The increasing distance of offshore wind farms from coastal areas has created an urgent need for the development of long-term extra high voltage direct current (EHVDC) cables. Factory joints are commonly used to connect sections of submarine cables, forming extensive cable systems. Therefore, studying factory joint is crucial for advancing long-length cable lines. This study investigates the physicochemical and dielectric insulation characteristics of XLPE samples under various vulcanization pressures, highlighting the effects of these pressure changes on the properties of 500 kV EHVDC cross-linked polyethylene (XLPE) cable joints.

Commercially available 500 kV EHVDC XLPE pellets were used to prepare the XLPE samples via hot-press method. Initially, a specified quantity of XLPE pellets was distributed between two iron plates. The pellets were preheated at 120℃ for 5 minutes and then heated at 180℃. Cross-linking was subsequently performed under different vulcanization pressures of 1.3 MPa, 1.6 MPa, 1.9 MPa and 2.5 MPa respectively. The fabricated XLPE specimens underwent physical characterization through Fourier-transform infrared spectroscopy (FTIR), differential scanning calorimetry (DSC), X-ray diffraction (XRD), and gel content analysis. While electrical measurements included current density analysis, pulsed electro-acoustic (PEA) analysis, and DC breakdown test.

The physiochemical results indicate that increasing vulcanization pressure enhances the crosslinking degree of XLPE samples, transforming the material from a linear molecular structure to a 3D network structure and breaking macromolecules into smaller, mobile molecules. The increased mobility of these small molecules leads to improved crystallinity, resulting in a higher crystallinity structure. Additionally, the recrystallized macromolecular chains have higher melting temperatures, raising the overall melting temperature of the samples. However, higher vulcanization pressure also produces crosslinking by-products that are difficult to decompose and volatilize. The combination of high temperatures and pressures causes thermal expansion forces perpendicular to the lamellae, increasing lamella spacing, creating more amorphous regions, and effecting the insulation performance of the samples.

Regarding electric insulation performance, the DC breakdown strength and space charge injection threshold strength of the fabricated XLPE samples initially increase and then decrease with the increase in vulcanization pressure. Conversely, conductivity current and average space charge density first decrease and then increase. An optimal vulcanization pressure of 1.9 MPa was identified, at which the XLPE samples exhibited improved electrical insulation properties. Below this pressure, the increased trap energy levels inhibit carrier transport, thereby reducing the number of free carrier paths and hindering the formation of conductive channels, ultimately increasing the breakdown strength of the XLPE samples. However, at vulcanization pressures above 1.9 MPa, the increased crosslinking byproducts create more shallow traps, which lower space charge injection and accumulation, ultimately distorting the sample's internal electric field. Additionally, the increased lamella spacing creates more amorphous regions, reducing the carrier transport barrier and further decrease the breakdown strength of the prepared XLPE samples.

Based on the results, it can be concluded that appropriately increasing the vulcanization pressure of factory joints improves the physicochemical and electrical properties of XLPE. However, excessively high vulcanization pressure can have a detrimental impact on the electrical insulation properties of cable factory joints.

Metal oxide surge arresters are crucial for overvoltage protection in power systems, determining the insulation level of electrical equipment, with their core component being the ZnO varistor. However, modern stable ZnO varistors exhibit an anomalous decrease in power loss during aging, contradicting the increase in power loss predicted by the classical ion migration model. This discrepancy poses challenges for the condition assessment and life prediction of ZnO varistors due to a lack of theoretical foundations, thereby presenting a potential threat to the power system. Consequently, the study of the anomalous aging mechanism of stable ZnO varistors has been identified as a major challenge for the varistor community by CIGRE in both 2013 and 2017.

In this paper, stable ZnO varistors are subjected to accelerated DC aging at elevated aging temperatures to investigate their long-term stability transition. With increase in aging temperature, power loss trend transitions from a continuous decrease at 120℃ to an initial decrease followed by an increase at 150℃, and a sustained rise at 180℃. The decreasing power loss trend can be fitted by a double exponential decay function, while the increasing power loss is proportional to the square root of the aging time t^0.5. After transitioning to a mixed stable type at 150℃, the aging of stable ZnO varistors becomes irreversible. In-situ high-temperature dielectric measurements reveal that the interface space charge polarization relaxation process shifts to higher frequencies with decreased relaxation time and activation energy decreasing from 0.583 eV to 0.560 eV, indicating the destruction of the grain boundary structure. Low-temperature dielectric tests show that intrinsic point defects of zinc interstitials undergo irreversible consumption after aging. Upon transitioning to an instable type at 180℃, the "crossover" phenomenon of the forward current-voltage (I-U) characteristics disappears at 180℃, and both forward and reverse I-V characteristics shift towards increased leakage current region as a whole. Severe deterioration in reverse electrical parameters was observed, as breakdown voltage U_1mA decreases from 200.5 V to 92.9 V, the nonlinear coefficient α decreases from 16.3 to 2.0, and the leakage current rises from 19.5 μA to 479.3 μA. More importantly, offline physical and chemical structural tests show a reduction in the diffraction angles of ZnO crystal planes and decreased peak intensities. Additionally, a significant decrease in the binding energy of the Zn_2p orbital is observed, with Zn_2p3/2 and Zn_2p1/2 orbitals decreasing from 1 022.5 eV and 1 045.9 eV to 1 022.1 eV and 1 045.2 eV, respectively. This indicates the reduction of zinc interstitials and confirming that the interface states cannot maintain stability at high temperatures, thus becoming neutralized and consumed with zinc interstitials.

These findings demonstrate that the essence of the decreasing power loss in stable ZnO varistors lies in the stable interface states at the grain boundary, which, however, cannot maintain stable at certain high temperatures. The interface states would then neutralize with the zinc interstitials due ion migration, subsequently leading to the reduction of zinc interstitials and the destruction of the ZnO lattice, resulting in significant deterioration of ZnO varistors. Therefore, optimizing the high-temperature stability of the interface states is crucial for enhancing the long-term stability of ZnO varistors.

AC/DC hybrid system has become an effective solution for large-scale new energy consumption because of its characteristics of multiple power sources, multiple drop points, large capacity and cross-regional flexible transmission. However, the wide-frequency oscillation problem of AC/DC hybrid power system with high proportion of power electronics is prominent, which threatens the safe and stable operation of the system. At present, the stability analysis of hybrid system is mainly faced with the problem of how to take into account the AC and DC sections of the hybrid system at the same time and how to cover the system equipment with different impedance characteristics, so as to realize the unified analysis of the AC/DC hybrid network with multiple power electronic equipment. To address the above issues, this paper establishes a unified immittance network model for system-level stability assessment of hybrid systems.

First, based on the voltage/current source type device characteristics of the network devices, the impedance/ admittance forms of each network module of the system are standardized to avoid the problem of solving for the right half-plane poles in the process of system stability analysis. Secondly, from the perspective of the AC and DC ports of the system equipment, the AC/DC hybrid system can be divided into mono immittance subsystem and hybrid immittance subsystem. Finally, based on the interaction relationship between system immittance networks, a unified immittance network model containing complete oscillation information of the hybrid system is established. Under the premise of covering the stability information at each AC and DC port of the hybrid system, the unified immittance network model reduces the dimension of the system network model, at the same time expands the system network matrix from a single AC or DC system to an AC/DC system. Furthermore, combined with the derivation and change process of the immittance network, and through the modularization of the system expansion, this paper makes the immittance network applicable to the analysis of objects with different topologies,and extends the unified immittance network to general AC/DC hybrid systems. In addition,in contrast to the matrix model that merely aggregates a single impedance feature or a single admittance feature, the unified immittance network encompasses the AC/DC systems with various impedance characteristics, accomplishing the mutual unification of the new energy unit with admittance properties and the load network with impedance properties. This facilitates the precise establishment of the network mathematical model when confronted with the AC/DC system that concurrently incorporates equipment with current/voltage source characteristics, and realizes the unified coverage of equipment models with dissimilar impedance characteristics.

Based on the immittance network model, a unified immittance network stability criterion covering the interaction relationship of each AC/DC port of the hybrid system is derived. Combined with the expansion analysis of the immittance network, the criterion is extended to the general AC/DC hybrid system. The stability criterion comprehensively covers the stability problems of the AC/DC system, and avoids solving the problem of poles in the right half-plane, simplifying the analysis process. Furthermore, an oscillation traceability method for the hybrid system is given based on the immittance network model. Finally, based on the application examples of the AC/DC hybrid system, the immittance network model and its stability analysis method proposed in this paper are evaluated and verified.

Electric vehicles (EV) have the characteristics of both traffic and mobile load, and their charging behavior will have an interactive impact on the power grid. With the rapid increase in the number of electric vehicle and the continuous improvement of their penetration rate, charging guidance for large-scale EVs has become an important measure to alleviate the contradiction between local limited charging resources and strong charging demand. Therefore, considering the influence of future traffic information changes on navigation strategy, this paper proposes a fast guidance strategy for electric vehicle charging based on dynamic traffic inference.

First of all, a dynamic traffic information prediction model based on spatio-temporal self-supervised learning (ST-SSL) is established. A self-supervised learning (SSL) module for spatial and temporal heterogeneity of traffic data is designed to achieve accurate prediction of multi-period traffic flow information. Secondly, a multi-time dynamic impedance modeling method for urban road network considering future traffic information changes is designed, a charging navigation strategy considering multi-demand scenarios and multi-navigation objectives of users is established, and a solution method based on dynamic Dijkstra algorithm is proposed to realize the selection of the optimal charging station and the planning of the optimal navigation path. Based on the global charging navigation results, the service range of urban charging stations is dynamically evaluated, to achieve rapid charging guidance for electric vehicles. Finally, taking the actual road network of a certain area in Los Angeles as an example, the accuracy of the prediction model and the effectiveness of the guidance strategy are proved, which can effectively perceive the dynamic traffic information and quickly realize the service range division of urban charging stations and the charging guidance for electric vehicles.

In this paper, a fast guidance strategy for electric vehicle charging based on dynamic traffic inference is proposed, based on the case simulation results, the main conclusions can be obtained as follows. (1) The model based on ST-SSL can make full use of the spatial and temporal heterogeneity of traffic data, improve the prediction effect of traffic flow information, and provide an effective data basis for the construction of dynamic traffic impedance. (2) The proposed multi-scenario and multi-objective charging navigation strategy based on dynamic impedance can effectively perceive traffic information and take into account the diversified needs of users, effectively reduce the cost of charging navigation for different users, and reasonably guide the load distribution of electric vehicles. (3) The proposed dynamic Dijkstra algorithm can recommend the optimal path according to the future traffic information, which can be used as a navigation algorithm to plan the driving path, and can also recommend the customized optimal charging station according to the needs of users. (4) The division of charging station service range based on the global charging navigation results can effectively evaluate the service range of charging station, and provide an important reference for the construction planning of charging station. Based on the evaluation results, the charging navigation strategy is quickly assigned to each node, which effectively reduces the computing resource consumption of charging navigation.

With global climate warming, the frequency of heavy rainstorm events is increasing. During heavy rainstorms, composite suspension insulator strings are prone to reduced creepage distance utilization and potential rain flashover accidents due to the sudden increase in rain intensity and rain columns bridging the skirt gaps. Since the rod diameter and skirt diameter of composite suspension insulators are smaller than those of large-diameter composite pin insulators, and there are significant differences in skirt structure and creepage distance, the rain flashover characteristics of suspension insulators cannot be equated with pin insulators. Hence, there is insufficient data on the rain flashover characteristics of composite suspension insulators.

This paper conducts artificial rain flashover tests under strong rainfall conditions using five different skirt structures of composite insulators. It investigates the effects of rain intensity and water conductivity on insulator AC flashover characteristics by analyzing the rain column length at the skirt edge, simulation models, arc development paths, and critical leakage currents. A formula for calculating the unit insulator height AC flashover voltage under the combined effects of rain intensity and water conductivity is proposed and verified for accuracy.

Results indicate that the unit insulator height rain flashover voltage is negatively correlated with rain intensity and water conductivity, following an exponential decay function. The voltage gradient reduction can reach 41.9% and 43.8% respectively due to these factors. The rainfall intensity of 9 mm/min can be considered the tipping point between non-extreme and extreme rainfall. Additionally, composite insulators with larger skirt diameters and skirt spacings are more significantly affected by rain intensity, while the impact of water conductivity is consistent across different skirt parameters. When the rainfall intensity is low, the shorter rain column causes the arc to develop along the path of “rain column-air gap-small skirt surface”. When the rainfall intensity is high, the rain column is longer, and the vertical air gap within the rain column is directly broken down, leading to the arc development path of “rain column-vertical air gap-large skirt surface”.

In addition, the maximum length of the rain column at the skirt increases with rain intensity and varies with skirt diameter and spacing. The arc development path differs significantly among insulators with different skirt parameters. Insulators with larger skirt diameters and spacings exhibit a combined air gap and skirt surface discharge path, with reduced rain column bridging the skirt gap due to effective skirt coverage, resulting in higher creepage utilization. Therefore, insulators with larger skirt diameters and spacings can effectively enhance the insulation strength. Due to the influence of the number of charged particles in the water and the area covered by the water film, the average critical leakage current increases with rain intensity and water conductivity, which increases the energy gained by the electric arc and promotes the development of local arcs and the formation of discharge channels. During heavy rainstorms, to reduce surface leakage current on composite insulators and prevent rainwater from bridging the skirts to form a “rain pillar-vertical air gap” continuous arc path, it is recommended to use composite insulators with larger inter-skirt spacing and skirt diameters in regions prone to frequent heavy rainfall.

The purpose of the wind-solar complementary system (WSCS) is to couple wind power and photovoltaic (PV) in a complementary way to strengthen the ability to generate power continuously in the medium and long term. However, due to the uncertainty of natural resources, the power output of WSCS is still unstable. In recent years, hybrid energy storage systems (HESS) have been used to match the WSCS to reduce the volatility of system output, but there are still some problems leading to the system's economic cost making it difficult to control. For example, the coupling relationship between wind and solar is linear, and the premise of the fluctuation smoothing strategy is to meet the power demand of load-side or grid-connected. In order to solve the mentioned problems, this paper proposes a method that HESS smooths fluctuations of wind-solar coupling power considering multi-scenario planning. By constructing a nonlinear coupling relationship between wind and solar and optimizing the capacity allocation of power source-side hybrid energy storage, the system accommodation characteristics for power fluctuations are improved.

Firstly, the marginal distributions of the two power sources are constructed using KDE based on the historical data of wind power and PV, and the joint distribution is obtained by preferably using the Gumbel-Copula functions. The multi-scenario set obtained by random sampling of the joint distribution is able to reflect the intensity of the fluctuation changes. Secondly, the FFT and its IFFT are used to analyze the spectral analysis of the unstable power in scenarios set to determine the power borne by each energy storage unit. In this part, since the multi-scenario ensemble originates from a joint distribution, the correlation of each scenario in the ensemble is consistent, which means the cut-off frequency that distinguishes battery and super-capacitor does not change with the change of scenario. Finally, an optimization model is established with the objective function of minimum the cycle operating cost of HESS, and the capacity configuration of the HESS is calculated using an improved PSO. The result of capacity configuration provides room to accommodate fluctuations in power source-side output, which reduces the instability of the system.

The results of the simulation example show that the increase in frequency deviation before the HESS configuration is much larger than that after the HESS configuration. The change rate of RMSE for calculating the frequency deviation before and after the configuration of HESS ranges from 60.0% to 83.5%, the change rate gradually increases with the increase of the number of scenarios in the ensemble. This suggests that the role of HESS in regulating frequency increases as the number of scenarios increases. Meanwhile, with the increase in the number of scenarios in the set, the maximum growth in the rated power and rated capacity of the batteries is 77.1% and 54.9%, respectively. And that of the super-capacitors is 40.0% and 42.4%. However, this makes the increase in equipment cost of the super-capacitor more prominent. Further, the configuration of HESS makes the power fluctuation of the system at adjacent moments smoother. The fluctuation accommodation range of the power source-side within the time intervals of 10 h, 60 h, and 240 h is enhanced by 12.9%, 7.4%, and 6%, respectively. The amplification of the fluctuation accommodation range decreases with the longer of the time intervals. Nevertheless, the HESS still has rechargeable power characteristics when the power source-side output is zero.

From the simulation results, the following conclusions can be drawn: (1) Different numbers of scenarios in the set have consistent correlation, so the dividing frequency of the battery and the super-capacitor does not change with the number of scenarios, which makes the HESS can be effective for the fluctuating power to modulate frequency. (2) The more complex the frequency variations of the fluctuating power at the wind-solar coupling output, the more pronounced the capability of HESS modulating frequency. (3) The results of the HESS configuration reflect that super-capacitors and batteries have greater advantages in the rated power and rated capacity, respectively. (4) Even if the wind-solar coupling output is close to 0 or 0 after the configuration of HESS, the system is still able to ensure that there is a certain margin to counteract the fluctuating impact of sudden power changes.

Accurate prediction of the battery state of charge (SOC) is of great significance to improve the utilization efficiency and safety performance of the battery, and the monitoring of the battery state of charge is very important to help prevent overcharge and overdischarge accidents. The traditional SOC prediction methods are highly dependent on the mechanism model and statistical model, and have problems such as sensitive outliers and limited practical accuracy. In this study, a CNN-LSTM-AM (convolutional neural network - long short term memory neural network - attention mechanism) model is proposed to predict SOC variation trend through battery measurable variables.

The model first uses a one-dimensional convolutional neural network to extract spatial features of measurable variables, including battery current, voltage, temperature and average voltage, and then sends them to bidirectional long and short time memory for time series analysis. Finally, the attention mechanism is introduced to screen key features, reduce the redundancy of feature data, and improve the accuracy and generalization of the model. In addition, CNN-LSTM-AM model adopts rime optimization algorithm to optimize the hyperparameters in the training process, which effectively improves the training efficiency and reduces the training cost.

The actual evaluation on CALCE (Center for Advanced Life Cycle Engineering) data set of lithium iron phosphate shows that the attention mechanism can effectively improve the training performance of the prediction model, and the rime optimization algorithm adopted can help reduce the model hyperparameters, so as to obtain higher prediction accuracy. The performance of CNN-LSTM-AM model was tested under different temperature conditions, and both RMSE and MAE were less than 1%, which was sufficient to confirm the feasibility of the model to predict SOC. In addition, even if the initial SOC is uncertain, the proposed CNN-LSTM-AM model can still accurately track SOC trend changes, and the overall prediction accuracy reaches RMSE<1.5% and MAE<1.5%. The RMSE and MAE results of the network proposed in this study are smaller than those of CNN-LSTM and CNN-LSTM-AM. It shows strong robustness and generalization ability. Finally, in order to comprehensively compare the performance of different SOC prediction methods, the CNN-LSTM-AM model proposed in this study is compared with other experimental results. It can be seen that the method proposed in this study has significantly lower RMSE compared with AT-CNN-LSTM. At the same time, considering that the proposed method uses less training set data, we can also see the advantages of the designed network. Compared with EI-LSTM-CO(extended input-LSTM-constrained output), it can be found that the error is close. In addition, EI-LSTM-CO performs some post-processing on the predicted SOC, which can also reflect the superiority of the proposed method.

The following conclusions are drawn from the simulation analysis: (1) A CNN-LSTM-AM model is proposed and applied to the SOC prediction task of battery, which can effectively capture important input features and improve the prediction accuracy. (2) Design a rime optimization algorithm, which can automatically search the optimal solution of CNN-LSTM-AM model, effectively reduce the time cost of hyperparameter optimization. (3) The influence of different ambient temperatures and initial SOC values on the prediction accuracy of CNN-LSTM-AM was studied, and the performance of CNN-LSTM-AM was compared with that of traditional prediction models to verify its strong robustness and high generalization ability.

For bundled conductors, the shadowing effect of upwind sub-conductor will affect the airflow and droplets distributions of downwind one, resulting in the difference in icing characteristics. Traditional icing calculation process generally ignored these differences and, hence, only giving an identical icing mass result of each sub-conductor. This affects the study of the aerodynamic characteristics and deicing methods of icing bundled conductor. Although some scholars have pointed out that the shadowing effect between sub-conductors will influence the icing process, there is no quantitative study. Therefore, this paper further explores the shadowing effect and relevant influencing factors of bundled conductors through numerical simulation and test research. Furthermore, based on the analysis of the shadowing effect and the superposition principle, a rapid calculation method of ice mass accreted on the bundled conductor is proposed.

Firstly, the distributions of airflow and droplets around bundled conductor are solved by Eulerian-Eulerian two-phase flow model. Secondly, combined with the mass and thermodynamic balance equations, the icing mass and shape accreted on bundled conductor under various icing environments are obtained. Then a new parameter called shadowing coefficient is defined to investigate the shadowing effect and influencing factors as well. The results show that: Shadowing effect is weakened with increasing absolute value of shadowing angle and bundled-spacing, but intensified with the increase of median volume diameter (M_VD) of droplets; Meanwhile, the shadowing effect experiences a growth and then drops down along with the increase of wind speed, and reach to the max at 15 m/s range 5~20 m/s.

Based on the superposition principle and shadowing effect analysis, a rapid calculating method for ice mass on bundled conductor is proposed. Where iced bundled conductor is regarded as a linear combination of non-shadowed sub-conductor (single conductor) icing intensity and shadowing coefficient, so the icing intensity of various types of bundled conductor can be obtained only requiring the icing intensity of single conductor and the shadowing coefficient in the corresponding environment. Then the rapid icing calculation formulars of 3,4,6,8-bundled conductor under various shadowing angle is given by geometry analysis, respectively, which simplifies the calculation of the icing mass on bundled conductor.

Finaly, a 4-bundled conductor nature icing test was carried out at the Xuefeng Mountain Energy Equipment Safety National Observation and Research Station to validate the accuracy of the numerical simulation and rapid calculation method. Results show that under the environment parameters of ambit temperature T_f = -2℃, M_VD = 25.4 μm, liquid water content L_wc = 0.61 g/m³, wind speed V = 10 m/s and shadowing angle θ = 2°, the difference in icing intensity between rapid calculation and test results was within -4.01% to -19.77%, the icing thickness differences of sub-conductors were between 1.66% to -6.36% and the differences in shadowing coefficient were between 4.05% to 5.33%, which well verifies the accuracy of the rapid calculation method proposed in this paper.

Insulator is one of the most common and widely used electrical components in transmission lines, which plays a critical role in electrical insulation and mechanical support, ensuring that the current flows along the specified path and reducing electromagnetic interference with the surrounding environment. Since insulators are installed outdoors, they are exposed to wind, sunlight, rain, ice, frost and other bad weather for a long time, and their surface defects are inevitable. If the insulator appears self-explosion or drop string, which will cause leakage due to the loss of insulation, leading to electric shock accidents, thus resulting in huge economic losses. Relying on computer vision and deep learning technology, insulator defect detection from massive UAV aerial images has become an urgent problem for power operation and maintenance. However, the backgrounds of aerial images from overhead transmission line corridors are complex. Under different lighting conditions, shooting angles, shooting distances, etc., the scale of insulators in aerial images varies greatly, and the insulator strings are prone to occlusion, the defect area of the insulator is much smaller than the insulator itself. Therefore, there are numerous difficulties in detecting insulator defects in practical applications.

In recent years, compared with the traditional object detection methods, deep learning methods can quickly and accurately identify insulators and their defects from power inspection images. There is still a lack of comprehensive review of the latest progress in insulator defect detection in existing literature, without introducing object detection algorithms such as anchor free algorithm, YOLOv7, Transformer, and knowledge extraction techniques. In view of this, this article summarizes and analyzes a large number of visual methods for insulator defects detection, systematically reviews deep learning methods for insulator defect detection in drone aerial images, aiming to select appropriate detection methods for specific insulator defects and provide valuable reference for researchers engaged in transmission lines fault diagnosis.

This paper reviews the research progress of deep learning methods for insulator defect detection in UAV aerial images. Firstly, the current research status of transmission lines inspection based on deep learning is briefly reviewed. Then, the insulator defect detection methods based on deep learning are explained, mainly from the target detection models, lightweight network models, cascade detection models and other methods are summarized, which is conducive to the comparison between different deep learning methods and more helpful for power inspection personnel to select appropriate deep vision detection methods for fault diagnosis of specific electrical component. And the target detection models based on two-stage algorithms, one-stage algorithms and anchor-free algorithms are elucidated. The lightweight network models based on model pruning, knowledge distillation, low-rank decomposition, network quantization and the target detection model based on Transformer are summarized. Next, the self-built and public datasets for insulator defect detection are introduced. Due to the lack of training samples and unified dataset for insulator defect detection, scholars mostly conduct defect detection research through self-built datasets in different detection scenarios. Finally, the challenges faced by insulator defect detection methods based on deep learning are elucidated, including insufficient defect samples, low defect detection accuracy, difficulty in detecting small target defects, and feature extraction, etc. Based on existing deep learning techniques and recent research ideas, several important research directions in the future are pointed out, including expanding insulator defect samples, establishing unified performance evaluation indicators, small and zero sample learning, new defect detection frameworks, multi-level detection of small defects, deep fusion of multiple learning technologies, cloud-edge-end collaborative fusion, and improving network model stability and real-time performance.