96 results about "Short-term memory" patented technology

The invention discloses a power transmission line fault diagnosis and operation and maintenance scheduling method and system based on networking learning, and relates to the technical field of power transmission line fault diagnosis operation and maintenance scheduling. Related data is extracted to construct a high-risk equipment area and a visual high-risk area thermodynamic diagram, a visual risk grading diagram is constructed in combination with electrical quantity data, and meanwhile, an intelligent recognition storage network and a fault type classification recognition model are constructed in combination with a convolutional neural network-long and short-term memory network hybrid model; the model is optimized through networking learning and an attention mechanism, maintenance teams and resources are autonomously allocated in combination with an operation and maintenance management system, then autonomous optimization and closed-loop operation are achieved, full-process coverage of fault sensing, intelligent decision making and efficient response is achieved, the response time after a line fault occurs is remarkably shortened, and the maintenance efficiency is improved. And the fault handling and operation maintenance capabilities of the power grid system are comprehensively enhanced.

The invention relates to the field of power system operation and maintenance, in particular to a power transformation equipment operation and maintenance risk online assessment method and system. A power transformation equipment operation and maintenance risk online evaluation system comprises a data acquisition module, a weight configuration module, a sequence risk evaluation module, a collaborative decision game module and a disposal scheme output module. According to the method, a multi-source data fusion and dynamic threshold mechanism is introduced, real-time state quantity, historical maintenance records, operation modes and external weather information are uniformly mapped to a convolution-long and short-term memory network, key features are adaptively amplified in a feature weighting layer, redundant features are weakened, and collaborative recognition of short-term fluctuation and long-term degradation is achieved; compared with a traditional fixed threshold value or single monitoring quantity model, the method can keep sensitive and steady risk early warning capacity under the complex working conditions of severe weather, heavy load operation and the like, the false alarm rate and the missing report rate are greatly reduced, potential faults are locked in advance, and sudden power failure events are avoided.

The invention discloses an oral cavity data monitoring and early warning method and system based on deep learning, and the method comprises the steps: solving a problem that the focus recognition is inaccurate because the image collection process of an oral cavity endoscope of a patient is affected by various environment and equipment parameters; according to the method, synchronous binding of collected images and equipment parameters and intelligent image preprocessing are proposed, historical time sequence images and variation trend features are combined, lesion space and time features are extracted through combination of a lightweight convolutional neural network and a long-short term memory network, feature weighted reasoning is realized by using a context awareness attention fusion network, and the focus space and time features are extracted. According to the method, the model parameters and the risk threshold value can be dynamically adjusted, the focus recognition accuracy and the early warning individuation are improved, and clinical grading treatment and intelligent health management are effectively assisted.

The invention belongs to the technical field of mechanical equipment wear state monitoring. The invention provides a gearbox health state stage identification method and system based on oil characteristics, and the method comprises the steps: carrying out the preprocessing of an obtained oil abrasive particle image, and extracting the oil characteristics according to the preprocessed oil abrasive particle image; according to the oil characteristics and a pre-trained long-short-term memory neural network model, obtaining a gearbox health state classification result; training of a long-short-term memory neural network model: fusing the oil characteristics of each oil abrasive particle image sample to obtain a one-dimensional health factor, and obtaining a health index sequence according to the one-dimensional health factor of each oil abrasive particle image sample; and carrying out breakpoint detection on the health index sequence to obtain three breakpoints, and adding health state labels to all the oil abrasive particle image samples according to the obtained breakpoints. According to the method, accurate division of the wear stages and capture of the evolution trend can be automatically completed, and finally decision judgment exceeding artificial experience is formed.

The invention relates to the technical field of dynamic visual target motion tracking control, in particular to a dynamic visual target motion tracking control method and system based on deep learning, and the method comprises the steps: synchronously collecting continuous multi-frame target scene image data through a visual multi-frame collection module; and performing time sequence association and memory fusion on target features in continuous multi-frame target scene image data through a cross-frame feature memory fusion module, and constructing a target feature model. According to the invention, the current and historical stable features are dynamically fused through the cross-frame feature memory fusion module, time sequence association is realized in combination with the long and short-term memory network, and the problem of slow feature model updating under target deformation and shielding is solved; the deformation-shielding bimodal recognition module accurately recognizes a scene state, provides a basis for the multi-branch Kalman filtering prediction module, enables the multi-branch Kalman filtering prediction module to call a corresponding branch, corrects a prediction equation through a compensation factor, and improves the position prediction accuracy.

The invention discloses an oil chromatography data trend classification method and system based on depth feature enhancement and an attention mechanism, and the method comprises the steps: carrying out numeralization conversion, deletion detection and grouping trend calculation on oil chromatography original gas component data, and generating a basic feature vector; executing multi-scale sliding statistics, change rate and subsequence feature enhancement, and calculating comprehensive similarity and attention weight based on a template library to generate a weighted similarity vector; splicing the enhanced feature and the weighted similarity vector into a time sequence input sequence, and outputting an oil chromatogram trend classification result after attention expansion and long and short term memory network processing. According to the method, structured processing and basic trend extraction of data are realized, adaptive matching and weighted aggregation of historical operation modes are realized, and a multi-dimensional dependency relationship and time sequence dynamic change are captured, so that accurate classification of oil chromatogram trends is realized.

The invention relates to the technical field of florescence prediction methods, in particular to a florescence prediction method based on a GOD model, which comprises the following steps: acquiring historical daily average temperature data of a target area, and preprocessing the historical daily average temperature data; based on the preprocessed historical daily average temperature data, constructing and training a long and short-term memory network model for predicting day-by-day temperature in a future time period, and outputting a day-by-day temperature sequence; on the basis of the predicted day-by-day temperature sequence, a growth day model is adopted to calculate a daily growth day and an accumulated growth day, and threshold values of an initial flowering stage and an end flowering stage are determined; the florescence length is obtained by subtracting the last florescence from the early florescence, and temperature data predicted by combining the GOD model with the LSTM model is adopted, so that high-precision phenological prediction is realized, and reliable support is provided for flower appreciation tourism and agricultural planning; the method can improve the prediction precision of the flowering phase.

The invention discloses a rainfall downscaling method and system based on a deep learning network model fusing rainfall priori knowledge, and the method comprises the steps: firstly collecting the topographic data and low-resolution day-by-day rainfall data of a target region, and taking the data as input data; a short-term high-resolution precipitation field generated in a mesoscale weather forecast WRF mode is used as training truth value data; according to the method, the function of accurately downscaling the rainfall data in combination with the convolutional neural network and the long and short term memory network is realized, the spatial-temporal correlation of rainfall is fully considered in the downscaling process, and meanwhile, a likelihood function combined with coupled censored data, Box-Cox conversion and time variation variance Gaussian distribution is adopted as a rainfall loss function; the method not only can represent zero expansibility, skewness and heterovariance characteristics of rainfall, but also can improve the rainfall downscaling precision and quantify the uncertainty of rainfall downscaling, and is suitable for wide popularization and use.

The invention discloses an express customer service voice robot multi-round task dialogue system based on a large model, which relates to the technical field of artificial intelligence and natural language processing and comprises a dialogue domain classifier, a long and short term memory module, a sub-agent module, an information source module, a dialogue management module and a large model fine tuning and data support module. The dialogue domain classifier is used for routing user query to the corresponding sub-agent module; and the long-short-term memory module is used for constructing a layered persistent memory mechanism. According to the express customer service voice robot multi-round task dialogue system based on the large model, by introducing the large language model, accurate understanding of fuzzy and spoken expression of a user is achieved, the generalization ability of intention recognition is enhanced, a multi-level memory mechanism is adopted, continuity and stability of a dialogue state are ensured, and the dialogue efficiency is improved. Even if the user jumps or asks a question, the smoothness of the conversation can be kept, and the maintenance cost is greatly reduced.

The invention discloses a dynamic load identification method and device based on a physical guidance convolution long-short-term memory network, and relates to the field of data processing, and the method comprises the steps: constructing a dynamic load identification model based on the convolution long-short-term memory network, and training the dynamic load identification model, and obtaining a trained dynamic load identification model; loss functions used in the training process of the dynamic load recognition model comprise a data driving loss function and a physical guidance loss function, the physical guidance loss function is constructed based on a kinetic equation converted from a motion equation of a linear system under the action of power, and an overall transfer matrix is used; acquiring the acceleration and / or strain response time domain signal of each test point acquired in each time window when the to-be-identified dynamic load is applied to one or more action points on the structure, and inputting the acceleration and / or strain response time domain signal into the trained dynamic load identification model to obtain the predicted dynamic load corresponding to the one or more action points. The problem that current load identification is high in dependence on data is solved.

The application discloses a power transmission line fault diagnosis and operation and maintenance scheduling method and system based on network learning, relates to the technical field of power transmission line fault diagnosis and operation and maintenance scheduling, and comprises the following steps: acquiring electrical quantity data, equipment operation state data, ring network topology and time sequence data and the like, extracting relevant data to construct a high-risk equipment area and a visual high-risk area heat map, combining the electrical quantity data to construct a visual risk classification map, combining a convolutional neural network-long short-term memory network hybrid model to construct an intelligent recognition storage network and a fault type classification recognition model, optimizing the model through network learning and an attention mechanism, combining an operation and maintenance management system to autonomously allocate a maintenance team and resources, and then realizing autonomous optimization and closed-loop operation, so that the application realizes full-process coverage from fault perception, intelligent decision and efficient response, significantly shortens the reaction time after line fault occurrence, and comprehensively enhances the fault disposal and operation and maintenance capability of the power grid system.

The invention relates to the technical field of information integrated management, in particular to a hospital information integrated management system, which comprises the following steps of: acquiring communication time delay, bandwidth and queuing cost among service nodes, and performing comprehensive calculation and optimal path screening on multiple paths by using a Dikstra algorithm; according to the method, cost-minimized dynamic path planning is realized, a shortest cost path can be efficiently determined under a complex network topology, and failure range calculation and multicast synchronization are executed through hit judgment and a conflict key value updating algorithm in combination with a segmented transmission and block allocation strategy. In the access behavior analysis, trend prediction is performed by counting the access frequency of medical records and prescription sequences and combining a long-short-term memory network model, and the access priority and the capacity threshold are dynamically adjusted, so that the balance of task scheduling and the real-time performance of response are kept during load fluctuation, a traceable cluster recovery index is formed, and the service life of the cluster is prolonged. And stable transmission and intelligent scheduling of cross-system data can still be kept in a complex environment.

The invention discloses a video intelligent monitoring system and method based on cross-modal time-frequency fusion, and relates to the technical field of new energy power monitoring and safety management, and the method comprises the following steps: obtaining multi-modal physical data; detecting a waveform characteristic of the electrical signal, and when the waveform characteristic of the electrical signal meets a preset electrical abnormal condition, performing high frame rate sampling and frame rate compensation on the visible light video to obtain video data with enhanced time sequence continuity; converting the video data, the thermal infrared image, the acoustic signal and the electrical signal to a time-frequency domain to obtain a multi-modal video feature, and inputting the multi-modal video feature to a cross-modal fusion network to obtain a fusion feature vector; and performing anomaly type prediction on the fusion feature vector based on a long-short term memory network in combination with a semantic map model to obtain an anomaly type recognition result and a risk level judgment result. According to the method, the problem that the multi-modal data lacks a unified time reference and fusion mechanism can be solved.

The invention relates to the technical field of water quality monitoring, and provides a water quality prediction method based on multi-source data and a graph attention long-short term memory network, and the method comprises the steps: obtaining the annual multi-source data of a target monitoring section, carrying out the preprocessing of the multi-source data, obtaining the to-be-detected data, and carrying out the detection of the to-be-detected data; and inputting the to-be-detected data into the trained water quality prediction model, and obtaining a water quality prediction result of a preset number of days. The water quality prediction model is trained based on multi-source sample data of a plurality of monitoring sections, a river channel topological graph structure and water quality label data, and the river channel topological graph structure is determined according to an upstream and downstream relationship of a target monitoring section in a target regional river network and is constructed based on a graph attention network and a long and short term memory network; comprising a two-channel graph attention network module, a time sequence feature extraction module and a prediction output module, the upstream and downstream relation of a monitoring section is fully considered, multi-source data are fused, spatial features and time features are fully acquired, and the accuracy of the obtained water quality prediction result is improved.

The invention discloses a fault diagnosis method and device for electrical equipment, electronic equipment and a storage medium. According to an electromagnetic interference index, the data acquisition frequency and range are dynamically adjusted, high-frequency acquisition is pertinently started for state abnormal equipment so as to accurately capture abnormal signals, and multi-source data such as partial discharge pulses are fused. An equipment state feature sequence is constructed in combination with historical operation data, and high-correlation sensitive features are screened, so that the limitation of a single parameter is avoided; meanwhile, early faults are recognized and early warning is output through a pre-trained long-short-term memory neural network, a protection constant value is rechecked in cooperation with a rule engine and a convolutional neural network, a standby loop is automatically switched or the constant value is adjusted when abnormity occurs, and a complete monitoring-early warning-response closed loop is formed; the technical effects of improving the equipment state evaluation comprehensiveness, enhancing the early fault recognition capability, improving the early warning accuracy, increasing the response speed of the protection system and guaranteeing the operation safety and reliability of the electrical equipment and the power grid of the power plant are achieved.

The invention discloses a transformer health assessment method and system based on voiceprint recognition, relates to the technical field of power equipment state monitoring, and constructs a non-intrusive assessment method by taking a voiceprint signal of a transformer as a data source of health assessment. A transformer voiceprint health recognition model is obtained through convolutional neural network-long short-term memory network combined modeling, the model is trained through voiceprint signal data generated in the operation process of transformers in different health states, the voiceprint signal data, collected in real time, in the operation process of the transformers are recognized, and real-time and accurate recognition and evaluation of the health states can be achieved. An adaptive noise cancellation algorithm removes environmental noise, and environmental noise interference factors are greatly reduced. And performing cross validation in combination with transformer operation parameters, and correcting a health state recognition result. And outputting a health assessment report containing a health state recognition result and a maintenance suggestion based on the recognition result, thereby providing effective support for operation and maintenance decision while realizing real-time, accurate and intelligent assessment of the health state of the transformer.

The invention discloses an attention reminding algorithm for intelligent driving based on multiple modes, and relates to the technical field of intelligent driving and active safety, and the algorithm comprises the steps: firstly, collecting data in real time through a multi-mode sensor group, carrying out the alignment, and then calculating the quantization features, such as the fixation deviation degree, the steering wheel disturbance entropy and the heart rate variability; then, mapping scores by adopting an S-type function, analyzing a historical sequence by utilizing a long short-term memory network to generate a dynamic weight, and performing adjustment and weighted calculation on a single-mode score in combination with an environment complexity weight to obtain a comprehensive attention score; and finally, according to the continuous driving duration, calculating a linear decreasing dynamic reminding threshold value, executing dual logic judgment in combination with the environmental risk level, and triggering graded feedback through sound and light or vibration. According to the method, the problems of poor adaptability and high false alarm rate of traditional single-mode monitoring in a dynamic environment are effectively solved, the robustness of the system is improved, and adaptive evaluation and accurate early warning of driver distraction behaviors in a complex scene are realized.

The present application relates to the technical field of machine learning, and more particularly to a medium and low voltage power supply reliability evaluation method and system based on a HLNN model. The method comprises the following steps: obtaining power supply use time sequence data corresponding to medium voltage lines, distribution transformers and low voltage users, and constructing a hierarchical long short-term memory neural network (HLNN) model for hierarchical multi-scale feature analysis to obtain line-transformer-user corresponding hierarchical different time scale power failure features; based on line-transformer-user corresponding hierarchical different time scale power failure feature analysis, the topology connection construction of line-transformer-user corresponding power supply space topology relationship is analyzed to generate a line-transformer-user power supply topology relationship graph model; based on the line-transformer-user power supply topology relationship graph model, low voltage power failure influence evaluation and user number hierarchical calculation are performed, and medium and low voltage power supply reliability evaluation is simultaneously performed to obtain line-transformer-user corresponding medium and low voltage power supply reliability measurement. The present application can ensure the accuracy of power system reliability calculation.

The invention discloses a fresh food logistics cold chain dynamic monitoring method and device and a collaborative monitoring system, and belongs to the technical field of cold chain logistics. The storage monitoring model and the logistics monitoring model are selected through the server according to the knowledge graph corresponding to the first monitoring parameter, compared with a local maintenance model, the problem that adaptability becomes poor along with use of a local maintenance processing model and an existing processing model is avoided, and the accuracy of fresh product monitoring is improved; the storage monitoring model is generated by the long and short term memory network, and the logistics monitoring model passes through the multilayer sensor, so that the control method is prevented from depending on simple rule response, the reliability of fresh product processing is improved, the safety, traceability and efficiency of fresh product distribution are improved, and the risk of food deterioration and loss is effectively reduced.

The invention relates to a multivariable time sequence prediction and dynamic optimization method and device, which are applied to real-time data prediction and control optimization in an industrial process. Time sequence data of a plurality of sensor modules are collected, deep learning models such as a long and short term memory network model are adopted for real-time prediction, and parameters of a control system are dynamically adjusted, so that production efficiency is improved, resource use is optimized, and energy consumption is reduced. The method is suitable for wide industrial processes and has high application value.

This invention relates to the field of data processing technology and discloses a method for searching similar typhoon paths based on multimodal fusion. The method includes: acquiring and preprocessing a typhoon dataset to generate candidate typhoon trajectory data; using dynamic time warping to search for similar typhoons to the target typhoon in each candidate typhoon trajectory data; using dynamic time warping combined with Bézier curve parameterization to search for similar typhoons to the target typhoon in each candidate typhoon trajectory data; using dynamic time warping combined with path bifurcation identification to search for similar typhoons to the target typhoon in each candidate typhoon trajectory data; using dynamic time warping combined with a long short-term memory network model and a random forest model to search for similar typhoons to the target typhoon in each candidate typhoon trajectory data; and finally, performing decision-level fusion on the searched similar typhoons to generate the optimal similar typhoon. This method improves the accuracy and effectiveness of similar typhoon search by considering multiple influencing factors and multimodal fusion.

This invention discloses a method and system for assessing PCCP (Pipeline Conduit Pipeline) wire breakage risk based on fiber optic sensing and a time-series model, belonging to the field of fiber optic sensing technology. The method includes: acquiring static attribute data of the pipeline and dynamic monitoring signals of wire breakage; establishing a basic prediction model for the static attribute data using a random forest algorithm, outputting an initial risk score and prior value for failure time; establishing an enhanced time-series model by combining a long short-term memory network to dynamically model the wire breakage characteristics, outputting a real-time updated predicted failure time; performing real-time iterative updates, calculating the feedback error based on the difference between the predicted and actual failure times when an actual wire breakage event is detected, triggering online learning, local retraining, and adaptive adjustment of feature weights; and correcting the prediction error through error-assisted regression to improve overall prediction stability and fault tolerance. This invention enables dynamic assessment and prediction of PCCP pipeline wire breakage risk, providing intelligent decision-making basis for operation and maintenance.

This invention relates to the field of renewable energy output prediction technology, and discloses a method, equipment, and medium for renewable energy output prediction based on multidimensional index correlation analysis. The method acquires historical renewable energy output data and preprocesses the data; establishes a multidimensional index system and performs correlation analysis and screening of the multidimensional indicators to determine key influencing indicators; the correlation analysis includes analyzing the linear and nonlinear relationships between the indicators and renewable energy output; constructs a hybrid prediction model based on convolutional neural networks and long short-term memory networks, using the preprocessed data corresponding to the key influencing indicators as input to train and predict future renewable energy output data; wherein, convolutional neural networks are used to extract the interaction features between multidimensional indicators, and long short-term memory networks are used to capture time-series features. This invention improves prediction accuracy and provides reliable technical support for grid connection of renewable energy with high penetration rates.

The invention relates to the technical field of environmental information and industrial carbon emission, in particular to a cloud analysis and scheduling method for cement telegraph pole carbon footprint calculation, which comprises the following steps: constructing a cloud unified data access layer, and collecting raw material, energy and production logistics full-link data in real time; on the basis of life cycle evaluation, eight process stages are divided, and a high-precision carbon footprint accounting model is established; performing multi-dimensional sensitivity analysis on the key process parameters through Monte Carlo simulation to generate a carbon emission scene probability map; outputting a low-carbon scheduling instruction according to the optimal scene, and driving an advanced plan scheduling system to execute emission reduction and production scheduling; and a long-short-term memory neural network is introduced to realize model self-learning. According to the invention, real-time monitoring, accounting and active optimization of carbon emission in the whole life cycle of the cement telegraph pole are realized.

The invention provides an intelligent middle-screen voice assistant family scenario personalized reply system, and belongs to the field of large model interaction application. The method comprises the following steps: firstly, establishing a family user group, establishing an independent identity mark for each user, and setting a special voice assistant; secondly, a long-short-term memory mechanism needs to be established, key information of a current session and a current intention is reserved, information abstract extraction is performed on historical dialogue records, and a dynamic memory updating mechanism is established based on dimensions such as user preferences and intention occurrence frequency; then establishing a memory visual storage and update mechanism, so that a user can conveniently open and close a memory function, manually query or update a relationship between personal information of the user and family members, and lock important information so that the important information is not covered by a daily memory dynamic update mechanism; and finally, information with the highest relevancy in the long and short-term memories is matched through the small model in each round of session to serve as large model associated memories to be transmitted, token calling is reduced to the maximum extent, and the reply response speed is guaranteed.

The present application relates to the technical field of industrial soft measurement, especially a long and short term memory network industrial soft measurement method for desulfurization process flue gas, comprising: collecting input and output data in the desulfurization process of a power plant to form a historical sample database; preprocessing the collected sample data; establishing a robust LSTM model with Huber loss and l1 regularization and performing outlier detection on the data; traversing the historical data with a sliding window, taking the difference between the data after the window and the statistics in the reference window to obtain a time series S1; calculating the upper and lower edges of the S1 box plot as the normal range; if the difference between the current value and the statistics in the reference window is not within the normal range, it is abnormal; applying the above LSTM model to SO2 concentration prediction in the desulfurization process of a thermal power plant. The simulation results show that the proposed model has good anti-interference ability for outliers and non-Gaussian noise, reduces the influence of redundant variables, thereby improving the prediction performance of LSTM, and has good economy and applicability.

The invention discloses an infantile autism spectrum disorder detection method based on a sequence transfer diagram, which comprises the following steps: firstly, carrying out channel selection on time sequence data extracted from an fMRI signal, then coding the time sequence data into MTF image data, and locking time and space modes in the data at the same time, thereby breaking through the limitation of traditional static analysis. Secondly, inputting the generated MTF image data into a parallel convolutional neural network for learning, and automatically learning difference features between different ROIs by using the strong feature extraction and mode recognition capabilities of the MTF image data; and finally, inputting all the features into a long and short term memory module with an attention mechanism to model a nonlinear interaction relationship among different ROIs, thereby constructing a high-precision classification model. According to the method, the fMRI data is coded into the MTF image, so that the problem that complex time features are ignored and the problem of linear hypothesis limitation during data processing in the existing method are effectively solved.

This invention belongs to the field of wastewater treatment control, specifically relating to a collaborative control system for wastewater treatment equipment based on the Internet of Things (IoT). The system includes a multi-dimensional water quality sensing device, an edge computing gateway device, a long short-term memory neural network prediction device, a fuzzy logic decision controller, an aeration frequency converter, and a cloud-based monitoring and collaboration platform. This invention collects water quality parameters in real time through the sensing device. After preprocessing by the gateway, the neural network prediction device deeply analyzes the dissolved oxygen evolution trend. The fuzzy logic decision controller combines the predicted trend with real-time feedback to perform inference calculations, driving the aeration frequency converter to dynamically adjust the air supply. By deeply integrating time-series prediction capabilities with expert decision-making experience, this invention constructs a predictive sensing and decision-making closed loop, solving the problem of large time-delay control lag in the wastewater biochemical treatment process. This ensures stable effluent compliance while reducing aeration energy consumption and improving system robustness.

The invention discloses a power distribution network line environment meteorological data prediction method and system, and belongs to the technical field of meteorological prediction, and the method comprises the steps: S1, carrying out the space-time alignment and quality control of collected historical environment meteorological data of a power distribution network line; s2, taking the lightweight long-short-term memory network as a learning framework of a time sequence evolution rule in the historical environmental meteorological data, and taking the physical constraint as a learning constraint of the time sequence evolution rule in the historical environmental meteorological data to construct a hybrid prediction model; and S3, inputting the historical environmental meteorological data into the hybrid prediction model for multi-scale space-time prediction to obtain an environmental meteorological field of each interval of the power distribution network line, obtaining a confidence interval of each interval according to the environmental meteorological field, and obtaining predicted environmental meteorological data of each interval according to the confidence interval. The technical problem that the environmental weather prediction accuracy is difficult to improve in the prior art is solved.