In order to meet the requirements of the scenario of power supply and power protection in the power system, and improve the emergency response capability and power supply reliability of the system, this paper applies a number of key technologies and methods such as compact weight reduction of transformer, shock absorption and isolation of equipment, verification of mechanical impact resistance, anti-corrosion of cabin roof, rapid layout of lightning protection device, modular design of secondary equipment and skid-mounted mobile substation digital twin. A large-capacity skid-mounted mobile intelligent substation with a main transformer capacity of 63 MV∙A and a voltage level of 110 kV is designed. Taking the large-capacity skid-mounted mobile intelligent substation that is actually applied in a new project of 110 kV temporary substation in Leshan City, Sichuan Province as an example, the effectiveness of this design is verified. The practice shows that this kind of power equipment has good promotion value and application prospects.

Aiming at the problem of the lack of historical data on arc faults in most photovoltaic power stations, this paper proposes a photovoltaic system series arc fault diagnosis method based on ultrasonic sensors and isolation forest after collecting arc ultrasonic signals and analyzing their characteristics. Firstly, arc ultrasonic signals are collected and their characteristics and advantages are analyzed. Secondly, the S-transform is used to convert the transient voltage signal of the ultrasonic wave during the occurrence of series arc faults to the time-frequency domain. Then, the Teager energy operator is used to amplify the spectral differences. Subsequently, the time-frequency entropy is used to extract the time-frequency domain features of arc faults. Finally, arc faults are diagnosed based on dynamic thresholds and isolation forest without the need for historical data. Experimental results show that the proposed method can accurately identify series arc faults, with a diagnosis accuracy rate of 97.25%, and has strong anti-interference ability.

The running state of lithium batteries has the problems of insufficient accuracy of ontology state estimation and high difficulty of battery pack fault state diagnosis. A joint state estimation method based on electrothermal coupling model considering the influence of multiple factors is proposed, and a multi-sensor fault diagnosis method based on detection window and correlation coefficient is designed. For the state of lithium battery, firstly, the electrothermal coupling model of lithium battery is constructed based on the equivalent circuit model method. Secondly, the mechanism of joint estimation of state of charge (SOC) and state of health (SOH) is analyzed. Using extended Kalman filter (EKF) and particle filter (PF), combined with online parameter identification method, an online joint estimation model covering the whole life cycle of lithium battery is constructed to achieve accurate joint state estimation. For the battery pack fault state, the multi-sensor fault diagnosis method based on detection window and correlation coefficient realizes the accurate diagnosis and location of short circuit and open circuit faults. The experimental results verify the effectiveness of the proposed method, which can accurately reflect the operation state of the battery, and has certain engineering application value.

Establishing the correlation between macrolevel indexes and power supply reliability is essential to analyzing the power supply reliability of distribution network on a large scale (provincial and municipal). However, the existing methods rely on the detailed distribution network topology, which is difficult to meet the requirements of regional power supply reliability analysis. This paper firstly establishes a macrolevel index system of power supply reliability for planning, construction, operation and management of distribution networks. Then a calculation method of power supply reliability is proposed by taking into account the automated isolation of distribution networks and pre-scheduled outages. On this foundation, the calculation method of power supply reliability based on the macrolevel index system is proposed by dividing the equivalence assessment sub-models and calculating the parameters of the sub-models. Finally, a case study is carried out to verify the proposed method. The results show that the relative deviation between the calculation results of the proposed method and the actual data is within 11%. The method proposed in this paper can analyze power supply reliability in the region by using only the macrolevel indexes, which provides a certain degree of accuracy and operability for the planning of distribution networks, reliability management and reliability target setting.

The clustering of distribution networks optimizes resource allocation and achieves load balancing through node partitioning. Existing clustering indicators primarily rely on modularity and power balance metrics, neglecting the impact of electric vehicle (EV) schedulable characteristics on distribution network flexibility. To address this, a new sub-indicator, which is bilateral EV schedulable capacity matching load demand and EV response, is defined, and a comprehensive indicator is constructed using a combined weighting method. Simulations based on the IEEE 33-node system are conducted with various indicator types, EV penetration levels, and time period scenarios. By comparatively analyzing the impacts of these factors on clustering results, the practicality and effectiveness of the proposed method are validated.

Based on the 2×660 MW units of a power plant in Ximeng area, this paper studies the problem of power system subsynchronous oscillation/resonance, and puts forward the joint suppression measure based on supplementary excitation damping control (SEDC) and generator terminal subsynchronous damping control (GTSDC). The mechanism of the two systems is verified by experiments, and the oscillation attenuation characteristics of the independent scheme and the joint scheme under different working conditions are studied. The results show that the joint suppression measure based on SEDC and GTSDC achieves efficient collaborative control through parameter matching and multi time scale coordination, and significantly improves the suppression effect and response speed. Two actual fault cases verify the effectiveness and reliability of the joint suppression measure. The research also found that the sig-nal-to-noise ratio suppression effect of the speed signal is very important, so the suggestions of optimizing the reliability of the speed measurement system are put forward. The reseatch of this paper shows that the joint suppression measure based on SEDC and GTSDC can effectively solve the problem of unit subsyn-chronous oscillation, improve the dynamic stability rate of power system, and provide a technical refer-ence for the treatment of subsynchronous oscillation of similar units.

Modern power systems may experience untypical forced wideband oscillations, and traditional methods are difficult to identify the source of such forced oscillations. This paper proposes a universal identification method for forced oscillation sources based on voltage oscillation ratio (VOR), using wide-frequency measurement data provided by the broadband measurement system. The VOR is the ratio of the oscillation voltage amplitude relative to the steady-state voltage amplitude, which can effectively reflect the physical characteristics that the forced oscillation source has the maximum relative amplitude of oscillation voltage. This method is suitable for identification of low-frequency oscillation and sub/super-synchronous oscillation. This method is applicable in different voltage levels and can reduce the impact of measurement errors of transformers and broadband measurement devices. The effectiveness of this proposed method is verified through simulation cases and on-site cases.

A small-signal stability preventive control method based on convolutional neural network (CNN) sensitivity analysis is presented in the paper, to improve the developing speed of small- signal stability preventive control measures. For poor or negative damping low frequency oscillation modes (i.e., the damping ratios are smaller than a threshold), first, an optimization model with small- signal stability constraints is established; second, the sensitivities of the damping ratios with respect to control variables (the active power of adjustable generators) based on CNN model of damping ratio prediction are calculated and then the optimization model is transformed into a quadratic programming model by linearizing small-signal stability constraints through sensitivities; finally, the adjustment amounts of generator active power are obtained. Several iterations are needed to make the damping ratios meet specific requirements. Analysis results of WEPRI 36-node case show that the effective control measures can be obtained by the presented method, which is more precise than that of the support vector machine method. The computing speed of the presented method is faster than that of the traditional eigenvalue analysis method. The ideas presented in this paper can also be applied to transient stability preventive control.

The current methods for predicting the state of health of lithium batteries often suffer from low accuracy. This paper introduces a method for state of health prediction using a seagull optimization algorithm optimized deep extreme learning machine. Key health feature parameters, such as constant voltage charging and discharging times during battery cycles, are selected and their correlation with the battery state of health is analyzed using Pearson correlation. The proposed model predicts subsequent state of health values by learning from samples. Experiments conducted with battery data compare the proposed method with single extreme learning machine, single deep extreme learning machine, and other literature. Evaluation metrics, including maximum absolute error and root mean square error, demonstrate that the seagull optimization algorithm optimized deep extreme learning machine model achieves higher accuracy and faster prediction times, with errors below 1.1%, indicating superior prediction accuracy and applicability.

The assignment of work orders is an essential part of the power operation and maintenance management system. Timely and accurate assignment methods can effectively improve the efficiency of work order circulation. Based on in-depth analysis of the characteristics of work order management in the field of power operation and maintenance, this article proposes an automatic assignment model for power operation and maintenance work orders based on fit degree. The model adopts criteria importance though intercrieria correlation (CRITIC) weighting method and fuzzy comprehensive evaluation to calculate the compatibility between operation and maintenance personnel and work orders, determine the optimal candidate for work order assignment, and then achieve automatic assignment of work orders. The practical application results show that compared to manual assignment, this automatic assignment model can effectively improve the efficiency and accuracy of work order assignment.