control of systems with operating condition-dependent dynamics, including control moment gyroscopes, often requires operating condition-dependent controllers to achieve high control performance. The aim of this paper ...
详细信息
Isolated renewable power to ammonia (IRePtA) has been recognized as a promising way to decarbonize the chemical industry. Optimal sizing of the renewable power system is significant to improve the techno-economic of I...
详细信息
Visual-based object detection has become a crucial component in the realm of autonomous vehicles. However, conducting reliable testing for such systems remains unresolved. In this paper, we advocate for the applicatio...
详细信息
ISBN:
(数字)9798350348811
ISBN:
(纸本)9798350348828
Visual-based object detection has become a crucial component in the realm of autonomous vehicles. However, conducting reliable testing for such systems remains unresolved. In this paper, we advocate for the application of causal inference to investigate the pivotal environmental factors influencing detection accuracy. Through the integration of diffusion models, we address the specialized conditional generalization of hazardous testing images. Our approach involves the construction of observational data to attribute key factors and fine-tune the diffusion model. Additionally, we introduce an optimal prompt words search method that strikes a balance between test coverage and level of challenge. Subsequently, leveraging these optimal prompts, we propose a cost-effective testing image generation through both "Text2Scene" and "Image2Scene" fashions. The experimental results indicate that, on the generalized dataset, the performance of object detection algorithms is the poorest, with the average detection accuracy decreasing from 0.81 to 0.285. Moreover, retraining object detection models on our generalized critical test cases can ultimately enhance algorithm performance, achieving a median accuracy improvement of up to 8.13%. Overall, our research proposes a novel approach to generalize test cases, thereby contributing to the advancement and deployment of safer autonomous vehicles.
Spatial frequency (SF) is a characteristic of an image that could dissociate course and fine shape information. Physiological and psychophysical studies widely investigated the role of various SF contents in image pro...
详细信息
Spatial frequency (SF) is a characteristic of an image that could dissociate course and fine shape information. Physiological and psychophysical studies widely investigated the role of various SF contents in image processing. Inspired by the primate brain structure, deep neural networks improved various computer vision tasks such as image classification. Physiological studies show that low SF (LSF) contents of an image could be processed faster to provide feedback to facilitate object recognition. However, this knowledge has not been considered in designing neural network structures. This study introduces SFNet, a new neural network structure that employs an LSF-based feedback mechanism. SFNet is a two-stream structure where one stream is used for LSF processing to provide feedback for image classification. The other stream combines the LSF-based feedback and the HSF processing to form the final decision. The role of the proposed LSF-based feedback in image classification is investigated utilizing the CIFAR100 dataset. The results show that SFNet improves the performance in the presence of SF filtering compared to the equivalent structures.
The capacitated location-routing problem involves determining the depots from a set of candidate capacitated depot locations and finding the required routes from the selected depots to serve a set of customers whereas...
详细信息
Aiming at the truck scheduling problem in the open-pit mine scenario, a truck scheduling model based on real-time ore blending is established, and an adaptive evolution algorithm for truck scheduling based on DCNSGA-I...
Aiming at the truck scheduling problem in the open-pit mine scenario, a truck scheduling model based on real-time ore blending is established, and an adaptive evolution algorithm for truck scheduling based on DCNSGA-III is proposed. In the established scheduling model, the real-time grade variance of the crushing plant is minimized as one of the optimization objectives, and the Q-learning algorithm is introduced to adaptively select one of the most effective operators during the search process. Experiments show that the proposed method can effectively control the grade fluctuation of the ore flow and better scheduling schemes are obtained in comparison with algorithms equipped with the traditional search operator selection methods.
To facilitate responsive and cost-effective computing resource scheduling and service delivery over edge-assisted mobile networks, this paper investigates a novel two-stage double auction methodology via utilizing an ...
详细信息
Power to hydrogen is a promising method to consume surplus renewable energy existing in the world. However, due to bottlenecks in hydrogen storage and transportation, converting hydrogen into ammonia which is a kind o...
详细信息
The paper presents the schemes for manufacturing a microoptoelectromechanical (MOEM) accelerometer for measuring small accelerations. MOEM-accelerometer includes three subsystems: mechanical, optical and electronic. T...
详细信息
ISBN:
(纸本)9781665476294
The paper presents the schemes for manufacturing a microoptoelectromechanical (MOEM) accelerometer for measuring small accelerations. MOEM-accelerometer includes three subsystems: mechanical, optical and electronic. The mechanical subsystem includes an inertial body, mounted on a spring suspension in a case. The optical subsystem includes a laser, moving and fixed waveguides. The electrical subsystem includes a photodiode and electronic components. To measure microdisplacements that correspond to measured microaccelerations, an optical system based on the optical tunnel effect was used. The work considers the schemes of the optical transducer for linear and angular displacements of the mechanical subsystem. The moving waveguide together with the inertial body are combined into the mechanical sensing element that diverges in the case of acceleration. A technological process for manufacturing a MOEM accelerometer based on the “silicon on insulator” technology with additional layers of nitride and silicon oxide for optical functional elements is presented. Depending on the character of the movement of the sensing element, the functional schemes of the MOEM-accelerometer were developed with various changeable parameters: optical coupling length, gap, overlapping area between the moving and fixed waveguides. The article analyzes the advantages and drawbacks of the proposed schemes of accelerometers from the perspective of their manufacturing feasibility and the predicted accuracy. The highest sensitivity $(6.25 \times10^{6} \text{m} ^{\text{-1}}$) belongs to schemes with changeable gap between the waveguides. The dynamic displacement ranges of them is ± 80 nm. Lower sensitivity $(1.25 \times10^{6} \text{m} ^{\text{-1}}$) belongs to schemes with changeable overlapping area. The dynamic displacement range may reach ± 300 nm. Schemes with changeable optical coupling length possess the highest dynamic range which directly depends on a chosen optical coupling length and amoun
In this paper, a deep residual network based on convolutional block attention module (CBAM) is proposed, which is utilized for feature extraction of partially occluded face expression data. The proposed method overcom...
In this paper, a deep residual network based on convolutional block attention module (CBAM) is proposed, which is utilized for feature extraction of partially occluded face expression data. The proposed method overcomes the problem of localized occlusion face feature extraction by focusing on the regions and channels containing important information in the occluded face data through CBAM. Multi-task cascaded convolutional networks (MTCNN) are firstly utilized to localize the key regions of face emotion, and then deep emotion features are extracted by CBAM-ResNet network. The final emotion labels are generated. The effectiveness of this paper's method is verified on the RAF-DB dataset and the occluded CK+ dataset. The experimental accuracy in the RAF-DB dataset is 76.3%, which is 3.74% and 1.64% higher than the accuracy produced by the method of RGBT, and the WLS-RF, respectively. Application experiments are carried out in the real teaching scenario, which verifies the applicability of the algorithm in the real teaching scene.
暂无评论