In the garment production industry, garment cutting size matching plan is an important step in the process, which plays a decisive role in production management and cost control. In this paper, we first model the size matching problem of garment cutting, then use the improved fast particle swarm algorithm (APSO) to optimize the multi-objective optimization solution, and finally verify the performance of the APSO algorithm and the actual effect of garment size matching with cases. Comparing the test results of APSO, PSO and LDWPSO algorithms in the six test functions of Griewank, Ackle, Levy, Rastrign, Schwefel and Sphere, it can be seen that: with the improvement of the problem dimensions, the APSO algorithm used in this paper can still maintain a better optimization accuracy, and the optimization accuracy and stability are significantly improved compared with the PSO and the LDWPSO algorithms. LDWPSO algorithms. In the actual case, the APSO algorithm is more reasonable in the size combination and the number of layers of fabric, for four different types of apparel orders have obtained a superior optimal solution set, cutting production error is far less than the enterprise requirements. At the same time, compared with other optimization methods, the APSO algorithm has better optimization accuracy and solving efficiency, and can obtain a more superior cutting and bed splitting scheme. The algorithm proposed in this paper can effectively optimize the cutting size matching process, reduce fabric waste and production equipment investment, and has good application value and reference significance.

The global urbanization process is rapidly increasing, and a reasonable and scientific analysis of the relationship between urban land expansion and land resources plays an important role in the rational allocation and coordinated development of land resources. This paper constructs a spatio-temporal geographic weighted regression model coupled with geospatial and temporal coordinates, and incorporates temporal and spatial non-stationarity into the model. Then, using the method of hypothesis testing, the temporal non-stationarity and spatial non-stationarity of the spatio-temporal geographic weighted regression model are examined, and at the same time, the multiple covariance test and the variance expansion factor method are proposed to carry out further statistical inference of the model. As the degree of urban sprawl increases, the land resources weaken year by year from the center to the surrounding area.The global Moran’s I for the three periods from 2003 to 2023 are 0.6289, 0.7159, and 0.7368, respectively, which show a trend of increasing year by year. It shows that land resources are strongly influenced by urban expansion, and the spatial distribution of land resources shows spatial aggregation. Several variables, such as building volume rate, population size, regional economic development, regional cultural level, infrastructure construction and urban fallow area, have significant effects on the spatial differentiation of land resources. The above differentiation characteristics provide insights into the rationalization of urban expansion and the scientific allocation of land resources.

In the process of sharing accounting information using cloud computing technology, the integrity of the data is related to the security of the transmission and utilization of accounting information. For this reason, this paper studies the algorithm optimization based on the multi-branch path tree LBT. Multi-branch path tree LBT adopts distributed data storage method to reduce the number of hash operations. The data integrity auditing scheme is designed for different phases of cloud auditing, and the dynamic update process of cloud data is optimized to improve the data integrity verification effect. This algorithm can still maintain a high challenge success rate after more than 300 challenge data blocks, and the total overhead of the experimental computation does not exceed 8 ms, and the verification efficiency is also better than the comparison algorithm. Therefore, the research idea of this paper has validity and has improved effect on data integrity verification in the process of cloud computing smart accounting informatization.

This paper deeply analyzes the innovative application and intelligent upgrading steps of Artificial Intelligence Generated Content (AIGC) in Civic and Political Education. Based on metadata, we construct an automated generation model of Civics education resources, divide the meta-properties of education knowledge resources, set up a knowledge tracking model DT-BKT to obtain students’ mastery of Civics knowledge, adopt personalized recommendation model to realize the high adaptability of education resources based on students’ Civics learning, and combine the functions of each model to build a Civics education content intelligent generation and adaptability system. Knowledge tracking experiments show that the AUC and R2 indexes of the DT-BKT model in this paper are better than those of other comparative models, and it can better simulate the response of learners on the dataset. Facing different groups of learners is able to recommend Civics courses that meet the learners’ abilities. For active learners and potential learners, the average difficulty of the recommended client layer is higher by 0.08~0.15 and 0.06~0.085 respectively, while the overall difficulty difference for inactive learners is between -0.01~0.015, and the recommended difficulty is in line with the characteristics of the learner groups.

In this paper, a K⁃Means clustering algorithm based on improved differential evolution (AGDE⁃KM) is proposed to design the adaptive operation operator, design the multi-variation strategy and introduce the weight coefficients in the variation stage to regulate the searching ability of the algorithm and accelerate its convergence speed. The Gaussian perturbation crossover operation based on the best individual of the current population is introduced, and the optimal solution output from the improved differential evolution algorithm is used as the clustering center to realize the cluster analysis of students’ sports performance data. Afterwards, the hierarchical recognition algorithm and support vector machine are used to recognize students’ sports patterns, and the wavelet transform algorithm is used to extract and select the students’ sports feature quantities, so as to improve the accuracy of students’ sports pattern recognition in sports teaching. In the process of physical education teaching, AGDE ⁃ KM algorithm is more pertinent to the clustering effect of students’ sports performance, and its explanatory degrees of Calinski-harabasz metrics, profile coefficients, and Dunn metrics are 860.0276, 0.3928, and 0.0486, which are 19.0382, 0.0435, and 0.0099. In addition, the AGDE⁃KM algorithm achieves 95.7625%, 99.75%, and 99.85% of the mean value of step recognition accuracy for different testers in the 50m, 800m, and 1000m events, respectively, which is a good recognition effect.

Music conductors rely on the visual impact of gestures and emotions for the interpretation and expression of musical works. In this paper, we utilize spatio-temporal two-stream convolutional neural network and replace the original VGG-16 network with ResNet-34 network with deeper network structure to construct a conductor recognition model for improving music conductor level. The Dropou optimization is applied in the fully connected layer to reduce the overfitting phenomenon, and the network structure is designed to fuse the temporal and spatial networks in advance with the feature maps, in view of the defects that the network structure of dual-stream convolutional neural network is shallow and the temporal and spatial networks do not learn the temporal and spatial information correlation. After the construction is completed, the model is applied in the teaching of a music college. The spatio-temporal information fusion convolutional neural network proposed in this paper is compared with other existing methods, and it is found that the optimized design helps the convolutional neural network to learn better, and better emotion and action effects can be obtained. It has better recognition accuracy on the dataset and obtained the highest accuracy of 74.3% on the CoST dataset. The results of the dimensions of music perception ability of the conductor students in the experimental class are better than the reference class, and the dimensions of pitch and intensity are more than 20% ahead of the control class, which proves that the model in this paper is more powerful to promote the development of music perception of the conductor students.

In the context of rapid research and development of unmanned equipment products, how can we better design an environment sensing system suitable for unmanned equipment combat missions and combat tasks from the perspective of actual combat has become an important research topic. This paper explores the optimization scheme of unmanned equipment environment sensing system based on blockchain technology, proposes PBFT (DTPBFT) consensus algorithm based on C4.5 decision tree optimization, and combines with the full homomorphic encryption algorithm to put forward the shared data encryption scheme of unmanned equipment environment sensing system. The experimental results show that the classification accuracy of C4.5 decision tree is as high as 94.37%, which is better than other classification algorithms, indicating that the use of C4.5 decision tree can effectively improve the accuracy of the classification of the consensus nodes and the security of the PBFT algorithm. In the case of the same number of nodes, the throughput size of the DTPBRT algorithm proposed in this paper is always higher than that of the PBFT algorithm, and the consensus latency is higher than that of the PBFT algorithm only when there are Byzantine nodes inside the system, but the DTPBRT algorithm is able to effectively remove the Byzantine nodes inside the system, which verifies the superiority of this paper’s algorithm. Comprehensive encryption and decryption time-consuming and throughput data, this paper’s scheme in general can realize high data sharing efficiency and ensure the security of data sharing, which can provide technical support for the data security of unmanned equipment environment sensing system.

Teaching optimization algorithm is a new type of group intelligence algorithm, which simulates the teaching process of teachers, and this paper improves the algorithm to realize the improvement of music teachers’ teaching ability. Aiming at the shortcomings of the teaching optimization algorithm which is easy to mature prematurely, has low solution accuracy and converges to the local optimum, this paper proposes a teaching optimization algorithm which integrates the improved Tennessee whisker search. The algorithm combines Tent mapping and inverse learning strategy to initialize the population and improve the quality of the initial population. Tennessee whisker search is performed on teachers to improve their teaching ability. Incorporating the hybrid variation operator into the individual student variation formula allows the algorithm to quickly jump out of the local optimum dilemma. The experimental results show that the hybrid teaching optimization algorithm based on BASTLBO proposed in this paper has good solution accuracy and robustness in finding the optimum on different types of optimization problems. The algorithm in this paper can achieve better teaching ability results than the unimproved TLBO algorithm and the teaching optimization algorithm incorporating the hippocampus strategy, and the objective function on two different indexes is reduced by 8.75% and 7% compared with that of the TLBO algorithm, respectively, and the hybrid teaching multi-objective optimization model designed in this paper has stronger practicality.

Accurately capturing the behavioral factors of different types of customer groups and adopting targeted service strategies is the key to business competition in the hotel industry. In this paper, we combine the variance Boston matrix and PSO-based K-means algorithm to achieve hotel customer attribute segmentation based on customer behavior, customer value and word-of-mouth reliability, and then use deep learning algorithms to construct a hotel customer behavior prediction model. The feature fusion layer and SENet are incorporated into the residual network in order to utilize the feature expression ability of different layers and the spatial coding ability between different channels to enhance the hotel customer behavior predictive ability. Downloading the public dataset from the online wine travel platform for example analysis, it is found that the classification of this paper’s algorithm before customer segmentation has a correct rate of 83.75%, which is higher than the rest of the baseline models. After customer segmentation this paper’s algorithm achieves the highest recall rate in all customer categories, and the recall rate is as high as 84% on category 1 customer groups, and the superiority of the designed algorithm is verified. This study facilitates hotel management to target customer service and retention according to different customer groups.

In order to improve the consistency of on-chain-off-chain interaction of private data supported by blockchain and reduce the redundancy of data storage performance, this paper applies an efficient data interaction method of prefix hashing with improved red-black tree index to store public indexes and improve the efficiency of retrieval and interaction of blockchain data. Under the idea of generalization, anonymous region (AR) is used to hide the real location of participating nodes and protect the privacy of realized nodes. To reduce the computational overhead of the selection process, a cooperative sensing location privacy preserving optimization mechanism, LPPOM, is proposed. The scheme in this paper has a slow growth of data size on the chain with higher storage efficiency, larger throughput, and shorter query time (0.1899ms). The time cost consumed when the number of privacy chains is 15, 30, and 60 only increases by 0.2309-0.4855ms compared to the single chain system, indicating that the scheme scales well. When the file size is within 200 and the number of encrypted attributes is less than 4, its total encryption time meets the user’s privacy data encryption needs (between 66.1765-236.7081ms). The IPFS read/write module is able to satisfy the people’s daily use needs under the public network conditions, and its read/write speed is between 0.1568 and 0.2639MB/ms (file <100M).