18results about How to "Realize intelligent optimization" patented technology

The application relates to the technical field of data analysis, and discloses a karst fissure identification-based open-pit mine blasting parameter optimization method and system. The method comprises the following steps: obtaining karst fissure distribution characteristic data through multi-source detection, establishing a three-dimensional fissure network topology structure, analyzing static breaking agent inflation stress propagation rules, performing multi-objective optimization solving based on stress propagation direction and intensity distribution to obtain an initial blasting parameter combination, and obtaining dynamic adjustment blasting parameters through real-time tracking and regulation according to fissure dynamic evolution in the blasting process. The application solves the problem that static blasting parameters cannot be real-time regulated according to the dynamic evolution state of the fissure network in the blasting process, resulting in deviation between the blasting effect and the expected target.

This application discloses a method, apparatus, and computer program product for processing outbound call services. Relating to the field of artificial intelligence, the method includes: executing an outbound call task to a user; processing the target service according to a preset service processing procedure; collecting the user's voice signal during the target service processing; extracting voice features and voice content from the voice signal; determining the current service processing node corresponding to the voice content in the preset service processing procedure; determining the service path map corresponding to the preset service processing procedure; determining a path deviation index value based on the current service processing node and the service path map; inputting the voice features into a target model to obtain the user's behavior recognition result; determining a risk index value for the user interrupting the target service based on the behavior recognition result and the path deviation index value; and adjusting the preset service processing procedure according to the risk index value. This application solves the problem of poor accuracy in recognizing user intent in outbound call services in related technologies.

The invention discloses a communication performance optimization method, electronic equipment and a computer readable storage medium, and belongs to the technical field of communication. The communication performance optimization method is applied to a dual-computer communication system in which a master processor and a slave processor are in communication connection based on a preset communication protocol. The communication performance optimization method comprises the following steps: dynamically obtaining to-do tasks in a task list; determining a task category and a cache space category of the to-be-handled task according to the urgency degree and the complexity of the to-be-handled task; determining an execution position and a task priority of the to-be-handled task according to the task category; and dynamically adjusting data in the cache space according to the cache space type. The working performance of the wearable device can be improved through the dual-machine system.

The invention provides an e-commerce commodity category adaptive recognition and semantic evaluation method, which comprises the following steps of: extracting path hops and associated edge semantic type labels between target commodities from a commodity relation graph, and extracting all shortest path sets through graph traversal to obtain a hop distance between the commodities and an edge type combination; performing multi-layer information propagation on the hop count distance between the commodities by adopting a multi-layer graph convolutional neural network through the information transmission loss value, and determining a commodity semantic embedding vector after propagation of each layer; according to the propagated commodity semantic embedding vector of each layer, the vector distance between commodity semantic embedding is extracted, the similarity is evaluated, and the optimized information propagation aggregation layer number is obtained based on the similarity convergence trend; and evaluating the coincidence degree of the commodity category inference result and the actual commodity category attribution, and identifying the category inference deviation to obtain the correction direction of the self-adaptive identification of the e-commerce commodity category.

The embodiment of the invention provides a link generation method, a reinforcement learning method, a data processing method, equipment and a storage medium. The method comprises the steps of obtaining a first task demand of a first data processing task; according to a first task demand, matching a plurality of required first operators from the plurality of alternative operators, and generating a first to-be-optimized data processing link based on the plurality of first operators; according to the first to-be-optimized data processing link, the performance estimation of the first to-be-optimized data processing link and the current operation states of the plurality of alternative operators, optimizing the first to-be-optimized data processing link by using a trained reinforcement learning model to obtain a first target data processing link; wherein the first target data processing link is used for processing the first data processing task.

The application discloses a multi-position camera control method and system based on cooperative scheduling optimization, and the method comprises the following steps: acquiring real-time images and device positions of multiple photographic devices in a shooting area; according to the real-time images, identifying shooting object data corresponding to the shooting area based on an object recognition algorithm; determining a scheduling path corresponding to each photographic device according to the shooting object data and the device positions; and determining a cooperative scheduling scheme corresponding to at least two photographic devices based on a dynamic programming algorithm according to the scheduling paths corresponding to all the photographic devices. It can be seen that the application can realize intelligent optimization of a multi-device cooperative shooting path, improve shooting coverage efficiency and cooperativeness, and reduce the risk of shooting quality decline caused by path conflicts or insufficient coverage.

The invention provides a cross-border Internet of Things equipment fault diagnosis and prediction system based on artificial intelligence, and the system comprises a data collection module which is used for collecting the operation data of cross-border Internet of Things equipment, and a data preprocessing module carries out the cleaning, standardization and feature extraction of the collected data, so as to facilitate the subsequent analysis and prediction. According to the method, machine learning and deep learning are utilized to more accurately identify the fault type of the equipment and predict the fault occurrence probability in the future; the efficiency of fault diagnosis and prediction is improved, data acquisition, preprocessing, diagnosis and prediction tasks are automatically completed, manual intervention is reduced, and the error rate is reduced; according to fault diagnosis and prediction results, operation parameters of the equipment are intelligently optimized, the fault occurrence probability is reduced, the service life of the equipment is prolonged, and the operation efficiency of the equipment is improved; fault diagnosis and prediction results and equipment optimization suggestions are displayed through a user interface, and a user is helped to better understand and master the running state of equipment.

This invention provides a backdoor defense method for federated learning based on gradient consistency and dual unsupervised clustering, comprising: obtaining gradient updates submitted by multiple clients participating in federated learning in the current training round; for each client, predicting the gradient update of the current round based on historical global model parameters and the client's historical gradient updates; calculating the difference between the actual gradient update and the predicted gradient update of each client to generate a malicious score for the client; adaptively determining the optimal number of clusters based on the malicious scores of all clients using the Gap statistic; performing dual unsupervised clustering detection with a coarse-to-fine approach based on the optimal number of clusters: first, removing clients with high malicious scores using K-means clustering, and then removing residual malicious clients from the remaining clients using DBSCAN clustering, thus selecting benign clients; and performing federated learning global model aggregation based on the gradient updates of the selected benign clients.

This invention discloses a multi-model fusion method, apparatus, device, storage medium, and program product. It acquires current environmental state data and prediction results from multiple deep reinforcement learning models; calculates initial weights for each deep reinforcement learning model based on reinforcement learning update rules, using the current environmental state data, historical performance metrics of each deep reinforcement learning model, and the prediction results; calculates a deterministic metric for each deep reinforcement learning model based on deep evidence theory, using the prediction results; calculates final weights for each deep reinforcement learning model based on the initial weights and the deterministic metric; and weights and fuses the prediction results of all deep reinforcement learning models according to the final weights to obtain the optimal decision action. Using this invention, intelligent optimization through multi-model fusion can be achieved, providing users with reliable decision support and improving the overall performance and adaptability of the system.

This invention discloses a photovoltaic (PV) operation and maintenance (O&M) agent workflow and intelligent agent integration system, relating to the field of new energy intelligent O&M technology. It includes an edge sensing module, an event bus module, a multi-agent collaborative scheduling module, a self-evolving knowledge graph module, and a digital twin verification module. The edge sensing module collects current-voltage curves, infrared thermal imaging images, meteorological environmental parameters, and equipment operating status data of PV modules, and performs local analysis based on a lightweight anomaly detection model to output structured O&M events. By constructing a closed-loop O&M system integrating edge sensing, multi-agent collaborative scheduling, self-evolving knowledge graph, and digital twin verification, it achieves real-time and accurate identification of abnormal events in PV power plants, intelligent optimization and dynamic execution of task plans, automatic accumulation and iterative optimization of disposal knowledge, and high-fidelity simulation verification before execution, thereby improving O&M response speed and decision-making accuracy.

The invention discloses a dual-mode communication intelligent data transmission method based on edge calculation, and the method comprises the following steps: S1, collecting multi-mode environment data by an edge terminal, generating an input feature vector, and transmitting the input feature vector to an edge server; s2, inputting the feature vector into an improved hybrid expert model, activating an expert sub-network, and outputting a candidate communication action set; s3, constructing an initial node of the Monte Carlo tree, and performing multi-step simulation and evaluation; s4, according to the simulation result, backtracking the expected income, and selecting an optimal communication action sequence as a data transmission scheme; s5, issuing the transmission scheme to the edge terminal, executing mode switching, link shunting and data scheduling, and collecting feedback information; s6, the feedback information is used for optimizing the hybrid expert model and the Monte Carlo tree strategy, and a dynamic closed-loop optimization system is constructed. According to the method, the improved hybrid expert model and Monte Carlo tree search are fused, and intelligent decision and optimization of data transmission of dual-mode communication under edge calculation are realized.

The invention, which relates to the technical field of new energy application and intelligent operation and maintenance, discloses a photovoltaic wireless charging base station comprising a structure supporting module, a photovoltaic power generation unit, an intelligent control module, an energy storage module, a wireless charging module and a cloud service platform. The energy storage module comprises three independent pluggable battery packs, namely a main power supply battery pack C0, a first standby battery pack C1 and a second standby battery pack C2; the intelligent control module executes a multi-battery-pack intelligent scheduling algorithm, and controls the charging and discharging priority and automatic switching of each battery pack; the battery pack adopts a quick release structure and can be quickly extracted for emergency battery replacement; and the cloud service platform realizes remote monitoring and operation and maintenance management through the wireless communication module. According to the invention, zero-construction rapid deployment and unmanned intelligent operation and maintenance can be realized, and the reliability of energy supply is ensured through a dual guarantee mechanism of wireless charging and emergency battery replacement.

The invention provides a terahertz communication platform OMC system signaling tracking and analyzing method. The method comprises the following steps: acquiring an original signaling code stream reported by a managed device through an OMC system; performing fingerprint feature extraction on the original signaling code stream to generate a signaling fingerprint; matching the signaling fingerprint with a preset analysis routing table, and selecting a corresponding signaling analysis strategy according to a matching result; calling a signaling tracking dynamic library to analyze the original signaling code stream through a signaling analysis strategy to obtain an analysis result; and displaying and storing the analysis result.

The invention provides an intelligent fault diagnosis and emergency disposal method for a tower crane jacking hydraulic system, and belongs to the technical field of tower crane jacking hydraulic systems. Pressure, temperature, flow, vibration, displacement and acoustic sensors are arranged at multiple key parts of the tower crane jacking hydraulic system to collect multi-dimensional operation data; a feature noise self-adaptive filtering mechanism is established, the filtering intensity is dynamically adjusted according to the signal-to-noise ratio, weak fault signals are enhanced, a multi-head attention mechanism network is used for conducting fault mode intelligent recognition and accurate classification, the fault severity degree is quantized through a mahalanobis distance fault degree evaluation method, and self-adaptive adjustment of a confidence coefficient threshold value is achieved; and a multi-stage alarm response mechanism is constructed to execute corresponding monitoring strategy adjustment and emergency disposal operation according to the fault type and severity, so that the technical problem of low diagnosis accuracy caused by difficulty in effective identification of weak fault signals in the fault diagnosis process of the tower crane jacking hydraulic system is solved.

This invention relates to a tree-structured dynamic access control strategy and device for multimodal power grid data, belonging to the field of data access control technology. It solves the problem that existing technologies struggle to accurately identify and respond in real-time to complex abnormal behaviors. The strategy includes: preprocessing multimodal power grid data to construct a time-series tensor of multimodal features; extracting and modeling the time-series tensor using a multi-scale convolutional neural network and a bidirectional long short-term memory network to generate a global feature representation of the time-series tensor; generating a detection score for the multimodal power grid data based on the global feature representation through a fully connected layer; calculating the difference between the detection score and a dynamic threshold; and dynamically adjusting the minimum access permissions granted to users based on the path length of the difference in a multi-branch tree. This invention detects and evaluates data from different time spans and long-term / short-term dependencies, significantly improving the detection accuracy of abnormal access behaviors.

The application relates to the fields of artificial intelligence and digital medicine, and discloses a method, device, medium and equipment for generating an intelligent activity process, which comprises the following steps: acquiring a crowd label, an activity time and an activity type of an activity crowd; generating an intelligent activity process and setting a timing task according to the crowd label, the activity time and the activity type of the activity crowd; and executing corresponding actions according to the execution sequence of the intelligent activity process after the timing task is triggered. The embodiments of the application can generate a branch structure of an activity process by combining a crowd label, and form a final intelligent activity process through the adjustment of a user, so that flexible configuration and intelligent optimization of the activity process are realized, diversified and personalized processes are realized, and the technical problems of single activity form and low matching degree between a target crowd and an activity process are solved.

The present application belongs to synthetic aperture radar technical field, especially to a kind of GEO star ground bistatic SAR aperture optimization method and device based on resolution optimum, the specific process of this method is as follows: based on the resolution ellipse obtained by multiplying two main lobes when GEO star ground bistatic SAR two-dimensional imaging, construct resolution ellipse long axis model;Select core imaging area, according to ground plane range resolution and azimuth resolution, construct ground plane range and azimuth resolution mean model of core imaging area;Set the traversal step of trajectory, traverse to obtain the long axis of ellipse and resolution mean of full track section, select trajectory section that meets the requirement of set long axis of ellipse and resolution mean;Set two constraint conditions of core imaging area without blur and GEO star ground bistatic SAR visibility, complete imaging arc segment final selection using constraint condition, thus complete imaging aperture optimization based on double constraint joint screening.

The application provides a water jet propeller flow field reduction method based on a POD coupling neural network, and the method comprises the following steps: performing CFD simulation based on a boundary condition; performing eigenvalue orthogonal decomposition on data of a training set, extracting POD modes, and sorting the modes in descending order of energy; constructing a corresponding neural network model according to the data structure of the features in the low-dimensional feature set; optimizing the neural network model, evaluating the optimized reduction model, obtaining corresponding high-dimensional flow field data for a target water jet propeller working condition, converting the high-dimensional flow field data into corresponding low-dimensional features, inputting the low-dimensional features into the evaluated reduction model, and obtaining a reduction result of the target water jet propeller working condition. The application extracts eigenvalue flow structures in high-dimensional CFD simulation or experimental data, constructs a low-dimensional approximate model with extremely low calculation cost, restores a high-resolution full flow field in a short time, and directly outputs engineering parameters such as pressure drop, flow velocity distribution and vortex characteristics.