40results about How to "Solve vanishing gradient" patented technology

The invention provides a lithium battery capacity prediction method based on empirical mode decomposition and data fusion driving, and the method comprises the steps: S1, obtaining historical capacity degradation data of a certain lithium battery, and forming a source data set; s2, iteratively screening the source data set by using empirical mode decomposition, and decomposing historical capacity degradation data of the source data set into a residual sequence and a plurality of intrinsic mode functions; s3, constructing a long-short-term memory neural network source prediction model, and pre-training the constructed source prediction model by using the residual sequence and the intrinsic mode function data obtained in the step S2; s4, performing parameter fine tuning on the pre-trained source prediction model to realize transfer learning, and obtaining an optimal prediction model; and S5, predicting the future capacity of the target lithium battery by using the optimal prediction model obtained in the step S4. The method can effectively predict the future capacity and residual life of the lithium battery.

The invention discloses a sewage treatment process optimization method and device based on deep learning, and the method comprises the steps: carrying out the sample generation based on a generative adversarial network algorithm of hybrid differential privacy, and employing a reverse feature refinement strategy to adjust the data features in the sample generation process; performing feature extraction by adopting a full-connection neural network based on an adaptive wolf pack optimization algorithm; adopting a self-encoding neural network based on mask enhancement to selectively hide partial features of input data in an input layer by adopting a random mask to perform data compression and reconstruction so as to realize feature dimension reduction; and adopting a random forest algorithm based on model capacity self-adaptive pruning to perform classification prediction on the data after feature dimension reduction to obtain a classification result of the sewage treatment degree. According to the method, the sewage treatment process is integrally optimized and improved, the sewage treatment process can be monitored and dynamically optimized in real time after improvement, the sewage treatment efficiency is improved, the sewage treatment effect is improved, and it is ensured that the sewage treatment process meets the environmental protection standard.

The invention discloses a tooth instance segmentation method based on nonlinear features, and belongs to the technical field of image processing and deep learning, and the method comprises the steps: firstly collecting an oral X-ray film of a patient, carrying out the image quality evaluation of the X-ray film, carrying out the denoising of an unqualified image, making a mask image according to an FDI tooth position representation method, and carrying out the segmentation of a tooth instance; a tooth instance segmentation network with nonlinear features is constructed and trained for instance segmentation of tooth images, and a backbone network of the tooth instance segmentation network is composed of an encoder module, a reverse mechanism module, a dense expansion module, a decoder module, a splicing module, an output module and a receptive field feature convolution kernel attention mixing module. According to the invention, the method can be used for generating a more accurate instance segmentation result, overcomes the limitation of a traditional method when tooth pictures are complex and tooth types are various, carries out the index evaluation of a training result after training is finished, outputs a segmentation result after a set reference index is met, and achieves the high-precision recognition and segmentation of teeth.

The invention discloses an audio gender recognition method based on an LSTM architecture neural network, and the method comprises the following steps: 1, carrying out the audio preprocessing, and carrying out the sampling rate conversion, noise elimination and voice segment positioning; 2, feature extraction: extracting features from the audio signal, and taking the features as the input of an LSTM network; 3, constructing and training an LSTM (Long Short Term Memory) model; and 4, gender classification: after model training is completed, performing gender classification on a new audio signal by using the constructed model. The invention aims to overcome the limitation of the prior art and improve the accuracy and adaptability of gender recognition, especially in noise environment and cross-age recognition tasks. By introducing an advanced deep learning technology and an innovative feature engineering strategy, the speaker audio gender recognition method is devoted to promote the development of a speaker audio gender recognition technology, so that the speaker audio gender recognition technology can be applied in a wider range of scenes.

The present invention relates to the technical field of sleep diagnosis, and in particular to a sleep staging method based on deep learning, the method comprising: collecting a number of target physiological signals, preprocessing each target physiological signal to obtain a number of standard physiological signals; synchronously aligning and integrating each standard physiological signal in turn to obtain a standard sleep data set; determining the actual correlation between each physiological signal to be integrated in the standard sleep data set, and determining an actual signal combination strategy according to the actual correlation; generating an actual sleep data set according to the actual signal combination strategy, and extracting features from the actual sleep data set to generate actual feature parameters; inputting the actual feature parameters into a deep learning model to determine the actual sleep state; analyzing the actual sleep state, and performing corresponding sleep intervention according to the analysis results. The present invention realizes low-cost non-intrusive sleep monitoring, thereby improving people's physical health.

The present invention relates to the field of network traffic classification, and in particular to a network traffic classification method based on an attention mechanism. Aiming at the characterization problem of different important features, this paper proposes a new classification model, which first pre-processes the original network traffic, then extracts spatial features through a CNN layer, extracts temporal features from an LSTM layer, then fuses the extracted multi-scale features, and uses an attention mechanism to extract multi-modal network traffic significant features to improve the output quality of dynamic structures and improve feature representation capabilities. Through the design of the present invention, the classification accuracy can be ultimately improved.

The invention discloses a residual neural network-based transient attraction domain construction method for a network construction grid-connected interconnection system. The method comprises the steps of S1, establishing the network construction grid-connected interconnection system; s2, on the basis of the residual neural network, constructing a Lyapunov function of the networking grid-connected interconnection system; and S3, according to the Lyapunov function, constructing the transient stability attraction domain of the networking grid-connected interconnection system, and according to the application, the Lyapunov function of the heterogeneous system can be established, and the transient stability attraction domain of the system can be constructed based on the Lyapunov function.

The invention provides a vehicle driving obstacle distance detection method and system based on a vehicle-mounted depth camera, and relates to the technical field of vehicle obstacle detection. Comprising the following steps: acquiring RGB image information, depth point cloud information and 3D coordinate information of a to-be-identified obstacle during vehicle driving; geometric features of key points in the RGB image information are extracted, and obstacles in the key points are recognized; aligning and integrating the 3D coordinate information and the depth point cloud information according to a timestamp, and combining the identified obstacle to obtain a 3D coordinate with the timestamp of the obstacle; and learning spatial features of different scales of the 3D coordinates with the timestamps of the obstacle, and performing feature extraction to obtain the distance between the obstacle and the vehicle body. According to the method, the RGB image information, the depth point cloud information and the 3D coordinate information of the to-be-recognized obstacle in the running process of the vehicle are acquired by using the depth camera installed on the vehicle body, and classification and regression are performed after different feature information is extracted by using the hybrid network model, so that the effects of efficiently and accurately recognizing the obstacle and predicting the distance between the obstacle and the vehicle body can be considered.

The present invention relates to the field of high-speed wireless communication, and in particular to a deep learning-assisted OFDM channel estimation and signal detection method. The present invention first generates a data set through an OFDM system, then uses a self-normalizing network (SNN) to build a channel estimation model, and uses a convolutional neural network (CNN) and a bidirectional gated recurrent unit (BiGRU) to build a signal detection model; secondly, the data set is input into the neural network, the model is trained, and the network is optimized using an Adam optimizer and a mean square error loss function, and finally compared with LS, BiLSTM, GRU and DNN+BiLSTM models. The present invention effectively solves the problems of low channel estimation and signal detection accuracy, severe network loss and poor bit error rate performance in OFDM systems with a small number of cyclic prefixes and pilots. The simulation results show that in the case of fewer cyclic prefixes and pilots, the performance of this method is better than that of LS, BiLSTM, GRU and DNN+BiLSTM models.

The invention discloses a pilot fatigue classification method based on an EEG signal, and the method comprises the following steps: carrying out the frequency band decomposition of the EEG signal, and extracting differential entropy features corresponding to different frequency bands; a linear dynamic system (LDS) is used for filtering features to remove components irrelevant to fatigue, a deep ResTaylor Network (DRN) model is constructed, the problems of overfitting and gradient disappearance in high-degree polynomial processing are effectively reduced through a residual connection mechanism, and adaptive adjustment of model parameters is carried out in combination with an Adam optimization algorithm; and training the DRN model based on the training set, and verifying on the test set to obtain a pilot fatigue classification result. Through the pilot fatigue state classification method and device, accurate classification of the pilot fatigue state can be achieved, and compared with the prior art, the pilot fatigue state classification method and device have higher classification accuracy and robustness.

The invention proposes a power distribution network fault tracing method based on a graph convolutional neural network, and relates to the technical field of electric power, and the method comprises the steps: S1, constructing a power distribution network graph model, simulating a fault condition during the access of a high-proportion distributed power supply through the power distribution network graph model, so as to obtain power flow simulation data, and carrying out the data preprocessing; s2, constructing a graph convolutional neural network model containing a high-proportion distributed power supply according to the preprocessed power flow simulation data; s3, introducing the residual error into the image convolutional neural network model to construct a residual error-containing image convolutional neural network model; s4, carrying out the fault tracing of the power distribution network based on the residual image-containing convolutional neural network model; according to the invention, the node dependence problem in a complex power grid topological structure can be solved, the fault repair efficiency is improved, and the reliability and safety of power supply are realized.

The invention relates to the field of petrochemical engineering and the field of computers, in particular to an ethylene cracking furnace tube abnormal working condition prediction method and an ethylene cracking furnace tube abnormal working condition prediction system, and the method comprises the following steps: obtaining ethylene cracking furnace tube parameters as sample data; preprocessing the ethylene cracking furnace tube parameter sample data to obtain preprocessed sample data; classifying the preprocessed sample data to obtain a classification result, wherein the classification result comprises a data set with a label and a data set without a label; training a prediction model by using the classification result to obtain a trained prediction model; acquiring real-time parameter data of the furnace tube of the ethylene cracking furnace; and preprocessing the real-time parameter data of the furnace tube of the ethylene cracking furnace, and inputting the preprocessed data into the trained prediction model for prediction to obtain a prediction result. According to the embodiment of the invention, the abnormal working condition of the furnace tube of the ethylene cracking furnace is predicted, and favorable conditions are provided for ethylene production.

The application discloses a sea surface weak target detection method based on a residual network and a hypersphere constraint. The method can solve the problems of training underfitting or overfitting caused by the fact that the target sample is much smaller than the sea clutter sample under the sea clutter background, and the specific steps include: 1, performing short-time Fourier transform on each piece of data obtained after dividing the radar echo signal sample to obtain a time-frequency spectrum; 2, constructing an anomaly detection network model combined with a support vector data description based on a variational autoencoder of a residual network; 3, inputting the time-frequency spectrum into the anomaly detection network model for end-to-end training; 4, performing online detection of the target based on the trained model; and 5, evaluating the performance of the application from the constructed multi-dimensional index. The application can effectively enhance the separability between the target and the sea clutter, and improve the weak target detection precision under the condition of a real small sample.

The present invention proposes an epileptic EEG signal monitoring method based on a spatiotemporal attention mechanism, which belongs to the field of signal processing. It includes: obtaining multi-channel EEG signals and preprocessing them; constructing a graph attention network and a transformer network, using the output of the graph attention network as the input of the transformer network, and constructing a fusion network; constructing a training set based on the preprocessed multi-channel EEG signals, and using the training set to train the fusion network to obtain a trained fusion network; inputting the multi-channel EEG signals into the trained fusion network, first extracting spatial features through the graph attention network, and then inputting the EEG signals with aggregated spatial features into the transformer network to extract temporal features, and obtaining classification results. The present invention solves the problems of gradient disappearance and gradient explosion during long sequence training, and the problem of information loss during two-dimensional EEG processing, selects the most critical information from epileptic EEG signals, and then selects the optimal features to improve the accuracy of epilepsy detection.

The present invention relates to the field of artificial intelligence, and specifically refers to a caregiver behavior recognition and intervention system based on artificial intelligence, including integrated equipment, data acquisition and preprocessing modules, behavior recognition and analysis modules, intervention and response modules, user interface modules and security and privacy modules. The present invention proposes that in the behavior recognition and analysis module, a COA-CNN-LSTM model is used to analyze the preprocessed data, identify and classify the behaviors of caregivers and patients. The COA-CNN-LSTM model combines the characteristics of convolutional neural networks and long short-term memory networks, and improves the recognition accuracy of the behaviors of caregivers and patients. In the behavior recognition and analysis module, a VGG-ICNN model is used as a basic architecture to construct a behavior recognition model, thereby improving the feature representation capability of the behavior pattern, better understanding and predicting the behaviors of caregivers and patients, and improving the intelligence level and service quality of the system.

The present application provides a smart energy industry internet data management method, system, medium and server, which relates to the technical field of data management, wherein the method includes the steps of first data acquisition and processing, first modeling, first model training and first production forecasting; the system includes a data acquisition and processing module, a modeling module, a model training module and a production forecasting module. When predicting the production value, the present application uses a white box model to extract features based on physical laws and chemical reaction principles, further improving the prediction accuracy.

The application provides a kind of based on feature enhancement and composite model's ultra-short-term wind power prediction method, belongs to ultra-short-term wind power prediction technical field, this method includes: by calculating the mutual information of multiple meteorological characteristics and wind power output and normalizing, screening to obtain key features, and constructing wind power prediction fitness function, by minimizing wind power prediction fitness function, obtain enhanced input features;By introducing one-dimensional convolutional neural network and bidirectional long short gate recurrent unit, construct the residual network of bidirectional long short gate recurrent unit and convolutional neural network fusion, and introduce space-time attention mechanism, construct to obtain composite residual network ultra-short-term wind power prediction model;And use multiple meteorological characteristics and enhanced input features to predict to obtain ultra-short-term wind power prediction result;The application solves the problem that the influence of meteorological multi-scale fluctuation on output cannot be accurately grasped, prediction is disconnected, disturbance is not considered and prediction accuracy continues to decline with step increase.

This invention provides a traffic flow speed prediction method and system based on directed hypergraphs and attention mechanisms, belonging to the field of traffic state prediction technology. Based on directed hypergraphs, it constructs the relationships between nodes in a road network; aggregates directed hyperedge information to characterize the complex spatiotemporal features of the road network; builds an encoder and decoder architecture; integrates an attention mechanism, constructs an attention module, and introduces an improved dense connection structure to enhance the method's accuracy. This invention effectively overcomes the shortcomings of methods based on traditional graph structures, achieves the extraction and fusion of temporal and spatial features, and solves problems such as gradient vanishing in deep neural networks. It is applicable to multi-step traffic flow prediction in real road networks and achieves good prediction accuracy.

This invention provides a method, apparatus, equipment, and medium for predicting oil and gas field flow data. The oil and gas field flow correction method includes the following steps: S1, acquiring and preprocessing wellhead data of oil and gas wells; S2, identifying influencing factors of wellhead flow data by analyzing the correlation between the wellhead data and wellhead flow data; S3, identifying influencing factors of wellhead flow data by analyzing the correlation between the wellhead data and wellhead flow data; S4, establishing a hybrid virtual metering model of oil and gas wells based on the aforementioned mechanism model, and labeling the characteristic variables of the hybrid virtual metering model according to the influencing factors; S5, predicting wellhead flow data based on the hybrid virtual metering model. This invention, by using existing data for mining and analysis, can achieve accurate measurement of wellhead flow, replacing the current complex back-matching mechanism, and can enhance the core competitiveness in the field of efficient oil and gas field development.

The present application relates to a GNSS model adaptive optimization method for enhanced navigation scene perception in high-precision positioning, including obtaining GNSS observation data, extracting and calculating GNSS data features, constructing a time-series GNSS feature vector of GNSS data features through the SFFS algorithm; constructing an LSTM neural network with time memory function based on the GNSS feature vector; training the LSTM neural network in combination with GNSS observation data and scene labels preset for navigation scenes to obtain an LSTM model; inversely calculating the residuals of pseudorange and carrier phase; fitting the functional relationship between the residuals and preset influencing factors to construct GNSS random models for different classification scenarios; constructing threshold models for different classification scenarios based on preset different classification scenarios and GNSS observation data, the threshold model including carrier-to-noise ratio, satellite altitude angle and ratio value. The present application has the effect of improving the availability, accuracy and continuity of GNSS positioning services in complex scenarios to meet the GNSS positioning needs in dynamic and complex environments.

The invention relates to a natural gas long-term demand quantity prediction method and system, processing equipment and a storage medium, and the method is characterized in that the method comprises the steps: obtaining the time series data of a part of historical natural gas demand quantity of a to-be-measured region; encoding the obtained time sequence data, inputting the encoded time sequence data into a pre-trained natural gas demand prediction large model, and generating a natural gas demand prediction result of the to-be-detected region; wherein the natural gas demand quantity prediction large model is obtained by training a large language model based on a plurality of time series data including natural gas demand quantity, the method combines the context learning ability of the large language model, can flexibly adapt to sudden changes and fluctuations of a natural gas demand market, provides a dynamically adjusted and optimized prediction result, and improves the prediction efficiency. The method can be widely applied to the natural gas field.

The present invention provides a natural language knowledge extraction method and device based on the collaboration of large and small models, which relates to the field of data processing technology. After converting a natural language text file into a vector format, it is input into a pre-constructed small model, and a recognition result is output, thereby determining the event indicated by the text file. Among them, the small model optimizes the parameters of the neural network based on the circular energy flow optimization algorithm, which enhances the adaptability and stability of the network at different training stages. In addition, it is constructed using an extreme learning machine classification algorithm based on a dynamic weight adjustment mechanism to improve the accuracy of classification. The training sample set for training the small model is constructed after event extraction from the initial sample set by a large natural language model. The present invention uses a large model to perform preliminary event extraction and refined training of the small model, which can efficiently process a large amount of data while maintaining high accuracy, especially in the extraction of specific events and entities.

This invention proposes a method for denoising vibration signals from structural health monitoring based on a hybrid neural network. This method, implemented according to embodiments of the invention, eliminates the need for prior signal knowledge and manual parameter settings. It effectively removes various noise types from structural health monitoring vibration signals, significantly improves the signal-to-noise ratio, and reduces the root mean square error, achieving efficient and automated noise reduction of vibration signals.

The invention discloses a flue gas desulfurization slurry pH value prediction method based on a time convolution gated recurrent neural network, and the method comprises the steps: analyzing a flue gas desulfurization process, and combining a large amount of process monitoring data, a maximum mutual information analysis method is adopted to carry out correlation analysis on process variables such as flue gas pressure, inlet flue gas temperature, flue gas oxygen content, flue gas fly ash content, inlet flue gas sulfur dioxide concentration, inlet flue gas flow and the like, and key variables influencing the pH value of slurry are obtained. And then, combining the time convolution neural network with the gating circulation unit to form a time convolution gating circulation network. And finally, completing the prediction of the pH value of the slurry in the desulfurization process by using the selected key variables and the convolution gating circulation network. The pH value is monitored and adjusted in real time, shutdown can be prevented, and system reliability and economic benefits are improved. Therefore, accurate prediction of the pH value of the slurry has important significance on stability, efficiency and environmental protection performance of the flue gas desulfurization process.

The application provides a WGAN-based FRP sheet and concrete interface bonding slip model generation method, which can automatically, quickly and accurately obtain the bonding slip model and has a wide application prospect. After the automatic and rapid acquisition of the bonding slip model, the bonding performance between the FRP sheet and the concrete can be accurately reflected, and a safe and reliable reinforcement component design calculation method can be established. The WGAN is used to predict the strain, so that the training process can be simplified and the training can be stable. The WGAN can replace the traditional experimental analysis to quickly and accurately establish a bonding strength model. The LSTM is used as the generator of the WGAN model, so that the gradient disappearance and explosion problems of the RNN are solved, and the strain data related to time can be more accurately predicted. The CNN is used as the discriminator of the WGAN model, so that the quality and convergence speed of the generated samples are improved.

The invention provides a multi-site air quality prediction method based on adaptive hierarchical graph convolution, and belongs to the technical field of air quality prediction. The invention discloses a multi-site air quality prediction method based on self-adaptive hierarchical graph convolution, which is used for solving the problem that the real relationship between sites cannot be accurately represented by the current PM2.5 prediction method. According to the method, a hierarchical mapping graph convolution architecture is introduced, and different self-learning adjacency matrixes are used on different hierarchies, so that unique space-time dependence among different sites is effectively mined. And then connecting upper and lower adjacent matrixes based on an attention aggregation mechanism, and accelerating a convergence process. And finally, combining the hidden space state with a gating circulation unit to form a unified prediction architecture, capturing a multi-level space dependency relationship and a multi-level time dependency relationship, and providing a final prediction result. The invention provides a new thought and method for air quality prediction, and is helpful for better monitoring and controlling air pollution.

This invention relates to a method for diagnosing UAV motor faults based on dilated convolution residuals. The method comprises a data preprocessing module that converts one-dimensional vibration signals collected during UAV motor operation into two-dimensional grayscale images; a feature extraction module that extracts features from the input grayscale images to generate multi-channel feature maps; and a fault classification module that compresses the feature dimensions of the multi-channel feature maps, performs dimensionality mapping, and outputs the probabilities of various fault types, which constitute the fault diagnosis results. This invention improves fault identification accuracy by expanding the receptive field through dilated convolution and focusing on key features using a CBAM attention module, effectively capturing early, weak fault signals in the motor. The grayscale image conversion method in the data preprocessing stage does not require manual preset feature parameters, and the model can adaptively learn motor fault features under different operating conditions (such as different speeds and loads), with a generalization error of less than 5%.

The invention belongs to the technical field of gas cylinder pressure detection, and provides an aviation gas cylinder high-precision pressure detection method based on computer assistance. According to the method, firstly, comprehensive data are collected through monitoring equipment, a high-precision sensor and pressure detection equipment, feature engineering processing is carried out on the comprehensive data, and high-quality feature data are extracted; afterwards, an AE-MSGRU model is adopted for pressure prediction, the model captures short-term, middle-term and long-term dependency relationships of gas cylinder pressure data through a multi-stage GRU layer, key points in the pressure data are accurately recognized in combination with an attention mechanism, and time sequence characteristic data are further optimized; through the above steps, the method effectively improves the precision and reliability of the pressure detection of the aviation weather cylinder, ensures the safety of the gas cylinder under different use conditions, and provides powerful technical support for the pressure detection of the aviation weather cylinder.

The present invention relates to the field of traffic management technology, and in particular to a method and system for emergency control of highway toll stations based on deep learning. The method includes obtaining vehicle data information, including obtaining historical vehicle data and real-time vehicle data of a highway section, preprocessing the obtained vehicle data, including cleaning and smoothing the obtained vehicle data, and normalizing the processed vehicle data; constructing a prediction model based on the preprocessed vehicle data, training the model according to the constructed prediction model, inputting the obtained real-time vehicle data into the trained prediction model, and obtaining the prediction result of the traffic volume in the future time; generating an emergency control strategy according to the prediction result, and the present invention can obtain vehicle data in real time and input the trained prediction model in time, so as to alleviate traffic congestion to the greatest extent and improve traffic efficiency.

This invention discloses an intelligent active fault-tolerance method for gradually changing faults in navigation sensors, comprising: constructing and training a standard signal prediction module using long short-term memory neurons as basic units to predict navigation signals; constructing and training a gradually changing fault diagnosis module using a multi-layer stacked convolutional neural network model, which extracts gradually changing fault trend information based on the residual between the predicted signal and the measured signal, and diagnoses the gradually changing fault in its early stages when its impact on the system is within a set range; if no gradually changing fault is diagnosed, the measured signal is used as the navigation signal; when a gradually changing fault is diagnosed, the predicted signal from the standard signal prediction module is used as the navigation signal, thus achieving active fault tolerance for gradually changing faults in the navigation signal. This invention can effectively predict lidar altitude information and solves the problems of slow gradually changing fault diagnosis and the limited application scenarios of traditional threshold methods.