159 results about "Algorithm optimization" patented technology

The application discloses to the technical field of power network, specifically is a kind of power network vulnerability analysis system and method based on virtual attack and defense deduction, comprising: data acquisition module, for responsible collection various data of power network, the various data include network topology structure data, equipment configuration data, operating state data, security log data;Threat intelligence integration module is used to collect, integrate network security threat intelligence from the world, and with threat intelligence platform establishes data interface, realizes the automatic synchronization and update of threat intelligence.The application can simulate new attack means, monitor power network security state in real time by continuous virtual attack and defense deduction, combined with machine learning algorithm optimization attack and defense strategy.Even if there is unprecedented attack mode, the application can quickly identify potential threats through attack and defense confrontation, greatly improve the discovery ability of new and unknown vulnerabilities, and respond to network security risks in time.

The application relates to a mechanism and data driving-based engine cylinder cover milling surface quality prediction method and belongs to the technical field of automatic prediction. The method comprises the following steps: determining key evaluation indexes and influence factors of cylinder cover milling surface quality; collecting process parameters and surface roughness values, combining with a cylinder cover milling mechanical state to construct a milling force and thermal mechanism model based on a semi-analytical method and a heat source method, obtaining state variable data, and storing the state variable data into a historical database after preprocessing; constructing a surface roughness prediction model based on ADE algorithm optimization SVR, taking process parameters and mechanism model output state variable data as data driving model input, taking surface roughness values as output, and obtaining an optimal parameter combination of SVR by training historical data; and predicting the milling surface roughness by using real-time process parameters and mechanism model output state variable data. The method has the advantages of strong model state representation capability and low state variable data acquisition cost, and can realize accurate prediction of cylinder cover milling surface quality.

The present application relates to the technical field of agricultural remote sensing and machine learning, and particularly relates to a leaf area index estimation method based on improved XGBoost. The method comprises the following steps: acquiring unmanned aerial vehicle multi-spectral images and sample leaf area index measured values, and constructing a vegetation index map sample data set after preprocessing; performing feature extraction and fusion on the vegetation index map by using a deep learning network; constructing an improved beaver optimization algorithm, generating an initial population by using a two-stage initialization strategy, updating the position of the architect subpopulation by using an elite directional felling strategy, recombining individuals and the global optimal solution by using a vertical and horizontal cross strategy; optimizing the XGBoost hyperparameters by using the improved beaver optimization algorithm, and establishing a leaf area index estimation model. The leaf area index estimation method based on improved XGBoost combines deep learning feature extraction, improved swarm intelligence optimization algorithm and integrated learning regression modeling, and is helpful to improve the prediction accuracy and stability of the leaf area index estimation model.

The application relates to the technical field of quantifiable test indexes of new energy vehicles, and particularly discloses a method for constructing a typical road spectrum working condition based on a genetic algorithm and new energy platform data, data acquisition, invalid data section elimination rules including static data section elimination and abnormal continuous duration elimination; data abnormal value identification rules: speed abnormal values are defined and a correction method is set; data smoothing and denoising rules: a median filter is used to preliminarily denoise a speed sequence; characteristic parameter calculation: including kinematic characteristic parameters and new energy light truck characteristic parameters; construction of an initial section library; multi-objective genetic algorithm optimization; road spectrum synthesis and post-processing: optimal sections filtered by the algorithm are spliced, and smoothing processing is carried out to ensure the continuity of a speed curve; multi-objective genetic algorithm selection: selecting an NSGA-III algorithm as a multi-genetic objective algorithm condition; the application realizes rapid construction and iteration of the road spectrum, and is high in efficiency.

This invention discloses a power distribution IoT communication and equipment management system based on dynamic optimization algorithms. It adopts a four-layer architecture: a perception layer, an edge layer, a platform layer, and an application layer. The end devices in the perception layer are used to collect power distribution data and execute control commands. The edge devices in the edge layer, and their managed sub-devices, are used for protocol conversion, data preprocessing, and executing dynamic optimization algorithms. The platform layer is used for message parsing, algorithm optimization, device management, and data storage. The application layer provides a visual user interface. Through dynamic optimization algorithms and a full lifecycle device management scheme, the system significantly improves the reliability of power distribution IoT data transmission, optimizes network resource utilization, enhances equipment operating efficiency, ensures the real-time performance of critical services, and features strong compatibility, security, and concurrent processing capabilities.

This invention discloses a multi-parameter coupling-based deep peak-shaving full-condition optimization control method for coal-fired power units. Addressing the problems of existing technologies, this invention constructs a multi-parameter coupling index system, calculates the comprehensive coupling strength between each parameter and the load, and selects core parameters. Dynamic clustering is used to finely divide the full-condition operation from 20% to 100% rated load into multiple operating zones. An LSTM-GRU dual-channel deep learning prediction model is established for each operating zone. A three-layer hierarchical control architecture is designed, comprising load optimization scheduling, boiler-turbine coordination control, and combustion optimization execution. The boiler-turbine coordination layer uses multivariate model predictive control and online self-tuning of multi-objective weights through fuzzy inference. The combustion execution layer uses the NSGA-III multi-objective evolutionary algorithm to optimize air distribution. A constraint-adaptive tuning mechanism based on safety margin is established to form a closed-loop iterative optimization across all operating conditions. This invention can significantly improve the load response speed, main parameter stability, and combustion economy of the unit during wide-load operation.

This invention discloses a method and system for transformer partial discharge pattern recognition based on an optimized probabilistic neural network. The method first acquires the transformer partial discharge signal; then, it establishes a two-dimensional spectrum of the partial discharge phase distribution pattern based on the partial discharge signal and extracts discharge statistical features from this spectrum; finally, it inputs the discharge statistical features into the optimized probabilistic neural network for pattern recognition to obtain the partial discharge pattern. The optimized probabilistic neural network uses a pollination algorithm to optimize the smoothing factor, and the switching probability in the pollination algorithm is a nonlinear function that decreases with the number of iterations. This invention can more accurately and efficiently classify and identify different partial discharge patterns of transformers, providing a data foundation for transformer fault diagnosis and resolution.

This invention discloses an industrial carbon emission detection and prediction method and system. It collects multi-source heterogeneous data from industrial production processes, and sequentially performs standardization, noise removal, outlier removal, and data fusion on the multi-source heterogeneous data to obtain initial carbon emission data. A hybrid detection and prediction model is constructed, employing an encoder to extract multi-factor correlation features from the multi-source data and a long short-term memory network to extract temporal fluctuation features of the carbon emission data. An attention mechanism is introduced to weight the correlation features and temporal fluctuation features. An improved IWOA whale optimization algorithm is used to optimize the hyperparameters of the hybrid detection and prediction model to obtain a target hybrid detection and prediction model. The initial carbon emission data is input into the target hybrid detection and prediction model, and the prediction results are output, including the current carbon emission detection value and the carbon emission prediction sequence. This improves the accuracy of carbon emission prediction results and the efficiency of carbon emission detection.

The application discloses a double-reliability RVFL bat optimization wind power interval prediction method and system, specifically for: preprocessing wind farm historical power data, constructing a data set; establishing a basic RVFL, a cascaded enhanced RVFL and a direct connection enhanced RVFL candidate model, expressing the upper and lower bounds of the prediction interval as a linear combination of enhanced features and output weights; in the first stage, the hidden layer parameters are fixed, the output weights are solved by using an opportunity constraint optimization model containing a training set and a validation set coverage constraint, and the optimal basic model is selected according to the validation set SCORE index; in the second stage, the model parameters are used as elite prior solutions, a mixed initialization bat population is constructed by combining random exploration individuals, and the global network parameters are optimized; and the prediction interval is output on the test set and evaluated. The application improves the generalization coverage ability by double-reliability constraint, enhances the scene adaptability, and optimizes the network parameters by using a mixed initialization bat algorithm.

The embodiment of the application provides a kind of automatic driving algorithm optimization degree determination method, device and electronic equipment, can be determined according to the driving data of test vehicle based on the first automatic driving algorithm on the road, determine multiple first sample pairs, wherein the first sample pair includes first sample standard entropy and the average takeover times corresponding to first sample standard entropy, the first linear equation can be obtained by regression processing to multiple first sample pairs, then according to the driving data of test vehicle based on the second automatic driving algorithm on the road, determine multiple second sample pairs, wherein the second sample pair includes second sample standard entropy and the average takeover times corresponding to second sample standard entropy, according to multiple second sample pairs and first linear equation, the optimization degree of second automatic driving algorithm compared with first automatic driving algorithm can be determined. By quantifying automatic driving algorithm, the optimization degree of automatic driving algorithm can be evaluated.

The application relates to the field of ultra-wideband radar imaging and optical computing technology, in particular to a light computing real-time processing method for ultra-wideband radar imaging. The method comprises core steps of signal receiving and electro-optical modulation, optical pulse compression, initial phase compensation based on Kalman filtering, envelope alignment, coherence recovery and two-dimensional grid projection. Based on the ultrafast linear processing advantage and algorithm optimization principle of light computing, the application utilizes optical finite impulse response filters and adjustable delay line technology to disassemble and reconstruct the traditional imaging method, and reduces the end-to-end processing delay to the sub-millisecond level. The application breaks through the real-time processing bottleneck of ultra-wideband radar imaging, provides effective theoretical support and method basis for developing a full real-time ultra-wideband two-dimensional fine imaging system, and is widely applicable to application fields with high real-time requirements such as military reconnaissance and geological exploration.

The application relates to a reinforcement learning optimization method and device for an end-to-end target detection algorithm. The method comprises the following steps: obtaining image data to be processed, extracting image data features of the image data to be processed by using a hyperparameter optimization model, selecting corresponding hyperparameters in combination with corresponding task information, performing algorithm optimization on an end-to-end algorithm by using the hyperparameters, obtaining an optimized end-to-end algorithm, performing target recognition by using the optimized end-to-end algorithm, evaluating a target detection recognition result, selecting corresponding rewards, updating the hyperparameter optimization model based on the rewards in combination with a reinforcement learning algorithm gradient, and obtaining an optimized parameter optimization model. The method of the application does not need manual design for optimization, does not need to set a fixed manual detection threshold, greatly improves the algorithm optimization efficiency, and significantly improves the generalization of the algorithm.

This invention relates to the field of intelligent reliability engineering technology for electromechanical products, specifically an adaptive failure prediction method for motors based on digital twin-based intelligent optimization. Through physically-driven intelligent modeling, a physical information neural network is used to automatically learn time-varying coupling coefficients, achieving an organic fusion of data-driven approaches and physical constraints. Failure paths are dynamically reconstructed by using a particle swarm optimization algorithm to identify dominant failure modes in real time and dynamically adjust the weights of the failure propagation chain. A prediction-verification-correction closed loop is implemented, leveraging digital twins and Bayesian inference to achieve continuous self-evolution of the model and constantly improve prediction accuracy. Intelligent design of test schemes is employed, using a multi-objective genetic algorithm to optimize and accelerate test conditions, improving test efficiency and failure mode coverage. Probabilistic lifetime prediction outputs a remaining lifetime distribution with confidence intervals, quantifying prediction uncertainty.

This invention discloses a visual detection method for industrial production safety targets that integrates progressive feature pyramids. It acquires images of key areas in industrial settings, optimizes the target detection model through a lightweight feature extraction network and a dynamic threshold decision mechanism, introduces an adaptive feature pyramid network (AFPN) and an improved detection head structure, and enhances multi-scale feature fusion capabilities for detection. A hierarchical early warning system is constructed to achieve rapid response, accurate push notifications, and closed-loop management. A visual control system is configured to output the early warning results. This invention does not require additional dedicated detection equipment and is compatible with existing surveillance cameras or conventional industrial cameras in industrial settings. It can be implemented simply through algorithm optimization and architecture upgrades, effectively improving the feature capture efficiency for safety hazard targets of different distances and sizes, and significantly reducing category confusion and judgment errors in complex industrial scenarios.

This invention relates to the field of passive underwater target identification technology, and in particular to a propeller noise cyclic coherent demodulation method based on an improved real-number genetic algorithm. Based on cyclic stationarity analysis theory, an optimal coherent weighted quadratic envelope spectrum is constructed by replacing the fixed integration frequency band with a weighting function, and an adaptive detection threshold is constructed based on the false alarm probability. The observed spectrum is modeled as the sum of harmonics and noise, and a sparse non-negative optimization problem is solved to extract the harmonics. Using the total harmonic intensity of the envelope spectrum as the cost function, an improved real-number genetic algorithm is used to optimize and search for the optimal weighting coefficient vector. The demodulated spectrum is obtained by coherently weighting and integrating the normalized cyclic modulation spectrum. This invention achieves autonomous allocation of weighting coefficients and adaptive selection of demodulation frequency bands, effectively improving the signal-to-noise ratio gain of cyclic demodulation, and is applicable to passive target identification in data backtracking and review systems.

The application provides a control system of an LCD attaching device, and belongs to the technical field of control of LCD panel manufacturing equipment. The control system comprises the following modules: a vision module, which acquires a substrate pose and constructs a nonlinear target function, and realizes accurate alignment through ant colony algorithm optimization; a pressure acquisition module, which acquires pressure data and constructs a continuous pressure cloud chart; an extraction and calculation module, which analyzes the cloud chart, extracts a characteristic region, and constructs a risk index; a network analysis module, which outputs a speed adjustment amount and a pressure compensation amount through a fuzzy neural network; and a correction and determination module, which realizes micro-control correction of the attaching speed and pressure. The application realizes high-precision pose alignment and pressure acquisition, quantifies the attaching risk, dynamically regulates and controls the attaching parameters, effectively corrects local deformation of the substrate, and improves the LCD attaching yield.

The application discloses a kind of unmanned plane path planning method and related device based on triangle segmentation technology, method includes: based on the preset triangle route method obtains multiple unmanned plane inspection routes, forms unmanned plane inspection route set;With unmanned plane inspection route set as initialization particle group, route optimization calculation is carried out using preset genetic algorithm, obtain optimal unmanned plane route, preset genetic algorithm is based on the preset triangle segmentation algorithm optimization route selection, based on preset collision crossover strategy cross operation.Not only can select suitable genetic individual, but also can reduce calculation amount, while guaranteeing the reliability of route planning, also can minimize calculation pressure.Therefore, the application can solve the technical problems that the route planning method based on genetic algorithm in the prior art is limited by large calculation amount or large error, resulting in poor efficiency in engineering application process.

This invention discloses a lightweight AI underlying optimization method and system that significantly saves national computing resources, involving the fields of AI underlying algorithm optimization, domestic computing power scheduling, and embedded AI chips. Based on proportional filtering logic and STE dynamic lossless quantization encoding, this invention achieves end-to-end underlying reconstruction of AI models, incorporating domestically produced lightweight chips and inference models, and setting thresholds for redundancy removal, computing power utilization, accuracy loss, and computing power reduction. Through feature filtering, operator reconstruction, weight quantization, chip operator alignment, and accuracy self-calibration, model compression and efficient computing power scheduling are achieved, eliminating reliance on foreign technologies. This system can reduce AI computing power consumption in key national sectors by more than 65%, achieve a computing power utilization rate of ≥85%, increase inference speed by 2 times, and reduce accuracy loss by ≤0.5%. It is applicable to government, industry, cloud, and edge computing scenarios, contributing to the intensive, green, and autonomous use of computing power, and filling the gap in domestic AI underlying lightweight optimization technology.

The present application relates to a transformer fault diagnosis method based on improved sparrow search algorithm optimization SVM, through the pretreatment and feature engineering of transformer oil chromatographic data; according to the data of pretreatment and feature engineering, the model based on oil chromatographic data is constructed, and the model parameters are optimized and trained; the input transformer oil chromatographic data is carried out transformer fault diagnosis based on the optimized and trained model.The present application improves the sparrow search algorithm by introducing the best point set strategy, the golden regular update rule, the differential mutation disturbance and the reverse learning mechanism, so as to comprehensively improve the parameter optimization process of support vector machine, the global exploration ability of the algorithm is enhanced, the risk of falling into local optimum is avoided, and the problem of premature convergence is effectively avoided.In addition, the improved algorithm pays attention to multi-link optimization, and the parameter setting is simple, the algorithm is low in use difficulty, so that the optimized support vector machine (SVM) can effectively improve the transformer fault prediction precision.

The application provides a multi-target path collaborative optimization method based on space-time feature enhancement and graph neural network guidance, comprising the following steps: step 1, constructing a basic data system supporting path optimization; step 2, generating a node embedding vector capable of guiding path search; step 3, reconstructing the relevance of the distribution task in the feature space, and generating a high-quality initial solution set through heuristic guidance; step 4, realizing fine iteration of the distribution scheme through an improved genetic operator; and step 5, establishing a fast lookup table mechanism based on time axis discretization preprocessing and state recursion. The method effectively improves the real adaptability and algorithm optimization accuracy of the path planning scheme, realizes the collaborative optimization of distribution efficiency and driving safety, greatly reduces the calculation complexity in the dynamic scene, can generate a Pareto optimal solution considering multi-target requirements, meets the real-time scheduling and decision-making requirements of complex urban distribution scenes, and has good engineering application value.

This invention relates to the field of rock mechanical property testing technology, and in particular to a machine learning method and system for predicting rock shear strength parameters. The method includes the following steps: determining rock shear strength parameter prediction indices and rock shear strength parameters, and establishing a sample database; improving the sparrow search algorithm using an improved Tent chaotic mapping, dynamic adaptive weights, Levy flight strategy, and Cauchy-Gaussian hybrid mutation mechanism; training a Stacking ensemble model using the sample database, and simultaneously optimizing the hyperparameters in the Stacking ensemble model using the improved sparrow search algorithm to obtain a rock shear strength prediction model; using the rock shear strength prediction model to predict rock shear strength parameters, and simultaneously using the SHAP interpretability method to evaluate the contribution of each rock shear strength parameter prediction indices. This invention can achieve high-precision prediction of rock shear strength parameters and effectively avoids the problems of long cycles and high costs in existing rock shear strength testing.

The present application relates to the technical field of semiconductor testing equipment, and particularly relates to a full-automatic high-precision probe station, which utilizes multi-physical field coupling sensing and compensation to significantly improve positioning accuracy: through comprehensive collection of multi-physical field data such as temperature field and stress field, combined with coupling algorithm optimization compensation strategy, multi-factor interference is effectively offset, positioning repeat accuracy can reach ±0.05 mu m, and the testing demand of advanced semiconductor technology is met; dynamic three-dimensional modeling technology enhances dynamic adaptability: a dynamic three-dimensional modeling system of multi-physical field data influence size can update the testing target morphology in real time, avoids testing deviation caused by target deviation or surface undulation, and improves the adaptive ability of the equipment to complex testing scenes. High degree of automation improves testing efficiency: the full process of data collection, modeling, compensation and testing can be completed without manual intervention, the calibration process is simplified, the testing period is shortened, and it is suitable for batch detection scenes.

The application discloses a kind of multi-machine cooperation SLAM methods based on active deep reinforcement learning.The method comprises the following steps: running ORB-SLAM2 program to robot, and the initial motion trajectory pose graph of multiple machines is obtained by pose estimation of image acquisition through camera;Based on the obtained robot motion trajectory pose graph, more accurate pose is obtained by using deep reinforcement learning TD3 algorithm training to optimize trajectory;On the basis of reinforcement learning algorithm, active perception strategy is introduced to optimize the pose of multiple machines simultaneously, and according to the real-time SLAM estimation probability value P, the corresponding robot is selected to optimize the pose information by TD3 algorithm;The pose information and actual distance information of each robot are transmitted between robots, and the back-end optimization of SLAM trajectory is carried out using TD3 algorithm, to eliminate the cumulative error effect.The application can effectively eliminate the error accumulation in SLAM system, improve the positioning and mapping accuracy of SLAM, and there is no loop restriction, which increases the robustness of SLAM system.

This invention discloses a machine learning-based method and system for investment promotion recommendation in smart industrial parks, relating to the field of smart industrial park management technology. The method includes: collecting multi-source heterogeneous data from enterprises and park infrastructure within the smart park, extracting and embedding features to construct an enterprise feature matrix and a park resource feature matrix; constructing a multi-task prediction model using machine learning algorithms based on the enterprise feature matrix and park resource feature matrix; constructing an investment promotion recommendation evaluation function and introducing an improved slime mold optimization algorithm to optimize the weight vector, resulting in an optimized investment promotion recommendation evaluation function; and generating investment promotion recommendation results using the multi-task prediction model and the optimized investment promotion recommendation evaluation function based on the enterprise feature matrix and park resource feature matrix. This invention effectively solves the problems of low accuracy in virtual-real fusion, lack of adaptability in teaching models, single interaction methods, and single evaluation dimensions by deeply mining data features and adaptively optimizing evaluation weights.

The application relates to a virtual simulation test method and system for automatic driving social interaction capability, a mixed background traffic flow generation module with diversified human driving strategies is constructed, dynamic two-way interaction between background traffic participants and the tested AV is realized, and the simulation scene authenticity and test comprehensiveness are improved; with the help of a quantitative evaluation system based on human social rules, abstract social interaction capability is converted into quantifiable indexes, and clear bases are provided for AV interaction algorithm optimization; through the construction of a virtual simulation test platform with deep integration of multiple modules, the test complexity and cost are reduced, the test cycle is shortened, and the risk of extreme scene test is avoided; meanwhile, the virtual simulation test platform has good flexibility and expandability, can adapt to the test requirements of different scenes and different levels of AVs, can accurately simulate human implicit interaction behavior, reduces the deviation of the AV driving strategy, improves the driving safety and the traffic system efficiency, and provides reliable test support for large-scale commercialization of AVs.

This invention discloses a data glove for a teleoperation system and its control method. The data glove body is worn on the user's hand, and a signal transmission network is constructed between hand, wrist, and arm data units mounted on the glove body. Complete kinematic chain data from the shoulder joint to the finger joints is collected by each data unit and transmitted through the signal transmission network, including quaternion pose data and gesture data. The quaternion pose data is processed to obtain a spatial pose matrix, and the gesture data is processed using a gesture mapping algorithm to obtain a gesture sequence number. A master-slave mapping algorithm maps the workspace of the glove to the workspace of the slave device, thereby enabling the master device to freely control the slave robotic arm and execute specific complex task processes. This invention, through modular sensor layout and algorithm optimization, achieves precise, intuitive, and natural control of the slave robotic arm by the operator, reducing operational complexity and improving the user experience.

The present disclosure relates to an optimization system and method of an artificial intelligence (AI) algorithm for predicting a lesion in a coronary artery based on a fractional flow reserve (FFR), and more particularly, to a technology capable of providing an AI algorithm of which prediction accuracy of an FFR is improved.

The present application belongs to the technical field of building material microstructure analysis, and in particular to a foam light soil microstructure analysis method based on image recognition; a foam light soil forming test piece is selected, and a flaky analysis sample is obtained after surface treatment; an imaging device is used to collect multi-view images of the sample, and two-dimensional images meeting the analysis accuracy are obtained; the collected images are subjected to denoising, segmentation and quality optimization processing, and the boundary features of the pores and matrix are highlighted; the pore area is extracted through an image segmentation algorithm, the pore profile is automatically identified and microstructure characteristic parameters are collected, and an analysis model is constructed based on the characteristic parameters; the present application includes sample preparation, multi-view image acquisition, optimization preprocessing, feature identification and analysis steps, and through multi-angle collection and algorithm optimization, the pore features are accurately quantified, the information loss and quantization errors of traditional methods are avoided, and the present application has the advantages of low cost, high efficiency and accurate analysis of foam light soil microstructure features.

This invention discloses a method for enhancing the brightness temperature resolution of a microwave radiometer, along with related devices, equipment, and storage media. It relates to the field of spaceborne microwave radiometer technology and aims to solve the problems of insufficient brightness temperature resolution, sidelobe-induced reconstruction distortion, and the difficulty in balancing resolution and stability in spaceborne real-aperture microwave radiometers. The method includes: acquiring relevant data; constructing a convex optimization model with sidelobe suppression constraints and solving for initial weights; iterating with the initial weights, adaptively updating until a termination condition is met, and outputting the optimal weights; using the optimal weights to reconstruct the original brightness temperature data, outputting a result with enhanced resolution and no obvious oscillation artifacts. This invention requires no hardware modification, balances sidelobe suppression and resolution through algorithmic optimization, takes into account both accuracy and stability, adapts to the real-time requirements of spaceborne systems, provides support for meteorological and oceanographic monitoring, and discloses corresponding devices, equipment, and storage media.