Based on the complexity and nonlinear characteristics of market volatility, this paper proposes a market volatility prediction model that combines MA filtering method, autoregressive moving average (ARMA), and long-short-term memory (LSTM) neural network. And the back-propagation (BP) neural network is utilized to quantitatively solve the problem of corporate strategy formulation, and a corporate strategy formation model is established to determine the corporate strategic choice through the corporate strategic environment and strategic capabilities. In the ablation experiment, the combined model MA-ARMA-LSTM reduces its MSE, RMSE, MAE and MAPE by 0.0007, 0.0131, 0.0074 and 1.57%, respectively, compared to the ARMA model. Compared with common market volatility prediction models, the combined model has the smallest error in each assessment index. The output of BP neural network for corporate strategy selection is consistent with the expert ranking, which is verified to be in line with the actual business situation, indicating that the method in this paper can provide a reasonable corporate strategy.

This paper constructs a multi-agent simulation model to study and prevent juvenile delinquency. A multi-agent reinforcement learning model is constructed according to reinforcement learning theory to simulate the behavioral decision-making process of minors in different social environments. By introducing the NashQ algorithm, it simulates the minors’ strategic choices when facing the temptation of crime. In the simulation experiments, the NashQ algorithm meets the convergence requirements of the model, and only 1/3 of the training times are needed to achieve the stability of the simulated environment. Among them, family factors, school factors and social factors all affect the stability of the prevention effect. Good family environment, high quality teaching conditions and healthy social atmosphere can effectively prevent juvenile delinquency.

From World War II to the Cold War (1945-1991), the U.S. military-industrial complex went through a process from its rise to its full expansion, which had a profound impact on the global political and economic landscape. In this paper, computer simulation techniques are used to construct a vector autoregressive model (VAR) to quantitatively analyze the impact of the military-industrial complex on the U.S. economy. Smoothness and cointegration treatment and Granger causality test are done on the collected sample data. After that, the VAR model between three sets of variables, namely, military expenditure as a share of GDP, consumption as a share of GDP, and investment as a share of GDP, is designed. Using impulse response function and variance decomposition to analyze the data, we get that the rise of the U.S. military-industrial complex can effectively promote the growth of the economy in the long term, and the development of the economy can also promote the development of the military-industrial complex, but the promotion effect is not obvious.

The existence of a large number of multi-source heterogeneous hosts and application service types in various zones of the power monitoring system leads to difficulties in extracting comprehensive host attack trace data and the problem of fine-grained deep threat detection. This study combines network attack traces extracted from multi-source logs and stores them in attack trace styles. An attack event description model based on key attributes and behavior sequences is constructed. Based on the vulnerability scoring system, an algorithm is designed to map a general attack graph into an absorbing Markov chain attack graph, which provides a computational basis for the analysis of network attacks by calculating the state transfer probability matrix of the attack graph. Finally, the performance of this paper’s method for multi-dimensional data feature extraction is explored in a python experimental simulation environment. The simulation results show that the average mapping time of LSTM model for 7 vulnerabilities is 117ms, while the average mapping time of this paper’s algorithm is improved by 37ms compared to the LSTM model.Meanwhile, the accuracy, stability, average false detection rate and positive and negative recall rate also achieve good results, which verifies the validity of this method in the practice of power monitoring system management.

The specialty of soil and water conservation and desertification control has gradually become a hot and popular discipline, and the educational practitioners of this specialty must also follow the trend and actively carry out educational reform. This paper utilizes genetic algorithm to conduct in-depth research on the problem of class scheduling, and on the basis of traditional genetic algorithm, an improved adaptive genetic algorithm is proposed to be applied to the class scheduling system. Through the adaptive adjustment of genetic parameters to improve the convergence accuracy of the genetic algorithm and accelerate the convergence speed, and finally after chromosome conflict detection and repeated iterative operations, the final optimal scheduling program is obtained. The improved adaptive genetic algorithm is applied in the course scheduling system of soil and water conservation planning and design in colleges and universities. After experimental verification, the improved new adaptive genetic algorithm, under the setting of different rules of scheduling conditions, under the setting of different rules of scheduling conditions, the fulfillment rate of students’ class selection reaches 100%, and the mean value of the overall rule fulfillment rate reaches 94.1%, and the overall fulfillment rate of the scheduling efficiency is improved to 96% by applying it to the intelligent class scheduling system. Finally, the professional classes were tested on the knowledge of soil and water conservation planning and design, and the remaining eight dimensions of professional knowledge were accompanied by questionnaires, and the achievement data of the test were statistically analyzed using SPSS22.0. The analysis results show that the test scores are quasi-normally distributed, and the actual pass rate of each question in the test paper is roughly close to the preset difficulty, which proves that the test paper is of good quality and the algorithm designed by the institute can basically meet the requirements.

In the context of information is mostly trivial, messy and disordered, under the context of information fragmentation, the creation path of new media art is also being affected by it. Based on the color sensual imagery, this paper adopts the gray correlation analysis method to research on the creation of new media art. Through the questionnaire survey, the cluster analysis algorithm is used to filter the color semantics, and the five most representative color imagery semantics are selected as the imagery scale in the quantitative space. Combined with the grey correlation analysis method to construct a new media art creation perceptual evaluation model, the new media art creation works as the object of color design practice, the constructed color design evaluation model well reached the product color scheme with the color screening, confirmation and evaluation of the preferred goals. The design practice based on the evaluation model of new media art creation. The results show that, combined with the gray correlation analysis, the color design evaluation model of new media art creation constructed under the intentional color system can effectively improve the color design efficiency of the work scheme, and give an intuitive and accurate reference standard for the selection of the color scheme of the work.

The development of digital technology has made the use of machine learning algorithms to protect cultural heritage has become a trend. In this paper, based on the random forest algorithm, the conservation model of tomb mural cultural heritage is recognized. The mural paintings in the tomb of Prince Zhanghuai are used as the data source to construct the tomb mural painting dataset, and the images in the dataset are processed, augmented and labeled. The features such as color, texture and shape in the mural images are extracted as one of the input information of the cultural heritage protection model of the tomb murals. Based on the random forest algorithm, a pattern recognition model for the protection of cultural heritage of tomb frescoes is constructed, and the feature vectors obtained from the feature extraction are used to calculate the split points of the decision tree. The classification results of multiple decision trees are weighted and averaged to obtain the final recognition results. The recognition accuracies of this paper’s model on the training set, test set and validation set are 99.45%, 95.46% and 92.58%, respectively. This is a significant improvement over other existing algorithms. Meanwhile, the algorithm consumes significantly less time than the ResNet18 deep residual network model before and after data enhancement, and is able to efficiently accomplish the task of recognizing the protection of cultural heritage of tomb chamber murals.

This paper discusses the application of virtual reality technology in enhancing college students’ selfefficacy and proposes an iterative optimization algorithm based on learning experience. By analyzing self-efficacy, the application of virtual reality technology machines in education, and combining relevant theories and empirical studies, the structural equation model of virtual reality technology influencing college students’ self-efficacy is constructed. The original structural equation model is optimized by using algorithms such as stochastic gradient descent method and stochastic average gradient, and the effectiveness of the algorithms is verified through experiments. This paper concludes that virtual reality technology can significantly improve college students’ self-efficacy, and the proposed iterative optimization algorithm can effectively improve the prediction accuracy and fit of the original structural equation model.

The rapid development of natural language processing technology makes machine translation play an increasingly important role in cross-lingual information exchange. In this paper, we propose an English long text translation paradigm based on the self-attention mechanism and introduce various improvement strategies to enhance the model performance. The model’s ability to process English long text is improved by introducing multi-head attention and hierarchical self-attention modules. The long text translation paradigm is optimized by using techniques such as residual linkage, layer normalization and dynamic memory network. A series of experiments are conducted to verify the effectiveness of the improved model on the English long text translation task. The English long text translation paradigm constructed in this paper outperforms the Transformer model and other related variants on both CPU and GPU. And Transformer outperforms this paper’s model in terms of n-gram accuracy in real translation experiments. The BLEU scores of the improved model on News and other datasets are significantly improved compared with the original baseline model, which verifies the effectiveness of the improvement strategy of this paper and provides a reference for the solution of the problem of English long text translation.

The lateral quantitative control of paper machine is the key to the quality control of paper machine. In this paper, the paper machine transverse quantitative control system is introduced, for the transverse quantitative control and the model exists of strong coupling, large time lag, multi-dimensional number of characteristics, combined with the predictive control theory, put forward based on the dynamic matrix control and Gram polynomials of the intelligent prediction method. Meanwhile, the speed chain control device of the paper machine control system is designed based on the self-immunity controller and simulated and analyzed. The simulation results show that the slope of the model size and computation in this paper’s method is 1.97, which is smaller than that of the traditional MPC’s 2.85, and has more computational efficiency without affecting the predictive control effect, which is more suitable for online operation. At the same time, the speed chain control system applying the self-resistant control algorithm is better than the traditional PID control in terms of steady state performance, dynamic performance and anti-disturbance performance. The method proposed in this paper facilitates the predictive control and speed chain anti-disturbance of the lateral dosing control system of paper machine and promotes the improvement of paper quality.