How to form a personalized shortest learning path for vocal skills based on learners’ individual characteristics is the key to improve the efficiency of vocal music teaching. In this paper, on the basis of dynamic key-value memory network, a gating mechanism is used to update students’ knowledge mastery status, and a knowledge tracking model based on dynamic key-value gated recurrent network is proposed to realize the accurate assessment of students’ vocal music level. On this basis, after searching the suboptimal path using the particle swarm algorithm, the shortest path is searched using the ant colony algorithm, which solves the shortcoming of the blindness of the initial search direction of the single ant colony algorithm, and constructs a recommendation model for optimization of the learning path of vocal skills. The results of simulation experiments show that the model AUC and ACC on the ASSIST2015 dataset are 0.7468 and 0.7654, respectively, which are much higher than the highest 0.7281 and 0.7528 in the baseline model. Path optimization was achieved for both ordinary and excellent vocal students, and the average optimization was 4.297 and 3.242 on ASSIST2009, and 3.819 and 3.044 on ASSIST2015.This paper makes an innovative exploration to improve the quality of vocal music teaching.

In the era of artificial intelligence, the technology of speech conversion has developed rapidly and has gradually become a hot topic of research in the field of speech processing. This paper explores the problem of speech signal extraction and generation based on Wave RNN model, and constructs a speech conversion generation model driven by artificial intelligence. First, the short-time Fourier transform is utilized to convert and preprocess the speech signal in the time-frequency domain. Second, a stepwise speech enhancement model is proposed to enhance the perceived strength of the speech signal. Then, a speech generation model based on improved self-attention mechanism and RNN is designed to realize the generation of speech signals. Finally, the model effect is evaluated for application. The time-frequency domain feature that mixes time-domain features and frequencydomain features is able to capture the characteristics of speech signals more comprehensively than a single time-domain feature and frequency-domain feature, which corresponds to a higher recognition accuracy and a lower training loss value. Meanwhile, after speech enhancement, the average accuracy of model A~D speech recognition is improved by 19%~25%, which indicates that the stepped speech enhancement model used in this paper can substantially enhance the perceptual strength of speech signals. In addition, the language conversion model in this paper outperforms other speech conversion models in both MCD and RMSE, and its advantage in rhyme mapping is obvious, and the pitch of the output speech is more accurate and natural. The model in this paper has high practical value in speech signal generation and conversion.

As the father of musical instruments, the piano is commonly used in solo, repertoire, accompaniment and other performance processes. In the process of piano playing, the quality of its sound is closely related to the playing skills. The article analyzes the structural composition of the piano as well as the physical mechanism of sound generation, and summarizes the characteristics of the four elements of piano music, namely pitch, intensity, timbre and duration, on the mathematical basis of the twelve equal temperament laws and the vibration equations of the strings. Subsequently, we analyze the time and frequency domain characteristics of the piano’s musical technique evolution, and calculate the main physical parameters that can affect the piano timbre. Finally, based on the theoretical study and characterization, the corresponding result evaluation experiments were conducted. It is concluded that by analyzing the root-mean-square and mean values of the vibration time-domain signals of piano soundboards excited at different points, it can be seen that, for different structures of piano soundboards, there are excitation points that can maximize their vibration signals. At the same time, the time-domain characteristic index crag factor is analyzed, and it is found that there is no obvious pattern in the crag factor value of the vibration signal of the soundboard with different point excitations.

The aim of this study is to develop a near-infrared photothermally controlled nano-retarded release system loaded with the anticancer drug Adriamycin optimized based on numerical simulation calculations. Firstly, the instruments, agents and experimental methods for the preparation of selfassembled albumin-loaded nanoparticles were introduced.The cumulative absorption wavelengths of the albumin nanoparticles were investigated by UV and IR spectroscopy, and it was found that the maximal absorption wavelengths of DOX and BDC were distributed at 487 nm and 435 nm, and that the UV maximal absorption wavelength of CUR was 435 nm.In the in vitro slow-release performance, it was found that the cumulative release rate of DOX reached 97.36% when pH 5.0 was used, and that when CUR was used, the cumulative release rate of DOX reached 97.36%. The cumulative release rate of DOX reached 97.36% at pH 5.0, while it was only 59.15% and 30.81% at pH 6.0 and 7.0. The cumulative release rates of CUR at the three pH values were 58.69%, 29.98% and 16.81%, respectively, which were basically the same trend of the retardation curves of the two drugs. The nanoparticles degraded morphology showed the widest and narrowest particle size distribution in PBS buffer solution at pH=5.0 and 7.0, respectively. The loading capacity of the optimized model showed good consistency of effect on measured (11.03%) and predicted (10.87%) values.The photothermal conversion experiments of DOX nanoliposomes were found to have concentration and time dependent photothermal conversion effects. In this paper, from the optical characterization of albumin drugcarrying nanoparticles, it was found that UV light was able to excite PFNSNO for photodynamic therapy as well as NO release through the fluorescence resonance energy transfer process.

Giving full play to the vitality and autonomy of inter-governmental departments can improve the national governance system and enhance the modernized governance capacity. This paper conducts a relevant research on the coordination mechanism between multiple government departments. According to the network analysis method, the relationships of departments in different service items and fields are studied, and the causes of the problems of power distribution and coordination mechanism in the network of departmental relationships are explained, and the analysis of the influence mechanism of departmental coordination is completed by using the random forest algorithm. The analysis results show that the power of government departments in the fields of housing security, social insurance, labor, employment and entrepreneurship, public education, and health care is more concentrated, and the Ministry of Civil Affairs and the Ministry of Human Resources and Social Security have an active position in the coordination process. Cultural cognitive bias, imbalance of power and responsibility, lack of coordination system guarantee and insufficient support of coordination environment are the causes of problems in the coordination mechanism. In addition, ambiguous coordination responsibilities, imperfect institutionalized coordination and lack of supervision system are important factors affecting the multisectoral coordination mechanism.

Rural tourism, as an important part of the tourism service industry, the study of the spatio-temporal evolution and influence mechanism of rural tourism flow has also become a hot topic at present. This paper takes Jiangsu Province of China as the research area, proposes the heat measurement and identification method of rural tourism based on network data, constructs a heat measurement model, takes standard deviation ellipse analysis, average nearest neighbor index method, kernel density analysis as the core method of spatial analysis, and proposes the hotspot identification method on the demand of spatial relevance analysis, so as to provide the method and means for the analysis of the spatio-temporal evolution of the rural tourism flow. In the analysis of the influence mechanism of rural tourism flow, the QAP model is used as a research tool to explore the influencing factors of rural tourism flow.The value of rural tourism hotness was low during 2014-2017, and it has rapidly increased and maintained a high growth trend since 2018. The Gini coefficient of rural tourism hotness increased from 0.51 in 2014 to 0.72 in 2018, and then fell back to 0.65 in 2023, and the degree of spatial difference of rural tourism hotness showed a weakening trend, and the hotspot areas of rural tourism were increasing. The structure of tourism flow is affected by a variety of factors such as spatial proximity, tourism income, and the impacts produced by the factors change somewhat in different time periods.

Due to the complex structure of multi-dimensional anthropomorphic wind turbine and the harsh operating environment, in order to reduce its maintenance cost, it has become a popular research hotspot to get fast and effective condition diagnosis and fault early warning through big data mining and analysis of wind turbine condition monitoring. The article clarifies the basic mechanism and typical faults of multi-dimensional anthropomorphic wind turbine, and after analyzing the characteristic frequency of faults on the transmission chain of multi-dimensional anthropomorphic wind turbine, it proposes the anomaly detection method of wind turbine condition monitoring data based on the auxiliary eigenvectors improved density clustering (DBSCAN), which realizes the accurate identification of different types of normal data, valid anomalous data containing fault information, and invalid anomalous data in the monitoring data. It realizes the accurate identification of different types of normal data, valid abnormal data with fault information, and invalid abnormal data in monitoring data. Subsequently, the actual historical data of the wind farm is used as the experimental data set to realize the identification of the operating status of the wind turbine. Finally, the DBN-Dropout wind turbine fault identification method is proposed by combining Deep Confidence Network and Dropout technique. The experimental results indicate that the recognition rate of this paper’s model for nine faults is as high as 99.88%, and the superiority and accuracy of this paper’s model in feature extraction and fault diagnosis are verified by comparing its performance with other fault detection models.

In this paper, a personalized scheme recommendation method for dance movements based on ontological similarity is proposed. An ontology model of trainers is established, and in order to explore the interactions between trainers’ attribute features and their influence on core parameters, SWRL rules are established using Jena inference engine for the inference of core parameters of training programs. The similarity degree is calculated according to the different types of user variables respectively, and the artificial neural network model is used to determine the degree of similarity between different trainers, in order to complete the recommendation of personalized training programs for dance movements. And then the requirements of the system are summarized to achieve the framework construction of the personalized dance movement training program recommendation system to achieve the health management in the training process. The recommendation effects presented by the similarity calculation method of this paper have reached the design goal of this paper, and the personalized recommendation system of this paper has also significantly improved the physical fitness level and the performance effect of dance skills of the experimental group of dance trainees, and the success rate of the kicking back leg movement has reached 91.67%. However, the system’s function of improving health knowledge and health awareness needs to be further upgraded.

In this paper, the autoregressive moving average model (ARMA) and LSTM deep neural network are first introduced, and the time series are decomposed into high volatility components and low volatility components by MA filtering method. Then the time series forecasting model ARMA and deep neural network LSTM are combined on the basis of MA filtering method to form ARMA-LSTM combination model based on MA filtering method, and the application effect of this model in financial market volatility forecasting and risk response is verified through empirical evidence. The results show that the ARMA-LSTM_t model will achieve relatively good results in predicting the GDP_IG of the current year using the data of the 12 months of the current year and the last month of the previous year, and the training and prediction sets of the ARMA-LSTM combination model proposed in this paper have the best results. In addition, there is a positive relationship between investment-related indicators and GDP_IG, and the addition of investment network search data improves the estimation accuracy of the model, obtains smaller prediction errors, and improves the prediction accuracy of the ARMA-LSTM model in the short and medium term.

In the context of urban elderly human resource development, differential evolutionary algorithms can be used to optimize the development strategy and improve the efficiency of resource utilization. The study constructs a multi-objective scheduling optimization model for human resources based on an improved differential evolutionary algorithm, which searches for the optimal development strategy by simulating the mutation, crossover, and selection operations in the process of biological evolution. In addition, the model combines a multi-objective feature selection algorithm to capture the data information of urban elderly resource development more accurately and ensure the scientific and practicality of the strategy. The pareto front of this paper’s algorithm on the optimal solution test function is more in line with the real frontier, and the GD value is between 0.00171 and 0.0325, which has better convergence. The execution time of this algorithm for elderly manpower resource scheduling is shortened compared to the comparison algorithm, and the convergence of different task sizes is accomplished when iterating to 110~150 rounds. The ADE-MOFS algorithm has the lowest running cost and the shortest completion period on elderly manpower resource scheduling. The research in this paper shows new ideas and methods for the rational development and utilization of urban elderly manpower resources, which has important theoretical and temporal significance.