Adaptive Weight Coefficients Research Articles

A feasibility restoration particle swarm optimizer with chaotic maps for two-stage fixed-charge transportation problems

This paper delves into solving the two-stage non-linear fixed-charge transportation problem (two-stage NFCTP), where each arc is associated with fixed and variable costs that increase proportionally to the square of the units transported. The presence of fixed charges and non-linear components categorizes this problem as NP−hard, leading to computational challenges, inefficiencies, and the risk of local optima. To address these challenges, a feasibility restoration particle swarm optimizer with chaotic maps (CEPSO) is presented. The proposed algorithm introduces (i) non-linear adaptive inertia weight and acceleration coefficients to maintain better exploration and exploitation rates during the search. (ii) Ten chaotic maps are integrated into the acceleration coefficients to enhance optimization capabilities further. (iii) Feasibility restoration mechanisms, including constraint compliance adjustment and ratio adjustment procedures, are incorporated to ensure the feasibility of solutions generated by CEPSO. The algorithm’s performance is evaluated across small and large-scale NFCTPs, ranging from 35 to 1044 dimensions, and compared to existing PSO variants using various evaluation metrics. Experimental analyses demonstrate CEPSO’s superior optimization performance for two-stage NFCTPs, positioning it as an advanced framework in this domain and contributing to the novelty of this study. The related codes can be found using this link: https://github.com/ChauhanDikshit.

Multi-Objective Optimization Strategy for Commercial Vehicle Permanent Magnet Water Pump Motor Based on Improved Sparrow Algorithm

In order to achieve multi-objective optimization for a permanent magnet water pump motor in heavy commercial vehicles, we propose a strategy based on response-surface methodology and the improved sparrow algorithm (CGE-SSA). Firstly, the output capacity of the pump during actual operation was tested with an experimental bench to determine the design parameters of the motor, and then its modeling was completed using Ansys Maxwell 2022r2 software. Secondly, the response-surface model was established by taking the parameters of permanent magnet width, rib width, and slot width as optimization parameters and the output torque (Ta), torque ripple (Tr), and back electromotive force (EMF) amplitude as optimization objectives. Meanwhile, three methods—namely, circular sinusoidal chaotic mapping, improved golden sinusoidal strategy, and adaptive weight coefficients—were used to improve the convergence speed and accuracy of the sparrow search algorithm (SSA). Finally, the multi-objective optimization of the permanent magnet synchronous motor was completed using the improved sparrow algorithm. A comparative analysis of the motor’s output before and after optimization showed that the torque pulsation and reverse electromotive force of the motor were significantly improved after optimization.

A novel hybrid algorithm based on improved marine predators algorithm and equilibrium optimizer for parameter extraction of solar photovoltaic models

In the field of solar photovoltaic (PV) systems, the accurate and reliable extraction of parameters from PV models is crucial for effective simulation, evaluation, and control. Although various optimization algorithms have been widely used for parameter extraction in solar PV systems, the accuracy and reliability of the parameters extracted by these methods usually fall short of the expected standards. To address these shortcomings, a novel hybrid algorithm that combines the improved marine predators algorithm (MPA) with the equilibrium optimizer (EO), named IMPAEO, is proposed. In the IMPAEO, an elite opposition-based learning strategy is introduced to expand the exploration range of the algorithm to enhance population diversity; an adaptive weight coefficient is employed to improve the updating rule of the MPA, facilitating a balance between global exploration and local exploitation; the EO operator is integrated to enhance the performance of the MPA in the local exploitation phase to improve the solution quality and convergence speed; and the linear population size reduction (LPSR) technique is introduced to dynamically reduce the population size and improve the search performance. The effectiveness of the proposed IMPAEO is validated using the CEC2017 benchmark test suite, which is also applied to extract parameters of the single diode model, the double diode model, and three different PV modules. Comparative analysis with other algorithms demonstrates that the IMPAEO exhibits significant advantages in terms of solution quality, convergence speed, and algorithm stability. Consequently, the proposed IMPAEO not only proves to be efficient and reliable in parameter extraction for solar PV models but also offers a promising alternative in this field.

A short-term wind speed prediction method based on the IDBO-BPNN

Wind energy is known for its uncertainty and volatility, necessitating accurate wind speed prediction for stable wind farm operations. To enhance wind speed prediction accuracy, this study proposes a BP neural network (BPNN) short-term wind speed prediction model based on the Improved Dung Beetle Optimization (IDBO) algorithm. Addressing the issue of local optimization and reduced accuracy in the BPNN optimized by the Dung Beetle Optimization (DBO) algorithm, the circle chaotic mapping is utilized for population initialization to achieve a more uniform initial distribution. The improved sine-cosine algorithm, triangle wandering strategy, and adaptive weight coefficient are then employed to optimize dung beetle positions, balancing global exploration and local development capabilities and improving the algorithm’s search performance. Finally, the improved DBO algorithm optimizes the weights and thresholds of the BPNN, and the IDBO-BPNN prediction model was constructed. Simulation experiments were conducted based on wind speed data from a wind farm in Ohio, USA. The IDBO-BPNN model was compared with other prediction models, and error evaluation indexes were introduced to evaluate the experimental results. The findings demonstrate that the suggested model yields the most accurate predictions and achieves the optimal error evaluation indexes. MAE, MSE, RMSE, NSE and R2 of dataset 1 are 0.42247, 0.28775, 0.53642, 88.8785%, 89.161%, those of dataset 2 are 0.28283, 0.14952, 0.38668, 85.7383%, 86.577%, and those of dataset 3 are 0.45406, 0.39268, 0.62664, 84.3859%, 84.931%. In particular, compared with BPNN model, the five evaluation indexes of the IDBO-BPNN model promoted by 41.53%, 57.38%, 34.71%, 24.91%, and 11.44%, respectively in dataset 3. Therefore, that the proposed IDBO-BPNN model exhibits higher accuracy in short-term wind speed prediction, indicating its feasibility and superiority in the realm of wind energy.

An interpretable TFAFI-1DCNN-LSTM framework for UGW-based pre-stress identification of steel strands

Steel strands serve as the key load-bearing components of pre-stressed bridges, yet the identification of effective pre-stress for steel strands is a challenging task. In this study, a time and frequency adaptive fusion input-one dimensional convolutional neural network-long short-term memory (TFAFI-1DCNN-LSTM) framework was proposed for ultrasonic guided wave (UGW)-based effective pre-stress identification of steel strands. In this framework, the time and frequency domain signals of UGW were input into the network as two parallel branches, CNN and LSTM were utilized to extract spatial and serial-related features, and a trainable weight coefficient was introduced for adaptive fusion of time and frequency domain features. Firstly, ablation studies were conducted to investigate the influence of each key component in the proposed framework on the prediction results. Secondly, deep learning (DL) models of Res Net, Dense Net and Inception Net were introduced as comparisons, and the prediction results based on TFAFI-1DCNN-LSTM were compared with those based on the above DL models. Thirdly, the noise-robustness and performance under a few trainable samples of TFAFI-1DCNN-LSTM were also investigated. Finally, the outputs of various key layers of the proposed framework were subjected to dimensionality reduction and visualization to provide an intuitive demonstration of the feature extraction process, and the inherent mechanisms of the proposed framework were interpreted using local interpretable model-agnostic explanations (LIME). Results show that the architecture of TFAFI-1DCNN-LSTM is reasonably designed and all key components are necessary. The weights of features in time and frequency domain branches are rationally assigned due to the addition of an adaptive weight coefficient. Compared with other DL models, the proposed TFAFI-1DCNN-LSTM exhibits higher pre-stress identification accuracy, excellent noise-robustness, and superior performance under only a few trainable samples. The correlation between features extracted by the network and the pre-stress labels significantly increases with the deepening of the network, and the correlation between input UGW signals and predicted pre-stress values is effectively interpreted utilizing LIME, proving that the architecture of the proposed framework is reasonable and effective, the predicted results are reliable and credible.

A novel data-driven IBA-ELM model for SOH/SOC estimation of lithium-ion batteries

Lithium-ion batteries (Libs) have become the best power source for electric vehicles due to their high energy density and long cycle life. The state estimation is directly related to the safe and efficient use of Libs. In our previous research, a Bat Algorithm-Extreme Learning Machine (BA-ELM) model was constructed and promising results were achieved for state of health (SOH) estimation. To increase the engineering application value of BA-ELM for SOH/state of charge (SOC) estimation, optimization mechanism of improving the global exploration and local exploitation capabilities of the model was studied. First, three improved methods including a conversion mechanism of Artificial Bee Colony algorithm employing bees and scout bees, a mechanism for adaptive adjustment of the search frequency of bats according to the Euclidean distance difference, and an adaptive inertia weight coefficient method were proposed to construct a novel Improved Bat Algorithm (IBA)-ELM model. Second, a battery life degradation test platform was built to obtain the degradation dataset for Libs. The variation rules of nine battery variables and cycles were revealed. Four high-quality SOH-related features were extracted. Thus, the SOH estimation performance of six models was evaluated. Third, a Libs charge–discharge dataset under driving conditions was collected and processed. Four models were trained based on 72,048 samples from four conditions. Further, SOC estimation tests were conducted on 35,071 samples from completely different conditions. The results show that the root mean square error (RMSE) and mean absolute error (MAE) of SOH estimation by IBA-ELM decreased by 21.93 % and 23.79 %, respectively, compared to those by BA-ELM. The RMSE and MAE of SOC estimation decreased by more than 10 %. Error, radar, and violin plots demonstrate that the IBA-ELM could accurately estimate SOH/SOC in larger scale datasets, simultaneously exhibiting better generalization performance.

Research on LiDAR point cloud data transformation method based on weighted altitude difference map

Road surface detection plays a pivotal role in the realm of autonomous vehicle navigation. Contemporary methodologies primarily leverage LiDAR for acquiring three-dimensional data and utilize imagery for chromatic information. However, these approaches encounter significant integration challenges, particularly due to the inherently unstructured nature of 3D point clouds. Addressing this, our novel algorithm, specifically tailored for predicting drivable areas, synergistically combines LiDAR point clouds with bidimensional imagery. Initially, it constructs an altitude discrepancy map via LiDAR, capitalizing on the height uniformity characteristic of planar road surfaces. Subsequently, we introduce an innovative and more efficacious attention mechanism, streamlined for image feature extraction. This mechanism employs adaptive weighting coefficients for the amalgamation of the altitude disparity imagery and two-dimensional image features, thereby facilitating road area delineation within a semantic segmentation framework. Empirical evaluations conducted using the KITTI dataset underscore our methodology’s superior road surface discernment and extraction precision, substantiating the efficacy of our proposed network architecture and data processing paradigms. This research endeavor seeks to propel the advancement of three-dimensional perception technology in the autonomous driving domain.

An unsupervised multi-focus image fusion method via dual-channel convolutional network and discriminator

The challenge in multi-focus image fusion tasks lies in accurately preserving the complementary information from the source images in the fused image. However, existing datasets often lack ground truth images, making it difficult for some full-reference loss functions (such as SSIM) to effectively participate in model training, thereby further affecting the performance of retaining source image details. To address this issue, this paper proposes an unsupervised dual-channel dense convolutional method, DCD, for multi-focus image fusion. DCD designs Patch processing blocks specifically for the fusion task, which segment the source image pairs into equally sized patches and evaluate their information to obtain a reconstructed image and a set of adaptive weight coefficients. The reconstructed image is used as the reference image, enabling unsupervised methods to utilize full-reference loss functions in training and overcoming the challenge of lacking labeled data in the training set. Furthermore, considering that the human visual system (HVS) is more sensitive to brightness than color, DCD trains the dual-channel network using both RGB images and their luminance components. This allows the network to focus more on the brightness information while preserving the color and gradient details of the source images, resulting in fused images that are more compatible with the HVS. The adaptive weight coefficients obtained through the Patch processing blocks are also used to determine the degree of preservation of the brightness information in the source images. Finally, comparative experiments on different datasets also demonstrate the superior performance of DCD in terms of fused image quality compared to other methods.

Adaptive Reactive Power Optimization in Offshore Wind Farms Based on an Improved Particle Swarm Algorithm

To address the reactive power optimization control problem in offshore wind farms (OWFs), this paper proposes an adaptive reactive power optimization control strategy based on an improved Particle Swarm Optimization (PSO) algorithm. Firstly, an OWF multi-objective optimization control model is established, with the total sum of voltage deviations at wind turbine (WT) terminals, active power network losses, and reactive power margin of WTs as comprehensive optimization objectives. Innovatively, adaptive weighting coefficients are introduced for the three sub-objectives, enabling the weights of each optimization objective to be adaptively adjusted based on real-time operating conditions, thus enhancing the adaptability of the reactive power optimization model to changes in operating conditions. Secondly, a Uniform Adaptive Particle Swarm Optimization (UAPSO) algorithm is proposed. On one hand, the algorithm initializes the particle swarm using a uniform initialization method; on the other hand, it improves the particle velocity update formula, allowing the inertia coefficient to adaptively adjust based on the number of iterations and the fitness ranking of particles. Simulation results demonstrate the following: (1) Under various operating conditions, the proposed adaptive multi-objective reactive power optimization strategy can ensure the stability of node voltages in offshore wind farms, reduce active power losses, and simultaneously improve reactive power margins. (2) Compared with the traditional PSO algorithm, UAPSO exhibits an approximately 10% improvement in solution speed and enhanced solution accuracy.

A Chlorophyll-a Concentration Inversion Model Based on Backpropagation Neural Network Optimized by an Improved Metaheuristic Algorithm

Chlorophyll-a (Chl-a) concentration monitoring is very important for managing water resources and ensuring the stability of marine ecosystems. Due to their high operating efficiency and high prediction accuracy, backpropagation (BP) neural networks are widely used in Chl-a concentration inversion. However, BP neural networks tend to become stuck in local optima, and their prediction accuracy fluctuates significantly, thus posing restrictions to their accuracy and stability in the inversion process. Studies have found that metaheuristic optimization algorithms can significantly improve these shortcomings by optimizing the initial parameters (weights and biases) of BP neural networks. In this paper, the adaptive nonlinear weight coefficient, the path search strategy “Levy flight” and the dynamic crossover mechanism are introduced to optimize the three main steps of the Artificial Ecosystem Optimization (AEO) algorithm to overcome the algorithm’s limitation in solving complex problems, improve its global search capability, and thereby improve its performance in optimizing BP neural networks. Relying on Google Earth Engine and Google Colaboratory (Colab), a model for the inversion of Chl-a concentration in the coastal waters of Hong Kong was built to verify the performance of the improved AEO algorithm in optimizing BP neural networks, and the improved AEO algorithm proposed herein was compared with 17 different metaheuristic optimization algorithms. The results show that the Chl-a concentration inversion model based on a BP neural network optimized using the improved AEO algorithm is significantly superior to other models in terms of prediction accuracy and stability, and the results obtained via the model through inversion with respect to Chl-a concentration in the coastal waters of Hong Kong during heavy precipitation events and red tides are highly consistent with the measured values of Chl-a concentration in both time and space domains. These conclusions can provide a new method for Chl-a concentration monitoring and water quality management for coastal waters.

Energy-Saving Control Strategy for Multisleep States of CNC Machine Tool Considering Components Priority

As the representative energy-consuming equipment in the manufacturing industry, machine tools consume lots of energy during the non-machining period, which leads to low energy efficiency. The existing strategies do not consider the startup sequence among components, which causes the control parameters to fail to run during the practical startup process, and this paper conducts energy-saving research to address the issues. The relationship of priority among components was used as the constraint to obtain an energy-saving control strategy for multi-sleep states of machine tools considering components priority. Furthermore, the improved whale algorithm was applied to solve the energy-saving control strategy of the machine tool, and the search and convergence ability of the algorithm was improved by introducing the chaotic convergence factor and adaptive weight coefficient. Finally, the energy-saving strategy was utilized with a vertical CNC machine tool as an experimental example to mill the end face of the randomly arriving workpieces. The results show that this changed algorithm has a better solution speed and effect than the ordinary whale algorithm and some other improved whale algorithms. Compared with the practical control process and other control strategies, the control strategy considering components priority can obtain more stable productivity and better energy-saving effect.

Optimization of frequency dynamic characteristics in microgrids: An improved MPC-VSG control

For the power imbalance caused by the load switching in microgrids (MGs), which in turn causes the frequency crossing limit problem. In this paper, we propose an improved model predictive control (MPC) based on the existing MPC-VSG control, combining adaptive inertia damping control and adaptive weight coefficient control for joint control, and adjusting the inertia damping coefficient of VSG in real-time by adaptive inertia damping control. In MPC control, the measured parameters at the current moment are used to predict the state quantities at the future moment, and then the target function is designed with the angular frequency deviation and the VSG power output as performance indicators, and the weight coefficients in the target function are updated in real-time by detecting the frequency deviation. After solving the objective function to obtain the required power compensation, the power reference value of the VSG is updated to optimize the dynamic response of the frequency. Experiment and comparison analysis with two existing methods show that the proposed method can further optimize the frequency response and smooth the output power of other power supply components in the microgrid when they are disturbed.

An Enhanced Particle Swarm Optimization Algorithm via Adaptive Dynamic Inertia Weight and Acceleration Coefficients

The particle swarm optimization (PSO) algorithm counts among the most popular metaheuristic algorithms based on swarm intelligence. Since the publication of the first article on this optimization technique, researchers have developed many PSO variants with some improvement in its performance. The PSO optimization performance hinges on its ability to achieve a good exploration-exploitation balance. The most common method that helps to improve exploration-exploitation balance is modifying the PSO three controlling parameters, namely the inertia weight and two acceleration coefficients. In this paper a PSO variant that combines adaptive dynamic inertia weight and adaptive dynamic acceleration coefficients to enhance the exploration-exploitation balance of the PSO is proposed. The enhanced PSO algorithm called Adaptive Dynamic Inertia Weight and Acceleration Coefficient Optimization (ADIWACO) algorithm is tested on seven well-known standard test functions comprising four unimodal and three multimodal ones. The performance of the PSO is then compared with that of the standard PSO (SPSO) and four existing PSO variants. The experimental results comprising optimum value, runtime, mean value, standard deviation and convergence rate, and confirmed by the results of ranking statistics and Wilcoxon signed rank test conducted on the experimental results, indicate significantly better performance by the proposed PSO algorithm.

An extended conformable fractional order grey prediction model of per capita electricity consumption for daily life in China

Conformable fractional-order grey prediction models have attracted considerable attention due to their versatile modeling techniques. However, existing models often suffer from limitations in adaptability. To address this, this study proposes a new extended conformable fractional-order grey prediction model, namely the ECFGM(1,1) model. By integrating an adaptive weighting coefficient into the conformable fractional-order accumulation process, the model can effectively prioritize new information, thereby enhancing its rationality and adaptability. Moreover, the adjusted process can be tailored to either emphasize new information or adhere to traditional accumulation methods, which improves its adaptability. To verify the effectiveness of the ECFGM(1,1) model, ECFGM(1,1) is applied to two examples from the literature. The model evaluation results show that the ECFGM(1,1) model has higher fitting accuracy and predictive accuracy than the GM(1,1), CFGM(1,1), and NIPGM(1,1) models. Using the constructed ECFGM(1,1) for predictive analysis of the per capita electricity consumption for daily life in China, the results show that this model can capture the laws of its changes over time. Finally, per capita electricity consumption for daily life in China from 2022 to 2026 is predicted. The results show that by 2026, such consumption is estimated to reach 1165.35 KW.h.

Adaptive dynamic windowing approach based on risk degree function

Aiming at the problem that the dynamic window approach (DWA) has low planning efficiency and it is difficult to avoid fast-moving obstacles, an improved adaptive DWA based on risk degree function is proposed in this paper. The risk function is designed and introduced into the evaluation function of the traditional DWA to evaluate the risk of collision between dynamic obstacles and the robot, so that the robot can effectively avoid faster obstacles. Then, according to the fuzzy control principle, the adaptive weight coefficient is designed to improve the evaluation function, so that the mobile robot can move to the target point more efficiently. The simulation results show that compared with the traditional DWA, the adaptive DWA based on risk degree function reduces the time of completing the planning task by about 8%, and the path length after the completion of the planning by about 8%, it indicates that the improved algorithm has higher efficiency. After 50 repeated experiments, using the adaptive DWA based on risk degree function successfully avoids all obstacles to complete the planning task, which shows that this algorithm has higher security.

DYC Design for Autonomous Distributed Drive Electric Vehicle Considering Tire Nonlinear Mechanical Characteristics in the PWA Form

This paper presents a novel direct yaw moment control (DYC) strategy for autonomous distributed drive electric vehicle (DDEV) to improve path-following driving stability under critical maneuvers. Due to the fact that the tire mechanical characteristics show highly nonlinear under those critical conditions, the piecewise affine (PWA) identification method is chosen in this work to model the tire nonlinear mechanical characteristics since it shows best model accuracy/simplicity compromise for system control design. On this basis, to properly reflect the coupling behaviors between longitudinal motion and lateral motion, the vehicle dynamics model which considers the tire nonlinear mechanical characteristics in the PWA form is established. By analyzing the phase plane of vehicle nonlinear model, a novel external yaw moment calculation method based on global fast terminal sliding mode (GFTSM) control method is applied to the upper layer of DYC architecture for eliminating the errors of yaw rate and sideslip angle. Then, different weight coefficients are allocated to obtain the final external yaw moment. In the lower controller, the optimal four-wheel torque is distributed according to a concise and accurate algorithm in which adaptive weight coefficients are considered. The simulation results verify the effectiveness of the proposed control strategy in improving the driving stability performance of autonomous DDEV against two other control strategies under three extreme conditions.

Energy-Saving Adaptive Routing for High-Speed Railway Monitoring Network Based on Improved Q Learning.

In high-speed railway operational monitoring network systems targeting railway infrastructure as its monitoring objective, there is a wide variety of sensor types with diverse operational requirements. These systems have varying demands on data transmission latency and network lifespan. Most of the previous research focuses only on prolonging network lifetime or reducing data transmission delays when designing or optimizing routing protocols, without co-designing the two. In addition, due to the harsh operating environment of high-speed railways, when the network changes dynamically, the traditional routing algorithm generates unnecessary redesigns and leads to high overhead. Based on the actual needs of high-speed railway operation environment monitoring, this paper proposes a novel Double Q-values adaptive model combined with the existing reinforcement learning method, which considers the energy balance of the network and real-time data transmission, and constructs energy saving and delay. The two-dimensional reward avoids the extra overhead of maintaining a global routing table while capturing network dynamics. In addition, the adaptive weight coefficient is used to ensure the adaptability of the model to each business of the high-speed railway operation environment monitoring system. Finally, simulations and performance evaluations are carried out and compared with previous studies. The results show that the proposed routing algorithm extends the network lifecycle by 33% compared to the comparison algorithm and achieves good real-time data performance. It also saves energy and has fewer delays than the other three routing protocols in different situations.

Application of Improved Butterfly Optimization Algorithm in Mobile Robot Path Planning

An improved butterfly optimization algorithm (IBOA) is proposed to overcome the disadvantages, including slow convergence, generation of local optimum solutions, and deadlock phenomenon, of the optimization algorithm in the path planning of mobile robots. A path-planning grid model is established based on an improved obstacle model. First, the population diversity is improved by introducing kent mapping during population position renewal in the normal butterfly optimization algorithm (BOA) to enhance the global search ability of the butterfly population. Second, an adaptive weight coefficient is introduced in the renewal process of each generation to increase the convergence speed and accuracy. An opposition-based learning strategy based on convex lens imaging is introduced to help the butterfly population jump out of the local optimum. Finally, a mutation strategy is introduced to solve the path planning problem. On this basis, two path simplification strategies are proposed to make up for the shortcomings of planning paths in grid maps. The shortest path lengths solved by IBOA, BOA, and GA in the 20 × 20 map are 30.97, 31.799, and 31.799, respectively. The numbers of iterations for the shortest paths searched by IBOA, BOA, and GA are 14, 24, and 38 in that order. The shortest path lengths solved by IBOA, BOA and GA in the 40 × 40 map are 63.84, 65.60, and 65.84, respectively. The number of iterations for the shortest paths searched by IBOA, BOA and GA are 32, 40, and 46, respectively. Simulation results show that IBOA has a strong ability to solve robot path planning problems and that the proposed path simplification strategy can effectively reduce the length of the optimal path in the grid map to solve the path planning problem of mobile robots. The shortest paths solved by IBOA in 20 × 20 and 40 × 40 maps are simplified to lengths of 30.2914 and 61.03, respectively.

Dual-path multi-scale context dense aggregation network for retinal vessel segmentation

There has been steady progress in the field of deep learning-based blood vessel segmentation. However, several challenging issues still continue to limit its progress, including inadequate sample sizes, the neglect of contextual information, and the loss of microvascular details. To address these limitations, we propose a dual-path deep learning framework for blood vessel segmentation. In our framework, the fundus images are divided into concentric patches with different scales to alleviate the overfitting problem. Then, a Multi-scale Context Dense Aggregation Network (MCDAU-Net) is proposed to accurately extract the blood vessel boundaries from these patches. In MCDAU-Net, a Cascaded Dilated Spatial Pyramid Pooling (CDSPP) module is designed and incorporated into intermediate layers of the model, enhancing the receptive field and producing feature maps enriched with contextual information. To improve segmentation performance for low-contrast vessels, we propose an InceptionConv (IConv) module, which can explore deeper semantic features and suppress the propagation of non-vessel information. Furthermore, we design a Multi-scale Adaptive Feature Aggregation (MAFA) module to fuse the multi-scale feature by assigning adaptive weight coefficients to different feature maps through skip connections. Finally, to explore the complementary contextual information and enhance the continuity of microvascular structures, a fusion module is designed to combine the segmentation results obtained from patches of different sizes, achieving fine microvascular segmentation performance. In order to assess the effectiveness of our approach, we conducted evaluations on three widely-used public datasets: DRIVE, CHASE-DB1, and STARE. Our findings reveal a remarkable advancement over the current state-of-the-art (SOTA) techniques, with the mean values of Se and F1 scores being an increase of 7.9% and 4.7%, respectively. The code is available at https://github.com/bai101315/MCDAU-Net.

DQN regenerative braking control strategy based on adaptive weight coefficients

Aiming at the problems existing in regenerative braking control strategy based on Q-learning which include the dimensional disaster of state and action variables discretization and the return function weight coefficient determined empirically. This paper proposes deep Q-learning network (DQN) regenerative braking control strategy based on adaptive weight coefficients. Firstly, braking performance evaluation indexes are determined which are braking energy recovery efficiency and braking stability coefficient. Then, the state and action variables and return function are constructed respectively. Therein the braking demand power and state of charge ( SOC) are taken as state variables, braking torque proportional coefficient, and weight coefficients are taken as action variables. And return function is formulated by trading off braking energy recovery efficiency and braking stability. Finally, using the MATLAB/Simulink software, the simulation model of real working condition in Yubei district of Chongqing is established. The simulation results show that braking recovery efficiency of the proposed strategy is 7.4% higher than that of Q-learning strategy, and the average braking stability coefficient is decreased by 0.08. The results indicate the proposed strategy can better balance between braking energy recovery efficiency and braking stability than the conventional strategy.