Corporate ESG disclosure quality is a key condition to optimize industrial structure and a realistic path to reach sustainability performance. Based on the theoretical knowledge of Bayesian network model, the research program of corporate ESG disclosure quality and sustainability performance influence path is designed. According to the current status of enterprise development, 11 research variables are set, which contain explanatory variables, interpreted variables, and control variables. Mathematical statistics and Bayesian network modeling are adopted to parse the mutual influence mechanism between the two. In the forward Bayesian inference, the probability of enterprise sustainability performance being in a good state is 49.3%, and the probability of the explanatory variables being in a good state is increased to 58.7% by changing the state probability of other variables. In order to provide a comprehensive overview of the relationship, backward Bayesian inference was also performed, and when the probability of sustainability performance being in a good state was 100%, the probability of the board concurrent position being in a good state was the highest with a value of 72.3%. This study enhances the most effective corporate ESG disclosure quality control program for companies to maximize the possibility of sustainability performance.

In this paper, with the help of the real-time state observation property of the Kalman ϐilter method, we propose to use the Kalman ϐilter method for channel estimation of OFDM wireless communication system. The linear interpolation method is used to deal with the fading process of data symbol positions, and the Kalman ϐilter estimation expression of the fading process is obtained. And considering the computational complexity of the channel estimation algorithm, the channel estimation is optimized by adding the 1st order AR model into the channel model. The Doppler frequency is used as the simulation parameter to analyze the operational performance of the Kalman ϐilter channel estimation method under different Doppler frequencies. To further broaden the applicability of the proposed method in this paper, a MIMO-OFDM system is introduced, and numerical simulations are conducted to analyze the relationship curves between the outage probability and the SNR performance under the OFDM channel processing module for both the random channel and the random channel with OFDM modulation. In the massive MIMO multipath random transmission channel, the better the SNR performance of the channel, the smaller the probability of generating interruptions. Meanwhile, in the presence of the same non-ideal factors (hardware impairments, interference noise) interruption probability impairments of the channel, the SNR in OFDM-ideal state is about 10 dB more than the OFDM-hardware impairments simulation value.

This study analyzes the aerodynamics of fluttering flight of birds through their body structure characteristics. A convolutional neural network is combined with a bird-like flight aerodynamic model. By analyzing the symmetric and asymmetric motion laws of birds in flight, the three-dimensional model and equations of motion of the wing-fluttering motion are established, the aerodynamic simulation study of bird wing-fluttering flight under Computational Fluid Dynamics(CFD) and train it by convolutional neural network. When the model trained to 12 rounds, the loss values on both the training and validation sets converge to about 3.5%, the training effect is good. The predicted values of the lift-to-drag ratio by the model in this paper are close to the CFD calculated values, and the average relative errors of the validation set test set are 0.483% and 0.486%, respectively. In addition, the model predicts the pressure coefficient of the flow field better, and the prediction error of the vast majority of the positions is less than 1.2%. In conclusion, the convolutional neural network can significantly improve the performance of bird flight aerodynamic simulation model.

The environment near substations is complex, and electrocution accidents of operators occur from time to time during on-site operations, and the development of safety detection models for substation operations has received more and more attention. The article proposes a safety distance detection model for substation operation, which is mainly composed of binocular stereo matching perception model and safe area detection model. The binocular stereo matching perception is based on the PSMNet network model, combined with the parallax regression calculation to obtain the threedimensional coordinates of the operation area in the process of substation operation, and the threedimensional reconstruction of the substation operation process. The spatial context inference algorithm is utilized in the safe region detection model to detect the edge of the safe region, and the image segmentation of the safe region of the substation operation scene is performed by the improved OTSU algorithm. Then the three-dimensional coordinates obtained from binocular stereo matching perception and the three-dimensional coordinates of safe region detection are solved for the Euclidean distance, and then the safe distance detection of substation operation is realized. The EPE result accuracy of binocular stereo perception matching on the dataset is reduced by 0.71px compared with CRL, and the resulting mismatch pixel rate is between 0.83 and 1.48%. The average time-consuming image segmentation of the improved OTSU threshold segmentation method is 6.34ms, and the average relative error of the safety distance detection for substation operation is only 0.85%, and the maximum absolute error of the safety distance detection is only 0.13 m. Combining the spatial contextual reasoning algorithm with the deep learning technology can realize the effective detection of the safety distance for substation operation in multiple scenarios, and fully ensure the operation of the substation workers’ safety.

At present, digital twin technology has been developed in many fields and plays a very important role. In this study, digital twin technology is applied to remote control of power system to build a set of remote control system of power system, which contains perception layer, data layer, operation layer, function layer and application layer. In order to make the power system remote control system more reliable and effective, a power system fault diagnosis method based on MRPSODE-ELM is proposed using deep learning technology. The method combines PSO algorithm and DE algorithm to construct a multiple stochastic variation particle swarm differential evolution algorithm, and it is used for the optimization seeking ability of the number of neurons in the hidden layer of the limit learning machine. The experimental results show that the MRPSODE-ELM model performs superiorly in detecting different fault types in terms of accuracy, recall and F1 score, with the results of each index above 95%, and the fault diagnosis accuracy is improved by 4.77% and 3.36% over SVM algorithm and DNN algorithm, respectively, and possesses a smaller training time consumption. The fault detection method proposed in the study can be applied to the remote control of power systems based on digital twins.

The study proposes a multi-stage dynamic resilient recovery strategy based on multiple energy storage to cope with distribution network failures after a disaster in a coastal city, and the post-disaster recovery of the urban distribution network is planned in phases, which is divided into the first stage of emergency response, the second stage of energy storage recovery and the third stage of economic optimization. Then the post-disaster defense measures of the coastal city are improved by optimizing the recovery strategy. After the calculation example design, the post-disaster recovery and resource scheduling effects of this paper’s multi-stage dynamic recovery model are examined through simulation experiments. The multi-stage dynamic recovery model of this paper takes 261 minutes to recover the urban distribution network, which is shorter than the 273 minutes of the traditional recovery model, and the post-disaster resilience is improved. The proposed dispatching scheme based on the multi-stage dynamic recovery model in this paper uses only 13 vehicles, which is the least number of vehicles among all dispatching schemes. The traveling path of mobile emergency resources of this paper’s scheme is most consistent with the post-disaster restoration scenario. The combined level of load reactive power and active power restoration of this paper’s scheme is optimal.

According to the decision-making process of power grid investment, this paper sets the objective function and constraints, realizes the construction of optimization model, and selects genetic algorithm as the solution algorithm of optimization model. Under the requirement of evaluation index principle, 16 secondary indexes and 4 primary indexes are screened, thus forming the evaluation index system of power grid project investment efficiency. The experimental conditions are set to evaluate and analyze the optimization of investment decision and multidimensional benefits of power grid project respectively. Along with the reduction of voltage data, the diversity of optimal solutions for grid project benefits begins to materialize, and the diversity of optimal solutions of GA algorithm is higher than that of PSO algorithm, indicating that the use of genetic algorithm to calculate optimal solutions for grid investment benefits is more effective. In addition, the closeness of the seven projects to the optimal solution is 0.4613, 0.5044, 0.4681, 0.5398, 0.6342, 0.5759, 0.4116, respectively, of which project 5 has the best investment benefit.

The rapid development of the electric power market makes the scientificity and rationality of grid investment decision-making particularly important. In this paper, firstly, we design a grid investment benefit assessment method based on fuzzy comprehensive evaluation. And taking the grid investment benefit of M city in 2022 as an example, the fuzzy comprehensive evaluation method is used to quantitatively evaluate the grid investment benefit. Based on the evaluation results, the weaknesses of power grid investment in M city are found. Then the multi-level optimization strategy of grid investment is further proposed to achieve the maximization of investment benefits. The strategy considers the objectives and constraints of different levels, such as grid structure, power supply reliability, operation efficiency, and power sales revenue, and coordinates the interests between all levels by establishing a multi-objective optimization model to achieve the global optimization of the grid investment decision. Finally, after adjusting the allocation ratio and the allocation amount by the multi-level optimization strategy, the overall evaluation of the city’s grid investment efficiency in 2023 is improved from “average” to “excellent”. It shows that the multilevel optimization strategy designed in this paper can provide scientific guidance for grid investment decision-making.

Aiming at the shortcomings of traditional relay protection, an adaptive multi-area protection coordination model is studied and designed. Firstly, combining different control strategies such as master-slave control and sag control, a method of AC/DC distribution network trend calculation and network loss analysis based on the alternating iteration method is proposed and realized to ensure that the adaptive relay protection can act correctly. The proposed method is analyzed for AC/DC hybrid distribution network trend calculation, and the alternating iteration solution method is used for trend analysis and calculation, and the effectiveness of the proposed method is veriϐied by two examples of AC/DC hybrid distribution networks. Then the adaptive Agent with reinforcement learning is introduced, and its constructed multi-agent system has more system adaptive capability. The adaptive current interruption protection is compared with the traditional current interruption protection, and its protection principle and protection scope are analyzed, on the basis of which an adaptive coordinated protection method based on MAS grid is proposed to realize the MAS adaptive current interruption protection, and its simulation is veriϐied. The experimental results show that the method of this paper can signiϐicantly improve the ϐlexibility, effectiveness and stability of AC and DC distribution network operation.

As a key link in international trade, the price volatility of container transportation has a profound impact on the global supply chain, and uncertainty shocks are one of the main causes of price volatility. With this topic, this paper measures the level of uncertainty at the policy level through the uncertainty index construction method, which lays the foundation for subsequent research. Dynamic correlation and impulse effect analyses of container transportation market prices under uncertainty shocks are conducted using DCC-GARCH and SVAR models. China’s economic policy uncertainty index showed four stages of significant increase in 2001, 2008, 2015 and 2019. The overall price volatility of the container transportation market shows an upward trend, and in 2023, the transportation price is 23,835 yuan. Container transportation prices are affected by the uncertainty of China’s economic policies as well as China’s trade policies, with correlation coefficients ranging from -0.69 to 0.60. The influence of China’s economic policy uncertainty index on container transportation price does not have a long time lag effect.