The electric power steering(EPS) system is characterized by possibility o supplying variable assist steering torque according to the vehicle movement state and driver behavior. Based on this feature, a compensating torque algorithm was suggested for the vehicle with tire blow-out. The dynamic models were built for the EPS system and the tire blow-out, the impact torque on the steering wheel caused by the blow-out tire and ground was calculated. A compensating current was added to the EPS system assist steering current to attenuate the steering shocking by the impact torque. The algorithm was verified by simulation test, and the results showed that it can compensate effectively the impact torque on the steering wheel caused by the blow-out and improve the steering feel of driver.

A model of the accelerator pedal and a dynamic model of whole car with 15 degrees of freedom are established using AMESim software. The conduct awareness of novice drivers is analyzed. The logical threshold algorithm is applied to recognize the misuse of the accelerator pedal. Then the safety assistant system integrated with fuzzy-PID control algorithm for Anti-lock Braking System (ABS) is started-up to realize vehicle active safety. A co-simulation platform is built up using Matlab/Simulink software and AMESim software to test and verify the control algorithm. Results show that the control system has good real-time property, which can meet actual application requirement and improve the safety performance of vehicles.

The aerodynamic characteristics of a heavy-duty truck driving on a bridge were studied by the numerical simulation under 4 crosswind intensities with windspeeds 0,10,20 and 30 m/s. The changes of the air flow field around the truck under different crosswind intensities and under the influence of the bridge parapet were studied. It was found that when a truck driving on a bridge under crosswind, strong vortexs form between the truck and the bridge parapet due to existence of the parapet, and the flow field around the truck changes with crosswind intensities. The aerodynamic characteristic of the truck on a bridge is different slightly from that of the truck on ground, and the drag force is proportional to the crosswind intensity.

Five pictures of auto-inner ornament with different colors are taken and presented on a liquid crystal screen as visual stimuli. Five in-vehicle noise samples are acquired from five types of car under different conditions as evaluation objects. "Luxury" and "Sporty" are taken as the subjective indexes. First, semantic differential experiments are conducted to evaluate ten hearing characteristics of 25 groups of audio-visual combination. Then the subjective evaluation results of sound quality are comprehensively evaluated using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) based on entropy weight. The results show a tendency to consider "blue" as the strongest "luxury" color and "orange" as the weakest "luxury" color; "red" as the strongest "sporty" color and "pink" as the weakest "sporty" color. Statistic results show that colors as the visual stimuli can improve the sound quality of vehicle-interior.

A method to estimate the characteristic impedance and the wave number of porous material was proposed based on in-situ measurement. The feature of the PU probe which can measure the pressure and the particle velocity at one point simultaneously was used to measure the surface impedances of porous material at different boundary conditions by means of the principle of the 2-cavity method, and the characteristic impedance and the wave number were abtained by calculation. Different boundary conditions were compared and suitable boundary condition combination was determined. With the comparison of measurement data and calculation results, the method proved to be accurate and of practical use. Compared to the methods based on the impedance tube, it is more convenient.

An active diesel particulate filter (DPF) regeneration system was evaluated, which applies an in-cylinder late post injection (LPI) and a diesel oxidation catalyst (DOC) to increase exhaust gas temperature for DPF regeneration. The influence of LPI on the performance of diesel engine and temperature rise patterns of DOC were studied by the bench test. The results showed that, at medium and low load conditions (40% and 50% load), LPI produces higher level of unburned hydrocarbon (HC), which is mainly oxidized on DOC to raise exhaust gas temperature to initiate active DPF regeneration. The results should be of certain reference value for optimizing the matching of DOC and DPF, and provides basic information for developing DPF regeneration method and DPF regeneration control strategy.

A concept of Newtonian ideal gas flow was defined and an energy equation for the Newtonian ideal gas flow was derived. Based on the test on a LJ491QE1 gasoline engine, its gas exchange losses were calculated using the above-mentioned equation. The results showed that at low-load conditions of the naturally aspirated port fuel injection gasoline engine, the intake loss power accounts for higher percentage of its effective power output, and the exhaust loss for high-load conditions. The proposed quantitative theory of gas exchange loss of the internal combustion engine is significant for further study on actual cycle losses of the engine.

Changes of underhood flow field and temperature field of at high-speed conditions before and after closure of the left grill of a bus engine compartment were analyzed with numerical simulation and experiment. The results indicated that: the rate of the cooling air flowing into the engine compartment from the left grill is higher than the rate of the hot air flowing out of the engine compartment from the fringe of grill. Before closure of the left grill of the bus engine compartment, the hot air reflected by the back door of the engine compartment can be hampered strongly by the cooling air flowing into the engine compartment. Closure of the left grill weakens the cooling air hampering, therefore, under the vehicle bottom air driving, the hot air flows out of the engine compartment more smoothly. Accordingly, closure of the left grill can not adversely affect the underhood thermal environment.

To study the real-time vehicle load monitoring method, a 2 DoF 4-parameter 1/4 vehicle vibration model was built with an established random road roughness model as the input excitation. The dynamic performance of suspension was simulated using SIMULINK under different working conditions and a global linear relation ship between the dynamic displacement and load of the suspension was obtained. Based on this analysis result, a new method to monitor the vehicle dynamic loading status with multi-sensors was proposed. A dynamic load monitoring and early warning platform was designed and calibrated for a FAW heavy-duty truck with a single-chip computer MC9S12XEP100 as on-board end. The road tests were performed on an A-level road under different speeds and the pre-set position change of goods. The test results showed that the proposed method can real-time detect the change of vehicle loading status, providing a new way for early warning of abnormal vehicle loading status.

In order to break through the limitations of current dynamic traffic data from the Sydney coordinated adaptive traffic system(SCATS) applied only for dynamic signal control, from the perspective of information sharing, the dynamic traffic data collected from the inductive loop on SCATS were applied in the research of automated incident detection algorithm. The characteristics of the data collected from the inductive loop on SCATS were analyzed, and a new algorithm of automated incident detection (AID) based on the data was designed. The validity of the proposed AID algorithm was tested with measured data, and the new algorithm had a better performance on detection rate, false alarming rate and mean time to detection, it would be effective in improving the coordination in traffic signal control, traffic information guide and traffic command of roads covered by SCATS.

To identify the road network traffic flow reliability evolution and the potential trend under sudden disaster, the Rough set theory in combination with the urban traffic state identification and prediction knowledges were used to construct a knowledge representation system for the road traffic network reliability under sudden disaster. The logic reasoning of road traffic network reliability operation trend was realized through a series of operation processes, such as discretization of attribute data of the road traffic network elements, attribute and data reduction of the road traffic network reliability decision making table, etc. The experiment results demonstrated that the road traffic network reliability fits well with the traffic flow charactesistic under sudden disaster. The prediction accuracy of the proposed knowledge representation system for the road traffic network reliability reaches up to 90%.

Aims at the traffic guidance and evacuation problems under large public places emergency, to achieve the maximum volume of traffic and shortest evacuation time. Then set up the Bi-level programming of traffic guidance and evacuation and build the OCMPSO based on sort selection to study the Bi-level programming. Finally, simulate the large-scale activity in a Spots Centre. The results show that the model proposed in this paper can evacuate the vehicles under large public places emergency efficiently and safely.

Due to complicacy, randomness and nonlinearity of traffic system, traffic flow parameters are usually considered as the random time series. A method based on G-P algorithm that is helpful to convert the traffic flow time series to stationarity series was proposed. First, the autocorrelation function was adapted to evaluate the stationary of traffic flow under free traffic, congested traffic and jam traffic. Second, G-P algorithm was used to calculate the embedding dimension, reconstruct the phase space, and convert the traffic flow non-stationary parameters to stationary time series under 3 traffic states. At last, some examples illustrated the model and showed its practical applicability based on measured traffic flow data. The research can provide the input set for the traffic flow parameters analysis, fitting and prediction.

A phase scheme was built for the typical 3-leg signalized intersection containing 6 vehicular phases and 3 pedestrian phases. The concepts of master phase and the slave phase were introduced, and the mechanism of phase scheme design was explained from the phase numbering, phase combination, phase switching, and phase display sequence. For the 3 master phases with slave pedestrian phase, the numerical relationship among the minimum green time of the master phase for the pedestrian crossing safely the street and the phase scheme, the minimum pedestrian walk time interval and the needed intergreen time interval were analyzed. Based on this analysis, and absolute minimum green time calculation dedicated to the 3-leg intersection was proposed. It provides a strong support for the pretimed signal timing plan and the development of traffic response control strategy.

The optimization of average passenger waiting time for transit transfers with vehicle capacity constraints was studied. In a certain cycle time, the transfer waiting time for different transit lines with vehicle capacity constraints is minimized by adjusting transit schedules in a transit network structure. And a solution based on genetic algorithm was proposed to the optimal model. The results show that the model could be relaxed to the model proposed by Shafahi while the vehicle capacity is enough to transfer demands without the vehicle capacity constraints. The average transfer waiting time with vehicle capacity constraints could increase by 370% compared to Shafahi's result. The results illustrate that the vehicle capacity is a key factor to take into account in the optimization of transit transfer because it influences the passenger transfer time significantly.

Based on the space-time prism method, the traveller activity space was limited in a definite space-time prism, the accessibility of activity zone in space was defined under time constraints according to the activity plan of the traveller. A 2-step modeling strategy was proposed for the destination choice behavior. An attracting ratio of the zone was introduced, and a destination choice behavior model with a non-liner utility function was built, and the quantification process, the parameter calibration and the result check of the zone attracting ratio were discussed. The results showed that the destination choice set can be defined by the time-space constraints of the travel in combination with the GPS technology. The non-linear utility function with the zone attracting ratio can predict the destination choice behavior more accurately. The conbination of the theory and the real behavior of the traveller provides a new idea for modeling of the destination choice behavior.

Under a maximum welfare objective, an urban rail transit system is optimized subject to various financial constraints. For the system, the decision variables including fare, headway, route spacing, and service zone area are optimally solved to analyze unconstrained, break-even and subsidy cases. Numerical results and sensitivity analysis are presented in the paper. From the numerical evaluation, it is shown that the optimal welfare curve is very flat over a wide range of subsidies. Thus, with the welfare maximization objective, the break-even transit subsidy policy may be preferable for the urban rail transit system. The results derived from this study can support effective decision-making on urban rail transit systems in areas that may experience significant shifts in residential density, as well as geographic or physical changes in their street networks.

In order to simulate the pedestrian flows in the normal conditions realistically, the issue of the social force model, a simulated pedestrian continuously pushing over other pedestrians, was solved by introducing a pedestrian deceleration avoidance mechanism and analyzing the pedestrian required space. The improved model was simulated by program compiled with C++ language to validate its effectiveness. The simulation results showed that the pedestrian can avoid other pedestrians or obstacles effectively. The simulation data were fitted by means of the generalized regression neural network, and an obtained velocity-density diagram of the pedestrian flow was compared with its real data. The compared result showed when the pedestrian density in the range of 0~3.5 p/m², the proposed model can reproduce the basic behavior of the pedestrian flow.

Though Burgers model has been widely used to analyze viscoelastic creep properties of asphalt mixture, there are some disadvantages such as subjective determining time segmentations for parameters identification of the model. An optimum time segmentation identification method for Burgers model was proposed in this paper. The optimum time segmentations were determined according to the principle of least residual sum of squares. Its stability and versatility were verified by theoretical analysis and experimental test. The parameters of Burgers model can be identified more accurately by this method. It will be helpful to predict the viscoelastic deformation more accurately and evaluate creep performance of asphalt pavement.

In order to assess the working condition of the bridge structure system(BSS), based on the study results of the BSS components, a new method was proposed to assess the BSS reliability. The Markov process steady-state probabilities were used as the indexes for assessment of the BSS reliability, the relations between Markov process steady-state probilities and BSS components reliability indexes were established and a relational matrix R of the BSS reliability was deduced. For the real BSS, the system state transfer matrix may be established. Given the reliability index of every conponent of the BSS, the relational matrix R of the BSS reliability can be obtained. The steady-state probability of every state,i.e. the reliability assessment index, of the BSS can be derived by the Markov process steady-state probability calculation equation. An example showed that the proposed method is feasible and effective, providing a theoretical basis and practical method to BSS reliability assessment in the design and operation processes.

The fracture toughness characteristic of the high-strength steel at low temperature was studied. The crack tip opening displacement δ_m of construction steel Q460C with thickness 14 mm was tested at low temperature, and its fracture appearance was analyzed microscopically by an electronic microscope. It was found that at temperature -40 ℃, the fracture of the 3-point bending specimen of steel Q460C shows an obvious brittle fracture mechanism. The δ_m value of steel Q460C tends to decrease with the temperature reduction, which is less than those of steels Q235, Q345 and Q390, so the steel Q460C is characterized by the poorest fracture toughness among them. A Boltzmann function fitting analysis was performed to the test data, and the ductile-brittle transition temperature and its Changing behavior were obtained. The results showed an obvious cold brittle characteristic of the high-strength construction steel Q460C.

In order to increase the strength and anti-deformation capability of seaside saline soil, it can be reinforced by wheat straw or solidified by lime respectively or jointly. Unconfined compressive test for heavy compaction samples of 152 mm in diameter and lightly compaction samples of 102 mm in diameter, as well as triaxial shear test for samples of 61.8 mm in diameter have been carried out in succession of axial strain at failure of reinforced saline soil and reinforced lime-saline soil be calculated for analyzing its anti-deformation capability. The results indicated that anti-deformation capability of reinforced soil gradually weakens with the increase of water content of soil, but enhances gradually with the increase of reinforced length and reinforced ratio of wheat straw, and then decreasing after surpassing certain reinforced length and certain reinforced ratio. The optimal anti-deformation capability of the reinforced saline soil depends upon the reinforcement conditions by wheat straw. The suitable length is 50 mm for the heavy compaction sample, its suitable reinforced ratio is 0.25%. For the lightly compaction sample, they are 30 mm and 0.25% respectively. For the reinforced by lime soil sample of 61.8 mm in diameter with uniform reinforcement in the upper part, entirely, and in the lower part, they are 20 mm and 0.25% respectively. The joint reinforcement by wheat straw and line enhances the strength and anti-deformation capability of the seaside saline soil.

The microstructures, penetration and mechanism of the welding joint by twin wire Metal Inert Gas (MIG) welding with Cr₂O₃ activating flux were studied by using SEM, XRD and optical microscopy. Results show that, in Hot Affect Zone (HAZ), the grain size with the flux is bigger than that without the flux. The grain size increases with the flux but the growth rate gradually slows down. The penetration depth into the welding joint increases sharply with the flux first then slows down to a stable state. More welding heat input and less Mg₂Si content are the main reasons leading to the growth of grain size. The phase compositions of the joint with flux do not change compared to the joint without flux, which are still α-Al and Mg₂Si. The arc constriction and higher arc temperature are the main factors for the deeper penetration into the welding joint.

Based on the experiments of laser welded ultra-high strength steel with thickness of 1.8 mm, the effects of the welding speed on microstructure and microhardness of the weld seam were studied. The results showed that microstructures of the weld seam are the lathy martensites. The picric acid solution was used to display the boundary of the prior austenite grain, and the grain exhibits the columnar crystals. With the increase of the welding speed, the grain sizes of the prior austenite decrease and the fine cellular dendrites in the center of the weld seam change into the equiaxed grains. The microhardness of the weld seam increases with the welding speed.

The characters of flexible discrete clamp stretch forming (DCSF) machine were introduced. A FE model of flexible DCSF was built. Spherical surface and saddle surface were selected for simulation. The influences of stretching velocity, friction coefficient and clamp shape on the forming results were researched. The numerical simulation results showed that the stretching velocity 1 m/s is good for satisfactory forming efficiency and accuracy. The better the lubrication condition is, the more homogenous distributions the stress, strain and thickness are. The clamp shape is sensitive to liquidity of transitional area and clamping area, more uniform strain and smaller thinning value were got by circular clamp. The simulated results are validated by experiment. The obtained achievements are useful for the optimal design of DCSF machine and the practical stretch forming process implementation of sheet metal.

Based on the linear infinitesimal deformation thermoelasticity model Duhamel-Neumann equation, the effects of temperature factors on the residual stress of plastic parts manufactured by the rapid heat cycle molding(RHCM) were studied. An on-board blue-tooth pront shell mold and a set of accurate temperature control device were developed, and the effect of mold temperature in RHCM on warpage of the part was investigated. The results showed that the mold temperature should be under an upper limit. Under the same mold temperature, for the parameters set in the feasible range, their effects on the warpage of RHCM part decrease according the the following order: pressure keeping time, pressure keeping value, meet temperature, injection pressure, and cooling time. With the increase of mold temperature, part warpage decreases quasi-linearly. The warpage shows a modest decrease with the increase of meet temperature and injection pressure, and demonstrates a small V shape fluctuation with the increase of pressure keeping value, pressure keeping time and cooling time.

The friction and wear behaviors of the carbon fiber reinforced friction material (CFRFM) were investigated using a D-MS tester. The microstructures and constitutions of the CFRFM surface layer were measured using scanning electronic microscope and energy dispersive spectroscopy respectively; and their influence factors were analyzed according to the experiment results of thermo gravimetric analysis and tribological properties. The experimental results showed that the tribological properties of CFRFM do not reduce evidently, and friction coefficient has a small amount of increase during temperature rising from 100 to 300 ℃. The surface of CFRFM consists of three layers: iron-rich layer, which plays an important role in improving tribological properties of the material surface, thermodynamic loosened layer, and deformation reinforced layer. High heat conductivity, favorable to decrease the depth of the heat affected layer on the friction surface, has an important effect on the surface structure of CFRFM, under present experimental conditions, the depth of the heat affected layer is about 0.55 mm.

A fast crashworthiness optimization was performed for the car door as a major energy absorber during the car side collision. A mathematical model was built for the crashworthiness optimization, selecting the plate thicknesses of the major components as the design variables, increase of the energy absorbed by the door structure as the objective function. The uniform test design was taken to distribute the sample points rapidly and reasonably. The polynominal regression response surface method(RSM)was applied to build the precise approximate model of the original optimization problem. The particle swarm optimization(PSO) algorithm was used to optimize the approximate model. The results showed that the proposed method is feasible and effective in car crask worthiness optimization which is valuable for the vehicle passive safety analysis.

The parallel algorithm was presented for enhancing the statistics processing efficiency of thousands of billions aircraft load spectrum testing data. The data processing was multi-level parallel analyzed, and the coarse-grained, medium-grained, fine-grained level parallel processing modes were formulated. Two parallel algorithms were proposed ultimately, one is coarse-grained and medium-grained parallel algorithm based on aircraft type data flow, and the other is medium grain and fine-grain parallel algorithm based on flight data flow. Comparison test based on the small-scale SMP (Symmetrical Multi-processors) cluster shows that the algorithms can improve statistical processing efficiency greatly. A highest speedup of 5.82 in 8 cores can be achieved. The algorithms present a creative method for the complex scientific computation and enhance the computing efficiency in real testing data processing of load spectrum.

A cache system for vehicular central navigation system based on two-level map partitioning is designed and implemented. The cache system includes index files, data files and relevant programs. The structures of the index file and data file as well as the structure of cache for sparse matrix are designed. The cache system manages cache data by using FIFO algorithm. Experiments are carried out to verify the proposed cache system. Results show the cache system can accelerate the response of central navigation system and reduce the amount of data transmitted across the network.

A position tracking control scheme was proposed for a common class of electro-hydraulic servo system based on the Lyapunov direct method. A simple Lyapunov function was constructed for the system and specific methods were given to decide the related parameters and to get the control law. The system stability was ensured symptotically and the scheme could be applied to all electro-hydranlic servo systems with this kind of mathematical model. Taking a super low-speed hydraulic press system as an example, the proposed control scheme was applied successfully. The simulation result proved the validity of the proposed control scheme.

The control issue of strict-feedback nonlinear systems with non-periodical time-varying uncertainties and unknown control gain function is investigated. An adaptive iterative learning controller is developed. In the controller design, an iterative neural network is proposed to eliminate the influence of non-periodical time-varying uncertainties. The backstepping and robust adaptive control techniques are combined to solve the design difficulties brought by strict-feedback structure and unknown control functions. Stability analysis indicates that all state variables are bounded and the output tracks the desired trajectory in integral sense. Simulation results verify the effectiveness of the proposed scheme.

Sliding mode controller based on neural-network is presented. The controller drives the picking manipulator of two degrees of freedom to achieve position control. The neural-network controller mimics an equivalent control law in the sliding mode control. A robust controller is chosen to curb the system dynamics on the sliding surface, thus the asymptotic stability can be guaranteed and high-accuracy position tracking control can be achieved. An adaptive law based on Lyapunov stability theory is employed to estimate the uncertainties of the system. Numerical simulations illustrate the excellent position control performance of the picking manipulators, which can fulfill the requirements of engineering application.

In order to solve the hysteresis nonlinearity of magnetic shape memory alloy actuator, a hysteresis model that utilizes the principle of the PI model is proposed. This model is composed of a number of simple linear operators called linear-play operators. As the structure of the model is similar to the structure of the neural network, the method for training the weight of neural network is introduce. In order to improve the real-time characteristic of the control system, forgetting factor recursive least-squares algorithm is adopted to train the weight of the model. Experimental results demonstrate that the modeling error is 0.0015 mm, mean-square deviation is 2.2931?10^-4, and the maximal error rate is 0.1593%. The proposed modeling method is effective and could obtain higher accuracy.

Based on experimental measurement, the hysteresis modeling of Shape Memory Alloy (SMA) actuator is studies by combination of Preisach model. First, a testing apparatus is fabricated inspired by output displacement, which is generated from temperature-induced phase transformation occurring in SMA device due to the applied electric flow. As a prerequisite of the theoretical hysteresis modeling, the data sets are sampled from the experimental results by recording the input and output of SMA actuator during motion process. Second, the hysteresis phenomenon happened to input voltage, temperature and following displacement due to shape memory effect is discussed. Finally, with the introduction of Preisach model, a modified version of classical Preisach model and formulation of numerical implementation are derived to elucidate the hysteresis of temperature-displacement. Results show that the SMA hysteresis curve obtained by simulation with Matlab toolbox matches well with the experimental data, which can be taken as reference for future application.

To overcome the weakness of traditional control strategy for combustion system in power plant, this paper proposes a design of H_∞ robust controller which takes the radiant energy as the intermediate variation of cascade control. The procedure of the controller design is presented. The stability and robustness of the controller are analyzed by simulation. Finally, the designed controller is compared with the fuzzy PI controller which is based on self-turning for radiant energy. Simulation results show that the designed controller has good robust stability and robust performance.

A parameter identification method was proposed for the dynamic battery model based on the genetic algorithm(GA) and the extended Kalman filter(EKF) algorithm. An electrical equivalent circuit model was built for the lithium-ion power battery. In the model 2 RC networks describe the electrochemical activation process and the concentration difference polarization process of the battery respectively, and the hysteresis voltage characterizes the difference of equilibrium voltages between the charging and discharging processes. The coupled parametes of the model, i.e. the RC network parameters and the hysteresis coefficient were identified by the EKF algorithm with minimal mean square error estimation effect. For the problem of the EKF which is difficnlt to get the optimal filtering effect by test adjustment, the GA was used to optimizing the EKF noise matrix. The results of simulation and experiment showed that the proposed battery model parameter identification method meets the requirement of the electric vehicle on simulation precision.

Images of the same scene, respectively obtained from visible and infrared sensors, are different in the high-frequency details. Based on this character, first, wavelet transform is imposed on the infrared image and visible light image that are in register, and the edges of the two images are extracted under multi-scale. Second, local module square and its ratio are respectively considered as the activity measure and match measure, and the image edge feature is considered as fusion strategy. The fusion image is obtained through synthesis module and multi-scale inverse transform. Finally, the fusion image is enhanced and assessed. Under the wavelet domain, image edge extraction, image enhancement and image fusion are combined. Experimental results show that high quality visible and infrared images are obtained under the objective evaluation criteria in the absence of standard reference image.

A Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) algorithm for recognizing target variants is developed. This algorithm uses the local texture similarity between the variant and the original target for recognition. First, a SAR image registration algorithm based on clear edges is proposed. Then, the texture characteristic, which is obtained by combining the Gabor transform, LBP and spatial domain histogram, is employed to describe the SAR image. Finally, histogram sequence matching based on the large characteristic is used to perform recognition. The effectiveness of the proposed algorithm is verified by experimental results on MSTAR S2.

First, to get rid of the unstable and false corners, the candidate matching points are obtained with the corrective constraint matching method. Second, based on the stability of the corners in the process of affine transformation, the images are initially matched. Finally, by setting the condition of coplanar point projective constraint, the relative highest accurate matching points are selected as the best matching corners. Experiments show that the best matching points selected by the proposed method are robust to the added noise and interference, and ideal registration effect can be achieved even for real images with serious added noises and interference.

A new scheme to detect space-time interest points based on cumulative entropy difference is proposed for human action recognition and video analysis. First, periodic space-time features are detected. Second, the concept of cumulative entropy difference is put forward as the criterion to evaluate the interest points. Third, the points with high cumulative entropy are selected as the final key points. Finally, c-means cluster analysis is applied to find the features, and the prototype of the key video is obtained. The video is converted to the feature vector presented by the prototype; then the video classification is conducted. Experimental results show that the cumulative entropy difference evaluation can effectively remove the noise cuboids, and the method based on the evaluation can realize human action analysis and facial expression recognition.

The Smallest Univalue Segment Assimilating Nucleus (SUSAN) edge detector is prone to false-edge problem in noisy environment since it only utilizes the size of USAN area. To solve this problem, a novel edge detector based on feature fusion of the USAN area is proposed, which is called FUSAN edge detector. First, the FUSAN edge detector finds candidate edge points using the size information of USAN area. Then, directly and indirectly using the shape information of USAN area, the FUSAN edge detector computes two edge responses. Finally, the two responses are fused using simple strategies to get the final edge response. Comparing to SUSAN edge detector, FUSAN edge detector utilizes both the size and shape information of USAN area. Synthetic and Cameraman images are used to evaluate the performance of FUSAN edge detector. Experimental results show that FUSAN edge detector outperforms SUSAN edge detector.

An iterative receiver scheme cascading Turbo codes with Continuous Phase Modulation (CPM) is proposed for low Signal to Noise Ratio (SNR) and narrowband applications. According to Rimoldi decomposition model of CPM, the purpose of sharing-information and correcting-complementary is achieved by combining the inner iterations with outer iterations. Turbo iterative detection is introduced and Soft-Input Soft-Output (SISO) demodulation and decoding algorithms are adopted. For reducing the iteration delay, a novel dynamic iteration stopping algorithm is presented based on A Priori Probability Decision (APPD). The likelihood at different iteration time is derived and analyzed. Turbo-CPM system model is built and simulation is carried out. Simulation results show that the error correction performance of the proposed iterative scheme is much better than that of the pipelining cascade at very low SNR. Meanwhile, the proposed stopping algorithm can not only reduce the iteration delay, but also improve the system convergence and reliability.

In order to solve the problem of estimation error of keyword weight, an improved model of second KWS is proposed. This model is based on the Context-Dependent Maximum a Posteriori Probability (CDMAP) method. Meanwhile, a dynamic match phone-lattice algorithm is introduced. This algorithm can efficiently solve the problem of inherent phone recognition error, which adversely affects the recognition accuracy. Experiment results show that the system performance is greatly enhanced by the dynamic match phone-lattice algorithm for both robustness and detection accuracy.

To solve the femtocell pilot power allocation problem in next-generation cellular network, a novel optimization solution for femtocell pilot power based on immune memory clone algorithm is propose. The mathematical model of the femtocell pilot power allocation problem is established. The framework of the immune memory clone algorithm is given. Simulation experiments results show that, based on the network topology and communication flow distribution, the proposed optimization solution can effectively optimize femtocell pilot power allocation and femtocell radius, and it can be applied to solve real engineering problems.

The current communication network management benefit could not be exerted, which is aroused by disadvantaged construction and running management. In order to solve these problems, the concept of communication network pedigree and model is put forward, which is based on the relationship of network loading and network correlation. The reachable effect factor arithmetic based on DEMATEL & ISM is contrived. This method takes the complexity of systems into account and shows the definition that qualitative problem can be solved in quantitative analysis. The objectivity and efficiency of this method is verified through network pedigree analysis which includes network administrative level and effect degree.