To address the issue of missed and false detections of small objects of helmet wearing detection in construction scenarios， caused by reasons such as crowding， occlusion， and complex backgrounds， a cross-layer multi-scale helmet wearing detection algorithm with double attention mechanism based on YOLOv8n was proposed. Firstly， a small object detection head was designed to enhance the model’s ability to detect small objects. Secondly， the double attention mechanism was embedded in the feature extraction network to focus more on capturing object features in complex scenarios. Thirdly， the feature fusion network was replaced with the cross-layer multi-scale feature fusion structure S-GFPN （Selective layer Generalized Feature Pyramid Network）， which was improved with Re-parameterized Generalized Feature Pyramid Network （RepGFPN）， so as to enable multi-scale fusion of small object feature layer with other layers and establish long-term dependencies， thus reducing background information interference. Finally， the MPDIOU （Intersection Over Union with Minimum Point Distance） loss function was employed to address non-sensitivity issues related to scale changes. Experimental results on the public dataset GDUT-HWD show that compared to the YOLOv8n， the improved model increases the mAP@0.5 by 3.4 percentage points， and improves the detection accuracy for blue， yellow， white， and red helmets by 2.0， 1.1， 4.6， and 9.1 percentage points， respectively. The model also outperforms the YOLOv8n in five complex scenarios： density， occlusion， small objects， light reflection， and darkness， and provides an effective method for helmet wearing detection in real-world construction scenarios.

Aiming at the problem that the key entity information in the police field is difficult to recognize， a neural network model based on BERT （Bidirectional Encoder Representations from Transformers）， namely BERT-BiLSTM-Attention-CRF， was proposed to recognize and extract related named entities， in the meantime， the corresponding entity annotation specifications were designed for different cases. In the model ，the BERT pre-trained word vectors were used to replace the word vectors trained by the traditional methods such as Skip-gram and Continuous Bag of Words （CBOW）， improving the representation ability of the word vector and solving the problem of word boundary division in Chinese corpus trained by the character vectors. And the attention mechanism was used to improve the architecture of classical Named Entity Recognition （NER） model BiLSTM-CRF. BERT-BiLSTM-Attention-CRF model has an accuracy of 91% on the test set， which is 7% higher than that of CRF++ Baseline， and 4% higher than that of BiLSTM-CRF model. The F1 values of the entities are all higher than 0.87.

In order to achieve the high efficiency in automatic train operation, on the basis of the analysis of the train at different stages of operation, taking parking as the key stage, analytic hierarchy process was used to get quantitative description of the importance between each performance indexes and evaluation function of parking controls comprehensive performance in this stage, then the fuzzy manipulation rules of the online control were got. The offline operation of train under the rules was simulated for several times, the different schemes in sub-regional division and start braking point selection were scored to get the parking manipulation scheme which performance indexes are the best. Finally the simulation system was designed based on VC++ platform, and it has verified that the practical effect of the train running under the control algorithm has good parking precision, comfort and time saving.

To synthesize facial pose and expression images simultaneously from one image, a tensor-based subspace projection method for synthesizing multi-pose and expression face images was proposed. Firstly, the forth order texture tensor and shape tensor were created from the feature annotated images respectively. Then a tucker tensor decomposition technique was applied to build projection subspaces (person, expression, pose and feature subspaces). Core tensors, expressions, poses and feature subspaces were organized into a new tensor properly which was used for synthesizing new facial poses and expressions. The proposed method took full advantage of the intrinsic relationship among the facial affected various factors. The experimental results show that the proposed method can synthesize different facial expressions with kinds of poses of the face using a known facial expression and pose image.

Based on the review of recent development of evolutionary computation and the principle of free energy minimization of thermodynamics, a new thermodynamics evolutionary algorithm for solving Shubert function optimization problem was proposed. The numerical experiments were conducted to measure the performance of thermodynamics evolutionary algorithm. The results show that thermodynamics evolutionary algorithm is of potential to obtain global optimum or more accurate solutions than other evolutionary methods.