9 results about "Parallel design" patented technology

The invention relates to the technical field of image processing, in particular to a dual-branch coding desert segmentation model network structure based on structure state space duality, which adopts multi-dimensional dynamic convolution to replace traditional convolution in the initial stage of an encoder, introduces a mamba2 module based on the structure state space duality into the backbone design of the encoder, and improves the robustness of the encoder. The efficiency and adaptability of the model are remarkably improved, a double-branch parallel design is adopted, one branch uses cavity convolution to extract multi-scale context information, the other branch reinforces feature expression through a mamba2 module, the model is connected in series with a space attention module and a channel attention module between an encoder and a decoder, and the algorithm is more accurate. The method has the advantages that the method is simple and easy to implement, interference of irrelevant information on segmentation results is suppressed, a deformable large kernel attention module is introduced to the tail end of a decoder, and global and local modeling capability of the model in processing desert complex boundary regions is effectively improved by combining flexibility of deformable convolution and global receptive field characteristics of large kernel convolution.

The invention relates to the technical field of fault detection, in particular to an industrial fault detection method and system based on dynamic drift perception and diffusion enhancement. The method comprises the following steps: constructing an unsupervised fault detection DDA-DE model, and processing an industrial data flow by utilizing dynamic drift awareness DDA to establish a statistical distribution baseline and a drift threshold; determining an initial model parameter and an anomaly threshold by using diffusion enhanced anomaly detection DE; robustness enhancement of concept drift is carried out on the unsupervised fault detection model based on a diffusion strategy; according to the Mahalanobis distance real-time drift sensing algorithm based on industrial enhancement, the concept drift phenomenon can be detected more efficiently, and collaborative detection of data drift and abnormal events is achieved through the parallel design of the drift sensing algorithm and the fault detection classifier.

This invention discloses a method for inverting gross primary productivity (GPP) based on multi-level deep learning. The method comprises obtaining spectral, meteorological, and flux information from remote sensing images; generating a lookup table using the PROSAIL model to establish correspondences between spectral information and leaf structure, leaf biochemical, and soil information; constructing a GPP inversion model using multi-level deep learning; inverting GPP using the model and assessing the inversion accuracy. The method utilizes the model to convert spectral information into information such as leaf structure, expanding the spectral information dimension; designing four types of limiting factors to enhance model interpretability; constructing a multi-level deep learning network, utilizing it to automatically optimize the expression of limiting factors, and employing a parallel design to clarify the effects of these factors, thereby improving model accuracy.

The invention relates to a collaborative modeling method, device and system, computer equipment and a readable storage medium. The method comprises the following steps: acquiring a target model copy corresponding to each client; the target model copy is obtained by designing and processing the initial model copy of the target model in parallel by each client and feeding back to the server; the initial model copy is constructed by the server in response to a design request of the client; and according to the submission sequence of the target model copies, performing data merging on the initial model master based on each target model copy in sequence to obtain a target model master corresponding to the target model. By adopting the method, the model design efficiency can be improved.

The present invention discloses a method, system, and storage medium for designing and managing architectural models based on cloud computing technology, belonging to the field of cloud computing technology. The method includes periodically collecting relevant parameter information of the architectural model to obtain a basic data set; preprocessing and feature extraction of the basic data set to obtain a feature data set; calculating a complexity assessment value of the architectural model based on the feature data in the feature data set; judging the magnitude of the complexity assessment value and switching to a parallel design mode, a hybrid design mode, or a distributed parallel design mode based on the judgment result; and designing and managing the architectural model based on the current design mode. The present invention assesses the complexity of the architectural model by collecting and processing relevant parameter information of the architectural model, and selects an appropriate design mode for data processing based on the complexity, thereby achieving efficient distribution and management of design tasks on different computing nodes.

The invention discloses a virtual simulation experiment method for schematic diagram design of a hydraulic system. The virtual simulation experiment method comprises the following steps: (1) analyzing and confirming a basic loop needing to be designed according to design requirements; (2) designing and simulating an oil supply loop; (3) designing and simulating a speed regulation loop; (4) designing and simulating a reversing and rapid motion loop; (5) designing and simulating a speed change-over loop; (6) drawing a schematic diagram of the hydraulic system; (7) designing and simulating an electromagnet action control circuit; and (8) performing experimental evaluation. The method is particularly emphasized to cultivate the comprehensive evaluation ability required by the engineering decision of students, break through the teaching mode of a single standard solution, and provide the practice opportunity of multi-scheme parallel design and comparison. And the inherent problems of high equipment cost, high construction risk, limited student participation degree and difficult observation of oil flow direction and element action of the traditional experiment can be effectively avoided, and an interactive virtual experiment method is provided for the schematic diagram design experiment teaching of the hydraulic system.

The invention provides a collaborative design method for a general station and a general assembly plant in the ship industry. The collaborative design method provides a cross-organization closed-loop collaborative mode based on bidirectional mapping and data packet driving. Through a system architecture with a parallel front end and an independent rear end, a general station and a general assembly factory can carry out parallel design and review on the same platform, the method establishes a design ID-construction ID bidirectional traceability mechanism throughout the whole process, modification of any party can be automatically pushed through a data packet, influence analysis is triggered, and an accurate closed loop of change is realized. The mode not only breaks through data flow and business flow of design, process and manufacturing, but also fundamentally solves three core pain points of data splitting, flow serialization and change response low efficiency brought by a traditional'wall-throwing 'mode, and promotes ship research and development to be transformed to efficient collaboration and flexible parallel.

The invention discloses a multi-scheme parallel design whole-process visual traceability system and an implementation method. The method comprises the following steps: 1) firstly deploying a data engine micro-service; 2) deploying a service system, and calling a data engine SDK (Software Development Kit) to complete initialization of a system queue; 3) calling a data engine SDK by the back end of the service system, and transmitting data among different schemes in the parallel design process; and 4) customizing a connection line and a control at the front end of the business system to call the data engine SDK, and returning to a data rendering interface, so that the whole business process is traced and visualized. The method is realized by adopting a micro-service architecture, a data engine service provides an SDK (Software Development Kit) to the outside, and circulation of design process data among different schemes and real-time updating of design process parameters are realized through the data engine SDK; by customizing connection lines and controls, the interface is deeply rendered, clear connection lines exist in the whole argumentation process, and full-link visual tracing of multi-scheme parallel design is achieved.

The application discloses a concrete proportioning multi-objective parallel design method based on sparse Gaussian process Bayesian optimization, and the method comprises the following steps: taking a regression model represented by a preset machine learning algorithm as a meta-learner, respectively constructing meta-prediction models of data-driven concrete mechanical properties, and finally integrating the meta-learner to form a high-precision concrete mechanical property prediction model; judging whether the current concrete mechanical property pre-training prediction model meets engineering requirement information, if yes, the step can be skipped, the optimization target of the concrete proportioning is determined, that is, the weights of the mechanical property parameters and the cost of unit volume of concrete are set; entering the sparse Gaussian process Bayesian parallel optimization process, predicting the mechanical property parameters according to the generated initial proportioning, calculating the cost of unit volume of concrete under the proportioning, performing multi-round optimization, and finally obtaining the optimal proportioning; the method constructs a high-precision concrete mechanical property prediction model through pre-training or a newly-built data set, adopts a residual neural network, integrated learning and other models to integrate the advantages of multiple algorithms, and improves the accuracy of the concrete mechanical property prediction such as compressive strength, tensile strength and elastic modulus.