Physical education teaching resources are an important part of teaching resources, and it is necessary to adopt a sustainable development approach to ensure the rational utilization of resources. In this paper, firstly, the factors affecting the allocation of physical education teaching resources in colleges and universities are analyzed by using principal component analysis and systematic cluster analysis, and the validity of the method is verified. Secondly, it constructs the influential element model of regional physical education teaching resources allocation efficiency level based on Tobit regression, and explores the locational factors affecting the distribution of physical education teaching resources. Finally, relevant countermeasure suggestions were put forward based on the analysis results. Using principal component analysis to downscale the 17 indicators of the influencing elements of physical education teaching resource allocation in the statistical data, four principal components were obtained, whose cumulative contribution rate was as high as 90.22%, which was greater than 85%, i.e., it had a 90.22% degree of explanation for the original data. Then, the dimensionality-decreased data were clustered and realized to evaluate and rank the allocation of physical education teaching resources in 23 sample universities. In addition, the results of Tobit multiple regression analysis showed that factors such as regional geographic location, regional population density, regional economic development and the scale of investment in physical education teaching resources all have different degrees of influence on the allocation efficiency of regional physical education teaching resources.

Existing translation teaching content has certain deficiencies, this paper discusses the computational methods to optimize the translation teaching content by combining the semantic association network model. A domain translation model with joint semantic information is proposed, which constructs a bilingual mapping relation of domain-specific word vectors to obtain the semantic k-nearest neighbors of words in a specific domain,so as to estimate the domain intertranslation degree of words and improve the adaptive ability of the domain translation model. Then a semantic similarity computation model (SRoberta-SelfAtt) incorporating Robert’s pre-training model is proposed. The model incorporates a self-attention mechanism to extract the association of different words within the text, and acquires richer sentence vector information. The proposed domain translation model is able to obtain more accurate translation results while spending less time. Compared with the stability of the iterative process of the basic model, the SRoberta-SelfAtt model has higher iterative stability. The Roberta-based semantic similarity computation model can effectively improve the performance of the word vector model. The experimental results show that the domain translation model with joint semantic information and the SRoberta-SelfAtt model are more practical for the task of optimizing translation teaching content.

Promoting the output and transformation of scientific and technological achievements of higher vocational colleges and universities is not only the topic of promoting the high-quality development of education in higher vocational colleges and universities, but also the way to deeply implement the innovation-driven development strategy. Taking higher vocational colleges and universities in four municipalities directly under the central government as research samples, this study first utilizes the DEA model to measure the transformation efficiency of scientific and technological achievements of higher vocational colleges and universities in four municipalities directly under the central government in the period of 2014-2023, and combines with the literature analysis method to dig out the key influencing factors of their transformation energy efficiency. Then, the fuzzy set qualitative comparative analysis method (fsQCA) is used to carry out empirical research on the transformation efficiency due to inputs and outputs of scientific and technological achievements of the studied higher education institutions and the interactions between their influencing factors, so as to analyze the grouping path of the improvement of the energy efficiency of the transformation of scientific and technological achievements of the higher vocational colleges and universities. In the analysis of the results of measuring the efficiency of the transformation stage of scientific and technological achievements, the efficiency of the transformation stage of scientific and technological achievements of local higher vocational colleges and universities in D city is generally at a high level, with an average value of 0.427. Meanwhile, regional development factors (consistency 0.9081>0.9) and policy factors (consistency 0.9322>0.9) are the necessary conditions for the efficient transformation of scientific and technological achievements of higher vocational colleges and universities, and they are the key influences to improve the energy efficiency of scientific and technological achievements transformation.

Shaanxi folk women’s red has beautiful graphic patterns, which is a treasure of Chinese folk culture. In order to better realize the inheritance and innovation of folk women’s red, this paper refers to the idea of multi-objective optimization, and innovatively designs the composition of ornaments through genetic algorithm and bipartite continuous pattern design method. In order to find out the deep meaning and cultural value of Shaanxi needlework decoration and the unique aesthetic, emotional and life experience of women hidden behind the decoration. In addition, further research on Shaanxi needlework decoration art through multi-objective optimization will not only help to deeply understand the common characteristics of national art, but also help to deeply understand the characteristics of folk art itself. The research shows that the composition scheme designed in this paper has been positively evaluated by experts and consumers, and can promote the inheritance and innovation of Shaanxi folk needlework.

In the long-term teaching practice, various disciplines have accumulated a large number of teaching resources but cannot function fully and efficiently. For this reason, this study constructs a knowledge mapping of college disciplines based on deep learning. First of all, the overall construction of the atlas is planned, the core concepts of the discipline are identified, the relationships between the knowledge points are defined, and the resources corresponding to the knowledge entities and attributes are expanded. Then deep learning is utilized for the entity construction of the subject knowledge graph, the neural network models BiLSTM+CRF and BiLSTM+Attention are used for the subject entity identification and relationship extraction, and finally the subject knowledge fusion and storage is carried out, and the effectiveness of the designed algorithms is verified on the dataset. The data show that the knowledge representation of knowledge graph is conducive to demonstrating the logical meaning between learning materials, facilitating learners to correlate what they have learned previously with what they are learning now, fusing old and new knowledge, and facilitating learners to meaningfully construct knowledge.

In order to realize the intelligent operation and maintenance of electrochemical energy storage power station and make the working process of the power station battery more efficient, stable and safe, this paper establishes a safety monitoring system of electrochemical energy storage power station through multimodal fusion sensing technology. The multi-sensor fusion technology and multi-sensor calibration process are proposed, and the Kalman joint filter fusion algorithm is obtained based on the traditional Kalman filter extension, which fuses the collected multi-modal sensing data to realize the real-time detection of the state information of each battery of the energy storage power station. Simulation experiments are carried out to verify the reliability of the Kalman joint filter fusion algorithm, and the deviation value of this algorithm in the filter fusion processing is only 0.1426, which is lower than that of the comparative sliding average filtering algorithm. The RMSE values of X-axis and Y-axis in the motion target tracking experiments are less than those of the comparative mean drift algorithm 0.189 and 0.1412, and in the speed, they are less than those of 0.0062 and 0.0073, which are better in terms of accuracy performance. And in the application practice of battery safety monitoring system for electrochemical energy storage power station, the error between SOC estimation and actual value is less than 5% in either DST condition or UDDS condition, and the internal resistance 0R change curve is similar to the actual value of the internal resistance, and the estimation error is less than 4%.

The article constructs binocular vision 3D image structure by feature extraction and data acquisition of animated images, setting the base modeling points multi-level, establishing texture mapping modeling relationship, then designing key frame interpolation algorithms such as segmented cubic spline interpolation and quaternionic spherical linear interpolation, and applying geometric algebra to 3D animation modeling, and using a conformal geometric algebra approach to describe the 3D model as well as the dynamic model. Calculation results. The 3D animation modeling using the method of this paper reduces the error of 36.8mm compared with the same type of method, so the effect of using the method of this paper is better than 1other algorithms in 3D human body modeling. In the subjective evaluation of the visual effect of 3D animation video, 19 people think that the video has a strong sense of spatial three-dimensionality, and on the whole, the majority of people think that the animation video developed using the method of this paper is clear, realistic, has a sense of spatial three-dimensionality, smooth movement of the object, and the use of the lens is comfortable, which has a better visual communication effect.

Human-computer interaction scenarios have a broad prospect in the field of English learning. In this paper, a human-computer dialogue interaction system for English learning scenarios is designed based on deep reinforcement learning and artificial intelligence interaction technology. Firstly, a speech enhancement method based on collaborative recurrent network is proposed to optimize the speech analysis module. On this basis, we design the framework of human-computer interaction system, and construct a human-computer dialogue interaction system for English learning scenarios that contains three modules: natural language understanding (NLU), knowledge retrieval enhancement, and natural language generation (NLG), in which knowledge retrieval enhancement utilizes ChatGPT for document reordering design. In the speech enhancement simulation experiments, the mean value of network congestion for the speech enhancement method designed in this paper is 0.073, which achieves at least 50% performance improvement, reduces speech distortion and optimizes the signal-to-noise ratio at the same time. The system is experimentally analyzed for two tasks, conversation state tracking and conversation reply generation, and outperforms the baseline model on both tasks. Finally, a subjective evaluation is conducted, and the system in this paper scores 3.766, which is obviously a smoother human-computer interaction experience, and the English learning interaction experience has a greater advantage compared with the other methods. This paper provides innovative ideas and feasible methods for combining cutting-edge information technology with interactive English teaching.

Based on the relevant theoretical basis and research experience, this paper constructs a three-in-one, subjective and objective evaluation index system of inclusive preschool public service quality of “flexible”, “green” and “soft” quality. Subsequently, the GA-BP neural network quality assessment model based on machine learning algorithm was constructed by utilizing BP neural network analysis and hierarchical analysis to assign weights to the indicators. It was applied in a scientific operation process to synthesize subjective and objective data to understand the quality of public services of inclusive preschool education, and to propose an improvement path in combination with the IPA analysis model. The results show that the weights of the three first-level indicators are 0.428, 0.4231 and 0.1489, respectively, and the weights of the four second-level indicators, including special child care and migrant child care, are more than 0.1, while the weights of the other second-level indicators are all less than 0.1. Among the third-level indicators, the weights of the reasonable sharing of the cost of pre-school education among the government, families, kindergartens and the society, and the synergy of education and rehabilitation are more than 0.05, while the weights of the other third-level indicators are all less than 0.5, and the weights of the other third-level indicators are more than 0.1. In addition, the difference between the actual evaluation results and the simulation evaluation results of the teaching quality of universal preschool education public services is relatively small. And the error between the real values of GA-BP model is extremely small, and its average error is only 0.483.

Smart grid technology is developing rapidly around the world and is gradually applied to the operation and maintenance management of power systems, and its main advantage lies in its integration capability, which can effectively realize the high efficiency, security and reliability of power system operation and maintenance. This paper explores the integration of grid operation and maintenance by integrating computing and information theory using multidimensional data mining and analysis methods. The operation data of smart grid is first preprocessed, including resampling and PCA dimensionality reduction of multidimensional data signals. Then, a CNN-based power operation state prediction model and an R-CNN-based grid fault diagnosis model are constructed to ensure the stable operation and timely maintenance of the smart grid, and the predicted and actual values of the smart grid operation state of the CNN model are basically consistent with each other, with the MAE, MSE, and RMSE of 0.00104, 0.00014, and 0.012, respectively, and the prediction results are good. The effect is good. Compared with CNN and SVM, the performance of R-GNN model is better, and after PCA dimensionality reduction, the fault identification rate of R-GNN model is as high as 98.91%. And the delay of the R-GNN method for fault diagnosis is only 0.04s, while it can realize the comprehensive and accurate localization of the fault area. This paper provides methodological reference for the utilization of multidimensional data mining and analysis technology to realize the operation and maintenance integration of smart grid.