In order to achieve the accuracy of mango grading, a mango grading system was designed by using the deep learning method. The system mainly includes CCD camera image acquisition, image preprocessing, model training, and model evaluation. Aiming at the traditional deep learning, neural network training needs a large number of sample data sets; a convolutional neural network is proposed to realize the efficient grading of mangoes through the continuous adjustment and optimization of super-parameters and batch size. The ultra-lightweight SqueezeNet related algorithm is introduced. Compared with AlexNet and other related algorithms with the same accuracy level, it has the advantages of small model scale and fast operation speed. The experimental results show that the convolutional neural network model after super-parameters optimization and adjustment has excellent effect on deep learning image processing of small sample data set. Two hundred thirty-four Jinhuang mangoes of Panzhihua were picked in the natural environment and tested. The analysis results can meet the requirements of the agricultural industry standard of the People’s Republic of China—mango and mango grade specification. At the same time, the average accuracy rate was 97.37%, the average error rate was 2.63%, and the average loss value of the model was 0.44. The processing time of an original image with a resolution of 500 374 was only 2.57 milliseconds. This method has important theoretical and application value and can provide a powerful means for mango automatic grading.

Accurate traffic sign recognition (TSR) is critical for enhancing the safety and reliability of autonomous driving systems. This study proposes an optimized YOLOv5-based framework to address challenges such as small-scale detection, environmental variability, and real-time processing constraints. Three key innovations are introduced: (1) k-means++ clustering for anchor box optimization, achieving a 77.55% average IoU (vs. 75.95% for traditional k-means) to enhance small-target detection; (2) comprehensive comparative analysis of YOLOv5 variants (s/m/x), revealing precision-speed trade-offs (99.3–99.5% mAP@0.5 vs. 32–45 ms inference time) for deployment flexibility; and (3) systematic hyperparameter tuning to maximize robustness across diverse scenarios. Leveraging the CCTSDB dataset (13,830 annotated images), experiments demonstrate the framework’s superiority: it attains 98.1% mean average precision (mAP), 98.6% recall, and 99.3% precision, outperforming Faster-RCNN and SSD by 5–8% in mAP while maintaining 45 FPS throughput. The YOLOv5s variant achieves optimal balance with 99.3% mAP@0.5 and 32 ms per-image inference, validated through rigorous statistical analysis (Tukey HSD). Robust performance in challenging conditions (e.g., small sample, backlit sample, foggy scenes) is evidenced by detection confidence exceeding 0.90. These results highlight the framework’s applicability in latency-sensitive intelligent transportation systems.

As one of the important load-bearing components of a truck, the drive axle housing must meet the requirements of stiffness and strength. The traditional design method uses redundancy design to meet the performance requirements. The joint design between the three-dimensional mathematical model and finite element model is adopted, and the optimal design of the drive axle housing is realized based on topology optimization and multiobjective optimization. Firstly, the static analysis of the drive axle housing of a rear axle drive truck was carried out with four typical working conditions. It was concluded that the four working conditions all operate under the yield limit of the material, and it was found that the maximum equivalent stress of the four working conditions occurs at the step of the half-shaft casing. Among the four working conditions, the most critical one is the maximum vertical force working condition. Then, based on the maximum vertical force working condition, the fatigue life analysis is conducted, and the minimum fatigue life appears at the transition position of the half-shaft sleeve and the arc transition position of the main reducer chamber. The remaining parts can meet the design requirements. The overall safety factor of the drive axle housing is mainly between 1 and 5 when operating under this working condition. Then, through modal analysis, the first to sixth natural frequency and vibration modes of the drive axle housing are extracted. Based on the modal analysis, the dynamic characteristics of the drive axle housing are further studied by harmonic response analysis and random vibration analysis. Finally, two kinds of lightweight optimization schemes for the drive axle housing are given. Topology optimization reduces the mass of the drive axle housing by 17.4%, but the overall performance slightly decreases. Then, the five dimensional parameters of the drive axle housing are selected as design variables. The mass, maximum deformation, equivalent stress, service life, and the first-, second- and third-order natural frequencies are defined as objective functions. Through the optimal space-filling design method, the experimental designs are performed and the sample points are obtained. Based on the results of experiment design, the multiobjective genetic algorithm and response surface method are combined to optimize the objective functions. The analysis results show that the mass is reduced by 4.35%, the equivalent stress is reduced by 21.05%, the minimum life is increased by 72.28%, and the first-, second-, and third-order natural frequency are also increased to varying degrees. Two different optimization strategies are provided for the design of the drive axle housing.

To solve the problem between temperature and thermal stress of a titanium alloy piston, a multi-objective optimization method combined response surface methodology and experiment design is performed to calculate an optimal design of the titanium alloy piston. Firstly, the thermo-mechanical coupling analysis, static and dynamic characteristics analysis are carried out. The analysis results show that the thermal load has a significant influence on the piston. Secondly, the five dimensional parameters are chose as the design variables through sensitivity analysis. The piston thermo-mechanical coupling stress, deformation, mass, and the first ring groove are selected as the objective functions. By using the optimal space-filling design method, the sample points between design variables and objective functions are gained. Then, the mathematical relationship between them is obtained by the response surface method. Finally, the multi-objective genetic algorithm is employed to optimize the mathematical model. After optimization, the maximum coupling deformation decreases by 3.05%, the maximum coupling stress decreases by 27.85%, the mass reduces by 5.46%. The maximum temperature of the first ring groove reduces by 12.54%. This study can provide a reference for the piston optimization design.

To improve the applicable performance of the engine and accomplish the determination of optimization, the crankshaft, as one of its essential components, is optimized based on experimental design and response surface method (RSM). First, the static analysis of the crankshaft is carried out. Over the solution, the maximum equivalent stress and deformation distribution of the crankshaft and the area with the most concentrated stress are obtained to provide the direction for subsequent optimization. Then the modal analysis is carried out to define the portions with large amplitude to prevent resonance. At the same time, based on the modal analysis results, the harmonic response of the crankshaft is examined to ensure that the crankshaft can avoid resonance and fatigue caused by forced vibration at different engine speeds. It also provides a theoretical basis for the further optimization design of the crankshaft. Then, through the experimental design and response surface optimization analysis, the optimization purpose of reducing the maximum equivalent stress, maximum deformation, and lightweight of the crankshaft is accomplished. Through the design expert software, the response surface model and fitting equation of the crankshaft are solved by the RSM, which provides a reference basis for optimizing the crankshaft. Finally, according to the optimization scheme, the crankshaft is reconstructed, and the optimized crankshaft model is compared with the original model. After optimization, the maximum equivalent stress is reduced by 9.43%, the maximum deformation is reduced by 3.68%, and the mass is reduced by 1.30%. At the same time, the first three natural frequencies are reduced by 10.28, 10.28, and 10.34%, which achieves the purpose of optimizing the crankshaft design. The research results provide effective theory and evidence for the structural design of the crankshaft.