The random noise in the desert zone has low-frequency characteristics in addition to nonlinear and
non-Gaussian characteristics． It exists in both analog records and actual desert seismic records． The effective
signal is largely submerged in the noise posing great difficulties to the processing of subsequent data． Based on
these problems，this paper combines the CEEMDAN (Complete Ensemble Empirical Mode Decomposition with
Adaptive Noise) with the ＲOSL(Ｒobust Orthonormal Subspace Learning)． The CEEMDAN algorithm is used to
decompose the desert seismic data． All the modalities decomposed are combined into a new record，and the new
record is subjected to low rank decomposition． Then superimpose all the modes of each channel in the obtained
sparse part to obtain the denoising result． The combination of the two solves the problem that the single low rank
processing has no obvious effect on the desert seismic data，and avoids the problem of choosing the modal of the
CEEMDAN algorithm decomposition． The simulation experiment and the actual data processing prove that the
proposed algorithm has obvious advantages in suppressing the low frequency random noise． The amplitude of the
effective signal can be guaranteed to be more than 85%． The suppression of the surface wave in the actual data
is relatively thorough．

In order to improve the overall efficiency and imaging quality of ghost imaging，the DCT (Discrete
Cosine Transform) which is the quasi-optimal transform in image processing is applied to reconstruct ghost
imaging images with high quality under low sampling rate． The two-dimensional DCT matrix is used as the ghost
imaging detection matrix，and the two-dimensional transformation of the target object is realized by simple
multiplication． The low-frequency information of the image is concentrated in the upper left corner of the bucket
detection value matrix． Simulation and experimental results show that the reconstruction of ghost imaging objects
based on two-dimensional DCT is of higher quality than that based on one-dimensional DCT and Hadamard
transformation． When rapid imaging is needed，only 12． 5% sampling rate is needed to basically restore the
target． This method can reconstruct high quality ghost imaging objects with low sampling rate． It is convenient for
ghost imaging generalization and application．

The construction of parity check matrices for QC-LDPC-C (Quasi-Cyclic Low Density Parity-Check
Convolutional) codes needs to be free of length-4 cycles，and the computational complexity of direct
construction will increase exponentially without considering the characteristics of matrices． The construction for
QC-LDPC-C codes based on periodic filling with offsets of sub-matrices is proposed． With the periodicity of the
base check matrix，the proposed method fills the deterministic sub-matrices to realize fast encoding; during the
construction for random sub-matrices，the optimized selection with offsets of sub-matrices is utilized to ensure
that the obtained matrix after each selection for sub-matrix position and periodic expansion satisfies the
expended structure without length-4 cycles． The base check matrix whose expanded matrix is free of length-4
cycles can be obtained，and the girth of expanded check matrix is at least 6． LDPC-C codes with different
parameters are used to compare with the constructed QC-LDPC-C codes． Experimental results show that the
codes constructed with proposed method can achieve better performance and lower encoding and decoding
complexity．

 In order to reduce the harm to the system caused by high PAPＲ(Peak Average Power Ｒatio) of OFDM
(Orthogonal Frequency Division Multiplexing) signals，a new algorithm ( LS-ACE-POCS) is obtained by
integrating the LS(Least Square) estimation method into the traditional ACE-POCS algorithm． The least square
estimation is used to calculate an optimal factor in new algorithm，then the optimal factor is multiplied by the
de-peak signal，and generates a new de-peak signal，the new algorithm for one iteration can be compared with
the traditional ACE-POCS algorithm for 10 iterations． The Matlab simulation results show that under the same
simulation environment，the LS-ACE-POCS algorithm's ability to reduce PAPＲ is far superior to the traditional
ACE-POCS algorithm，Specifically，the PAPＲ of the new algorithm after one iteration is 0． 223 dB lower than
that of the traditional algorithm in the tenth iteration，in terms of BEＲ(Bit Error Ｒate) performance，a new
algorithm of bit error rate performance slightly superior to the traditional algorithm，In terms of time complexity，
the new algorithm is much better than the traditional algorithm and greatly reduces the complexity of the system．

 The traditional CＲ(Cognitive Ｒadio) optimization algorithm realizes spectrum pooling interference
suppression at the expense of sacrificing the bandwidth and data transmission rate of cognitive users． For
improving the spectrum utilization further，the optimization algorithm based on Tabu is proposed． The Tabu
search is used to obtain the modulation wavelet packet basis with minimal ICI (Inter-Carrier Interference)
energy． The simulation results show the new algorithm can reduce the ICI energy and improve the BEＲ(Bit Error
Ｒate) of the system better comparing with the algorithm of deactivating adjacent subcarriers，the algorithm of
optimal deactivating subcarriers and the algorithm of ICI suppression based on genetic algorithm． The
optimization algorithm doses not occupy the bandwidth of un-LU ( Licensed User) and sacrifice the data
transmission rate of un-LU．

In order to solve the communication problem between the 32-bit MCU(Microcontroller Unit) and the DC
(Direct Current) governor，a communication method and implementation process based on the Modbus bus protocol
is proposed． The 32-bit MCU selects the STM32F103 chip as the communication host，and the DC governor uses
the ETD DC790P as the communication slave． The method can realize the communication structure of a host multi-
slave，and realizes communication between a 32-bit MCU host and a plurality of DC governor slaves． The host
sends a command to the slave through the protocol，and the slave receives the command to make an action and
responds to the host through the protocol． Ｒepeated experiments have proved that the communication method based on Modbus protocol can accurately realize the communication between the host and multiple slaves，and can
accurately realize the control of the motor speed and torque of the 32-bit MCU． And the Modbus implementation
method has the advantages of high stability，strong anti-interference ability and high portability．

Aiming at the problem that the selection and optimization of mechanical fault parameters are neglected
in the current research on mechanical fault diagnosis，which leads to poor accuracy，this paper proposes a sparse
feature similarity measurement method for mechanical fault based on quantum genetic algorithm optimization．
According to the first signal nonlinear mixing，then demixing． With kurtosis as the objective function and
quantum genetic algorithm，the separation matrix parameters and nonlinear de-mixing parameters of blind source
separation are optimized to realize blind source separation of mechanical faults． Based on fault signal processing，
the similarity measurement of mechanical fault sparse features is realized by combining quantum genetic algorithm
with LSSVM ( Least Squares Support Vector Machine)． LSSVM selects the model parameters during the
mechanical fault diagnosis，the quantum genetic algorithm is used to optimize the model parameters of LSSVM．
The parameter selection problem of LSSVM is transformed to an optimization problem，and the optimized LSSVM
classification model is used to realize the classification of mechanical fault sparse feature similarity mode． The
experimental results show that the method can achieve high efficiency blind source separation，high accuracy in
mechanical fault diagnosis and good performance．

In order to realize the intelligent perception oriented CNC (Computer Numerical Control) system，
based on the characteristics of the network communication and open architecture of the intelligent CNC system，
we design a secondary development platform of the intelligent CNC system and the reconfigurable hardware
structure of the system，and put forward the multi-channel signal processing technology，which ensures the real-
time processing，reduces the interference between the signals and improves the processing efficiency． In order to
satisfy the error caused by nonlinearity and time delay in CNC system，the adaptive control of servo motor of CNC
machine tool is realized based on the improved MFAC(Model-Free Adaptive Control) data driving algorithm．
With the SSB field bus developed by ourselves to collect sensor signals，the design of data frame format and
communication protocol is completed to ensure the low power consumption and high reliability of the information
acquisition system． The improved MFAC data-driven algorithm is tested by the test and verification platform
based on blue sky CNC，and good test results are achieved．

Traditional image-based visual servo control needs to calculate Jacobian matrix and inverse of Jacobian
matrix，which is complex in structure，large amount calculation and unsatisfactory real-time performance． The
BP neural network optimized by PSO(Particle Swarm Optimization) genetic algorithm realizes the vision servo
control of“eye on hand”robot by learning the mapping relationship between image feature space and robot
motion space． By optimizing the weight and threshold of BP neural network，the disadvantages of long training
time and slow convergence speed are prevented． The experimental results show that the optimized algorithm has
high efficiency． The designed controller can make the robot end actuator reach the expected position in a shorter
time． The average error between the actual value and the expected value of the motion position of the image
feature points is about two pixels，which has good convergence speed and control accuracy． The relevant
conclusions can provide the basis for the optimization of robot visual servo control and improve the application
performance of the algorithm．

 In order to improve the efficiency of temperature control and deployment，a new optimization algorithm
for the operation and maintenance of integrated power warehouse is proposed． Based on the solution of space
dividing original idea，design solution space segmentation processing mechanism，application solution space
segmentation processing mechanism，the current comprehensive utility tunnel power operations，the different
voltage grade evaluation，the current utility tunnel power storehouse magnetic field，electromagnetic field data
capture are calculated，the natural convection into simultaneous solving differential equation，energy equation is
solved simultaneously to calculate the conduction data of the current power operation and maintenance of the pipe
gallery． Finally rely on the solution space division，the above to calculate the magnetic field calculation of power
storehouse data and the analysis of convective parameters，delimit the molecular space respectively，clear number
and individual number space，the optimal solution is obtained and the operation and maintenance upgrade is
realized． Experimental data show that the local maximum temperature has been reduced by nearly 27%，and the
absolute temperature has been reduced by 32%．

 Ｒecently，land use classification based on machine learning algorithm and remote sensing data has
been a hot topic for scholars at home and abroad． In this study，Sentinel-2 remote sensing image is used as the
data source to classify land use of Longjiang County． In order to achieve high level remote sensing image
classification and effectively remove the information redundancy of high dimensional features，the CatBoost
algorithm is used to reduce the dimension of all feature sets． The feature set is used to train ＲF algorithm and
AdaBoost algorithm． The results show that the Kappa coefficient of CatBoost，ＲF and AdaBoost algorithm are all
above 0． 77，and the Kappa coefficient of CatBoost algorithm is up to 0． 911 4; CatBoost classification is an
effective method of land use classification，which provides a fast and feasible method for the classification of land
types．

 Particle swarm optimization algorithm is introduced into a nonlinear financial risk model to solve the
problem that particle swarm optimization algorithm has low search ability and particle is easy to fall into local
optimization at the later stage of iteration． based on the optimization of inertia weight and the variation of
individual position of each particle，an improved particle swarm optimization algorithm is proposed． PSO
(Particle Swarm Optimization) is used to select the optimal control parameters to reduce the total risk value of
the financial system to the greatest extent． The simulation results show that the improved particle swarm
optimization algorithm is superior to the traditional particle swarm optimization algorithm in terms of global
optimization and search speed．

 In order to improve the recognition rate，the DAG-SVM(Directed Acyclic Graph Support Vector
Machine) algorithm is applied to character recognition，and an OCＲ(Optical Character Ｒecognition) system
based on DAG-SVM，which can be divided into three components is proposed． Algorithms of image pre-
processing，image segmentation and OCＲ are researched． Image pre-processing，the first part，contains image
filtering，graying and tilt correction． And the second component is image segmentation，which contains
binaryzation and segmentation from two sides． The last part is character recognition，whose main algorithms is
researched and DAG-SVM is choosed． The system is simulated by Matlab，and the results show that the
recognition rate is 92． 86% ．

At present，China's fault detection technology for engine vibration parameters still uses traditional
manual experience and technician test methods． It also has certain subjectivity． A wireless three-axis vibration
test system is designed based on MEMS(Micro-Electro-Mechanical System) accelerometer and Bluetooth wireless
communication，and equipped with a host computer，which can be used for engine fault detection and status
diagnosis． First，the vibration curve is obtained by a sensor installed on the outside of the engine． Then it is sent
to the corresponding upper computer through a wireless module，and the corresponding vibration parameters are
judged to determine the operating state of the engine． The vibration system test and actual field experiments show
that the designed vibration test system performance index meets the engine vibration detection requirements，and
can be applied to the monitoring and diagnosis of the engine online fault． The designed wireless three-axis
vibration test system has the characteristics of easy installation，fast and accurate measurement，etc．

To study the dynamic changes in urban expansion in Taiyuan，extracting urban built-up areas using
DMSP/OLS( Defence Meteorological Satelite Program/Oprational Linescan System) nighttime lighting data
combined with statistical data method，weighted center of gravity transfer method is proposed to extract the
center of gravity of a city． The main research areas are the spatial and temporal characteristics of land
expansion，the type of urban expansion，and the shift of city center of gravity based on light intensity． The
results show，Taiyuan City's urban built-up area continues to expand and the city's center of gravity is moving
southward． From 1992 to 2012，the main expansion trend of Taiyuan City is westward and southwest and
enhance the dominant expansion type for filling． From 2002 to 2012，the intensity of expansion is especially
obvious，the main expansion trend is south and southeast and is leading type of extension．

 In order to improve the ability of early warning and analysis of pipeline safety hazards，according to the
requirements of GB32167-2015 oil and gas transmission pipeline integrity management specification and the actual
characteristics of pipe-intensive fields，Type Ⅲ pipelines are further classified． And combined with the geographical
and surrounding residential areas，the high-consequence area of the pipeline is effectively identified，the pipeline
integrity evaluation system is improved． Through Internet + data-driven model a complete oil field pipeline data
management platform is established to evaluate comprehensively the risk of oilfield pipeline operation with a visual
presentation． It improves the accuracy of risk assessment and ensures to predict the risk timely and accurately．

 In order to improve the accuracy and precision of binocular stereo vision device，a calibration
technology based on binocular stereo vision is proposed． According to the characteristics of two cameras in
binocular stereo vision，using the calibration plate of circular symbol array as the image source，through the
preprocessing of the image，under the constraint of radial error，the method of extracting feature points by using
geometric relation of circle and least square data processing is used to obtain coordinate of identification point and
homography matrix of image． The ideal external and internal parameters are inferred． The accurate parameters of
binocular stereo vision device are obtained． This test provides accurate data basis and reliable correction
conditions for the device to acquire measured data．