The electromechanical braking system of vehicle with motorized wheel is analyzed using vehicle braking stability theory. Referring to existing control strategies for regenerative braking system, a new regenerative braking system control strategy for in-wheel motor vehicle is proposed. Using the proposed strategy, simulations of in-wheel motor vehicle and normal electric vehicle are carried out under New European Driving Cycle (NEDC) working condition. Simulation results show that the proposed strategy is applicable to in-wheel motor vehicle, and makes the in-wheel motor vehicle markedly superior to normal electric vehicle in terms of braking stability and energy recovery capability.

A braking force distribution control strategy is proposed using a hierarchical control structure based on electronically controlled braking system of bus (EBS). In the upper layer, a deceleration control strategy is designed to ensure that the vehicle braking deceleration is only dependent on the brake pedal distance but independent of the vehicle load condition. In the lower layer, a strategy of braking force distribution between front and rear axles based on ideal braking deceleration of the upper layer is taken to fulfill driver's need of whole bus braking force and to realize smooth axle load transfer. The proposed control strategy is tested using a hardware-in-the-loop simulation test bench of EBS system. Result show this strategy can improve the driving comfort and safety of bus.

A human-seat-suspension model with seven degrees-of-freedom was built on a 1/4 car model. The vertical acceleration, suspension work space and dynamic tire load were selected as the objective functions, and parameters of the suspension and seat cushion were selected as the design variables according to the results of parameter sensitivity analysis. Dimensionless method was introduced to make the optimization computation independent of the working conditions. Then a multi-objective optimization model was built to optimize the ride dynamic behavior of vehicle. Non-dominated Sorting Genetic Algorithm-II (NSGA-II) was employed for optimization computation and Pareto optimal set and Pareto frontier were obtained. Results show that seat cushion with small stiffness and large damping contributes to improve ride comfort without worsening other behaviors. In Pareto optimal set, the suspension stiffness is linearly directly proportional to suspension damping, which decides that optimal ride comfort and suspension work space can not be obtained simultaneously. Tire stiffness can be approximately determined based on dynamic tire load.

In this paper, the shift quality evaluation methods of stepped automatic transmission are analyzed first. Then the evaluation criteria of the shift quality are established based on the power-on up-shift of Dual Clutch Transmission (DCT). Finally, the shift quality objective evaluation method of DCT under ideal shift process is proposed. Comparison tests show that the proposed objective evaluation method is in good agreement with the subjective evaluation methods. The model provides a new shift quality evaluation approach for stepped automatic transmission.

A two-stage clustering method is proposed to construct the driving cycle of garbage truck. In this method, first the concept of kinematics interval is put forward to describe segment's instantaneous characteristics, upon which an advanced hierarchical clustering is carried out. Second, proper segments are chosen from each category according to the results of the second stage clustering. Finally, the representative driving cycle is fitted out. The proposed method is applied to construct driving cycles of a typical garbage truck in Xi'an City. For comparison a driving cycle is also constructed using traditional clustering method, actual driving cycle is measured and error analysis is carried out. Results show that compared to the traditional clustering method, the two-stage clustering method can improve the probability distribution of speed by 59.4%, the probability distribution of acceleration by 81.7%, and the joint probability distribution of speed-acceleration by 22.4%.

A two-target model based on effective path is established to solve the Feeder Bus Network Design Problem (FBNDP). First, the effective path concept is defined on the basis of complex network betweenness considering urban traffic congestion, which is the basis of the design of feeder bus network. Taking into account the constraints such as the line length, non-linear coefficient, transport capacity etc., the two-target model is established, in which the minimization of passenger travel time and maximum of the transport efficiency are taken as the targets. Then referring to idea of "Combined Each Route, then Form the Network", the solution approach to the model is proposed. Finally, the model is applied to the Light Rail Transit 4 of Changchun. The results show that the feeder bus network designed by the model in the presence of traffic congestion can reflect the residents' travel selection strategy better.

In order to set the lane change warning parameters being more suitable to the characteristics of domestic drivers, the influences of relative speed and distance between the subject vehicle and its neighboring vehicles on the lane change behavior are investigated using road experiment data. Then the safety distance that allows the driver to perform the lane change operation under different speed levels is determined. Besides, to identify the driver's subjective basis for risk assessment, different characteristics of the driver in cancelling and performing lane change are analyzed. Finally, the warning rules for the neighboring vehicles in different speed levels are proposed. The results show that the warning proportion in the road experiments is basically consistent with the actual situation, which demonstrate the feasibility of the proposed warning rules.

A new theoretical formula for soft ground bearing capacity is derived. This formula is based the unified strength theory, and the effects of consolidation, intermediate principal stress and the coefficient of lateral pressure are taken into consideration. According to this formula, an improved calculation method of the critical filling height of embankment is proposed. In addition, the calculation method is applied to railway subgrade filling test data to check its feasibility. The calculated results show reasonable good agreement with experiment results. The new formula may be used as the theoretical basis for determining the safe filling height at different phases of railway subgrade works.

The aggregate skeleton composition of the stone mastic asphalt with the nominal maximum aggregate size of 13.2 mm (SMA-13) and its contact properties were studied using a three-dimensional discrete element method. A user-writing program was coded with FISH language to extract the contact forces in different sized aggregates (> 2.36 mm). The roles of different sized aggregates and the contact properties between these aggregates were investigated based on the statistical analysis of large amount of data. The results of the study show that aggregates with sizes of 13.2 mm, 9.5 mm and 4.75 mm form the aggregate skeleton of SMA-13, while 2.36 mm aggregates mainly fill the voids of the skeleton. Furthermore, aggregates of sizes 9.5 mm and 4.75 mm are the main body of the skeleton, which bear nearly 60% of the vehicle load on asphalt pavement.

Base on previous research on soil arching effect between adjacent piles, the concept of load transfer ratio of soil arching was proposed, and an analysis model was established. Through centrifuge model test, the Ito's calculation method of lateral pressure on pile was proved to be applicable to deep foundation supported with sparse row piles. Based on this approach, the influence factor of the load transfer ratio of soil arching was comprehensively analyzed with an assumed case of sparse row piles. Results show that the earth pressure passing to the pile by soil arching can account for over 60% of the total pressure on the pile. Therefore, soil arching effect between adjacent piles in sparse row pile retaining structure can not be ignored. The load transfer ratio of soil arching approximately exponentially decreases with the increase in pile space. When the pile space is constant the load transfer ratio decreases with the increase in pile diameter. In addition, if parameters are unchanged, the load transfer ratio exponentially increases with the cohesion and internal friction angle of the soil.

The dynamic responses of an actual Ferris wheel system induced by failure of some stay cables are investigated. Whether the failure of some stay cables can cause further failure of other stay cables, or lost of the lateral stiffness of the wheel, even collapse is discussed. Constructive results are obtained by means of dynamic analysis. The variation periods of the lateral displacement of the wheel and the stay cable force are almost equal to the first natural vibration period (T_l) of the wheel. It is about half of T_l after the failure of the stay cables, the effect of the resilience of the failure stay cables on the dynamic behavior of the wheel becomes obvious and reach to an upper limit. The predominant failure model is the failure of three stay cables in one side at the same time. In this situation, the maximal tension force of the stay cables laying in one side reaches 64% of it breaking force; while the stay cables laying in the other side are almost in slack state. In conclusion, any specific failure mode of the stay cables would not cause the further failure of other stay cables, which might result in collapse of the wheel. Numerical results show that the collapse resistance of the Ferris wheel is strong enough.

A diesel/Liquefied Natural Gas (LNG) duel fuel system is established by installation of a LGN Port Fuel Injection (PFI) system on a 4-cylinder diesel engine, in which the original common rail system and control unit are kept. The main principle is to utilize injected diesel as the ignition source of the natural gas. The system has four modes: bypass mode, following mode, duel fuel mode and individual mode. Calibration and optimization are conducted on a test bench, and experiment results show that the fuel mass substitution rate is in the range of 62% to 87%; the duel fuel system has higher efficiency under middle and heavy load in comparison to the original diesel engine.

Accumulator is an energy storage device in parallel hydraulic hybrid systems, and its main parameters greatly influence the energy recovery effect. The models of the main system elements were built and simulated in AMESim software. Simulation curves were obtained under different conditions, which were used to study the effects of the main parameters on the accumulator system performance. In order to verify the simulation results, experiments were conducted on the test bench restructured on real vehicles according to the simulation results. It is shown that the experimental results are in agreement with the simulation results. The pre-charge pressure and volume size of the accumulator system have the most significant influence on energy saving. Brake energy recovery and recycling rate are inversely proportional to the speed. Pre-charge pressure has great impact on the braking distance. After the accumulator volume reaches a certain value, the changes of energy recovery and reutilization do not significantly improve further with the volume increase.

A fault recognition method based on Hilbert-Huang Transform (HHT) and the fuzzy C-means clustering is proposed. This method takes advantages of HHT in processing nonlinear and non-stationary signals. The vibration acceleration signals collected from the axial piston pump housing are processed by HHT. First, the signals are treated by means of the Empirical Mode Decomposition (EMD) and the Ensemble Empirical Mode Decomposition (EEMD) respectively. Then, the Intrinsic Mode Function (IMF), which is most sensitive to faults, is chosen according to the short-term maximum entropy spectrum analysis, and the sensitive IMF is re-decomposed. Third, the fault characteristic vector is constructed based on the characteristic energy of the local marginal energy spectrum. Finally, the faults are recognized using the fuzzy C-means clustering algorithm. The results show that, compared with EMD, the interactive time based on EEMD is reduced greatly, and the fault recognition accuracy is substantially improved by the proposed recognition method.

A new curve fitting method based on moving least square is constructed for closed discrete points. According to the moving least square theory, the fitting point is obtained between discrete points and the influence domain with defined radius is determined in order. Considering the geometric features of discrete points, a new approach to determine the weights of neighbor points is proposed for the local approximants by constructing a point associated with the chord length, which makes the weights of neighboring points vary decreasingly. The discrete points of blade-section profile are fitted and the results confirm and validate the proposed fitting method.

The degree of interaction between dependent systems determines the error between the theory reliability and field reliability. So the dependent coefficient between the dependent systems is an important reference for design reliability and distribution reliability. However, the existing method can only estimate the dependent coefficient of a simple chain which includes two systems. In this work, the interaction in failure chains is classified after investigating large amount of failure data. Based on the types of failure chains, a model of dependent coefficient for complex system is developed using the function relationships among the dependent failure rate, independent failure rate and dependent coefficient of subsystems. A case study is carried out to verify the developed model. Results show that the model can estimate the dependent coefficient of a complex system, and improve the theory. It is very important to more precisely design the reliability growth and to optimize the repair and maintenance schedules.

A two points Receptance Coupling Subsystem Algorithm (RCSA) model is proposed at the joint site of tool subsystem and holder subsystem of an end milling processing system. The displacement receptance of tool point was obtained according to the RCSA model. Harmonic response analysis was carried out to get the displacement receptance of tool subsystem and holder subsystem using ANSYS software. Matlab was adopted to write the program of two points RCSA. Data from ANSYS were used to calculate the displacement receptance of the tool point. A finite element model of the system, including tool subsystem and holder subsystem, was established in ANSYS, and displacement receptance of tool point was obtained. The results of the RCSA model and finite element model were compared and relative error of the two models was calculated. The relative error is less than 0.7% in the frequency range concerned. Experiment was conducted, and the experiment results and theoretical results are in good agreement.

In order to accurately identify the early fault feature of helical gear, a new diagnosis algorithm is proposed based on the angular domain synchronous average and order tracking analysis. First, employing the concept of time domain synchronous average, an algorithm of angular domain synchronous average is put forward to reduce the noise interference. Using optical encoder as the external trigger, the vibration signals is angular domain are obtained by resampling the vibration signals of time domain according to the encoder pulse. By angular domain synchronous average processing, the irrelevant periodic signals and noise interference are reduced. Then, the fault features of helical gear rotating at low speed are extracted by the order tracking analysis. Experiment results of pitting fault diagnosis verify the effectiveness of the proposed algorithm.

The thermal deformation of the friction pair of hydrostatic bearing of heavy vertical lathe has great influence on the shape of oil film under the conditions of high speed and heavy load. The thermal deformation will result in great changes of the oil flow state, basic assumption and boundary conditions. Thus, it is difficult to use the theoretical lubrication performance formula to predict precisely the lubrication characteristics of the oil film. To solve this key problem, the thermal deformation of rotating workbench and base is simulated based on fluid-structure interaction method, and the law of thermal deformation distribution is revealed. The relationship between the thermal deformation of hydrostatic thrust bearing and rotational speed of the workbench is established. A model to predict the lubrication performance of oil film is proposed considering the thermal deformation. A test rig is established and experiments are carried out. The experiment results are basically in agreement with the simulation results. This work may provide reasonable theoretical basis for design of heat dissipation structure, lubrication and solving deformation problem of hydrostatic thrust bearing.

Taking the combined roller of GM1400×800 high pressure grinding rolls as the object, a combined roller geometric model of an interference connection is proposed. The key structural parameters of the combined roller bearing torque were obtained. With the determined unilateral interference quality range and selected appropriate value, the influence of drum-shape error on combined roller contact surface bearing torque of high pressure grinding rolls was analyzed. The results show that proper drum-shape error can not only reduce stress concentration on both ends of the combined roller, but also make the radial stress distribution on the contact surface more uniform and increase the average radial stress value. When the unilateral interference quantity equals to 0.45 mm, the torque capacity of the combined roller increases monotonously with the drum-shape error if the error is lass than 0.07 mm, and when the error equals to 0.07 mm the capacity reaches the maximum value. Hereafter, the torque capacity gradually decreases as the error increases. Therefore, to reduce the stress concentration on both ends of the combined roller, the drum-shape error should be controlled within the range of 0.03 mm to 0.07 mm.

The mechanical properties of self-piercing riveted (SPR) lap joints in dissimilar metal sheets of aluminum 5052 and copper H62 were investigated by experiments. The force-displacement and load-life curves were obtained to characterize the joint properties. A mathematical statistic method was used to analyze the validity of the experiment data. The strength, stiffness, impact resistance, failure modes and failure mechanism of the joints were studied. Results show that the stiffness varies with load of the joint, but fatigue strength improves with the increase in stiffness. The static strength of SCC is the highest, SAA2 second, while SAC is slightly higher than SAA1. The static strength of SPR joints are influenced by the length of rivet, thickness and property of the sheets, and the effect of the upper sheet property is prominent. The effect of the lower sheet on impact resistance is prominent. The displacement at failure is related to the sheet thickness, and the failure mode is the rivet pull-out from the lower sheet. The failure strength of SCC is the highest, SAA2 is better than SAC. The fatigue strength is influenced by the slight slip, the thickness and property of the sheets. Joints of the same metal mainly fracture in the lower sheet, while joints of dissimilar metals fracture in the soft material.

Smoothed Multiscale Finite Element Method (SMsFEM) is proposed to solve the problems associated with heterogeneous and multiscale materials. SMsFEM is used to study the fliud flow in fractured madia with a dual porosity model. Using the global fine-scale solution at initial time to determine the boundary conditions of the basis function, local information of material heterogeneity and coupled global information, such as fluid flow in porous and fractured media, are simultaneously reflected in the multiscale finite element basis functions. As a result, an accurate solution can be achieved in the coarse scale. SMsFEM introduces the strain smoothing technique into conventional Multiscale Finite Element Method (MsFEM), therefore, simplifies the assembly of the macro matrix and overcomes the overly-stiff phenomenon existing in the traditional finite element method. Simulation results indicate that the proposed SMsFEM is highly consistent with the fine-scale solutions from standard finite element method. It is also demonstrated that SMsFEM has better efficiency and accuracy than MsFEM.

In order to reduce the longitudinal extension of rectangular tube in the roll forming process, a three-dimensional elastic-plastic nonlinear Finite Element Model (FEM) of the roll forming is established with the actual factory production process. The metal flow law in the roll forming of rectangular tube and the distribution of the longitudinal elongation rate are obtained by analyzing the simulation results. The obtained results are consistent with the field measurements. The effects of the roller speed, the thickness of the tube wall and the location of the roll diameter on the elongation are analyzed using the proposed FEM. Then these processing parameters are optimized, which improve the continuous rolling performance of rectangular tubes. This research may help to predict the final dimensional accuracy and determine the processing parameters of the roll forming.

Thermal-mechanical simulation tests of 316LN steel were carried out on Gleeble-1500D thermal-mechanical simulator with temperature in the range of 950 ~ 1200 ℃ and strain rates of 0.005, 0.05, 0.5 and 1 s^-1. The fractures were observed using a Scanning Electron Microscope (SEM). The plasticity and fracture mechanism of 316LN steel at elevated temperature were studied. The results show that the fracture of 316LN steel at elevated temperature is ductile fracture. The size and depth of dimples increase with the temperature and strain rate. The plasticity also increases with temperature and strain rate. Meanwhile, the relationship models between fracture strain and deformation conditions (temperature and strain rate) and between plastic indexes (elongation and area reduction) and deformation conditions were established using regression method, respectively. Employing these models, the fracture strain, elongation and area reduction of 316LN steel under certain conditions can be calculated. This study provides reference to the design of forging technology of 316LN steel.

To overcome the problems in wheat harvesting machine such as loss, damage, high impurity rate, fracture and adhesion, the friction and tensile properties of wheat were studied by experiments. The 'Jinan wheat 20' was used as subject. The moisture content of the wheat was 36.16% measured with a PM-8188 grain moisture tester. The static friction coefficient was measured on a mechanical static friction test bench. The tensile curve was obtained using a QT-1186 electronic testing machine. Results show that the static friction between rubber and wheat is higher than other materials. So the bionic reel and threshing mechanism made of rubber could increase the static friction force. The tensile strength gradually increases from the bottom to top. Using horizontal rotation rub threshing process and bionic cutting parts could improve the threshing performance and reduce the impurity rate with lower power consumption.

3D model reconstruction and partition of the plantar surface of ostrich didactyl foot were achieved using reverse engineering. Discrete element method and finite element method were employed to simulate the interactions between the plantar surface of ostrich didactyl foot and sand. It was shown that the flowing extent of sand particles under the plantar surface along the foot breadth direction was greater than that along the foot length direction. This may indicate that the plantar surface morphology of ostrich foot can preferably retard the slippage during ostrich foot moving ahead on sand. Within the plantar surface of ostrich foot, the middle groove in the third toe plantar surface has most obvious effect of sand flow fixation, under which the forces on the sand particle groups present the grid-patterned stable tight distribution. A mathematical model of the middle groove in the third toe plantar surface was established, which may provide theoretical reference for bionic design of the plantar surface of ostrich foot.

This paper studies the energy management of pure electric vehicles. By analyzing the structure of the power battery system and the components and parts of high voltage system, the power flow relationship of the energy management system is confirmed. Aiming at the controlling strategy of common energy management system, the power distribution coefficient of the system is introduced based on the characteristics of the vehicle and battery. An optimal control method based on fuzzy controller is designed, analyzed and validated by simulation. Finally, comparison experiments of the control strategy are carried out on a drum test bench. Results show that this optimal control method can reduce the energy consumption ratio effectively, and under the condition that the total energy of the battery is steady, the vehicle can keep ideal economic performance and ideal dynamic property.

This paper presents a strategy of vehicle stability control based on the Electric Hydraulic Braking (EHB) system of the vehicle. First, the yaw rate deviation and the mass-centre sideslip angle rate threshold are determined using fuzzy inference methods. Then, the logic threshold PI control method is used to calculate the additional yawing moment. Finally the additional yawing moment is realized on the EHB system. In addition, the fuzzy PI self-tuning algorithm is used to optimize the control of the target pressure of the wheel cylinder of EHB system. Simulation and experiment results show that the fuzzy PI self-tuning algorithm can obtain good control effect in the whole working section of EHB system. When the vehicle loses stability in steering, the control strategy can implement corrective action in time to improve the safety of the vehicle.

In target tracking using a single feature, the tracking robustness can not be guaranteed in the case of complex environments, and the increase in particle number will lead to in low efficiency of the algorithm. To overcome these problems, a multiple feature fusion strategy is chosen to ensure sustained and stable tracking. Using this strategy, the weights of features will be adjusted adaptively to the environmental changes, and the number of particles is adaptive to improve the efficiency of the algorithm. Experimental results show that the proposed algorithm can effectively solve the target rotation, target occlusion, background confusion and many other issues with high robustness.

In order to improve the efficiency of the robotic excavator meanwhile to obtain its smooth running trajectory, a new trajectory planning method based on Non-uniform Rational Basis Spline (NURBS) was proposed. A quintic NURBS curve was employed to interpolate joint positions of the robotic excavator. Using the vector-valued function, by Leibniz formula, higher order derivative vector of NURBS curve was obtained to set the start and stop speeds and accelerations; and to make up the trajectory curves with continuous jerk, which passes specified joint positions. The effect of NURBS weight factor in sectional trajectory planning was analyzed. Nonlinear minimization problems under constraints were solved using sequential quadratic programming; then time-optimal trajectories were planned. Simulation and experiment results show that the proposed trajectory planning method can provide time-optimal trajectories, which satisfy kinematic constraints and the sectional trajectory optimization is effective.

A robot system proposed in this paper can be used to automatic spraying for aircraft and other products with large-scale free-form surfaces. The mechanical structure and control system structure of the spray system are described. The spraying operation process is analyzed in details, and the technologies related to the spraying operation planning, such as aircraft pose calibration and spray gun trajectory planning, are studied. Based on CATIA secondary development, the control software of the spraying system is developed, which seamlessly integrates all the software operations corresponding to the whole spraying process into ACTIA platform. Spray tests on the workpiece demonstrate that the robot system is running stable and reliable with high automation, which can effectively improve the coating quality.

To solve the problem of actuator failure of high-speed trains, an adaptive fault-tolerant control scheme is presented. First, a nonlinear uncertain dynamic model for high-speed trains under normal operating state is built, and corresponding model with actuator failure is analyzed. Then, to realize fault-tolerant control, a module using high-speed train model based state observer for fault detection and isolation is designed. The adaptation law can be reconfigured according to the output of fault detection and isolation based on train operating state, which forms a new framework of adaptive fault-tolerant control via selection of different tuning function β_. The stability of the fault-tolerant control system of high-speed trains is proved theoretically. Numerical simulations of a train model verify the effectiveness of the presented scheme.

To ensure the correctness and uniqueness of the diagnosis, the definition of diagnosability is presented and an algorithm is proposed to test the diagnosability of Higher-order Discrete Event Systems (HDES). To diagnose HDES, it is needed to construct a diagnosis engine of HDES, which is also used for testing the diagnosability. The state of diagnosis engine includes sets of faults, thus, it is not necessary to traverse the whole graph to search faults when the behavior has been reconstructed. Furthermore, in the premise of no influence on the diagnosis results, the transition between states is constrained, thus, to avoid the generation of unnecessary nodes.

In this paper, the Constrained Optimization Problem (COP) is converted into a bi-objective optimization problem with preference. Then the problem is solved with a Guiding Multi-objective Differential Evolution (GMODE) algorithm. The other methods based on Pareto dominance treat both objectives as equal importance without bias to either objective. In contrast, the proposed GMODE algorithm is guided by α-domination to search with dynamic bias to different objectives, which overcomes the drawback of the methods based on Pareto dominance and improves the convergence speed of the algorithm. Numerical experiments on several well-known benchmark functions and comparison with the other three state-of-the-art methods demonstrate that the GMODE algorithm is competitive with, in some cases superior to the other methods in terms of the quality, efficiency and robustness.

Focusing on non-linear separable network data with unknown specification, an unsupervised feature selection algorithm based on Support Vector Machine (SVM) was proposed, termed UFSSVM. The proposed algorithm first maps the non-linear network data into a high dimensional feature space using a non-linear mapping function; then it performs unsupervised feature selection in the high dimensional feature space. Compared with traditional unsupervised feature selection algorithms, the proposed algorithm can automatically get the relevant features just using the original network packet without the preprocessing step to get the original feature set. The performance of the proposed algorithm is examined by simulations and with real network data set. Experiment results illustrate the feasibility and effectiveness of the proposed algorithm in feature subset selection.

In this paper, we propose a single batch test algorithm on Cost-sensitive Uncertain Nave Bayes for Uncertain Data (SBT-CSUNB). We define the influence of an uncertain attribute on the total cost in cost-sensitive Nave Bayes Classifier, and put forward a method to fine an optimal batch test strategy. Experiment results on UCI Database demonstrate that the proposed algorithm can effectively reduce the total cost, and the performance is stable with different parameters and under high uncertain rate.

On the basis of machine cogmatics, a multi-information transformation model is put forward for extending the transformation and unity theory among massages, knowledge and intelligence. An individual knowledge and behavior module is proposed with its behavior description specification. Then, the construction principle of Daemon SoftMan is discussed with the knowledge and behavior unification description using the Generalized Knowledge Representation. Finally, the key implementation of Daemon SoftMan is described based on Linux system, including the boot of Daemon SoftMan, the initialization of SoftMan society and the creation of SoftMan.

To solve the problem of natural language processing for Chinese standard literature, the deep learning technology is employed to build a statistical language model. The Hierarchical Log-Bilinear language model is improved and the unsupervised learning and supervised learning are integrated. In order to accomplish the machine learning, the stacked restricted Boltzman machines are taken to train words' distributed representations, which are taken as the input to a supervised feedforward neural network. The proposed is evaluated using more than one million standard literature bibliographic data. Experiment results show that this model can effectively improve the model's ability to learn the probability of words' distribution.

A REEM algorithm is proposed, which takes the residual energy and communication cost of sensor nodes as two parameters. The role-relationship and member-relationship between nodes are constructed in the algorithm, and information transformation is conducted through these two relationships. The role-relationship and member-relationship are dependent on the residual energy and communication cost of the nodes. The energy-balance of the wireless sensor network is obtained by the conversion of the role-relationship and member-relationship. The algorithm can make use of the node residual energy as much as possible, extend the life cycle of the network, and realize maximum information collection.

A new facial feature point tracking method based on Mind Evolutionary Algorithm (MEA) was presented. The integrated feature model was formed by Scale Invariant Feature Transform (SIFT) feature in tracking region. Then the feature points were determined. The existing region of the feature points were restricted by the space distribution of the integrated feature model. The MEA was used to compute the best-case coverage solution by similartaxis procedure and dissimilation procedure in target frames. It can more precisely determine the restriction region of tracking feature points, and improve tracking precision and efficiency. The original feature and replaced feature were introduced to make the tracking steadier and faster in complicated condition. The experiment results indicate that the new method can stably track 95% feature points in complicated condition, while the tracking speed can be kept withing 25 frames per second.

According to the interaction between local structures of RNA, a new algorithm, named LIFold, is presented to predict RNA secondary structure with pseudoknot. For a given RNA sequence, first, the optimal RNA secondary structure without pseudoknot is generated based on the method of free energy calculation. Then, the pseudoknot stems are obtained using local structure pairing interaction, and an energy calculation model with pseudoknot is established on the basis of the optimal structure. Finally, the RNA secondary structure with pseudoknot is obtained using optimization algorithm. Using the test data of HotKnots, the sensitivity and Positive Predict Value (PPV) reach 84% and 80% respectively. Using the test data based on PseudoBase database, the sensitivity and PPV reach 78% and 73% respectively. Comparing to HotKnots, ILM, PknotsRG IPknot and FlexStem softwares, the accuracy of the proposed algorithm, LIfold, is higher.

Standard K-medoids algorithm has the disadvantages of easy-to-fall into local optima and the clustering effect is commonly influenced by the initial cluster centre. To overcome these shortcomings, a modified K-medoids algorithm is proposed which is based on the genetic simulated annealing algorithm. By combing the genetic algorithm and the annealing algorithm, the global search ability and convergence speed of the proposed algorithm are greatly improved. Comparison experiment results show that the modified algorithm can effectively overcome the shortcomings of the Standard K-medoids algorithm that the clustering efficiency, convergence speed and clustering quality are improved.

To overcome disadvantages of long time consumption and low efficiency when k-mer frequency is used for DNA segment recognition, the attributes reduction of rough set theory is adopted to reduce the k-mer frequency. Signal reduction experiment in the whole genome of 30 microbial strains was carried out. Results show that using this method can reduce 72.27% of the original high-dimensional genetic signals, and increase the accuracy by 0.62%, meanwhile, the running time is shortened by 73.3%.

To satisfy the personalized requirement of privacy preservation, an algorithm for preserving trajectory privacy is proposed, which is based on the technique of Canonical Correlation Analysis (CCA). In this algorithm, the trajectories, which are considered to be insensitive by the data producer, are published directly. The protection of privacy is only imposed on the trajectories, which are considered to be sensitive by the data producer. To this end, a latent variable is first obtained from both the insensitive and sensitive trajectories. Then based on the latent variable, a lot of trajectories are generated to replace the sensitive trajectories. The advantage of this algorithm is that it can not only respect the privacy preserving wish of the data producer, but also obtain high quality trajectories. Experiments on real-world data set of trajectories were carried out, and the results demonstrate the effectiveness of the proposed algorithm.

A novel search strategy for local region based on hierarchical region detection is proposed. All local regions of interest satisfying candidate detection condition are extracted using this strategy. Then the fine plate location, including gradient enhancement, redundant edge removing and plate segmentation, is performed sequentially. Experiments results show that the proposed method possesses strong robustness to environmental noise and provides high accuracy of real-time license plate location.

In order to improve the intelligence of the supervisory system, thus to transfer the supervisory control from device-oriented mode to custom-oriented mode, a monitor model, named Semantic Knowledge Based Supervisory Control Model (SKBSCM), is presented. In this model the Resource Description Framework (RDF) is used to describe data, and semantic knowledge based rules are employed for reasoning. The model automatically integrates the collected data and the user preferred information, perceives users' need, and takes the initiative to generate the most reasonable solution on the environment context. The proposed SKBSCM model is applied to the smart-home field. Experiments on some daily living activities show that the model greatly improves the users' experience feeling.

A new frequency estimation method for harmonic signals embedded in strong chaotic interference is proposed. This method is based on the frequency doubling, wavelet packet theory and cross spectrum theory, and it is composed of three steps. First, using the cycle characteristic of harmonic, the sampling data are equidistantly taken value of the harmonic signals and then construct a new data sequence. The doubling harmonic is deviated from the chaotic centre frequency. Second, the wavelet packet decomposition algorithm is applied to analyze the energy distribution characteristics of the harmonic signals and chaotic signals. Finally, according to the results of second step, the components, on which the harmonic signal energy concentrates, are extracted, then, the cross spectrum method is applied to estimate the frequencies of the harmonic signals. Simulation results demonstrate the effectiveness of the proposed method.

A distributed tracking and fusion algorithm based on Unscented information filter is proposed for nonlinear target tracking in wireless sensor networks. In this algorithm, the unscented transformation is combined with extended information filter to handle the nonlinearity of the target motion and measurement in the framework of information filtering. The Kalman consensus filter is used as distributed fusion structure to combine the estimate of each local sensor node in the sensor networks with constrained topology and limited bandwidth. The efficiency and the superiority of the proposed algorithm are demonstrated by simulation results.

The tracking algorithm based on locality sensitive histograms lacks the function of anti-occlusion and fails to track targets stably under long-time occlusion. To overcome this problem, a real-time tracking algorithm using multilayer template based on locality sensitive histograms is proposed. The algorithm first divides the template into a number of top regions; then estimates the occluded state of each top region; and stops the template updating of the occluded top regions. Because the different reliability of each top region, different update levels are taken to maintain the robustness under varied target conditions. The proposed algorithm is tested with two video sequences and results show that the algorithm achieves stable tracking for target size of 60pixel×60pixel. As compared with the locality sensitive histograms tracking, the proposed algorithm can track stably under illumination change and occlusion.

It is easy for digital image to contain noise. To overcome this problem, an adaptive filter algorithm based on Alpha-trimmed mean filter and Mahalanobis distance is proposed. This algorithm combines the advantages of the Alpha-trimmed mean filter and the feature of the Mahalanobis distance. First, the reference sequence is obtained by the Alpha-trimmed mean filter. Then, the Mahalanobis distance between the reference sequence and the comparing sequence is calculated, which is used to determine the weight coefficient of the pixels in the filter window. Finally, the weighted result is taken as the filter output. The experiment results show that the filtering effect of the proposed algorithm is very high for the image containing Gaussian noise, salt and pepper noise or mixed noise, and can keep the details of the original image.

In order to improve the diagnosis efficiency of faulty circuits with tolerance and to ameliorate the characteristic description of faulty symptom, first, the faulty features of the circuit with tolerance are created by information entropy method. Then, a feature extraction method based on discrete Particle Swarm Optimization (PSO) algorithm is proposed to obtain the optimal feature subset of the faulty features. Finally, the optimal feature subset is used to train the classifier to diagnose the faults of circuit with tolerance. Experiment results validate this proposed method.