The detection of composite insulator defects in substations still relies on inspection by operators，which is a heavy workload and prone to leakage due to visual fatigue.To reduce the computational resource consumption and shorten the training time，the region convolutional neural networks（RCNN）was improved by reorganizing the convolution kernel，and a detection method was proposed for insulator crack shape features. The method can meet the premise of insufficient training sample data，but also can get better convolutional neural networks （CNN）training effect，and finally achieve accurate crack recognition. In the training phase，the RGB three-channel decomposition method was used to expand the training data set，the median filtering method was used to remove the noise，the improved convolutional kernel was used to train the CNN. In the test phase，the images were decomposed by RGB three-channel decomposition and input to CNN to get the exact crack center coordinates and length. The non-maximum suppression（NMS）algorithm was used to de-weight the images to get the final crack recognition results. The example analysis shows that the propose method can still achieve good recognition accuracy， and accurately identify the specific location of cracks under the premise of insufficient training samples.

During the operation of high-voltage DC power grid，due to factors such as load characteristics and voltage fluctuations，the instantaneous power characteristics such as capacitor current and harmonics are more complex，which increases the difficulty of identifying faults in the power grid system. Therefore，a bipolar short circuit fault identification technology for high-voltage DC power grids considering instantaneous power characteristics was proposed. By establishing a topology model of the high-voltage DC power grid，the bipolar short circuit fault of the high-voltage DC power grid was anlyzed. Based on the instantaneous power charac-teristics of the endpoints of the high-voltage DC power grid，the area where the bipolar short circuit fault occurs was determined. Using the maximum value positioning method of dual terminal fault current to obtain bipolar short circuit fault location，the fault identification of bipolar short circuit in high-voltage DC power grid was completed. The experimental results show that the proposed method has high accuracy in locating and identifying bipolar short circuit faults in high-voltage DC power grids，and the efficiency of fault identification is improved.

In order to ensure the minimum carbon emission and optimal power distribution of the hybrid energy storage microgrid under the constraint of carbon footprint，a distributed coordinated control algorithm of the hybrid energy storage microgrid under the constraint of carbon footprint was proposed. Based on the whole life carbon footprint of mixed energy，the distributed coordinated control objective function of hybrid energy storage microgrid with minimum carbon emissions and optimal constant volume of mixed energy was constructed，and the constraint conditions were determined. On this basis，combining with the uncertain characteristics of hybrid energy storage microgrid，the objective function was rewritten to form a two-stage brodding optimization and coordination control model. Adopting column and constraint generation algorithm to solve the model，obtain the optimal solution of the objective function.The test results show that，after the application of this method，the carbon footprint coefficient is lower than 9.0，the power distribution result of ultracapacitor is about 4.8 MW，the maximum value of network loss power and maximum voltage deviation is 0.42 MW⋅h and 0.067 V respectively. The indirect and direct carbon emissions are significantly reduced，and the charged state of the hybrid energy storage system is effectively improved.

With the increased penetration rate of new energy year by year，it is difficult to accurately predict the randomness and fluctuation characteristics of its output，causing a severe challenge to the operation，planning and scheduling of electrical power system. Therefore，modeling for the uncertainty of new energy has attracted more and more attention. To obtain the time sequence characteristics of new energy output scenario more effectively，a new energy scenario generation method was proposed based on data drive，and combined self-attention mechanism with generative adversarial network discriminator with gradient penalty through applying the SA/WGAN model. Through building a deep learning model based on the combination of two models，effectively highlight the timing sequence characteristics of new energy output scenario and enhancing the nonlinear fitting capability in scenario generation. The example results show that，compared with the scenario generation results of original WGAN and WGAN-LSTM，the new energy generation scenario of proposed model can not only effectively improve the accuracy，but also possess the advantages of stable WGAN-GP training results and quick SA calculation speed，which can achieve a more efficient generation of scenarios that is close to the distribution of real new energy scenario.

Harmonic loss is one of the main causes of safety accidents in excitation transformers. If the loss value is too large，it can easily affect normal substation and transmission. Therefore，a method based on wavelet packet decomposition and reconstruction was proposed to calculate the additional loss of rectification harmonics in excitation transformers. By obtaining wavelet packet decomposition and reconstruction coefficients，the additional losses of two types of harmonics in excitation transformers were analyzed. The open circuit and short circuit experiments were used to obtain the equivalent circuit parameters of the excitation transformer，and the wavelet packet decomposition and reconstruction algorithm was used to calculate the resistance and reactance under different harmonic frequencies，and compare them with the reference value to obtain the additional loss value of the rectification harmonic of the excitation transformer. Finally，a certain type of excitation transformer was selected and used the proposed method to calculate its additional loss value under different harmonic frequencies. The results show that the calculated loss value is very close to the actual results，verifying the high practical value of the proposed method.

Making full use of the regulation of flexible resources can effectively suppress the fluctuation of renewable energy generation and ensure the economic and safe operation of power system. An optimal dispatching method for power system considering the flexibility of battery energy storage systems （BESS）and pumped hydro energy storage （PHES）was proposed. Based on the establishment of the flexible ramp-up （FRU）and flexible ramp-down （FRD）requirement model，a mixed-integer linear problem （MILP）based temporal day-ahead security-constrained unit commitment and re-dispatch with the objective of minimizing the cost of energy，reserve，and flexible ramp products （FRP）was formulated for the studies，which was solved using CPLEX solver. The IEEE-RTS-24 node system was taken as an example and the proposed model was studied on 4 cases to verify that the proposed method can effectively improve the operational flexibility of power system and reduce operating cost.

To adapt to the construction of a novel power system with new energy as the main body，it is necessary to continuously improve the new energy consumption capacity of various voltage-level power grids.Through the research on the influencing factors and improvement methods of new energy consumption capacity of distribution network，a new energy capacity evaluation method of a regional distribution network based on multi-factor limitation was proposed，which analyzed the consumption capacity from the high-voltage power grid，medium-voltage feeder and low-voltage platform area，respectively. In addition，an optimal method of improving the investment efficiency based on unit new energy was also proposed. Through analysis，it is found that medium-voltage lines are mainly affected by voltage crossing and thermal stability constraints. In contrast，high-voltage power grids are mainly affected by main transformer capacity and higher-level transmission lines. Finally，a practical calculation tool was formed through the constraints and calculation models of new energy consumption.Combined with the countermeasures and methods of different factors，the study has important reference significance for China's novel power construction.

In order to improve the stability and economy of photovoltaic-storage combined power station，a capacity optimization method of photovoltaic-storage system based on the whole life cycle was proposed. The basic model of the station was analyzed. The capacity optimization model of energy storage system was established based on whole life cycle theory. And the revenue and expenditure of the system were considered. The actual case was simulated and analyzed by the three schemes. It can be obtained that the optimal method of system capacity with the maximum net present value as the goal has high economy. The net present value is 809 thousand yuan and 738 thousand yuan higher than the method of taking power quality as the goal and not allocating energy storage，respectively. If the capacity of the energy storage system exceeds 3 MW，it will not give full play to the maximum benefit，it will lead to the saturation of system power sales revenue and assessment cost，then the battery loss cost is increased. The method of optimizing the system capacity by taking the maximum net present value as the goal can reduce assessment cost and battery loss cost，it can also increase the system power sales revenue and maximum net present value，and improve power quality.

After the distributed wind farm stations are integrated into the distribution network，due to the characteristics of wind farm short-circuit current amplitude control and frequency offset，the traditional three-stage current protection of distribution network is difficult to accurately identify the fault on the wind power transmission line. Based on the similarity of fault current traveling waves，a fast protection method for wind power transmission lines in distribution networks was proposed to ensure fast and safe operation of distribution networks. First，the differences in the supply fault currents at the two ends of the transmission line were analyzed. A similarity pilot protection approach was proposed by exploiting the difference in the current travelling waveforms during the initial phase of the fault. The wavelet mode maximum was used for the start of protection. With the improved longest common subsequence algorithm，the waveform similarity at both ends was measured for fault discrimination.Finally，the simulation results demonstrate the excellent speed and reliability of the proposed method.

Microgrid is an effective method to integrate a large number of distributed generators into the power grid. Aiming at the optimal dispatch problem of microgrid，an optimal dispatch method based on proximal policy optimization algorithm was proposed.Firstly，the optimal diapatch model of microgrid was constructed by considering the operation cost of microgrid and operation constraints of various equipment. Secondly，the problem was formulated as a reinforcement learning framework，and the elements of reinforcement learning such as state，action and reward function were designed. Finally，the solution flow based on the proximal policy optimization algorithm was designed，and the effectiveness of the proposed method was verified by simulation.