14results about How to "Solve the degradation problem" patented technology

The invention discloses a vehicle speed prediction method and system in a dynamic vehicle following scene considering a traffic environment, and the method comprises the steps: extracting vehicle following case data through setting a standard and a threshold value of a vehicle following process in a natural driving data set; using a target detection algorithm to obtain traffic environment characteristics related to the vehicle following case data, extracting vehicle dynamics parameters in the vehicle following case data, fusing the vehicle dynamics parameters and the traffic environment characteristics, and performing normalization processing to obtain fusion characteristics; removing redundant variables in the fused features by adopting a combined feature selection method; and dividing the fusion features from which the redundant variables are removed into a data set, training a pre-established Informer-FDR vehicle speed prediction model, and obtaining a vehicle speed prediction result by using the trained vehicle speed prediction model. The method effectively reflects the dynamic interaction characteristics of the driver and the traffic environment and potential traffic conflicts, enhances the understanding of the model for the complex driving environment, and improves the real-time performance and accuracy of vehicle speed prediction.

The invention discloses a method and a system for monitoring pressure abnormity of a fuel oil common rail pipe of a marine main engine. The method comprises the following steps: acquiring historical operation parameter data and constructing a data set; the historical operation parameter data comprises historical host load data, historical host rotating speed data and historical fuel oil common rail pressure data; historical operation parameter data are preprocessed, an SVDD model is constructed and trained based on the historical operation parameter data, the model obtains a historical normal data hyper-sphere in a three-dimensional feature space, and the center and the radius of the hyper-sphere are obtained; similarly, real-time operation parameter data collected in real time are preprocessed and then input into the model, the distance between the real-time operation parameter data and the center of the hypersphere is obtained, and the value of the distance is compared with the value of the radius of the hypersphere in historical normal data; and when the numerical value of the distance is large, it is judged that the real-time operation parameter data is within an abnormal range, the fuel oil common rail pressure of the ship main engine is in an abnormal state, and an alarm signal is triggered. The method has the effect of low error rate.

The invention discloses a data-driven heavy-load locomotive primary steel spring damage detection method, and belongs to the technical field of heavy-load locomotive primary spring damage detection. The detection method comprises the following steps: S1, obtaining a vertical vibration response signal of a framework of the heavy-load locomotive, and preprocessing the vertical vibration response signal to obtain a sample set; s2, dividing the sample set into a training set, a test set and a verification set, and performing standardization processing; s3, constructing a primary spring damage detection model; s4, training and adjusting the primary spring damage detection model by adopting the training set and the verification set to obtain a final primary spring damage detection model; and S5, performing evaluation verification on the final primary spring damage detection model by adopting the test set to realize primary steel spring damage detection of the target heavy-load locomotive. The method can realize intelligent grading detection of the damage of the primary spring of the heavy-duty locomotive, has the characteristics of accuracy, high efficiency, strong generalization ability and the like, and is expected to be developed into a long-term monitoring method for the damage of the primary spring.

The application relates to the technical field of deep learning and recommendation system, and particularly discloses a hybrid expert model sparse inference method and system for generative recommendation, which comprises the following steps: obtaining original input data, and obtaining an input vector through an embedding layer; obtaining hybrid expert weights through normalization and maximum value selection operation on the input vector; calculating attention-enhanced features according to the hybrid expert weights based on a hierarchical attention mechanism; inputting the input vector and the attention-enhanced features into a hybrid expert model to calculate a recommendation result; wherein, according to the attention-enhanced features, expert weights are calculated based on a parallelized gating mechanism activated by a Sigmoid activation function; the activated expert layer is selected according to the expert weights, and the input vector is used to calculate the recommendation result. The application can achieve a recommendation accuracy comparable to or even higher than that of an advanced dense model under low time overhead and low calculation complexity.

The present application relates to the field of optical interferometry, and aims at the phase distortion problem in off-axis interferometric quantitative phase imaging, and provides an interferometric microscopic phase distortion elimination method based on PACUNeXt3+ network. The method comprises the following steps: 1. using Zernike polynomials and test target pictures to simulate and generate a data set; 2. establishing and training a PACUNeXt3+ neural network model; 3. inputting the interference graph I or of the sample to be measured into the trained neural network, and outputting the background interference graph I' r corresponding to the sample without sample information; 4. using the two interference graphs I or and I' r to reconstruct the sample phase distribution φ o (x, y) without phase distortion. The present application has high precision and speed, can eliminate the secondary or high-order phase distortion in interferometric quantitative phase imaging, and has great application prospect in the field of phase imaging.

The present invention discloses a method for object segmentation based on a residual module and an attention mechanism and a related device, which relates to the technical field of remote sensing object segmentation. The method comprises the following steps: obtaining a remote sensing image to be segmented; inputting the remote sensing image to be segmented into a trained object segmentation model to obtain an object segmentation result; wherein the object segmentation model is a network model obtained by combining a residual module and an attention module based on a U‑Net neural network. The present invention adopts a U‑Net model combining a residual network structure and an attention mechanism. The main idea is to use an improved residual structure to avoid the degradation problem of a deep network model. In addition, a hybrid attention mechanism and a cross attention mechanism are introduced, so that the model has the ability to connect to long-distance contextual information, can make more full use of the global information of the image, and enhance the network's adaptive ability.

This invention discloses an image segmentation method and medium for densely populated live cell regions. The method includes acquiring multiple first data pairs and preprocessing each first data pair. Each first data pair includes a single-modal master image of a live cell region and a paired real auxiliary modality image. A feature extraction network is trained using the preprocessed first data pairs, and the feature extraction network outputs a predicted auxiliary modality image corresponding to the single-modal master image. The preprocessed first data pairs and the predicted auxiliary modality image are input into a pre-trained segmentation model, and the dynamic weights of the predicted auxiliary modality image are adjusted to optimize a second loss function to obtain optimal weight parameters. The single-modal master image of the live cell image to be segmented is input into the trained feature extraction network, and combined with the optimal weight parameters, it is input into the pre-trained segmentation model to obtain the segmentation result. This method is particularly effective for accurate segmentation in densely populated regions.

This invention discloses a soft-constraint target tracking method and system based on genetic resampling, belonging to the field of target tracking technology. The method includes: constructing a target tracking model based on the target being tracked; generating multiple particles required for particle filtering based on the initial state value of the target being tracked; updating each particle over time based on the target tracking model to predict the state and weight of each particle; resampling each particle using a genetic algorithm; calculating the corrected likelihood function of the resampled particles based on the measurement equation; correcting the predicted weights based on the corrected likelihood function; and updating the predicted state based on the corrected weights to obtain the final estimated state of the target being tracked. This invention can integrate known nonlinear inequality soft constraints in the tracking of ground moving targets, improving the tracking accuracy of ground targets under such soft constraints.

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.

The invention relates to the technical field of load identification, in particular to a sequence mapping load identification decomposition method based on a multi-scale residual neural network, which comprises the following steps that: a load identification model is constructed, and the network structure of the load identification model is a multi-scale residual network based on expansion convolution; acquiring an aggregation power data set and an electric appliance level power data set in the public data set, and preprocessing the aggregation power data set and the electric appliance level power data set to obtain a preprocessed training set and a preprocessed test set; and training and testing the load identification model by using the preprocessed training set and test set to obtain a trained load identification model. According to the method, the load identification model of the multi-scale residual network based on the expansion convolution is constructed, so that mixed data features can be learned in a more targeted manner, and the load identification effect of the load identification model is improved.

This application discloses an occlusion detection method and apparatus combining dynamic degradation decomposition and physical restoration, relating to the fields of computer vision and industrial safety monitoring technology. The method includes: extracting multi-scale features from industrial scene images using a dynamic degradation decomposition backbone network; explicitly decomposing and suppressing the dynamic degradation patterns of the industrial scene images in manifold space; generating degradation correction prompts and strategy prompts; inputting these prompts into a physically guided dehazing neck network; and using the physical inductive bias of an atmospheric scattering model to decouple and restore the multi-scale features, generating a multi-scale restored feature map; and using a context relation module to perform semantic reconnection operations on the multi-scale restored feature map to generate a context-enhanced feature map, which is then input into a decoupled detection head to predict the location and category of industrial personal protective equipment (PPE) in the industrial scene image. This solves the problem of low accuracy in detecting the wearing of industrial PPE in high-risk industrial scenes, which poses safety risks in industrial production.

This disclosure relates to the field of face recognition technology, and provides a training method and apparatus for a face recognition model that avoids the long tail problem of data. The method includes: constructing a face recognition model; obtaining a training dataset, and executing the following loop to train the face recognition model in multiple rounds: sampling the current round of training from the training dataset using a dynamic sampler to obtain a sample set used for the current round of training; inputting the sample set into a feature extraction network to obtain a feature vector set corresponding to the sample set; inputting the feature vector set into a normalization network to normalize the feature vectors in the feature vector set; calculating a classification loss using a classification network and a contrastive loss using a contrastive network based on the feature vector set processed by the normalization network; updating the model parameters of the face recognition model based on the classification loss and contrastive loss; incrementing the training round number corresponding to the current round of training by one; and ending the loop when the training round number equals a preset round number.

This invention discloses a method, system, device, and medium for hierarchical perception modeling of API behavior based on multimodal feature fusion, belonging to the field of API modeling technology. The method includes: processing API call behavior data through multimodal feature fusion to generate a fused feature matrix and modal weights; performing multi-level semantic encoding based on temporal sequence and dependency relationships to form a behavioral semantic graph and hierarchical feature representation; calculating behavioral deviation to generate risk scores, anomaly lists, and grading results; monitoring new data and adaptively updating the model, outputting model iteration versions and drift reports; and performing interpretable analysis based on risk scores, weight distribution, and semantic graphs to generate a visualized risk association view. This invention improves the accuracy of panoramic understanding and anomaly identification of complex API behaviors through multimodal feature fusion and hierarchical semantic modeling, ensuring stable detection performance in dynamic environments and effectively solving the model degradation problem.

This invention discloses a weather-interference-adaptive point cloud 3D target detection method and apparatus, relating to the field of target detection technology. The method includes: constructing a weather-interference-adaptive point cloud 3D target detection model, comprising an interference sensing network and a hybrid expert target detection model; converting raw point cloud data into distance images and inputting them into the interference sensing network for interference level classification to obtain interference levels; matching the corresponding expert model in the hybrid expert target detection model using a sparse activation mechanism based on the interference levels; inputting the raw point cloud data into the matched expert model and outputting the corresponding detection results; based on the detection results and interference levels corresponding to the raw point cloud data, employing a staged training strategy, training the detection model through a constructed total loss function to obtain a trained model; inputting the point cloud data to be detected into the trained model and outputting the final 3D target detection result. Using this invention can improve detection accuracy.