46results about How to "Short training time" patented technology

The invention discloses a lower limb exoskeleton rehabilitation robot joint torque prediction method and device based on an electromyographic signal, and the method comprises the steps: firstly collecting the electromyographic, joint angle, angular velocity and joint torque data of a user, and carrying out the modeling of four muscles related to a knee joint through a neuromusculoskeletal model; and identifying physiological parameters used in the model by using a sparrow optimization algorithm, and outputting an optimal solution to obtain an estimated torque. Then, the real torque and the estimated torque are endowed with the weights of # imgabs0 # respectively, and a fusion torque is obtained; and training a neural network model by using the collected data and the fusion moment, and meanwhile, enabling the neuromusculoskeletal model to participate in the training process of the neural network model by adopting a method of bringing the neuromusculoskeletal model into a loss function, so that the constraint of the model is increased. And finally, storing the trained model for subsequent calling. The method has the advantages of high precision and strong generalization, and model training can converge more quickly under the condition of less data volume.

The invention relates to the technical field of data processing, and discloses an education data safety emergency response and recovery system and method based on artificial intelligence. The method comprises the following steps: firstly, acquiring education-related data, generating an initial education data matrix, calculating sensitivity, similarity and association strength, and performing risk assessment on the initial education data matrix; secondly, performing network anomaly detection on the processed education data matrix, and starting an emergency response scheme when the education data is invaded by a network in an abnormal state; a BP neural network is trained to obtain a BP neural network model, and education data risk prediction is realized; and finally, according to an educational data risk prediction result, missing or damaged educational data is recovered, network security situation assessment is carried out, and advanced security emergency response is realized. By analyzing and evaluating the education data, the purpose of safety emergency response and recovery is achieved, and the method is accurate and objective.

The invention discloses a bearing fault diagnosis method integrated with a support vector dynamic neural network, and the method comprises the following steps: obtaining vibration data of a key position of a bearing of a rotating machine, carrying out the data preprocessing of the vibration data, and obtaining a preprocessed vibration signal data set; constructing a support vector machine model with a quadratic programming problem form based on the vibration signal data set, and designing a corresponding penalty function and a Lagrange function to simplify the support vector machine model; designing a variable parameter dynamics solving equation, and establishing a final integrated support vector dynamic neural network bearing fault diagnosis model to solve the support vector machine model; and simultaneously training a plurality of support vector dynamic neural network bearing fault diagnosis models based on a bagging classifier method, carrying out bearing state diagnosis on the vibration signal data set, and outputting a diagnosis result. According to the invention, the accuracy of the machine learning method, the convergence proof of the neural network solver and the interpretability of the intelligent fault diagnosis model are improved.

The invention belongs to the technical field of electric power prediction, and particularly relates to an electric power load prediction method based on relation enhancement supervised learning, which comprises the following steps: acquiring historical data related to electric power load from an electric power system monitoring station, preprocessing the data, and analyzing the correlation between enterprises so as to establish an enterprise correlation graph; a neural network model GL-STGNN is established to train the neural network model GL-STGNN, the predicted data of the power load in a future set time period overcomes the defects in the prior art, frequency domain and time domain multi-dimensional global time sequence characteristics extracted by a global GRU model are fused, a local model of a node is introduced through a node embedding module, and the overall performance of the power load is improved. According to the method, global-local feature fusion is completed through splicing, so that global features and local effects are fully considered, related features are captured to the maximum extent, and the model has the accurate prediction capability.

The invention discloses a secondary screen cabinet wiring terminal alternating current and direct current voltage identification circuit and a separation and classification method, and relates to the technical field of power electronics and signal identification. According to the invention, firstly, a pre-amplification and AC / DC separation circuit mainly composed of a pre-amplification module, a high-pass filtering module and a low-pass filtering module is used for completing amplification and AC / DC separation after voltage signals of a secondary screen cabinet wiring terminal are acquired; converting the time sequence of the AC voltage signal into image data through a Grubby angle difference field; and then optimizing by using the image data through establishing and training a model based on a convolutional neural network, and continuously iterating to identify and classify the optimal parameter of the AC voltage signal of the secondary screen cabinet wiring terminal. Finally, after model training is completed, the test set is input into the model to determine the accuracy of the model.

The invention discloses a microbial agent for promoting cucumber growth and a preparation method and application thereof, and belongs to the technical field of agricultural microorganisms. The microbial agent is prepared from the following components: achromobacter denitrificans, bradyrhizobium angularis and Vilnelandan nitrogen-fixing bacteria in a volume ratio of 2 to 1 to 1. Bacteria with a growth promoting effect are reasonably combined for use, so that the growth promoting effect of the mixed flora on cucumbers is better than that of a single strain, and the adaptability of the synergistic effect of the strains in the actual soil rhizosphere complex environment is higher. By determining the number of strains and mixing according to a specific proportion, germination of cucumber seeds is stably and effectively promoted, root elongation is promoted, the flowering and fruit setting rate is increased, and meanwhile the plant height, plant biomass and chlorophyll content of cucumbers in the seedling stage are increased. Besides, the microbial agent disclosed by the invention is beneficial to improving the growth quality and yield of plants and reducing environmental pollution, has wide application potential in the fields of agriculture and ecology, and is worthy of popularization.

The invention discloses a multi-microgrid coordination control method based on deep learning, which relates to the technical field of multi-microgrid coordination control and comprises three parts of multi-microgrid model construction, decision mathematical model construction and deep reinforcement learning-based model construction. Wherein the multi-micro-grid model is a novel micro-grid model composed of a distributed power supply for wind power generation, a public energy storage system, a group of constant-temperature control loads such as a park constant-temperature system, a heating system and a refrigerating system, and a group of park loads; wherein the decision mathematical model adopts a Markov decision process; wherein an AC algorithm is adopted for construction based on the deep reinforcement learning model. According to the multi-microgrid coordination control method based on deep learning disclosed by the invention, the minimum economic operation cost of the multi-microgrid is taken as an optimization target, the economic dispatching optimization of the microgrid is realized, the algorithm training time is shorter, the economic benefit of a dispatching strategy given by an algorithm model is higher, and the optimization efficiency is higher. And the influence of wind power volatility on safe and economical operation of the micro-grid can be effectively reduced.

The present invention belongs to the biological field, and specifically relates to a strain of Lentilactobacillus buchneri that ferments and produces a variety of floral and fruity flavor substances and its use. In view of the fact that there are fewer species of Lentilactobacillus buchneri in the existing liquor brewing process, the research on the flavor substances, especially aroma substances, produced by the fermentation of Lentilactobacillus buchneri is blank, and the present invention provides a strain of Lentilactobacillus buchneri and its use. It can simultaneously produce floral and fruity alcohol substances, including ethyl elaidate, n-octanol, n-hexanol, 1-nonanol, phenylethyl alcohol, methyl heptenone, and geranyl acetone. Therefore, the strain can be used to prepare natural floral and fruity flavor substances, used for the optimization of fermented grains microorganisms, and strengthen the abundance of aroma-producing microorganisms in the liquor brewing process, and has good industrial application prospects.

The application provides a wheelchair control system based on motor imagery electroencephalogram signals and surface electromyogram signals, and relates to the technical field of brain-computer control, comprising a data acquisition module, a system mode selection module, a signal processing module and a decision module; the data acquisition module is used for acquiring physiological signals of a wheelchair user and sending the physiological signals to the system mode selection module; the system mode selection module is used for selecting a processing mode of the physiological signals; the signal processing module is used for pre-processing the physiological signals and extracting signal features for identification; the signal processing module sends an identification result to the decision module; the decision module is used for fusing multiple instruction information and converting the instruction information into a final wheelchair control instruction, and outputting the control instruction to the wheelchair. The application can improve the operation accuracy of the control system, reduce the fatigue of the user when performing motor imagery behavior, improve the user experience and the application performance of the wheelchair control system, and expand the user range of the brain-controlled wheelchair.

The invention relates to the field of robot control, aims to overcome the defects in the background technology, and provides a masked palm robot-imitating deep reinforcement learning control method so as to shorten the training time, improve the learning efficiency and improve the learning effect. According to the technical scheme, the method comprises the following steps of 1, constructing a D3QN model; 2, setting a space state and a reward function; step 3, acquiring a bionic swimming track of the rear foot joint of the masked civets; 4, the bionic swimming robot is controlled to move according to the bionic swimming track; and step 5, training the D3QN model. According to the invention, appropriate actions can be quickly learned in a real underwater environment, the environment is better adapted, the training efficiency and effect are also improved, and experimental results show that the robot can reach a stable swimming state in synchronous and staggered bionic swimming modes.

The invention discloses a graph neural network accelerated training method for a GPU, a storage medium and an electronic device, and the method comprises the steps: obtaining an initial graph, segmenting the initial graph into N sub-graphs, and enabling the initial graph to be a static graph corresponding to a moment in a dynamic graph sample; calculating loads of the sub-graphs are estimated, the N sub-graphs are distributed to the M GPUs based on the calculating loads of the sub-graphs, training of the graph neural network is carried out, and at least two sub-graphs are distributed to each GPU; incremental data is obtained, the subgraph is updated based on the incremental data, and the incremental data is used for representing the change of a static graph in the dynamic graph sample scanned at the current moment relative to the static graph currently processed by the GPU; and distributing the N sub-graphs containing the incremental data to M GPUs and carrying out graph neural network training. According to the technology, the calculation and resource utilization rate is improved, the training time consumption is reduced, and the training efficiency of the graph neural network during training is improved.

The invention discloses a 3D target detection deep neural network model. The 3D target detection deep neural network model comprises a data acquisition module which is mainly used for acquiring 3D target data required by network model training; the data classification module is mainly used for carrying out pre-classification on the obtained 3D target data; a data training module; the data storage module is used for preferentially storing the trained data and taking the data as subsequent samples; and a sample set storage module. According to the 3D target detection deep neural network model, the data classification module classifies the data to be trained in advance, so that the situation that the training time is influenced due to differences of multiple pieces of data is avoided, and the processing efficiency of the deep neural network model is improved; and secondly, the sample set storage module takes the trained data as samples to carry out preferential storage for subsequent training, so that the data demand is reduced, and normal training is ensured at the same time.

The invention relates to a dynamic gesture recognition method based on hand skeleton data, and belongs to the field of medicine, the method comprises the steps of S1, data acquisition and processing, S2, dynamic gesture recognition based on hand skeleton data, S3, C3d model optimization adjustment, C3d model data input size adjustment, and maximum pooling layer addition. The method can better adapt to the recognition requirements of the special gestures of the stroke patient, the size change is matched, and the final feature dimension is not affected. The dynamic gesture recognition based on the hand skeleton data plays a very important role by adjusting the size of the input data, as the features contained in the hand skeleton data are much simpler than those contained in RGB data, the model calculation amount is greatly reduced, and redundant information in the aspects of background, illumination change, hand appearance and the like is also eliminated, so that the dynamic gesture recognition accuracy is greatly improved. And the model can focus on key features more quickly, so that the training time is remarkably shortened.

The invention provides a method for improving spore production of natamycin producing bacteria, belongs to the technical field of streptomyces culture, and mainly provides an FtsZ gene, and the nucleotide sequence of the FtsZ gene is shown as SEQ ID NO.1; the FtsZ gene is applied to promotion of spore production of streptomyces gilvosporeus; the invention provides a genetically engineered bacterium for improving the sporulation quantity of streptomyces gilvosporeus. The genetically engineered bacterium is obtained by introducing an FtsZ gene and an erythromycin strong promoter ermEp into the streptomyces gilvosporeus through a carrier, a strain of streptomyces gilvosporeus F607-F is provided, the streptomyces gilvosporeus F607-F is preserved on October 28, 2024, and the preservation number of the streptomyces gilvosporeus F607-F is CCTCC M 20242352; the invention also provides a solid culture medium for promoting spore production of streptomyces gilvosporeus. The genetically engineered bacterium F607-F provided by the invention has the spore production advantage and can produce natamycin at high yield.

The invention discloses a social network link prediction-oriented time sequence diagram network parallel training acceleration method. The method comprises the following steps: firstly, calculating a redundancy score and an old score for each user interaction edge in a training set, and calculating a comprehensive information loss score according to the redundancy score and the old score; secondly, according to a preset core set retention proportion alpha, selecting all user interaction side comprehensive information loss scores and upper alpha quantiles of repetitiveness as threshold values, and discarding user interaction with the comprehensive information loss scores lower than the threshold values, so that a simplified core set is obtained through single-time preprocessing; then carrying out adaptive batch division on the obtained core set, and dynamically determining an acceptable maximum user interaction number in each batch according to a redundancy and old comprehensive information loss score; and finally, training the time sequence diagram neural network based on the divided batches to realize social network link prediction. The method not only improves the training efficiency, but also greatly improves the precision of the model.

The present invention discloses a method for simplified identification and performance prediction of uplink interference in a wireless system, comprising the following steps: step 1, determining the interference model required for simplified identification and performance prediction of interference; step 2, collecting user data generated when the wireless system is running; step 3, simplified interference identification; step 4, wireless performance prediction. The advantages of the present invention are that simplified interference identification is proposed, with signal user indication vector, interference user indication vector and power offset vector as features, UL-SINR measurement as labels, and supervised machine learning algorithm is used for training to obtain link-level interference relations, and a complete and accurate uplink interference relationship modeling scheme is provided without the need for additional physical equipment and user complete measurement reports, thereby realizing low-complexity, fast, accurate and complete link-level interference identification.

The application provides a wall-penetrating target behavior recognition method based on channel state information, which comprises the following steps: preprocessing the collected CSI signals; performing PCA data dimension reduction on the preprocessed CSI signals, removing redundant signals and irrelevant information, and extracting optimal subcarriers; performing first-order difference processing on the signals extracted after PCA dimension reduction, and then using a method based on a buffer sliding window to segment effective feature signal segments; converting the effective feature signal segments into feature images with time-frequency domain features through STFT, and inputting the feature images into a pre-trained SE-ResNet18 convolutional neural network for behavior recognition classification. The method can penetrate the wall to realize the behavior recognition of the human target behind the wall. Compared with the traditional deep learning network, the method adopts a small sample transfer learning method combined with a pre-trained model, and the recognition accuracy can reach 91.67% under the conditions of fewer training times, fewer iteration times and shorter training time, so that the behavior recognition classification task can be effectively completed.

This invention relates to a prediction-based robotic arm path planning method, apparatus, electronic device, and medium. The method involves real-time acquisition of hand position data and storage in a raw dataset, which also includes the current position data of the robotic arm, target position data, and actual obstacle position data. Based on the raw dataset, an Informer model is trained to obtain model weights and output predicted hand position data. The predicted hand position is used as a virtual obstacle, and a predicted repulsive potential field is obtained based on the predicted hand position data. Based on the predicted repulsive potential field, the repulsive potential field of the actual obstacle, and the gravitational potential field of the target position, the current resultant force is calculated to optimize the robotic arm's trajectory, control the robotic arm's movement, and perform sorting. The target position is the endpoint of the robotic arm's movement. Compared with existing technologies, this invention has advantages such as higher prediction accuracy, faster calculation speed, and higher human-machine collaboration efficiency.

The present application discloses a fault diagnosis method based on random walk and machine learning. First, the vibration signal data of the moving parts are collected and preprocessed. For the preprocessed vibration signal, the fault feature extraction algorithm is used to extract the feature parameters, and the obtained feature parameters are subjected to principal component analysis PCA. Then, a random walk optimization algorithm is constructed, and the random walk optimization algorithm is used to optimize the hyperparameters of the machine learning model. Then, the sample data after PCA dimension reduction is divided into a training set and a test set, and the training set is put into the optimized machine learning model for training. The test set is used for fault prediction to obtain the prediction accuracy under different features. Finally, a comparative analysis is performed to select the feature with the highest accuracy as the best feature. This method solves the problem that the traditional algorithm takes too long to find the optimal solution and cannot find the global optimal solution. While ensuring the prediction performance of the model, the prediction accuracy of the model is significantly improved, thereby improving the efficiency of fault diagnosis.

The present invention discloses a training method, terminal and storage medium for accelerating a pulse neural network, the method comprising: obtaining a pulse sequence and an initial synaptic weight for training, generating an input signal of a pulse neuron according to the pulse sequence and the initial synaptic weight; obtaining an actual output pulse sequence generated based on the input signal; obtaining an expected output pulse sequence, updating the initial synaptic weight according to the actual output pulse sequence and the expected output pulse sequence, so as to realize training of the pulse neural network. The present invention reads image data through a host device and converts it into a pulse sequence and sets initial weights. Then, the most time-consuming parameter update task in the pulse neural network training is handled by a special acceleration device, which improves the speed of parameter update and accelerates the training process of the pulse neural network. Thereby solving the problem that the training process of the pulse neural network in the prior art is time-consuming.

The invention discloses an installation and maintenance personnel equipment image detection method and system based on a gating kernel self-attention Transform and an inverse residual multi-scale convolution module, and relates to the technical field of image detection. The method comprises the following steps: acquiring image data of a worker on an installation and maintenance site, and preprocessing the image data to obtain target image data; performing multi-scale feature information extraction on the target image data by adopting a preset inverse residual multi-scale convolution module to obtain corresponding feature information; on the basis of the feature information, feature coding is carried out by utilizing a gating kernel self-attention Transform, so that corresponding enhanced feature information is obtained; and respectively inputting the enhanced feature information into a plurality of target detection heads with different scales for feature decoding so as to obtain corresponding target categories and corresponding target sizes, and completing image detection. According to the invention, the safety of the current staff can be evaluated efficiently and accurately, the possibility of potential safety hazards is reduced, and the standardization of installation and maintenance staff is improved.

The present invention relates to a model construction method and training method for multi-task language recognition, including: using an unsupervised pre-training method to perform a first pre-training on a wav2vec model; establishing a training sample set containing multi-language sample data according to task requirements including language recognition and data effective recognition; using sample data in the training sample set to perform a second pre-training on the wav2vec model pre-trained for the first time; adaptively fine-tuning the wav2vec model pre-trained for the second time according to the task requirements to construct a final multi-language recognition model; and using sample data in the training sample set to train the multi-language recognition model again, so that the multi-language recognition model can simultaneously recognize the language and voice effectiveness of input voice data. The present invention realizes the simultaneous recognition of two tasks, language recognition and effective voice recognition, and improves the accuracy and recognition efficiency of language recognition.

This invention relates to a remote sensing extraction method for pear orchard areas based on the Re-UNet model, which overcomes the shortcomings of inaccurate classification results and low efficiency in pear orchard area extraction from remote sensing images compared with existing technologies. The invention includes the following steps: acquiring a remote sensing image dataset; constructing the Re-UNet pear orchard area extraction model; training the Re-UNet pear orchard area extraction model; acquiring and preprocessing the remote sensing images of the pear orchard areas to be segmented; and obtaining the remote sensing extraction results for the pear orchard areas. Based on the UNet semantic segmentation model, this invention solves the overfitting problem that easily occurs in small datasets. It also incorporates spatial and channel attention mechanisms and a residual module, further enhancing the feature transfer and cumulative integration characteristics of pear orchard areas in high-resolution remote sensing images, effectively reducing the "salt and pepper" phenomenon and misclassification, and improving the overall segmentation accuracy.

The present disclosure discloses a training method, device and system of a large model-based risk identification system, relating to the technical field of artificial intelligence such as machine learning and natural language processing. The specific implementation scheme comprises: adopting a first training sample, fine-tuning a pre-trained large language model, so that the large language model learns to identify the intent information of the query sequence of the user; adopting a second training sample, training a risk identification system including the fine-tuned large language model and a risk identification model, so that the risk identification system learns to identify the risk information of the user; wherein during the training process, only the parameters of the fine-tuning layer in the fine-tuned large language model are adjusted.

The invention relates to a speech language category recognition method, which comprises the steps of providing a target speech audio, and loading the target speech audio into a pre-constructed language recognition model so as to perform language recognition processing on the target speech audio by using the language recognition model, when the language recognition model carries out language recognition processing on the target voice audio, the language recognition processing at least comprises voice feature extraction processing, language embedding extraction processing and recognition classification processing which are executed in sequence, and after the voice feature extraction processing is executed, corresponding voice feature information is generated based on the target voice audio; after language embedding extraction processing is executed, corresponding language embedding information is generated based on the voice feature information; and after recognition and classification processing is executed, mapping the language embedded information into language classification information representing a language corresponding to the target voice audio. According to the invention, efficient recognition of multiple languages can be realized, and the accuracy and recognition efficiency of language recognition are improved.

The application provides a medical image segmentation method and system based on a CNN-Transformer parallel encoder, and relates to the technical field of medical image segmentation, and the specific steps comprise: inputting a standardized medical image into a parallel encoder, extracting local texture and global semantic feature maps, and aligning the resolution and spatial position through a spatial correlation matrix; screening lesion features based on an anatomical structure prior feature set, dynamically allocating weights to obtain fusion features; taking a lesion gold standard mask as a label, obtaining an initial lesion probability feature map through a decoder, and calculating a bias value; iteratively optimizing a parameter combination to output an optimal configuration parameter combination, and obtaining a final lesion probability feature map, which is classified and activated, thresholded, and outputted as a segmentation mask. The application effectively avoids the local optimal solution of parameter optimization, combines a bias value threshold-driven rapid screening mechanism, fine-tunes parameters for different modal images without full retraining, and effectively breaks through the bottleneck of poor adaptability of the prior art.

The invention belongs to the field of biotechnology, and specifically relates to Enterococcus gallinarum producing 3,4-dimethylbenzaldehyde and its application. In order to enrich the microbial sources of 3,4-dimethylbenzaldehyde production, the present invention screens and obtains a strain of Enterococcus gallinarum with a strong ability to produce 3,4-dimethylbenzaldehyde from nature, and the deposit number is CGMCC NO.25148. The strain has potential application prospects and application value in preparing microbial agents, freeze-dried bacterial powder, pest attractants, etc. containing 3,4-dimethylbenzaldehyde.

The embodiment of the invention provides a parasitic parameter extraction neural network distributed training method and device, and relates to the technical field of integrated circuits. According to the method, a local batch of each processing device is obtained by using a data pushing process in each training batch, the data reading efficiency is improved, the memory is saved, and parallel training is performed on a parasitic parameter extraction neural network on the processing device by using a Horovod parallel framework to obtain a local gradient. Local BN layer statistical parameters of each processing device are obtained to calculate global statistical parameters, the global statistical parameters are distributed to each processing device to perform BN layer data updating, local gradients are updated at the same time, all local gradients are synchronized by using a Horovod framework to determine model updating parameters, and the model updating parameters are updated; and training is continued until the trained parasitic parameter extraction neural network is obtained. On the basis of a Horovod parallel framework, cross-card-batch normalization synchronization operation is realized, the situation that training is not converged due to the fact that local batches are too small is avoided, the training precision can be maintained, and the training efficiency can be improved.

The present invention relates to an algorithm for using a GRU neural network to assist navigation during a GNSS signal interruption. When the GNSS signal is available, the speed, heading and output of the IMU of the INS will be used as input features for training the GRU model, and the position increment of the GNSS is the output of the model. Once the GNSS signal is lost, the information of the INS will be fed into the GRU model to generate a pseudo-GNSS increment. After accumulating the pseudo-GNSS increment, the pseudo-GNSS position will be sent to the Kalman filter to correct the INS navigation solution result. In addition, considering that it is difficult to obtain a large amount of GNSS observation data with good observation in complex scenarios, the number of observable satellites in some areas may be 1 to 3. At this time, the use of a tightly coupled model can still perform combined navigation, which guarantees a more effective positioning result to a certain extent and can provide better training data for the neural network. Therefore, the present invention selects a GRU-assisted tightly coupled mode.

The invention provides a WOA-GRU-based hydrogen atomic clock fixed parameter control method and device. The method comprises the following steps: acquiring and preprocessing a historical operation parameter time sequence of a hydrogen atomic clock; constructing a parameter time sequence prediction model taking a gating circulation unit as a core; performing global automatic optimization on the hidden layer size, the discard rate and the learning rate of the GRU network by adopting a whale optimization algorithm; training a GRU model by using the optimized hyper-parameters; and finally, obtaining a group of collaborative optimal fixed parameter values capable of enabling the hydrogen atomic clock to stably operate for a long time by utilizing the trained model in a rolling prediction mode. The GRU network with a more concise structure is adopted to replace a traditional LSTM network, intelligent global search of WOA is combined, the training efficiency and the model stability are remarkably improved while the prediction precision is guaranteed, and the method is particularly suitable for a hydrogen atomic clock control scene with limited data volume and high parameter noise and has good application prospects. Accurate, stable and intelligent fixed control of internal parameters of the hydrogen atomic clock is realized.