Abstract:Optical wireless communication, as a high-speed, high bandwidth, and high security communication technology, has received widespread attention. However, noise, as an important limiting factor of system performance, can cause system signal distortion and signal-to-noise ratio to decrease, thereby affecting communication quality. Therefore, in order to reduce the impact of noise on signals and design and adjust communication systems more effectively, it is necessary to understand and simulate the behavior of noise, and noise models are an important tool for studying noise behavior. This article systematically studies the noise and noise models in optical wireless communication systems. Firstly, the article analyzes each type of typical noise introduced in optical wireless communication systems from the perspectives of signal source, channel, and destination. Then, the research progress on corresponding noise and its noise models at home and abroad was summarized, and the mechanism of noise generation was analyzed, and corresponding noise models were provided. In addition, the characteristics and limitations of relevant models were summarized. Finally, key suppression technologies for various types of noise were summarized, and further research directions in this field were discussed, which can provide theoretical support and inspiration for the development of optical wireless communication technology and system design and optimization in this field.

Abstract:The stainless steel casing structure of large gas turbine engine contains a large number of R angle regions, which are prone to micro-cracks, pores, inclusions and other defects in the manufacturing process because of its large wall thickness and curved surface structure. Due to the shape limitation of the curved surface of the R angle, the radiographic film cannot be arranged, and the R angle area with large thickness is difficult to be penetrated by the ray, so that the sensitivity of the radiographic testing is reduced, and defects are missed. Therefore, the ultrasonic immersion phased array testing for the internal defects in the R corner region of the large stainless steel case is proposed.The phased array probe is arranged on the inner ring curved surface of the casing, and the array element is controlled to transmit ultrasonic waves to form a focused sound field in the R angle area by modifying the curved surface focusing rule. The numerical simulation model is established to analyze the focusing performance of the sound field before and after the modification of the focusing law, and the influence of the water distance on the focusing sound field is also analyzed. Based on the modified focusing rule and the optimized water distance, the ultrasonic phased array testing of the R focal region of the casing is carried out. The results show that a focused sound field can be formed in the R angle region by modifying the curved surface focusing rule and optimizing the water distance, and the image quality of the sector scanning in this region is significantly improved; the transverse hole defect with the equivalent size of Φ 1. 5 mm can be displayed with good resolution, and the quantitative relative error is as low as 6. 7% by use of -6 dB quantitative method.

Abstract:The automated detection of spikes in electroencephalogram is currently a prominent area of research, with significant implications for epilepsy diagnosis. There are primarily two types of existing detection methods: signal analysis and machine learning. The former is sensitive to outliers, while the robustness of the latter’s algorithms to different data has not been fully verified. Additionally, traditional spike detection methods based on single-channel EEG are susceptible to artifact interference. In response to the limitations of existing algorithms and considering the electrophysiological characteristics of spikes, we propose a spike detection algorithm based on multi-channel data weight fusion and wavelet decomposition. Firstly, a multi-channel weight fusion method is designed using amplitude and waveform trends as feature values to enhance single-channel data according to the discharge characteristics of epileptic spikes. Secondly, the algorithm introduces wavelet decomposition to effectively extract local features from the signal and enhance its ability to detect signals with mutation characteristics. Finally, clinical EEG data collected from epileptic patients verify that the algorithm can achieve accurate detection of interictal spikes at a diagnostic accuracy rate exceeding 92.3%. Compared with traditional single-channel EEG spike detection methods, this approach offers advantages such as high accuracy and simple calculation, making it an effective technology for interictal spike detection in epilepsy.

Abstract:In response to the limitations of traditional pulse sensors such as discomfort, inconvenience in wearability, and low precision, this study presents the development of a wearable pulse sensor based on P(VDF-TrFE) flexible piezoelectric film. The objective is continuous monitoring of human pulse signals, aiming to offer robust support for cardiovascular disease prevention and treatment. Firstly, P(VDF-TrFE) flexible piezoelectric films were prepared using the solution casting method to serve as the sensor substrate. Conductive electrodes were then printed on the surface of these films using screen printing technology. Additionally, a mesh shielding layer was incorporated into the design, and both square and circular array sensors were fabricated. These sensors were used for experimental comparison to evaluate their pulse signal acquisition performance. Secondly, to counteract the challenge of low-frequency weak pulse signals vulnerable to various noise interferences, a precision signal conditioning circuit was designed, integrating signal amplification and filtering functions to achieve high-fidelity, low-noise pulse wave signals. Experimental results demonstrate excellent dielectric, piezoelectric, and ferroelectric properties in the prepared P(VDF-TrFE) films, with ad33value reaching -25 pC·N-1, enhancing the sensor’s ability to rapidly and accurately capture low-frequency pulse signals. The designed flexible pulse sensor significantly outperforms conventional rigid sensors by conforming more effectively to the contours of human skin, thereby enhancing the sensation-free wearing experience. This design not only meets but salso exceeds the requirements for wearability and comfort, making it an ideal choice for continuous, unobtrusive health monitoring applications. Particularly, the circular array sensor exhibits higher sensitivity and clarity in detecting continuous pulse wave signals containing most physiological characteristic points compared to the square sensor, thus achieving superior detection performance. Moreover, the designed signal conditioning circuit effectively mitigates 50 Hz power frequency interference and high-frequency noise interference, successfully amplifying the average peak voltage from 0.069 V to 5.467 V. This results in clear and stable pulse waveforms while retaining the main features of pulse signals. The system achieves high sensitivity, stability, and accuracy in acquiring human pulse signals while suppressing noise interference. Consequently, the wearable pulse sensor based on flexible piezoelectric films developed in this study holds promise for effective detection and acquisition of human pulse wave signals, with wide-ranging applications in medical health monitoring and wearable device research fields.

Abstract:Aiming at the problem that the imaging features are inconspicuous and the proportion of objects is small in the optical remote sensing small ship detection under the complex marine scenes, such as sea-lean boundary and near-shore rocky reefs, an improved small ship detection method based on YOLOv8s is proposed. Firstly, the prediction layers are modified based on the introduction of shallow feature maps in the neck layers, which balances the weights of shallow locational information and deep semantic information, and enhances the attention of the model to small objects. Secondly, the C2f-FE module is adopted to utilize the channel grouping and the cross-channel information interactions, enhance the feature extraction of small ships, and reduce the model parameters, which merges the FasterNet Block and the efficient multi-scale attention mechanism. Finally, the dynamic detection head module is employed to improve detection capability of the model on different spatial scales and object tasks at different prediction layers. The experimental results show that compared with the original YOLOv8s model, the improved model reduces the number of parameters by 42.3%, the detection accuracy mAP50 and mAP50:95 values are improved by 4.2% and 2.2% on the MASATI dataset, and mAP50:95 values are improved by 1.7% and 1.4% on the DOTA-Ship and DOTA-Small Vehicle datasets, respectively. It can be concluded that the improved model not only achieves lightweight and accurate detection of small ships, but also satisfies the high-accuracy detection for the generalized of small objects in remote sensing scenarios.

Abstract:A gearbox oil state recognition method based on PCA feature optimization and AdaBoost ensemble learning is proposed to address the problems of low accuracy and limited generalization ability in traditional gearbox oil analysis methods. Firstly, the multi parameter oil data is cleaned using box plots and SMOTE interpolation to improve the quality of the oil data; Secondly, PCA is used for oil product feature optimization to obtain a subset of oil product feature optimization that is helpful for identification. While effectively integrating multi parameter information of oil, it can significantly reduce the time complexity of model operation; Then, a basic model for oil state recognition is established using BP neural network, and the GWO wolf pack optimization algorithm is introduced to optimize the model. A weak classifier GWO-BP with optimal initial weights and thresholds is constructed, and an adaptive boosting AdaBoost algorithm is adopted to combine multiple weak classifiers GWO-BP, integrating them into a strong classifier with strong robustness. Finally, the experimental data was applied for verification and analysis. The experimental results showed that the proposed method had the best performance, with an average recognition rate of 99.30 ± 0.16% and an average time of 32.77 ± 1.27 seconds. It could quickly, efficiently, and accurately identify the oil state of the gearbox lubricating oil, laying a good foundation for realizing online oil state recognition of gearboxes and having important engineering application value.

Abstract:Magnetic flux leakage (MFL) internal detection is the core technology of pipeline internal detection, which is crucial to ensuring the safe transportation of pipelines. Due to the long-term underground or deep sea environment of pipelines, there are many small defects on the surface of pipelines. Due to the limited information available on small defects, traditional deep learning defect detection methods have difficulty achieving satisfactory detection results for small defects. A composite backbone network-based signal detection method for small magnetic leakage defects is proposed. First, a data enhancement method called background compression is proposed to compress background signals and thus enhance key features of small defects. Secondly, an adaptive positive and negative sample allocation strategy is designed to address the issue of uneven positive and negative sample allocation for small defects in the region proposal network. Finally, a multi-branch high-resolution feature extraction network for small defects is proposed, which uses a multi-branch composite structure to obtain high-resolution features for feature fusion, thereby improving the network’s utilization of small defect texture information. The proposed method is validated using pipeline data from a test site, and the experimental results show that the proposed method is effective, achieving a detection accuracy of 90.3%, with an 8.4% mAP improvement compared to the best results.

Abstract:The third-generation power semiconductor device-gallium nitride high electron mobility transistor (GaN HEMT) has been widely used in the fields of power electronics and communication electronics due to its excellent voltage and temperature tolerance. GaN HEMT devices usually work under harsh external conditions such as high temperature and high power. In order to avoid the sudden failure of GaN HEMT devices from affecting the normal operation of power electronic equipment, it is of great significance to carry out active real-time state detection. By designing and conducting repetitive experiments under different temperature and drain-source voltage conditions, the energy of the device stress wave is extracted and analyzed to explore the effects of temperature and drain-source voltage on the GaN HEMT. Aiming at the problem that the device stress wave acquisition process is susceptible to noise interference, a stress wave denoising algorithm based on variational mode decomposition (VMD) of goose optimization algorithm is proposed. The experimental results show that the proposed GOOSE-VMD signal processing method can achieve good noise reduction while preserving the characteristics of stress wave signals to the greatest extent possible; there is a good positive correlation between the device stress wave energy and drainsource voltage; the energy of stress waves decreases with increasing temperature, but when the temperature reaches 82.05℃, the energy of stress waves increases with temperature.

Abstract:Rocket sled tests hold significant experimental value in the development of aerospace, weaponry, electronics, and nuclear weapons. To establish a precise measurement control network for the track and accurately obtain the spatiotemporal parameters during rocket sled tests, a combined measurement method based on a distance-angle mixed intersection adjustment model is proposed. Firstly, a mixed intersection adjustment model was constructed utilizing angle measurements from total stations and distance measurements from laser trackers. Principles for constructing the measurement error matrix were defined, and the global coordinates were optimally estimated using a nonlinear least squares method. Secondly, the Monte Carlo method was employed to simulate and analyze the measurement equipment layout and the accuracy of the mixed intersection adjustment model. Simulation results indicated that positioning the measurement equipment centrally within the measurement range minimized the overall coordinate measurement errors of position markers, thereby reducing initial value errors in the adjustment model and enhancing the model’s solution accuracy. Finally, experimental verification was conducted at a rocket sled test site. Within a measurement range of 669 m, the standard deviation of the position marker distances in the track measurement control network was found to be 0.19 mm, validating the feasibility of the mixed intersection adjustment model for long straight track measurements. This method offers significant reference value for full-range measurement tasks.

Abstract:Wi-Fi wireless sensing technology has become a research hotspot in the field of perception, which can realize intelligent perception of human activities and the surrounding environment. The existing wireless sensing models have a large number of parameters, which makes it difficult to sense in real-time in scenarios with limited computing power such as mobile devices. To this end, a classification and recognition model based on a mixture of a lightweight feature extraction module based on depth-separable convolution and a stacked gated recurrent unit is proposed. Firstly, a lightweight feature extraction module based on depth-separable convolution is constructed to capture the spatial features of human gestures and keep the temporal nature of the features unchanged; then the spatio-temporal features of human gestures are learned using a two-layer stacked GRU network; finally, the performance of the model is validated using the open-source dataset Widar, and the BVP features in the CSI information are extracted to improve the recognition of cross-domain scenes accuracy, and a weighted loss function is utilized to solve the sample imbalance problem. The results show that the proposed model achieves an accuracy of 77.6% in cross-domain scenarios with a parameter count of only 236.891 K. Compared with other existing Wi-Fi gesture recognition models, the proposed model greatly reduces the parameters and computational complexity of the model while its performance remains basically unchanged, which lays a foundation for the popularization of the Wi-Fi wireless sensing technology in practical applications.

Abstract:Six dimensional acceleration sensor can be widely used for dynamic holographic detection in the field of humanoid robotics, so as to guarantee the flexibility and stability of robot motion. The existing six dimensional acceleration sensor have the problems of slow response speed, narrow response range and other poor dynamic characteristics, which limit the sensor’s sharp and wide-range response to the realtime dynamic position information of the measurement carrier. To address this problem, a study on the compensation of the dynamic performance of the six dimensional acceleration sensor in the time-frequency domain is carried out. The dynamic model of the sensor is established using differential equations, a high-precision dynamic model parameter identification algorithm based on NDE-FLNN is proposed, and the dynamic compensation model is further derived to compensate for the response speed of the sensor and to enhance the dynamic performance of the sensor in the time domain. After that, the dynamic compensator of each channel of the sensor is designed based on the zero-polarity configuration method, which eliminates the original poles and introduces new poles to expand the response range of the sensor and improve the dynamic performance of the sensor in the frequency domain. The experimental results show that compared with the DE-FLNN algorithm, the improved NDE-FLNN algorithm is able to identify the dynamic model parameters of the sensor with higher accuracy, and the adjustment time of each component of the sensor compensation is reduced to about half of the original one, which is within 150 ms, and the operating bandwidth is expanded from 22 Hz to 84 Hz, so the dynamic performance of the sensor in the time-frequency domain has been significantly improved.

Abstract:To address the problem of insufficient grasp detection accuracy of existing grasp detection methods in complex unstructured grasping scenarios due to the conflict of angle training labels and the non-consistency between graspable regions and object regions, this paper proposed an adaptive feature fusion grasp detection network, AFFGD-Net. The network firstly adopted the angle prediction method based on the partition method, which encoded the angle values into two parts, namely, angle category and offset for learning and prediction. The conflict angle values were divided into the same category to reduce the conflict of angle training labels, and the offset was used to compensate for the loss of accuracy in the classification part to improve the prediction accuracy of the network for grasp angle. Secondly, the adaptive receptive field block ARFB and attention skip connection module ASCM are introduced. ARFB enhanced the network’s ability to characterise the features of multi-scale graspable regions, and improved the grasp detection ability of multi-scale objects by adaptively fusing features of different scales. ASCM recovered the edge features of the graspable regions by adaptively fusing the low-level spatial features and the high-level semantic features, which improved the network’s grasp angle and grasp width prediction accuracy. Finally, the effectiveness of the proposed network was verified by experiments. The accuracy of AFFGD-Net reached 98.9% and 97.7% in the image segmentation and object segmentation test modes in the Cornell dataset, respectively, and 95.2% in the Jacquard dataset. The detection speed of the network reached 111 FPS, which showed good real-time performance. The experimental results showed that AFFGD-Net outperformed the existing methods in terms of both accuracy and real-time crawl detection, confirming the effectiveness of the proposed method.

Abstract:To solve the map construction problem of the quadruped robot for natural gas pipeline inspection, an ISCSO-FastSLAM algorithm optimized by the improved sand cat swarm algorithm is proposed. Firstly, the Cauchy variation strategy is introduced to improve the ability of the sand cat swarm algorithm to jump out of the local optimum and accelerate the convergence speed, and the adaptive genetic parameters are added to improve the stability of the sand cat swarm algorithm. Then, the predicted particle set of the FastSLAM algorithm is updated by improving the optimal solution of the position prediction output of the sand cat swarm algorithm to improve the estimation accuracy. Meanwhile, the low weight particle optimization strategy is used to replace the original resampling step in particle filtering to ensure the diversity of particles. Then, different simulation environments are constructed to compare the different algorithms, and the simulation results show that the ISCSO-FastSLAM algorithm constructs the map more accurately than the WOA-FastSLAM algorithm, and the estimation errors of the robot position and the environmental signposts are reduced by 17.1% and 23.3%, respectively, under the simulation environment of 20 m×20 m. Finally, the quadruped robot is used to conduct map construction experiments in a residential area of 60 m×100 m, and the experimental results show that, compared with the FastSLAM algorithm and the WOA-FastSLAM algorithm, the ISCSO-FastSLAM algorithm is able to construct a more accurate map of the residential area inspection, and the estimation errors of the key inspection locations such as valve wells and regulator boxes are reduced by 16.2% and 6.0%, respectively.

Abstract:Coated pipelines are widely used in industries such as chemical, petroleum, and gas. Cracks and localized corrosion caused by pipeline corrosion may pose significant safety hazards, making the detection of pipeline defects extremely important. However, the thicker coating on the pipeline makes it difficult for conventional non-destructive testing to detect defects in the pipeline. Pulse eddy current non-destructive testing technology, due to its strong excitation energy and excellent penetration ability, can detect defects in the pipeline without removing the coating. This study aims to analyze the effectiveness of defect detection under different signal characteristics. Firstly, a three-dimensional finite element model of a pipeline with a coating layer is established using simulation software. Secondly, differential voltage peak, differential voltage peak time, differential voltage zero crossing time, and differential signal fundamental frequency amplitude are selected as signal features to analyze the relationship between signal features and defects. Finally, a more suitable evaluation signal is selected. The simulation results show that the peak value of differential voltage and the amplitude of fundamental frequency increase with the increase of defect arc length and with the increase of defect depth. The peak time and zero crossing time are only related to depth, and increase with the depth of the outer surface and decrease with the depth of the inner surface. And by fitting signal features with defects, select signals suitable for defect resolution. This study helps to optimize the application of pulsed eddy current non-destructive testing technology in defect detection of coated pipelines, improving the accuracy and efficiency of detection.

Abstract:Aiming at the deficiency of similar motion recognition in the current UWB radar attitude recognition research domain, a motion recognition model integrating time-frequency analysis and random forest (RF) is proposed. A time-frequency analysis method of two-dimensional smoothed pseudo Wigner-Ville distribution (2D-SPWVD) based on smoothed pseudo Wigner-Ville distribution (SPWVD) is proposed to extract the time-frequency features of the preprocessed human motion echo signals. Principal component analysis (PCA) was employed to reduce the dimension of the feature vectors, and the top 30 principal components with a high cumulative contribution rate were selected as new feature vectors to be input into the RF classification model optimized by sparrow search algorithm (SSA) for the identification of five distinct human similar drop actions in the presence of obstacles. The experimental outcomes demonstrate that the pretreatment algorithm can effectively enhance the SNR of the action echo signal, and the PCA-SSA-RF classification model can effectively distinguish five different human fall movements, overcome the particularity of data and the interference of obstacles, with an accuracy rate as high as 96.6%. In the fall detection task within the real-time data stream, the average classification accuracy of the model reaches 93%, and it is profoundly compared with RF, PSO-RF and other diverse classical classification models, featuring high accuracy and short overall time, and possessing both accuracy and classification efficiency. The superiority and effectiveness of the proposed method are verified.

Abstract:In semiconductor manufacturing, wafer defect classification is an important step in ensuring product quality. However, due to the diversity and complexity of wafer defects, the existing hybrid wafer defect classification network still has shortcomings in accuracy. To solve this problem, a hybrid wafer defect classification network based on global and local multi-scale feature fusion—MLG-Net was proposed. MLG-Net consists of three main modules: feature extraction module, global branch, and local branch. The network aims to better extract and utilize the global semantic information and local detail features of wafer defect images, which are combined with multi-scale feature fusion technology to form a more comprehensive feature representation, which helps the classifier to make more accurate judgments in the face of complex mixed defects, thereby improving the classification accuracy. To verify the effectiveness of MLG-Net, a large number of experiments were carried out on MixedWM38, a dataset containing 38 mixed types of defects, and the classification accuracy reached 98.84%. The results show that MLG-Net is superior to the six mainstream wafer defect classification methods in terms of comprehensive performance. This result demonstrates the importance and effectiveness of global and local feature fusion in dealing with hybrid wafer defect classification tasks.

Abstract:To address the challenges of low contrast and small defect sizes in some X-ray images of tires, which make detection difficult, an improved model based on generative adversarial networks (GANs) is proposed to enhance the accuracy of tire defect detection. Initially, issues with traditional generators are analyzed. Building upon the GANomaly model, the proposed approach incorporates the attention mechanism module (NAM), flow alignment module (FAM), and PatchGAN to enhance feature extraction and image reconstruction capabilities. The NAM enhances the model’s focus on defect areas through normalization, while the FAM accurately maps features from low-resolution to high-resolution feature maps, ensuring information consistency and effective fusion across multiple scales. PatchGAN, with its local discriminator, improves the model’s ability to recognize local features. Validation tests on a self-constructed dataset of four tire defect types demonstrate significant improvements in key metrics, achieving an AUC of 96.4% and an AP of 95.8%. These results indicate enhanced feature extraction and image reconstruction capabilities, leading to improved defect detection accuracy.

Abstract:In order to improve the transmission efficiency of the dual active bridge converter, a dual-objective weight optimization strategy for current stress and backflow power based on the new extended phase shift control is proposed. Firstly, the new phase-shift ratio is redefined according to the angle of the high level of the primary-side output voltage and the phase-shift angle between the primary and secondary-side output voltages to simplify the mathematical model of the transmission power, and then three operation modes are classified by the relationship between the phase-shift angles and the corresponding mathematical models of the current stress and the transmission power are derived; Based on it, the mathematical model of the backflow power is deduced and the characterization is carried out; Then, two operating modes covering the full power are selected and the current stress and backflow power weight optimization functions are established, the optimal shift ratio combinations are solved according to the polarity regularization method and the soft-switching characteristics are analyzed; Finally, a simple and fast closed-loop control strategy is designed by combining the optimal shift ratio combinations with the proposed soft-switching conditions. A prototype is built for experimental verification, and the optimization strategy is compared with the traditional one in terms of current stress, backflow power and transmission efficiency. The experimental results show that the dual-objective optimization strategy improves the system efficiency by 20% in low-power mode and 11% in high-power mode compared with the traditional extended phase-shift control, which verifies the feasibility and effectiveness of the design scheme.

Abstract:To address the issues of low accuracy, slow convergence, and difficult data acquisition in intelligent search algorithms for solar cell parameter estimation, we propose a method that combines second-order Bézier curves with an enhanced Squirrel Search Algorithm. First, the optimum Bézierr control point is found on the line that passes through the maximum power point and is parallel to the line of the open circuit voltage point and the short circuit current point. This approach leverages the relationship between control point positions and battery fill factor to achieve precise modeling of the I-V characteristic curve without the need for experiments. This method not only accurately describes the output characteristics of HIT cells but also effectively reduces the impact of measurement noise on parameter identification. Secondly, we introduce Sobol sequences, reverse learning, and chaos theory to improve the standard squirrel algorithm. Sobol sequences are integrated into the initialization process as quasi-random samples, and a reverse learning strategy enhances population diversity and search space coverage. Additionally, a tent chaotic mapping perturbs the optimal solution, enhancing the algorithm’s capability to escape local optima. The improved squirrel optimization algorithm is applied to heterogeneous junction solar cell parameter estimation and compared with other intelligent optimization algorithms. The results showed that the improved algorithm achieved root mean square errors of 0.028 25, 0.017 458, and 0.023 61, respectively, indicating the highest accuracy. This demonstrates the effectiveness and accuracy of the algorithm in the parameter identification of heterojunction solar cells, providing a reliable and precise new method for solar cell parameter identification.

Abstract:The influence of high-speed airflow field on the contact pressure and arc state between the pantograph carbon slide plate and the catenary during the operation of high-speed train was analyzed. By calculating the contact pressure and arc state models that are more in line with the actual state, an experimental model of pantograph arc considering the influence of high-speed airflow field is established. In this paper, a 1D-2D-GAF-PCNN-GRU-MSA fault detection model based on the improved black-winged kite algorithm （IBKA）was proposed. The gram-angle field (GAF) was used to convert the one-dimensional contact voltage signal into a two-dimensional image, and the feature recognition was carried out by the parallelizing convolutional neural network (PCNN). In addition, the one-dimensional timing signal is captured by the gated recurrent unit (GRU). The features of the one-dimensional time-series signal and the two-dimensional image are fused to make up for their respective limitations. In view of the parameters in the model, such as the learning rate that is difficult to determine, the number of neurons in the network layer of the gated recurrent unit, and the improved black-winged kite algorithm is integrated to optimize the parameters to make the model more reasonable. Finally, the multi-head self-attention mechanism was fused to improve the accuracy of the model. The proposed model and other three models were tested on three sets of pantograph-net arc models with different experimental conditions, and it was verified that the proposed model had strong robustness and high accuracy.

Abstract:In order to eliminate the disturbance noise in the process of gas coal rupture signal acquisition, a quantum optimization noise reduction model for gas coal rupture signal based on Improved quantum swarm algorithm (IQPSO) optimized variational mode decomposition (VMD) was proposed. In view of the fact that VMD is limited by the number of decompositions and the selection of penalty parameters, which affects the noise reduction effect, the IQPSO algorithm is used to optimize the optimization process of VMD parameters, and the decision weight coefficient and adaptive control factor are introduced into the QPSO algorithm to improve the particle decision adaptability and parameter search ability of the algorithm. The VMD algorithm with parameter optimization is used to decompose the rupture signal of gas-containing coal, the effective correlation coefficient of each signal component is calculated to identify the critical point of noise, and the wavelet transform is used to process the high-frequency noise and reconstruct the remaining components to obtain the denoised gas-containing coal rupture signal. The noise reduction model is compared with the EMD, VMD, PSO-VMD, SSA-VMD, GWO-VMD models through the simulation signal and field measured signal. The experimental results show that the signal-to-noise ratio of the proposed model is increased by more than 20%, the root mean square error is reduced to less than 0.03, and the energy proportion is more than 90%, which is better than other noise reduction models, and the adaptability and decomposition efficiency are strong, which can effectively retain the local characteristics of the signal and have a better noise reduction effect on complex signals in the field.

Abstract:In the evolution of contemporary communication systems, user experience plays an increasingly vital role, making quality of experience (QoE) a widely utilized metric that intuitively reflects the end-users’ perception of wireless services. To addresses the access and allocation problems within Het-CRN in smart home environments with multi-service muli-channel, a QoE-driven wireless resource allocation scheme is proposed. This scheme combines the improved simple additive weighting (SAW) and analytic hierarchy process (AHP) method to comprehensively evaluate the user preferences, service requirements, and channel parameters that affect user experience to obtain the objective and subjective weights of different services and further calculate the comprehensive weighs. In this scheme, a Markov model is established to describe the Het-CRN system state based on queuing theory. The model can effectively analyze the behavior under different user loads. Thus, the performance of different access and allocation algorithms can be evaluated by using the proposed scheme. Numerical results show that the proposed comprehensive weighting method significantly improves the user satisfaction of different services and significantly improves the quality of user experience compared to the SAW and AHP methods. By analyzing the performance results in conjunction with the relative standard deviation, it is further demonstrated that the comprehensive weighting method exhibits higher precision on key performance indicators such as throughput, delay, and rejection rate, and more accurately meets the actual user needs.

Abstract:The temperature control system of high and low temperature test chamber has nonlinear and time-delay. The traditional PID control has high overshoot and long adjustment time, but the effect of fuzzy PID control is affected by the formulation of quantization factor and scale factor. In order to improve the response speed and stability of the temperature control system of the test chamber, a method of temperature control of the high and low temperature test chamber based on DBO algorithm was proposed to optimize the fuzzy PID quantization factor and scale factor. Firstly, the transfer function of the heating model of the high and low temperature test chamber was established, and the traditional PID, fuzzy PID, PSO optimized fuzzy PID and DBO optimized fuzzy PID models were built in MATLAB/Simulink for simulation. In addition, the PLC, touch screen and temperature control box were used to build experimental devices to carry out actual temperature control experiments. The simulation results show that the overshoot of DBO optimized fuzzy PID is reduced by 1.02% and the adjustment time is reduced by 106 s compared with that of PSO optimized fuzzy PID. The experimental results show that the overshoot of the fuzzy PID optimized by DBO is reduced by 1.1% and the adjustment time is reduced by 120 s compared with that of the fuzzy PID optimized by PSO, which verifies that the DBO algorithm has a better effect than PSO in optimizing the quantization factor and scale factor of fuzzy PID. The temperature control effect of the optimal quantization factor and scale factor optimized by DBO at different temperatures is tested, which indicates the feasibility of optimizing the fuzzy PID control scheme by DBO algorithm.

Abstract:To address the problem of poor bearing fault diagnosis due to the large amount of deep network feature parameters and the unbalanced number of fault category samples, this paper proposes a lightweight bearing fault diagnosis method under unbalanced data. Firstly, the one-dimensional vibration signals collected by the sensors are reconstructed into a two-dimensional grey scale map as model input.Secondly, an asymmetric multi-scale feature extraction module is designed to extract features from the input signal using convolution and null convolution at different scales, and a part of the features are mapped to the original space for removing noise and restoring the original data structure. Next, the extracted rich feature information is fed to the channel position bi-weighting module to bi-directionally weight the key channel and key position features using inverse channel convolution and local averaging. Then, a depthwise separable convolution (DSC) dense residual structure is designed to increase the feature fusion of each layer of the network while keeping the network lightweight and optimize the backpropagation performance through shortcut paths. Finally, the focal loss function is used to adjust the learning process of the model according to the importance of different fault categories, thus better adapting to the unbalanced data distribution.