Abstract:Bone is the second most commonly transplanted tissue in the world, with at least four million surgical procedures using bone grafts and bone substitutes every year. However, the limitations of traditional treatments have affected current treatment options, and the clinical demand for bone grafts has continued to increase due to the high incidence of trauma, cancer, infection, and arthritis. Autografts and allografts are commonly used in the clinical treatment of bone defects, but chronic inflammation, disease transmission, and immune rejection have hindered their development. In addition, metal-based material scaffolds are the most widely used implants, while there are also problems such as stress shield, infection, and inflammation, leading researchers to find new material scaffolds to replace them. Therefore, developing bioactive three-dimensional (3D) scaffolds that can be adaptively expanded and filled to promote bone regeneration has become a key area of focus in bone tissue engineering (BTE). In recent years, manufacturing methods including 4D printing of shape memory materials have been used to create new methods to replace traditional bone grafts. This paper mainly reviews the classification of traditional polymer materials and new shape memory polymer materials, the main manufacturing methods, mechanical properties, biocompatibility, and the latest applications of polymer scaffolds in BTE. Furthermore, the importance, current challenges, and future development directions of 4D printing technology in BTE are summarized.

Abstract:Titanium alloy is employed in aerospace, aviation, navigation, and other fields because of its excellent comprehensive properties, such as high specific strength, strong heat resistance, excellent corrosion resistance, and good low temperature performance. However the high cost of titanium alloy seriously restricts its large-scale application. Based on Ti-6Al-4V alloy which has the best comprehensive properties, several Ti-Al-V-Fe alloys with different contents of V and Fe were designed under the guideline of using cheap alloy elements instead of expensive alloy elements. With the help of the stability factor K_β, the stability of β phase and the effect of β stabilizing elements were analyzed. The calculation models for critical cell coefficients ofβstabilizing elements of quenched alloys and annealed alloys were established by theoretical derivation. On this basis, the calculation methods for volume fraction of α phase and β phase were established, and the theoretical calculation of the volume fraction of α phase and β phase in Ti-Al-V-Fe alloy was realized. Taking Ti-6Al-3V-1Fe alloy as an example, the theoretical volume fraction of α phase and β phase in the alloy was calculated. Then, Ti-6Al-3V-1Fe alloy ingots were prepared by vacuum non-consumable arc melting furnace. Samples were taken and heat treated. Finally, the volume fraction of α phase and β phase in the alloy was determined by X-ray diffraction. It was found that the measured results were in good agreement with the theoretical calculation results, indicating that the established model is feasible to predict the volume fraction of α phase and β phase.

Abstract:Network architecture is an important factor affecting the performance of convolutional neural networks. Due to the low efficiency of the traditional manual design of network architecture, the method of automatically designing network architecture through algorithms has attracted more and more attention. Although the approach of differentiable architecture search (DARTS) has the capacity of designing networks automatically and efficiently, there are still problems owing to its super network construction and derivation strategy. An improved algorithm was proposed to overcome these shortcomings. First, the coupling problem caused by sharing architecture parameters in the super network was disclosed by quantifying the changes in the number of the skip candidate operations during the algorithm search process. Next, aiming at the coupling problem of the super network, a super network with multi-level cells was designed to avoid the mutual influence of cells at different levels. Then, in view of the “gap” between the super network and the derived architecture, the entropy of architecture parameters was introduced as the loss term of the objective function to inspire the training of the super network. Finally, architecture search experiments were conducted on CIFAR-10 dataset, and architecture evaluation experiments were conducted on CIFAR-10 and ImageNet respectively. Experimental results on CIFAR-10 show that the proposed algorithm removed the coupling problem between cells at different levels and improved the performance of the automatically designed architecture, which achieved classification error rate of only 2.69%. The architecture had the classification error rate of 25.9% on ImageNet, which proved its transferability.

Abstract:Crowd evacuation guidance systems are of great significance for protecting lives and reducing personal and property losses during disasters in buildings. Existing crowd evacuation guidance systems require the manual design of models and input parameters, incurring significant workloads and potential errors. An end-to-end intelligent evacuation guidance method based on deep reinforcement learning was proposed, and an interactive simulation environment based on the social force model was designed. The agent could automatically learn a scene model and explore the path planning strategy by interacting with simulation environment and through trial and error with only scene images as input, and then directly output dynamic signage information, thus achieving the crowd evacuation guidance efficiently. Aiming to solve the “dimension disaster” phenomenon of deep Q network (DQN) algorithm caused by high dimension action space and complex network structure in crowd evacuation, a combined action-space DQN algorithm was proposed. The algorithm grouped the output layer nodes of the Q network according to action dimensions, significantly reduced the network complexity, and improved the practicality of the system in complex scenes with multiple guidance signs. Experiments in different simulation scenes demonstrate that the proposed method is superior to the static guidance method in evacuation time and on par with the manually designed model method. It shows that the proposed method can effectively guide the crowd, improve the evacuation efficiency, and reduce the workload and artificial errors of manually designed models.

Abstract:The existing human target device-free indoor localization algorithm based on received signal strength (RSS) is difficult to give consideration to artificial workload, time consumption, and positioning accuracy under the circumstance of environment changes. In view of this problem, this paper proposes a device-free localization algorithm based on transfer clustering and fusion variational auto-encoder (FusVAE) in changing indoor environment. After environment changes, a small amount of RSS samples without labels were collected. Then, a semi-supervised fuzzy C-means clustering based on metric learning (SFCMML) was proposed to accurately cluster and label the samples, and the original model was retrained, where only a small amount of artificial work and process time was required to achieve a high localization accuracy for the original model in the new environment. In addition, aiming at the problem that the RSS samples collected in the new environment were in small quantity, the FusVAE was constructed based on coordinate fusion to generate RSS samples in the new environment for data enhancement, which could enrich the quantity and quality of the RSS samples, improve the generalization ability of the model, and enhance the positioning accuracy. Experimental results show that under the circumstance of environment changes, the average positioning accuracy of the proposed algorithm reached 88.6%. Compared with the algorithms of the same type in the same field, the proposed algorithm had higher positioning accuracy and better environmental adaptability, which is more applicable to device-free indoor localization in changing environment.

Abstract:Iris recognition is an effective and widely used biotechnology, which has higher security performance than face recognition and fingerprint recognition. However, the overall performance of the recognition system is largely affected by the iris segmentation accuracy. In order to effectively improve the iris segmentation accuracy, based on the analysis of the physiological structure of iris, the literature in the relevant fields at home and abroad was reviewed, and the advantages and disadvantages of various algorithms were analyzed. A new accurate iris segmentation algorithm was proposed, which overcomes the hypothesis of concentric circles of traditional segmentation algorithms. Drawing on the idea of completed local binary patterns (CLBP) algorithm and fusing the grayscale information and structural information of the image, the shape-sensitive detection operator was proposed to effectively eliminate the two major factors that affect segmentation accuracy, i.e., the interference of eyelid and eyelashes. In addition, a segmentation process was proposed, which is divided into two parts: coarse iris segmentation and precise segmentation. Coarse segmentation includes outer contour and pupil rejection, and precise segmentation includes eyelid and eyelash rejection. Finally, a series of comparative tests were conducted to investigate accuracy and calculation efficiency on the iris datasets CASIA-IrisV3-Interval and CASIA-IrisV1 published by the Institute of Automation, Chinese Academy of Sciences. After using the proposed segmentation algorithm, the accuracy on the OSIRIS Version 4.1 iris recognition system reached 97.14% and 98.28% respectively, and the running time was significantly reduced, up to 0.699 s and 0.758 s respectively.

Abstract:In order to improve the hardness and wear resistance of TiN films, single-layer TiN-Cu composite films and five groups of TiCu/TiN-Cu nano-composite multilayer films with modulation period (Λ) of 5.9~62.1 nm were prepared on cemented carbide substrates by multi-arc ion plating. The microstructures and mechanical properties of the films were characterized by scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS), X-ray diffraction (XRD), nano-indentation tester, scratch tester, friction-wear tester, and other machines. The influences of modulation period on the structures and mechanical properties of the nano-composite multilayer films were studied. Results show that compared with the single-layer TiN-Cu composite film, the TiCu/TiN-Cu nano-composite multilayer film effectively inhibited the growth of the grains, the layer was well formed, and the film was uniform and dense. The TiN grains in the film grew in face-centered cubic structure along (111) direction. With the decrease of the modulation period, the crystallinity of the film reduced, while the hardness of the film increased first and then decreased. When the modulation period was 13.7 nm, the optimal comprehensive properties were achieved, where the hardness and H³/E² value of the film reached 42.6 GPa and 0.689, the friction coefficient reached the minimum value of 0.17, and the adhesion was 49.2 N, closing to the maximum value of 53.1 N, indicating that the film possesses ideal hardness and wear resistance. Through multi-arc ion plating, the TiCu/TiN-Cu nano-composite multilayer films were prepared. By adjusting the modulation period, the mechanical properties of the films were effectively improved and the application domains were expanded.

Abstract:To resist the wormhole attacks in wireless sensor networks (WSNs) and improve network performance, a wormhole attack detection strategy integrating node creditworthiness and path hops (WADS-NC&PH) was proposed. Based on the abnormal situations of the neighbor number of the nodes under wormhole attacks, the suspicious nodes whose neighbor number exceeded the threshold were screened, and the nodes in their exclusive neighbor set were allowed to communicate with each other. The number of path hops was recorded, and the path whose number of hops exceeded the wormhole threshold was marked as the path to be tested. The Bayesian trust model was utilized to calculate the direct trust value of the intermediate node on the path to be tested, and combined with the trust factors such as the number of neighbors, processing delay, node energy, and packet forwarding rate, the indirect trust value of the node was judged, thereby obtaining the comprehensive trust value of the node. By integrating the number of path hops with the comprehensive trust of intermediate nodes, the path trust evaluation of the path to be tested was calculated. Based on the path characteristics of the nodes attacked by wormholes, the trust threshold was set reasonably, and WADS-NC&PH was proposed to detect wormhole attacks in WSNs. Simulation results show that WADS-NC&PH had a significant effect on the detection of wormhole attacks. Even in the face of highly attacked networks, this strategy could still effectively detect wormhole attacks and remove false links, improving the security and reliability of WSNs.

Abstract:In the process of image restoration, in order to solve the problems of incomplete separation of color and texture optical properties and single dictionary design in image inpainting of sparse representation, which leads to the structural incoherence and blurring effect of mural image inpainting results, a mural inpainting method based on robust principal component analysis (RPCA) decomposition of block nuclear norm and entropy weighted clustering sparse representation was proposed. First, the proposed RPCA image decomposition algorithm based on block nuclear norm was used to decompose the mural image into a structure layer and a texture layer, and the block nuclear norm was used to perform texture correction operations, which could overcome the problem of incomplete separation of structure and texture by RPCA. Then, the entropy weighted k-means method was proposed to cluster the structure layer image and construct sparse sub-cluster dictionaries, and the reconstruction of the structure layer image was completed by sparse value decomposition and split Bregman iterative optimization method. Finally, the image of texture layer was repaired by using the bicubic interpolation algorithm, and the repaired structure layer and texture layer were fused to complete the repair of the damaged murals. Experimental results of digital restoration on the real Dunhuang murals show that the proposed method could effectively protect the edges, textures, and other important features in the mural image. In terms of visual quality and quantitative evaluation such as peak signal-to-noise ratio (PSNR), the proposed method had better performance than the comparison algorithms, and the restoration efficiency was higher.

Abstract:In view of the problems of poor clarity, low contrast, and dark color of traditional image defogging algorithms, an improved defogging algorithm was proposed for processing images with sea fog. First, the dark channel image and the minimum image of foggy image were obtained. The foggy image was converted into the HSV color space to calculate the color decay rates of each pixel, which were then sorted in descending order, and the minimum value of the first 10% values was taken as the threshold of the bright and dark parts. On this basis, the dark part (I_{HSV_dark}) of the foggy image was calculated. The variational function was introduced to determine whether the pixel in the image was from the bright area, and the variational dark part (I_{VAM_dark}) based on the variational function was obtained. Then, the two dark parts were combined to calculate the dark part image (I_dark), which was used to estimate the ambient light value of the dark area. The pixel values were sorted in descending order, and the average value of the degraded image pixels corresponding to the top 1‰ pixel values was selected as the value of A_dark. Next, a method for removing textures was proposed based on the multi-level weight relative total variation model. The minimum image was filtered as the rough estimate of the transmittance image, which was adjusted by the transmittance function to weaken the defogging of the bright image and enhance the defogging of the dark image. Finally, a minimum variance median guide filter algorithm was proposed to optimize the adjusted transmittance, and the clear image was acquired based on the foggy image degradation model. Experimental results show that the information entropy, average gradient, contrast, and fog aware density evaluator (FADE) of the restored image obtained by the proposed algorithm were significantly improved compared with traditional algorithms.

Abstract:Due to the long-run accumulation and the complex and changeable external environment, the performance level of a system is uncertain. Accurate quantitative description may cause the interval estimation of the system performance level to be too narrow, which already cannot fulfill the engineering calculation in practice. Meanwhile, failure-correlation is another important factor affecting the reliability of the system. Ignoring failure-correlation can lead to the overestimation of the system reliability, which affects the distribution of maintenance force and the storage of spare components, and is not conducive to the long-term stable operation of the system. Therefore, taking the parallel repairable system of main and auxiliary units composed of multi-state components as research object, the failure transfer rate, repair transfer rate, and state performance level of the components were regarded as fuzzy numbers, and the failure-correlation of the system components was investigated. The reliability analysis of two typical kinds of main and auxiliary parallel systems were carried out. The model for fuzzy multi-state main and auxiliary repairable parallel system with failure-correlation and the fuzzy state transfer differential equations were established, and the fuzzy state probability of the system was obtained. By using the α-level truncation set and the Zadeh-expansion principle, the truncated set interval of the fuzzy state probability and the steady state availability of the system was determined. The steady measures of the system were obtained, and the influence of the fuzzy degree of component parameters on the steady measures was given by numeral application, which provides reference for the study on the failure-correlation of multi-state systems under complex conditions, and is conducive to the optimization of maintenance support in engineering design.

Abstract:A method to improve the localization accuracy of intelligent vehicles was proposed by using map representation model from three-dimensional (3D) point cloud polarization. The point cloud polarization image was used as the node in this model, and the global position representation of the node could be realized through high-precision Global Positioning System (GPS) and Euler angle. The 2D and 3D features of the point cloud were then extracted from the polarization image to realize the multi-scale feature representation of the node, and the numerical description and virtual reconstruction of the road scene were realized through a series of polarization nodes. During the localization process, the 3D laser point cloud was in real-time acquired for polarization representation, and multi-scale feature matching was carried out with map nodes to realize the map localization of the intelligent vehicles. Specifically, map nodes were first filtered through GPS matching or topological localization based on the stability condition of the GPS signal of the intelligent vehicle, and a localization candidate set was obtained for the coarse localization. Then, the nearest map node was detected by matching the 2D point cloud features of the polarization image from the candidate set to complete the node localization. Finally, for the metric localization, the position of the intelligent vehicle was calculated by using the 3D point cloud feature matching results and the global position of the nearest map node. The experiment was carried out in two typical scenarios. The node localization accuracy rate was 98.7%, the average localization error was 21.4 cm, and the maximum localization error was 42.9 cm. Results show that the proposed algorithm has the advantages of high positioning accuracy, strong robustness, low cost, and simple calculation process.

Abstract:The semantic expression of ship behaviors is the basis to realize the intelligent recognition and knowledge reasoning of the water traffic situation. In order to achieve the semantic expression and abstraction of the spatial-temporal motion features of ships and improve the semantic conversion method of ship trajectories, a cognitive computing model of ship behavior semantics was proposed. By integrating the spatial-temporal data of ship trajectory and navigation environment information, based on the ship trajectory unit, the spatial-temporal behavior of the ship was abstracted to atomic behavior, topological behavior, and traffic behavior according to the theory of space topology, considering the motion states, spatial topological characteristics, and behavioral semantic characteristics. The conceptual modeling, semantic description, and formalization expression were carried out at different levels from spatial-temporal trajectory to semantic behavior of the ship. Finally, different types of navigation behaviors of ships in port waters were verified based on the semantic model. Results show that the model could be used to model and express ship behaviors with different motion features and spatial topological features in port waters, and five typical traffic behaviors were extracted, which indicates that the model is reasonable and applicable to the cognitive modeling of ship behaviors. The model can provide theoretical and methodological basis for semantic recognition and knowledge calculation of ship behaviors to realize the semantic calculation and cognitive reasoning of advanced ship behaviors.

Abstract:To establish a rolling force model with reliable prediction accuracy, a cosine velocity field was proposed according to the characteristics of metal flow in rolling process, and corresponding energy analysis was carried out based on the proposed velocity field. The proposed velocity field could strictly satisfy the volume constant condition, the boundary condition, and the geometric equation, which indicates that the velocity field can satisfy the kinematically admissible condition well. In the modeling process, the inner product and addition method of vector component was adopted to derive the internal deformation power during rolling, and the integration problem of the nonlinear specific plastic work rate of Mises criterion was solved. In addition, the mathematical expressions of friction power and shear power were derived based on the proposed velocity field. The analytical model of force-energy parameters in rolling process was obtained in terms of the variational principle of rigid-plastic material. The prediction accuracy of the rolling force model was verified by using the measured data from a domestic factory. Comparison results show that the deviations between the predicted rolling force and the measured value were within 7.55%, indicating a high accuracy. The established model was compared with the classic Sims model and Tselikov model, and a good superiority was found. In order to investigate the parameter variation during the rolling process of thick plates, the influences of reduction, shape factor, friction factor, radius-thickness ratio, and roll radius on the stress state coefficient and the position of the neutral point were analyzed.

Abstract:In the lossless compression schemes of the H.26X series video coding standards, the residuals obtained by intra-frame prediction still retain strong spatial correlation and directly participating in entropy coding will lead to the decrease in coding efficiency. Unlike natural images, intra-frame prediction residuals contain rich edge features. To utilize the special spatial correlation of residuals, further reducing its spatial redundancy and improving the intra-frame coding efficiency, a universal lossless intra-frame coding algorithm based on residual median edge detection was proposed. The algorithm first performed edge detection on the intra-frame prediction residuals step by step. By analyzing the numerical features of the adjacent residuals, the median edge detection algorithm was applied to obtain new prediction value. Then, new prediction residual was obtained by comparing prediction value and original residual value. Finally, according to whether the energy of the current coding unit was reduced, it could quickly determine whether to use the new residuals for entropy coding, so as to ensure that the coding of the new prediction residuals could improve the compression ratio. Experimental results show that the coding units processed by this algorithm had lower spatial redundancy and residual energy, thereby reducing the bit-rate after entropy coding. Applying the proposed algorithm in H.265 and the latest H.266 standards, the average energy of residuals decreased by 67.9%, and the average bit-rate reduced by 7.04% and 5.98% respectively, while the codec time changed slightly, with significant practical value.

Abstract:An experience replay DQN (ERDQN) data transmission scheduling algorithm was proposed for network transmission application scenarios that require low latency and high reliability in the fields of in-vehicle networks, telemedicine, and industrial control. The main purpose and task of this algorithm was to reduce network delay and improve the stability of network transmission. The ERDQN algorithm optimized the queuing strategy at the sender side based on the deadline-aware transport protocol (DTP), and gave fully consideration to the priority of data blocks and Deadline, taking them as important factors for calculating the order of entering the waiting queue, which avoided the problem of losing Deadline of data blocks and reduced the queuing delay of network transmission. Meanwhile, in the congestion control, the current network transmission state was used as the feature vector to predict the parameters of the next network transmission state and give different reward factors for evaluation. Through the iterative learning of the ERDQN network, the optimal parameters were automatically adjusted to fit the current network transmission. The average transmission rate was high and stable during the subsequent network link transmission process, which alleviated the problems of network congestion and transmission stability and reduced the network transmission delay. Experimental results show that the average queuing delay and transmission delay of the ERDQN algorithm were much lower than those of the traditional congestion control algorithm (Reno algorithm), and much higher than the traditional congestion control algorithm in terms of quality of experience (QoE), which could minimize the network transmission rate fluctuation, reduce the packet loss rate, and provide stable and reliable transmission.

Abstract:To solve the problem of the arrival-departure track utilization planning for high-speed railway stations, an optimization model of train platform utilization based on time segment and multi-objective optimization and an improved fast non-dominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ) were proposed. First, considering the coupling relationship between the arrival-departure tracks and the routes in the throat area, the study period was divided according to the arrival and departure distribution characteristics of the trains. Taking the utilization balance of the train platform, strong planning robustness, and passenger service quality as the optimization objectives, an optimization model of train platform utilization was established by integer programming method. Then, an elite strategy with feasible solutions was designed by introducing constraint domination principle. At last, the actual operation data of a station on the Beijing-Shanghai high-speed railway was taken as an example for simulation verification. Calculation results show that the occupancy cost of the train platform was reduced by 16.98% compared with the original scheme in the normal peak period, which significantly improves the quality of passenger service. During the rush hour, the variance of the occupation time of the platform and the total conflict coefficient were reduced by 48.57% and 29.81% compared with the original scheme, which significantly improves the equipment utilization rate and the robustness of the arrival-departure track utilization plan. In comparison with the overall optimization method, the time segment optimization method could be more targeted in reducing the function value of the optimization objectives. Therefore, considering the influence of train arrival and departure distribution factors on the train platform utilization, it can effectively improve the optimization factors and provide support for making station operation plans during different busy periods.

Abstract:The optimal external magnetic field parameters for stable discharge of a novel HiPIMS and obtaining high-quality film were explored, and the modulation of the external magnetic field on the discharge and deposition characteristics of electro-magnetic fields synergistically enhancing high power impulse magnetron sputtering ((E-MF) HiPIMS) was studied. The effects of coil current of external magnetic field on the discharge of HiPIMS vanadium target as well as the microstructure and properties of vanadium films were investigated. The discharge behavior of HiPIMS was monitored, and the influences of external magnetic field on the phase structure, surface morphology, cross-section morphology, friction and wear resistance, and corrosion resistance of vanadium films were studied by XRD, SEM, AFM, friction and wear tester, and electrochemical corrosion method. Results show that with the increase of the coil current, the substrate ion current increased gradually. When the coil current was 6 A, the maximum substrate ion current density reached 209.2 mA/cm². The phase structure of vanadium film was only V (111), but the intensity of its diffraction peak increased with the increase of coil current. The surface of vanadium film exhibited typical round-pit shape. The surface roughness decreased first and then increased, and the minimum was only 10 nm. At low coil current (<4 A), the vanadium film showed a compact and fine crystal growth structure. With the increase of coil current, the deposition rate of the film increased gradually. When the coil current was 4 A, the friction coefficient of the vanadium film was the smallest, the wear resistance was the best, and the vanadium film samples had the best corrosion resistance.

Abstract:The cooperative control problem of heterogeneous multi-agent systems with uncertain denial of service (DoS) attacks was studied. The openness of network environment will lead to the increasing complexity of network attacks, among which the research on a class of uncertain network attacks is of significance. Due to the difficulty of modal acquisition under uncertain attacks, the mismatch between the controller mode and the system mode will occur. First, in normal circumstance, all agents were time synchronized and they communicated with each other over a fixed sampling period. When an attack occurred, the hold input mechanism was adopted and the attack duration was assumed to be bounded. The complex dynamic system model was constructed by using Markovian switching system method. Next, the original high-dimensional system was transformed into two low-dimensional closed-loop error systems by using the decoupling technology, and sufficient conditions to ensure the output consistency of the heterogeneous multi-agent system were obtained based on the Lyapunov stability theory. Then, the gain of the controller was obtained by solving a series of matrix inequalities through related matrix transform methods. Finally, the effectiveness of the proposed method was verified by the simulation of the mobile stage robot system. Compared with the existing results, the attack probability considered in this study could be uncertain or even completely unknown. The designed asynchronous controller has better compatibility as it covers the common synchronous controller as well as the model independent controller as special cases.

Abstract:To reduce the misoperation of submariners in the complex deep-sea environment and improve the reliability of human factors, the optimization of the cabin space of manned submersibles is a key method. A series of optimization research and evaluation have been carried out for the layout of small and confined space by constructing simulators. Taking Jiaolong as an example, a simulation model for the operating attitudes of submariners was built based on multi-objective game. In the virtual cabin environment, the multiple targets, tasks, and factors such as the joint attitude, comfort, and balance of submariners were regarded as a game. According to the cooperative game framework model of multi-agent system (MAS), a behavior mode hierarchical action meant to complete multiple differential games. The game algorithm adopted the method of gradient descent and searched the sub-behavior of the previous step size as the benchmark, so as to quickly obtain the optimal working state of the virtual submariner. On the basis of negotiation, the Pareto optimal solution set of the optimal equilibrium state satisfying multiple interests could be obtained by accumulating the results of the optimal equilibrium state through the differential game of multiple behavior modes. Simulation results show that the submariners could modify their behaviors under the high-intensity and long-time working environment, exhibiting good applicability, which verifies the feasibility and effectiveness of the proposed technology. The series of dynamic data obtained by simulation provides a means to optimize the cabin layout of submersibles and improve the reliability of human factors.