31 results about "Parallel learning" patented technology

The invention discloses a marine environment real-time monitoring and early warning system based on machine learning, and relates to the technical field of machine learning. Comprising the steps that an ocean multi-source sensing module collects ocean environment data in real time through a sensor and a combined collection scheme; the multi-source feature extraction module performs time domain, change rate and frequency domain feature analysis on the data to construct a unified multi-dimensional feature vector; the multi-model fusion prediction module outputs a marine environment state vector through dynamic weighting and deviation correction based on a parallel learning architecture of a deep neural network, a long-short-term memory network and a one-dimensional convolutional neural network; and the ocean risk identification and early warning module generates graded and classified early warning information through double study and judgment of a sea condition classifier and an abnormal event detector. According to the method, comprehensive acquisition, deep feature mining, high-precision prediction and accurate early warning of marine environment data are realized, the problems of low prediction precision, risk identification lag and the like in the prior art are effectively solved, and reliable guarantee is provided for marine operation safety.

The invention discloses a hydraulic arm safety control method and device based on parallel learning and a high-order CBF, and the method comprises the steps: building a kinetic equation of a hydraulic mechanical arm through a Lagrange method, integrating unmodeled dynamics, structural parameter change and external interference into an uncertain item, and describing the generalized uncertainty of the uncertain item in a linear parameterization form; for the problem of insufficient excitation in a complex environment task, a parallel learning mechanism is introduced, and historical data and real-time data are combined to realize parameter identification. For high relative order safety constraints (such as obstacle distance constraints and joint limiting constraints) in a task space, an obstacle function family is constructed. A dynamic error buffer function is introduced, and a high-order adaptive control barrier function condition is designed. The high-order self-adaptive control obstacle function constraint is embedded into a real-time quadratic programming solving problem, input obtained through optimization solving can keep the track precision, meanwhile, a joint instruction is automatically corrected to prevent constraint failure, and minimum intervention type safety control is achieved.

The invention discloses a water plant inlet water quality change identification method and system, and the method comprises the steps: collecting water plant inlet water turbidity, pH value, dissolved oxygen content, COD concentration, ammonia nitrogen concentration, total phosphorus concentration and other parameters, removing abnormal values, and constructing a water quality parameter time sequence data set; causal association among parameters is mined through a causal inference-based hybrid model, a weight matrix is generated, after the matrix and time sequence data are fused, bidirectional time features are extracted through a bidirectional long-short-term memory network in combination with an attention mechanism, attention weights are given, and then a gradient elevator integrated model is input; a plurality of base learners are used for parallel learning and dynamic weighted fusion of local results to obtain a preliminary recognition result, and finally the preliminary recognition result is mapped to a water quality change type space to determine a change type. The system comprises six units, a complete and efficient treatment flow is formed from water quality parameter collection to final recognition result output, and the actual requirements of a water plant for high-precision and high-reliability recognition of inflow water quality changes are practically met.

The invention discloses a malicious traffic detection method fusing CNN-LSTM, and belongs to the technical field of network security. Comprising a multi-layer feature fusion mechanism of modular design, multi-path parallel learning design and joint optimization of regularization and feature compression. The method has strong automatic feature extraction capability, hidden complex features can be autonomously learned from a large amount of network traffic data, and the detection efficiency and the detection effect are improved; the method has the advantages that the false alarm rate is obviously improved, the complex nonlinear relation can be better processed, and the method has higher learning ability for the attack mode which is difficult to capture by the traditional method; the method has strong generalization ability and can well adapt to a dynamically changing network environment; even in the face of unknown threats, effective detection can be carried out through the similarity of the feature modes; and the data feature capturing capability, concurrency, expression capability and robustness of the model are effectively improved.

The application discloses a kind of FFC self-learning assembly methods based on parallel reinforcement learning, establishes including real physical assembly system and simulation system parallel learning system, and physical assembly system and simulation system are run in parallel, experience data in assembly process is gathered to experience pool of parallel learning system, simulation system is trained by experience data provided by physical assembly system in experience pool, and feedback guides real physical system to execute assembly task;Wherein, the physical information of physical assembly system is stored into parameter server for simulation system to train after being handled by Softmax classifier of parallel learning system.The self-learning assembly method of the application not only has the advantages of high assembly efficiency and learning efficiency, but also has the advantage of high assembly success rate.

The invention provides a multi-quad-rotor unmanned aerial vehicle distributed anti-collision optimal formation control method based on reinforcement learning, and the method specifically comprises the steps: firstly constructing a multi-unmanned aerial vehicle distributed optimal formation model through graph theory knowledge and an unmanned aerial vehicle dynamics model; according to different distances between the unmanned aerial vehicle and the obstacle, the flight area is divided into three parts, and the speed is limited after the unmanned aerial vehicle detects the obstacle, so that the anti-collision function of the unmanned aerial vehicle is realized; secondly, in order to convert a distributed formation control problem into a trajectory tracking problem, a distributed observer is designed to estimate the state of a virtual leader; then, a newly defined value function is corrected through a Lyapunov-like control barrier function, so that anti-collision distributed optimal formation control is realized; and finally, designing a new Hamiltonian-Jacobian-Bellman equation by using the corrected value function, and determining optimal control input by using single-evaluation reinforcement learning in combination with a gradient descent method and a parallel learning algorithm.

The application discloses a kind of restaurant scene under the identification method and safety monitoring system of character interactive behavior, belong to video analysis technical field, the method includes: determine each video monitoring collection area picture, data pre-processing etc.;Establish restaurant staff face and human target, customer human body, restaurant common article label library;Establish the label library of the body related action behavior of interaction with restaurant common article and the interaction between people, judge the safety / unsafe of its behavior and establish label;Build double-task double-flow parallel learning network, extract the feature information of target identification and the feature information of character interactive action behavior;Process the feature information extracted by fusion model network, build a network model for restaurant customer safety monitoring and employee service work content understanding, and judge whether there is unsafe employee and customer behavior, establish a kind of intelligent monitoring system, and have certain help effect to the automation level of video intelligent monitoring system.

The application discloses an emergency operation and maintenance anomaly detection system and method based on double-network time sequence staggered learning, relates to the technical field of emergency operation and maintenance anomaly detection, and performs directional collection, time alignment, anomaly identification and correction on emergency operation and maintenance full-link information system node operation data to construct a standardized operation and maintenance time sequence dataset; further, short-range time sequence segments and long-range time sequence segments are extracted according to time advancing relations and are respectively input into a first detection network and a second detection network to realize parallel learning of local fluctuation characteristics and continuous evolution characteristics; on this basis, collaborative anomaly discrimination is completed through double-network staggered feedback, and in combination with the continuation relation of the anomaly segments in adjacent time sequence intervals, anomaly event credibility evaluation, grade division and early warning information output are completed, so that the accuracy, reliability and graded early warning capability of emergency operation and maintenance anomaly detection are improved.

The invention discloses a dialogue type recommendation method and system based on multi-source collaborative enhancement, and relates to the technical field of artificial intelligence and recommendation systems, and the method comprises the following steps: S1, multi-source feedback collection and matrix construction: extracting hidden feedback data from a dialogue context and extracting dominant feedback data from an external platform, a user and project interaction matrix is constructed based on implicit feedback, a user scoring matrix is constructed based on dominant feedback, and normalization processing is performed on the dominant feedback to improve cross-user comparability; s2, multi-source collaborative weight learning: respectively taking the user and item interaction matrix and the user scoring matrix as collaborative data input, and executing EASE learning in parallel to obtain an implicit weight matrix and an explicit weight matrix to represent an item similarity structure; according to the method, multi-source collaborative enhancement of the dialogue type recommendation candidate set is realized through EASE parallel learning of correlation weights of multi-source items, density statistics-based adaptive weighted fusion and combination of dialogue context dynamic popularity adjustment.

This invention discloses a deep learning-based method for ultra-short-term photovoltaic (PV) power prediction. First, historical data is preprocessed to eliminate irrelevant variables and accelerate model training. Then, the advantages of three clustering algorithms are combined to obtain a more reasonable dataset partitioning. Next, particle swarm optimization is used to optimize the parameters of variational mode decomposition. Finally, parallel learning of CNN and GRU networks is employed to identify local and temporal features of the data, enabling the network structure to fully leverage the input data. Finally, a deeper learning process is achieved through CNN neural network concatenation and fusion, resulting in high-precision prediction. The PV power prediction method of this invention demonstrates excellent performance, significantly outperforming other traditional models in predicting PV power under different weather conditions.

The invention provides an engineering facility target identification method and system based on multi-source fusion data, and the method comprises the steps: obtaining a multi-source image of a to-be-processed region, carrying out the dense matching of the multi-source image through a multi-view stereo matching technology, and generating a three-dimensional point cloud; inputting the three-dimensional point cloud into a pre-constructed target recognition model for engineering facility target recognition to obtain the category of the engineering facility target; the target recognition model is obtained by training an initial target recognition model through a pre-constructed training sample, and the initial target recognition model is constructed based on a multi-feature parallel learning module for extracting primary features, a multi-feature integration module and an up-sampling decoding module for semantic recognition; according to the method, the mechanism of front attention feature extraction, feature integration and semantic recognition of human vision is simulated through the target recognition model, precise recognition of the engineering facility target is achieved, the multi-level feature integration module ensures that each level feature always participates in semantic construction, and the recognition result is more accurate.

This invention relates to the field of asphalt pavement inspection technology and discloses a method for optimizing the configuration of mechanical equipment for asphalt pavement construction. The method includes: step S2: collecting various data during the asphalt pavement construction process and preprocessing the collected data to generate dynamic sensing data; step S3: constructing a digital twin model based on the dynamic sensing data; step S4: deploying a multi-agent decision engine on edge computing nodes, and based on the prediction results of the digital twin model, solving the dynamic reconfiguration scheme of the machine group in real time through a multi-agent collaborative algorithm to generate control commands including roller operation adjustment and paver speed adjustment; step S5: issuing the control commands to the construction machine group for execution; and step S6: performing a parallel learning process on a cloud server, performing counterfactual inference on historical construction data, constructing a construction knowledge graph, and continuously optimizing the digital twin model and the multi-agent decision process based on actual construction feedback. This application improves collaborative capabilities.

The invention discloses a thermal infrared target tracking Siamese network tracker, which relates to the technical field of computer vision and comprises a fine-grained feature parallel learning convolution block, a multi-layer fine-grained feature fusion module, a Siamese residual error refining module and a significance loss function. Wherein the fine-grained feature parallel learning convolution block captures key global features from a shallow layer, the feature diversity is enhanced, and fine-grained information loss in residual connection is reduced; the multi-layer fine-grained feature fusion module effectively integrates depth and shallow features through bilinear matrix multiplication; the Siamese residual error refining module uses a residual error learning technology to correct a saliency map prediction error, and combines depth supervision to gradually optimize a prediction result; the saliency loss function restrains saliency prediction and guides the network to pay attention to highly distinguished fine-grained features. Experiments show that the tracker is excellent in performance on a plurality of benchmark tests.

The embodiment of the invention provides an image processing model training method and device and an image processing method and device.The training method comprises the steps that a training sample set is acquired, and each training sample in the training sample set comprises a sample image, a corresponding sample task and a labeled image; performing integration processing on the sample image in each training sample and the corresponding sample task to obtain sample input information corresponding to each training sample; and training a pre-constructed image processing model based on the sample input information and the annotated image until the training is completed to obtain a target image processing model. According to the scheme, multi-task processing capability can be integrated through a single model, multi-task parallel learning is supported, the training complexity is reduced, the training efficiency is improved, and the image processing result quality is improved.

The invention discloses a multi-user cooperative anti-interference frequency using method based on spectrum situation data learning, which is characterized in that a user inputs a spectrum matrix obtained by sensing at the current moment into a pre-trained neural network, the satisfaction degree of the user to different channels is output, and the user selects the channel frequency with the highest satisfaction degree; on the premise of not needing information interaction among users and not depending on a central coordination mechanism, aiming at diversified external interference problems in a dynamic electromagnetic environment, each user can autonomously avoid interference and optimize spectrum use based on independent perception and distributed learning. According to the method, through a multi-agent parallel learning architecture, each user can realize cooperative anti-interference only by depending on local sensing, mutual interference and malicious interference among the users are effectively avoided while the spectrum utilization rate is improved, and the method is suitable for a complex communication scene with strong antagonism and rapid spectrum change.

The invention relates to a network abnormal traffic detection method based on multi-modal feature fusion and deep reinforcement learning, and the method comprises the steps: collecting network traffic data, carrying out the feature extraction of the network traffic data, and obtaining continuous features and discrete features; performing normalization and imaging processing on the continuous features to obtain a two-dimensional image; the two-dimensional image is input into a trained encoder, an image feature vector is acquired, and the encoder adopts a SimCLR framework to carry out self-supervised pre-training; performing coding conversion on the discrete features to obtain statistical feature vectors; and splicing the image feature vector and the statistical feature vector into a composite feature vector, inputting the composite feature vector into a deep reinforcement learning model, and obtaining a flow detection result through multi-agent parallel learning and an integrated decision mechanism. According to the method, the abnormal traffic can be detected, discriminated and subjected to action execution, and the characteristic problem of unknown abnormal traffic is effectively solved.

The invention discloses a gearbox fault early warning system based on a wind power plant cluster operation and maintenance cooperative reasoning network, relates to a wind turbine generator early warning technology, and provides a scheme for solving the problem that the prior art is not suitable for the marine environment. The data preprocessing module is used for converting oil monitoring data of a plurality of offshore wind plant gearboxes into state indexes; the multi-agent cluster learning module captures complex fault modes in the data from different angles through parallel learning to form diversified prediction perspectives; the collaborative reasoning module integrates prediction results generated by the multi-agent cluster through a meta-learning mechanism; and the early-stage fault early-warning decision module converts the collaborative reasoning result into a specific early-warning decision, and provides probability confidence and graded early-warning suggestions. The method has the advantages that complex fault modes in data are captured from different angles through parallel learning, various prediction results generated by a multi-agent cluster are effectively integrated, and the method has important engineering significance for ensuring safe and reliable operation of a wind power plant.

The application discloses a kind of based on digital twinning turnout switch machine monitoring and fault diagnosis system and method, belong to rail transit field, including based on edge orientation and edge edge federal union's cloud edge collaborative mode to the data of turnout switch machine is collected and is governed;Establish turnout switch machine twinborn model, including three-dimensional model, behavior model and diagnostic model, and will switch machine work twinborn data integration and fusion in three-dimensional model, realize three-dimensional visualization;Based on one-dimensional power signal twinborn data and two-dimensional GASF image parallel learning fault diagnosis method diagnoses whether turnout switch machine fails.This application uses the above-mentioned based on digital twinning turnout switch machine monitoring and fault diagnosis system and method, effectively improve the switch machine operation and maintenance efficiency, reduce the operation and maintenance cost of switch machine, can be more intuitive, efficient, intelligently realize switch machine health management.

This invention proposes a multi-branch parallel learning method and system for identifying individual frequency-hopping radio stations, achieving effective identification of individual frequency-hopping radio stations. By constructing a deep residual convolutional network with multi-branch parallel learning, data processing is performed based on statistical features and decision thresholds to obtain the final identification result of the individual frequency-hopping radio station. The identification method of this invention yields accurate results and high efficiency, providing a solid foundation for communication security.

The embodiment of the invention provides a power consumption demand prediction method based on a hybrid model library and dynamic weight optimization, and belongs to the technical field of power load prediction. The power consumption demand prediction method comprises the following steps: acquiring original data about historical power consumption load, temperature and holiday and festival data, and preprocessing the original data; performing nonlinear conversion and quantitative coding on the preprocessed historical electrical load, temperature and holiday and festival data so as to construct a depth feature reflecting the physical driving strength and the social activity modulation effect of the historical electrical load, the temperature and the holiday and festival data; and inputting the depth features into a hybrid model library to drive a plurality of heterogeneous prediction models in the hybrid model library to perform parallel learning and prediction according to the depth features so as to obtain preliminary prediction results of the heterogeneous prediction models. According to the power demand prediction method, the problem that key factors such as temperature, holidays and festivals cannot be depicted sufficiently in existing feature engineering can be solved, and cross-service scene high-precision and high-robustness power demand prediction can be realized.

The invention relates to a method for detecting fraudulent transactions in a time sequence dynamic graph. The method is used for solving the problem that organized fraudulent behaviors cannot be effectively detected in the prior art. The method comprises the following steps: acquiring a new transaction event set, updating the new transaction event set to a historical event sequence, and acquiring event sequence characteristics up to a previous moment; updating the global transaction relation graph of the previous moment by using the transaction relation in the new transaction event, and updating the local entity interaction graph of the previous moment by using the transaction entity relation in the new transaction event; and carrying out parallel learning on the global transaction relation graph and the local entity interaction graph, respectively obtaining embedded features of a transaction order and embedded features of a transaction entity node, splicing the embedded features and the embedded features, aggregating the spliced embedded features and current event time sequence features, obtaining risk identification features, and using the risk identification features to identify the gang fraud probability. According to the method, cross-level and cross-relation complex graph features can be captured through parallel graph learning, recognition of gang behaviors is enhanced, and a system can find and prevent cross-transaction and cross-account fraudulent behaviors.

The invention discloses a real-time analysis method for torsional vibration of a rotating machine driven by cross-domain dynamic excitation and a related device, and relates to the technical field of monitoring of rotating mechanical performance, and the method comprises the following steps: firstly, constructing a universal node model of a rotating machine shaft system to obtain lumped parameters; meanwhile, real and virtual sampling points are arranged and spliced into virtual and real sampling grid nodes, so that the problems of poor generalization and incomplete sampling information of a traditional model are solved; acquiring actual working condition parameters of different sampling timestamps, and generating a training data set through a finite element method; training a torsional vibration analysis model comprising a multi-physics field distribution parallel learning module, a cross-domain characteristic driven torsional excitation resolving module and a dynamic directed graph-based aggregation output module, wherein the multi-physics field module captures the dynamic change of a temperature and pressure field under a variable working condition, and the torsional excitation resolving module extracts cross-domain coupling characteristics by means of an attention mechanism to improve the precision; the aggregation output module guarantees real-time performance; and finally, shaft system torsional vibration analysis is completed by using real-time working condition parameters and the trained model, and torque and relative torsion angle distribution is output. The method gives consideration to real-time performance and accuracy, and can provide support for dynamic control and predictive maintenance of the rotating machinery.

Embodiments of the present invention provide a divide-and-conquer algorithm which divides expanded data into a cluster of machines. Each portion of data is used to train logistic classification models in parallel, and then combined at the end of the training phase to create a single ordinal model. The training scheme removes the need for synchronization between the parallel learning algorithms during the training period, making training on large datasets technically feasible without the use of supercomputers or computers with specific processing capabilities. Embodiments of the present invention also provide improved estimation and prediction performance of the model learned compared to the existing techniques for training models with large datasets.

This application relates to the field of machine learning technology and discloses a method and apparatus for measuring multiphase flow based on filter enhancement. A differential pressure flowmeter module collects first data, and a capacitance tomography sensor collects second data. A multilayer perceptron fuses the first and second data to obtain fused data, reducing the impact of different sensor data on model performance. Multiple filter enhancement modules adaptively filter the fused data channel by channel, attenuating data noise and mitigating overfitting. A multi-scale convolutional neural network processes the filtered data to obtain multi-scale feature vectors, enabling parallel learning of rich flow information at different scales from random flow points. Finally, multiple fully connected layers process the multi-scale feature vectors, outputting multiple single-phase flow values ​​for the multiphase flow. Multi-task learning allows for simultaneous estimation of the single-phase flow of the multiphase flow.

This invention discloses a gearbox fault early warning system based on a collaborative reasoning network for wind farm cluster operation and maintenance, relating to wind turbine early warning technology. It addresses the problem that existing technologies are unsuitable for offshore environments by proposing this solution. A data preprocessing module converts oil monitoring data from gearboxes in several offshore wind farms into status indicators; a multi-agent cluster learning module captures complex fault patterns from the data from different perspectives through parallel learning, forming diverse prediction perspectives; a collaborative reasoning module integrates the prediction results generated by the multi-agent cluster through a meta-learning mechanism; and an early fault warning decision module transforms the collaborative reasoning results into specific early warning decisions, providing probability confidence levels and tiered early warning suggestions. The advantage lies in its ability to capture complex fault patterns from different perspectives through parallel learning, effectively integrating various prediction results generated by the multi-agent cluster, which has significant engineering implications for ensuring the safe and reliable operation of wind farms.

The invention discloses a digital twinning-based turnout switch machine monitoring and fault diagnosis system and method, and belongs to the field of rail transit, and the method comprises the steps: carrying out the collection and treatment of the data of a turnout switch machine based on an edge guiding and edge-edge federation combined cloud edge cooperation mode; establishing a turnout switch machine twinborn model which comprises a three-dimensional model, a behavior model and a diagnosis model, and integrating and fusing switch machine twinborn data into the three-dimensional model to realize three-dimensional visualization; the fault diagnosis method based on one-dimensional power signal twinborn data and two-dimensional GASF image parallel learning is used for diagnosing whether the turnout switch machine breaks down or not. According to the turnout point switch monitoring and fault diagnosis system and method based on digital twinning, the operation and maintenance efficiency of the point switch is effectively improved, the operation and maintenance cost of the point switch is reduced, and health management of the point switch can be achieved more visually, efficiently and intelligently.

The application discloses a hydraulic arm safety control method and device based on parallel learning and high-order CBF, adopts the Lagrange method to establish a hydraulic mechanical arm dynamics equation, integrates unmodeled dynamics, structural parameter changes and external disturbances into an uncertain term, and adopts a linear parameterization form to describe the generalized uncertainty; in view of the problem of insufficient excitation in a complex environment task, a parallel learning mechanism is introduced, historical and real-time data are combined to realize parameter identification; for high relative order safety constraints (such as obstacle distance constraints and joint limiting constraints) in a task space, a family of obstacle functions is constructed; a dynamic error buffer function is introduced, and a high-order adaptive control obstacle function condition is designed; the above high-order adaptive control obstacle function constraint is embedded into a real-time quadratic programming solving problem, and the obtained input can automatically correct joint instructions to prevent constraint failure while maintaining trajectory accuracy, so that the "minimum intervention" safety control is realized.

The application discloses a kind of remote sensing image space-spectrum fusion methods for spectrum non-covered waveband, first, full color image and multispectral image are preprocessed, construct the sample library for network training of reduced resolution and full resolution;Then using supervised-unsupervised parallel learning strategy, construct waveband self-guided attention module, cross-resolution cross-channel interactive learning module, multiscale residual block and the fusion network of reconstruction module, establish the loss function containing supervision constraint, dual-domain spectral constraint and cumulative frequency constraint, input low and high full color-multispectral sample pair for training until network convergence. Through the network trained, the high spatial resolution multispectral image is obtained by fusing the high-resolution full color image and low-resolution multispectral image to be processed. The application effectively enhances the spatial detail information of the spectrum non-covered waveband after fusion while maintaining the fusion effect of the spectrum covered waveband, so that all multispectral bands can achieve balanced fusion effect.

The invention relates to the field of power grid load control, discloses a virtual power plant local load regulation and control method and system based on multi-stage dynamic accurate response, and aims to solve the problems of multi-device data isomerism, inaccurate device state evaluation and static regulation and control strategy in the prior art. The method comprises the steps that a virtual power plant multi-source heterogeneous equipment unified information model is constructed, real-time and historical data are integrated, and space-time alignment is achieved; equipment operation features are extracted, the equipment state is diagnosed through a graph neural network and a series-parallel learning model, and an equipment fault feature knowledge base is constructed; and in combination with an evaluation result and a load demand, establishing a multi-level regulation and control priority system, generating a dynamic regulation and control strategy and performing closed-loop regulation and control. The system comprises a corresponding information model construction and data integration module, an operation feature extraction and state evaluation module and a multi-stage dynamic regulation and control strategy generation and regulation and control module. According to the invention, information islands are eliminated, the diagnosis precision and regulation adaptability are improved, and the operation efficiency and stability of the virtual power plant are guaranteed.