A microstructure gradient electric field-based insulating oil carbon particle intelligent detection method and device

Abstract

A kind of microstructure gradient electric field-based insulating oil carbon particle intelligent detection method and device, characterized in that, the method includes the following steps: electrode array is installed in the oil tank of transformer by special support, ensure that electrode is in full contact with insulating oil, connect signal processing unit, display alarm module and remote monitoring module;Through touch screen, set monitoring period, alarm threshold and alarm level, configure the network parameters of remote monitoring module, ensure data upload;High-frequency alternating electric field is applied to electrode system and real-time acquisition electric field signal, uses deep learning algorithm to analyze electric field signal, identifies the type of carbon particle and calculates its concentration;When the concentration of carbon particle detected exceeds the set threshold, the system will trigger audible light alarm and send SMS notification.

Description

Technical Field

[0001] This invention relates to the field of electrical engineering, and more specifically, to a method and apparatus for intelligent detection of carbon particles in insulating oil based on a microstructure gradient electric field. Background Technology

[0002] With the rapid development of power systems, oil-immersed transformers, as important power equipment, are widely used in the conversion and transmission of electrical energy. They employ special insulating oil as the insulating medium to ensure safe operation. However, the presence of carbon particles in the insulating oil can negatively impact the transformer's insulation performance, thereby affecting the equipment's reliability and safety.

[0003] Traditional detection methods primarily rely on physical or chemical analysis. These methods typically suffer from insufficient sensitivity and complex operation, failing to meet the real-time monitoring requirements for minute carbon particles. For example, common existing detection methods involve analyzing insulating oil using optical or chemical methods. While these methods can detect carbon particles, their sensitivity to small particles is relatively low, and the detection process is cumbersome, making online real-time monitoring impossible. Furthermore, traditional methods are susceptible to various factors during detection, such as environmental conditions like temperature and pressure, leading to instability in the monitoring results and consequently affecting the assessment of the transformer's safety status.

[0004] Currently, some new monitoring devices exist on the market, such as detection systems based on image recognition or acoustic signals. However, these methods largely rely on complex algorithms and equipment, have not yet achieved widespread application, and still have shortcomings in terms of environmental adaptability and stability. Furthermore, existing technologies lack comprehensive and systematic monitoring of carbon particles in insulating oil and fail to fully consider the response characteristics of electric fields to carbon particles, leading to insufficient assessment of potential risks in transformers.

[0005] Therefore, there is an urgent need for a new detection method that can efficiently and accurately monitor the presence and concentration of carbon particles in insulating oil based on the electric field disturbance method, thereby improving the safety and reliability of transformers, providing timely warnings of potential risks, and ensuring the stable operation of power equipment. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and device for intelligent detection of carbon particles in insulating oil based on a microstructure gradient electric field.

[0007] The present invention adopts the following technical solution.

[0008] The first aspect of this invention relates to an intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field. The method includes the following steps: installing an electrode array inside the transformer tank using a dedicated bracket to ensure full contact between the electrodes and the insulating oil; connecting a signal processing unit, a display alarm module, and a remote monitoring module; setting the monitoring cycle, alarm threshold, and alarm level via a touchscreen; configuring the network parameters of the remote monitoring module to ensure data upload; applying a high-frequency AC electric field to the electrode system and acquiring the electric field signal in real time; analyzing the electric field signal using a deep learning algorithm to identify the type of carbon particles and calculate their concentration; and triggering an audible and visual alarm and sending a text message notification when the detected carbon particle concentration exceeds a set threshold.

[0009] The electrode array is installed inside the transformer's oil tank using a dedicated bracket, ensuring full contact between the electrodes and the insulating oil. It connects to a signal processing unit, a display and alarm module, and a remote monitoring module. The array includes electrodes with a needle-like tip and a planar shape. The needle-like tip has a diameter of 50 μm, and the planar electrode has a size of 5 mm × 5 mm. The electrode spacing is adjustable. The electrode surface is coated with a multilayer graphene nanomaterial. The electrode array is fixed inside the transformer's oil tank using a high-strength insulating material bracket. The installation angle of the electrodes is adjustable, ranging from 0° to 30°.

[0010] Applying a high-frequency alternating electric field to the electrode system and acquiring the electric field signal in real time includes: the electric field is applied as a high-frequency alternating electric field with a frequency range of 1kHz to 50kHz and a voltage range of 10V to 100V.

[0011] A high-frequency alternating electric field is applied to the electrode system and the electric field signal is acquired in real time. The electric field signal is analyzed using a deep learning algorithm to identify the type of carbon particles and calculate their concentration. This includes: performing time-frequency transformation on the signal using the Chirp-Z algorithm; extracting and classifying features using a CLDNet fusion neural network model; and automatically identifying and classifying the carbon particle concentration, particle size, and morphology in the electric field signal.

[0012] A high-frequency alternating electric field is applied to the electrode system and the electric field signal is acquired in real time. The electric field signal is analyzed using a deep learning algorithm to identify the type of carbon particles and calculate their concentration. This includes automatically adjusting the parameters of the processing algorithm according to changes in the actual environment, such as temperature, humidity, and oil flow rate.

[0013] By using CNN for feature extraction and classification, the concentration, size and morphology of carbon particles in the electric field signal are automatically identified and classified. The output carbon particle information includes concentration, particle size distribution and morphology; the particle size distribution is from micrometer to nanometer level; the morphology includes spherical, plate-like or fibrous.

[0014] When the detected carbon particle concentration exceeds the set threshold, the system will trigger an audible and visual alarm and send an SMS notification, including: when the detected carbon particle concentration exceeds the set threshold, the system will remind the user through an audible and visual alarm.

[0015] The second aspect of this invention relates to an intelligent detection device for carbon particles in insulating oil based on a microstructure gradient electric field, implemented using the intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field described in the first aspect of this invention. The device includes an installation module, a parameter setting module, a data analysis and alarm module, and a data storage and remote monitoring module. The installation module is used to install the electrode array inside the transformer's oil tank using a dedicated bracket, ensuring full contact between the electrodes and the insulating oil, and connects to a signal processing unit, a display and alarm module, and a remote monitoring module. The parameter setting module is used to set the monitoring cycle, alarm threshold, and alarm level via a touchscreen, and configure the network parameters of the remote monitoring module to ensure data upload. The data analysis and alarm module is used to apply a high-frequency AC electric field to the electrode system and collect the electric field signal in real time, using a deep learning algorithm to analyze the electric field signal, identify the type of carbon particles, and calculate their concentration. The data storage and remote monitoring module is used to trigger an audible and visual alarm and send a text message notification when the detected carbon particle concentration exceeds a set threshold.

[0016] A third aspect of the present invention relates to a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to perform the steps of the method described in the first aspect of the present invention.

[0017] A fourth aspect of the present invention relates to a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in the first aspect of the present invention.

[0018] The beneficial effect of this invention is that, compared with the prior art, the intelligent detection method and device for insulating oil carbon particles based on microstructure gradient electric field in this invention achieves ultra-high sensitivity detection at the 0.01ppm level and real-time particle morphology identification through a triple breakthrough of field strength enhancement, frequency domain coding and intelligent analysis, completely solving the defects of traditional detection technology in terms of accuracy and early warning capability.

[0019] The beneficial effects of the present invention also include:

[0020] 1. The system sensitivity is improved to 0.01ppm (8 times higher than the traditional electrochemical method), the particle size detection limit is broken down to 0.1μm, and intelligent classification of carbon particle morphology is achieved (false judgment rate ≤3%); the time-frequency coding excitation strategy makes the separation degree of multi-particle size response spectrum ≥80%, which effectively solves the problem of small particle missed detection in the traditional single-frequency detection method.

[0021] 2. Gradient field strength combined with online self-calibration technology ensures an accuracy of ±0.5%FS even under 50m / s oil flow impact. The system exhibits excellent environmental adaptability, operating within a full temperature range of -40℃ to 120℃, and boasts an IP68 protection rating. Simultaneously, blind source separation technology effectively suppresses common-mode interference, improving the signal-to-noise ratio by ≥20dB, thereby guaranteeing detection stability under complex operating conditions.

[0022] 3. The visual interface improves fault location efficiency by 80%, and the trend prediction model reduces unplanned downtime by more than 30%; the system supports edge-cloud collaborative computing, and the incremental learning mechanism enables the model to continuously evolve (monthly error correction), further improving the intelligent operation and maintenance level of the equipment. Attached Figure Description

[0023] Figure 1 This is a system diagram of an intelligent detection device for carbon particles in insulating oil based on a microstructure gradient electric field.

[0024] Figure 2 This is a schematic diagram illustrating the working principle of an intelligent detection device for carbon particles in insulating oil based on a microstructure gradient electric field.

[0025] Figure 3 Here is a diagram of the electrode structure;

[0026] Figure 4 Here is a flowchart of the holographic signal analysis algorithm;

[0027] Figure 5 This is a schematic diagram of an adaptive signal processing algorithm.

[0028] Figure label:

[0029] 1-Insulating support; 2-Platinum metal electrode plate; 3-Impurity particles; 4-Wire; 5-Power supply; 6-Signal processing unit; 7-Signal transmission medium; 8-User terminal; 21-Interdigital electrode; 22-Nano coating; 221-Nano zinc oxide coating; 222-Graphene coating. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer and more accurate, the technical solutions of this invention are described in detail below through several specific embodiments. The embodiments used in this invention are merely illustrative and are not intended to limit the scope of this invention.

[0031] Addressing the shortcomings of existing technologies, such as insufficient field strength of traditional electrodes (<5×10⁻⁶), 4 The invention provides a method and device for intelligent detection of carbon particles in insulating oil based on a microstructure gradient electric field, which addresses issues such as missed detection of tiny carbon particles due to V / m, difficulty in capturing multidimensional features of particles due to single-frequency excitation, and measurement drift (>±3% FS) caused by offline calibration.

[0032] In a first aspect, this invention relates to an intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field, the method comprising the following steps:

[0033] Step 1: Install the electrode array inside the transformer's oil tank using a dedicated bracket, ensuring full contact between the electrodes and the insulating oil. Then connect the signal processing unit, display alarm module, and remote monitoring module.

[0034] This array employs asymmetric interdigitated electrodes fabricated using MEMS technology. The electrode linewidth ranges from 30 to 100 μm (adjustable to ±1 μm), and the spacing gradient ranges from 10 to 200 μm. A nano-zinc oxide / graphene heterostructure coating (coating thickness of 100-500 nm) is deposited on the electrode surface via magnetron sputtering, creating a localized electric field strength ≥1.8 × 10⁻⁶ in the insulating oil. 5 A gradient electric field of V / m. This electrode array is integrated into the transformer oil circuit through an insulating ceramic support, possessing resistance to oil flow impact and capable of withstanding oil flow rates up to 50m / s; at the same time, it has a built-in temperature-pressure dual compensation sensor with an accuracy of ±0.1% FS, which can effectively eliminate environmental interference and ensure stable operation of the system under complex working conditions.

[0035] The carbon particle detection device of the present invention mainly includes an electrode system, a signal processing unit, a display and alarm module, a data storage unit, and a remote monitoring module, and its working principle is as follows: Figure 2 As shown.

[0036] The electrode system of this invention is manufactured using MEMS (Micro-Electro-Mechanical Systems) technology, and utilizes an asymmetric interdigitated structure to enhance the sensitivity and responsiveness of the electrodes, such as... Figure 3 As shown. The electrode linewidth is 50 μm, and its spacing gradient can be adjusted within the range of 20 μm to 200 μm to meet the detection requirements of carbon particles of different sizes.

[0037] The asymmetric interdigitated electrode design, fabricated using MEMS technology, enables the formation of complex electric field distributions between the electrodes. The advantage of this asymmetric design lies in its ability to provide a stronger electric field gradient than symmetric electrodes under different operating conditions, which is crucial for the capture and detection of carbon particles. The electrode linewidth is 50 μm, and the electrode spacing is adjustable from 20 μm to 200 μm. This design allows the electrodes to adapt to carbon particles of different sizes, ensuring accurate detection of carbon particle concentration across various particle size ranges. By adjusting the spacing, the gradient of the electric field distribution also changes, further improving detection accuracy.

[0038] The electrode surface is coated with a heterogeneous coating of nano-zinc oxide and graphene (coating thickness 100–500 nm). Nano-zinc oxide possesses strong charge-trapping capabilities, effectively enhancing the electrode surface's ability to capture carbon particles, while graphene exhibits high conductivity and excellent mechanical strength, enhancing the electrode's durability and stability. This coating combines the high dielectric constant of nano-zinc oxide with the conductivity of graphene, enabling the electrode surface to more accurately sense changes in the electric field during monitoring.

[0039] Electrode integration method: These micro / nano electrodes are integrated into the transformer's insulating oil container using a dedicated insulating support. The electrode surface is in full contact with the insulating oil, enabling effective detection of the presence and concentration of carbon particles in the oil during the application of an electric field.

[0040] During implementation, users set the monitoring cycle through an augmented reality (AR) interface, ranging from 1 minute to 24 hours, and can dynamically adjust it according to the transformer's load rate and ambient temperature and humidity. The temperature and humidity range supports environmental variations from -40℃ to 120℃. The system's built-in operating condition matching algorithm analyzes oil flow velocity (1–50 m / s) and historical degradation trends to automatically recommend the most suitable monitoring frequency for the current operating conditions. For example, under high load or high oil flow velocity, the system will automatically increase the monitoring frequency to respond more promptly to changes in carbon particle concentration.

[0041] To improve detection sensitivity, this invention employs a time-frequency hybrid coding technique to apply a wideband electric field of 0.1–10 MHz. A 12th-order pseudo-random phase-modulated signal (phase jump ±180°) is generated based on the Chirp-Z algorithm and output to a micro / nano electrode array via a 250 MS / s high-speed DAC. The dwell time for each frequency band is 0.1 to 1 second and is adaptively adjustable to ensure system stability and responsiveness under various operating conditions.

[0042] Step 2: Set the monitoring cycle, alarm threshold and alarm level via the touch screen, configure the network parameters of the remote monitoring module, and ensure data upload.

[0043] The dynamic interference immunity system features online self-calibration, automatically performing zero-point and range calibration daily. Combined with Kalman filtering technology, it compensates for temperature drift, ensuring the system's error does not exceed ±0.5% FS across the entire temperature range of -40℃ to 120℃. Simultaneously, blind source separation technology is employed to suppress electromagnetic noise and common-mode interference, improving the signal-to-noise ratio by ≥20dB, ensuring stability and high accuracy in complex electromagnetic environments.

[0044] The system transmits data back via a Time-Sensitive Network (TSN), is compatible with the IEC 61850 protocol, and has a latency of ≤1ms. The storage unit constructs a spatiotemporal matrix of carbon particle growth, and uses an LSTM prediction model to analyze the 72-hour degradation trend with an accuracy of ≥92%. Daily automatic dual-frequency calibration (1kHz / 1MHz) and Kalman filter calibration ensure drift ≤±0.5% FS within 5 years.

[0045] Step 3: Apply a high-frequency AC electric field to the electrode system and collect the electric field signal in real time. Use a deep learning algorithm to analyze the electric field signal, identify the type of carbon particles, and calculate their concentration.

[0046] The time-frequency hybrid coding excitation module generates a 0.1-10MHz broadband swept frequency signal based on the Chirp-Z modulation algorithm and incorporates pseudo-random phase jumps (phase change ±180°, randomly distributed), which are output to the electrode array via a high-speed DAC (sampling rate 250MS / s). Each excitation cycle contains 12 sets of non-uniformly spaced frequency-hopping signals, which can synchronously excite the polarization migration and relaxation effects of carbon particles, thereby improving the separation of response signals for carbon particles of various sizes (0.1-10μm) to 80%. This design effectively improves the detection capability for particles of various sizes, especially showing significant advantages in the detection of microparticles (nanoscale).

[0047] The specific implementation method includes: generating a 12th-order pseudo-random phase modulation signal (phase jump ±180°) based on the Chirp-Z algorithm, and outputting it to the micro / nano electrode through a 250MS / s high-speed DAC; the dwell time of each frequency band is 0.1 to 1 second (adaptively adjustable), simultaneously exciting the polarization migration (1kHz to 1MHz) and relaxation effect (1-10MHz) of carbon particles, achieving full particle size coverage of 0.1-10μm; the gradient field strength on the electrode surface is ≥1.8×10 5 V / m, environmental interference is corrected in real time through a temperature-pressure dual compensation module (accuracy ±0.1% FS).

[0048] The multimode resonant signal analysis unit employs a CLDNet fusion neural network model for holographic signal analysis. The signal processing flow includes: feature extraction, using a 7-level adaptive convolution kernel to decompose the acquired ΔC / ΔQ (capacitance / charge distortion) signal into the fundamental frequency, third harmonic, and relaxation term; intelligent decision-making, analyzing time-frequency ridge features through a bidirectional LSTM network and outputting the following three-dimensional parameters: concentration range of 0.01-500 ppm, dynamic error ≤ ±0.5%; particle size distribution of 0.1-10 μm, classification accuracy of 0.05 μm; morphology classification as spherical, sheet-like, and fibrous, with a classification accuracy (F1 value) ≥ 97%.

[0049] A 7-level adaptive wavelet packet decomposition (Daubechies9 wavelet basis) is used to remove power frequency noise and reconstruct the fundamental frequency, 3rd / 5th harmonics, and relaxation cofactors. Then, time-series modeling is performed: time-frequency ridge features are extracted using a bidirectional LSTM network to construct a 12-dimensional dynamic tensor. Finally, dynamic decision-making is performed: outputting carbon particle concentration (0.01-500ppm, ±0.5% accuracy), particle size distribution (0.05μm resolution), and morphological classification (F1 score ≥97%). Furthermore, if the output confidence level is <95%, the system triggers an incremental learning mechanism to update the model parameters through resampling.

[0050] In the signal excitation unit, the system uses an FPGA as the control core, with a built-in Chirp-Z algorithm processing module. It achieves wideband (0.1MHz~10MHz) frequency domain sweeping by online generation and updating of a 12th-order pseudo-random phase sequence (phase jump ±180°). The FPGA sends the digital baseband signal to a 250MS / s, 14-bit resolution DAC, which is then output to the micro-nano electrode array after cascaded amplifiers. The amplification link includes two stages of RF power amplifiers with a bandwidth flatness of ±1dB and an output range of ±50V. An adjustable capacitor / inductor impedance matching network is configured to ensure optimal matching with the 50Ω electrode impedance. The initial dwell time for each frequency band is set to 0.5s, and can be dynamically adjusted within the range of 0.1s~1s according to oil temperature and preset sensitivity parameters through an "adaptive dwell time algorithm" to achieve a balance between high SNR and fast response.

[0051] In the feature denoising stage, the signal processing unit uses a 7-level adaptive wavelet packet decomposition technique to denoise the acquired electric field signal. Specifically, the signal x(t) is decomposed into a series of frequency band signals x i (t), where each frequency band corresponds to a specific scale. Wavelet packet transform is performed using the Daubechies9 wavelet basis, using the formula:

[0052]

[0053] Among them, h i Let represent the filter for the i-th wavelet packet, * denote the convolution operation, and N be the number of decomposition levels. In this way, the low-frequency and high-frequency noise components of the signal can be separated, thereby eliminating power frequency noise and environmental interference. The reconstructed signal retains the main components related to changes in carbon particle concentration, while the noise component is effectively removed.

[0054] Next, in the time-series modeling stage, the signal processing unit uses a bidirectional LSTM (Long Short-Term Memory) network to analyze the time-frequency ridge features of the signal. To capture long-term dependencies in the signal, the LSTM network uses the following state update formula:

[0055] f t=σ(W f ·[h t-1 ,x t ]+b f )

[0056] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0057] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0058] c t =f t *c t-1 +i t *tanh(W c ·[h t-1 ,x t ]+b c )

[0059] h t =o t *tanh(c t )

[0060] Where, x t It is the input signal at the current time step, h t-1 and c t-1 These are the output and memory state from the previous moment, respectively, W f W i W o W c It is the weight matrix in the network, b f ,b i ,b o ,b c σ is the bias term, sigmoid activation function, and tanh is hyperbolic tangent activation function. Through this process, the LSTM network can extract long-term time dependencies from the signal, further extract carbon particle response features from the signal, and generate a temporal dynamic feature representation of the electric field signal.

[0061] In another embodiment, in the signal acquisition and preprocessing unit, the micro-nano electrode feedback signal is sampled by a 250MS / s, 12-bit ADC, and then the FPGA performs framing and time-stamping of 1024 samples, embedding a time-frequency hybrid coding tag. Primary denoising is performed within the FPGA using Daubechies-8 wavelet decomposition at 8 levels to remove noise components exceeding the 0.1MHz to 10MHz bandwidth. The denoised data is then transferred to the ARM Cortex-A53 main control CPU via DMA. On the CPU, the system first calls the ARIMA(p,d,q) model (p=2, d=1, q=2) to perform predictive residual analysis on each frame of signal, automatically determining the start t0 and end t1 of each frequency band; then, it calculates the mutual information (MI) time-series curve, eliminating abnormal spikes exceeding the threshold; finally, it performs max-min normalization on each signal segment, outputting zero-mean, cell variance data blocks.

[0062] The time-frequency feature extraction unit is deployed on a GPU accelerator board, performing parallel short-time Fourier transform (STFT) and Chirp-Z transform fusion operations for each signal block. Specific steps include: selecting a switchable window function (Hamming window or Blackman window) of length N = 512 for the signal block, and calculating respectively:

[0063]

[0064] Where φ n This is the linear frequency modulation phase quantity generated online for the Chirp-Z controller. The system further performs cubic spline interpolation and smoothed polynomial fitting on the obtained amplitude Af(t) and phase φf(t) curves to extract 16 time-frequency features, such as amplitude peaks, slope change points, and second derivative extrema. The feature vectors of each frequency band are concatenated to form an overall input vector X with dimension M≈256.

[0065] The deep fusion signal processing unit is based on the original CLDNet network structure of this invention, and sequentially includes a feature denoising layer, a temporal modeling layer, and a dynamic decision layer. The feature denoising layer employs a two-layer multi-head self-attention mechanism (MHSA, Head=8), with each layer followed by a residual connection and LayerNorm. The temporal modeling layer consists of three layers of bidirectional LSTM (HiddenSize=256) to capture the spatiotemporal correlation of features in different frequency bands. The dynamic decision layer deploys XGBoost (tree number=100, Depth=6) regression branches and a Bayesian optimization classifier in parallel, simultaneously outputting regression values ​​and discrete categories for carbon particle concentration C, particle size distribution D, and morphological index M. The entire network is trained offline on a manually prepared calibration sample set (2,000 samples, concentration range 0-100ppm, particle size 50-500nm), and undergoes minor iterative fine-tuning after each batch measurement to adapt to changes in the online environment. The overall input vector X is input to the trained deep fusion signal processing unit.

[0066] The results output and feedback control unit visualizes the C, D, and M data sets in real time via a touchscreen UI, displaying concentration-time curves, particle size histograms, and speciation pie charts. When C exceeds the user-defined threshold Cthr or the network detects an abnormal morphology probability > 0.9, the system automatically triggers an audible and visual alarm, records the event log, and activates the reference oil sample closed-loop injection module to complete an automatic calibration test. All measurement data are stored in CSV format on a local SSD and encrypted and uploaded to the cloud platform via a 4G / 5G communication module, enabling remote monitoring and historical data retrieval.

[0067] To improve the accuracy of time-frequency signals, the system will further perform deep fusion processing on the time-series features to construct a 12-dimensional dynamic tensor T. 12 To represent the complex response of a signal across various frequency bands and time scales:

[0068]

[0069] Among them, F i It is the time-frequency characteristic of the i-th frequency component, g i (x t ) is the corresponding time series response function, α i These are weighting coefficients, reflecting the contribution of different frequency components to the overall features. Through the combination of these multi-dimensional features, the system can more comprehensively understand the time-frequency ridges of the signal and extract accurate carbon particle information. In the final dynamic decision-making stage, the system combines the features after time-series modeling and noise reduction processing, and outputs carbon particle concentration, particle size distribution, and morphology classification through an optimized neural network. Carbon particle concentration C n The calculation formula is:

[0070]

[0071] Among them, |F i | 2 This represents the energy of the i-th frequency band, k1 is a constant coefficient, and α i and β i These are the weighting coefficients and response exponents for each frequency band, reflecting the concentration contribution of carbon particles at different frequencies.

[0072] The formula for calculating particle size distribution is:

[0073]

[0074] Where γi and δi are the weighting coefficients of particle size distribution, g i (x t) is the time-frequency response function, representing the particle size response corresponding to each frequency band. Particle size distribution, through analysis of multi-frequency signals, precisely distinguishes carbon particles from the micrometer to the nanometer scale.

[0075] The morphological classification output is optimized using the following formula:

[0076]

[0077] Among them, P i Let d represent the probability of the i-th form. i It is the particle size in this morphology, d avg λ is the average particle size of all particles. i and η i It is a weighting coefficient for morphological classification, and this formula enables accurate classification of spherical, sheet-like, and fibrous morphologies.

[0078] Specifically, to eliminate the impact of temperature and pressure fluctuations on the accuracy of electric field measurement, this invention incorporates a dual compensation module for temperature T(t) and pressure P(t) in the main circuit. Firstly, during the equipment manufacturing phase, the system undergoes testing on a constant temperature and pressure test bench to measure the electric field under various conditions: temperature T (–40~85℃) and pressure P (0.8~1.2 atm). m (t) Perform multi-point calibration and use the least squares method to fit the baseline compensation coefficients α0 and β0;

[0079] During operation, four Pt1000 platinum resistance temperature sensors and two MEMS piezoresistive pressure sensors sample at 1 kS / s. After Sigma-Delta A / D conversion, the data is buffered via FPGA channels and sent to the calibration unit in real time. This unit, combined with a one-dimensional Kalman filter, processes the measured E... m (t) is recursively corrected, and least squares fine-tuning is triggered every 10 seconds to update α(t) and β(t). Compensation E corr (t) is calculated as follows:

[0080] E corr (t)=E m (t)-α(t)[T(t)-T0]-β(t)[P(t)-P0]+K KF (E m (t),t)

[0081] Where T0 and P0 are reference temperature and pressure states, K KF The Kalman gain function is used, and its state equation and measurement equation are set with reference to the standard one-dimensional model. Through this module, the system can control the electric field drift caused by temperature and pressure errors within ±0.5%.

[0082] Step 4: When the detected carbon particle concentration exceeds the set threshold, the system will trigger an audible and visual alarm and send an SMS notification.

[0083] The intelligent early warning and interactive platform features an AR (Augmented Reality) interface that dynamically displays a carbon particle deposition heat map (60fps refresh rate). It predicts the risk of insulating oil degradation within 72 hours using a transfer learning model and triggers a three-level alarm system in advance, including audible and visual alarms, SMS notifications, and cloud-based alarms (false alarm rate ≤0.1%). Furthermore, the system transmits data back to the main control system via the TSN protocol, supporting seamless integration with the IEC 61850 standard protocol to ensure compatibility with existing power equipment management systems and high-efficiency data transmission.

[0084] Users can set the monitoring cycle (adjustable from 1 minute to 24 hours) through an augmented reality (AR) interface, and dynamically optimize the detection frequency based on the transformer's load rate and ambient temperature and humidity (supporting a range of -40℃ to 120℃). The system has a built-in operating condition matching algorithm that can automatically recommend the optimal monitoring scheme based on oil flow rate (1 to 50 m / s) and historical deterioration trends.

[0085] The AR interface dynamically displays a thermal map of carbon particle deposition (60fps refresh rate), with areas exceeding the standard marked by 3D pulsed light spots, and supports gesture-based zoom analysis. The early warning system includes a three-level alarm linkage strategy: Primary warning (concentration exceeding threshold by 50%): triggers a 105dB audible and visual alarm; Intermediate warning (concentration exceeding the standard for 3 consecutive cycles): automatically pushes SMS messages and equipment work orders; Advanced warning (CPHI index > 0.8): initiates cloud-based expert consultation and links with oil regeneration equipment (via OPC UA protocol). The alarm system has a built-in environmental adaptive module that dynamically adjusts the alarm volume according to background noise (30-90dB).

[0086] The signal processing unit outputs carbon particle concentration, particle size distribution, and morphology classification with an accuracy of ±0.5%. If the confidence level of the system output is below 95%, an incremental learning mechanism is triggered to resample the signal using new samples and update the parameters of the neural network model to achieve self-optimization of the model.

[0087] Through the aforementioned complex signal analysis process, this invention can efficiently and accurately extract relevant information about carbon particles, while improving the system's adaptability and robustness in complex environments, providing strong data support for carbon particle detection and equipment early warning.

[0088] The AR interface dynamically displays a thermal map of carbon particle deposition at a refresh rate of 60fps. Areas exceeding the standard are marked with 3D pulsed light spots and support gesture controls for zooming in and out, facilitating in-depth analysis and localization.

[0089] The early warning system includes a three-level alarm linkage strategy:

[0090] Primary warning (concentration exceeds threshold by 50%): Triggers a 105dB audible and visual alarm to notify the user of the potential risk of carbon particle accumulation.

[0091] Intermediate warning (exceeding the limit for 3 consecutive cycles): Automatically pushes SMS messages and equipment work orders to remind maintenance personnel to pay attention to the equipment status.

[0092] Advanced warning (CPHI index > 0.8): Initiate cloud-based expert consultation, provide remote diagnosis and decision support, and can link with oil regeneration equipment for processing (linked via OPC UA protocol).

[0093] To adapt to varying environmental noise levels, the system incorporates an environmental adaptive module that dynamically adjusts the alarm volume based on background noise variations (e.g., changes within the 30-90 dB range). The algorithm's principle is as follows: Figure 5 As shown. This module performs adaptive noise adjustment using the following formula:

[0094]

[0095] Among them: A adjusted The adjusted alarm volume; Abase is the basic alarm volume (volume under standard noise conditions); N is the current ambient noise level; N0 is the noise baseline value; ΔN is the noise tolerance range; α is the adjustment coefficient used for sensitivity adjustment.

[0096] In addition to detecting carbon particle concentration, the system can also trigger an abnormal mode detection mechanism based on changes in parameters such as oil flow rate, temperature, and pressure. When the system detects that changes in these parameters exceed normal ranges, indicating a potential risk of equipment degradation, it will activate an early warning mechanism. This detection mechanism is based on the following formula:

[0097]

[0098] Where: X is the current monitored parameter value (such as oil flow rate, temperature, etc.); X0 is the historical standard value; σ is the standard deviation of the parameter. When the deviation value exceeds the set threshold, the system will trigger an early warning mechanism to further analyze the potential risk of carbon particle accumulation.

[0099] The alarm system has a built-in environment adaptive module that can dynamically adjust the alarm volume according to the background noise (30-90dB) to ensure that the alarm can be clearly received in different noise environments.

[0100] The system transmits data via a Time-Sensitive Network (TSN), is compatible with the IEC 61850 protocol, and has a data transmission latency of ≤1ms. The storage unit constructs a spatiotemporal matrix of carbon particle growth and combines it with an LSTM prediction model to analyze the degradation trend over 72 hours. The model accuracy reaches over 92%.

[0101] The system automatically performs dual-frequency calibration (1kHz / 1MHz) and Kalman filter calibration daily to ensure drift ≤ ±0.5% FS over 5 years. The Kalman filter formula is as follows:

[0102]

[0103] Where P(t+1) is the prediction error covariance matrix for the next time step. Let P(t) be the system state transition matrix, P(t) be the error covariance matrix at the current time, and Q be the process noise covariance matrix.

[0104] To ensure the system's accuracy during long-term operation, especially in response to environmental changes and new failure modes, this invention introduces an incremental learning mechanism. When the system's prediction accuracy decreases, a self-optimization process is triggered, retraining the model using new monitoring data. The core idea of ​​incremental learning is to update the model using the following formula:

[0105]

[0106] θ t+1 =θ t +Δθ t

[0107] Where: θ t The model parameters at the current time step are η, the learning rate is η, and L(θ) is the learning rate. t D t ) is the loss function; This is the gradient of the loss function.

[0108] By applying the aforementioned multimodal early warning and data fusion, and predictive maintenance technologies, this invention can effectively improve the equipment health management capabilities of substations. Through precise fault early warning and predictive maintenance, the system minimizes equipment failure risks, optimizes maintenance processes, and enhances the operational safety and reliability of substations.

[0109] The second aspect of this invention relates to an intelligent detection device for carbon particles in insulating oil based on a microstructure gradient electric field; implemented using a method for intelligent detection of carbon particles in insulating oil based on a microstructure gradient electric field according to the first aspect of this invention; wherein the device includes an equipment installation module, a parameter setting module, a data analysis and alarm module, and a data storage and remote monitoring module; the equipment installation module is used to install the electrode array in the transformer tank using a dedicated bracket to ensure full contact between the electrodes and the insulating oil, and connects a signal processing unit, a display and alarm module, and a remote monitoring module; the parameter setting module is used to set the monitoring cycle, alarm threshold, and alarm level via a touch screen, and configure the network parameters of the remote monitoring module to ensure data upload; the data analysis and alarm module is used to apply a high-frequency AC electric field to the electrode system and collect the electric field signal in real time, analyze the electric field signal using a deep learning algorithm, identify the type of carbon particles, and calculate their concentration; the data storage and remote monitoring module is used to trigger an audible and visual alarm and send a text message notification when the detected carbon particle concentration exceeds a set threshold.

[0110] The carbon particle detection device includes a micro-nano composite electrode array, a time-frequency hybrid coding excitation module, a multi-mode resonant signal analysis unit, a dynamic anti-interference protection system, and an intelligent early warning and interaction platform. The carbon particle detection device mainly includes an electrode system, a signal processing unit, a display and alarm module, a data storage unit, and a remote monitoring module, aiming to further improve the safety and reliability of substation transformers.

[0111] The electrode system employs an improved micro / nano electrode array, with the electrode material being a highly conductive platinum alloy to enhance the uniformity and stability of the electric field. To improve the sensitivity and detection accuracy of the electrodes, the electrode array adopts a tip-planar structure, meaning that some electrodes in the array are tip-shaped while others are planar. This design enhances the electrodes' ability to capture carbon particles.

[0112] Electrode size and spacing: The tip of the needle electrode has a diameter of 50 μm, the planar electrode has a size of 5 mm × 5 mm, and the electrode spacing is adjustable, ranging from 50 μm to 300 μm. By adjusting the distance between the needle tip and the planar electrode, the detection effect of carbon particles of different sizes can be optimized.

[0113] Electrode materials and surface coating: The electrode surface is coated with a multilayer graphene nanomaterial. This coating effectively improves the electrode's charge trapping ability and ensures its stability after long-term use. The excellent conductivity of graphene makes the electrode more sensitive to changes in electric field.

[0114] Electrode Installation: The electrode array is fixed inside the transformer tank by a high-strength insulating material bracket, ensuring full contact between the electrodes and the insulating oil. The electrode installation angle is adjustable from 0° to 30° to accommodate different types of transformer containers.

[0115] The electric field is applied as a high-frequency alternating electric field, with a frequency range of 1kHz to 50kHz and a voltage range of 10V to 100V. By adjusting the frequency and voltage, the strength and uniformity of the electric field are effectively optimized, ensuring sensitive detection of various carbon particles.

[0116] The signal processing unit employs a multi-level processing algorithm combining deep learning and time-frequency analysis to accurately analyze the acquired electric field signals. The key optimization points of this algorithm include:

[0117] Multi-band time-frequency analysis: By using the Chirp-Z algorithm to perform time-frequency transformation on the signal, it is possible to accurately capture minute fluctuations in the electric field signal and identify the influence of different types of carbon particles.

[0118] Deep Convolutional Neural Networks (CNNs): CNNs are used for feature extraction and classification, enabling automatic identification and classification of carbon particle concentration, size, and morphology in electric field signals. CNN networks process signal features through multiple convolutional layers, perform dimensionality reduction via pooling layers, and finally output relevant information about carbon particles through fully connected layers.

[0119] Dynamic adjustment algorithm: This signal processing unit can automatically adjust the parameters of the processing algorithm according to changes in the actual environment, such as temperature, humidity, and oil flow rate. For example, if the oil flow rate is high, the system will automatically increase the monitoring frequency to respond more promptly to changes in carbon particle concentration.

[0120] The algorithm's computation time is less than 0.05 seconds, meeting the requirements for real-time monitoring. The final output of carbon particle information includes concentration, particle size distribution (micrometer to nanometer scale), and morphology (spherical, flake, or fibrous).

[0121] The alarm display module features a 7-inch high-definition touchscreen design, providing a user-friendly interface. Users can set the monitoring cycle (from 1 minute to 24 hours), adjust alarm thresholds, and view real-time monitoring data and historical trends via the touchscreen.

[0122] Display Interface: The user interface features a simple and intuitive design, including functions such as real-time monitoring data, historical data curves, alarm settings, and system configuration. Users can directly set and operate the device via the touchscreen, enhancing its operability.

[0123] Alarm function: When the detected carbon particle concentration exceeds the set threshold, the system will alert the user with an audible and visual alarm. The alarm volume is adjustable (60dB to 90dB), and the alarm light color is customizable (red, yellow, green). In addition, the system supports SMS notification; when the carbon particle concentration is abnormal, an SMS will be automatically sent to the user's preset mobile phone number, containing the carbon particle concentration, detection time, and device location.

[0124] The remote monitoring module has been expanded to support remote data access and control via a cloud platform. Monitoring data is uploaded to the cloud server via 4G / 5G networks, and users can view the device's operating status, historical data, and analysis reports in real time through a mobile app or web browser.

[0125] Cloud storage and data analytics: The cloud platform has a storage capacity of 1TB, supporting long-term storage of monitoring data. The cloud platform integrates data analytics tools, enabling trend prediction and fault diagnosis, and providing equipment maintenance recommendations based on carbon particle concentration trends.

[0126] Remote control functionality: Users can remotely control the device via a mobile app or web browser, including starting / stopping monitoring, adjusting monitoring cycles, and modifying alarm thresholds. The remote monitoring module supports multi-user access control, allowing multiple users to simultaneously view and analyze the same set of data.

[0127] The specific implementation steps include:

[0128] Equipment Installation: Install the electrode array inside the transformer's oil tank using a dedicated bracket, ensuring full contact between the electrodes and the insulating oil. Connect the signal processing unit, display and alarm module, and remote monitoring module.

[0129] Parameter settings: Set the monitoring cycle, alarm threshold, and alarm level via the touchscreen. Configure the network parameters of the remote monitoring module to ensure data can be uploaded to the cloud.

[0130] Electric field application and signal acquisition: The electrode system applies a high-frequency AC electric field and acquires the electric field signal in real time. The signal is then analyzed by the signal processing unit.

[0131] Data Analysis and Alarm: The signal processing unit uses deep learning algorithms to analyze the electric field signal, identify the type of carbon particles, and calculate their concentration. When the detected carbon particle concentration exceeds a set threshold, the system will trigger an audible and visual alarm and send an SMS notification.

[0132] Data storage and remote monitoring: Monitoring data is automatically uploaded to the cloud server, and users can view the data and analysis reports in real time through an app or web browser. The system performs trend analysis based on historical data and provides maintenance suggestions.

[0133] A third aspect of the present invention relates to a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to perform the steps of the method described in the first aspect of the present invention.

[0134] A fourth aspect of the present invention relates to a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in the first aspect of the present invention.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions of the present invention still include content that can be modified or equivalently replaced in the specific implementation of the present invention. Any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A smart detection method for carbon particles in insulating oil based on a microstructure gradient electric field, characterized in that, The method includes the following steps: The electrode array is installed inside the transformer's oil tank using a dedicated bracket to ensure full contact between the electrodes and the insulating oil, and is connected to the signal processing unit, display and alarm module, and remote monitoring module. The electrode array employs asymmetric interdigitated electrodes manufactured using MEMS technology. A nano-zinc oxide / graphene heterogeneous coating is deposited on the electrode surface via magnetron sputtering, creating a localized electric field strength ≥1.8 × 10⁻⁶ in the insulating oil. 5 Gradient electric field of V / m; The monitoring cycle, alarm threshold and alarm level can be set via the touch screen, and the network parameters of the remote monitoring module can be configured to ensure data upload. A high-frequency AC electric field is applied to the electrode system, and the electric field signal is acquired in real time. A wideband electric field of 0.1–10 MHz is applied using a time-frequency hybrid coding technique. A 12th-order pseudo-random phase modulation signal is generated based on the Chirp-Z algorithm and output to the micro / nano electrode array via a 250 MS / s high-speed DAC, simultaneously exciting the polarization migration and relaxation effects of carbon particles. Deep learning algorithms are used to analyze the electric field signal, identify the type of carbon particles, and calculate their concentration. Specifically, a CLDNet fusion neural network model is used for holographic signal analysis, and a 7-level adaptive wavelet packet decomposition is used to remove power frequency noise. A bidirectional LSTM network is used to analyze the time-frequency ridge features, outputting the concentration, particle size distribution, and morphology of the carbon particles. When the detected carbon particle concentration exceeds the set threshold, the system will trigger an audible and visual alarm and send an SMS notification.

2. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 1, characterized in that: The electrode array is installed inside the transformer's oil tank using a dedicated bracket to ensure full contact between the electrodes and the insulating oil. It is then connected to a signal processing unit, a display and alarm module, and a remote monitoring module, including: In the electrode array, some electrodes are needle-shaped, while others are planar. The tip diameter of the needle electrode is 50μm, the size of the planar electrode is 5mm×5mm, and the electrode spacing is adjustable; The electrode surface is covered with a multilayer graphene nanomaterial, and the electrode array is fixed in the transformer tank by a high-strength insulating material support. The installation angle of the electrode can be adjusted from 0° to 30°.

3. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 2, characterized in that: The process of applying a high-frequency alternating electric field to the electrode system and acquiring the electric field signal in real time includes: The electric field is applied as a high-frequency alternating electric field with a frequency range of 1kHz to 50kHz and a voltage range of 10V to 100V.

4. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 3, characterized in that: The process of applying a high-frequency alternating electric field to the electrode system and acquiring the electric field signal in real time, analyzing the electric field signal using a deep learning algorithm, identifying the type of carbon particles, and calculating their concentration includes: The signal is transformed using the Chirp-Z algorithm; By using the CLDNet fusion neural network model for feature extraction and classification, the concentration, size and morphology of carbon particles in the electric field signal are automatically identified and classified.

5. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 4, characterized in that: The process of applying a high-frequency alternating electric field to the electrode system and acquiring the electric field signal in real time, analyzing the electric field signal using a deep learning algorithm, identifying the type of carbon particles, and calculating their concentration includes: The processing algorithm parameters are automatically adjusted based on changes in the actual environment, such as temperature, humidity, and oil flow rate.

6. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 5, characterized in that: Feature extraction and classification using CLDNet are employed to automatically identify and classify the concentration, size, and morphology of carbon particles in electric field signals, including: The output carbon particle information includes concentration, particle size distribution, and morphology; The particle size distribution ranges from micrometers to nanometers. Its shape can be spherical, sheet-like, or fibrous.

7. The intelligent detection method for carbon particles in insulating oil based on a microstructure gradient electric field according to claim 6, characterized in that: When the detected carbon particle concentration exceeds a set threshold, the system will trigger an audible and visual alarm and send an SMS notification, including: When the detected carbon particle concentration exceeds the set threshold, the system will alert the user through an audible and visual alarm.

8. A smart detection device for carbon particles in insulating oil based on a microstructure gradient electric field; characterized in that: The method is implemented using the intelligent detection method for insulating oil carbon particles based on a microstructure gradient electric field as described in any one of claims 1-7; wherein... The device includes an equipment installation module, a parameter setting module, a data analysis and alarm module, and a data storage and remote monitoring module; The device installation module is used to install the electrode array in the transformer's oil tank using a special bracket, ensuring that the electrodes are in full contact with the insulating oil, and connects to the signal processing unit, display alarm module, and remote monitoring module. The parameter setting module is used to set the monitoring cycle, alarm threshold and alarm level via the touch screen, configure the network parameters of the remote monitoring module, and ensure data upload. The data analysis and alarm module is used to apply a high-frequency AC electric field to the electrode system and collect the electric field signal in real time. It uses a deep learning algorithm to analyze the electric field signal, identify the type of carbon particles, and calculate their concentration. The data storage and remote monitoring module is used to trigger an audible and visual alarm and send a text message notification when the detected carbon particle concentration exceeds a set threshold.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-7.

Images