The cogeneration virtual power plant with hydrogen storage provides a new path for hydrogen storage to participate in grid scheduling and consume new energy by combining the cogeneration characteristics of the aggregated resources. First, the cogeneration characteristics of hydrogen storage energy are fully considered, and the day-ahead scheduling optimization model that minimizes the comprehensive operating cost of the virtual power plant is established to make decisions on the day-ahead output plan of the aggregated resources under the condition of meeting the heat and electricity load demand.Then,the day-ahead hydrogen demand forecast model is established by considering the load-side hydrogen demand, including the hydrogen fuel demand of hydrogen buses and hydrogen cars. Next, the hydrogen demand is embedded as a constraintinto the virtual power plant day-ahead scheduling optimization model to obtain the optimal operation strategy of the virtual power plant under the consideration of hydrogen load demand. Finally,a simulation analysis in a park is conducted, and the results show that the operating cost of the optimal scheduling strategy of the virtual power plant considering the revenue from hydrogen sales is reduced by 24.6% compared with the traditional virtual power plant scheduling strategy only considering the characteristics of hydrogen storage cogeneration.

To solve the problem that the unreasonable layout of distributed photovoltaic(PV) will cause irreversible impact on the distribution network, a distributed PV optimal allocation model is proposed which considers the uncertainty of PV output. Meanwhile, the optimal access location and access capacity of distributed PV are solved by taking the maximum access capacity and the lowest network loss as the objective function. The improved sparrow search algorithm is used to solve the model,which has excellent global and local search ability and good individual update mechanism, and can find the optimal solution more quickly and accurately for such models. The simulation results of the model using IEEE-33 node distribution network show that full connection of distributed PV can be achieved through using the improved sparrow search algorithm to solve the model, network losses of the distribution network can be reduced, the power quality of the distribution network can be improved,and optimized allocation of distributed PV resources can be achieved.

Aiming at the problems caused by the large-scale integration of distributed power sources into the distribution network in the form of microgrids,such as the increase of system operation uncertainty, the frequent action of protection devices,and the complex and changeable network topology,a reliability assessment method for active distribution network considering the unknown network topology is proposed. Firstly, a reliability assessment model for distribution system containing microgrids is constructed, and the reliability index of distribution network with unknown topology is modeled by considering the relationship between active distribution network reliability and network topology. Next,a reliability assessment method for distribution network based on the sequential Monte Carlo simulation methodis is proposed,which is applied to the calculation of different reliability indices. Finally, an active distribution grid with microgrid access in a certain region is used as the content for calculation,and the reliability indicators are not signicantly different from the actual indicators, which indicates that the proposed method can effectively assess the reliability of active distribution networks with unknown topology, and the selection of microgrid system parameters can be guided according to the pre-determined reliability indices.

Under the dual-carbon background, as the focus of new energy consumption,microgrid is deeply affected by the uncertainty of source load. In order to solve this problem and reduce the carbon emissions generated during system operation,this paper proposes a two-stage robust optimization scheduling method for microgrid consideringthe stepped carbon trading mechanism,which improves the system's ability to cope with the fluctuations of new energy output, and ensures the economic operation of the system based on the reduction of carbon emissions. Firstly, the microgrid micro-source mathematical model and the laddered carbon trading mechanism are constructed,and the two-stage robust optimization model is established. The KKT condition is used to reduce the subproblem to a mixed-integer programming problem,and the nonlinear constraints are linearized by the large M method, and interative solution is performed using coloum-and-constraint generation. Finally,the validity of the proposed optimal scheduling model is verified by the arithmetic example,and the effects of the stepped carbon trading mechanism on the economic cost of the system and the carbon emissions are analyzed.

In order to improve the power system line carrying capacity and power quality, a power supply quality improvement scheduling method and regulation strategy based on distributed energy storage is proposed. Firstly, a distributed energy storage scheduling scheme is established to realize the change from passive assistance to active support of energy storage equipment in improving the carrying capacity of the power system. Then, the centralized state regulation strategy and decentralized state control method of distributed energy storage are studied,and equipment field application experiments are conducted. The results show that under centralized regulation, the equipment can effectively prevent the heavy overload of the line and the back transmission of power,and can reduce the peak valley difference of the substation area by 37.6%. Under decentralized governance,the problem of insufficient local power supply capacity can be effectively addressed. The distributed energy storage regulation strategy can serve as a flexible scheduling resource for the new power system, assisting in its stable operation.

To address the issues of the artificial jellyfish search (JS) algorithm for permanent magnet synchronous motor (PMSM), such as low parameter identification accuracy, slow multi-parameter identification, and to falling into local optimum easily, an improved artificial Jellyfish search algorithm is proposed. Firstly, Tent map and opposition-based learning strategy are designed to enhance the ability of jellyfish groups to approach optimal positions. Secondly, The jellyfish swarm motion and the ocean current motion of nonlinear decreasing time control function and bance algorithm are designed. Finally, to overcome the problem that the JS algorithm is prone to fall into local optimum leading accuracy degradation, Gaussian mutation is designed to help JS algorithm jump out of local optimum. The experimental results show the proposed algorithm has higher accuracy and faster convergence speed for PMSM parameter identification, and the identification accuracy can reach 99.15%.

In order to study the heat transfer characteristics of swirl burners under different air volume ratios, a 600 MW supercritical front and rear hedgeboiler in a power plant is simulated and calculated based on CFD technology, so as to obtain the cold state power field,hot state temperature field and component distribution under different air volumes, and the obtained data are processed and analyzed. The results show that when the central wind speed is less than 6 m/s, the central wind rigidity is poor, the average flow field velocity is low, and it cannot resist the return of high temperature flue gas. When the central wind speed is more than 8 m/s, it is easy to cause turbulence in the backflow zone, affecting the uniformity and stability of the power field. If the center wind increases 5 m/s the ignition point position will be postponed by about 6%, while the maximum temperature of coal powder ignition increases by about 34%. This is because the rigidity of the central wind is constantly increasing,thereby guiding the flue gas recirculation area to converge and form a high-speed airflow field, and quickly igniting the air-powder mixture to achieve full combustion.

Aiming at the current problem of insufficient intelligence for skid-mounted mobile substations, a comprehensive solution based on the internet of things (IoT), big data analytics, and artificial intelligence(AI) is proposed. By introducing automation control and remote monitoring technologies, the substation’s capabilities in data acquisition, fault prediction, and energy management are enhanced. Experimental results show that the intelligent system significantly improves power supply reliability and operation and maintenanceefficiency while reducing costs. The intelligent strategies not only increase the adaptability of substations but also provide technical support for the construction of future smart grids.

In order to reduce the risk of composite insulator defects on the safe operation of the power grid, the ultrasonic detection technology is carried out for the composite insulator defects with a radius of curvature of 13 mm, and the ultrasonic characteristics of the three defects are swept and analyzed through the special 3 mm single crystal, 5 MHz straight probe and sheath three different types of composite insulators simulating uneven thickness,porosity, and debonding defects. The results demonstrate that the method can effectively detect the defects of uneven thickness, porosity and debonding, in which the measurement accuracy of sheath thickness is 0.3 mm, the detection sensitivity of porosity defects is 2 mm in diameter, and the positioning accuracy is 0.4 mm and the detection of debonding defectsis characterized by the disappearance of the reflective wave at the interface of the adhesive. The detection method is highly efficient and can enhance the flexibility and practicality of on-site inspections.

The use of voltage feed-forward control under a weak grid will increase the negative phase shift of the output impedance of the grid-connected converter at the cross-cutting frequency,which may lead to the harmonic resonance of the grid-connected current. Aiming at the problem a resonance suppression strategy for the LCL-type grid-connected converter based on the three-frequency-band impedance remodeling is proposed. The strategy establishes a three-frequency-band equivalent model of output impedance on the basis of grid voltage feed-forward control, and uses the three-frequency band reshaping gain factor to reshape the output impedance of the grid-connected converter in different frequency bands, which reduces the negative phase shift of the output impedance in the mid-frequency band,and improves the stable operation capability of the grid-connected converter under a weak power grid.Simulation results show that the strategy retains the ability of traditional voltage feed-forward control to suppress grid background harmonics, improves the phase margin of the grid-connected system, and effectively suppresses the resonance phenomenon of the grid-connected current.

Automatically and accurately extracting power towers from airborne laser point cloud data in power distribution scenes is a critical step in routine power inspections. Due to the undulating terrainand dense vegetationin mountainous areas,the features are difficult to distinguish, and it remains challenging to extract targets from mountainous power scenes using existing methods. To solve this problem, a semantic segmentation method based on self-attention perception is proposed,which mainly includes color feature extraction,local position encoding,and multi-layer attention perception modules. When processing large-scale point clouds of distribution network scenes, this method can integrate RGB features as attached attributes of the point cloud's spatial features,thereby improving the accuracy of semantic segmentation. Experimental results show that the proposed method has significant performance advantages in power scenarios, achieving a point cloud segmentation accuracy of over 90%, and its effectiveness has been validated on the public S3DIS dataset.

A multi-modal image component localization method for distribution network inspection is proposed for the task of thermal fault discrimination of distribution network components. The method includes two key steps that is multi-modal image information coordination and object detection. Firstly, a self-adaptive image registration method is proposed to address the issue of misalignment between high-resolution visible light images and infrared images with temperature information,which can complete high-quality cross modal registration tasks. Secondly,the prediction information of the registered visible image is transferred to the infrared image by the prediction information transfer method to complete the detection of the infrared image. The results show that compared with direct detection of infrared images,the proposed method can improve detection accuracy by 18.4% and performs extremely well in terms of precision and recall.

Aiming at the issue of external damage to transmission lines caused by the construction of cranes and other machinery, a transmission line early warning and interlocking control system based on radio frequency radar is proposed to construct a three-dimensional multi-level warning zone under the transmission lines. Through calculations of target RCS value, distance, speed,longitude and latitude, spatial segmentation, clustering, dynamic and static separation, tracking and other filtering and recognition processing are carried out, which realizes the state monitoring,tracking and early warning of ultra-high machinery under transmission lines. The developed crane is equipped with autonomous linkage, monitoring and locking devices and has laser light curtain directional warning and voice prompt functions. On-site testing results show, the system can effectively prevent external damage to transmission lines and can provide early warning and active intervention for the intrusion hazards of ultra-high machinery such as cranes.

Aiming at the problem that existing research mainly focuses on the identification of electrical components in the secondary circuit of substations and neglects the determination of connection relationships,a method for jointly analyzing CAD electrical drawings using bitmap and vector graphics is proposed. Firstly, YOLOv8 object detection algorithm is used to identify electrical drawings in bitmap format,obtain the categories and positions of electrical components in the drawings, and then the vector graphic characteristics of CAD drawings are used to pick up the line segment information between connecting component symbols,generating a topology map with wires as nodes and wire connection points as edges. Finally,a depth first search method is used to traverse the topology map to determine the connection relationships between component symbols. This method combines the advantages of extracting component symbols from bitmap using object recognition methods and extracting line segments using vector graphics, resulting in high accuracy in electrical component recognition and accurate recognition of connection relationships.

In order to establish a comprehensive and scientific evaluation system for transmission and transformation projects, a comprehensive benefit evaluation method based on the improved TOPSIS-RSR method is proposed.Firstly, the Delphi method is used to filter the comprehensive benefit evaluation indexes of the transmission and transformation project,including economic benefit, carbon reduction benefit and social benefit.Secondly, the subjective and objective methods are combined with fuzzy analytical hierarchy process and entropy weight method to assign weights. Finally, the ultimate weights are obtained by applying the principle of minimum information.During the construction of the evaluation model, the rank-sum ratio (RSR) method is introduced into the technique for order preference by similarity to an ideal solution (TOPSIS), and the Euclidean distance in TOPSIS is replaced with standardized Euclidean distance, effectively avoiding evaluation discrepancies caused by different dimensions. The TOPSIS-RSR method before and after the improvementis is utilized to evaluate and rank 220 kV power transmission and transformation projects in four regions. By comparing the test values of the two models,the advancement of the improved model is verified.

Aiming at the deficiencies of unmanned arerial vehicle (UAV) inspection image denoising method based on the traditional wavelet threshold,an improved wavelet threshold denoising method is proposed. By improving the threshold selection method and threshold function, the denoising effect of UAV inspection image is more remarkable and suitable to solve the multi-resolution problem. By introducing regulating factors the improved threshold function avoids the discontinuity of hard threshold function which can lead to oscillations during image reconstruction and the fixed deviation of soft threshold function which can lead to loss of accuracy and distortion in image reconstruction. The improved threshold selection method avoids the problem that denoising is incomplete or the useful image information is lost too much caused by the traditional threshold selection method.

In order to improve the accuracy and efficiency of short-term power load forecasting, a load forecasting model named NRBO-TCN-GRU-Attention is proposed by combining the temporal convolutional network (TCN), gated recurrent unit (GRU) model, Newton-Raphson-based optimizer (NRBO) and the attention mechanism (attention). In this model, the NRBO is utilized to optimize the hyper-parameters. The TCN module extracts features from the load data and inputs the extracted features into the GRU module to capture the long-term dependencies in the load sequence. Next, the attention mechanism is utilized to reinforce the important features. Finally, the prediction results are output through the fully connected layer. The experimental results show that the proposed model outperforms other comparative models in four metrics on the two-day and one-week test sets, namely, coefficient of determination, meanabsolute error, meanabsolute percentage error and root-mean-square error, which validates the superiority and applicability of the proposed model.

Aiming at the problem that the output power of photovoltaic（PV） power is affected by a variety of factors, showing randomness and intermittency, which leads to the difficulty of accurately predicting PV output. A method of distributed PV output forecastmodel construction and optimization considering multiple real-time meteorological factors is proposed. Firstly, according to solar irradiance, temperature, wind speed, daily irradiation duration and relative humidity,the abnormal PV data return to by Newton interpolation, and the missing data are filled in to complete the preprocessing of PV data. On this basis,the characteristics of numerical weather prediction under uncertainties are quantitatively selected and the PV predictors are extracted based on fuzzy weighted convolutional neural network. Finally, a PV power output forecastmodel is constructed based on least square support vector regression, and the model is solved. The experimental results show that the research method has a very high precision for the prediction of distributed PV output.

To address the issue of insufficient accuracy and robustness in existing feature extraction methods due to the lack of large datasets in the intelligent design of power equipment, a WinCLIP+ method based on small-sample learning is proposed to enhance feature extraction performance under zero-sample and small-sample conditions. This method combines the advantages of both zero-sample and small-sample learning,utilizes a pre-trained CLIP model for multi-scale feature extraction,introducesa small number of design reference samples, and incorporatea reference correlation module alongside a multi-scale feature fusion mechanismto enhance the extraction capability of features of different types and scales. Furthermore, WinCLIP+ further enhances the robustness of feature extraction by integrating linguistically guided predictions and visual reference information. To verify the effectiveness of this approach,an anomaly detection task is conducted. The experimental results indicate that WinCLIP+ achieves a detection accuracy of 84% under zero-sample conditions and 91% under small-sample conditions, demonstrating its significant effectiveness and robustness for feature extraction in the intelligent design of power equipment.

Aiming at the problem that the transformer no-load closing vibration signals is nonlinear,non-stationary and transient, which leads to the difficulty of fault feature extraction, a method of transformer winding loosening fault diagnosis based on chaos theory is proposed. Firstly, the mutual information value method is used to determine the optimal delay time. Secondly, the pseudo-neighbour method is used to select the embedding dimension,and the phase space reconstruction of transformer no-load closing vibration signals under the normal and loose fault states of the winding is performed. The change rules in chaotic attractor of transformer winding vibration signals under different compression stateis are analyzed in detail. Finally, the relevant chaotic feature parameters, namely,the maximum Lyapunov exponent and correlation dimension,are extracted from the chaotic attractor,and the relationship between them and the compression state of transformer winding is studied. The research results show that the chaotic features of transformer no-load closing vibration signals can effectively reflect the transformer winding loosening fault and provide a new method for transformer winding loosening fault diagnosis.