A method for optimizing the state of a solid-sealed pole column for a primary-secondary integrated circuit breaker

Abstract

The present invention discloses a method for optimizing the state of a solid-sealed pole column for a primary-secondary integrated circuit breaker. The present invention aims to solve the problem of misoperation of the circuit breaker caused by traditional fixed-threshold criteria. The method uses a multi-physical-field coupling monitoring network to collect data on the temperature gradient, mechanical vibration frequency, strain distribution, and electric field strength of the solid-sealed pole column in real time. Combining the extraction of spatio-temporal characteristics of partial discharge signals, a comprehensive degradation index (CDI) is constructed. An improved particle swarm optimization algorithm is used to dynamically optimize the feature weights, and a bidirectional long short-term memory network (LSTM) is used to predict the degradation trend to generate a dynamic warning threshold (DWT) and a critical threshold (CWT). It realizes efficient data processing and iterative optimization of the model. At the same time, the mechanical strain, electric field distortion, and discharge evolution path are visualized through a three-dimensional digital twin model. The actual measurement shows that this method reduces the fault misjudgment rate to less than 5%, shortens the warning response time to within 10 seconds, and greatly reduces the partial discharge amount through the silver nanowire self-healing technology.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method for optimizing the state of a solid-sealed pole column for a primary-secondary integrated circuit breaker. Background Art

[0002] When a single-phase grounding fault occurs in a power system, multiple potential safety hazards may be triggered: First, the fault current will directly threaten the insulation performance of distribution and substation equipment, leading to equipment damage and even fires; Second, the system voltage imbalance will affect the stable operation of the regional power grid, resulting in a decline in power supply quality; More seriously, dangerous step voltages may be generated around the fault point, endangering human life and causing livestock casualties. To address such risks, modern power systems generally adopt primary-secondary integrated pole-mounted circuit breakers as the core protection devices, which effectively isolate the fault area and ensure system safety through real-time state switching functions (such as switching from closed to open during a fault).

[0003] Current technical solutions mainly rely on zero-sequence current / voltage monitoring to achieve fault judgment: By continuously collecting zero-sequence current (I0) and zero-sequence voltage (U0) data of the line and statically comparing them with preset fixed thresholds (such as I0 = 10A or U0 = 30V), when the monitored value exceeds the threshold, the circuit breaker is triggered to operate. However, this solution has significant limitations - the fixed threshold is difficult to adapt to complex working condition changes. For example, in a cable line with a large capacitive current, a zero-sequence current of up to 8A may be generated during normal operation; in the case of an instantaneous overvoltage caused by lightning strikes, the zero-sequence voltage may briefly exceed 25V but not be a real grounding fault. Such scenarios are extremely likely to cause misjudgment by the protection device, resulting in frequent mis-tripping of the circuit breaker, not only reducing power supply reliability but also potentially causing economic losses due to unnecessary power outages.

[0004] The essence of the misoperation problem lies in the lack of dynamic adaptability of the traditional threshold judgment method: It can neither distinguish real faults from transient interferences nor cope with the time-varying characteristics of line parameters (such as changes in insulation resistance caused by seasonal temperature rise). Therefore, developing an adaptive protection algorithm with intelligent discrimination ability to fundamentally improve the accuracy of fault identification has become a technical bottleneck that needs to be broken through urgently in the construction of intelligent distribution networks. Summary of the Invention

[0005] (I) Technical Problems to be Solved

[0006] To solve the above problems, the present invention proposes a method for optimizing the state of a solid-sealed pole column for a primary-secondary integrated circuit breaker, aiming to solve the problems in the prior art that rely on zero-sequence current / voltage monitoring to achieve fault judgment and the essence of the misoperation problem lies in the lack of dynamic adaptability of the traditional threshold judgment method.

[0007] (II) Technical Solutions

[0008] A method for optimizing the state of a solid-sealed pole column for a primary-secondary integrated circuit breaker of the present invention includes:

[0009] Obtain multi-physical field coupling monitoring data inside the solid-sealed pole column, electric field intensity distribution data on the surface and inside the pole column, and capture the spatio-temporal characteristics of the partial discharge signal of the solid-sealed pole column;

[0010] The multi-physical field coupling monitoring data includes temperature gradient, mechanical vibration frequency and strain distribution. Analyze the multi-physical field coupling monitoring data to extract the first feature dataset, and construct a comprehensive degradation index through weighted fusion of the first feature dataset to obtain a prediction result;

[0011] Based on historical data and the prediction result, use a long short-term memory network to predict the degradation trend in the future period, generate dynamic warning thresholds and critical thresholds, and generate action instructions through the dynamic warning thresholds and the critical thresholds.

[0012] In the present invention, the analysis of the multi-physical field coupling monitoring data to extract the first feature dataset includes:

[0013] Perform empirical mode decomposition on the temperature gradient signal to extract the intrinsic mode function, calculate the short-time Fourier transform energy spectrum entropy for the mechanical vibration frequency signal, construct a covariance matrix for the strain distribution and extract the maximum eigenvalue to form a multi-dimensional time-frequency domain fusion feature vector.

[0014] In the present invention, it also includes a step of visualizing and reconstructing the degradation state. Map the multi-physical field coupling monitoring data to a three-dimensional digital twin model of the solid-sealed pole column, and display the spatial and temporal evolution trajectories of the mechanical strain concentration area, the high-risk area of electric field distortion and the partial discharge pulse through the superposition of heat maps, and mark the area where the dynamic threshold is exceeded.

[0015] In the present invention, the capture of the spatio-temporal characteristics of the partial discharge signal of the solid-sealed pole column includes: using an improved matching pursuit algorithm to perform sparse decomposition on the ultrasonic signal, the basis function library includes Gaussian modulated sine waves and impulse response waveforms, and localize the spatial coordinates of the discharge source and the energy diffusion path through time-delay-Doppler joint parameter estimation.

[0016] In the present invention, the method is implemented based on an edge-cloud collaborative computing framework. The edge node performs data preprocessing and feature extraction, and the cloud platform performs iterative training of the degradation model. The two transmit the compressed degradation index sequence through a feature-level differential privacy encryption channel.

[0017] In the present invention, the calculation formula of the comprehensive degradation index is: , where represents the weight of the heat accumulation index, reflecting the influence of temperature on the aging of the pole column represents the heat accumulation index, represents the vibration offset, Represents the weight of the vibration offset, characterizing the contribution of mechanical stress to performance degradation, Represents the electric field distortion rate, Represents the weight of the electric field distortion rate, measuring the impact of abnormal electric field distribution on insulation performance, Is the partial discharge energy entropy, The weight of the partial discharge energy entropy, quantifying the correlation between the complexity of the discharge pattern and insulation defects;

[0018] Thermal accumulation index Reflects the heat accumulation effect during the long-term operation of the pole column, used to predict the thermal aging of insulating materials, partial discharge energy entropy The higher the entropy value, the more complex the discharge pattern and the greater the risk of insulation defects.

[0019] In the present invention, the weight coefficient , , , Is dynamically adjusted by an improved particle swarm optimization algorithm, and its fitness function fuses the KL divergence of historical fault samples and the environmental correction factor of real-time working conditions. The weight adjustment period is an integer multiple of the preset time window and satisfies the constraint conditions .

[0020] In the present invention, it further includes a fault root cause analysis module. When a critical threshold overrun event is detected, it automatically associates historical multi-physical field data to generate a fault evolution tree diagram, and calculates the contribution weight of each physical field parameter through a causal reasoning engine, outputting the dominant failure factor and the associated confidence level. The contribution weight is calculated based on the Shapley value algorithm.

[0021] In the present invention, the action instruction includes a hierarchical response strategy:

[0022] When the first-level instruction is triggered, the pole column is in a healthy state and only records data;

[0023] When the second-level instruction is triggered, the partial discharge suppression module is started to reduce the electric field distortion by adjusting the surface potential distribution of the pole column;

[0024] When the third-level instruction is triggered, the mechanical stress compensation device is started to actively cancel the vibration energy using piezoelectric ceramics and notify the operation and maintenance personnel for emergency intervention;

[0025] The level of the action instruction is determined by the warning threshold and the critical threshold.

[0026] (III) Beneficial effects

[0027] Compared with the prior art, the beneficial effects of the present invention are:

[0028] (1) In the present invention, the feature weights are optimized based on an improved particle swarm optimization algorithm, and a dynamic threshold is generated by combining the confidence interval predicted by the LSTM network, effectively distinguishing true faults from transient interferences and greatly reducing the misjudgment rate; a three - level linkage protection mechanism (data recording - potential regulation - mechanical compensation) realizes precise fault isolation, shortens the response time compared with the fixed - threshold strategy, and avoids economic losses caused by unnecessary power outages. Description of the Drawings

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following - described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0030] Figure 1 It is a schematic diagram of the logical structure of the optimization method. Detailed Embodiments

[0031] Embodiment

[0032] As Figure 1 shown, the state optimization method for the solid - encapsulated pole of a primary - secondary integrated circuit breaker provided by the present invention includes the following steps:

[0033] S100. Obtain the multi - physical - field coupling monitoring data inside the solid - encapsulated pole, the electric - field intensity distribution data on the surface and inside the pole, and capture the spatio - temporal characteristics of the partial - discharge signals of the solid - encapsulated pole.

[0034] Embed a high - precision sensing network inside the epoxy resin encapsulation layer of the solid - encapsulated pole. Specifically, arrange 6 groups of fiber - Bragg - grating sensor groups at equal intervals along the axial direction of the pole. Each group contains 3 gratings in orthogonal directions to monitor the mechanical strain in the X / Y / Z axial directions respectively. The wavelength resolution of each grating is 0.1 pm, the temperature measurement accuracy reaches ±0.5 °C, and high - speed sampling of 2000 times per second is achieved through frequency - division multiplexing technology.

[0035] Specifically, the multi - physical - field coupling monitoring data includes temperature gradient, mechanical vibration frequency, and strain distribution.

[0036] For electric - field monitoring, 24 micro - capacitance coupling probes with a circular distribution are used and arranged around the conductor at a spacing of 5 mm. The surface of the probes is coated with a nano - scale insulating coating to avoid electric - field interference. The original voltage signals collected in real - time are processed by a differential - amplification circuit and then the three - dimensional electric - field intensity distribution map is reconstructed through a finite - element inversion algorithm, with a spatial resolution better than 1 mm³.

[0037] In terms of partial discharge detection, a 32-channel ultrasonic sensor array is evenly arranged on the outer surface of the pole column. Each unit uses a 40 kHz narrow-band filter to eliminate environmental noise. The coordinates of the discharge source are determined by the time difference positioning method. The time-frequency characteristics of the discharge pulse are extracted by combining the Morlet wavelet transform. The feature dimensions include 12 parameters such as pulse amplitude, rise time, and frequency band energy ratio.

[0038] S200. Analyze the multi-physical field coupling monitoring data to extract the first feature dataset, and construct a comprehensive degradation index by weighted fusion of the first feature dataset to obtain the prediction result.

[0039] At the data processing layer, an edge computing node needs to be constructed to achieve real-time fusion of multi-source data. After the temperature data is filtered by the Kalman filter to eliminate transient noise, the axial temperature gradient is calculated and the radial thermal accumulation index , where is obtained by integrating the deviation between the temperature curve in the past 30 minutes and the baseline value, while the axial temperature gradient is used for real-time diagnosis of local hot spots to guide the operation and maintenance personnel to locate the fault area.

[0040] After the mechanical vibration signal extracts the main frequency component through the fast Fourier transform, the main frequency offset Δf is calculated using a sliding window. The window length is 10 seconds and the step size is 1 second. When Δf exceeds 15% of the baseline value, the strain energy density (SED) calculation module is triggered. The SED value is dynamically updated according to the formula , where is the strain in each direction, is the corresponding stress. The electric field distortion rate is defined as the cosine similarity between the current electric field distribution and the standard distribution, and is obtained by normalizing the dot product operation between the reconstructed electric field vector and the standard template. The calculation period is 5 times per second.

[0041] The partial discharge energy entropy is calculated by the Shannon entropy of the frequency domain energy of all discharge pulses within 1 minute. The formula is , where is the proportion of the energy in each frequency band.

[0042] The dynamic weight adjustment mechanism of the comprehensive degradation index (CDI) adopts an improved particle swarm algorithm. When initializing, set , , , As the basic weight, the objective function is set to minimize the mean square error between historical fault data and CDI values. When the algorithm is running, the latest fault case library is downloaded from the cloud every 15 minutes, and 200 particles are generated through parallel calculation for iterative optimization. The inertia weight decreases linearly from 0.9 to 0.4, and the acceleration constants c1=1.5 and c2=1.8. After 50 generations of iteration, the optimal weight combination is output. For example, under the condition of increased humidity in the rainy season, the system automatically increases the partial discharge weight δ to 0.35, and reduces the mechanical vibration weight β to 0.15 to adapt to the impact of environmental changes.

[0043] S300, based on historical data and the prediction results, using a long short-term memory network to predict the degradation trend in the future period, generate a dynamic warning threshold and a critical threshold, and generate an action instruction through the dynamic warning threshold and the critical threshold.

[0044] Specifically, historical data includes multi-physics field coupling monitoring data (such as temperature gradient, mechanical vibration frequency, strain distribution) and extracted comprehensive degradation index (CDI). First, the original data is cleaned to remove outliers and noise, and the dimensions are unified through Z-score normalization. Time series data is segmented according to fixed time windows (such as every hour) to form input-output pairs, such as the CDI sequence of the past 24 hours as input and the CDI sequence of the next 6 hours as output target.

[0045] The dynamic threshold generation module adopts a two-layer LSTM network structure. The input layer contains 16-dimensional features such as CDI time series data, ambient temperature and humidity, and load current. The hidden layer is set with 128 neurons, and the output layer predicts the CDI change curve within the next hour. The training data comes from 120,000 sets of samples collected in 3 years. The Adam optimizer is used for model training. The initial learning rate is 0.001 and decays by 50% every 10 rounds.

[0046] The LSTM output is randomly perturbed multiple times, such as adding Gaussian noise, to generate the probability distribution of future CDI. The dynamic warning threshold (DWT) is set to the 95th percentile of the predicted value, and the critical threshold (CWT) is set to the 99th percentile. For example, if the mean CDI predicted for the next hour is 75 and the standard deviation is 5, then the DWT is 75 + 1.645×5 ≈ 83.2 and the CWT is 75 + 2.326×5 ≈ 86.6. The threshold is updated every 5 minutes to ensure that it adapts to the latest data trends.

[0047] The prediction results are simulated by Monte Carlo to generate confidence intervals. The dynamic warning threshold DWT takes the 95% quantile of the predicted value, and the critical threshold CWT takes the 99% quantile. When it is detected that CDI exceeds DWT for three consecutive times, the system activates the potential control module and applies reverse voltage through 64 micro-electrodes distributed on the surface of the pole. The adjustment amount is dynamically calculated according to the electric field distortion rate. The maximum adjustment voltage is ±5kV, and the response time is less than 20ms.

[0048] Action instructions include a graded response strategy:

[0049] Level 1 instruction (CDI ≥ DWT and < CWT), when the level 1 instruction is triggered, the pole is in a healthy state and only data is recorded;

[0050] Secondary instruction (CDI ≥ CWT): When the secondary instruction is triggered, the partial discharge suppression module is activated to reduce the electric field distortion by adjusting the potential distribution on the pole surface;

[0051] Level 3 instruction (exceeding CWT three times in a row): When the level 3 instruction is triggered, the mechanical stress compensation device is activated, using piezoelectric ceramics to actively offset the vibration energy, and notifying the operation and maintenance personnel to intervene urgently.

[0052] In particular, when the third level instruction is reached, the self-healing function of the sealed pole can also be set and activated.

[0053] In the mechanical stress compensation link, the piezoelectric ceramic array built into the support structure generates anti-phase mechanical waves according to the vibration spectrum characteristics. The control system analyzes the main frequency component of the vibration in real time, generates a drive signal synchronously through a digital phase-locked loop, and outputs it to the piezoelectric ceramic through a high-voltage amplifier. The amplitude adjustment accuracy reaches 0.1μm, effectively attenuating the vibration energy by more than 60%. When the CDI breaks through the CWT, the system immediately activates multi-level interlocking protection: first, a level 4 alarm signal is sent to the SCADA system, and the standby cooling fan is synchronously started to reduce the temperature by 10°C, and an axial preload is applied to the pole through a hydraulic mechanism to relieve mechanical deformation.

[0054] The self-healing function is achieved through microcapsules embedded in epoxy resin. The capsule shell is made of polyurethane material and encapsulates a silicon-based repair fluid containing silver nanowires. When the local discharge energy entropy exceeds the threshold for 5 consecutive minutes, the high-frequency electric field triggers the capsule to rupture. Driven by the electric field, the repair fluid migrates to the discharge area in a directional manner and bridges the insulation defects through silver nanowires. Actual measurements show that the local discharge can be reduced to less than 15% of the initial value within 30 seconds. Operation and maintenance personnel can view real-time three-dimensional status maps through AR glasses. The system automatically marks high-risk areas and provides maintenance priority recommendations, reducing the average troubleshooting time to 40% of traditional methods.

[0055] The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope of the present invention. The technical content claimed by the present invention has been fully recorded in the claims.

Claims

1. A method for optimizing the state of a sealed pole for a primary-secondary fusion circuit breaker, characterized in that: include: Obtain multi-physics field coupling monitoring data inside the sealed pole, the electric field intensity distribution data on the pole surface and inside, and capture the spatiotemporal characteristics of the partial discharge signal of the sealed pole; The multi-physical field coupling monitoring data includes temperature gradient, mechanical vibration frequency and strain distribution, the multi-physical field coupling monitoring data is analyzed to extract a first feature data set, and a comprehensive degradation index is constructed by weighted fusion of the first feature data set to obtain a prediction result; Based on historical data and the prediction results, a long short-term memory network is used to predict the degradation trend in the future period, a dynamic warning threshold and a critical threshold are generated, and an action instruction is generated through the dynamic warning threshold and the critical threshold; The analyzing the multi-physics field coupling monitoring data to extract a first feature data set comprises: The temperature gradient signal is subjected to empirical mode decomposition to extract the intrinsic mode function, the short-time Fourier transform energy spectrum entropy of the mechanical vibration frequency signal is calculated, the covariance matrix of the strain distribution is constructed and the maximum eigenvalue is extracted to form a multi-dimensional time-frequency domain fusion feature vector; it also includes a degradation state visualization reconstruction step, mapping the multi-physical field coupling monitoring data to the three-dimensional digital twin model of the sealed pole, and displaying the mechanical strain concentration area, the high-risk area of ​​electric field distortion and the spatiotemporal evolution trajectory of the local discharge pulse through the superposition of thermal maps, and marking the dynamic threshold exceeding area.

2. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 1 is characterized in that: The method of capturing the spatiotemporal characteristics of the partial discharge signal of the sealed pole includes: using an improved matching pursuit algorithm to sparsely decompose the ultrasonic signal, the basis function library includes Gaussian modulated sine waves and impulse response waveforms, and locating the spatial coordinates of the discharge source and the energy diffusion path through time delay-Doppler joint parameter estimation.

3. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 2 is characterized in that: The method is implemented based on the edge-cloud collaborative computing framework. The edge node performs data preprocessing and feature extraction, and the cloud platform performs iterative training of the degradation model. The compressed degradation index sequence is transmitted between the two through a feature-level differential privacy encryption channel.

4. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 3 is characterized in that: The calculation formula of the comprehensive degradation index is: ,in Indicates the weight of the heat accumulation index, reflecting the effect of temperature on the aging of the pole represents the heat accumulation index, represents the vibration offset, Represents the weight of the vibration offset, characterizing the contribution of mechanical stress to performance degradation, represents the electric field distortion rate, Represents the weight of the electric field distortion rate, which measures the impact of abnormal electric field distribution on insulation performance. is the partial discharge energy entropy, The weight of partial discharge energy entropy quantifies the association between discharge pattern complexity and insulation defects; Heat accumulation index Reflects the heat accumulation effect of the pole in long-term operation, and is used to predict the thermal aging of insulation materials and partial discharge energy entropy The higher the entropy value, the more complex the discharge pattern and the greater the risk of insulation defects.

5. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 4, characterized in that: The weight coefficient , , , Through dynamic adjustment using an improved particle swarm optimization algorithm, its fitness function integrates the KL divergence of historical fault samples and the environmental correction factor of real-time working conditions. The weight adjustment cycle is an integer multiple of the preset time window and meets the constraints. .

6. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 5, characterized in that: It also includes a fault tracing analysis module. When a critical threshold exceeding event is detected, it automatically associates historical multi-physical field data to generate a fault evolution tree diagram, and calculates the contribution weight of each physical field parameter through a causal reasoning engine, outputs the dominant failure factor and associated confidence, and the contribution weight is calculated based on the Shapley value algorithm.

7. The state optimization method of the sealed pole for the primary and secondary fusion circuit breaker according to claim 6, characterized in that: The action instructions include a hierarchical response strategy: When the first level command is triggered, the pole is in a healthy state and only records data; When the secondary instruction is triggered, the partial discharge suppression module is activated to reduce the electric field distortion by adjusting the potential distribution on the pole surface; When the third-level command is triggered, the mechanical stress compensation device is activated, using piezoelectric ceramics to actively offset vibration energy and notify the operation and maintenance personnel to intervene urgently; The level of the action instruction is determined by the warning threshold and the critical threshold.

Images