21 results about "Cross field" patented technology

The application discloses a kind of one-dimensional force electric characteristics of concrete test piece and system, it is related to the cross field of detection and intelligent sensing technology of civil engineering material dynamic mechanical property.Testing method includes the following steps: S1: using one-dimensional Hopkinson pressure rod experimental device to the concrete test piece pre-embedded piezoelectric sensor is axially impacted loading, obtains incident wave signal, reflection wave signal and transmission wave signal;The current and voltage corresponding to the resistance obtained in parallel to the two ends of piezoelectric sensor;S2: obtaining the one-dimensional stress-strain curve of concrete test piece;S3: based on current and voltage, the charge amount generated by piezoelectric sensor is obtained;S4: the impact velocity of impact rod is changed and the stress-charge amount variation curve is obtained by repeating steps S1-S3.Getting the axial stress of concrete test piece and the charge amount output by piezoelectric sensor simultaneously, constructs stress-charge amount relationship curve, realizes the quantitative calibration of dynamic sensitivity.

The application discloses a shale lamina identification method based on an improved yolov8 large model, belongs to the cross field of geological exploration and computer vision image recognition technology, and comprises the following steps: labeling lamina in a shale slice image, dividing the lamina into dark lamina, bright lamina and shell lamina, respectively giving corresponding yolo format labels, and unifying image naming specifications; dividing the processed image dataset into a training set, a verification set and a test set; training the training set by using the improved yolov8 model; outputting the model mAP50 value based on the verification set, analyzing the precision, recall rate and F1 value, and if the accuracy standard is not reached, fine-tuning and optimizing the model; and predicting the lamina of the shale slice image by using the optimized model and visually outputting the position and category label of the lamina. The application identifies the lamina type by learning the lamina characteristics, and breaks through the limitation of lamina characteristics in the traditional method.

The application provides a data processing method and device, which can be applied to the fields of automation and artificial intelligence. The method comprises the following steps: performing semantic analysis on to-be-processed data, and determining a target field related to the semantics of the to-be-processed data from a plurality of preset fields; determining, based on the field characteristics of the target field, a cross field having a cross-research direction with the target field from the plurality of preset fields; clustering a plurality of field entities in the cross field to obtain a plurality of entity clusters, wherein the plurality of field entities in the cross field are determined by a large language model according to a knowledge base related to the cross field; determining, based on a target entity cluster and the target field, prompt information for the to-be-processed data by using the large language model, wherein the target entity cluster is an entity cluster having the highest relevance with the to-be-processed data; and instantiating the prompt information by using a task execution intelligent agent to obtain a processing result of the to-be-processed data.

The application discloses a mass spectrum combined characterization qualitative and quantitative method and system of unified hidden state, relates to the cross field of analytical chemistry, bioinformatics and artificial intelligence data processing, the method obtains original data and instrument / method metadata containing LC-MS / MS, encodes to form a condition vector and is used for conditional modulation, constructs wave encoder and particle encoder, respectively Tokenizes continuous chromatographic signal and discrete fragment peak set into wave encoder and particle encoder to obtain wave representation and particle representation, realizes cross-modal interaction through wave-particle fusion module, forms unified hidden state or object-level unified hidden state set, adopts two-stage training of self-supervised pre-training and task fine-tuning, and applies conditional quantitative operator and qualitative operator to unified hidden state or object-level unified hidden state set in the inference stage, and parallelly outputs quantitative results and identification score / probability, and is suitable for end-to-end qualitative and quantitative analysis in DDA / DIA scene.

The application discloses a method for improving simulation speed of physical quantity of cross-field diode, relates to simulation technology of vacuum electronic device, and is proposed in view of the long simulation time in the prior art. Before solving the electric field force corresponding to each electron particle, the numbering is sorted in sequence from the cathode to the anode according to the positions of all the electron particles at the current time, the numbering is corresponding to the positions of the electron particles, and then the electric field force corresponding to each electron particle is solved according to the partial solution of the diode potential. The method has the advantages that the solving step of the electric field force corresponding to the partial solution of the diode potential in the cross-field diode is improved, the fast sorting algorithm is combined, the simulation speed is greatly improved compared with the general meshless numerical algorithm, and the method is suitable for the vacuum space of the cross-field diode in which the electron particles are injected from the cathode to the space with an arbitrary initial speed and an arbitrary magnetic field value.

The application discloses a kind of comprehensive traffic low altitude global intelligent perception methods based on autonomous evolution, belong to the cross field of intelligent transportation and computer vision.Method includes: collection multi-source perception data and extract structured causal variable;Dynamic causal diagram is constructed, and attribution analysis of perception error is realized by intervention learning;Cross-scene model migration and few-sample self-adaptation are realized based on knowledge graph;Adaptive inference is carried out using dynamic graph neural network of environment perception;Autonomous evolution closed loop driven by causality is realized based on multi-source feedback.The application realizes the leap of perception system from "perception-optimization" to "understanding-evolution" in low-altitude complex traffic environment, significantly improves perception accuracy, scene adaptability and system explainability, and reduces artificial operation cost.

The application discloses a radar echo space-time extrapolation method based on asymmetric space modeling, belongs to the cross field of meteorological radar data processing and artificial intelligence technology, and first constructs an asymmetric input space sample containing a target region and a surrounding extended region by preprocessing original meteorological radar observation data, builds an asymmetric space modeling prediction network based on a 3D U-Net architecture, adopts a regional mask mean square error loss function to complete model training, and finally realizes accurate space-time extrapolation of the target region radar echo. The application solves the technical problems of information fragmentation and significant long-time prediction error accumulation of the traditional symmetric modeling method, improves the precision and stability of radar echo extrapolation under complex weather systems, and is suitable for 0-1 hour nowcasting and meteorological disaster accurate early warning scenes.

The application relates to the cross field of artificial intelligence and object recognition, and discloses a mode recognition-based unmanned aerial vehicle target recognition and tracking method.The method comprises the following steps: collecting visible light, infrared and point cloud data through a multi-modal sensor, performing space-time alignment and fusion to generate a multi-channel feature map; constructing an initial feature set based on more than 3 frames of observation, generating a class-invariant target prototype representation through a meta-learning encoder, and storing the target prototype representation in a dynamic template library; subsequently, target recognition is realized through normalized cosine similarity matching, and a pan-tilt closed-loop tracking is driven; and under high confidence, a sliding average weighting strategy is used to incrementally update the prototype representation online.The application does not need to retrain the backbone network, and can realize second-level new target adaptation with only a small amount of samples, thereby improving the autonomy, generalization capability and tracking robustness of the unmanned aerial vehicle in an open environment.

The application discloses a neutron irradiation sample box convenient for remote operation and a dismounting tool thereof, and belongs to the cross field of nuclear science and technology application and equipment manufacturing. The sample box comprises a sample cover, a sealing sheet, a sample inner grid and a cavity. The sample cover is in threaded connection with the cavity. The sealing sheet is sealed by plastic deformation. The sample inner grid is divided into partition compartments for placing samples. The dismounting tool comprises an installation base, a screwing handle, a lower jaw, a circular arc knife edge, a sample box separation nut and an arched sample box positioning block and the like. The sample box is convenient for remote operation of adding and removing samples by special design. The non-welding sealing is adopted to guarantee sample irradiation conditions. The dismounting tool can be remotely operated in the whole process. The application further increases innovative features such as radiation-resistant adhesive, a pressure sensor and a fine adjustment device, improves sealing stability and operation accuracy, and solves the problems of inconvenience in remote operation, sealing and sample adding and removing of the traditional sample box.

The application provides a sea ice sample augmentation method, system and device, relates to the cross field of sea ice sample processing and computer vision, and the method comprises the steps of: labeling and preprocessing a ship-based sea ice image to obtain a standardized real sea ice scene graph and a corresponding sea ice label graph; constructing an improved conditional generative adversarial network comprising a generator and a double-head discriminator, wherein the double-head discriminator comprises a spatial domain discriminator and a wavelet domain discriminator; inputting the sea ice label graph into the generator to generate a sea ice scene graph; combining the generated graph and the label graph into a first input group, combining the real graph and the label graph into a second input group, inputting the two input groups into the two discriminators respectively to obtain spatial domain and wavelet domain discrimination results; generating a composite loss function according to the discrimination results, updating the parameters through back propagation; repeating the process until the loss converges to obtain a trained model; and inputting a sea ice label graph to be augmented into the model to output a sea ice augmented sample by the generator. The application can generate high-quality sea ice samples with global structure and detailed texture, and can alleviate the shortage of samples.

The application relates to the cross field of software engineering and artificial intelligence, and in particular to an AI-based test failure root cause analysis and automatic repair method, which constructs a space-time anchored tensor based on multi-modal original data of test failure; the evidence matching degree of a reverse cause-effect chain is calculated in combination with objective evidence, a root cause result is truncated, an initial repair code is matched, a history track is extracted to eliminate pseudo-associated cases to evaluate risks, and a global configuration change with high risk is downgraded to a local private configuration; the influence domain safety degree and a repair cost penalty value are calculated based on real associated cases, and an access evaluation result is obtained through weighted comparison; retesting is performed, if the retesting fails, cross-generation residual features with repaired redundant features are extracted, new data is injected, the above steps are iteratively executed until the fault is eliminated. The application effectively filters out AI deduction illusions and pseudo-associated influences, and improves the accurate positioning and safe self-repairing capability of complex faults.

A vehicle fault-tolerant control method under abnormal conditions of a multivariate measurement sensor belongs to the cross field of electric digital data processing and intelligent vehicle control technology. The method first performs redundant modeling and credibility evaluation on the key state of the vehicle based on the space-time correlation and physical consistency constraints of multi-source sensor data; then, when a sensor abnormality is detected, a state reconstruction model driven by redundant data is used to realize real-time compensation and correction of the failed observation; further, a fault-tolerant control strategy for the minimum functional requirements of the vehicle is constructed by combining a fast fault identification and a virtual sensor switching mechanism, so as to ensure that the vehicle can still maintain basic driving and safety control capabilities under the condition that part of the sensor function is damaged. The present application significantly improves the safety and robustness of intelligent driving vehicles in complex environments and abnormal conditions while ensuring the real-time performance of the system, and is suitable for intelligent vehicle control systems with multi-sensor fusion.

The method for intelligent evolution of water resource decision based on cross-modal analysis and adaptive learning relates to the cross field of wisdom water, natural language processing and artificial intelligence, and includes: cross-modal feature analysis on unstructured text data, application of a multi-user water resource mixed game model to generate an initial water right game allocation and scheduling scheme, and quantitative evaluation of the comprehensive benefits of the scheme by using a comprehensive benefit quantitative evaluation function; capturing the manual fine-tuning, interactive decision and selective tendency actions of the management user on the initial water right game allocation and scheduling scheme and converting them into a human preference reward signal of reinforcement learning; constructing a reinforcement learning comprehensive reward feedback function; and using a reinforcement learning algorithm to dynamically evolve and update a target weight strategy library, so as to realize the autonomous evolution of the multi-user water resource mixed game model and the accumulation of the strategy knowledge base. The method breaks through the data modal barrier and avoids the problem of strategy rigidity caused by static weights.

The application discloses a topsoil stratum modeling method based on MicroStation attribute identification and relates to the cross field of geological survey and computer-aided design. The application creates a three-dimensional model of a borehole in MicroStation, projects an orifice coordinate to a plane, adopts a BowyerWatson algorithm to perform Delaunay triangulation, and constructs a borehole profile triangular net. After manual adjustment, the strata are automatically connected according to corresponding rules, and the triangular area space is closed. Then, the stratum profile is generated into a geobody model along the elevation, the lens structure is subjected to shearing treatment, and finally, a complete three-dimensional topsoil geological model is formed. The application realizes three-dimensional visual expression of the topsoil stratum, can be arbitrarily cut and viewed, and only needs to be partially reconstructed when new survey data is added, so that the modeling workload is greatly reduced, the accuracy and updating efficiency of the geological model are improved, and an intuitive and reliable geological basis is provided for geotechnical engineering design and construction.

A current transducer (1) includes a magnetic field gradient sensor (2) and a primary conductor (5) having a first branch (6a), a second branch (6b), and a bridging branch (8) interconnecting one end of the first branch (6a) to one end of the second branch (6b), the first and second branches being separated by a gap (G), the first branch (6a) including a first junction section (24a) connected to the bridging section, the first junction section being configured to guide primary current obliquely and outwardly relative to a centerline (C), and the second branch including a second junction section (24b) connected to the bridging branch (8), the second junction section being configured to guide primary current obliquely and inwardly relative to the centerline, such that a magnetic cross-field generated by the primary current flowing in the bridging branch is canceled out by magnetic cross-fields generated by the primary current flowing in the first and second junction sections, respectively.

A kind of fire-burned trace land quick identification extraction method based on time sequence remote sensing vegetation index belongs to the cross field of electronic information technology and resource and environment technology, including the following steps: step one, obtains and pre-processes historical and latest remote sensing image, calculates spectrum index sensitive to fire-burn and constructs time sequence curve;Step two, extracts historical synchronous data, carries out trend fitting using non-parametric Theil-Sens method, obtains the prediction model of each pixel;Step three, calculates the residual between the actual value and predicted value of spectrum index of latest image and previous effective image, and further calculates residual change amount;Step four, carries out threshold binaryzation processing to the residual change amount image of each spectrum index, takes intersection to multi-index binary result and fuses iteration with historical monitoring result, finally determines the range of fire-burned trace land.The present application significantly improves the timeliness and accuracy of fire-burned trace land identification, and can provide effective technical support for forest fire prevention, ecological assessment and emergency management.

The present application relates to a kind of near infrared spectrum wood identification method based on wavelet convolution network, belong to the cross field of wood science, spectral analysis technique and artificial intelligence.The method includes: the near infrared spectrum of wood is pretreated, spectrum is decomposed using discrete wavelet transform, and complementary three-channel one-dimensional feature tensor containing original, trend and detail information is constructed;Build double-flow model, wherein the classification network adopts series convolution structure to extract deep features from tensor for classification;The feature level auto-encoder is used in the rejection network, and the features extracted by the classification network are received for reconstruction training.The model is trained using a two-stage strategy, first optimizing the classification network, then fixing its parameters and training the rejection network alone.In online identification, the feature reconstruction error of the sample to be tested is calculated and compared with the preset threshold, to realize the interception of unknown or counterfeit samples, and only the classification results of the samples passing the test are output.The present application effectively improves the identification accuracy and anti-counterfeiting ability.

The application relates to a marine ecological early warning and prediction method based on multi-source data fusion. The application is suitable for the cross field of marine environment monitoring and information technology. The technical scheme adopted by the application is as follows: a marine ecological early warning and prediction method based on multi-source data fusion comprises the following steps: acquiring multi-source heterogeneous marine environment data of a target sea area, including marine environment data from space-based, air-based, sea-based and shore-based monitoring and social perception; sequentially performing standardization processing, intelligent feature extraction and feature fusion on the multi-source heterogeneous marine environment data to obtain a fusion data cube; optimizing a marine ecological dynamics model through data assimilation based on the multi-source heterogeneous marine environment data; taking the fusion data cube and the assimilation-optimized marine ecological dynamics model as inputs, performing operation through a fusion prediction model coupled with mechanism and data driving, outputting a prediction result of a marine ecological disaster, and matching the prediction result into graded early warning information based on an early warning index library and publishing the information.

The application discloses a kind of optimal odd complementary sequence pair search method, framework, storage medium and equipment, belong to the cross field of signal processing and artificial intelligence, which comprises: initialization sequence prefix, search is carried out by the selection, expansion, evaluation, back transmission steps of MCTS;Wherein, evaluation step uses CNN to directly estimate the value of current sequence prefix, avoid the high computing overhead caused by random completion in traditional MCTS;The state, strategy and return information generated in the search process are independently managed and fixed length processing by replay pool, guarantee the stability and generalization ability of CNN training under different length sequence input;And update CNN parameter in closed loop mode, continuously optimize search strategy.The application can efficiently and stably search out optimal odd complementary sequence pair satisfying Krammer lower bound in the field of radar signal design, wireless communication channel estimation and synchronization, etc., significantly improve the efficiency and engineering practicability of sequence search.

The application belongs to the cross field of artificial intelligence and intelligent operation and maintenance of industrial equipment, and particularly relates to a predictive health management method for key components of unloading machines, aiming to solve the problems of lag in early fault identification, poor working condition adaptability and insufficient life prediction accuracy in traditional operation and maintenance mode. The method collects vibration, load, current and temperature data through multi-source sensors and synchronizes them at millisecond level, extracts dynamic features by combining working condition adaptive wavelet packet decomposition and autoencoder noise reduction, and inputs the features into an LSTM network with attention mechanism to build a health index monotonically decreasing from 0 to 1. The application significantly improves the early damage detection rate and evaluation robustness, with a prediction error less than ±15%, a reduction of more than 40% in unplanned downtime, and support for edge real-time calculation and port-level cloud platform collaboration, promoting the transformation of the operation and maintenance system to prediction and collaboration, reducing the annual maintenance cost by 25%, and improving the comprehensive efficiency of equipment by 8 percentage points.

The application relates to the cross field of nuclear safety and membrane separation technology, and particularly relates to a ceramic membrane assembly abnormality diagnosis method and system for a nuclear-grade high-temperature dust-containing environment. The method comprises the following steps: S1: acquiring multi-source data of a ceramic membrane assembly, presetting a time acquisition window and a maximum storage time, constructing a coupling memory matrix of the ceramic membrane assembly, and extracting a real-time query feature vector; S2: calculating a transient drift sensitive factor of the ceramic membrane assembly in the query feature vector according to the query feature vector; calculating a real-time scalar sequence of the ceramic membrane assembly and a historical scalar sequence matrix according to the transient drift sensitive factor; and S3: calculating a single-dimensional drift indicator of the query feature vector and the coupling memory matrix according to the real-time scalar sequence and the historical scalar sequence matrix. The application acquires drift compensation through a historical scalar sequence and a real-time scalar sequence, and jointly diagnoses according to the drift compensation and thermal fatigue.