Abstract:Cantilever plates are common structures, which have been widely utilized in engineering practice, including aerospace, rail transportation, civil engineering and other fields. The deformation estimation is always one of the important research contents. This paper aims to solve the problem of bending of rectangular elastic cantilever plates and proposes a method to solve it by establishing the relationship between strain and deformation. The finite difference expression of structural transverse deformation is achieved under the framework of linearly elastic small deflection theory. The distributed optical fiber sensor with the ability of high-density strain measuring is used to obtain the strain input required for the expression, and a method of error correction for deformation measurement is proposed. The technical feasibility and measurement accuracy of structural deformation estimation based on distributed optical fiber sensing technology are evaluated. Some dial indicators are used in the metal cantilever bending tests. Experimental results show that the maximum deviation between results from the estimation and the dial indicators is 2. 0 mm, which accounts 9. 22% of the indication from dial indicators. The distributed optical fiber sensing technology is competent to perform deformation estimation, which has the potential application prospect.

Abstract:To enhance the output voltage amplitude of the magnetostrictive displacement sensor, the torsional strain of the waveguide wire is studied. Based on the related theories of electromagnetics and theoretical mechanics, a mathematical model of the torsional strain of the waveguide wire and a three-dimensional finite element simulation model are formulated. The factors that affect the torsional strain of the waveguide wire are analyzed. The corresponding output voltage of the sensor is obtained by experiments. Results show that the output voltage of the sensor is consistent with the change trend of the maximum torsional strain of the waveguide wire. The output voltage of the sensor and the maximum value of the torsional strain of the waveguide wire increase with the increasing of permanent magnet magnetization, permanent magnet geometric parameters, and permanent magnet placement angle. As the distance between the permanent magnet and the waveguide wire increases, it shows a trend of first increasing and then decreasing. Finally, it is determined that the permanent magnet has a length of 15 mm, a width of 10 mm, and a height of 5 mm. The length direction is magnetized 1T and the magnetization direction is perpendicular to the waveguide wire. The distance between the permanent magnet and the waveguide wire is adjusted to 25. 5 mm. And the torsional strain of the waveguide wire reaches a maximum of 3. 84×10 -4 mm, and the output voltage of the sensor also reaches a maximum of 0. 109 V. Research results provide guidance for reasonable selection of magnetization direction, magnetization size, geometric parameters, placement manner and distance between permanent magnet and waveguide wire.

Abstract:The accurate understanding of the hydrological characteristics of cave dripping water is the basis of studying the groundwater infiltration process in the karst vadose zone. To continuously and accurately obtain dripping water information, this paper designs a dripping water sensor of karst cave to meet the needs of karst hydrogeological study. The piezoelectric induction mode and PVDF piezoelectric film are selected as the induction material, and the heart sound sensor is used as the induction element. Through the analysis of the response principle of the structure design of the extended induction surface, it is verified that the three-layer design could interfere with each other, produce superposition signal and obtain good induction effect. The inductive surface materials of the top layer and the intermediate layer are determined. By simulation in the laboratory and field observation in cave, the precision size of the sensor is designed. The signal processing circuit, software and hardware of embedded system of the sensor are developed to realize the long time monitoring and recording. Cave field data analysis indicates that the induction height of sensor is ≥20 cm, the fastest rate of dripping water is 10 drops per second. It has advantages of precision, stability and waterproof. The sensor is suitable to high humidity cave environment and meets the needs of the related research works.

Abstract:Transverse sensitivity of the single-axis accelerometer is one of the most important performance metrics, which has significant influence on the calibration model and accuracy. In this study, a novel model considering the transverse sensitivity is proposed to calibrate the three-axis micro-electro-mechanical systems ( MEMS) acceleration sensors. The transverse sensitivity of the model is introduced into the calibration equation as a symmetric matrix. The established calibration function includes zero deviation, scale factor error, non-orthogonal installation error and transverse sensitivity. Finally, the 12-position calibration method is utilized to collect experimental data. The maximum likelihood parameter estimation algorithm is used to solve the calibration model to achieve sensor calibration. Experimental results show that the compensation error of the calibration mathematical model is smaller in terms of mean values and variance than current calibration mathematical model. The accuracy after compensation is higher and the stability is better, which show positive effectiveness on improving the accuracy of MEMS three-axis acceleration sensor. This study has important theoretical significance and engineering application value

Abstract:To deal with commutation error in conventional position-sensorless control technology of the brushless DC motor, a closed-loop correction method based on rotor angle observer is proposed. By the systematic analysis of the commutation error mechanism, a rotor angle observer model is formulated to achieve the rotor position in real time. The total commutation error of the motor could be normalized to the phase difference between the back EMF and the phase current for correction. Compared with the traditional position-sensorless control technology, the proposed method can accurately correct the commutation error in a wide range of motor speed and have the advantage of higher robustness. Extensive simulation and experiment results show that the proposed commutation error correction strategy can obtain accurate commutation points and significantly reduce the torque ripple. Especially, during the commutation period, the phase current ripple after correction is decreased from 42% to about 18% .

Abstract:In Lamb wave damage detection of aircraft structure, it needs to address the problem of the environmental temperature influence. In this study, a weighted time domain warping (WTDW) temperature compensation method is proposed. Without any prior parameters or models, the temperature influence on Lamb wave signal can be adaptively compensated by WTDW. Furthermore, the temperature over-compensation problem of traditional dynamic time warping (DTW) that can easily suppress the defect information in the compensated signals is addressed. Therefore, WTDW is conveniently applied to the practical Lamb wave damage detection. Firstly, based on the fundamental theoretical analysis of DTW, the adaptive temperature compensation theory of WTDW is studied. Then, associated with the linearly-dispersive signal construction ( LDSC) dispersion compensation damage imaging algorithm, the WTDWbased high reliability and high resolution damage imaging method under environmental temperature variances is proposed. Experimental results on an aluminum plate ( 800 mm × 650 mm × 2 mm) show that the proposed WTDW method can adaptively compensate the influences that the environmental temperature changes from the room temperature (23℃ ) to -20℃ ~ 50℃ . And the damage location error of the WTDW-based Lamb wave high reliability imaging is less than 2. 8 mm.

Abstract:The seismic interferometry is a microtremor survey algorithm for achieving the dispersion curve between two geophones. To obtain the dispersion curve, the traditional seismic interferometry in the frequency domain uses monotonically decreasing Bessel functions to fit the coherence coefficients. The effective dispersion points of calculation are less, and high frequencies are missing. To perform microtremor detection, the resolution of shallow structure is low, and there are exploration blind areas with a depth similar to the station distance. An improved seismic interferometry method is proposed, which uses the signal-to-noise ratio model of the measured signal to correct the coherence coefficients of two signals. The coherence coefficients are divided into different intervals by the monotonicity of the first type of zero-order Bessel curve. Piecewise fitting Bessel function is utilized to solve the surface wave phase velocity. The engineering practice shows that this method has a good detection effectiveness. The high frequency band is expanded, and the detection accuracy is improved. The problem of shallow exploration blind areas is effectively addressed.

Abstract:Aiming at the problem that the orthogonal frequency division multiplexing (OFDM) underwater acoustic communication is sensitive to Doppler frequency shift, a Doppler estimation and compensation algorithm for OFDM underwater acoustic communication based on non-uniform fast Fourier transform ( NUFFT) is proposed. The type-2 NUFFT technique is introduced into the OFDM underwater acoustic communication and utilized to perform Doppler factor grid compensation on the received signal. And the estimation of underwater acoustic channel is obtained through the sparse Bayesian learning. Then the minimum sparse energy of the underwater acoustic channel is used as the Doppler factor judgment criterion to obtain a high-precision Doppler estimation. Simulation results show that when the signal to noise ratio (SNR) is greater than 5 dB, the root mean squared error (RMSE) of the proposed method is better than 1×10 -5 . And the bit error rate (BER) of the system decreases with the decrease of the RMSE of Doppler estimation. The results of the sea trial show that the relative radial velocity between the two ships obtained with the proposed Doppler estimation method is consistent with the actual situation. And the average BER after decoding is better than 8. 33 × 10 -4 . Simulation and sea trial results demonstrate that the proposed algorithm can effectively estimate Doppler factor and reduce the system bit error rate.

Abstract:When the coupling power transfer ( CPT) is applied in the submerged buoy system, it can provide power for the communication beacon without electrical contact. However, the distance between the beacon and the submerged buoy will change. The system may be in the deviating resonance state and its efficiency will be reduced. In this article, the law of the coupling coil with the distance is analyzed. The model of the CPT system is formulated. The relationship between the coil efficiency and frequency in the deviating resonance state is studied. The system efficiency can be improved by using the search algorithm, which is evaluated by simulation. A prototype of the CPT system with variable distance is established to evaluate the search algorithm. Experimental results show that the search algorithm can improve the system efficiency under the change of the coil distance. Compared with the fixed-step perturbation observation algorithm, the maximum efficiency of the particle swarm optimization (PSO) is increased by 3% , which is less affected by the choice of the initial center frequency. The efficiency of the CPT system can be improved when the coil distance changes by using PSO.

Abstract:Dust deposition on photovoltaic (PV) modules seriously affects the stability of PV system output, resulting in the reduction of generated power and shortening the service life of the PV modules. The accurate assessment of dust accumulation concentration in the PV field will help to improve the accuracy of the power prediction model of PV power generation. In this paper, the dust particles collected in PV power station are taken as the research object. Firstly, the elemental composition, content, morphological characteristic and particle diameter distribution of the dust particles are analyzed. According to the actual power generation efficiency of PV modules and environment parameters, a soft measurement model of dust accumulation concentration is established to rapidly evaluate dust accumulation degree of PV power station. Secondly, in order to accurately obtain the relevant parameters of the model, several experiments on the influence of dust accumulation concentration on power generation performance are carried out to obtain the relationship between the output power of PV modules and irradiance, dust accumulation concentration, as well as module temperature. Finally, the accuracy and reliability of the model are verified in natural conditions. The results show that compared with other traditional methods the proposed model has better prediction performance, and the accuracy can reach 89. 6% .

Abstract:To eliminate the local artifacts in terahertz (THz) images caused by power fluctuation effect, a THz image enhancement model based on homomorphic filtering is constructed. However, the parameter values of the enhancement model have large differences and strong coupling, which brings great difficulties to determine the parameters of the enhancement model. Therefore, an adaptive teachinglearning-based optimization algorithm based on chaotic mapping and differential evolution is proposed to solve the optimal parameters of the enhancement model. Firstly, the standard Logistic chaotic mapping is improved, which increases the population diversity. Secondly, the update rate of fitness is introduced, the adaptive adjustment function of the inertial weight is constructed and the global and local optimization abilities are balanced, which is beneficial for the population to approach the optimal solution Thirdly, based on the idea of differential evolution, the teaching reform stage is proposed to avoid the algorithm falling into the local optima. Finally, the defect samples were prepared and terahertz non-destructive testing experiments were carried out. The results show that compared with the other three methods, the developed method has the best effect in eliminating local artifacts, and the two-dimensional entropy of THz images increases by 16% , 5% and 10% , respectively, and the average gradient.

Abstract:Aiming at the demand for scientific periodic replace of smart meters, a remaining useful life (RUL) prediction method based on the basic error data of smart meters is established. Firstly, the Person correlation coefficient is adopted to screen out the environmental stress that has great impact on the basic error data of the smart meter as the model input; then the Gaussian kernel, the Matern32 kernel, and the periodic kernel are adopted to match the basic error trend of the smart meter under the multi-stress environment; the Bayesian method and the Monte Carlo Markov Chain (MCMC) are used to solve the model. Experiment results show that smart meters from different companies have different environmental tolerances. Under typical environmental condition of high dry heat, the posterior upper quartile value of the smart meters from A company reaches the threshold, and the RUL is 43 months; the smart meters from company B has no general failures happened, however there will be a great possibility to failure in the next 47 months, and troubleshooting and error verification should be started. In the typical environment of high dry heat, the accelerated out-of-tolerance failure phenomenon of smart meters does not meet the 8-year verification period stipulated in the measurement regulations, and the periodic verification work should be dynamically adjusted.

Abstract:Remaining useful life (RUL) prediction of lithium battery is very important for the safe using of lithium batteries. Due to the capacity regeneration phenomenon and random interferences in the using process of lithium batteries, the prediction accuracy and generalization performance of a single model with a single-scale signal are relatively poor. Aiming at these problems, a new RUL prediction method based on variational mode decomposition (VMD) and integrated deep model is proposed. Firstly, VMD is used to decompose the lithium battery capacity data to obtain the global degradation trend of the signal and the local random fluctuation components in multiple scales. Then, the global degradation trend and various fluctuation components are modeled using multilayer perceptron (MLP) and long short-term memory (LSTM) neural network, respectively. Finally, the prediction results of the sub-models of various components are integrated to obtain the final remaining useful life prediction result of the lithium battery. Experiment results show that the proposed method possesses high prediction accuracy and stability.

Abstract:Partial discharge (PD) is the main cause of premature failure for the turn-to-turn insulation system in the inverter-fed motor. The prediction of PD inception voltage (PDIV) for the turn-to-turn insulation plays a significant role in the insulation design of inverterfed motors. Therefore, a PDIV prediction method for turn-to-turn insulation based on the deep belief network (DBN) is proposed in this paper. Firstly, a PD simulation model is formulated, which is based on Townsend theory. The PDIV of different simulation parameters for the turn-to-turn insulation is calculated. Secondly, the influence factors of PDIV on the turn-to-turn insulation are analyzed. The DBN is implemented to mine the non-linear relationship between the influence factors and the PDIV. Furthermore, the effectiveness of the proposed method is evaluated by simulation analysis and experiment. Finally, the principal influence factors of the turn-to-turn insulation are investigated by the mean impact value algorithm. The case study demonstrates that the max relative error of the proposed method is 5. 9% . It provides a novel idea for the condition assessment and insulation design of inverter-fed motors.

Abstract:The gear degeneration evaluation technology plays an important role in maintaining safety of various equipment operation. The traditional gear degeneration evaluation methods are susceptible to feature extraction and data pre-processing tricks. The methods based on the generative model use raw observations to perform evaluation. And the human factors can be effectively reduced. However, traditional generative models, such as variational autoencoder (VAE), are limited by poor performance in marginal probability density evaluation. In this study, multivariate invertible deep probabilistic learning (MIDPL) is proposed, which can establish the connection between a given distribution and an unknown observation distribution by stacking learnable invertible transformation. The marginal probability density evaluation of the multi observation sequence can be realized through the given distribution. The proposed MIDPL model is evaluated by gear degeneration experiments. Compared with VAE, the evaluation errors of MIDPL for gear pitting dataset and gear breaking dataset are reduced by 30. 92% and 69. 25% , respectively. The proposed MIDPL can achieve more accurate and stable degeneration evaluation.

Abstract:The measurement and error analysis of shearer cutting height are a significant research task for the longwall mining automation. In this paper, aiming at the body attitude sensor and rocker-arm swing-angle sensor measurement scheme as well as the body attitude sensor and displacement sensor of height-adjustment hydro-cylinder measurement scheme, two measurement models of shearer cutting height are constructed, respectively. Adopting the function error formula, the measurement error models are obtained. Taking the type MG1000 / 2660-WD shearer as an example, the changing laws of the shearer cutting height versus pitch angle, rockerarm swing-angle and displacement of height-adjustment hydro-cylinder are analyzed, and the positions where the shearer cutting heights have the maximum values in these two measurement schemes are obtained. According to the analysis result, the accuracies of the rocker-arm swing-angle sensor and displacement sensor of height-adjustment hydro-cylinder have a litter influence on the measurement error, and the accuracy of the body attitude sensor would determine the measurement error of the cutting height. Finally, taking the case where the measurement error of shearer cutting height is less than 5 cm as an example, the accuracy requirements of the sensors in these two measurement schemes are obtained as follows: The accuracy of rocker-arm swing-angle sensor is 0. 022° and the maximum dynamic error of the pitch angle of the body attitude sensor is less than 0. 16° ( within 1 hour) , the accuracy of the displacement sensor of height-adjustment hydro-cylinder is 1 mm and the maximum dynamic error of pitch angle of the body attitude sensor is less than 0. 14° (within 1 hour) .

Abstract:In order to detect the damage of the inner and outer walls of non-ferromagnetic metals, a medium and low frequency AC electromagnetic field detection model is designed. The coupling mechanism of magnetic flux leakage and eddy current effect in the detection of AC electromagnetic field is analyzed, and the electromagnetic detection device is made for experimental verification. Based on COMSOL finite element simulation, the two-dimensional simulation model is established. The simulation tests are carried out for surface defects and buried depth defects of different depth and sizes, and the relationship between voltage amplitude, phase of the coil and defect depth is obtained. Based on the simulation results, the electromagnetic detection device is made and the non-ferromagnetic metal test piece is tested. The results show that the manufactured electromagnetic detection device can penetrate the 18 mm stainless steel plate to detect a rectangular buried depth defect with a height of 5 mm and a length of 12 mm and a micro circular hole defect with a upper surface of 3 mm and Ф5 mm. It is concluded that the low frequency excitation signal has a good detection effect on the upper and lower surface defects of the test piece, while the intermediate frequency excitation signal is only suitable for the detection of upper surface defects.

Abstract:The existing soft measurement methods of absolute gas emission generally do not consider the influence of the historical data of gas emission. To address this issue, a soft measurement model of gas emission based on the long short-term memory (LSTM) in deep learning is proposed. The time series of historical data of absolute gas emission and its related influencing factors are utilized for prediction. Due to the gradient problem, the LSTM model needs to pay special attention to control the learning rate to prevent the severe decreasing of prediction results. The LSTM cell structure is adjusted, and the softsign function is introduced to solve the gradient problem through its first derivative with relatively gentle changes. In this way, the network convergence is faster and less prone to saturation. In view of the existence of many hyperparameters in LSTM, the quantum-behaved particle swarm optimization ( QPSO) algorithm is used to optimize the soft measurement accuracy of absolute gas emission. And the kernel-principal component analysis is utilized to reduce the dimension of measurement indexes to accelerate the convergence speed of the model. Comparing the improved model with the initial model, the improved model has higher accuracy and efficiency. The root mean squared error, mean absolute percentage error and goodness of fit determinant are 0. 080, 0. 82% and 0. 988, respectively. Comparing the proposed model with ELM, PSO-SVM, PSO-BP and GRU models, the proposed model has smaller error and better measurement results than other models. Experimental results show that the proposed soft measurement model of gas emission has better performance.

Abstract:To further improve the accuracy of the microwave waveguide method for detecting the unburned carbon content of fly ash, a method is proposed to calculate the ash level and relative group refractive index of fly ash in the waveguide. Through time domain transformation, the time of microwave reaching the interface of different media is achieved. Then, by combining with the transmission speed of microwave in different media in the waveguide, the reflection of microwave on the interface of different media in the waveguide is obtained. This paper uses a rectangular waveguide with a specification of 8 mm×15 mm×100 mm. Under the 12. 4-18 GHz frequency band, the above method tests the level and relative group refractive index of a series of ash samples with a mass of 0-12 g and a carbon content of 1% -10% . Results show that there is a critical point in the measurement process. When the actual ash level in the waveguide is larger than 47. 3 mm, the error between the measured level and the actual gray level is basically within ±2 mm. And the relative group refractive index of the ash sample in the waveguide is positively correlated with the carbon content of the fly ash. It does not vary with the mass of fly ash.

Abstract:A novel single probe bolt axial stress measurement method is put forward using the group speed ratio of different high order modes of cylindrical guided wave based on the propagation characteristics of high-frequency ultrasonic guided wave in metal rod parts. The Pochhammer-Chree equation considering the grain scattering attenuation is solved with numerical method. Then, the dispersion curves of group velocity and attenuation coefficient for guided wave are obtained, and the propagation laws of the guided wave in highfrequency region are analyzed. Next, the bolt axial stress measurement method based on the group speed ratio is derived based on the nonlinear acoustics and elastic mechanics theory. The ultrasonic stress measurement platform was built. The characteristics of the ultrasonic pulse-echo guided wave signals are discussed, the empirical wavelet transform algorithm is proposed to conduct mode decomposition of the signals and acquire the group velocity of a specific mode in the signal. A bolt axial stress calibration / measurement comparison experiment was carried out using the proposed method and the traditional method with the transit time ratio of the longitudinal wave and transverse wave. The results indicate that the average measurement error of the proposed method is only about 4% , which is significantly better than that of the traditional method ( average measurement error 6% ), and the proposed method has a more concise measurement process.

Abstract:The electromagnetic acoustic transducer (EMAT) has low energy conversion efficiency and weak excitation signal, although the electromagnetic acoustic resonance (EMAR) technology based on long-period excitation signal for generating acoustic resonance can improve the signal-to-noise ratio of the received signal, it may cause the main pulse widen, enlarge the blind area of ultrasonic detection and reduce the measurement accuracy. In this paper, a broadband excitation electromagnetic acoustic resonance technology ( BEEMAR) and a Halbach array longitudinal wave EMAT are applied, a single-cycle broadband excitation is used as the EMAT input signal, and the EMAR method is used to measure the plate thickness of 7075 aluminum alloy specimens. The probability of detection (POD) model is used to characterize the measurement accuracy for micro metal plate thickness. The experiment results show that in the POD curves obtained with BE-EMAR under three signal-to-noise ratios when the noise is not reduced, the corresponding maximum difference of a50 or a90/ 95 is 0. 05 mm, while the maximum difference under the same signal-to-noise ratio before and after the noise reduction is 0. 07 mm. However, in the POD curve obtained with the time of flight (ToF) method the maximum difference is 0. 21 mm. Therefore, there is no significant difference between the results obtained with BE-EMAR before and after noise reduction, and the scheme is insensitive to conventional noise signals. Before and after noise reduction, in the POD curve obtained with BE-EMAR, a50 and a90/ 95 both can be stabilized within 0. 55 mm, which proves that the BE-EMAR has better accuracy, stability and anti-noise ability than ToF method in thickness measurement.

Abstract:Aiming at the problem that the accuracy of the actual braking deceleration of the urban rail train is low under current theoretical deceleration open-loop control mode, a closed-loop braking control method is proposed based on parameter estimation. The actual braking force deviations caused by the additional running resistance when the train is running on ramps and curves and the change of friction coefficient of friction pair are taken as an equivalent total disturbance suffered by the train during braking. Then, taking the train deceleration and braking cylinder pressure as the inputs, the gradient estimation approach is adopted to solve the total disturbance. According to the estimated value of the total disturbance, the control target of train braking force is modified online to realize the closedloop control of the urban rail train braking force. In order to facilitate the programming in the actual electronic braking control unit of the braking system, the above control algorithm is discretized. The simulation and hardware-in-the-loop test results show that the parameter estimation algorithm can estimate both constant disturbance and variable disturbance, and the convergence rate of the estimated value is positively correlated with the estimator coefficient. However, when the coefficient increases to greater than 1, the convergence rate tends to saturation. This braking force closed-loop control algorithm can improve the tracking performance of the actual train deceleration to the target value. When the slope of the ramp changes according to sinusoidal law and at the same time the actual friction coefficient changes with the velocity, the deceleration deviation of the closed-loop control decreases from 0. 36 m/ s 2 to 0. 08 m/ s 2 . Moreover, when sliding occurs during braking, anti-skid control and braking force closed-loop control still can be compatible well, and the changes of the sliding axle speed, brake cylinder pressure and actual train deceleration meet the expectations.

Abstract:As an emerging robotics technology, soft exoskeleton incorporates soft drives and wearable structure, efficiently solves the problems of traditional rigid exoskeleton robots, such as heavy weight, poor compliance, low efficiency and poor wearing comfort. It is suitable for the walking assistance and movement rehabilitation of the patients with motor dysfunctions and has become one of the main developing directions in the field of active rehabilitation. This review focuses on the research development of innovation and application of soft exoskeleton in recent 10 years, collects and sorts out 60 target papers, which include 50 papers about cable drives, 7 papers about pneumatic drives and 3 papers about other drives. This paper analyzes and discusses the related researches from three aspects: soft drive structure design, control method and rehabilitation application. Currently, the soft exoskeleton research is mainly on cable drive and position control. The development direction is the equipment weight lightening and drive compliance, which combines electrophysiological signal research and human-computer interaction theory to reduce human metabolism. The related clinical trials show that the soft exoskeleton is beneficial to increase the symmetry and pace of the patient gait, which is closer to the normal gait. In the future, soft exoskeleton robots are expected to bring better active rehabilitation effects and wearing experience for patients with gait disorders.

Abstract:Aiming at the requirement that the operation tasks that cannot be completed by one person alone need additional assistance, the modular and reconfigurable supernumerary robotic limb is proposed and developed, which is composed of several basic modules connected in series with the same structure and function. In order to realize the module connection configuration and the shape reconfiguration capacities of the modular supernumerary robotic limb, the structure and function requirements of basic module units are analyzed. According to the requirements, the module unit based on steering engine driving is designed. Each unit is composed of two wedge-shaped sub-modules rotating coaxially with each other to form a rotational degree of freedom. The mechanical connection mechanism based on magnetic attraction is designed, which can realize the quick connection and separation of two modules. Furthermore, the connection method using double row metal contacts with interlacing disalignment is proposed to ensure the reliable electrical system connection of two module units. Four different connection orientations of two modules connected in series are clarified. The relationship between the configuration types of the supernumerary robotic limb and the number of modules is obtained. Facing to the supernumerary robotic limb with arbitrary number of modules and different connection orientations, the general forward kinematic model of the modular supernumerary robotic limb based on virtual axis is established. Experiments were carried out aiming at the rotation, carrying and fast connection capacities of basic modules, as well as the abilities of assisting the operator to carry wire and deliver telephone of supernumerary robotic limb in welding process, which verifies the assistant ability of the proposed new modular supernumerary robotic limb.

Abstract:For motion recognition based on the surface electromyography ( sEMG) , reducing the channel number of sEMG electrodes could simplify the target hardware implementation, and improve the rapid response performance. However, it also has the disadvantage of coarse accuracy. In this study, we propose a sEMG recognition method by combining the single-channel sEMG decomposition and the long short-term memory (LSTM) recurrent neural networks. Firstly, the single-channel sEMG signals are decomposed into motor unit action potential trains ( MUAPTs) . Then, features are extracted from the MUAPTs, and set as inputs to train the LSTM classification model. Experiments are conducted on 6 candidates with respect to the gesture recognition scenario. Five gestures are considered as outputs of the model. Experimental results of the proposed method are extensively compared with those obtained by other three schemes, including support vector machine ( SVM) with non-decomposition data, SVM with decomposed data, and LSTM with non-decomposition data. For the sEMG of Quadratipronator, the average classification accuracy is more than 90% using the proposed method. Compared with LSTM with non-decomposition data, SVM with decomposed data, and SVM with non-decomposition data, the accuracy of the proposed method is increased by 18. 7% , 4. 17% , and 11. 53% , respectively. These results verify the efficacy of the proposed method.

Abstract:There are two difficulties in the application of robot grasping in industry. How to detect the graspable object accurately and how to select the optimized grasp target among the detected multiple objects. In this paper the homoscedastic uncertainty is introduced into Keypoint RCNN to learn the weights of various losses, the attention modules are integrated into feature extractor, which composes the improved Keypoint RCNN model. A two-stage object grasp detection algorithm is proposed based on the improved Keypoint RCNN model. In the first stage, the improved model is used to predict the masks and keypoints. In the second stage, the masks and keypoints are used to compute the grasp representation and overlap rate of the object, the overlap rate represents the level of collision while grasping. According to the overlap rate, the optimized grasp target can be selected from multiple graspable objects. Comparison experiment indicates that the performances of the improved Keypoint RCNN model are improved in object detection, instance segmentation and keypoint detection compared with those of original model, and the average precisions (AP) on the self-built dataset reach 85. 15% , 79. 66% and 86. 63% , respectively. Robot grasping experiment proves that the proposed grasp detection algorithm can accurately calculate the grasp representation, select the optimized grasp and guide the robot to grasp the target with collision-free grasp.

Abstract:The gear structure is a key component of the transmission device, and the accurate detection of the addendum circle is an important basis for subsequent assembly. In the visual measurement of addendum circle, the traditional image processing method has low detection accuracy, and when the inclination angle of the gear structure is too large, the gear teeth will be occluded, which leads to the poor robustness of the algorithm. Aiming at the above problems, Detection method of addendum circle of gear structure based on machine vision is proposed. First, the sub-pixel corner detection of the gear teeth is performed based on the curvature scale space ( CSS) technology with adaptive threshold, second, the hyper least square method is used to fit the addendum ellipse, and finally the ellipse parameters are optimized by compensating for the quasi-eccentricity error. The experimental results show that the algorithm can not only extract the addendum circle that contains all the gear teeth images, but also can perform high-precision detection of the occluded images of the gear teeth. At the same time, it can compensate the elliptical quasi-eccentricity error caused by lens distortion. The measurement accuracy of the addendum circle center is 0. 056 mm, and the measurement accuracy of normal vector is 0. 068°, which meets the requirements of visual measurement of gear structure.

Abstract:The existing cable elongation at break measurement methods have defects of poor stability and low automation degree. To address these issues, a method based on the improved L-K optical flow is proposed. Firstly, black ring markers at both ends of the cable surface are used to extract feature points. Secondly, edge extraction and roundness calculation are utilized to obtain the initial centroid positions of the markers. The centroid with the feature points detected by the Harris algorithm is matched according to the principle of minimum Euclidean distance. In this way, the automatic selection of feature points is achieved. During the process of cable stretching, the improved L-K optical flow method is used to track the feature points and calculate the acceleration and spacing of the feature points in real time. An acceleration mutation criterion is proposed, which can accurately locate the frame where the cable breaks. The problem of poor accuracy of artificial judgment at the cable rupture moment is solved. Finally, according to the distance between the feature points before and after the cable deformation, the elongation at break of the cable is calculated. Experimental results show that the average error of cable elongation at break is 2. 78% , and the average speed of the algorithm is 21 frames/ s. Results indicate that the algorithm can measure the cable elongation at break quickly and accurately.

Abstract:The efficiency of traditional industrial CT image conversion to 3D printing G-code is low. To address this issue, a fast conversion method of industrial CT image to G-code based on the adjacent layer data matching is proposed. Firstly, the Canny operator is used to extract the contour edge of industrial CT image. Secondly, the contour bifurcation problem is processed to realize the geometric information data matching between adjacent layers. Thirdly, the contour interpolation between adjacent layers is carried out to meet the requirements of 3D printing layer thickness, so as to avoid the “ladder effect”. Finally, the G code for 3D printing is achieved by the filling coding. By using the proposed method, it takes 10. 5 s to convert the wheel CT image into G code, which is much less than other indirect conversion methods. There is no “ladder effect” in the 3D printed wheel hub, and the average dimension error rate is 0. 25% . Experimental results show that the conversion method does not involve intermediate format, and has high conversion efficiency. The conversion error is equivalent to the traditional method, which is suitable for parts with complex cavity structure.