4results about How to "Improve regional adaptability" patented technology

This invention discloses an adaptive scheduling method, system, and medium for distribution networks under disturbances caused by renewable energy grid connection. It relates to the technical field of distribution networks and includes: real-time data collection from renewable energy grid connection points to obtain a multi-source sensing dataset for disturbance event identification, resulting in multiple disturbance events; dynamic partitioning of the distribution network based on disturbance type labels and spatiotemporal characteristics to determine multiple autonomous control areas; distributed collaborative optimization using a regional dynamic equivalent model to generate multiple local scheduling strategies; and integration of these local scheduling strategies for global coordination and verification to generate and execute adaptive scheduling instructions for the distribution network. This invention solves the technical problems of delayed response to renewable energy grid connection disturbances, poor efficiency of centralized control, and difficulty in balancing local autonomy and global coordination in existing technologies. It achieves the technical effects of improving the distribution network's rapid response capability to disturbance events, the regional adaptability of scheduling strategies, and the overall operational safety and reliability.

This invention discloses a method for calculating the installed capacity of distributed generation (DG) in a distribution area based on the time-series power of distribution transformers. The method includes: classifying the target distribution area by DG type; matching the DG type classification results with corresponding time period filtering principles; filtering the corresponding pure load time periods according to the time period filtering principles; obtaining pure load time period data; classifying the pure load time period data by date type; extracting the load cycle features corresponding to each date type; fitting the load cycle features using a pre-trained LSTM neural network model; outputting the load prediction value corresponding to the pure load time period; calculating the DG output at the corresponding time based on the load prediction value; filtering the maximum DG output value for the pure load time period; assigning dynamic coefficients to the maximum DG output value; and outputting the actual installed capacity of the target distribution area. This method improves the accuracy of DG installed capacity calculation, enhances data robustness, is applicable to a wider range of application scenarios, and is low-cost and easy to implement.

The application discloses a soil heavy metal pollution ecological risk identification method and system, relates to the technical field of soil pollution identification, and couples a module to construct a dynamic model of a physical field and fuse to obtain a coupling model, takes output of the coupling model as dynamic coupling data; an optimization module constructs a double-layer optimization framework of upper-layer space layout parameter screening and lower-layer strength judgment parameter optimization based on the coupling model output, adopts a probability agent model in iteration of the double-layer optimization framework, constructs a Pareto front solution set to screen candidate schemes based on multiple groups of risk identification schemes generated by iteration, and outputs a soil heavy metal pollution ecological risk dynamic identification atlas and hierarchical management and control suggestions in combination with a dynamic checking mechanism. The identification system can output the soil heavy metal pollution ecological risk dynamic identification atlas and the hierarchical management and control suggestions, not only realizes accurate positioning and hierarchical management of risks in space, but also achieves balance between ecological safety guarantee and economic cost control at a decision-making level.

The application provides a regional pest and disease collaborative identification method and system based on privacy federated learning. The method comprises: acquiring pest and disease observation data collected by each regional local node, and performing standardization processing and privacy classification on the pest and disease observation data in the local node; performing privacy protection processing on the local model update quantity and the local pest and disease phenotype prototype according to the privacy constraint parameter; dividing each local node according to the regional ecological domain representation, and generating a shared identification model, regional adaptation parameters and a cross-regional pest and disease prototype library; performing collaborative correction on the shared identification model, the regional adaptation parameters and the cross-regional pest and disease prototype library to obtain a target collaborative identification model; inputting current pest and disease observation data of a target region into the target collaborative identification model to generate a regional pest and disease collaborative identification result. The application can improve the regional adaptation capability, balance privacy protection and identification accuracy, and improve the collaborative identification credibility.