To ensure the safety of new energy vehicles during the entire period of use, it is necessary to conduct health monitoring for the full life cycle of lithium-ion batteries. Aimed at the low learning rate due to the small capacity of training data set for the remaining useful life (RUL) prediction model based on neural network and the duplicate collinearity of the extreme learning machine(ELM) method, a method for augmenting the training data set is proposed. In addition, based on the improved ELM, an RUL prediction model for the full life cycle of lithium-ion battery is built. First, the early operation data of battery is extracted to formulate health factors, and the Akima interpolation method is used to augment the amount of training data. Then, the salp swarm algorithm is used to improve the ELM network, and the RUL prediction model for the full life cycle of lithium battery is established. Finally, the NASA battery data set is used to validate the model. Experimental results show that the proposed method for augmenting the training data capacity is effective, the capacity tracking capability of the RUL prediction model in full life cycle is strong, and the prediction error is small.

As power lithium-ion batteries play a key role in the electric vehicle industry, ensuring their working reliability has become a research hotspot at present. In this paper, the materials and manufacturing processes of lithium-ion batteries are reviewed. The battery state estimation and remaining useful life prediction methods are summarized in detail, and the advantages and disadvantages of these methods are discussed. From the perspective of battery management system, the relevant knowledges of equalization management system and thermal management system are sorted out in turn. From the perspective of electric vehicle hybrid energy storage system, the performance degradation mechanism under actual working conditions and the relevant technologies are elaborated upon. Finally, the status quo of key technologies related to the reliability of power lithium-ion batteries used in electric vehicles is summarized from four aspects, and the development possibilities in the future are forecasted.

In view of the high fluctuation and randomness of wind turbine output, which affects the safe and stable operation of power system as well as the accuracy of wind power prediction, a wind power prediction method based on the fluctuation characteristics of wind power is proposed. First, the fluctuation characteristics of wind power are analyzed in terms of time scale and unit scale, and the appropriate wind power data is selected for wind power prediction. Then, a wind turbine short-term power prediction model based on least squares-support vector machine (LS-SVM) is established. The adaptive variational mode decomposition (AVMD) is used to decompose the wind power data to achieve frequency division, and the improved particle swarm optimization(IPSO) is used to optimize the model parameters affecting the re-gression prediction in the LS-SVM model. Experimental results show that the prediction model has strong adaptability, and the effectiveness of the prediction method can be proved by prediction error evaluation indexes.

The influence of ash deposition on the output characteristics of photovoltaic (PV) modules and the deposition rule of pollution particles on the modules' surface are studied, which are helpful for formulating an ash removal scheme and improving the efficiency of photoelectric conversion. The PV array on the roof of a fan hall in North China Electric Power University is taken as the research object, and an artificial ash distribution experiment was carried out to explore the influence of different ash deposition amounts on the output power, current and voltage of modules. To determine the influence of one single factor on the deposition of pollution particles, a numerical model of particle deposition is established using COMSOL under the same conditions as those in the natural deposition test of PV modules, and the influences of wind speed, humidity, particle size and pollution concentration on the deposition of pollution particles on the surface of PV modules are simulated and analyzed. Test results show that ash deposition has little influence on the working voltage, but a great influence on the output power and working current. When the ash deposition density is 5.07 g/m², the changing rates of output power, current and voltage are 8.71%, 6.48% and 0.40%, respectively. Simulation results show that when other conditions are the same, the deposition amount of particles decreases first and then increases with the growing wind speed and particle size, and the minimum value is reached at the wind speed of 3 m/s and particle size of 15 µm. Under the same conditions, the particle deposition amount increases with the growing humidity and pollution concentration.

Compared with 4G communication, 5G communication uses key technologies such as large-scale antennas, ultra-dense networking, and high-frequency communication to greatly improve its performance, resulting in increased power consumption of single 5G base stations and multiple sites. As a result, the power supply requirements are also changed. At present, there are two kinds of power supply forms which can meet the new power supply requirements, i.e., near supply and HVDC remote supply. Aimed at the HVDC remote supply, the power supply architecture and key technology are sorted out, and a feasible operation mode is proposed by comprehensively considering the operating investment cost in the economic mode of peak-shaving and valley-filling, the busy and idle periods of 5G base station service load, and the peak-valley time-of-use electricity price. On the basis of the construction which has already been invested under the HVDC remote supply scheme, the investment cost is small, and the economic benefit is obtained by using the time-of-use electricity price difference between peak and valley periods, thus reducing the power consumption cost of base stations to a certain degree. Finally, the development of 5G base station power supply is forecasted, providing reference for the research or design of its power supply.

As a commonly used islanding detection method, the active detection method can reduce the non-detection zone by increasing the disturbance intensity, but it will also increase the degree of harmonic pollution to the grid at the same time. Aimed at this problem, a hybrid fuzzy low-harmonic islanding detection method without non-detection zone is proposed in this paper. First, the non-detection zone is judged, and the passive islanding detection method which has no harmonic pollution to the grid is used preferentially. If this zone is within the non-detection zone of the passive detection method, then a fuzzy low-harmonic islanding detection method with fuzzy disturbance control is used to disturb the output current in its second and fourth quarter periods. Finally, the corresponding simulation model is built to verify the effectiveness of the proposed method.

In this paper, a Boost converter model in the pseudo-continuous conduction mode(PCCM) of inductive current based on the definition of R-L fractional-order is studied. On this basis, the state-space averaging model of the converter is derived. Then, the expressions for the step-up ratio, DC static operating point, inductive current ripple, and output voltage ripple are derived. Results show that compared with the corresponding expressions derived on the basis of the definition of Caputo fractional-order, the derived expressions for DC static operating point and step-up ratio under the definition of R-L fractional-order are related to both the duty cycle and the orders of the capacitor and inductor. The expression for output voltage ripple is not only related to the order a of fractional-order inductor, but also related to the order ? of fractional-order capacitor. Therefore, the orders of inductor and capacitor obviously affect the DC component of state variables and the stable-state characteristics of factional-order PCCM Boost converter. Finally, a mathematical model and a circuit model of the R-L fractional-order PCCM Boost converter are built in MATLAB/SIMULINK, and simulation results show that the analysis result of the R-L fractional-order PCCM Boost converter model is more stable than that of the other PCCM Boost model, and the corresponding error is smaller.

Aimed at the problem of DC voltage fluctuations caused by load switching, power fluctuations and dou-ble-frequency injection in DC microgrids, a dynamic compensation control strategy for a Buck-type bidirectional DC-DC converter based on model predictive control (MPC) is proposed. First, the corresponding discrete state space matrix is es-tablished, and the input current is used as the disturbance. Second, the model-based inner-loop current predictive control and outer-loop voltage control of the Buck-type bidirectional DC-DC converter is designed. Third, a dynamic compensa-tion control structure based on a residual generator is designed for the current disturbance, and the dynamic compensa-tion controller Q(z) is solved. At the same time, the recursive least squares algorithm is used for parameter identification to reduce the influence of model uncertainty on the dynamic compensation control strategy. Finally, a comparative ex-periment was designed on the PSCAD/EMTDC simulation platform to verify the effectiveness of the proposed control strategy. Experimental results show that the compensation control structure can effectively solve the problem of DC bus voltage fluctuations and enhance the robustness of the entire system without changing the original predictive control.

Aimed at the problems of current spikes and voltage fluctuations in the charging process of existing multi-level inverters, a novel quasi-resonant five-level inverter is proposed in this paper. A carrier-stacked pulse width modulation scheme is adopted, and the switched capacitor technology is combined with the traditional five-level inverter, which reduces the accumulated level at the previous stage and realizes the output of five-level voltage. In addition, the capacitor charging process and the series inductance resonate to eliminate current spikes and voltage fluctuations, while reducing the heating problem of the capacitor and prolonging the service life of the circuit. Finally, simulation verification is carried out, and results prove the correctness and feasibility of the scheme for the proposed inverter structure.

A design method for small and medium-power DC/DC converters suitable for low-orbit commercial aerospace is given in this paper. A synchronous rectifier flyback topology and surface mounted devices are adopted, which can meet the demands for common single-and multi-output DC/DC converters applied in aerospace, with advantages of low cost, high performance, high reliability, and mass production. To further improve the conversion efficiency and power density, gallium nitride(GaN) FETs are also used. In addition, the losses of main power devices in the topology under different operating conditions are calculated and compared. Finally, a DC/DC converter with a wide range of input voltage (23-47 V) and output of 5 V and 30 W was built for verification.