Abstract:Laser-projection displacement meters (LDM) are widely used for bridge deflection monitoring. A comparative calibration method is proposed to fulfill the LDMs’ on-line and in-situ calibration requirement. Taking a laser displacement meter with traceable calibration as a reference sensor (RS), the in-situ measured value series by the LDM under calibration (SUC) and the RS are firstly acquired simultaneously under the circumstances of the same time-varying excitation source. Then, by matching the two series based on feature points sectioning method, the phase differences caused by the variation of excitation source and diverse response characteristics of sensors are solved, and sequentially comparative measured values with respect to the certain inputs are obtained. Finally, performing the statistical analysis to the matched data series produces the calibration results of the LDM under calibration. An on-line calibration simulation system is designed and set up according to the real in-situ conditions, accompanying with a verification subsystem using laser interferometer as a reference to verify the validity of the proposed method. The experimental results show that the proposed method has a high consistency with the method of simultaneously comparative measurement by using laser interferometer as a reference. Especially, the proposed method can be used under in-situ conditions of bridge deflection monitoring application.

Abstract:As an important bearing component of the aircraft, the multilayer metal fastener structure is subjected to continuous groundairground cyclic loading. The fatigue cracks occur around the hole of fastener, such as rivets and highlock bolts due to the stress concentration. The traditional nondestructive testing method, however, is difficult to detect those cracks under insitu conditions. The remote field eddy current testing technology breaks the skin effect limitation in principle, and has great advantages in detecting deep hidden defects. Therefore, this paper proposes a new design of remote field eddy current sensor with different structure from the traditional method. The excitation coil is placed coaxially with the detection coil, which is located inside the excitation coil with a magnetic field shunt between them. This kind of structure can greatly reduce the sensor size. The optimal design of the new sensor is obtained by utilizing the finite element simulation. This method analyzes the size of the excitation coil, structure material and composition of the magnetic field shunt. Experimental results show that the new sensor can detect perforation of the fasteners by the 2×02×4 mm (length×width×depth) at a depth of 4mm. With the increasing of the defect length, the signal amplitude is increased greatly.

Abstract:The characteristic and distribution status of space charge directly changes the intensity and distribution of the electric field inside dielectric material and seriously affects the electric performance of electric device. In recent years, nano materials and micronano electric devices develop rapidly, detecting and grasping the characteristic information of space charge in nanoscale even smaller scale becomes a problem to be solved urgently. Based on the basic principle of pulsed electroacoustic method is new space charge distribution test method was designed and implemented based on THz wave and elasticoptic sampling technology. Based on the stressinduced birefringence effect, the elasticoptic sampling sensor was designed and manufactured and its performances are tested. A space charge test system is built and used to test the custom silicon PN junction samples. The results show that the changing rule of space charge region width versus bias voltage is consistent well with the basic electrical characteristics. The method adopts the entire optical technical means, which breaks the limitation of traditional electronic test technology to system bandwidth. Experiment results show that the method designed in this paper can effectively and reliably improve the space charge test resolution to nanoscale.

Abstract:A multiparameter oil contaminant detection sensor is designed. Based on singlespiral inductive sensor, the sensor adds another solenoid coil, which can perform inductance detection and capacitance detection. Inductance detection distinguishes between ferromagnetic particles and nonferromagnetic particles in oil; capacitance detection distinguishes between water and air in oil. Compared with traditional singlecoil sensor, this design not only realizes the multiparameter detection of oil contaminants, but also adopts the spiral coil with smaller wire diameter and more turns, which increases the detection sensitivity of the sensor. The experiment platform built with the proposed sensor is used to carry out test experiment. The iron particles with diameter greater than 20 μm and the copper particles with diameter greater than 90 μm can be detected during inductance test; the water drops with diameter greater than 90~100 μm and the air bubbles with diameter greater than 160~170 μm can be detected during capacitance detection. This study and design provides a new method for the rapid detection of oil contaminants, which has certain significance for the field of mechanical equipment fault diagnosis and life prediction.

Abstract:Laserprojection displacement meter (LDM) is widely utilized for bridge deflection monitoring. A comparative calibration method is proposed to fulfill the online and insitu calibration requirement of LDM. First, a laser displacement meter with traceable calibration is adopted as the reference sensor (RS). The insitu measured value series by the RS and the LDM under calibration are acquired simultaneously under the same timevarying excitation source. Then, two series are matched based on feature points sectioning method. The phase differences caused by the variation of excitation source and diverse response characteristics of sensors are solved. The sequentially comparative measured values with the certain inputs are obtained. Finally, the statistical analysis is performed to the matched data series, which can produce the result of LDM under calibration. An online calibration simulation system is designed and set up according to the real insitu condition. A verification subsystem using laser interferometer as a reference is adopted to evaluate the proposed method. Experimental results show that the proposed method has a high consistency with the method of simultaneously comparative measurement by using laser interferometer as a reference. Especially, the proposed method can be used under insitu conditions of bridge deflection monitoring application.

Abstract:When highprecision cloud detection is implemented on visible spectral remote sensing images, the variability of the cloud form and the similarity between the cloud area and the earth object will reduce detection accuracy. To address this problem, this paper proposes a weighted multilevel scale fused network (WMSFNet), which can be trained endtoend without manual intervention. Firstly, the sensitivity to the cloud form is reduced by learning cloud area and earth object in turn. Meanwhile, WMSFNet can automatically extract highlevel spatial features through the fully convolutional network. In this way, cloud and earth object can be distinguished at the pixel level. A multilevel feature fused structure is designed to combine semantic information with spatial information from different levels. The detection of segmentation boundaries can be enhanced. Experiments performed on several real remote sensing images demonstrate that the proposed method can reach pixel accuracy of 9539%, which is better than other stateoftheart semantic segmentation methods. The error rate of cloud fraction is less than 1%, which provides a new solution for cloudcontaminated remote sensing images.

Abstract:In order to improve the predictive ability of auxiliary engine faults, combined with the advantages of unsupervised learning methods in deep learning, an early fault warning method for power plant auxiliary engine based on improved stacked autoencoder network is proposed. The method takes the historical normal data of the auxiliary engine as the training set, utilizes the nonlinear expression ability of the stacked autoencoder (SAE) network to express the relationship between the variables of the auxiliary engine, and introduces the batch normalization (BN) algorithm to optimize network performance. For the input observation vectors, the SAE network gives the corresponding reconstruction vectors, constructs the similarity based on the fusion distance to represent the deviation between the observation vector and reconstruction vector. When the auxiliary engine starts to deviate from the normal state, the deviation between the observed value and reconstructed value increases, and the similarity drops to the warning threshold, which indicates that the engine fault appears. The normal data and fault data of the medium speed coal mill of a certain thermoelectric unit are used to conduct test and verification respectively. The results show that the SAE network with BN algorithm introduced has lower reconstruction error. The proposed fault warning method can make early warning before the coal mill is tripped, which indicates that the method can effectively make fault warning of auxiliary engine fault and has certain engineering application value.

Abstract:To achieve highspeed ultrasonic imaging for irregular twolayered medium, one kind of frequency domain synthetic aperture focusing technique (SAFT) based on visual source (VS) technique is proposed. Firstly, VS technique is adopted to construct a series of VS points at the interface of an unknown testing object with irregular twolayered structure. Secondly, the expression of irregular interface is obtained by the interpolation processing with the coordination of VS points. Thirdly, the velocity model is formulated, which represents the velocity distribution in the mesh imaging region according to the expression. Finally, the velocity model is used to accommodate lateral velocity variation for each migration interval during the implementation process of the nonstationary phase shift migration imaging. The performance of the method is evaluated using the convex surface and the sinusoidal curve surface, respectively. Meanwhile, the imaging results and the operation efficiency of the frequency domain VSSAFT are compared with the time domain VSSAFT. Experimental results show that the defect location and the surface shape are displayed accurately in two kinds of VSSAFT images. In the single CPU core condition, the calculation time is only 047～053 s for the frequency domain VSSAFT. However, the time domain VSSAFT need consume 9121～9354 s.

Abstract:The visualinertial simultaneous location and mapping (VISLAM) is mainly based on visual and inertial navigation information fusion. It is a tedious work to calibrate the cameraIMU extrinsic parameter offline. The tracking accuracy is affected when the mechanical configuration of the sensor suite changes slightly due to the impact or equipment adjustment. To solve this problem, one kind of VISLAM algorithm with automatic calibration and online estimation of the cameraIMU extrinsic parameters is proposed. In the algorithm, the first step is to estimate the cameraIMU extrinsic rotation with the handeye calibration and the gyroscope bias. Secondly, the scale factor, gravity and cameraIMU extrinsic translation are estimated without considering the accelerometer bias. Thirdly, these parameters are updated with the gravitational magnitude and accelerometer bias. Finally, the cameraIMU extrinsic parameters are put into the state vectors for online estimation. Experimental results using the EuRoC datasets show that the algorithm can automatically calibrate and estimate the cameraIMU extrinsic parameters. The errors of extrinsic orientation and the translation are within 05 degree and 002 meter, respectively. This can help improve the rapid utilization and accuracy of the VISLAM system.

Abstract:Aiming at the problem of distinguishing split defects from mineral lines on the wood surface, a method based on local binary difference excitation pattern (LB_DEP) is proposed. Firstly, the potential defect regions are segmented with image preprocessing, then linear split and mineral line are screened using geometric parameters. Based on local binary pattern (LBP) and Weber′s law, an LB_DEP histogram reflecting the correlation relationship between image texture structure positions and difference excitation is established. Finally, the histogram features of LBP and LB_DEP are extracted, which are fused with feature data to form the feature vector that is used as the input of the SVM classifier to classify defects. Two feature extraction methods are proposed, namely ‘Hchisquare’ and ‘HPCA’, which are both evaluated on the selfbuilt data set. The experiment results show that for the two feature extraction methods the recall rates of 0937 and 0958, as well as the precision of 0950 and 0965 are obtained, respectively. Compared with other similar researches, the recall rate and precision are improved by at least 3% and 5%, respectively, and the time consumption is also at the level of milliseconds, which indicates the advantages and effectiveness of the proposed method.

Abstract:The spectral analysis method of oilwater mixture has become a research hotspot of current oilwater twophase flow measurement. However, in traditional oilwater mixture spectral analysis, the spectral feature extraction is generally achieved with the dimensionality reduction techniques, such as principal component analysis, successive projection algorithm and etc. The number of extracted spectra is mostly more than 10, which makes the oilwater two phase measurement fiber optic sensor expensive to manufacture and difficult to implement. In order to improve the practicability of the oilwater twophase measurement fiberoptic sensor based on spectral analysis, it is necessary to realize the ultrasparse representation of the oilwater twophase infrared spectrum. In order to achieve this goal, a spectral ultrasparse representation method with oilwater mixture spectral selfcross correlation (SCC) is proposed. In order to verify the effectiveness of the method, an experiment device for measuring the water content of oilwater mixture using infrared spectrum technique was established. From the 6 bands selected with the SCC algorithm, the bands of 1 050 nm and 1 650 nm were selected according to actual production process, and the dynamic experiments were carried out. The experiment results show that bands of 1 050 nm and 1 650 nm respond well to the mixed flow pattern of oil and water, and the two bands exhibit significant crosscorrelation. Obviously, this study will help improve the service performance of industrial fiberoptic sensors.

Abstract:Up to now, in most network cable factories, the detection of twistedpair pitch is still completed by manual sampling test. In china, most of the existing twistedpair pitch detection methods based on machine vision are remained in laboratory stage. In real production line, the detection results of these methods cannot satisfy the actual industrial detection requirements. Aiming at this problem, a realtime detection method is proposed for twistedpair pitch based on edge reconstruction. The method firstly extracts the edge data of the twistedpair with fast twisted pair edge detection method, and then reconstructs the twisted pair edge with Hammerstein model and hybrid particle swarm optimization (HPSO) algorithm. By this way, the method can filter out noise and obtain smooth twistedpair edge curve with precise twisted point position. On this basis, the twistedpair pitch data can be obtained finally through finding the twisted point positions. Through testing, the method can detect 125 m cable product per minute, the detection error is controlled between -161%~158%, all the specifications can meet the requirements of industrial realtime detection, and this method has high practical application value.

Abstract:The star sensor is the most accurate attitude measure instrument on the satellite. It determines the threeaxis attitude of the satellite by imaging and recognizing stars. It is mainly consisted of optical system, electronic system and information processing system. Star sensor on the conventional satellite has large weight and volume, which is difficult to meet the mission requirements of the booming micronano satellite. It has become a major obstacle which limits the accuracy of micronano satellite positioning. Compared with the rapid development of electronic miniaturization, integration and information processing technology, the optical system has become the bottleneck of star sensor miniaturization. To solve this problem, this paper proposes a large relative aperture microsmall optical system based on aspherical 1/13, which can realize 17° full field of view with 5 lenses. MTF is better than 05 at the Nyquist frequency point. The glass combination with the same partial or close dispersion coefficient and large difference of dispersion coefficient is selected. It is effective to correct the wide spectral chromatic aberration from 500~880 nm, which can achieve full field distortion ≤0013%. This technology can help design the nanostar sensor lens with a focal length of 25 mm. Its weight is 15 g, which is only 1/5 of the domestic nanostar sensor lens. Experimental results show that the optical system can meet the requirements determined by the angular secondlevel star point center after distortion and other indicators test. It provides one kind of core guarantee for realizing the high precision nanostar sensor.

Abstract:The insitu test of coded solar sensors is difficult to carry out. To solve this problem, one kind of LED solar simulator with motion function is proposed. The white LED is applied to simulate the solar light signal, and the motion equipment is employed to simulate the solar light vector signal. The paper introduces the structural composition and the working principle of the solar simulator. The simulation method of solar light and solar light vectors is studied very carefully. The required number of LEDs is determined by the radiant power. The linear array space of the LEDs is determined by the Sparrow criterion. The optical parameters of the cylindrical mirror are calculated by the matrix method, and the simulation analysis is performed by using Lighttools software. According to the structure and functional requirements of the sensor, the elements of motion equipment are confirmed, including ultrasonic motor, double rocker mechanism, mounting bracket and encoder. The vector adjustment accuracy of the device is analyzed. Experimental results show that the irradiation spot is 50×10 mm for the radiation simulator with a working distance of 50 mm. The irradiance is higher than 393 W/m2 and the irradiation nonuniformity is better than ±73%. It can simulate three solar vector angles of -13°, 0°, and 38°. The simulation accuracy is better than ±0008 3°, which meets the requirements of solar sensor testing. This method provides a powerful tool to ensure the quality of aerospace products.

Abstract:The structure parameter errors result in the regularity of position accuracy of the offset orthogonal articulated arm coordinate measuring machine (AACMM) in different configurations. The paper studies the relationship between the structure parameter error and the position accuracy of 6joint AACMM using the definite integration. The position accuracy prediction model is formulated. The inverse kinematics algorithm of offset orthogonal AACMM is proposed and the position accuracy distribution of AACMM in different configuration of specified operating point can be described. The configuration flexibility and weight of position accuracy in different configuration are explained. Evaluation experiments are summarized as follows. The comparison test of configuration flexibility is carried out; the position accuracy distribution of AACMM in different operation points is described; by using the distribution feature, singlepoint tests are implemented under different working points of different configuration. Experimental results show that the proposed method has the attribute of high configuration flexibility and is more suitable for the operation of AACMM. In further, it can be used to guide the selection of operation point and measurement configuration in operating the AACMM.

Abstract:In order to solve the selfpowering problem of wireless sensors in intelligent wheels, a piezoelectric selfpowered intelligent wheel is proposed, which enhances the practicality and safety of existing wheel energy harvesting technology. The piezoelectric cantilever beam is fixed in the spoke, which generates periodic excitation with the gravity effect of the free end mass block in rotation. In order to avoid direct collision with the spoke, a safety limiting structure is designed. Through establishing the electromechanical dynamics model of the system, the parameters of the device are designed for the R16sized rim. An experiment platform was built to achieve the load optimization and study the influence of the limiting pitch on the harvesting performance. In order to fully evaluate the performance of the piezoelectric selfpowered intelligent wheel, a prototype experiment was carried out on a real vehicle. The results show that the harvesting device designed in this paper can obtain the energy of 643~8660 μW at 50~70 km/h, which provides the possibility of establishing a selfpowered intelligent wheel system without maintenance.

Abstract:In the real environment, there is a crosssensitivity problem between fiber grating strain and temperature measurement. This paper proposes a simultaneous measurement method of two parameters using chirp fiber grating sensor. By sealing chirp fiber bragg grating (CFBG) on the equalstrength beam, the difference between the sensitivity of the center wavelength and the bandwidth of the CFBG reflection spectrum to temperature and strain is used to form a coefficient decoupling matrix, which can achieve simultaneous measurement of the corresponding change and temperature. At the room temperature, the sensitivities of CFBG center wavelength and bandwidth with strain are 079 and 138 pm/με. The corresponding linearity values are 0998 8 and 0999 3. Among the temperature range of -20~60℃, the sensitivities of the central wavelength and bandwidth of CFBG with temperature are 2274 and 2397 pm/℃. The linearity values are 0999 8 and 0997 0. Experimental results show that the simultaneous measurement of strain and temperature can be achieved using one single CFBG.

Abstract:To measure profile parameters of automotive synchronous belt, the problems include low efficiency and the influence of human factors. This paper proposes a noncontact measurement scheme based on laser triangulation. The measurement device of automobile synchronous belt tooth profile is designed. The profile data acquisition of the synchronous belt is realized by the laser displacement sensor coincidence with the transmission speed of the synchronous belt. The improved segmentation algorithm is proposed to achieve feature segmentation of the collected profile data. By fitting these data after segmentation, the profile curve features are reconstructed and the profile parameters of the measured teeth are obtained. Based on the laser triangulation method, the profile parameters of synchronous belt are measured. The automatic measurement of profile parameters of automobile synchronous belt is realized, which can enhance the measurement efficiency and avoid the influence of human factors. The measurement accuracy is also ensured. It is expected that this method can be applied in the practical production of synchronous belt.

Abstract:When guidelines for uncertainty method (GUM) are utilized to evaluate the uncertainty of the pitot tube measuring wind speed, the measurement results often contain deviation. To confirm the applicability of this method and modify the deviation, GUM and Monte Carlo method (MCM) are firstly adopted to evaluate the uncertainty of the measured wind speed. Then, the applicability of GUM is verified by the methods given by JCGM 101:2008 GUM supplement 1. Finally, the wind speed measurement results obtained by GUM and MCM are compared. Results show that GUM method is still suitable for the uncertainty evaluation of wind speed measurement using pitot tube. The standard uncertainty of GUM evaluation results generates a negative deviation. The deviation rate relative to its measured wind speed value is -0611 8×10-5. The modified value is “0611 8×10-5 the measured wind speed value”. This value is smaller than the measured wind speed. For the practical application, it can help decide whether to modify it according to the actual situation.

Abstract:Ballistocardiogram (BCG) signal is a physiological signal that reflects the mechanical characteristics of the heart it can achieve continuous acquisition measurements without electrode binding. However, the BCG signal is weak and highly susceptible to interference, and is often submerged in noise during measurements. In order to effectively identify BCG signals, this paper proposes a BCG denoising method based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) combined with permutation entropy (PE). Firstly, the collected BCG signal is decomposed with CEEMDAN to obtain a series of intrinsic mode function (IMF) from high to low frequencies. Secondly, the value of each IMF component is calculated with PE and the threshold range of the useful signal is determined, thereby the high frequency noise and baseline drift in the signal are filtered out. The experiment results show that the amplitudefrequency characteristics of the signal after noise reduction are significantly improved, and compared with the traditional method the signaltonoise ratio is significantly improved, which proves that the proposed noise reduction method has obvious effect and can effectively restore the BCG signal characteristics.

Abstract:VE (virtual environment) force feedback system based on haptic handle has been widely used in humanmachine interaction field. Stability is the prerequisite for the system to work properly. Firstly, the force feedback device, the virtual interaction environment and operator are modeled independently as independent modules, then the closedloop control model of the force feedback interaction system is modeled based on impedance reconstruction method. The Hurwitz criterion is used to derive the system stability conditions, the frequency response function is used to analyze the influences of the stiffness and damping of virtual environment, the signal hold time interval of the zeroorder holder and the impedance of the operator on the system stability, and the critical stability conditions of the influence factors, such as virtual stiffness, virtual damping, operator impedance and etc. are given. The developed force feedback haptic handle is utilized to construct an experiment system and experiment study is conducted. The system stability under different conditions is analyzed for different frequencies of force feedback and different virtual environment impedance. The experiment results prove the correctness of the theoretical analysis.

Abstract:To recognize and track fingertip quickly and accurately, one kind of fingertip detectingtrackingsupervising algorithm based on Kinect depth information is proposed. The detection process mainly uses the depth data to realize the fingertip recognition. The fingertip detection result is combined with Kalman filter to determine the steady state of the fingertip. The tracking process mainly takes the steady state of the fingertip in detection process as the initial condition. Fast discriminatiwe scale space tracking algorithm is adopted as the supervising process to realize the realtime tracking of the fingertip. During tracking the movement of the fingertips of each frame in real time, the supervising process supervises the realtime tracking results of the fingertips in each frame. By matching the features of the fingertips to correct the deviation of the tracking results, the accuracy of the tracking results can be improved. Hence, the robustness of the whole system can be realized. In this paper, multiple sets of experiments have been implemented on fingertip detection, recognition and tracking. Both of them can accurately detect fingertips and track the movement of fingertips in real time. Experimental results show that the proposed method has strong robustness and accuracy.

Abstract:To solve the difficult problem of automatic identification rock mass microseism and blasting vibration signals, a feature extraction and classification method based on empirical wavelet transform_Hankel matrix_singular value decomposition (EWT_Hnakel_SVD) is proposed. Firstly, EWT spectrum segmentation method is improved to adapt the transient and diversity of microseism signals. Its effectiveness is demonstrated by using simulation signals. Then, the improved EWT is used to decompose the microseismic and blasting vibration signals. Five principal components of f1～f5 are obtained by correlation analysis, which are utilized to formulate the Hankel matrix. The maximum singular value and singular entropy of each Hankel matrix are calculated. Finally, the genetic algorithmoptimized support vector machine (GASVM) is adopted to classify the microseism and blasting signals. Experimental results show that the singular entropy of the blasting vibration signal component fl～f4 is too much singular entropy of the rock mass microseismic signal component fl～f4, and the maximum singular value of the blasting vibration signal component fl～f5 is greater than that of the rock mass microseismic signal component fl～f5. The improved EWT recognition is better than traditional EWT and empirical mode decomposition. GASVM recognition effect is better than support vector machine, logistic regression and Bayes discriminant method. The method based on EWT_Hankel_SVD and GASVM classification can reach accuracy rate 94%.

Abstract:Door control system is one of the most important subsystems in the subway vehicle. Due to the complex mechatronic structure, frequent open and close movement, and crowded passenger flow environment, high failure rate of door system persists. To accurately detect the incipient fault, a bigdatadriven optimal feature selection algorithm and a random forests (RF) based incipient fault diagnosis method are proposed in this paper. Firstly, multiphase timedomain fault features are extracted from door′s position, drivenmotor′s speed and current signals. Secondly, the irrelevant and redundant features are removed and the optimal fault features are retained by using distance evaluation technology. The selected optimal fault features are adopted as the input of RF classifier. The fault labels are utilized to formulate an intelligent fault diagnosis model. Finally, the fault diagnosis model can realize the online automatic recognition of different incipient faults in the door subsystem. Experiments are conducted on the bench testing door system of Hangzhou line 4. Results show that the proposed method can extract the early features of incipient faults. Compared with several existing methods, the diagnostic accuracy and robustness of the proposed method are greatly improved after optimal feature selection.

Abstract:The intermittency and randomness of wind make the operation state of wind turbine change frequently. As a result, the false positive ration and false negative rate in anomaly detection of equipment are serious. The costs of operation and maintenance in the wind power industry are high. To solve this problem, one kind of Knearest neighbor fault detection method based on dynamic feature matrix is proposed in this work. It constructs a dynamic feature matrix based on mutual information to describe the dynamic characteristics of wind turbine. The weighted knearest neighbor fault detection method is introduced to address the influence of the characteristic contribution and cumulative mutual information in dynamic feature matrix. The dynamic threshold can help reduce false alarm caused by the sudden change of operation state. This paper takes examples of the common sensor faults and actuator faults in the 5MW offshore benchmark of National Renewable Energy Laboratory and the pitch system faults in SCADA system. The fault detection results of the proposed method are compared with PCA, KPCA, FDkNN and PCkNN, respectively. Experimental results demonstrate that the proposed method can accurately detect the fault information. Compared with other methods, it can achieve better fault detection results.

Abstract:To generate the test sequence, it has to achieve the description of temporal constraints. At present, there is no effective method to realize this objective. Therefore, a modeling method for sequence testing with temporal constraints is proposed in this article. The allowable intervals of state transition are proposed, which can describe the continuous appearing number of a former state when there is a state transition. Another problem of sequence testing is how to evaluate the target coverage degree effectively. This problem is solved by introducing kernel functions to evaluate the target coverage degree of test sequences. Finally, the real cases are modeled based on the proposed method. The availability and feasibility of the method are validated.

Abstract:The vibration signal of planetary gearbox has the complexity of frequency component and timevarying. To solve this problem, the fault diagnosis method based on deep learning with timefrequency fusion and attention mechanism is proposed. Firstly, the wavelet packet decomposition is used to transform the original vibration signal into two dimensions of frequency band and time, which are adopted as input data. Then, the convolutional neural network is applied to fuse the frequency band characteristics of the data. The bidirectional gated recurrent unit is employed to fuse the timing features. The attention structure is adopted to weight and merge the features of different time point adaptively and dynamically. Finally, the classifier is used to identify the endtoend fault diagnosis of the planetary gearbox. Experimental results show that this method has higher accuracy than the existing deep learning fault diagnosis model. It can accurately diagnose various health states of planetary gearbox.

Abstract:The geological information cannot be fully dynamically perceived in the shield tunneling process, which makes it difficult to accurately predict the ground settlement. To solve this problem, one kind of dynamical stratum identification model using the adaptive complex stratum changes is proposed in this paper. This method is based on the extreme gradient boosting (XGBoost) using shield construction parameters to implement the inverse deduction of stratum changes. In this way, the changing rule of construction parameters can be clarified when the stratum changes. The fusion model of error back propagation algorithm (BP) and support vector regression (SVR) for ground settlement prediction is formulated to obtain the intrinsic relationship of the ground settlement at different distances from the initial excavation face, stratum conditions and parameters of shield driving. The proposed method is validated by the 590ring data of a metro construction. Compared with the traditional method, it can achieve higher prediction accuracy.

Abstract:Aiming at the formation control problem of AUVs based on leader in a known path, a new formation controller combining AUV path control and formation coordination control is proposed. Firstly, in the controller the backstepping adaptive dynamical sliding mode control method is adopted to achieve the path following control of the AUV, which transforms the position, attitude and timevarying velocity tracking to virtual velocity control, makes the AUVs to reach the expected position and velocity, effectively avoids the singular value problem of backstepping control, nicely realizes the control of the uncertain model, and at the same time improves the accuracy of follower collaborative positioning. Secondly, based on the path tracking control, the formation coordination controller converts the position error control of the leader and follower into speed error control of the follower, enables the followers to quickly reach the desired positions and makes all the AUVs achieve and maintain the expected formation. Simulation experiment was carried out to verify feasibility of the proposed control strategy, the result shows that the proposed algorithm improves the response speed, control accuracy and stability of the formation. The test experiment on the lake with three AUVs was conducted, and the result proves the effectiveness of the proposed control strategy, which can be effectively applied in actual situation.

Abstract:This paper takes the aerial treepruning robot that is used for clearing the tree barricades near highvoltage lines as study object, and the attitude control problem of the aerial robot is addressed. Firstly, the attitude dynamics model and control allocation matrix of the aerial treepruning robot with new structure are established. Then, in order to overcome the uncertainty of the inertia matrix, the sliding mode method is used to design the attitude controller. At the same time, a nonlinear function is introduced to improve the traditional boundary layer method and enhance the controller performance. Then, a control allocation matrix switching strategy is proposed to solve the dynamics model change problem of the aerial robot in the cutting operation process. Finally, the controller was implemented on the simulation platform. The experiment results show that the sliding mode controller designed in this paper has good attitude control performance and can effectively overcome the inertia uncertainty of the robot body. Through switching the control allocation matrix, the attitude stabilization of the aerial robot in the cutting operation process is realized.