Z. Zhang’s scientific contributions

Each arm of the modular multilevel converter (MMC) employs a large amount of sub-modules. Consequently, the possibility of an internal component failure is very high. After removal of the fault parts, the three-phase arms will be in the asymmetry running operation. Therefore, the ac current cannot distribute symmetrically between up and down arm of the faulty phase, causing direct current fluctuations. First, we derived the equivalent circuit of the MMC. Then, we analyzed ripple components of the arm current and direct current, and proposed a dc ripple current suppressing control strategy. The core idea is to introduce an additional component in the arm voltage reference which can offset the fault voltage. The feasibility of the suggested control strategy is verified by the PSCAD/EMTDC software. The simulation results show that(1) the dominant components of the dc ripple current are the fundamental and second harmonics; (2) by applying the proposed strategy, the dc ripple current can be suppressed less than 1% in the test case; (3) the capacitor voltage values can be adjusted dynamically to ensure that the sub-module switching calculation number is equal to the real required number. © 2016, High Voltage Engineering Editorial Department of CEPRI. All right reserved.

The traditional hybrid high voltage direct current (HVDC) system has three key issues, i.e., DC side fault self-clearance problem, the transmit power of the voltage source converter (VSC) is much smaller than that of the line commutated conveter (LCC), and DC low voltage ride-through problem caused by AC grid fault at the rectifier side. Thereby, we proposed a new LCC-C-MMC hybrid HVDC scheme, of which the inverter side consisted of series/parallel connected combined modular multilevel converters with improved clamp double sub-modules (C-MMCs). Then we researched the steady operation control characteristic of the proposed hybrid system, and analyzed the reason why the transmission power was interrupted by AC grid faults at the rectifier side. To solve DC low voltage ride-through problem, we proposed an improved control strategy with bypassing valve group, namely backup constant voltage control. Furthermore, we analyzed the equivalent circuit and arm charging characteristic during applying process of the bypass valve. The feasibility of the suggested control strategy was verified by the PSCAD/EMTDC software. The simulation results show that the induced third harmonic injection modulation and reactive power dynamic adjusting strategy can expand the steady operating area and transient operating range, therefore, the system achieves excellent operation characteristics under 10% AC voltage drop; when the fault is very serious (such as 50% AC voltage drop in the test case), by applying the bypassing valve strategy, the converter unit at high voltage terminal is switched off to make the system operated under the half-voltage mode, then, the healthy converter unit is controlled to transmit power continuously; during the arm bypassing period, the series connected damping resistors can effectively suppress the charging current, thus reduce the amplitude of module increased capacitance. © 2016, High Voltage Engineering Editorial Department of CEPI. All right reserved.

The combined C-MMC(Modular Multilevel Converter applying Clamp double submodule) with converter units connected in series and parallel is adopted to achieve the high-voltage and large-capacity power transmission. Two basic patterns, series-type and parallel-type, are classified for the switch-on/off process of converter unit in the practical project and the technical difficulties of each type are analyzed. The configuration scheme of bypass/disconnecting switches and the current limiting scheme of series resistors in clamp double submodule are designed. A strategy based on the bypass pair control is proposed for the series-type unit switch-on/off, which takes the arms of one phase as the bypass pair while blocks the arms of other phases during the unit switch-on/off to smoothly transfer the direct current between the converter unit and the bypass switch. For the blocked arms, the resistor connected to the clamping diode in series effectively suppresses the charging current and damps the increase of module capacitor voltage. The effectiveness and feasibility of the proposed strategy are verified by the results of PSCAD/EMTDC simulation. © 2015, Electric Power Automation Equipment Press. All right reserved.

Acoustic model adaptation algorithm aims at reducing the recognition performance degradation caused by the mismatch between training and testing data. Among the many adaptation techniques based on deep neural network( DNN), retraining is the most straightforward way. However it is prone to over-fitting, especially when adaptation data is sparse. In this paper, two typical acoustic adaptation methods, namely linear transformation network adaptation and Kullback-Leibler divergence regularization adaptation, are experimentally explored for taskadaptation purpose. An elaborate comparison is made, and results show that KL divergence regularization technique achieves better performance under different amounts of adaptation data.

In order to study the optimal modification of ventilation and dust control system, it is necessary to master the dust diffusion law and confirm the optimal velocity in a raw-coal postern. For the above popurse, a representative postern in Muguajie Coal Preparation Plant was taken as the research background. Based on the gas-solid two-phase flow theory, the discrete phase mode was used to simulate the dust concentration in the postern. Through comparing with on-the-spot measurement data of dust concentration, the simulating results were proven. The results indicated that the dust concentration reached the highest near the tail and guide chute, and it reduced gradually in radial direction. When the air velocity in the postern was 2 m/s, dust removal efficiency was above 97.2%, and the dust concentration was below 4 mg/m3.

For the high concentration of dust source, it needs to arrange multi-nozzle for co-inhibition which usually accompanied by interference phenomenon. The mutual collision and re-breaking of droplets plays an important role in the process of removaling dust. Meanwhile, the number and location of nozzles affect the whole effect of atomization remarkably. In order to comprehend the series of dynamic events in the process of droplet movement, the UDF method on the basis of collision model, breakup model and evaporation model were used to make numerical simulation for spray field. The simulation results show that the increase of nozzle numbers can appropriately improve the uniformity of particle diameter distribution, then to improve dust removal efficiency. With the increase of atomizing pressure, the initial kinetic energy of droplets increased, which had a good dust removal effect for the respirable dust of less than 5 μm.

Objective: To identify the clinical profile and neuro-imaging findings of the right temporal lobe variant of semantic dementia (RTLV), and to understand the relationship between anatomic and function by comparison to the left temporal lobe atrophy. Methods: Of 70 patients with diagnosis of frontotemporal dementia in our memory clinic, three patients with right temporal lobe atrophy were identified based on the MRI scans. We obtained the profile of cognitive function, behavior and personality changes in these 3 patients and compared them with those in 5 semantic dementia (SD) patients with predominant left-sided temporal lobe atrophy. We also underwent 18F-fluoro-2-deoxy-D-glucose positron emission tomography (18F-FDG-PET) with statistical parametric mapping analysis in these 3 patients. Results: In RTLV patients, the most prominent cognitive deficits were impairment of prosopagnosia, and getting lost. Variety behavioral symptoms including apathy, social disinhibition, stereotypy, compulsive behaviors were more prevalent in early course for patients with RTLV, which were different from SD patients with semantic loss began with anomia or single-word comprehension. 18F-FDG-PET scan showed decreased metabolism mainly in right temporal lobe and in other brain regions with varying degrees. Conclusions: The RTLV is clinical heterogeneous. Prosopagnosia, behavioral changes and getting lost are the main features in the early course. It therefore, might be proposed a separate clinical entity.

The effects of several kinds of non-phosphate additives including sodium chloride, sodium citrate, and trehalose on the water-holding capacity of scallop closed shell muscle were studied with detecting thawing loss rate, cooking loss rate, the gel water holding capacity (WHC) and elastic by a single factor test and L9 (34) orthogonal test. The results showed that they had been improved the water-holding capacity when they were used alone. When the optimal concentrations of soaking solution of sodium chloride, sodium citrate and trehalose were 0.5%, 1.5% and 0.6%, the thawing loss rate and cooking loss rate were respectively reduced from 10.54% to -5.80%, 52.06% to 39.11%, WHC and elastic were improved respectively from 3.70% to 6.03%, 2.53 mm to 3.04 mm. In addition, sodium citrate had the largest impact on water-holding capacity followed by sodium and trehalose. The above results showed that compound non-phosphate additive can improve water-holding capacity and texture of scallop closed shell muscle. It can replace phosphate as water-holding agent application into production.

In order to improve the extraction yield and content of astaxanthin, supercritical CO2 fluid extraction was used to extract astaxanthin from the head of Pacific White Shrimp, high performance liquid chromatography (HPLC) was adopted to determine the content of astaxanthin after saponification, the influence of extraction parameters on the extraction yield and astaxanthin content was determined by single factor test, then response surface method was employed to optimize technical parameters. The results showed that the influence of extraction parameters on the extraction yield of astaxanthin was not significant (p>0.05), but significant on the astaxanthin content (p<0.05). When the particle size of material was 40 mesh, moisture content was 9%, the theory optimum conditions for extraction astaxanthin with supercritical CO2 extraction were confirmed as follow: pressure 403.95 Pa, extraction temperature 39.95°C, CO2 flow rate 1.16 L/min, under the conditions, the astaxanthin content was 796.3 μg/g. In application, the astaxanthin content in validation test was 789.61 μg/g when the extraction pressure was 400 Pa, the extraction temperature was 40°C and the CO2 flow rate was 1.2 L/min. The results can provide a reference for extraction and purification of astaxanthin.

The deactivation of the catalyst for removing residual carbon in hydrotreating residue oil was studied by means of BET, ICP, elementary analysis, Raman spectrometry and TEM. It was shown that the coke forming on the spent catalyst was as much as 24.57%(w), and the coke distribution along the radial of the catalyst pellets was V-shape, which led to the pore diameters deceasing and the diffusion resistance increasing. It was found that the MoS2 phases were longer and the number of its piles was more after running. It was suggested that the resistance to coke formation and the stability of MoS2 phase at high temperature should be enhanced to improve the stability of the catalyst. During running, temperature runaway and hot spot should also be avoided.