Aiming at the problem that the sample covariance can not replace the statistical covariance when the
number of samples is small，a novel target detection method based on Riemannian manifold is proposed for
monostatic MIMO( Multiple-Input Multiple-Output) radar． It takes advantage of the fact that the THPD( Toeplitz-
Hermitian Positive Definite) matrix can form Riemannian manifold in signal space． The single-snapshot THPD
covariance matrices of received signals and noises are respectively generated through the burg algorithm，and the
Riemannian mean of the noise THPD covariance matrices is calculated． The Riemannian distance between the
THPD covariance matrices of received signal and noise is taken as the detection statistic． The method can
increase the dissimilarity between the received signal and noise on the Riemannian manifold． The simulation
results indicate that compared to the traditional method which uses the Euclidean distance，the proposed method
can significantly improve the target detection performance under low SNR ( Signal-to-Noise Ratio) and singlesnapshot．

In order to solve the performance degradation problem that may be introduced by typical analog
circuits in the design stage，the research progress of circuit virtual verification technology is analyzed at home and
abroad，the mechanism models such as HCI ( Hot Carrier Induced ) ，NBTI ( Negative Bias Temperature
Instability) and TDDB ( Time Dependent Dielectric Breakdoun) which cause the performance degradation of
analog circuits are deeply studied． Furthermore，three key tasks are carried out which contains the design of
performance degradation simulation process model，the derivation of the failure physical model and the extraction
of model parameters． Based on the research，a typical analog operational amplifier is selected to simulation the
performance degradation by Cadence spectre software． The performance degradation of the device proves that the
proposed simulation method is feasible at 25 ℃ after two years and ten years of operation． This conclusion can
provide preliminary support information for the improvement design work in the design stage of the analog circuit．

Aiming at the great influence of atmospheric turbulence on the performance of space optical
communication system，a scheme of atmospheric turbulence compensation based on mode diversity combined with
MRC ( Maximum Ratio Combining) algorithm is proposed． Different modes are used as independent channels for
spatial diversity reception，and the maximum ratio combining algorithm is used to realize the diversity and union
of multi-mode channels． In the aspect of MRC algorithm implementation，the phase and signal-to-noise ratio
measurement in different branch weighting coefficients are transformed into the same frequency and phase
adjustment in digital domain and the calculation of signal-to-noise ratio weight coefficients，which simplifies the
hardware implementation difficulty． Applying VPItransmissionMaker simulation platform and combining with
Matlab programming，a simulation system is built to verify the feasibility of this method． The simulation results
show that this method can compensate the effect of atmospheric turbulence on the performance of air-optical
communication system，and the compensation effect for strong turbulence is slightly better than that for weak
turbulence． The more the number of mode diversity，the better the compensation effect．

Freeways need to be closed or speed-limited in ice and snow conditions，which may lead to problems
such as reducing road capacity and vehicle safety． In order to solve this problem，this paper proposes a variable
speed limit method for freeway under ice and snow conditions based on the characteristics of road alignment，ice
and snow conditions and traffic flow characteristics． A velocity generation model including road alignment and
snow conditions is established，and a fuzzy control system is introduced to predict the snow condition． The
simulation experiment was carried out by Vissim software in some sections of the freeway． The results show that
the variable speed limit method proposed in this paper can dynamically adjusts the speed limit value according to
the snow condition． It reduces the vehicle delay compared with the control groupⅠ ( fixed speed limit) and
improves traffic efficiency and road capacity compared to control groupⅡ( road closure and snow removal + no
speed limit) ． It shows that compared with the traditional method，the proposed method performs better ensuring
the efficiency and safety of highway vehicles under ice and snow conditions．

In order to solve the problem that the headlights of the high-speed EMUs( Electric Multiple Units) can
not be restarted during the driving，the faults are locked in the 110 V bus system and the network control system
activates the circuit breaker after removing the influence of the high-voltage power supply line，380 V bus and
load side． Using computer control，frequency conversion voltage regulation，sensor measurement and other
technologies，combined with host computer control，a set of test systems is built． After verification，the system
can quickly find out the source of the fault．

In order to ensure the structural reliability and flight safety of the aircraf，it is necessary to carry out
the dynamic reliability andysis and life prediction of the wing structure of the near space vehide． Based on the
force analysis of the wing，a new dynamic reliability analysis method of the wing for a near space vehicle is
presented with considering of random loads and strength changes． Initially，the calculation of the shear stress，
tension and compression stress，and equivalent stress of the wing section is analyzed． Then，the number of
aerodynamic loads is characterized by Poisson stochastic process． Based on the structural strength analysis of the
wing and strength interference theory，a new scheme of describing aerodynamic loads with truncated normal
distribution is proposed，a dynamic reliability model of the wing is established，and the reliability indexes are
provided． The influence of strength degradation and flight dynamics on the reliability is analyzed，to present the
basic requirements for ensuring the vehicle’s structural reliability and provide the reference for designing
structural reliability controller． The simulation results show that in order to ensure the structural reliability of the
aircraft in actual flight，we should try our best to avoid excessive increase of flight speed，negative angle of attack
and high angle of attack，and appropriately increase the flight height if conditions permit．

In order to detect weak signal，a phase sensitive detector based on voltage frequency conversion is
proposed． The voltage to be measured is converted into frequency through the voltage frequency conversion
circuit，and the frequency signal is counted and digital filtered by the single-chip microcomputer according to the
reference signal，so as to realize weak signal detection． The related detection principle is analyzed，the hardware
and software design and implementation of the system are completed，the verification experiment and error
analysis are carried out，and the correctness and feasibility of the scheme are verified． The system is simple in
principle，low in cost，reliable，convenient and practical，and has a broad application prospect in weak signal
detection．

In order to save the time and cost of FPGA ( Field Programmable Gate Array) upgrade，we design a
FPGA program method that uses UART ( Universal Asynchronous Receiver /Transmitter) to replace the traditional
JTAG( Joint Test Action Group) method． The design is mainly composed of Xilinx FPGA，UART chip，Flash
chip and serial port connection line． By integrating the MIcroBlaze processor，ICAP ( Internal Configuration
Access Port) ，IP( Intellectual Property) core and UART control module into the FPGA chip，the system on the
programmable chip is set up． At the same time，MultiBoot double mirror technology is adopted to achieve that
even if the update fails，the backup image can still be loaded to ensure the normal work of the system，so as to
ensure the stability of the design． Experimental results show that this design can replace the traditional FPGA
upgrade method and save the upgrade time and cost． This design has the advantages of high update efficiency，
low maintenance cost，high stability，and can be used for FPGA remote update．

Aiming at the background of wide application of high-resolution scientific camera and large domestic
demand，a domestic high-resolution scientific CMOS( Complementary Metal Qxide Semiconductor) camera based
on GSENSE400 image sensor of Gpixel company is designed． The camera system includes data acquisition，
control and Camera-Link output design based on FPGA( Field Programmable Gate Array) ． The FPGA mainly
includes SPI ( Serial Peripheral Interface) configuration module，CMOS timing driver module，data acquisition
module，camera link data conversion module and serial communication module． According to the timing logic of
CMOS，the timing design of CMOS driver is implemented in FPGA． According to the amount of HDR( High-
Dynamic Range) image data output by the camera，in order to simplify the design of FPGA data transmission
module，using a DS90CR287 chip，the base mode of Camera-Link is selected for image data transmission，and
the serial port control of the camera is realized． The imaging test is implemented for the designed camara system．
It can output 24 frames continuously in HDR mode with 2 048 × 2 048 pixels，low readout noise，high
sensitivity，high dynamic range image，and basically meet the requirements of scientific level imaging conditions．

In order to study the signal generation and signal processing methods of the rotational accelerometer
gravity gradient instrument，a semi-physical simulation system for the gravity gradient instrument is designed．
First，the gravity gradient anomaly caused by an anomalous body in the ground is calculated according to the
measurement principle of the gravity gradient instrument． Using the output formula of the accelerometer，the 12
accelerometers’output signals of the full tensor gravity gradient instrument are simulated by LabVIEW． The
simulation signal is demodulated to obtain the full tensor gravity gradient value． A single disc is taken as an
example for semi-physical simulation． After the acceleration signal generated by computer A passes through the
hardware circuit，the median filtering and demodulation of the conditioned accelerometer signal by computer B
will obtain the abnormal value of gravity gradient with an accuracy of 100E． The design of this simulation system
can demodulate the gravitational gradient of the full tensor，which breaks through the limitation of demodulating
single disc signal． The introduced semi-physical simulation increases the reliability of the algorithm and the
credibility of the simulation． The physical manufacturing has reference significance．

In order to solve the problems of high performance requirements and complex hardware circuits in
existing AC ( Alternating Current) voltage amplitude measurement methods． A voltage /frequency converter is
used to modulate AC voltage amplitude information into a digital frequency signal． By analyzing the digital
frequency signal，the AC voltage amplitude before modulation can be calculated． The mathematical relationship
between the output frequency of the voltage-frequency converter and the input AC voltage amplitude during the
conversion period is deduced in detail． An AC voltage amplitude detection circuit is designed which contains only
a voltage /frequency converter chip and a double operational amplifier． The experiment results show that the
method is correct and the circuit design is reasonable and effective． The method utilizes the easy integration
characteristics of digital frequency signals，and each test value is used to calculate the amplitude of the signal to
be measured． The simple circuit design reduces the design cost and improves the measurement stability．

SVM ( Support Vector Machine) is a linear classifier with the largest interval defined in the feature
space． Values of these parameters determine the learning performance and generalization ability． In order to
solve the problem of SVM parameter selection，the modified vortex search algorithm is used to find the optimal
fitness function． The simulation results show that the modified vortex search algorithm is an effective SVM
parameter selection method． It is beneficial to jump out of the local minimum，and its performance is not lower
than the vortex search algorithm．

To overcome the shortcomings of the traditional popular graph-based sorting algorithms，only using the
image boundary as the background query，the accuracy of the query selection directly affects the results of the
algorithm，an improved algorithm is proposed． Based on the detection results of current algorithms，the selection
of foreground and background seeds is optimized． Firstly，super-pixel segmentation is performed on the image to
make full use of the middle-level information of the image． Secondly，the image saliency map is calculated with
the popular ranking algorithm． Finally，the saliency result is processed to select better query points and obtain
the final saliency map． Compared to eight algorithms on CSSD( Complex Scene Saliency Datase) and ECSSD
( Extended Complex Scene Saliency Datese) datasets，the experimental results show that the proposed algorithm
has higher detection accuracy．

Characterized by rapid urban population growth and rapid expansion of urban land use，the urban heat
island effect is becoming more and more significant． To quickly extract urban surface temperature information and
explore the spatial and temporal distribution of urban heat island effect，based on the medium-resolution
conventional and easily available radiation information of Landsat images in 2000-2018，the Changchun，a
typical northern city with a significant heat island effect is taken as example to be explored． Using the single
window algorithm to calculate the surface temperature of the images in different years and generate the thematic
map of Changchun surface temperature． The experiment is combined with the ground observation temperature
data in the corresponding period to identify the temporal-spatial changes of the heat island effect in Changchun
and to explore the regular of change and influence factors of the intensity of the heat island effect in typical areas．
The results show that the temporal change characteristics of the heat island effect in Changchun during the study
period are strong at night and weak by day; strong in autumn and winter，while weak in spring and summer．
Spatial variation show obvious spatial differences． But a whole， the high temperature zone increases．
The relationship between the heat island effect of the land type: Building land ＞ Green land ＞ Forest land ＞Water
body．

The traditional network resource behavior detection method can not accurately determine the symbol
observation of resource behavior，resulting in poor detection effect and severe limitation of overall detection．
Therefore，a new detection method of public resource abuse based on hidden markov model is proposed． The
hidden markov model is constructed to determine the current observation symbol sequence through data analysis．
And the behavior detection is based on the information of Windows operating system． A set of sensitive file
information is constructed and distributed it in system sensitive folders． On this basis，the current hidden markov
model parameters are determined． The probability difference value of corresponding sequence is calculated，and
the actual threshold value is set to determine the resource abuse behavior according to the current information
security requirements of public resource network． The simulation results show that the monitoring method is
reallistic and effective，and has higher generalization and application performance．

In order to solve the problem that the traditional compression sensing reconstruction methods of core
image are prone to losing details at low bit rates，a block compression sensing reconstruction algorithm based on
K-SVD( K-Singular Value Decomposition) dictionary learning is proposed． Firstly，according to the theory of
block compression sensing，the core image is divided into multi-blocks，and the corresponding blocks are
observed by Gaussian random matrix． Then，the initial solution is obtained by MMSE ( Minimum Mean
Squareerror Estimation) estimation and the adaptive threshold is calculated by global threshold formula with the
lifting wavelet decomposition． Finally，the sparse representation method of K-SVD dictionary learning combined
with Landweber iteration is used to achieve compression and reconstruction． The experimental results show that
the reconstructed image can retain the texture information of the core image better than the traditional methods at
the same sampling rate，and the PSNR( Peak Signal to Noise Ratio) of the reconstructed core image is increased
by about 0． 1 ～ 0． 8 dB．

In order to solve the problem of environmental risks caused by sudden environmental risks，an
environmental risk assessment model is proposed． Certain area in Northeast China is taken as an example． By
collecting data on relevant environmental risk enterprises，using gridded method an environmental risk assessment
model is built． Based on the ArcGIS tool，spatial analysis of regional grids is performed to map regional
environmental risks． Through analysis and screening，it is determined that the high environmental risk areas in
the study area are concentrated in the east near the river and the central region． After detailed investigation of the
industrial structure and ecological layout of environmental risk enterprises in the district，combined with the
preparation of environmental risk maps． Finally，a targeted decision-making plan for urban environmental risk
prevention and control and emergency rescue is proposed．

In the process of driving，violations such as looking at mobile phones and talking with passengers
cause great hidden dangers to safe driving． In order to solve such problems，we propose a multi-angle behavior
recognition method，which can synchronously capture video of driving behaviors from three angles，construct a
multi-angle driver behavior data set，and use deep convolutional neural network to carry out recognition and
classification． The experimental results show the advantages of 3D CNN over 2D CNN in accuracy． And in
comparing input clip frames experiment，we found the stacked video frames will affect accuracy and in R2plus1D
model ( 3D convolution filter is decomposed into a separate component of time and space) ，realized driver
behavior based recognition accuracy of 87% from multiple perspectives．

The process of behavior recognition can be regarded as the combination of feature extraction and
classifier to a large extent． Compared to static image object recognition，video feature extraction of human
behavior recognition is more susceptible to such factors as dynamic background，acquisition device motion，
perspective and illumination，so it poses great challenges to researchers． Based on the systematic classification of
behavior recognition feature extraction，according to the different types of behavior recognition feature extraction
methods and common behavior recognition database，the behavior recognition feature extraction is systematically
classified to expound the latest research progress． And the current research difficulties and possible future
research directions are discussed．

In order to solve the problem of difficult to borrow bicycles during the rush hour of the public bicycle
system，and improve the operating efficiency and user satisfaction of the public bicycle system，the operation of
the existing public bicycle system in a certain city was used as the research object． The borrowing and return
rules of the system analyze the operating characteristics of public bicycles，study the time characteristics and
turnover characteristics of public bicycles，and provide a data basis for the establishment of scheduling models．
Finally，a better public bicycle scheduling model is established through the ant colony algorithm． The ant colony
algorithm，establishs a static dispatching route optimization model，and obtains the optimal dispatching route for
dispatching vehicles． Thus，the problem of difficult to borrow bicycles during peak hours is effectively solved，
reducing the time required for dispatching and improving work efficiency．