34 results about "Sample condition" patented technology

The invention relates to the technical field of computer vision, and discloses a hybrid supervision collaborative learning-based power transmission line identification method under a small sample condition, and the method comprises the steps: obtaining an aerial image of an unmanned plane, and carrying out the preprocessing and marking of the image, and obtaining a data set; a self-supervised learning framework of a self-adaptive mask enhancement strategy is proposed, linear geometric priori knowledge is obtained through a lightweight U-Net segmentation model, a linear element self-adaptive mask mechanism driven by weak supervised learning is constructed, and a masked image is generated; a scale-width-angle collaborative enhancement multi-dimensional line feature attention module is provided, the multi-dimensional line feature attention module is fused into a sparse convolution encoder to extract visible region features, and the perception and feature extraction capability of the model on the slender structure of the power transmission line is enhanced; a masked area is reconstructed through a decoder, a composite loss function optimization strategy with weight collaboration is provided, a segmentation task under the condition of foreground and background extreme imbalance is met, the overall convergence speed of the model is increased, and power transmission line recognition precision rise under the condition of small samples is achieved.

The invention relates to the technical field of fault diagnosis, and discloses a bearing zero sample fault diagnosis method based on envelope order spectrum multi-dimensional feature extraction. According to the method, simulation fault data is generated through a bearing dynamical model, and the problem of sample scarcity is solved; the simulation data and the health data are combined to train a multi-scale residual attention network; for a to-be-diagnosed signal, a diagnosis result is obtained through the network model and a physical characteristic threshold rule based on the envelope order spectrum; and finally carrying out weighted fusion on the two to obtain a conclusion. According to the method, the dependence on a real fault sample in the prior art is overcome, high-precision diagnosis under a zero sample condition is realized, and the reliability and interpretability of the model are improved through decision-making layer fusion of physics and data.

The application discloses a kind of few sample X-ray defect intelligent detection methods based on diffusion generative model, belong to industrial nondestructive testing and artificial intelligence technical field;The method first acquires preprocessed X-ray image dataset, constructs defect condition embedding that retains physical characteristics;Build the condition diffusion generative model that introduces rectified flow reparameterization and consistency model distillation mechanism optimization, trains high-fidelity defect sample in stages generation;Build multiscale Transform detection network, through mixed sample training, dynamic loss optimization, course learning and feature consistency constraint, by the intermediate feature of detection network is fed back to generative model to adjust defect condition weight, realize the joint training of generative model and detection network;Finally input image to be detected completes defect positioning classification.The application can generate multiple types of real defect samples, eliminate sample imbalance and domain bias, with high detection accuracy and real-time performance under few sample conditions.

This application relates to the field of individual radiation source identification, specifically to a method and apparatus for individual radar radiation source identification in small-sample open-set scenarios. One specific implementation of the method includes: inputting the acquired target input signal into a pre-trained optimized feature extraction model to obtain the original target activation vector; determining the target activation distance vector based on the original target activation vector; inputting the target activation distance vector into a preset extreme value theoretical model to obtain a target reliability score; correcting the original target activation vector based on the target reliability score to generate a corrected target activation vector; normalizing the corrected target activation vector based on a preset normalization formula to generate a normalized probability; and determining target identification information based on the normalized probability, a preset threshold, and a preset identification formula. This implementation can achieve individual radiation source identification under small-sample conditions.

This utility model relates to the field of sample transfer technology, specifically disclosing a sample rack, which includes a base and several support frames. The base has several mounting slots spaced apart along a first direction; each support frame corresponds to one of the mounting slots, and the support frames can be placed in or removed from the mounting slots; each support frame has several receiving slots spaced apart along a second direction, in which several sample tubes are placed. When inspecting samples, one support frame can be removed to observe the sample tubes within, improving inspection efficiency. Furthermore, compared to the sample tubes themselves, the support frame is easier to place and remove, reducing the risk of tubes falling and improving the safety of sample condition inspection.

The application discloses a load interval prediction method based on conditional residual distribution modeling and local adaptive calibration, and the method comprises the following steps: constructing a prediction residual sequence based on a point prediction result and real load data, and forming a basic data set for interval prediction modeling; establishing a conditional residual distribution model based on a mixed density network, outputting mixed weights, mean values and variance parameters corresponding to residual distribution under different sample conditions, and constructing a load initial prediction interval; introducing a local adaptive conformal calibration method, using calibration samples to construct a non-uniformity score, using a Gaussian kernel function combined with a near neighbor truncation strategy to calculate the local similarity weight between the to-be-tested sample and the calibration sample, determining the local adaptive calibration radius through the weighted non-uniformity score quantile, and dynamically correcting the upper and lower boundaries of the initial prediction interval. The application can effectively improve the adaptability of the load interval prediction result to error heterogeneity and local time-varying characteristics.

This application relates to a method, apparatus, and electronic device for assisting in process optimization. The method includes: obtaining a condition parameter space and an optimization objective; sampling the condition parameter space to obtain multiple sampled condition points; predicting the multiple sampled condition points using a preset model to obtain an estimate of the optimization objective for each sampled condition point; processing the estimated optimization objective for each sampled condition point using a preset acquisition function to determine recommended condition points from the multiple sampled condition points; and outputting the recommended condition points. By automatically generating recommended process condition parameter values ​​through Bayesian optimization, the process optimization process can be guided, shortening the optimization cycle, obtaining better process conditions with fewer experiments, saving material costs during optimization, and reducing reliance on the prior knowledge of technical personnel.

The application discloses a kind of train operation and freight loading state detection methods for receiving and sending train, it is related to railway transport safety monitoring technical field, method includes: obtaining the original data of receiving and sending train, original data is divided according to preset sample condition, obtain small target data, sample sparse data and multimodal data;Small target anomaly detection module, few sample anomaly detection module and multi-source anomaly detection module are respectively detected to small target data, sample sparse data and multimodal data;Small target detection result, sample sparse detection result and multi-source heterogeneous detection result are comprehensively judged and handled, and the final detection result of train operation and freight loading state is output.The application can solve the problem that small target sample, sparse sample learning multimodal data recognition effect is poor, and comprehensively covers receiving and sending train monitoring point, improves detection accuracy and intelligent safety monitoring level.

The application discloses a few-sample wafer defect image background removing and data enhancement method based on frequency-space fusion, relates to the technical field of wafer defect detection, and can be applied to wafer defect identification, small-sample machine learning model training and industrial visual detection system construction in a semiconductor manufacturing process. In order to solve the problems of few-sample wafer defect sample scarcity, incomplete process texture background suppression, low-quality synthetic defect sample and the like in the prior art, the application realizes three core stages of background removing and defect enhancement by constructing a few-sample wafer defect library and adopting an FSF-TBS module to complete few-sample wafer image defect transplantation, accurately extracts wafer defect features, efficiently suppresses and removes process background, and simultaneously generates high-quality defect enhancement samples. The application can significantly improve the signal-to-noise ratio of a wafer defect image, effectively alleviate the plight of defect sample shortage, improve defect identification precision under small-sample conditions, and is suitable for semiconductor industrial visual detection and small-sample model training scenes.

This invention discloses a method, apparatus, equipment, and medium for substation defect detection in small-sample scenarios, relating to the field of substation defect detection technology. The method includes: a regression network spatially expanding and grid-sampling candidate prediction boxes, embedding prototype features using category prototype vectors and an attention mechanism, and constructing a mask-stacking region propagation mechanism to output target box prediction results; a target classification network filtering candidate defect category sets based on candidate prediction boxes and category prototype vectors, and generating candidate defect category confidence scores through multi-source embedding fusion by constructing intra-class response embeddings, inter-class ranking embeddings, and background semantic embeddings; and introducing a text-image alignment network to assist in scoring the semantic consistency of candidate prediction boxes within a preset interval, obtaining calibrated candidate defect category confidence scores. This method can improve defect detection accuracy under limited sample conditions.

The application provides a high-end equipment weld seam radiographic defect semantic reasoning method and system under a few sample conditions. First, U-Net is used to extract the weld seam area and adaptively slide window and cut, and the anisotropic diffusion filtering and the variance guided CLAHE enhancement bottom layer features are combined; a controllable diffusion model of a boundary box perception symbol distance and a category condition is proposed to generate a high-fidelity synthetic defect image. Second, a light-weight visual detection model is used to output defect positioning and classification results. Then, an industrial weld seam knowledge graph is constructed. Finally, a large-small model coordination center of a physical-semantic dual state space is constructed, visual features are mapped to a large language model for semantic review, batch statistics and trend early warning are realized by analyzing the lead type identification, and a customized diagnosis report containing defect type qualitative analysis, mechanism tracing, physical troubleshooting and trend early warning is output. The application realizes a leap from bottom layer visual perception to deep layer causal cognition.

This invention discloses an automatic heat-insulating sample delivery device for a drop hammer test, relating to the field of impact testing technology. It includes: a sample heat-insulating mechanism comprising a heat-insulating box containing a heat-insulating liquid; and a sample delivery mechanism comprising a delivery track and a robotic arm assembly moving along the delivery track. The heat-insulating box and the drop hammer detection station are both located directly above the displacement path of the robotic arm assembly. The heat-insulating liquid is industrial alcohol, and its chemical composition and mass percentage include: ethanol content ≥95.0%, methanol content ≤0.06%, water content ≤5%, and impurities controlled ≤0.5 mg / L. This invention regulates the temperature of the heat-insulating liquid through a temperature control component, achieving rapid cooling and heating of the industrial alcohol, facilitating the setting of different sample conditions, and ensuring a stable industrial test temperature.

The invention discloses a training method and device of an action generation model, equipment, a medium and a program product, and belongs to the field of action generation. The method comprises the following steps: acquiring a priori condition representation and a priori sequence; sample condition representation and a sample sequence with personalized features are obtained; inputting the priori condition representation into the action generation model to obtain a priori prediction sequence; inputting the sample condition representation into the action generation model to obtain a sample prediction sequence; and training an action generation model based on a priori loss and a sample loss, the priori loss being obtained based on the priori prediction sequence and the priori sequence, and the sample loss being obtained based on the sample prediction sequence and the sample sequence. According to the method disclosed by the invention, prior knowledge learned during pre-training is reserved during model training, and the situation that personalized customization of the action generation model is affected due to the fact that the action generation model obtained through training has an overfitting problem can be avoided.

This invention relates to the field of fault diagnosis technology and discloses a zero-sample fault diagnosis method for bearings based on multi-dimensional feature extraction of envelope order spectrum. This method generates simulated fault data through a bearing dynamics model to address the problem of sample scarcity; it combines the simulated data with healthy data to train a multi-scale residual attention network; for the signal to be diagnosed, diagnostic results are obtained through the network model and physical feature threshold rules based on the envelope order spectrum, respectively; finally, the two are weighted and fused to arrive at a conclusion. This invention overcomes the dependence of existing technologies on real fault samples, achieves high-precision diagnosis under zero-sample conditions, and improves the reliability and interpretability of the model through the fusion of physical and data decision layers.

The invention discloses a mobile phone application performance automatic test system and method based on artificial intelligence, and relates to the technical field of electric digital data processing. The system comprises a sampling condition monitoring module, an automatic sampling adjustment module and a test feedback module. The mobile phone application performance is automatically tested based on the artificial intelligence algorithm in the current test period, the sampling condition of the test data is monitored in real time, then whether automatic sampling adjustment is executed or not is determined according to the sampling frequency adaptation judgment result, and if yes, automatic sampling adjustment is executed. If yes, the data noise condition of the test data is monitored after automatic sampling adjustment, test feedback is carried out after data noise degree judgment, whether sampling secondary adjustment is carried out or not is determined, and otherwise, the data sampling condition continues to be monitored, so that the adaptation degree of the test condition of the test data and the sampling frequency is improved; the problem that in the prior art, in the process of testing the application performance of the mobile phone, the adaptation degree between the sampling frequency and the corresponding testing condition is low is solved.

The application discloses a mobile phone application performance automatic test system and method based on artificial intelligence, and relates to the technical field of electric digital data processing.The system comprises a sampling condition monitoring module, an automatic sampling adjustment module and a test feedback module.The application performs automatic test on the mobile phone application performance based on an artificial intelligence algorithm in the current test period, monitors the sampling condition of test data in real time, then determines whether to perform automatic sampling adjustment according to the result of sampling frequency adaptation determination, if yes, monitors the data noise condition of test data after automatic sampling adjustment to perform data noise degree determination, and then performs test feedback, and simultaneously determines whether to perform secondary sampling adjustment, otherwise, continues to monitor the data sampling condition, thereby improving the adaptation degree of the test condition of test data and the sampling frequency, and solving the problem of low adaptation degree of the sampling frequency and the corresponding test condition in the process of testing the mobile phone application performance in the prior art.

This invention discloses a multimodal gearbox fault diagnosis method based on deep transfer learning, addressing the problem of poor gearbox fault diagnosis capability under unlabeled sample conditions, belonging to the field of gearbox fault diagnosis technology. The method includes: collecting raw vibration signals under different operating conditions, defining them as source and target domains; fusing multimodal information in the time and frequency domains using a data-level fusion method and dividing the collected signals into samples; constructing a deep multimodal adversarial transfer network model, extracting fault information features from the source and target domains through iterative adversarial training to adapt to the joint probability distribution of the source and target domains, and utilizing the abundant fault label information in the source domain to ensure accurate fault category discrimination; finally, obtaining a trained transfer diagnosis model for the target domain. This method is applicable to gearbox fault diagnosis under different operating conditions, i.e., transfer diagnosis between different operating conditions or different fault types, exhibiting high accuracy and good generalization performance.

This invention relates to the field of deep learning technology, specifically to a method and system for pre-training a small-sample data annotation model. The method and system include: a general training module, an adaptation training module, a pseudo-label expansion training module, a pseudo-label adjustment module, a human annotation module, and a data annotation module. This invention utilizes a large amount of unlabeled data within the task domain to pre-train the data annotation model for general features, then combines this with a small amount of human-labeled data for task-specific adaptation training. Furthermore, it introduces iterative expansion of the training set using high-confidence pseudo-labeled data, human correction feedback for low-confidence samples, cross-round abnormal pseudo-label identification, propagation domain correlation analysis, and a model correction mechanism based on parameter perturbation direction, enabling the data annotation model to continuously optimize and update in real time under small-sample conditions.

The invention discloses a rolling bearing fault diagnosis method and system under a small sample condition, and is suitable for fault diagnosis under the small sample condition. According to the method, based on historical small sample fault data, an improved 1D-DDPM model is adopted to generate fault samples, and the fault samples and real samples are mixed to construct a fault diagnosis data set. In the data set, all the test samples are real samples, and the training samples are formed by mixing the real samples and the generated samples. And meanwhile, the 1D-CNN and the channel attention are combined to construct a fault diagnosis model, so that the fault diagnosis accuracy under the limited sample condition is improved. The system comprises a data acquisition module, a preprocessing module, a data generation module, a data mixing module, a fault diagnosis model training module, a fault diagnosis module and a database storage module, and can be used for training and applying a fault diagnosis model under limited sample conditions. According to the method, a more accurate and efficient fault diagnosis model under a small sample condition is constructed by combining the strong sample generation capability of the 1D-DDPM model and the strong feature extraction capability of the improved 1D-CNN model.

The invention discloses an action diffusion model training method and device, equipment and a storage medium, and relates to the field of artificial intelligence. The method comprises the following steps: acquiring a sample action sequence and a sample condition representation; adding first noise to the sample action sequence to obtain a sample noise sequence; inputting the sample noise sequence and the sample condition representation into an action diffusion model, and outputting a predicted action sequence by the action diffusion model; adding second noise to the sample action sequence to obtain a sample noise adding sequence, and adding third noise to the prediction action sequence to obtain a prediction noise adding sequence; and training the action diffusion model by taking increasing the similarity between the sample noise adding sequence and the prediction noise adding sequence as a training target. According to the method, the action diffusion model can learn the intermediate process of restoring the sample action sequence from the sample noise sequence under the guidance of the sample condition representation, so that a relatively high action generation rate is obtained.

This invention discloses a single-channel source separation and anti-interference method based on meta-learning under small sample conditions. The method comprises the following steps: In the training phase, a meta-learning training dataset is first created, including the target communication signal and various interference signals; then, a subset is randomly extracted from the dataset to construct a task set, which is divided into a support set and a query set; next, a signal separation network is built, and a meta-learning objective function and optimization strategy are created, including inner and outer loops; finally, the task set signals are input into the network for training, and the optimal pre-trained model is saved. In the fine-tuning phase, a fine-tuning dataset is first created and tasks are extracted; then, a fine-tuning objective function and optimization strategy are created; finally, small sample signals are input into the pre-trained model for fine-tuning, and the optimal model is saved. This invention achieves effective separation of communication signals and interference signals under unknown or limited sample conditions, realizes effective interference identification and suppression, improves communication anti-interference capability, and ensures communication stability and reliability.

The invention discloses a diagnosis method for composite faults of rotating machinery. Comprising the steps of 1, data processing; step 2, establishing a twin comb filter network model SCFN; wherein the comb filter network CF-Net extracts features under the conditions of background noise and limited samples. Meanwhile, a twin neural network SNN is used to learn a relationship between samples. Step 3, constructing a multivariate loss function; and the feature representation generated by the SCFN feature extraction module is constrained through combined use of the comparison loss and the binary cross entropy loss. Wherein the contrast loss directly guides the embedding space to enhance the feature discrimination, and the binary cross entropy BCE loss supervises the prediction of the similar probability at the last layer of the network. And step 4, fault diagnosis. In a word, the method mainly solves the problem of composite fault diagnosis of the rotating machinery under small sample and noise conditions. In addition, data does not need to be preprocessed, and end-to-end fault diagnosis is achieved.

A specimen condition assessing device includes a processor that executes a program for performing image processing on an image of a target specimen container and a memory for storing a result of the image processing. The processor performs processing of inputting the image, processing of detecting a boundary position from the image, processing of determining a type of upper and lower regions of each boundary position and determining a boundary position of each inclusion, and processing of assessing whether or not the target specimen container is testable on the basis of the boundary position of each inclusion.

This invention proposes an adaptive feature clustering-based method for diagnosing coupled faults in industrial machinery under low-labeled sample conditions. It relates to the field of intelligent fault diagnosis for industrial equipment and addresses the technical problems of existing fault diagnosis methods, such as high dependence on labeled data, poor multi-source signal feature fusion, insufficient deep feature mining of coupled faults, and weak adaptability to complex operating conditions. First, heterogeneous kernel density adaptive normalization is performed on multi-sensor signals. Then, three-dimensional weighted primary and secondary features are adaptively mined to generate pre-trained weights. A diagnostic clustering model is constructed through iterative fusion of channel attention to complete fault identification. Finally, the model is optimized, and the source and evolution of the fault are quantified. This invention is suitable for low-labeled industrial scenarios, provides high accuracy in coupled fault diagnosis, exhibits strong robustness under conditions such as sample imbalance and limited channels, can quantify the source of faults, reveal evolution patterns, and provide accurate operation and maintenance suggestions. It also has good scalability and is easy to implement in engineering.

This invention discloses a method and system for small-sample fault diagnosis of industrial control valves based on uncertainty quantification, belonging to the field of industrial process equipment fault diagnosis technology. The method includes: collecting and preprocessing fault data on valve stem displacement and medium flow rate of industrial control valves; extracting expected, entropy, and hyperentropy uncertainty cloud features using a cloud model inverse cloud generator; generating cloud samples based on cloud features using a forward cloud generator to amplify the data; constructing a multi-objective cloud feature optimization function with distribution, intra-class compactness, and regularization constraints; and solving for the optimal cloud feature parameters using the NSGA-II algorithm. This invention accurately quantifies the uncertainty of fault data, generates high-quality enhanced samples, and improves the diagnostic accuracy and robustness under small-sample conditions using a lightweight multi-constraint model, achieving automated and real-time fault diagnosis of industrial control valves, and is applicable to fault diagnosis of industrial control valves in the process industry.

The invention discloses a multi-level chain compression and multi-dimensional fusion text source discrimination method and device. The method comprises the following steps: S1, performing multi-level information chain compression on an input to-be-discriminated text to obtain main discrimination features, and extracting multi-dimensional statistical features from the to-be-discriminated text; s2, performing standardization processing on the main discriminant features and the multi-dimensional statistical features, and performing auxiliary feature weight distribution to obtain auxiliary features with weights; and S3, setting a weight ratio of the main discrimination features to the auxiliary features with weights, carrying out fusion calculation to obtain a comprehensive score, and comparing the comprehensive score of the to-be-discriminated text with the discrimination threshold to complete text source discrimination. According to the method, efficient and accurate discrimination of the text source under the zero sample condition is realized through multi-stage information chain type compression and multi-dimensional feature fusion, the problems of poor cross-domain adaptability, diversified model sources, insufficient annotation data and the like of an existing method are solved, and the practicability and reliability of academic text detection are improved.

The invention relates to the technical field of management systems, in particular to an intelligent sample management system based on LIMS system linkage, which comprises the following steps: by interacting with an LIMS system, determining a placement position, placing a sample at a designated position with a red light, accurately storing the sample, detecting the taken sample, and updating the state of the sample in the LIMS system to be in detection; the detected scanning sample is put back to the original position, the remaining amount is recorded through the weighing module, data are transmitted to the system, the LIMS system monitors the storage time of the remaining sample in real time, and when the sample is not scanned and sampled, the bottom plate weighs and senses weightlessness, and the sample management frame gives out an alarm sound, the condition of the sample needing to be processed when abnormal operation exists is obtained. According to the invention, the defects of laboratory sample management, including unclear sample identification and no unique number, are effectively overcome; the sample receiving registration information is incomplete; and the problems of missing detection, false detection and wrong detection of the sample are solved.

This invention discloses an automatic method and system for extracting landslide areas from remote sensing images based on few-sample enhancement and multimodal fusion. The method includes: acquiring and preprocessing multimodal remote sensing data; constructing a foreground-aware few-sample enhancement module, expanding the training samples through geometric transformation, spectral perturbation, and random cropping enhancement strategies; constructing a dual-branch feature extraction network to extract spectral and topographic features respectively; achieving adaptive fusion of spectral and topographic information through a cross-modal attention fusion unit; training a deep learning model using a composite loss function including cross-entropy loss, Dice loss, and boundary constraint loss; and performing topographic constraint post-processing and morphological optimization on the model output to obtain the automatic extraction result of landslide areas. This invention can achieve high-precision landslide identification under limited sample conditions, effectively reducing the false detection rate and improving the model's generalization ability and adaptability to complex scenes.

This application provides a method for small-sample fault diagnosis of aero-engines based on a deep Siamese self-attention network. It combines time-series clustering and Siamese neural networks to diagnose small-sample aero-engine faults. This application addresses the multi-label classification problem under small-sample conditions by learning a similarity metric, effectively improving the accuracy of multi-fault classification under such conditions and mitigating the overfitting problem of deep neural networks. SANet is used as the feature extraction module of the Siamese neural network to learn richer temporal features, thereby improving the separability of different types of samples in the mapping space. SANet can simultaneously extract short-term and long-term dependencies of time series data, and outperforms RNN models in extracting long-term dependencies. This effectively alleviates the training problem caused by imbalanced samples and provides better diagnostic results for fault samples.