the stirring shaft rotates at a preset fixed speed, and the direction of rotation is from the first laser emitter to the second laser emitter.

6. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 5, characterized in that: The cup body is cylindrical; the cup cover is a circular cover matching the cup body.

7. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 6, characterized in that: The height of the oil inlet pipe is higher than that of the oil outlet pipe, which has the effect of utilizing gravity to form a natural flow, enhancing oil circulation, improving the uniformity of impurity distribution, and reducing the power consumption of the pump.

8. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 7, characterized in that: The shielding plate is also provided with a plurality of hemispherical concave portions, which have the effect of increasing fluid disturbance, generating microscopic turbulence, and improving the uniformity of impurity suspension.

9. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 8, characterized in that: The arrangement of the multiple hemispherical concavities is equidistant between the centers.

10. A device for rapidly evaluating the solid impurity content of transformer insulating oil according to claim 1, characterized in that: The insulating oil solid impurity concentration assessment model adopts a multi-layer perceptron structure.

Images