Showing papers in "Journal of Tsinghua University in 2010"

Least squares support vector machine algorithm

Abstract: The least squares support vector machine(LS-SVM) has been widely used in various fields for artificial intelligence.This paper describes the basic LS-SVM algorithm.The influence of the model parameters on the model precision,complexity and computing time are analyzed using statistical learning theory and numerical examples.The conclusions give a good guide for determining the model parameters.An improved LS-SVM algorithm is developed with the results showing that the improved algorithm having better overall performance and better efficiency for wider application.

Comparison of methods for improving SOC estimation accuracy through an ampere-hour integeration approach

Abstract: Accurate values of the ampere-hour integration parameters are necessary to accurately estimate battery SOC (state of charge) Different methods have concentrated on correcting various parameters However, there is insufficient information on the importance of each parameter on the SOC accuracy Tests were conducted on a 32 V/11 A·h power LiFePO4 Li-ion battery to accurately describe the effect of various parameters on the SOC estimation accuracy The results show that the initial SOC correction is vital to the accuracy

Real-time detection of driver drowsiness based on steering performance

Abstract: On-the-road driving experiments were conducted to collect data for different levels of drowsiness.An analysis of the steering wheel data was used to identify two measures,the maximum non-steering percentage and the maximum standard deviation of the steering wheel angle,calculated using the double-window method,found to be significantly different for different drowsiness levels.These were introduced into a linear drowsiness prediction model that also takes into account the drowsiness level history.Validation tests show that the method which is solely based on the driver's steering performance has a successful drowsiness detection of about 82%.

Quantifying energy consumption and greenhouse gas emissions of the primary aluminum industry in China

Abstract: The energy consumption and greenhouse gas emissions of the primary aluminum industry in China were quantified to give a more complete understanding of the Chinese primary aluminum industry,based on the methodology given by the International Aluminum InstituteThe results show that in 2005,the Chinese primary aluminum industry consumed 6578 Mt of standard coal,which accounted for 29% of the national total energy consumption which aluminum smelting consumed 677% of the total energy consumption and discharged 827% of the greenhouse gases in the industryElectricity contributed 719% of the total energy consumption and 692% of the greenhouse gas emissions for the primary aluminum productionThus,improving the electricity production structure can effectively reduce energy consumption and greenhouse gas emissions

Probability for disaster chains in emergencies

Abstract: A series of secondary and derived events may occur after an emergency, which is called a disaster chain. The disaster chain makes the emergency event more complex; therefore, research is needed to improve emergency management of disaster chains. Theories of disaster science state that an emergency event consists of four elements: hazard formative factors, hazard affected bodies, hazard producing environment, and mutual functions. This paper presents a method to describe the disaster chains. The model is then used to describe the disaster chain of an earthquake with the results showing that the model accurately describes the emergency event and the event chain.

Stabilization of linear predictive control systems with softening constraints

Abstract: Model predictive control(MPC) instabilities caused by a short predictive horizon can be stabilized using a predictive contractive constraint for a linear predictive control system with softening constraints.The state contractive constraint concept is extended to a predictive input/output contractive constraint with convergence factors.With the open-loop optimal performance index based on the Lyapunov function,Lyapunov's second theorem is used to show that the MPC closed-loop system with the designed constraint is nominally stable.Simulations with a cascade integral object indicate that the method can stabilize a closed-loop system and the convergence factors can be used to adjust the convergence speed.Smaller convergence factors lead to the faster closed-loop system convergence.

Cases for life-cycle energy consumption and environmental emissions in residential buildings

Abstract: Three residential buildings with different sizes were analyzed based on the model developed in the companion paper 'Models for life-cycle energy consumption and environmental emissions in residential buildings' to better understand the characteristics and distributions of energy use and environmental emissions in each life-cycle phase.The results show that the average total energy use is 1.7 GJ/m2 per year with CO2 as the most common emissions followed by SO2,NOx,CO,and then PM10 throughout the buildings' life cycle.The energy use and emissions in materials extraction and production phase and the construction phase are as high as 80% of the total life-cycle energy consumption and emissions,which indicates how life-cycle energy conservation can be impacted with emissions reduced.

Relationship between spatial distribution of thief crime and geographical factors

Abstract: The geographical weighted regression(GWR) model is used to analyze the relationships between the spatial distributions of crime and geographical factors to find the cause of crime spatial distributions.The GWR model can capture the spatial variations in the spatially non-stationary problems.The GWR procedure is used to build a local model to analyze the effects of population density,road network density,and distance from police stations on the crime rate at the census tract level to give a set of mappable parameter estimates and t-values of significance that vary over space.Analytical results demonstrate that the spatial relationship between crime and geographical factors is a spatially non-stationary process and that the GWR model can help improve the accuracy of parameter estimation.

Evaluation mechanism for user behavior trust based on DSW

Abstract: Security measures for dynamic behavior trust should be combined with authentication security control to further strengthen the network service security.This paper presents basic criteria of evaluating user behavior trust based on double characteristics of social trust and computer evaluation requirements of the trust,with an evaluation mechanism for user behavior trust then developed based on double sliding windows(DSW).The trustworthiness and scalability of the user behavior evaluation were obtained by initializing the sliding window and setting the window size,the sliding conditions,the distrust and expired record locations,and the updating and replacing window contents.The trustworthiness and scalability were analyzed by using the user behavior evaluation criteria and by being compared with the conventional methods.The results show that the algorithm uses a fine grain size evaluation mechanism with the trust evaluated object not being the service providers but the end-user.

Cavitation model for oil film bearings

Abstract: The cavitation model significantly influences the calculational results for oil film bearings.Present cavitation models mainly focus on the characteristics of the oil film pressure profile.The oil cavitation mechanism was used here to develope a cavitation model based on the solubility of air in lubricating oil.The model has been used in computational fluid dynamics(CFD) simulation of a journal bearing with the calculated results with different eccentricities agreeing well with experimental results,including load capacity,side leakage flow and oil film rupture position.The model is more accurate than the Half-Sommerfeld cavitation model with excellent convergence.

Experimental study of thermoelectric generation from automobile exhaust

Abstract: Experiments were carried out to improve the automobile's fuel efficiency by a thermoelectric generator that recovers waste heat from the exhaust to reduce energy use and,thus,protect the environmentA platform was built to test the performance of multiple thermoelectric generatorsA series of tests gave the output power data of the thermoelectric generators for various hot side temperatures(83-270℃) and load conditions(0-520 Ω)The experimental data was used to design a system using thermoelectric generators for waste heat recoveryThe economic analysis using Dongfeng EQ140-1 truck shows that for the truck running 12 h/d and 330 d/a with the diesel oil price of RMB 5/L,the thermalelectric generation system from automobile exhaust needs 82 a to recover the costs

Life cycle assessment of economy for biodiesel

Abstract: An economic analysis was performed for biodiesel produced from rape seed oil,Jatropha curcas L oil,and waste oil throughout the biodiesel life cycle including the planting,harvesting,and transport,pretreatment,and biodiesel production and distribution.The life cycle assessment(LCA) cost analysis showed that the planting or purchase cost was about 76%～82% of the total cost,and the production cost excluding feedstock(rape seed oil,Jatropha curcas L oil and waste oil) was 14%～21% of the total cost.The planting and production costs accounted for the two largest shares of the total cost.The price of methanol affects biodiesel production.The Jatropha curcas L oil and waste oil based biodiesel fuels are cheaper than fossil petroleum based diesel.The Jatropha curcas L oil compete better economically.

Urban emergency evacuation plans based on GIS and multi-agent systems

Abstract: Urban evacuation during emergencies was modeled using a multi-agent system and a geographic information system (GIS). The model was based on the Repast platform with the GIS data displayed by the OpenMap software package as the evacuation environment. Human agent and shelter agent models were designed with rules to model human decision making, such as shelter selection and route planning. Movement of agents along the road network was simulated by a network based traffic model. Simulations were carried out for a Chinese city scenario to study the effect of information distribution on shelter selection. The results show that information sharing during evacuations reduces the total evacuation time.

Calculation of leakage permeance in a surface-mounted permanent magnet machine

Abstract: No-load magnetic flux leakage factor calculation is time-consuming using the finite element method. The magnetic circuit method was used with four different magnetic flux leakage paths of the permanent magnet motor taken into account to eliminate the inaccuracy due to the motor structure changes by determining the integral flux leakage region according to the motor structure size,with a small calculation amount. Simulation results with the magnetic circuit and the finite element methods show the accuracy of the proposed method with changing arc coefficient,air-gap length and magnet height. The method without complex finite element numerical calculation is convenient and suitable for the permanent magnet machine design and optimization.

Writer identification based on improved microstructure features

Abstract: The method previously proposed by the authors for writer identification based on the grid microstructure feature is a well-performance text-independent method for multilingual handwritings Microstructure feature extraction was improved in this paper with the weighted Manhattan distance as the similarity measurement to describe finer writing characteristics The method improves the identification performance on Chinese handwritings with the top-1 identification accuracy of 954% and the top-20 accuracy of 100% in the HIT-WM database involving 240 writers Experiments were also made to analyze the effect of the text content dependency between handwriting samples on identification performance in practical applications

Aircraft boarding strategy based on cellular automata

Abstract: Aircraft boarding strategies seek to minimize the boarding time.This paper provided simulations and analyses of aircraft boarding strategies using a hyperbolic model to describe conflicts in the aisle when passengers are placing baggage in the overhead bins.An aisle-conflict model and a seat-conflict model are used to describe passenger motion based on a cellular automata model.Simulations for seven strategies and twenty-one policies show that some strategies do not improve the boarding efficiency and an orderly strategy is not necessarily better than a disordered one.On the contrary,random boarding may be the simplest but most effective method.

Separation of magnesium and lithium by solvent extraction using di-(2-ethylhexyl) phosphoric acid(D2EHPA)

Abstract: Lithium is difficult to directly separate from brine from the salt lake because of the high concentration ratio of Mg2+/Li+Solvent extraction was used to extract magnesium from the brine to solve this problem using di-(2-ethylhexyl) phosphoric acid(D2EHPA) as the extractant and kerosene as the diluentThe saponificated D2EHPA has good Mg2+ extraction performanceFor the same phase ratio and saponification ratio,the Mg2+ extration efficiency increases with the D2EHPA concentrationThe saponification ratio should not be higher than 80% and the D2EHPA concentration should not be higher than 50% to avoid an excessive viscous extractantTo obtain a high purity of Mg compound,MgCl2 was used to strip the Li+ and K+ in the loaded extractantHigher MgCl2 concentrations and phase ratios improve the stripping

Dynamic simulations and failure analysis of a clamp band system for spacecraft

Abstract: A plate element model of a clamp band was used to analyze the dynamic properties of the satellite-rocket separation system and the influence of component failure. Simulations using the ADAMS analyzed the effects of the pyrotechnic bolt shock, lateral spring stiffness, band preload, and the friction between band components on the dynamic separation envelope of the band clamp. Three typical failure modes, including failure of a single explosive bolt, connection failure of a V shoe with the band, and inactivation of a separation spring, were investigated using the plate element model. The simulations show that the dynamic envelope can be improved by increasing the lateral spring stiffness and decreasing the shock from the explosive bolts. Although the satellite separates from the launch vehicle with connection failure of a V shoe with the band and inactivation of a separation spring, these failures could have serious consequences, such as collisions between the V shoes and the satellite and separation attitude deflection of the satellite.

Field application of multiple model predictive control for superheat temperature in boilers

Abstract: This paper presents a multiple model predictive control (MMPC) for a 300 MW coal-fired power unit, where conventional proportional-integral-derivative (PID) control of the superheat temperature is not effective due to the load variations Instead of model switching, a linear-weighting strategy is used to relate multiple models to produce a smoother output The strategy is validated by a group of water spray characteristic transfer functions from field experiments and model identification from step response data from simulations The MMPC was implanted in C++ and has been installed on a 300 MW coal-fired unit for several months The results show higher average superheat steam temperatures, lower average temperature deviations and control energy use, and better adaptation to large load variations

Construction collision detection for site entities based on 4-D space-time model

Abstract: A 4-D space-time model and an improved construction collision detection algorithm were developed for site entities to improve construction safety management.The model uses the hierarchy bounding boxes algorithm and the 4-D technology and building information model(BIM).The 4-D construction site layout management system accurately describes the 3-D shapes of site entities using the boundary representation(B-rep) method.The algorithm helps managers to dynamically analyze the construction process and avoid collisions between site entities and structural elements to improve the planning and deployment of site resources to meet the construction needs.

Models for life-cycle energy consumption and environmental emissions in residential buildings

Abstract: Impact factors for energy consumption and environmental emissions of building related activities were analyzed with several models constructed to design an architectural scheme for real low-energy,environmentally-friendly buildings.Models were constructed based on life cycle assessment(LCA) to quantify the energy use and environmental emissions(CO2,SO2,CO,NOx and PM10) for each life-cycle phase from material extraction and production through construction,operation,and maintenance to end-of-life treatment.The findings lay a theoretical groundwork for a comprehensive understanding of the levels of energy use and pollutant emissions in residential buildings in China.

Analysis of embodied energy and transfer emissions of China's import and export trade

Abstract: For estimating recent export embodied energy and emissions in China,this paper calculated the export goods embodied energy and CO2 emissions in 2002,2005 and 2007 using an input-output model and China input-output tables in 2002,2005 and 2007.The results indicate that the import and export processing trade plays a significant role in calculating the embodied energy and CO2 emissions in China.The export embodied energy(carbon equivalent) increased from 209 Mt in 2002 to 591 Mt in 2005 in China,with its ratio to the total energy consumption increasing from 13.79% to 25.04%.The import embodied energy also shows a growth trend in China.The net export embodied energy has reached 261 Mtce in 2007,accounting for 9.3% of total energy consumption.China remains a major economy with a net export embodied energy.

Biodegradation characterization of a pyridine-degrading strain

Abstract: Pyridine is a intractable toxic organic and a N-heterocycle compound.The biodegradation characterization of pyridine was investigated in a pure culture using a pyridine-degrading strain isolated from the soil in a pesticide plant.The strain was identified as Paracoccus sp.on the basis of a partial 16SrRNA gene sequence analysis and designated as strain KT-5.Analysis of pyridine degradation indicated that KT-5 was able to utilize pyridine as the sole source of carbon,energy and nitrogen.High concentrations of pyridine(1.331 g/L) did not inhibit the growth of KT-5.KT-5 showed different degradation characteristics with different initial pyridine concentrations.When the pyridine concentration was below 900 mg/L,the initial KT-5 degradation rate was higher than the final degradation rate.When the pyridine concentration was above 0.9 g/L,the initial KT-5 degradation rate was lower than the final degradation rate.

Comparison of maximum power point tracking technologies for photovoltaic power systems

Abstract: Maximum power point tracking(MPPT) technologies and their applications were classified based on the differences of control algorithms with three typical methods analyzed,i.e.the perturb and observe algorithm,incremental conductance algorithm and fuzzy control algorithm.The control model for each method was established with the system simulated using MATLAB/Simulink.Experiments were made to verify the feasibility and performance of the methods.By analyzing and comparing the simulation and experimental results,the quantitative characteristics of each method were obtained,including the steady state accuracy,dynamic response,misjudgment correction and hardware requirements.Therefore,the results provide the applicable conditions of the three methods for choosing the optimal scheme in practical MPPT applications.

Nuclide removal from the primary coolant of a nuclear power plant by continuous electrodeionization

Abstract: The efficiency of nuclide removal from the primary coolant of a nuclear power plant by continuous electrodeionization was investigated using membrane stack experiments for different water concentrations and electric currents.The results demonstrate that the current should be greater than 0.1 A when treating the primary coolant with continuous electrodeionization(CEDI) to reach 99% efficiency.Proper pretreatment or a two-stage CEDI treatment was necessary for high nuclides concentrations of the nuclides.CEDI is efficient,works continuously and does not need acid-bases regeneration.

Reliability provision mechanism for large-scale de-duplication storage systems

Abstract: Data de-duplication technology achieves storage utilization by storing duplicate data chunks just once,but at the cost of error resilience or reliability.This paper presents a high reliability provision mechanism for large-scale de-duplication storage system(RLDS in short).RLDS packages variable data chunks into fixed sized objects,while exploits ECC code to encode the objects and distributes them among storage nodes in a redundancy group,which is dynamically generated according to actual failure domains.RLDS uses a distributed recovery process upon failures.Experimental results show that RLDS can provide the same storage utilization as RAID schemes,as well as comparable reliability to replication based schemes with more redundancies.The results also show that the average recovery time of RLDS based configurations is much shorter than that of RAID schemes and that RLDS can provide run-time dynamic load balance by recovering data of overloaded nodes to light load nodes.

Building fire smoke control strategies simulation using fire and evacuation information

Abstract: During high-rise building fires, the fire must be controlled long enough to provide sufficient time for occupant evacuation, if heating, ventilating, and air conditioning (HVAC) systems and controllable door operations can be used to slow smoke propagation on the fire floor. A smoke control strategy utilizing the HVAC system and door operations was analyzed for various fire locations and building occupancy. The fire dynamic simulator (FDS) was utilized to simulate the smoke propagation, with an in-house fire evacuation (CFE) model used to simulate evacuation for different operating scenarios. The smoke propagation for various HVAC strategies and door operations in two typical buildings were evaluated with the results showing that the control strategies can effectively limit smoke propagation: reduce smoke speed by about 15%, reduce smoke temperature by about 25%, and reduce toxic and harmful gas concentrations by about 37%, which should improve occupant evacuation.

Contact simulations of flexible bodies based on absolute nodal coordinates

Abstract: The absolute nodal coordinate formulation is an important modeling method for flexible multibody systems for predicting collisions of flexible bodies in multibody dynamics simulations.A meshing method was developed for an absolute nodal coordinate beam and a plate element.The impact force calculating using Hertz contact theory was found using contact information from the collision detection between geometries.A collision calculation module was embedded in a multibody dynamics solver to predict contact between the flexible bodies or a flexible body and a rigid body.The simulations show that the meshing method and collision force calculations accurately model the beam and plate dynamic responses and the impact forces in beam,plate and rigid body collisions;thus,further expanding the absolute nodal coordinate formulation for engineering applications.

Effects of particle volume fraction and magnetic field intensity on particle structure of magnetorheological elastomers

Abstract: The controllable mechanical and electrical properties of magnetorheological elastomers(MRE) mainly depend on the particle structure inside the matrix materialMany factors affect the particle structureThe effects of particle volume fraction and magnetic field intensity on the particle structure were studied experimentallyThe results show that the particle structure transformed from continuous long chains to net structures as the particle volume fraction increased with strong curing magnetic fieldsThe structure changed from discontinuous short chains to net structures in weak curing fieldsThe net structures easily formed in weak magnetic fields with the structure being more complex than in the strong fieldsIncreases of the field strength and the volume fraction both improved aggregation of the particle chainsThe synergetic effect of different particle sizes induced particle aggregation and more complex branch like structures

Fatigue tests of single layer and laminated glass plates with point supports

Abstract: Brittle fatigue failure,which easily occurs in glass plates with wind loads,must be well understood for glass curtain wall designs.This paper describes high frequency fatigue tests on single layer and laminated glass panels.The fatigue stress amplitudes were determined according to the current codes as well as in static loading tests.The maximum stress at the center of the glass panel was used to quantify the applied fatigue load by a loading test.The test results show no fatigue after 2 million-cycle loading.The remaining static strength after the fatigue tests showed no strength reduction after the 2 million cycle loadings.The experiments show that point-supported single layer and laminated glass panels both have adequate fatigue capacity.