15results about How to "Optimize network structure" patented technology

The invention discloses a configurable flexible job shop scheduling method based on a multi-agent game model, and the method comprises the steps: mapping a scheduling problem into a dynamic disjunction graph model to describe the relation between a process and equipment, representing process nodes, equipment nodes and the mutual relation through a node set and an edge set, and combining the flexibility of a dynamic disjunction graph to obtain a flexible job shop scheduling model. And matching of equipment and processes is dynamically adjusted according to changes of production requirements, so that the equipment utilization rate is optimized. And a clustering method based on structure entropy is adopted, and process nodes are clustered into sub-graphs, so that the search space of a scheduling problem is reduced, and the calculation complexity is reduced. Construction of an order agent and an equipment agent and a game and cooperation process between the order agent and the equipment agent are covered. And each agent optimizes a decision through a deep Q network, and cooperates or competes according to a respective target. An automatic optimization configurator is used for optimizing hyper-parameters and a network structure of a deep Q network in a multi-agent system, and scheduling performance and adaptability are further improved.

The invention discloses injectable and degradable composite drug-loading hydrogel as well as a preparation method and application thereof. The injectable and degradable composite drug-loading hydrogel comprises sodium alginate hydrogel, piezoelectric nanofibers and triamcinolone acetonide and / or ropivacaine, wherein the piezoelectric nanofibers and the triamcinolone acetonide and / or the ropivacaine are loaded on the sodium alginate hydrogel; the piezoelectric nanofibers are poly-L-lactic acid nanofibers. Through the synergistic effect of controllable release of the medicine and piezoelectric stimulation, the dual effects of continuous analgesia and neural restoration are achieved in the PHN treatment process, more effective treatment selection is provided for a PHN patient, complications caused by a traditional implantable electrode can be avoided, a more minimally invasive, convenient and safe treatment mode can be provided for the patient, and the treatment effect is good. And a theoretical basis is provided for future clinical practice.

The invention relates to the technical field of optical sensor calibration and deep learning, and discloses a medium and low orbit optical load thermal deformation sight line error correction method, which comprises the following steps: S1, constructing an in-orbit camera imaging model based on a fixed star, and obtaining a camera sight line vector according to the imaging model; s2, analyzing reasons of errors caused by sun illumination to camera sight vectors and constructing characteristic parameters; s3, constructing a camera thermal deformation error model, and calculating a thermal deformation error matrix RSTD under different characteristic parameters; s4, the BP neural network is improved, a Newton-Raphson optimizer is used for adjusting the network structure of the BP neural network, and a rime optimization algorithm is used for initializing network parameters of the BP neural network; and S5, taking the characteristic parameters and the thermal deformation error matrix RSTD as input data, training by using the improved BP neural network, and predicting a thermal deformation error. According to the invention, camera sight line offset errors can be reduced, and the problem of frequent on-orbit camera sight line correction is solved.

The invention relates to the technical field of high-performance polymer materials, in particular to low-dielectric and high-temperature-resistant modified cyanate ester resin and a preparation method thereof. Advantages and characteristics of bisphenol A type cyanate ester monomers and bisphenol M type cyanate ester monomers are combined, and the dielectric property and the thermal stability are remarkably improved. Diallyl bisphenol A is introduced as a network structure modifier, the phenolic hydroxyl group of the diallyl bisphenol A adjusts the curing reaction rate, and the allyl and bisphenol A structures of the diallyl bisphenol A finely regulate and control the rigid-flexible balance of a cross-linked network through chemical copolymerization, so that the curing process is controllable and the final network structure is optimized. Meanwhile, a toughening system is innovatively designed, epoxy-terminated liquid nitrile rubber is adopted for toughening, stable micron-scale rubber particles are formed in a resin matrix in a chemical bonding mode, and the fracture toughness is improved. The modified resin has excellent dielectric properties, high heat resistance and a wide process window, and meets the strict requirements of a new generation of low observability aircrafts on a bearing-stealth integrated composite material matrix.

The application relates to the field of ceramic materials, and particularly discloses high-thermal-conductivity alumina ceramics and a preparation process thereof. The high-thermal-conductivity alumina ceramics are prepared from 50-70 parts of alumina powder, 40-50 parts of alumina fiber, 8-15 parts of a sintering aid, 0.7-1.5 parts of a dispersant and 0.2-0.3 parts of a binder, wherein the sintering aid is beryllium-zirconium composite fiber. The high-thermal-conductivity alumina ceramics have excellent thermal conductivity and mechanical properties, and have important application prospects in the field of heat dissipation of high-performance electronic devices. In addition, the preparation process can reduce the sintering temperature of the alumina ceramics, form a more uniform and refined grain system in the alumina ceramics, and significantly improve the thermal conductivity of the alumina ceramics.

The present application relates to the field of photovoltaic power generation prediction, and discloses a short-term photovoltaic output prediction model construction method and a prediction method, which are used for improving the prediction accuracy of photovoltaic output. The short-term photovoltaic output prediction model construction method comprises collecting historical data of a photovoltaic power generation system, the historical data of the photovoltaic power generation system comprising meteorological environment data of multiple regions and corresponding photovoltaic output data; using an RBF-MeanShift model to cluster the meteorological environment data to obtain multiple meteorological clusters; optimizing an initial ResNet model based on a DGBCO algorithm to obtain a feature extraction model, and using the feature extraction model to extract features of each meteorological cluster to generate multiple training samples; inputting the multiple training samples into a DeepAR model for training to obtain a source domain short-term photovoltaic output prediction model; obtaining a target domain small sample, and inputting the target domain small sample into the source domain short-term photovoltaic output prediction model for transfer learning to obtain a target domain short-term photovoltaic output prediction model.

The invention discloses a lithium ion battery SOC estimation method based on a GRU neural network. The method comprises the following specific steps: S1, establishing a GRU neural network model; s2, optimizing hyper-parameters of the GRU neural network by using an immune genetic algorithm; s3, optimizing the neural network model by using a multi-head attention mechanism and a residual structure; and S4, estimating the SOC based on the optimized neural network model. According to the method, a multi-head attention mechanism is introduced to fully mine input data features, the multi-head attention mechanism overcomes the defects of single-head attention in processing a complex model, feature information can be better extracted from multiple aspects, model over-fitting is prevented, the immune genetic algorithm is adopted to optimize GRU recurrent neural network hyper-parameters, and the accuracy of the model is improved. And the GRU can be used for well storing time sequence characteristics of important data and mining time sequence information, so that the time sequence information becomes a good candidate for SOC prediction.

The application discloses a quad-rotor formation obstacle avoidance control method based on an improved DDPG algorithm, and belongs to the technical field of unmanned aerial vehicle formation control. The method adopts an improved DDPG reinforcement learning algorithm to plan an obstacle avoidance path of the quad-rotor. When the improved DDPG reinforcement learning algorithm is executed, the priority weight of each quadruple experience in the experience replay pool is initialized. After the quadruple experience in the experience replay pool is sampled and trained, the priority weight of each quadruple experience is recalculated based on a TD error, and a Sum_Tree structure is updated. The method effectively alleviates the training instability caused by hyperparameter sensitivity, the misleading of policy updating caused by overestimation of Q values, and the problem that key experiences are not sufficiently learned, accelerates the convergence speed of the quad-rotor formation obstacle avoidance training, and improves the obstacle avoidance effect.

The invention relates to the technical field of nondestructive testing and deep learning, in particular to a prefabricated concrete sheet member quality detection method based on lightweight Transform and GPR data. Comprising the steps of GPR image data input and preprocessing, lightweight convolutional neural network feature extraction, feature dimension reduction, position coding and serialization, Transform encoder global feature modeling, Transform decoder target query and detection, target classification and bounding box regression, post-processing and result output and the like. According to the method, the lightweight backbone network is adopted, the parameter quantity and the calculation quantity are reduced through the deep separable convolution, the inverse residual structure and the SE attention mechanism, and meanwhile, the detection precision of small targets and dense targets is improved; the problems that in the prior art, detection precision and calculation efficiency are difficult to consider at the same time, and a lightweight model is insufficient in small target and dense target detection capacity are solved.

The application discloses a preparation method of an alumina aerogel heat insulation felt, and comprises the following steps: adding an aluminum source precursor into a ketone substance to form a precursor solution through sufficient dissolution; adding an alcohol solvent containing a silicon source precursor, an amine substance and a catalyst into the precursor solution to form an alumina sol through uniform stirring; adding a surfactant into the alumina sol to form an alumina sol containing a light shielding agent through uniform stirring; adding light shielding agent particles into the alumina sol to form the alumina sol containing the light shielding agent; immersing a ceramic fiber felt into the alumina sol containing the light shielding agent to perform gel aging, so as to obtain a fiber felt / gel composite; and performing drying and heat treatment on the fiber felt / gel composite, so as to obtain the alumina aerogel heat insulation felt. The alumina aerogel heat insulation felt prepared by the method has the advantages of high temperature resistance, excellent heat insulation performance and good flexibility.

This invention relates to a conductive silver paste-specific glass powder for automotive rear window defroster lines, its preparation method, and its application. The glass powder comprises the following components in the indicated mass ratios: Bi₂O₃: 45-65%, SiO₂: 15-30%, B₂O₃: 5-12%, Al₂O₃: 1-5%, K₂O: 2-8%, Li₂O: 1-5%, CaO: 0.5-3%, and Sb₂O₃: 0.5-3%. It is processed according to the following steps: raw material selection, pretreatment, precise mixing, high-temperature melting, rapid water quenching, fine processing, and grading. Compared with existing technologies, this invention has advantages such as adaptability to high-temperature short-time sintering, high acid resistance, low sheet resistance, high welding strength, and high adhesion.

The invention discloses bio-based flame-retardant polyether polyol and a synthesis method thereof, and relates to the technical field of new chemical materials and biomass utilization. The bio-based flame-retardant polyether polyol is obtained by carrying out arbuzov reaction on halogenated starch liquefied polyol and phosphorous acid triester, and the halogenated starch liquefied polyol is prepared from bio-based starch, a liquefying agent, a halogenated dihydric alcohol liquefying auxiliary agent and a catalyst. The bio-based flame-retardant polyether polyol disclosed by the invention can be directly used for preparing flame-retardant polyurethane foam, the polyurethane foam can show excellent flame-retardant property under the condition of no additional flame retardant, and meanwhile, the optimization of a network structure in the flame-retardant polyurethane foam is also facilitated, so that the flame-retardant polyurethane foam has excellent flame-retardant property. The flame-retardant polyurethane foam with low apparent density and high compression strength can be obtained, and the light weight and the structural stability of the polyurethane foam are further improved.

This application discloses a method for predicting harmful algal blooms based on Tent-GWO-GRU, comprising: acquiring and preprocessing historical water quality data of the target watershed; constructing a gated recurrent unit neural network (GRU) model based on the historical water quality data; optimizing the Grey Wolf Algorithm (GWO) based on the Tent chaotic mapping algorithm; and optimizing the GRU model parameters and making predictions. This application applies a time delay to the index data, fully considering the lag in algal bloom growth and evolution, thus improving the feasibility of the method. The introduction of the Tent chaotic mapping algorithm improves the method for generating the initial wolf pack in the GWO algorithm, resulting in a more uniform distribution of the initial wolf pack, significantly improving the algorithm's fitness value, and making it easier to find the global optimum. This application uses the Tent-GWO optimization algorithm to optimize the hyperparameters and network structure of the GRU model, which not only improves the model's stability but also achieves higher prediction accuracy.

The invention relates to the technical field of network optimization, in particular to a network optimization method for a power line carrier communication system, which comprises the following steps of: intelligently dividing communication areas to ensure the optimal communication quality among nodes in the same communication area and reduce signal interference and attenuation; impedance matching is completed in the same communication area, channel attenuation is reduced, network configuration is dynamically adjusted according to a network link communication quality matrix, and it is ensured that the network is always in the optimal state; by optimizing and reconfiguring the proxy sites in the industrial automation network, the delay and loss in the data transmission process can be reduced, so that the expansibility and flexibility of the network are ensured; the overall efficiency and stability of the network are improved by analyzing the network hierarchical structure and the traffic demand, formulating a load balancing strategy, optimizing the network hierarchy and reducing unnecessary nodes and links. The invention is used for solving the technical problem of insufficient optimization of industrial automation network communication in the existing scheme.

The invention discloses a face recognition method, device and equipment and a medium, and the method comprises the steps: obtaining a to-be-recognized image, and inputting the to-be-recognized image into a pre-trained recognition model; the network structure of the recognition model comprises a feature extraction layer, a clustering layer, a first classifier layer and a second classifier layer; wherein the first classifier layer comprises a plurality of first classifiers; the first classifier is trained by using training data in a visual angle and a dynamic loss function in the visual angle; the second classifier is trained by using training data between visual angles and a dynamic loss function between the visual angles; and performing feature extraction on the to-be-recognized image through the feature extraction layer, performing clustering through the clustering layer, comprehensively applying the first classifier layer and the second classifier layer, performing common analysis from dimensions in a view angle and between view angles, and generating a face recognition result of the to-be-recognized image. According to the invention, the accuracy of face recognition is improved.