Abstract:As the key technology of intelligent coal mining, gangue recognition technology has become a technical challenge in the field of intelligent coal mining. This paper firstly introduces the importance of coal gangue recognition technology and its impact on production safety and economic benefits. Subsequently, it points out such existing problems and challenges as the difficulty of identifying coal lumps and rock seams with different shapes, colors and depths, and the influence of noise as well as complex background in the process of identification, etc. In the next section, it describes in detail the key problems of coal gangue identification technology for integrated face mining. Next, the main methods of coal gangue recognition technology in comprehensive discharge working face are elaborated in detail: ray method, visual method, and vibration and sound signal method. By analyzing the principle, technical characteristics, advantages and disadvantages of gangue recognition methods, the current application status of gangue recognition technology in the comprehensive discharge working face, as well as the applicable conditions and problems of various methods are comprehensively evaluated. Finally, this paper discusses the future trend of this technology by emphasizing multi-sensor fusion, deep learning, intelligent decision-making and real-time monitoring as the core of current technological development.

Abstract:Aiming at the problems of large measurement delay and unpredictable production abnormalities that cannot be predicted in the manufacturing process of “ Chip / FPC on Glass” ( C/ FOG) for liquid crystal display ( LCD) panels, this paper proposes a virtual metrology method for C/ FOG manufacturing. This method uses sensors installed on the production machine to collect process state data during the production process and constructs a virtual metrology model based on multi-scale one-dimensional convolution and channel attention network (MS1DC-CA). Where the state data features in different scale ranges are extracted through multiple scale convolution kernels. In the preprocessing of original data containing missing values, an improved K-nearest neighbor interpolation method based on the particle swarm optimization algorithm (PSO-KNN Interpolation) is proposed to fill in the missing values. This method can reduce the interference introduced by filling values while retaining the features. Finally, it′s found that the actual defect rate is concentrated between 0. 1% and 0. 5% according to the experimental comparative analysis carried out for the data collected in actual production. The fitting mean square error of virtual metrology model is 0. 397 7 , which is lower than other existing fitting models. It also outperforms other existing fitting models under the evaluation indexes including mean absolute error, symmetric mean absolute percentage error and goodness of fit, which prouides the good predictive performance.

Abstract:To address the problems of different data distribution of rolling bearings under different speed conditions and low fault diagnosis accuracy caused by insufficient label samples in practical engineering applications, the domain adaptation modules are integrated into masked autoencoders (MAE). An improved masked autoencoders (IMAE) method for semi-supervised fault diagnosis of rolling bearings is proposed. Firstly, the two-dimensional time-frequency graph of the response signal is obtained by applying continuous wavelet transform (CWT) to the vibration signal of the rolling bearing. Then, the mask of the time-frequency graph is randomly masked, and the mask autoencoder is pre-trained with unlabeled samples to obtain the complex intrinsic features of the data. The reliance on labeled samples is reduced. Secondly, the domain adaptation module is introduced into the pre-trained encoder, and a small amount of labeled source domain data are used to fine-tune the IMAE, and the maximum mean difference is minimized in Hilbert space to reduce the data distribution difference between the source domain and the target domain caused by different rotational speeds. Finally, the semisupervised fault diagnosis of rolling bearing is realized under the Softmax classification layer. Through the experimental evaluation of the rolling bearing data set, the detection accuracy of the proposed method is more than 94% , which proves the feasibility and effectiveness of the proposed method.

Abstract:The mechanical structure of conventional circuit breaker is complex, and the faults caused by it are attriltuted to multi-source, so it is necessary to analyze the multi-source faults. However, the traditional multitask diagnosis method cannot deal with the interference between tasks well, which leads to the decrease of fault recognition rate. To solve this problem, a multi-fault diagnosis method based on wide-area features of vibration-current and soft sharing mechanism is proposed. Firstly, TKEO and DTM are used to achieve the accurate segmentation of vibration signal segments of the opening and closing process, and on this basis, the wide-area features of the vibration signal associated with the contact action and the attachment current signal are fused to synthesize color image samples to enrich the fault characterization information. Then, a multi-fault parallel diagnosis model is constructed based on the soft sharing mechanism of multitask learning, and automatically adjusts the weight ratio of the loss function of two tasks is automatically adjusted by adaptive weighting method to eliminate the mutual interference between tasks, thus improving the performance of fault diagnosis. Finally, examples are analyzed based on the two processes of closing and breaking respectively, and the results show that the classification accuracy of proposed method in this paper reaches 99. 78% and 99. 85% for two tasks respectively, which can effectively realize the multi-fault diagnosis of conventional circuit breakers.

Abstract:To explore the influence of various factors on thermal error in the feed system of CNC machine tools, an accurate thermal error prediction model is formulated. Thermal error measurement experiments are implemented on the feed system at a feed speed of 10 m/ min and ambient temperature of 20℃ to obtain the temperature rise and thermal error of the key points of the feed system. To improve prediction accuracy, Tent chaos is used to improve the squirrel search algorithm. The improved algorithm is utilized to optimize the neural network and establish a thermal error prediction model. The data obtained from thermal error measurement experiments are used for validation, and the results show that the prediction error of the neural network before improvement is 12. 23% , while the prediction error of the improved model is 8. 92% , indicating a significant improvement in accuracy. The prediction model is used to analyze the thermal error at the same position under different feed speeds. The results show that the temperature of key temperature measurement points in the feed system and the thermal error at each point of the lead screw increased with the increase in feed speed. Therefore, the proposed prediction model can accurately predict the thermal error of the feed system and provide a theoretical basis for error compensation

Abstract:To address the problems such as difficult data transmission and storage due to the large amount of operational status monitoring low-value density data, poor real-time performance of fault identification due to bandwidth impact, and the difficulty of deploying effectively large and deep learning models to edge-side hardware, this study proposes a gearbox edge intelligent fault diagnosis method based on multiplicative-convolutional network (MCN). Firstly, motivated by the merits of feature representation in signal filtering and feature extraction in deep learning, a lightweight MCN model is formulated. Secondly, a set of end-side edge intelligent processing unit prototype is made by using the embedded microcontroller unit. The system can be deployed directly at the edge of the gearbox, where the parameters of the MCN-based edge model can be trained and updated on the cloud side and deployed to the edge. The edge-side completes data acquisition, processing,and fault status identification, which can consume a large amount of sensor data directly. The experimental results show that MCN has an average recognition accuracy of 99. 75% , and the gearbox edge intelligent diagnosis system deployed with MCN can accurately identify the fault state at 0. 696 s.

Abstract:The accuracy of geomagnetic vector observation data will be affected by Instrument orientation error, and the existing correction methods require known geomagnetic field modulus or referring to standard instruments, and are difficult to be applied in downhole and marine situations where manual installation and adjustment of instruments are impossible. A large dynamic fluxgate magnetometer and a vector correction method based on Euler rotation transformation are proposed by the paper. The directional error angle of fluxgate can be obtained and the measurement is self-calibrated by this method without referring to standard instrument. The results show that when the orientation error of experimental instrument is set to the large angle of four different quadrants, the correlation coefficient with the corresponding component of station is still above 0. 99 after correction, and the length of confidence interval in the Bland-Altman diagram is reduced by more than 86% and the RMS error is reduced to less than 15% , which proves the effectiveness of this method. Correspondingly, the observation data quality of geomagnetic stations can be improved, and reference solutions for the orientation of downhole and marine geomagnetic observations can be also provided.

Abstract:Security robots frequently operate in the severe environments with dim and smoke, etc. Such an environment can be detected using millimeter wave, but its point cloud is sparse. Correspondingly the multi-millimeter waves point clouds can be fused to improve the ability to perceive environment. Accurate structural parameters are required for point cloud fusion to address the error in obtaining structural parameters through measurement. By analyzing the coordinates of multi-millimeter wave point cloud, particle swarm algorithm is utilized to search the structural parameters of millimeter wave radar, and the point cloud fusion as well as the construction of environment map are carried out according to the search results. Simultaneously, the evaluation critical of sparse point cloud map is proposed to quantitatively assess the millimeter wave sensing effect. Experiments were carried out in a darkened environment with a security robot, the results of which are as follows. The number of point clouds increases. The number of map boundary holes decreases by 55% on average. The boundary noise rate is reduced by 12. 9% on average. The dispersion of the object point clouds decreases by about 0. 06 on average. There is a decrease in the offsets of center positions in all experiments, when compared to the multi-millimeterwave perception system where the structural parameters were obtained by the measurement method.

Abstract:Machine learning plays an important role in WiFi fingerprint localization techniques. To address the problem that the effect of signal fluctuation on fingerprint recognition is often ignored and how to extract broader representation information from samples, this article proposes a WiFi localization algorithm based on improved contrastive learning and parallel fusion neural network. Firstly, the algorithm utilizes the improved CL to improve fingerprint discrimination, which increases the differentiation between different categories of fingerprints while reducing the differences between fingerprints of the same category. Secondly, a parallel fusion network based on CNN and LSTM is established. Compared with the traditional serial fusion method, the network can extract more effective features from the original samples. In addition, a flatten layer is added after the pooling layer to further consider the intermediate layer information of the network. Thus, a wider range of feature information is utilized to improve the generalization performance of the model. The results show that the proposed algorithm improves the localization performance by 26% over other localization algorithms.

Abstract:Gas flow measurement is widely used in respiratory monitoring, pipeline transport, and other fields. In this study, the thermal boundary layer parameters in the one-dimensional temperature distribution model of MEMS calorimetric sensors are carefully analyzed, and the corresponding empirical corrections are made. On the basis of the one-dimensional temperature distribution model, a new semicorrected theoretical model of the sensor voltage output with respect to the gas flow rate is proposed for MEMS calorimetric sensors with two pairs of upstream and downstream temperature measuring resistor chip structures. The theoretical model can be applied to different types of single-medium gases. Meanwhile, N2 and CO2 flow measurement experiments are implemented to compare with the theoretical model, which shows that the proposed theoretical model can accurately predict the flow of different gases, and the RMSE is 0. 15% for the CO2 measurement medium. In addition, combined with the theoretical analysis, a high-precision measurement model with a simple fitting form and a better applicability to deterministic gases is proposed, and the RMSE is 0. 05% for the CO2 measurement medium

Abstract:The accuracy and reliability of state estimation are seriously affected by measurement outliers and time-varying noise in complex environments. To address these issues, a robust adaptive factor graph optimization (FGO) integrated navigation algorithm based on variational Bayesian is proposed. First, the variational Bayesian inference is introduced into the FGO framework based on a priori and a posteriori two-stage updating to estimate the time-varying measurement noise covariance. Secondly, the mean innovation between neighboring keyframes is used to construct measurement covariance prediction as an outlier judgment to achieve robust estimation. Simulation and field tests based on INS / GNSS integrated navigation show that the proposed method can effectively estimate the timevarying measurement noise covariance in the presence of outlier interference, and reduce the horizontal position error by 26. 7% and 39. 8% compared to the M-estimation and sliding window adaptive FGO algorithms, which takes into account the accuracy and robust performance. It has an excellent adaptation to complex scenarios.

Abstract:Because of these shortcomings of the existing time-grating angular position sensor with multi-phase excitation, such as the accuracy, stability and dynamic performance improved difficulty in a small volume because of the difficulty of arranging further pole pairs, an angular position measurement method for capacitive type time-grating with single-phase excitation is proposed. In this method, the single-phase excitation is used to couple into four orthogonal signals in the space field, which can be constructed travelling-wave in time and space field simultaneously by a circuit. Then, an angular position measurement can be realized. In this article, these shortcomings of time-grating with multi-phase excitation, the measurement principle of time-grating with single-phase excitation, and the sensor prototype are introduced. The effectiveness of this measurement principle is evaluated by experiments. Experimental results show that the accuracy and dynamic performance of single-phase excitation sensors are better than multiphase excitation sensors with the same size and the same number of electrodes. the prototype accuracy of ±20″ and stability of 10″ were obtained. Speed fluctuation of ±1. 25% and the following error is ±2. 5″ under 400 rpm can meet the requirements of direct-drive motors.

Abstract:Geometric errors are important error sources of the computer numerically controlled ( CNC) five-axis machine tool. The conventional measurement methods and instruments are expensive and have a long measurement period. To address these issues, a fast identification method for geometric errors of a five-axis machine tool based on a double ball-bar is proposed. For the errors of linear axes of the machine tool, a spatial error model for linear axes is formulated, which is based on the multi-body system theory and homogeneous coordinate transformation. By measuring two circular trajectories of a double ball-bar (DBB) at different positions on the same plane, 4 key linear errors of the linear axes are identified. For the rotary table and swing axis of the five-axis machine tool, 12 geometric errors of the rotary axes are ascertained using the axial, radial, and conical measurement modes of the DBB, combined with the polynomial model. The experimental results of angle positing error show that the maximum error of the proposed method is 0. 001 8°, compared with that of the laser interferometer method. By using the identification results of the machine tool space error, the error of the test track is reduced from 16 μm to 4 μm, which is 25% before compensation. It evaluates the effectiveness of the method. The proposed method of geometric error identification for five-axis machine tools is convenient, and suitable for the industrial field.

Abstract:Magnetic particle imaging (MPI) is an emerging human body functional layer scanning imaging technology that uses biofunctionalized superparamagnetic nanoparticles as tracers. It typically requires the construction of a field-free point / line based on a gradient field and spatial scanning of it to achieve tracer localization. Its spatial resolution is proportional to the gradient field strength. However, high-strength gradient fields need large-volume electro / permanent magnets to construct, making them unable to be used in clinical requirements like minimally invasive lymph node surgery for localization. This study, without using a gradient localization field, proposes a narrow-band magnetic nanoparticle depth positioning theory based on Langevin functions and calculations of the spatial distribution of excitation-magnetization fields. It derives the functional relationship among the magnetic responses produced by magnetic nanoparticles at different depths and the spatial azimuthal angle. A half-peak width-distance detection and true concentration analysis model is formulated. Furthermore, a magnetic particle depth layer scanner (a non-gradient one-dimensional magnetic particle imager) is developed. The vitro experimental measurements show that its spatial positioning resolution is 15 mm with an error of 5. 21% , and the concentration model restoration error is 2. 61% . Compared with the approved European Sentimag static magnetic field magnetic particle positioning device, it has significantly superior performance and can fully meet innovative clinical application requirements like intraoperative lymph node positioning.

Abstract:In response to the close-range large-scale measurement requirements in on-orbit photogrammetry, we propose an optimal projection model identification and calibration method for large field-of-view photogrammetric cameras using starlight constraints. First, a piecewise starlight geometric projection model with an adjustable coefficient is formulated. Subsequently, a general multistation self-calibration bundle adjustment algorithm is developed for the piecewise starlight projection model. By combining the bundle adjustment algorithm with the northern goshawk optimization, we synchronously optimize the projection model adjustment coefficient, camera intrinsic and extrinsic parameters, and lens distortion coefficients. This optimization process continues until the star image points reprojection root mean squared error reaches the global minimum, resulting in the optimal projection model and its parameters. Experimental measurements show that, after calibrating the camera with a large field of view using starlight, the reprojection root mean squared error of star image coordinates is 1 / 9 pixel. In consecutive frame starlight calibration experiments, random errors in camera parameters using the Kalman filter are effectively eliminated. This method can identify the optimal projection model and calibrate all imaging parameters during the camera starlight calibration process, with the ability for consecutive frame calibration and parameter correction.

Abstract:In recent years, particle size and concentration measurements have been widely studied at home and abroad. An ultrasonic attenuation model based on the Monte Carlo method is extended to a high concentration for online measurement of desulphurization limestone slurry fineness in coal-fired power plants. The attenuation coefficients of the high-concentration Monte Carlo model ( HCMCM) are consistent with experimental measurement values. The model matrix constructed by HC-MCM can be used for the inversion of particle size distribution to reduce the calculation time. The relative deviation between the calculated value by the differential evolutionary algorithm and the set diameter and distribution width is less than 1% . A transmission and reflection dual-mode ultrasonic measurement device and software measurement system have been developed. The transmission sensor and ultrasonic probe are made to be suitable for the fineness measurement of high-concentration slurry, and have completed the 168-hour online measurement experiments relying on a coal-fired power plant. The results show that the trend and value of the attenuation spectra obtained by the ultrasonic transmission and reflection methods are basically the same, and the relative deviation between the volume median diameter DV50 and the measurement results of the laser particle analyzer is less than 8% . Its particle size satisfies the fineness requirements of finished limestone slurry in coal-fired power plants.

Abstract:In response to the complicated measurement environment of the aircraft hydraulic pipes, and the problem of the difficult layout of the sensors of the traditional wired measuring equipment, this article proposes a new kind of strain measuring device for hydraulic pipes. Based on the frequency response function method, dynamical models of the pipe and the measuring device are established to study the mapping relationship between their strains. The device is detachable and specifically designed for realizing non-contact measurement of the pipe′s strain. Combined with theoretical calculation and finite element simulation, this work studies the measuring device′s influence on the mode and strain of the pip and optimizes the design of the measuring device including size parameters. The accuracy of finite element simulation and the effectiveness of the non-contact measurement method are evaluated by experiments whose results show that the influence of the measuring device on the pipe is weak, with a 6. 2% relative average deviation of strain reconstruction below 300 Hz. The method and device are expected to solve problems for rapid and accurate measurement of the hydraulic pipe strain in the aircraft under complicated measurement environments.

Abstract:Many conventional transdermal energy transmission systems are designed as open-loop systems. It is crucial for patients to maintain stillness during the charging process since any movement can lead to misalignment or flipping of the coil. This can adversely affect the power received and disrupt the normal charging procedure. Regarding the above-mentioned issues, the LCC-S topology was used to achieve constant voltage characteristics at the receiver end. Then, a method based on mutual inductance estimation and primaryside power compensation was employed to dynamically adjust the transmit power at the transmitter end and effectively control the voltage at the receiver end. The simulation and in vitro experimental results demonstrate that this system can maintain a constant output voltage at the receiving end throughout the entire charging process of the lithium battery. The system allows for limited positional changes of the receiving coil, including axial displacement within 25 mm, lateral displacement within 22 mm, and rotation within an angle of 80°. This enables patients to engage in moderate movement during the charging process, greatly improving the stability and reliability of the transcutaneous energy transfer system. It has significant implications for the further application of artificial anal sphincter systems.

Abstract:Carbon fiber-reinforced plastics are prone to delamination defects during drilling and processing, and the delamination defects seriously affect the mechanical properties of the components, and there are serious safety hazards. For the detection of delamination defects at the hole edge of carbon fiber-reinforced plastics, a 1 / 4 matrix total focusing method based on the segmented annular arrays is proposed. Four structures of segmented annular arrays, R4×S8, R3×S12, R3×S16, and R4×S12, are designed. The characteristics of the focused sound field of each array are analyzed by numerical simulation comparisons, and the segmented annular array probe is finally determined to be the 5 MHz R4×S12 probe. By using this probe, the full-matrix capture of the upper, middle, and lower layer defects of the hole-edge delamination specimens of carbon fiber composites is performed, and the full-matrix 3D imaging and 1 / 4-matrix 3D imaging are implemented by using the VTK toolkit, respectively. The results show that the 1 / 4-matrix method has higher defect contrast than the full-matrix method for 3D imaging. The defects size characterization error of the 1 / 4-matrix method is smaller, with an error of no more than 6% . Compared with that of full-matrix imaging, the signal-to-noise ratio enhancement of the 1 / 4-matrix imaging is in the range of 3. 43~ 7. 61 dB, which effectively improves the image quality

Abstract:The damage to aluminized 321 steel, which is the main material of solar thermal power heat exchange tube, will lead to the shortening or even fracture of the life of the heat exchange tube. Therefore, the damage detection must be carried out. The damage characteristics of aluminized 321 steel are analyzed by the acoustic emission (AE) method, and the online dynamic monitoring of heat exchange tube performance is realized. The damage degree of aluminized 321 steel is characterized by using the AE Ib-value feature, and the self-organized mapping (SOM) neural network algorithm is used to cluster the AE characteristic parameters to analyze the damage mode of the material. The results show that the number of AE events in the mechanical plastic stage increases sharply, and the peak values of energy and ringing count indicate the fracture of the specimen. In addition, before the failure of the specimen, the Ib-value is significantly reduced and the density becomes dense, indicating that the variation characteristics of the Ib-value can be used as an early warning signal for the critical failure of the material. Four clusters and their corresponding characteristic frequencies are obtained by clustering analysis of the characteristic parameters through the SOM algorithm. The fracture morphology of the specimen is observed by scanning electron microscope ( SEM). The four clusters correspond to four types of damage modes, including hole growth and coalescence, micro-crack nucleation, macro-crack propagation, and fibrous fracture. This study aims to explore the damage evolution behavior of metal pipes and provide a basis for damage analysis and health monitoring of pipes.

Abstract:In this article, a spatial filter structure damage scanning monitoring method based on the annular array is proposed. The annular piezoelectric sensor array is arranged in the center of the structure to realize the omnidirectional scanning monitoring of the structure, which eliminates the influence of the blind area and far field on the monitoring effect in the active Lamb wave spatial filter monitoring method. The circular array consists of four linear arrays with different angles in the diameter direction. Each linear array uses the active Lamb wave spatial filtering monitoring method to monitor and image the structural damage. The imaging results of each array are fused by the PCA-wavelet transform to identify the location of damage. Thus, the omnidirectional damage scanning monitoring of the structure is realized, and the blind area, far-field, and false imaging of single array damage monitoring are eliminated. Then, more accurate damage monitoring results are obtained. The effectiveness and practicability of the method are verified by experimental research.

Abstract:The air-conditioning condenser is a key component of air-conditioning equipment. Its elbow connection makes it difficult to apply traditional contact damage detection methods due to various types of damage and complex geometric shapes. This article proposes a non-contact laser ultrasonic guided waves non-destructive testing method based on an energy mapping transfer network. Firstly, the approximate waveforms of the ablation signal and the thermoelastic signal are extracted through wavelet decomposition. An autoencoding energy mapping function is designed to map the thermoelastic signal feature space to the ablation signal feature space. The mapped thermoelastic signal close to the ablation signal is obtained. Then, the feature space of thermoelastic and ablation signals are aligned through the energy mapping transfer network. The network model uses the sum of domain conversion error and sample label error as the feature space alignment error value. Finally, the performance of the proposed method is evaluated through detection experiments on the air-conditioning condenser damage such as leakage, delamination, and cracks. The results show that the damage accuracy of the proposed method is 93. 09% , which is 7. 23% higher than the traditional laser thermoelastic excitation detection method.

Abstract:Aiming at the serious threat to train safety posed by the railway intrusion obstacles, while the general object detection methods based on deep learning struggle to break the barrier of data-driven training, the few-shot object detection methods have weak detection ability and low robustness for multi-scale obstacles in complex railway environments, this paper presents a high robust multi-scale fewshot railway intrusion obstacles detection model ( FRL-Net). The model utilizes the meta-learning strategy to capture rich feature information by designing the multi-scale few-shot obstacle feature extraction module, which can enhance the model′s ability to express the features of few-sample objects at different scales. The precise reweighting module is used for optimizing the meta-feature at different scales, and the few-shot railway obstacle detection optimization module is proposed to further enhance the few-shot railway obstacle detection performance of the model. The experimental results show that the proposed model achieves the mAP of 81. 8% in the 7-way 30-shot few-shot railway obstacle detection task, which is 3. 2% higher than that of FSRW. It is more suitable for detecting few-shot multi-scale railway obstacles in actual railway environments.

Abstract:In view of the complex coupling of the linear tilt multi-rotor plant protection UAV model and the interference of natural wind and plant protection actuators during the process of plant protection, a model compensation linear active disturbance rejection control algorithm is proposed in this study. Firstly, the kinematics and dynamics models of UAVs are formulated and the effectiveness of the centrifugal nozzle on the torque during the process of plant protection is analyzed. Then, the control decoupling and control allocation strategies are designed, which is based on the under-actuated characteristics of the UAV. Meanwhile, the model compensation-extended state observer is introduced to estimate other total disturbances inside and outside the system while regarding the fixed part of the operating model as the known disturbance. Therefore, the design of the pose controller is completed. The simulation results show that the maximum angle deviation and recovery time of the proposed algorithm are less than 0. 2° and 0. 55 s, respectively, during the process of plant protection. Compared with the PID algorithm and the linear active disturbance rejection algorithm, the maximum angle deviations are reduced by 3. 4° and 0. 9°, respectively, and the recovery time is reduced by 3. 1 and 0. 99 s, respectively, which shows that this algorithm has strong robustness. The flight experiments indicate that the UAV can track the plant protection trajectory well, and the maximum position steady-state error is less than 15 cm, which can meet the needs of plant protection.

Abstract:To enhance the surgical quality of bone milling surgery robots, it is required that the robot can perceive and control the milling state in real-time, (primarily including milling depth and milling angle). To address this problem, this article proposes a method to utilize a binaural microphone system to assist in milling state control. Firstly, the relationship between the milling state and the acoustic signal is modeled. Then the calibration experiments of milling depth and milling angle are completed to obtain the parameters of the proposed model. Finally the PD controller is used to control the milling state of the robot in real-time by combining it with the established model function. The experimental results show that when the desired depth of milling is 0. 5 mm, the experiment with angle control improved by 7. 0% in terms of milling depth deviation from the desired value and 34. 1% in terms of milling stability compared to the experimental results without angle control. It proves that in the extreme case of the desired depth of 0. 8 mm, the addition of angle control also has the positive effect of improving the milling effect. After experiments, it evaluates that the proposed method can effectively improve the work quality of the bone milling surgical robot.

Abstract:The traditional narrowband feedback active noise control (ANC) system uses the parallel adaptive filter to generate a reference signal. However, it still presents a poor quality of reference signal synthesis, high difficulty in setting the initial filter weights, and insufficient ability to suppress nonstationary narrowband noise. To solve the aforementioned problems, this article proposes a novel robust narrowband feedback ANC system where a cascaded-parallel adaptive filter is introduced to synthesize reference signal. In this way, the quality of reference signal synthesis is improved. In further, the overall noise reduction performance (NRP) is enhanced. The new system can not only solve the poor quality of reference signal synthesis in the traditional one but also reduce the complexity of the initial setting of filter weights and meanwhile improve the ability to cope with the non-stationarity of target noise. The experimental results show that the NRP of the new scheme is increased by 7. 89 and 9. 18 dB in the first half and second half respectively, as compared with the traditional one. Both simulated and experimental results show that the proposed system does present nice robustness and better NRP as compared with the traditional one.

Abstract:To solve the problem of low accuracy and difficulty in estimating the actual response of the servo system model in the tilt vector dynamic system under the influence of the external torque generated during rotor dynamics, this article incorporates the external moment of the servo as the disturbance noise into the identification link, constructs the system model, and proposes a parameter identification method for the tiltrotor servo system based on improving ecosystem particle swarm optimization ( IESPSO). To ensure the stability and safety of the test, a tilt vector dynamical system identification platform was designed to carry out the parameter identification test. The experimental results show that the IESPSO method has obvious advantages of convergence speed and estimation accuracy compared with the PSO algorithm, the ESPSO algorithm, and the recursive least squares method under the influence of the external moment noise generated during rotor dynamics.

Abstract:Blink detection is crucial in various practical application scenarios, such as eye disease detection, human-computer interaction, fatigued driving prevention, etc. To address the serious effect on the extraction of blink signal from the accompanying interference induced of the human body′s micro-scale movement, we propose a blink detection system, mmBlinkSEN, which can overcome the effects of accompanying interference and recover the blink waveform effectively. Inspired by the fact that blink and accompanying interference are mixed in a non-linear manner, a self-supervised deep contrastive learning method with a non-linear independent component analysis framework is proposed to separate blink and accompanying interference. A separation network ES-Net1 is designed, which is based on temporal correlation. The network takes two positive and negative sample sequences with temporal correlation and temporal uncorrelation as input to the network. The internal feature extractor inside the ES-Net1 is utilized to recover the temporal structure of the blink and the accompanying interference signal. Thus, the separation of the non-linear mixed signal is achieved. This article implements the mmBlinkSEN prototype system based on TI′s AWR1642 millimeter wave radar platform and validates the effectiveness of mmBlinkSEN with 14,000 sets of data. Experimental results show that mmBlinkSEN detects blink frequency with up to 88% accuracy in the presence of accompanying human interference.

Abstract:To solve the problem of requiring global optimization, real-time obstacle avoidance, and safe and reliable trajectory in surface ship path planning, a surface ship path planning method based on A ∗ algorithm and DWA algorithm is proposed. Firstly, the heuristic function dynamic weighting strategy is introduced to improve the search efficiency of A ∗ algorithm. Then, considering the motion characteristics of surface ships, an Angle weakening strategy of the path Angle node is adopted to reduce the angle and shorten the global path length. Finally, the trajectory evaluation function of the DWA algorithm is improved based on the influence of the global factors and track safety constraints, and the algorithm fusion is completed by providing subentry points of global path to guide the DWA algorithm to carry out local planning. Experimental results show that the total steering Angle of the proposed algorithm is reduced by 45. 6% and 46. 0% , respectively, compared with the existing fusion algorithms, which verifies the effectiveness and feasibility of the proposed fusion algorithm, and has more advantages over other traditional algorithms.

Abstract:In response to the issues of misjudgment of location area attribution and interference from outliers in fingerprint-based positioning of a large number of access points scenes, a multi-scale signal fusion indoor positioning algorithm is proposed, which incorporates dual-source signals. During the fingerprint online positioning phase, the spatiotemporal information of PDR signals is utilized to expand the number of reference points belonging to the location area, thereby alleviating the negative effects caused by misclassification in neighboring areas. Additionally, multiple distances and chi-square distances are used instead of the traditional Euclidean distance, in combination with spatial domain physical distance scales, to implement nearest neighbor selection at multiple scales. In this way, the interference from outliers is overcome effectively. We introduce a dynamic adaptation of the K value. Based on this, the dynamic linked fusion between Wi-Fi and PDR pre-positioning is established, which further enhances the accuracy of the positioning algorithm. Experimental results show that, under the same conditions of introducing dual-source signals, the proposed method exhibits superior overall performance compared to other multi-scale or dynamic K-value algorithms, with an average positioning accuracy surpassing other algorithms by 6. 6% to 23. 1% .

Abstract:Object state estimation always suffers low accuracy in complex traffic flow scenario due to noise interference and vehicle driving state changing. To solve these problems, a Kalman filter-based multi-object full lifecycle state estimation method is proposed for millimeter-wave radar, which includes both parameter initialization and online updating. Firstly, the Kalman filtering-based model is designed for multi-object full lifecycle state estimation in complex traffic flow scenario. Then, a data-driven approach is innovatively proposed for the observation noise covariance matrix initialization in Kalman filter. Furtherly, a variational Bayesian method is applied to update the Kalman filter parameters online for further enhancing the accuracy of multi-object full lifecycle state estimation. Finally, experimental data collecting from real vehicles are utilized to analyze the proposed method. The results show that the mean square error of this method is 0. 153 in multi-object state estimation, which is reduced by 36. 2% when compared with that of traditional Kalman filter. The comparison results evaluate the effectiveness of the proposed method on vehicle perception.