Abstract:In response to the technical challenges posed by the varying sizes, haphazard arrangement, and overlapping occlusion of prohibited items in security X-ray images, we propose an enhanced HRNet-based multi-scale feature fusion network model. This model aims to achieve automatic segmentation and recognition of prohibited items in images. In the encoding stage, we leverage the multi-resolution parallel network architecture of HRNet to extract multi-scale features, addressing the diverse scale of prohibited items in security X-ray images. In the decoding stage, a multi-level feature aggregation module is introduced that uses data-dependent upsampling instead of bilinear interpolation. upsampling to reduce information loss during aggregation, thus ensuring a more comprehensive representation of the features of the features extracted in the coding stage for a more complete characterisation of objects. In the overall architecture of the network, a de-obscuration module based on the attention mechanism is embedded to strengthen the edge-awareness ability of the model, alleviate the problem of serious overlapping occlusion of items in security X-ray images, and improve the segmentation and recognition accuracy of the model. By experimenting on the public dataset of PIDray security check images, the results show that the average intersection ratio of 73.15%, 69.47%, and 58.33% are achieved in the three validation subsets of Easy, Hard, and Hidden, respectively, which are 0.49%, 1.17%, and 5.69%, respectively, and the overall average intersection ratio is improved by about 2.45%.

Abstract:In order to accurately determine the degree of abrasion of abrasive belts in grinding screw rotors, the degree of abrasion of abrasive belts is identified according to the law of change of the color characteristics and texture characteristics of the surface image in the abrasion process of abrasive belts. The texture features and color features of the surface image of the abrasive belt after the grinding process are extracted, and the abrasive belt wear degree is classified according to the surface roughness of the screw rotor in different grinding time periods. The classification performance of the support vector machine is greatly affected by its own kernel function and penalty function, so it is proposed to optimize the kernel parameter and penalty parameter of the support vector machine by using the aquila optimizer optimization algorithm, and to establish the wear degree model of AO-SVM sand belt image recognition. The experiment is completed by utilizing a self-developed special belt grinding device for screw rotors. The grinding parameters are set as follows: the linear speed of the grinding belt is 10 m/s, the axial feed speed of the workpiece is 50 mm/min, the cylinder pressure of the tensioning wheel is 0.35 MPa, the cylinder pressure of the active wheel is 0.5 MPa, and the grinding time is 25 min. The recognition accuracy of AO-SVM for the abrasive belt wear degree model reaches 92.5%, which is improved by 5.0% and 3.6% compared to the random forest algorithm (RFC) and the XGboost classification algorithm, respectively, and the convergence speed is fast. The degree of abrasive belt wear can be identified by the AO-SVM model through the color feature change and texture feature change of the surface image of the abrasive belt, which can effectively avoid excessive abrasive belt wear and damage to the workpiece, and provide theoretical guidance for judging the degree of abrasive belt wear and the time to change the abrasive belt when the abrasive belt is used to grind the screw rotor.

Abstract:Aiming at the existing remote sensing image target detection methods with low detection accuracy for small-size aircraft targets, inaccurate feature information transfer and insufficient information interaction, a remote sensing image aircraft target detection method based on discriminative feature extraction and context-awareness is proposed. A backbone network with a discriminative feature extraction module is designed to enhance feature extraction for multi-scale aircraft targets; an adaptive feature enhancement module is introduced to selectively focus on small targets and optimize the transfer of feature information and information interaction; and a feature fusion up-sampling module is designed to perform up-sampling operations on the feature maps to improve the accuracy of high-level semantic information. The detection accuracy on the DOTAv1 dataset reaches 95.2%, which is 3.7% to 18% higher than that of mainstream algorithms such as YOLOv5s, SCRDet, ASSD. In addition, the detection speed and the number of model parameters of the proposed method are 147 frames per second and 13.4 M, respectively. Compared with the current mainstream algorithms, the proposed method has strong competitiveness and meets the real-time detection requirements of aircraft targets in the background of remote sensing.

Abstract:The recognition of vessel names (license plates) plays a crucial role in waterway transportation systems. Addressing the challenge of identifying vessel names in inland waterways, where targets are relatively small and vessels are observed at significant angular inclinations on both sides of the waterway, we propose an automatic ship name identification (ASNI) framework based on the differentiable binarization (DB) natural scene text detection algorithm and the convolutional recurrent neural network (CRNN) text recognition algorithm. ASNI comprises three main components: ship name detection, text image correction, and recognition. The text image correction component consists of a ship name correction module and a super-resolution reconstruction module. Firstly, the framework utilizes the DB algorithm to perform adaptive scale fusion processing on the candidate regions of vessel names in images, generating feature maps. Feature mapping is used to predict and generate binary images to identify connected regions, thereby obtaining regions of interest (ROI) containing vessel names. Subsequently, after ship name detection, a ship name correction module is introduced to rectify irregular text within the ROI using perspective transformation. Furthermore, a super-resolution reconstruction module is designed to enhance the resolution of the corrected vessel name images. Finally, the CRNN algorithm is employed to recognize vessel names within the corrected text images in the ROI, yielding the ultimate results. Through training and testing on the ship license plate (SLP) dataset specific to inland waterways, experimental results demonstrate that the ASNI framework achieves an average accuracy of 87.50% in vessel recognition, representing a 3.12% improvement over the baseline framework. The framework presented in this paper effectively addresses issues related to low resolution and angular inclinations leading to inaccurate vessel recognition. Compared to the baseline framework, ASNI exhibits superior recognition performance.

Abstract:A machine vision based surface defect detection method for microchannel flow channel plates is proposed to meet the demand for automatic detection of complex surface defects in industrial automation production. This method focuses on common pit and damage defects in the CV holes and expansion valve holes of the flow channel plate. Firstly, the ROI region is extracted through Hoff circle detection to eliminate background interference. Gaussian filtering is used to filter the ROI image, and binarization and morphological corrosion operations are used to filter out interference noise to highlight defect features. Then, the Two-Pass algorithm and seed filling method are used to calculate the connected domain to achieve pit defect detection. Use circle search to find the inner and outer circles of the hole end surface, unfold the circular ring, and use Canny edge detection operator to search for the defect contour, screen the contour area to achieve the detection of damaged defects. Through comparative experiments, it has been verified that the method proposed in this paper has a higher detection rate in the detection of defect samples in the runner plate compared to traditional surface defect detection methods. The method proposed in this article has been validated to have a stable defect detection rate of over 92% on the surface of the flow channel plate, and the algorithm has fast processing speed and strong robustness, achieving fast, non-contact high-precision detection and meeting the requirements of industrial automation.

Abstract:In order to lighten the rust image segmentation network model and eliminate the interference of nonsingle feature background and similar feature backgrounds such as rust liquid, this paper replaces the encoded part of the U-Net network model with the MobilenetV3_large network, imports the pre-trained weights of the MobilenetV3_large network based on the ImageNet dataset, and replaces the ordinary convolution of the decoded part of the U-Net network model with a deep separable residual convolution. And add the attention-oriented AG module and the Dropout mechanism in the process of upsampling. Experimental results demonstrate that the improved U-Net network model designed in this paper exhibits significant advantages in rust image segmentation under non-uniform feature background and similar feature background interference such as rust liquids. The model size is reduced by 81.18% compared to the original U-Net network model, resulting in a decrease of floating point calculations by 98.34%. Additionally, the detection efficiency has improved by 3.27 times, increasing from less than 6 frames/s to 19 frames/s. While the network model is lightweight, the accuracy of the network model is 95.54%, which is 5.04% higher than the original U_Net network model.

Abstract:The location of target corner points in an image is the key data for implementing many computer vision tasks. In order to overcome the data redundancy problem arising from traditional detection algorithms, an edge-based corner point target detection method in the scale space is proposed. First, a grouped multilayer scale space is constructed, and multiple smoothed images are obtained after projecting the original image into it. At the same time, the defined edge operator is applied to detect all edges in the smoothed image to obtain multiple sets of pixel points stored in order, and the transformation to larger scales is stopped when the number of point sets is stable. Then, at the current scale, the indicator values of each element in the point set reflecting its corner intensity are calculated. The support set interval of corner points is detected according to the variation pattern of these indicator values, and the final target corner points are determined in this interval using a Gaussian fitting function. Experiments show that the method is able to detect the target corner points with significant features and their angles, where the accuracy of the synthesized images is at the pixel level and the average error to figure ratio in the application case is about 1.5/100.

Abstract:In order to tackle the challenges associated with small character regions, complex lighting variations, and poor image quality in the identification of slab numbers in steel continuous casting production lines, a two-stage algorithm is proposed for slab number detection and recognition, utilizing deep learning techniques. Firstly, datasets for slab number detection and recognition are prepared based on the collected slab images of a continuous casting production line. Secondly, in the slab number detection stage, an AD-PAN feature fusion structure based on the DBNet algorithm is designed to enhance the multi-scale feature fusion capability and expand the receptive field of the detection algorithm, thereby improving the localization accuracies of the slab numbers. Thirdly, in the slab number recognition stage, the proposed algorithm incorporates a SPIN correction network and a SVTR slab number recognition network for end-to-end training, enabling them to actively transform input brightness and improve color distortion among characters and between characters and backgrounds. Finally, comparative experiments are conducted on self-made datasets for slab number detection and recognition, and experimental results demonstrate that the algorithm proposed in this study is capable of effectively locating slab at different positions on the roller table and robustly recognizing slab numbers in complex backgrounds. The Hmean value for slab number detection is 97.92%, and the accuracy for slab number recognition is 97.33%, confirming the high accuracy of slab number detection and recognition achieved by the proposed algorithm.

Abstract:The structural deformation of subsea pipelines increases the risk of pipeline fracture, which can cause significant economic losses. This paper proposes a pipeline displacement and deformation detection method based on a spherical inner detector equipped with a multi-channel magnetic sensor array. The stress in the pipe wall and the change of the magnetic flux density inside the pipeline caused by the displacement deformation of the pipeline are analyzed according to the magneto-mechanical effect. The structure layout of the detector and the magnetic signal acquisition scheme are designed. The collection of spatial rotational magnetic flux density information inside the pipeline is realized, and the rotational magnetic signal is converted into translational magnetic signal in the pipeline coordinate system. By extracting the magnetic data on the contour lines inside the pipeline, multiple channels of signals are integrated to increase the space sampling coverage of the magnetic signal inside the pipeline. Finally, internal inspection tests were conducted on pipeline vertical and equivalent lateral loading-induced displacement deformation. The results showed that under longitudinal force deformation state, the magnetic flux density longitudinal component dispersion decreased by 136.1 μT on the contour line inside the pipeline; under equivalent lateral force deformation state, the magnetic flux density lateral component dispersion decreased by 123.1 μT on the contour line inside the pipeline. The magnetic flux density dispersion measured by the spherical internal detector can reflect the displacement deformation of the pipeline in different directions, and the larger the deformation of the pipeline, the smaller the magnetic flux density dispersion.

Abstract:In pipeline transportation, the measurement results of turbine flowmeter are easily deviated by unsteady flow. In order to guide the selection and correction of turbine flowmeter under pulsating flow condition, it is necessary to estimate the maximum pulsation error of turbine flowmeter under pulsation conditions. Ignoring the change of fluid pressure and temperature, based on dynamic characteristics of turbine flowmeter and conservation of energy, the upper limit formula of sinusoidal pulsating flow error is derived by calculating the work ratio of pulsating flow and steady flow to the impeller in one pulsating period. A CFD simulation model of turbine flowmeter was established based on 6DOF model and UDF. The air flow test platform is built, the working grade standard device is used as the reference, and two flowmeters are connected by a tube section with built-in rectifiers. Compared with the flow value under constant flow, the error limit under each pulsating condition is calculated. The maximum pulsation difference between the results of the formula and CFD simulation is 0.281%, and 0.224% between the results of the formula and the air flow experiments. The upper limit formula for pulsation error is proved its accuracy when compared to the results of CFD simulation and air flow experiment. The pulsation error upper limit formula is proved to be accurate, and can be directly used to estimate and correct the value under the condition that the pulsation frequency is greater than 10 Hz.

Abstract:Extracting feature parameters from the recovery voltage polarization spectrum is currently a widely adopted method for evaluating the status of transformer oil-paper insulation. However, the polarization spectrum is prone to anomalous feature data due to factors such as working condition interference and artificial errors, which seriously reduces the accuracy of the evaluation. In response to the above issues, this paper proposed a recovery voltage data cleaning method based on local outlier factor (LOF) and improved K-nearest neighbor (IKNN). Firstly, Maximum recovery voltage Urmax, the initial slope Sr and dominant time constant tcdom of the recovery voltage polarization spectrum were selected as aging feature parameters, and anomalous feature data in the non-standard polarization spectrum were identified and filtered out based on the LOF algorithm. Secondly, the Fuzzy C-means (FCM) clustering algorithm was used to reduce the interference of noise points on the KNN algorithm, and the correlations between various features were highlighted by weighted Euclidean distance scale. Then, a data filling model architecture based on IKNN was constructed to fill in missing feature data. Finally, multiple sets of measured data were incorporated to validate the effectiveness of the proposed data cleaning method. The results indicate that the accuracy of status evaluation after data cleaning has increased by about 50% compared to the original data, which effectively improves the quality of transformer recovery voltage data and lays a solid foundation for accurate perception of transformer operation status.

Abstract:To guarantee the safety control performance of monorail cranes operating in complex track conditions within deep mines, enhancing the accuracy and reliability of dynamic inclination recognition for monorail cranes is necessary. Therefore, this paper proposes a dynamic inclination recognition method of monorail crane based on DFFRLS-AUKF algorithm. Firstly, an adaptive smoothing filtering algorithm is used to filter the acceleration and velocity data collected in real-time to avoid the interference of environmental noise and ensure the integrity of the data. Secondly, the track curvature model is established to achieve the accurate analysis of the entire working conditions of the track, and based on the filtered data, a reliable track curvature value is obtained by combining the dynamic recursive least squares of forgetting factor (DFFRLS) algorithm with the dynamic forgetting factor. Finally, based on the calculated track curvature, the unscented Kalman filter (UKF) is improved by using the Sage-Husa noise estimator, which achieves the self-adaptation of the dynamic Adaptive adjustment of dynamic inclination recognition, and the accuracy of emotional inclination recognition is improved. Experiments show that the proposed DFFRLS-AUKF algorithm improves the dynamic inclination recognition accuracy by 25.25% and 39.5% on average compared with the traditional algorithm during the testing of monorail crane in monorail section 1 and monorail section 2, which demonstrates that the DFFRLS-AUKF algorithm has good accuracy and reliability under different track conditions, and effectively guarantees the safety of monorail crane driving under complex track conditions.

Abstract:In the background of strong impulse noise, the performance of fractional low-order statistics delay estimation method is degraded and the prior knowledge of noise is required. In order to solve the problem, a new time delay estimation method based on second-order fractional loworder covariance is proposed. Firstly, the bounded nonlinear sigmoid function is used to process the signal with impulse noise, so that the additional impulse noise can be fully compressed without affecting the time delay information carried by useful signals. Then, the second-order fractional low-order covariance operation is carried out on the processed signals of receival and transmission, that is, after obtaining the self-fractional low-order covariance of the transmitted signals and the mutual fractional low-order covariance of the received and transmitted signals, the mutual fractional low-order covariance of the two is calculated again, thus, the effect of impulse noise can be further suppressed. Finally, the effectiveness of the proposed method is verified by simulation experiments. The results show that the proposed method is free from the restriction that the fractional low-order covariance index is less than the characteristic index of Alpha stable distribution noise, and has higher estimation accuracy than the fractional low-order covariance method. The simulation experiment results show that under the generalized signal-to-noise ratio of -10 dB, the delay estimation takes 0.056 0 s and the accuracy reaches 97.76%.

Abstract:Considering the problem that the sparse features in the existing traffic anomaly detection models are easily ignored by the feature selection algorithms, a traffic anomaly detection method based on feature coupling generalization (FCG) was proposed. First, the DBSCAN density clustering algorithm was used to remove outliers in the data to reduce the impact of the anomalies on the subsequent FCG algorithm. Second, the minimal-redundancy-maximal-relevance (mRMR) algorithm was used to sort the data features, and the most influential features for classification were selected to generate the class-distinguishing features (CDF) in the FCG algorithm, in order to enhance the classification ability. The K-nearest neighbors (KNN) algorithm was used to fill in the missing values in CDF to maintain data integrity. Then, the data were grouped according to attack categories, and the features were sorted using the mRMR algorithm respectively, and the sparse features with instance-distinguishing ability in the data of each attack category were selected as the example-distinguishing feature (EDF) in the FCG algorithm. The degree of coupling between the two features in the anomaly detection data and the upper concept of EDF were used to transform EDF into more generalized features. Finally, the processed data were fed into the random forest (RF) model based on Bayesian optimization (BO) parameters for classification and identification. Through simulation experiments on the NSL-KDD dataset, the accuracy reached 91.79%, which verifies the proposed method has a good detection performance.

Abstract:This paper addresses the optimization design problem of electromagnetic conductivity sensors for low conductivity measurements. The working principle of electromagnetic conductivity sensors is analyzed, and a physical model is constructed. The influence of excitation signal parameters, magnetic core dimensions, core spacing, excitation coil turns, and receiver coil turns on the output voltage is considered. An improved model for electromagnetic conductivity measurement is proposed, taking into account the multiplicity of excitation frequencies. Specifically, the excitation frequency directly affects the output voltage and also modifies the magnetic permeability of the core through the magnetostrictive effect, thereby influencing the output voltage. Theoretical analyses of the main parameters, such as frequency, spacing, and turns, are conducted using a comparative approach with experimental results. It is found that it exists an optimal frequency range where the output voltage remains relatively stable despite fluctuations in frequency. By adjusting the spacing between the two magnetic cores, coupling voltage between them is reduced, resulting in a higher proportion of effective signals and improved accuracy. The accuracy of the theoretical model is verified through experimental validation. Furthermore, the optimized parameters are employed in the design of a conductivity probe, which is calibrated using conductivity standard solutions. Comparative experiments are conducted with a German-made conductivity meter from Bode to calculate the Pearson correlation coefficient, demonstrating the accuracy and high reliability of the optimized model.

Abstract:Nonlinear friction negatively impacts the dynamic and static performance of hydraulic servo systems in robotic excavators, leading to issues such as trajectory creep, flat peaks, and steady-state errors. The traditional LuGre friction model, which relies solely on velocity and internal bristle state variables that cannot be accurately measured, fails to comprehensively describe the complex friction characteristics of excavator hydraulic servo systems. Considering the position, velocity, and direction of the excavator hydraulic servo system, we propose an enhanced LuGre friction model and introduce a velocity threshold to address the instability issue of the elastic bristle average deformation state observer in the friction model. Secondly, to address the issues of traditional optimization algorithms getting stuck in local optimal solutions and having slow convergence speeds, the basic particle swarm optimization algorithm has been enhanced. This enhancement involves the introduction of inertia weight, asynchronous change, and elite mutation operations to accurately and rapidly identify the six unknown parameters in the improved LuGre friction model. Subsequently, using the identified friction model, a friction compensation controller based on the principle of structural invariance is designed. Three different operating condition trajectory tracking experiments were conducted on a 23-ton excavator. The conventional proportional-integral-differential controller exhibits the highest tracking error, with the maximum tracking error for the triangular trajectory reaching 29.68 mm. In contrast, the feedforward friction compensation controller, which is based on the enhanced LuGre model, achieves a significantly lower error of 9.70 mm, representing a 67.31% reduction in error. The experimental results demonstrate that the proposed friction compensation controller significantly enhances the trajectory tracking accuracy of the excavator.

Abstract:The state of health (SOH) is an important index for battery management system, and accurate SOH estimation is of great significance for ensuring safe and stable operation of battery. Extracting reliable and effective health features to describe the aging state of battery and constructing accurate and stable estimation model are the main problems we face at present. In order to improve the accuracy of SOH estimation, a fuzzy entropy and particle filter (PF) based SOH estimation method for lithium-ion battery is proposed. Firstly, the fuzzy entropy value is extracted as the aging characteristic of the battery by analyzing the discharge voltage data during the aging process. Secondly, a non-parametric state-space model to describe the aging characteristics of lithium-ion battery is constructed based on the metabolic grey model (MGM) and the temporal convolutional network (TCN). Finally, the closed-loop SOH estimation of lithium-ion battery is realized by PF. In addition, the proposed SOH estimation method is validated using the NASA lithium-ion battery datasets and compared with other methods in the field. The results show that the maximum estimation error of the proposed method is about 5%, the estimation accuracy is improved by about 50% compared with similar methods, and the proposed method exhibits good robustness under different training cycles, which verifies the feasibility and superiority of the proposed method.

Abstract:Due to the presence of interference between users, power allocation problems in wireless communication networks are often non convex and require a huge amount of computation. The current graph neural network (GNN) has become an effective computational method used to solve this problem. In order to maximize network transmission speed while reducing computational complexity, a flexible duplex network graph representation method that incorporates device and communication connection attributes into GNN is proposed, and a corresponding flexible duplex graph neural network (FD-GNN) model is constructed. For the first time, the distance between node pairs, channel gain, and neighbors are introduced as dynamic thresholds into FD-GNN to adapt to dynamic environments. Excluding channel state information of neighbors in GNN, pruning edges in FD-GNN reduces computation time and network complexity. Simulation results show that the proposed threshold setting method based on channel gain neighbors has the best performance and reaches 97% of the weighted minimum mean square error (WMMSE), reducing the training time required by 24% compared to Full GANN. The proposed threshold based effectively reduces the time complexity of GNN operations and improves the effectiveness of the algorithm.

Abstract:The existence of virtual reality motion sickness is a key factor restricting the further development of the VR technology industry, the study of neural activity related to virtual reality motion sickness and its accurate detection is the premise to solve this problem, neural activity in resting-state virtual reality motion sickness missing from previous studies. Therefore, this study uses the resting Electroencephalogram（EEG）signals before and after the virtual reality motion sickness exposure task, and proposes the virtual reality motion sickness EEG characteristics as indicators to realize the detection of virtual reality motion sickness. First, the variational mode decomposition is performed on the EEG signals of five electrodes selected by statistical analysis in this paper, namely Fp1, Fp2, F8, T7 and T8, and extract the sample entropy, permutation entropy and center frequency from the selected modal components. Then, two stages of feature selection are performed by statistical tests and ReliefF algorithm. Finally, the selected feature vectors are sent to the support vector machine for classification, then the automatic detection of motion sickness in virtual reality is realized. The results showed that the accuracy, sensitivity and specificity of this method reached 98.3%, 98.5% and 98.1%, respectively, and the area under the ROC curve reached 1, it is superior to other methods, which proves the advantages and effectiveness of this method in the automatic detection of EEG signals in virtual reality motion sickness.

Abstract:Mild cognitive impairment (MCI) is a key stage in the diagnosis of senile dementia, and the characteristics of electrical brain (EEG) signals can reflect the cognitive status of MCI patients and help achieve early diagnosis. In the process of EEG feature extraction, most existing studies use fixed time windows to complete segmentation processing for each rhythm of EEG, ignoring the feature differences of different rhythms, thus affecting the diagnostic effect. In view of this problem, a new combined sliding window optimization algorithm is proposed, which improves the construction method of zero model by iterative amplitude adjustment Fourier transform (IAAFT), so as to evaluate the brain dynamic characteristics KPLI. By adopting a variety of sliding window combinations for EEG frequency band signals and guiding them with KPLI indicators, the best sliding window combinations suitable for different frequency bands are obtained. Based on the best sliding window combination, the phase lag index (PLI) is extracted from each band combination, the continuous wavelet transform (CWT) feature is performed, and the MCI diagnosis is realized through the ResNet-MLP dual-channel classification network. The results show that diagnostic classification was achieved for 88 subjects (32 MCI patients, 36 Alzheimer’s disease patients, and 20 healthy control) using a personalized combination band sliding window, and the classification accuracy was 82.2%, which is 10% higher than the classification of fixed window (72.2% classification accuracy is obtained). The results showed that based on the individualized EEG rhythm feature combination, the features of MCI could be better extracted, and the accuracy and specificity of the diagnosis of mild cognitive impairment could be improved, which was an effective EEG feature extraction method.

Abstract:The detection performance of resonant force sensor depends on the geometry dimensions, structure configuration and sensing mechanism of the resonant sensitive element. At present, the method of improving the detection performance by simply reducing the size has been in the bottleneck period. In order to study and develop a novel resonant force sensor, coordinate the contradiction between nonlinear vibration and resonant structure detection performance, explore a more sensitive sensing mechanism and improve its detection performance, a piezoelectric driven resonant magnetically coupled cantilever force sensor is proposed. Firstly, the design and theoretical modeling of the structure of the magnetically coupled cantilever beam are carried out. The influence of the external pressure on the vibration characteristics of the magnetically coupled cantilever beam structure is analyzed theoretically. With the increase of the pressure, the distance between the magnetically coupled cantilever beams decreases and the resonance frequency increases. Secondly, the experiment verifies the advantages of the bifurcation jump dynamic behavior, which increases the maximum amplitude by 2.8 times compared with the resonance of a single resonant beam. Then, two pressure detection schemes based on the bifurcation jump characteristics and frequency doubling response are studied, and the pressure detection is realized by using the critical frequency of the bifurcation jump and the high order response frequency when the mode is coupled. The sensitivity and linearity are analyzed. The experimental results show that the amplitude change of the detection scheme based on the bifurcation jump characteristic is obvious, which is about 5 times that of the detection scheme based on the frequency doubling characteristic, and it is easy to detect and overcome the adverse effects of nonlinear factors. The detection scheme based on frequency doubling response has high output sensitivity, which is about 4 times that of the detection scheme based on bifurcation jump, and large signal-to-noise ratio, which provides a certain reference value for designing resonant force sensors with different detection principles.

Abstract:Aiming at the data acquisition in large sparse sensor networks, a network architecture using the ubiquitous existence of mobile agents in environment to connect sparse sensors and a 2-dimensional grid random walk analysis model are proposed in this paper. The proposed sensor network model consists of three abstract layers, namely, bottom layer composed of wireless sensors, the middle layer composed of various transportation agents, and the top layer composed of access points/central repositories. The specific implementation principle is that the mobile transport agents located in the middle layer collect data from the wireless sensors distributing at the bottom layer and buffer the data, and after wandering transport, finally deliver the data collected from the wireless sensors at the bottom layer to the necessary access points at the top layer for necessary storage and processing, so as to achieve the data acquisition of the entire sensor network. The theoretical analysis and simulation experiment results show that the proposed sensor netwoks model based on mobile transport agents not only has robustness and scalability, but also has obvious advantages over base station network model and Ad-hoc network model in terms of sensor power consumption, data success rate and infrastructure invested cost.

Abstract:To solve the problem of low output accuracy of MEMS gyro in MWD, an online gyro error compensation method based on magnetic-gravity ephemera algorithm (MGMA) is proposed. Firstly, the error source of GYRO while drilling is analyzed and the error compensation model is derived. Secondly, the objective function of MEMS accelerometer is obtained by using the cross product of gravity vector. In addition, considering the adverse effects of strong vibration and impact on the accelerometer during drilling, the relative error constraint of the magnetic mode value is designed based on the strong anti-vibration ability of MEMS magnetometer. Then, on the basis of MA, the search upper and lower bounds are determined adaptively according to the relationship between the gyro and the magnetometer output, aiming at the constant change of gyro error parameters under the influence of harsh environment while drilling. The relative error of gravity mode value is used to design the inertia weight and balance the global exploration and local development ability of the algorithm. Finally, according to the relative error of the magnetic-gravity mode value, the variation perturbation strategy is introduced in the children to reduce the possibility of falling into the local optimal. The experimental results show that the gyro error after MGMA compensation is obviously reduced, and the well inclination error is reduced from 9.75° to 1.52°, and compared with PSO and MA algorithm, it has the advantages of fast speed and high precision.

Abstract:Aiming at the randomness and complexity of jointless track circuit faults, the single diagnosis model has the problems of one-sided extraction features and unreasonable empirical design of model structure. A fault diagnosis method based on intelligent optimization deep network is proposed. Firstly, the fault feature set is established by six voltage detection quantities of the track circuit signal centralized monitoring system. The convolutional neural network (CNN) is used to extract the feature space information, and the long short-term memory network (LSTM) is used to extract the time feature information, so that the features extracted by track circuit fault diagnosis have both spatial and temporal information. At the same time, the genetic algorithm (GA) is introduced to optimize the structure and parameters of the aforementioned deep neural network, and the output weight of the feature level of the two combined networks is further optimized by combining the Q-learning method in reinforcement learning. Finally, the multi-layer perceptron (MLP) is used to fit and correct the classification error of the deep network to improve the fault diagnosis accuracy of the model for the track circuit. The simulation results show that the recognition rate of the fault diagnosis of the track circuit using the intelligent optimized deep network model can reach 99.28% compared with the single model and the refined design combination model. The evaluation index is improved, and the fault diagnosis accuracy is higher. It is proved that the intelligent optimized deep network can further improve the fault diagnosis performance of the track circuit.

Abstract:UWB (Ultra-Wide Band, UWB) technology is one of the main approaches for obtaining the location coordinates of unmanned equipment while inspecting in large enclosed spaces with multiple obstacles. Currently, the design of UWB location base stations topology mainly relies on manual methods, which have many shortcomings such as low design efficiency, unknown design effect, and difficult determination of cost-effectiveness. To address these issues, this paper proposes a scientific and complete topological evaluation system of UWB location base stations, and a universal automated design method for the topology of UWB location base stations in large enclosed spaces. The method is based on simulation, genetic algorithm and topological evaluation system to automatically generate the optimal topology scheme suitable for practical application scenarios, including the optimal number and location information of base stations. The experimental results show that: 1) The topology generated by the universal design approach has a better comprehensive performance than the topology with the largest number of base stations, with an increase of 6.2% in overall performance; it also has a better comprehensive performance than the topology with the smallest number of base stations, with an increase of 21.2% in overall performance. 2) The topology generated by the universal design method has a better comprehensive performance than the serrated topology designed by humans, with an increase of 11.4% in overall performance and a design efficiency increase of 92.9%. Therefore, for the scenario of multiple obstacles in large enclosed spaces, the approach proposed in this paper can design a topology with better overall performance and higher cost-effectiveness in a shorter time, providing a foundation for unmanned equipment to conduct autonomous patrols and inspections in such large enclosed spaces.

Abstract:Testability verification and FMEA analysis of equipment consider single faults and use sampling methods to determine fault samples traditionally, without considering the different failure modes of redundant modules and the coverage rate of faults. Aiming at the requirements and difficulties of testability verification for launch vehicle electronic equipment, including redundant function verification requirements and testing coverage requirement etc., the functional degradation FMEA analysis method and evaluation method of coverage rate of fault modes was proposed on the basis of the general testability verification method of equipment, and a testability verification and evaluation method suitable for the launch vehicle electronic equipment was formed. An electronic equipment on the launch vehicle was taken as an example for modeling. The results proved that the method proposed in this paper effectively identified the redundant faults, and incorporate the fault modes of redundant modules into the evaluation of testability indicators. The faults of redundant modules were considered in the evaluation of testability indicators, and the verification of redundant functions of launch vehicle electronic equipment was assessed. The proposed fault coverage evaluation method and sample size supplement method effectively improve the coverage of testability verification for fault modes. It is suitable for testability verification and evaluation of launch vehicle electronic equipment, and can be extended to the electrical system of launch vehicle.