This thesis explores Latent Space Bayesian Optimization (LSBO) for the generation and optimization of de novo molecules and crystal materials. Our goal is to develop practical, sample-efficient de novo discovery algor...
详细信息
To efficiently plan the coverage path of UAVs during the visual inspection of building facades, the task is divided into two parts: viewpoint planning and path planning. First, a K-medoid clustering method is proposed...
详细信息
ISBN:
(数字)9798331521950
ISBN:
(纸本)9798331521967
To efficiently plan the coverage path of UAVs during the visual inspection of building facades, the task is divided into two parts: viewpoint planning and path planning. First, a K-medoid clustering method is proposed to generate the initial set of candidate viewpoints based on the point cloud model of the building facade. A greedy optimization algorithm is then employed to select the optimal subset of viewpoints, addressing the viewpoint planning problem. Second, an incomplete graph is constructed, and the objective function is defined for path planning, which is solved using the Gaussian Zenith-Ant Colony Optimization (GZ-ACO) algorithm. The path length and turning angles are considered optimization objectives. Simulation results demonstrate that the proposed viewpoint planning method reduces the number of viewpoints by 66.67% and 98.67%, respectively, compared to the iterative random sampling method and the displacement method. Furthermore, the GZ-ACO algorithm outperforms the traditional ant colony algorithm, achieving a 19.97% reduction in the objective function value and a 21.76% decrease in the average steering angle. The validity and feasibility of the proposed method are thus verified.
Digitalization, networking and intelligent management of Chinese language teaching resources are of great significance for improving teaching efficiency and promoting cultural exchange. However, in the face of massive...
详细信息
ISBN:
(数字)9798331536169
ISBN:
(纸本)9798331536176
Digitalization, networking and intelligent management of Chinese language teaching resources are of great significance for improving teaching efficiency and promoting cultural exchange. However, in the face of massive Chinese language teaching resources, the traditional classification method has been difficult to adapt to the characteristics of huge resources, diverse types and complex content. Because of its unique advantages in dealing with uncertainty and fuzziness, fuzzy clustering algorithm is applied to the design of this system to realize the intelligent and automatic classification of Chinese language teaching resources. The system architecture design includes four core parts: front-end user interface layer, back-end service layer, data processing layer and database storage layer. It adopts modular design, supports front-end separation mode and interacts through API. The functional modules of the system are divided into data preprocessing module, fuzzy clustering algorithm module, result display module, user management module and system maintenance module to ensure the maintainability, expansibility and efficiency of the system. In this study, the fuzzy C-means (FCM) algorithm is selected as the core algorithm, and the optimal clustering of data points and the center of each cluster are found by iteratively optimizing the objective function. The system test and evaluation show that FCM algorithm has achieved good clustering effect in the merging and classification of digital Chinese teaching resources, with contour coefficient of 0.75 and ARI index of 0.82, which shows high accuracy. The performance test results also show that the system has good response speed and stability, the throughput reaches 1050 requests per second, and the average utilization rate of CPU and memory is 57.3% and 68.9% respectively, which meets the requirements of high concurrency and big data processing. This system has high practicability and value in practical application, which ca
Transmission congestion is a major barrier between supplying cheap generation to load centers. Unified Power Flow Controller(UPFC) can help relive transmission congestion. For integrating a UPFC within the optimal pow...
详细信息
ISBN:
(数字)9798331504847
ISBN:
(纸本)9798331504854
Transmission congestion is a major barrier between supplying cheap generation to load centers. Unified Power Flow Controller(UPFC) can help relive transmission congestion. For integrating a UPFC within the optimal power flow, a linear programming based formulation can be used which is easily implementable. Moreover, determining the optimal location of the UPFC is vital for maximizing the ability to manage the line loading of the power system. In this study, a simple and efficient method has been proposed, namely the Discrete Teaching Learning Based Optimization(DTLBO) to find the optimal location of multiple UPFCs for the 39 Bus and 118 Bus System.
Non-Orthogonal Multiple Access (NOMA) is a nextgeneration communication technology that enables multiple users share the same wireless channels. To optimize the performance of NOMA, appropriate channel allocation for ...
详细信息
ISBN:
(数字)9798331506940
ISBN:
(纸本)9798331506957
Non-Orthogonal Multiple Access (NOMA) is a nextgeneration communication technology that enables multiple users share the same wireless channels. To optimize the performance of NOMA, appropriate channel allocation for each user is essential. Previous research has applied a fast optimization approach using the coherent Ising machine (CIM) to channel allocation in NOMA, multiplexing up to two users per channel, which can be formulated as a basic Ising Hamiltonian including up to second-order interactions. However, NOMA can support multiplexing more than two users per channel depending on channel conditions. This paper propose a method for optimizing NOMA systems with more than two multiplexed users per channel, which becomes a Higher Order Binary Optimization problem. We apply a method to reduce the order of the objective function, converting it to Quadratic Unconstrained Binary Optimization, and derive the Ising Hamiltonian for higher-order NOMA. We evaluate our proposed method through simulations using the CIM for optimization. Our results show that the proposed method leads to better solutions than conventional methods.
This paper delves into optimizing joint spectrum and power allocation between a matched primary and secondary link pair in cooperative spectrum sharing. Weighted sum energy efficiency (WSEE) is adopted as the objectiv...
详细信息
ISBN:
(数字)9798331542856
ISBN:
(纸本)9798331542863
This paper delves into optimizing joint spectrum and power allocation between a matched primary and secondary link pair in cooperative spectrum sharing. Weighted sum energy efficiency (WSEE) is adopted as the objective function to address the challenges of green communication. We propose a scheme based on deep reinforcement learning (DRL) to address this nonconvex spectrum and power allocation problem. Leveraging DRL, the spectrum and power allocation problem is autonomously optimized by only utilizing local information. Simulation results reveal that it achieves near-optimal performance and significantly enhances the network convergence speed with low computational overheads.
This study introduces a multi-objective cost management framework grounded in a multi-objective particle swarm optimization (MOPSO) approach. Initially, by establishing a virtual design framework for intelligent const...
详细信息
ISBN:
(数字)9798350389579
ISBN:
(纸本)9798350389586
This study introduces a multi-objective cost management framework grounded in a multi-objective particle swarm optimization (MOPSO) approach. Initially, by establishing a virtual design framework for intelligent construction that incorporates diverse aspects such as financial expenditure, timelines, and resource allocation, a comprehensive multi-objective cost management structure is developed. Subsequently, a refined version of the MOPSO algorithm is crafted to perform complex multi-dimensional optimizations encompassing cost, duration, and quality across various project scenarios, thereby fulfilling the distinct demands of numerous projects. The results from simulation experiments indicate that this algorithm achieves superior convergence precision and solution equilibrium in virtual design contexts, notably enhancing both the convergence rate and optimization precision of the model. Comparative analysis with traditional algorithms and alternative parameter configurations highlights the proposed algorithm's pronounced benefits in terms of objective function performance, allowing for a balanced optimization of multiple objectives and, consequently, achieving a more effective cost management outcome. This research offers an innovative and efficient approach to multi-objective cost management within intelligent construction's virtual design sphere, demonstrating strong potential for practical use and promising application opportunities.
DC motors find extensive application in various industries because of their accuracy in controlling torque and speed, which makes them crucial for machine operation, transportation and automation systems. Sliding mode...
详细信息
ISBN:
(数字)9798331543952
ISBN:
(纸本)9798331543969
DC motors find extensive application in various industries because of their accuracy in controlling torque and speed, which makes them crucial for machine operation, transportation and automation systems. Sliding mode control (SMC) offers enhanced resilience to system uncertainties and disturbances, despite the widespread use of conventional PID controllers. This study looks at how optimization algorithms, specifically the Chess Optimization Algorithm (COA) and Particle Swarm Optimization (PSO), can be used to fine-tune SMC parameters in order to make a linear DC motor work better. The study uses Integral Absolute Error (IAE) and Integral Square Error (ISE) as objective functions to look at things like rise time, overshoot, settling time and steady-state error that happen over time. The results show that COA is better than PSO because it has a shorter rise time (1.1245s for IAE and 1.4843s for ISE), less overshoot (7.0404x10 -5 % for IAE and 8.7887x10 -5 % for ISE) and less errors, even though PSO has slightly shorter settling times for ISE. The Chess Optimization Algorithm exhibits remarkable stability and accuracy, positioning it as a highly effective approach for optimizing intricate control systems such as SMC in linear DC motors.
Breast ultrasound is a non-invasive, economical method that is essential for the diagnosis of cancer. With the advent of deep learning in recent years, numerous CNN-based methods have been thoroughly studied for tasks...
详细信息
ISBN:
(数字)9798331506520
ISBN:
(纸本)9798331506537
Breast ultrasound is a non-invasive, economical method that is essential for the diagnosis of cancer. With the advent of deep learning in recent years, numerous CNN-based methods have been thoroughly studied for tasks including tumor localization and cancer classification. While prior single models performed well in both challenges, these approaches had several drawbacks, including long inference times, a need for a GPU, and the need for individual model fine-tuning. Our goal in this work is to create a new end-to-end multi-task architecture that can be used for both segmentation and classification. We obtained exceptional performance and time efficiency with our suggested method, achieving 79.8% and 86.4% in the segmentation challenge in DeepLabV3+ architecture.
We propose a data-driven approach for large-scale cellular network optimization, using a production cellular network in London as a case study and employing Sionna ray tracing for site-specific channel propagation mod...
详细信息
ISBN:
(数字)9798350368369
ISBN:
(纸本)9798350368376
We propose a data-driven approach for large-scale cellular network optimization, using a production cellular network in London as a case study and employing Sionna ray tracing for site-specific channel propagation modeling. We optimize base station antenna tilts and half-power beamwidths, resulting in more than double the 10%-worst user rates compared to a 3GPP baseline. In scenarios involving aerial users, we identify configurations that increase their median rates fivefold without compromising ground user performance. We further demonstrate the efficacy of model generalization through transfer learning, leveraging available data from a scenario source to predict the optimal solution for a scenario target within a similar number of iterations, without requiring a new initial dataset, and with a negligible performance loss.
暂无评论