Sleep staging has always been a hot topic in the field of sleep medicine, and it is the cornerstone of research on sleep problems. At present, sleep staging heavily relies on manual interpretation, which is a time-consuming and laborious task with subjective interpretation factors. In this paper, we propose an automatic sleep stage classification model based on the Bidirectional Recurrent Neural Network (BiRNN) with data bundling augmentation and label redirection for accurate sleep staging. Through extensive analysis, we discovered that the incorrect classification labels are primarily concentrated in the transition and nonrapid eye movement stage I (N1). Therefore, our model utilizes a sliding window input to enhance data bundling and an attention mechanism to improve feature enhancement after label redirection. This approach focuses on mining latent features during the N1 and transition periods, which can further improve the network model’s classification performance. We evaluated on multiple public datasets and achieved an overall accuracy rate of 87.3%, with the highest accuracy rate reaching 93.5%. Additionally, the network model’s macro F1 score reached 82.5%. Finally, we used the optimal network model to study the impact of different EEG channels on the accuracy of each sleep stage.

Medical three-dimensional (3D) positioning technology amalgamates traditional medical instrumentation with computational technology and medical imaging to facilitate the real-time visualization of anatomical lesions, surgical implements, and other relevant objects, thereby augmenting the physician’s diagnostic and therapeutic capabilities. Predominantly, optical positioning technologies are employed; however, they inherently suffer from limitations such as optical occlusion, precluding precise positioning. To surmount these challenges, the present manuscript introduces a phase-based 3D electromagnetic positioning technique predicated on spectral interpolation. Initially, four spectral lines exhibiting maximal signal amplitude are designated for the computation of a correction factor. Subsequently, the functional relationship between this correction factor and frequency offset is modeled via polynomial fitting, from which signal frequency and phase correction equations are derived. Experimental analyses executed within the ANSYS software (2021R2) environment substantiate the utility of the phase correction formula in calculating the phase disparity between the signals of the transmitting and receiving electromagnetic coils. This, in turn, enables the acquisition of the linear distance between these coils, which is instrumental in determining their spatial coordinates. The introduced 3D electromagnetic positioning methodology based on spectral interpolation effectively circumvents the issue of multiple solutions engendered by solving nonlinear equations inherent in traditional electromagnetic positioning systems. Simulation analyses corroborate that the proposed technique enhances phase detection accuracy by 1–2 orders of magnitude over conventional methods, achieving a positional accuracy within the effective measurement space of 3 mm—an improvement of at least 70% in comparison with established positioning paradigms.

Due to the influence of the environment on the water quality wireless sensor network, it is difficult to replace the node energy at any time, so it must make the most of the little energy available. This work provides a technique that combines cluster head selection, cluster structure creation, and data transmission into one optimal scheme. Firstly, we optimize the cluster head election probability threshold formula on the basis of LEACH and introduce overlap ratio in the competitive mechanism, to avoid excessive overlap between cluster heads. Subsequently, to alleviate the hot spot problem caused by multihop, the competitive mechanism of I-LEACH is optimized, which is based on the nonuniform competitive mechanism in the EEUC algorithm when electing the cluster head. Meanwhile, in the structural planning between cluster heads, the optimal path is searched based on the parallelism of genetic algorithm (variable path coding strategy); the combination of I-LEACH and EEUC-IGA, named Energy Balance Multihop Clustering Routing Protocol (BEBMCR), can avoid the emergence of hot paths and reduce running time. The simulation results show that BEBMCR still has a longer stable period and higher energy utilization rate under large-scale networks, and node energy consumption is more balanced.

In view of the problem of being easy to fall into local optimization when the gray wolf optimization algorithm handles complex optimization problems, this thesis starts from the two angles of perspectives of gray wolf search strategy and nonlinear control strategy to propose an improved gray wolf optimization algorithm based on beetle antennae search strategy (BIGWO). Firstly, it gives the leading wolves the ability of free exploration and communication by combining with the beetle antennae search strategy, which helps the improvement of the global search performance of the algorithm. Besides, it carries out the design of a nonlinear control strategy to balance the exploration and development of the algorithm. Then, the optimization test is carried out for 10 benchmark functions to verify the optimization ability of the algorithm. Finally, it makes the application of the algorithm into the actual optimization problem of vehicle planning task allocation with capacity constraints, so as to verify the ability of the algorithm to solve practical problems. According to the experimental results, it indicates that BIGWO algorithm has better optimization accuracy, stability and convergence speed, which is both suitable for the optimization of characteristic function and the practical optimization problems.

As technology continues to advance, the utilization of electrooculographic signals (EOG) in human-computer interaction and the biomedical field has progressively expanded. Notably, the processing and analysis of EOG signals are pivotal for achieving precise eye-tracking and visual control. Consequently, this paper introduces a methodology founded upon wavelet packet transformation and Generalized Learning Vector Quantization (GLVQ). This approach is tailored to the classification of saccadic and blink movements across four directions based on EOG signals. Initially, the method preprocesses EOG signals, employing endpoint detection techniques to extract pertinent segments associated with saccadic and blink movements. Subsequently, wavelet packet transformation is harnessed to analyze non-stationary and nonlinear characteristics, constructing a feature matrix used for classification through GLVQ. Experimental outcomes underscore the method's exemplary performance in addressing complex EOG signal classification tasks, achieving an accuracy rate exceeding 90%. In comparison to conventional approaches, the combination of wavelet packet transformation and GLVQ exhibits superior adaptive learning and classification capabilities, enabling enhanced adaptability to signal variations and complexities.