88results about How to "Improve training accuracy" patented technology

The invention relates to a power distribution network single-phase earth fault detection method based on a Gramb angle field and an improved convolutional neural network, and the method comprises the following steps: S1, analyzing the characteristics of an earth fault, obtaining measurement data from a power distribution network through a phasor measurement unit (PMU), and obtaining a two-dimensional virtual image from the obtained measurement data through a GAF; s2, through an improved convolutional neural network ICNN, analyzing the obtained two-dimensional virtual image by using a multi-scale feature extraction module and an attention module, and extracting and paying attention to key features between fault data; and S3, constructing a fault detection model of the single-phase earth fault of the power distribution network, dividing the fault detection model into an offline stage and an online stage, and carrying out fault detection according to different online conditions. According to the method, the fault diagnosis performance under different online conditions can be improved.

The invention discloses an in-structure sound field bounded uncertainty analysis method based on a physical information neural network. The method comprises the following steps: establishing a physical information loss function and establishing a physical information driven neural network; obtaining an input uncertain parameter value range of the internal sound field, obtaining a plurality of parameter levels, and training a sound pressure prediction model under the parameter levels; acquiring sound pressure of a sample point and establishing a data set; establishing a data driving residual error based on the data set, establishing an overall loss function by using the data driving residual error and a physical information loss function, and training a parameter-containing physical information driving neural network to obtain a parameter-containing internal sound field sound pressure prediction model; and establishing an interval model, and obtaining an internal sound field sound pressure uncertainty distribution range by using a global search algorithm and the parameter-containing internal sound field sound pressure prediction model. The method can solve the problem of prediction of the sound field in the structure containing uncertain parameters, and can effectively improve the calculation efficiency on the premise of ensuring that the calculation precision meets the engineering requirements.

The invention discloses an insulator defect detection method and system based on RT-DETR improvement. The method comprises the following steps: collecting a disclosed insulator defect data set and obtaining an image as a data set through a video interception frame; the method comprises the following steps: carrying out improvement based on a target detection model RT-DETR, changing a backbone network into a StarNet star network to carry out feature extraction, carrying out further improvement, removing a P5 detection head, adding a P2 detection head, and improving a star convolution module in the StarNet star network; a multi-head attention mechanism is changed into a partial self-attention mechanism PSA, and further improvement is carried out to obtain an S-PSA attention mechanism module; a RepC3 convolution module is improved into a RepMob convolution module; combining the loss function by using Inner-IoU and CIoU to obtain an Inner-CIoU loss function, and carrying out loss calculation; and training the improved RT-DETR, verifying the improved RT-DETR to obtain an optimal weight, and loading the weight into the improved RT-DETR to carry out insulator defect detection. According to the invention, the defect detection is carried out on the insulator by using the improvement based on RT-DETR, so that the detection is more accurate and efficient.

The embodiment of the invention provides an end-to-end automatic driving model training method and device, computer equipment and a medium, and relates to the technical field of automatic driving, and the method comprises the steps: obtaining the driving behavior data of a driver and the first self-vehicle planning data of an end-to-end model, and when the difference value of the two kinds of data exceeds a first preset threshold value, determining the driving behavior of the driver; triggering a data recording mechanism, obtaining preliminary screening data of a period of time before and after the current moment, obtaining second self-vehicle planning data based on the preliminary screening data and an end-to-end model, and carrying out comparative learning on the second self-vehicle planning data and the driving behavior data to obtain a similarity score; screening corresponding data fragments when the similarity score is lower than a threshold value in the primarily screened data, obtaining a virtual driving scene according to the data fragments, and inputting the virtual driving scene into the VLA model to obtain driving related data; and training an end-to-end model by taking the driving-related data as training data. According to the scheme, the quality of training data and the efficiency and precision of model training are improved.

The invention particularly provides a control method and device for self-cleaning of a water tank. The method comprises the steps of monitoring water quality data and cleaning data of each water tank in real time, selectively obtaining environment data of the water tanks according to the water quality data and the cleaning data of each water tank, and substituting the water quality data, the cleaning data and the environment data of the water tanks into a cleaning analysis and evaluation model when the environment data is obtained, obtaining a self-cleaning treatment scheme corresponding to the water tank, executing the self-cleaning treatment scheme on the water tank, obtaining cleaning water quality data of the water tank, and substituting the cleaning water quality data into the cleaning analysis and evaluation model again to obtain a self-cleaning evaluation result of the water tank. And feeding back the self-cleaning treatment scheme and the self-cleaning evaluation result to a user terminal corresponding to the water tank, and selectively cutting off water supply of the water tank based on the self-cleaning evaluation result of the water tank. And through real-time monitoring and timely self-cleaning treatment, water quality pollution is effectively prevented, and the safety of domestic water of a user is guaranteed.

The invention relates to the technical field of Internet of Vehicles, in particular to an Internet of Vehicles federal semi-supervised learning method and system, and the method comprises the steps: in each iteration, a vehicle client firstly generates a pseudo label of unlabeled sample data, then carries out the training of a local model, and uploads the trained model to a server; and after the server side aggregates the local models of all the vehicle clients, the local models and the supervision model trained by using the labeled sample data are aggregated again to obtain a global model of the current iteration, and the global model continues to be used for training the local model in the next iteration. According to the invention, the model training efficiency is improved, the real-time performance and adaptability of the system are further improved, and the coping ability of the vehicle in a complex and changeable traffic environment is enhanced.

The invention provides a health data processing method based on distributed privacy protection learning. The health data processing method comprises the following steps: constructing a federal learning framework to generate a machine learning model; performing an iteration process until the model converges or reaches a preset hyper-parameter; the iterative process comprises: independently training the machine learning model by local data; encrypting local data by a preset encryption algorithm, and calculating a gradient to obtain encrypted gradient information; sharing the gradient information based on multi-party computation; combining the gradient information by a preset aggregation algorithm; and updating the parameters of the machine learning model according to the aggregated gradient, and distributing the updated parameters. According to the invention, on the premise of fully protecting user data privacy, efficient and accurate distributed machine learning model training can be realized, so that the overall performance and security of a health data processing system are improved. The invention further provides a health data processing system based on distributed privacy protection learning and the electronic equipment, which have corresponding advantages.

According to the abnormal wear monitoring method for the disc cutter of the shield tunneling machine, tunneling parameters closely related to the wear state of the disc cutter are screened according to the abnormal wear monitoring requirement of the disc cutter in the tunneling process of the shield tunneling machine, and data missing caused by the collection problem is filled up. According to the long and short term memory self-encoder model based on the abnormal attention mechanism, prior correlation in an original sequence is obtained through a learnable Gaussian kernel, the distinguishability between normal and abnormal conditions is improved by obtaining correlation differences of the sequence, and accurate and reliable abnormal wear monitoring is conducted on the shield tunneling machine disc cutter in combination with the extreme value theory. According to the method, the tunneling parameters closely related to the running state of the disc cutter are identified more accurately, and the model training precision and the accuracy of abnormal wear monitoring of the disc cutter are improved. The method is of great significance to the application of abnormal wear monitoring of the disk cutter of the shield tunneling machine.

The invention discloses a boiler NOx emission estimation method based on a width learning system and integrated learning. The boiler NOx emission estimation method comprises the steps of 1, data processing; step 2, constructing an integrated learning prediction model; step 3, performing real-time estimation; according to the invention, the hybrid quantum dune cat swarm algorithm QSCSO is provided, a Levy flight strategy and a quantum strategy are added on the basis of the SCSO for improvement, the problems of low convergence precision, low optimization efficiency and the like caused by limitation of global search and easy falling into local optimum of the SCSO are solved, and a modeling method based on QSCSO-BLS is designed for individual learning device construction. According to the method, the training precision and generalization ability of the BLS model during modeling of the single-working-condition system are improved, meanwhile, the boiler NOx emission estimation method based on QSCSO-BLS and integrated learning is provided, and the problem that the model generalization ability is poor during modeling of the BLS full-working-condition system is solved.

The invention discloses a training method of a small sample image detection model and an image detection method, and belongs to the technical field of deep learning, and the base class training method comprises the steps: training an original image detection model through employing a support image set, a marked query image, and each target class in a class until the total loss is converged, obtaining a base class image detection model; support features and semantics of a target category are embedded and input into a variational auto-encoder, and compared with support features and reconstruction features of a specific category, potential vectors in a VAE intermediate process contain general features of more types of prototypes; potential vectors containing semantic information are sampled from the distribution, and feature aggregation is carried out on the potential vectors and query features, so that the estimated new class center is closer to a real class center; in addition, the reconstruction loss, the KL divergence, the classification loss, the regression loss and the semantic consistency loss in the VAE form the total loss, so that the training precision can be greatly improved.

The invention discloses an inversion mapping model training method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring to-be-trained seismic data and a wave impedance tag corresponding to the to-be-trained seismic data; training an initial learner based on the to-be-trained seismic data and the wave impedance tag corresponding to the to-be-trained seismic data to obtain an active learner model; obtaining non-labeled seismic data, and inputting the non-labeled seismic data into the active learning device model to obtain a wave impedance label of the non-labeled seismic data; and training a neural network model based on the unlabeled seismic data and the wave impedance tag of the unlabeled seismic data to obtain a trained inversion mapping model. According to the technical scheme, the wave impedance tag of the unlabeled seismic data is generated through active learning simulation, so that training of the inversion mapping model is completed under a small amount of labeled information, and the training precision of the inversion mapping model is improved.

The invention discloses a child health monitoring method and device based on a smart bracelet. The method comprises the following steps: S1, acquiring a training data set; s2, performing training processing on a preset children physiological monitoring model by using the training data set to obtain a trained children physiological monitoring model; s3, acquiring a child physiological parameter set by using the smart bracelet; and S4, processing the child physiological parameter set by using the trained child physiological monitoring model to obtain child health monitoring result information. The child health monitoring result information is used for representing the health degree information of the child. According to the invention, real-time and accurate monitoring of the health condition of the child is realized by using the bracelet.

The invention provides an image processing method and device, electronic equipment and a storage medium, and relates to the technical field of image processing. Comprising the following steps: extracting at least one foreground object image from each acquired initial image; determining scene depth information corresponding to each scene background image according to each collected scene background image; determining a target area in each scene background image according to each scene background image; and according to the scene depth information corresponding to each scene background image and the target area in each scene background image, embedding the at least one foreground object image into the at least one scene background image to generate a plurality of composite images. According to the method, expensive manual data annotation is replaced by a synthetic image generation mode, accurate label information can be provided for training of the downstream detection model due to the fact that the foreground object in the synthetic image has a clear identifier, and the training accuracy of the downstream detection model can be improved on the premise that time and labor consumed by manual annotation are reduced.

The invention discloses an electrical equipment fire detection method and system based on an improved Faster R-CNN. The method comprises the steps that flame and smoke pictures are acquired and marked; performing training improvement on the Faster R-CNN network, and performing dynamic picture proportion enhancement on the obtained flame and smoke pictures; and performing target detection on the acquired flame and smoke pictures based on the trained and improved Faster R-CNN network. When the improved Faster R-CNN is used for training, dynamic picture proportion enhancement is carried out on an input picture, the training precision of the neural network is improved, through improvement of the Faster R-CNN, the trained model can be used for a complex scene detection task, and the flame and smoke detection precision is better than that of a traditional model.

The invention discloses a microorganism image recognition method, device and equipment based on deep learning, a medium and a product, and relates to the field of microorganism detection. The method comprises the following steps: acquiring images of various microorganisms in the activated sludge, and labeling the types of the microorganisms in the images; carrying out data enhancement on the images, and forming an activated sludge microorganism image data set together with the marked types of microorganisms in the images; training a Faster RCNN network by using the activated sludge microorganism image data set to obtain a microorganism recognition model; and inputting the image of the to-be-detected microorganism into the microorganism identification model, and identifying the category of the to-be-detected microorganism. According to the method, the microorganisms in the activated sludge can be efficiently and accurately detected and classified.

The invention discloses a federal learning performance enhancement method and system based on a variational auto-encoder. The method comprises the following steps: initializing model parameters of the variational auto-encoder; constructing a federated learning optimization algorithm based on a variational auto-encoder, and initializing a neural network model; dividing the data set, and distributing a training set for each client; training a neural network model by using the training set, updating model parameters in each round of federated learning aggregation, and issuing the model parameters to the client; the client side trains encoder parameters, updates hidden space Z and decoder parameters and uploads the parameters to the server side; the server side adopts a distillation technology to take the decoder of the client side as a teacher model, and a global decoder is trained; processing the updated global decoder by using a cross attention mechanism; carrying out weighted average on model parameters, and continuously iterating to complete training; according to the method, the performance and the accuracy of the whole model framework are remarkably improved while the privacy of the customer is ensured.

The invention discloses a network security situation assessment method, system and device, and relates to the field of network security, and the method comprises the steps: obtaining a network traffic data set; the network flow data set comprises multiple pieces of network flow data in a set time period; based on a traffic classification model, classifying each network traffic data in the network traffic data set to determine an attack type of each network traffic data; the traffic classification model is obtained by training based on a training set by adopting a sparrow search algorithm in advance, and the training set comprises multiple groups of network traffic sample data and corresponding attack types; the traffic classification model is constructed based on a multi-scale convolution attention mechanism and a residual network; and calculating a network security situation value according to the attack type of each piece of network traffic data so as to determine a network security situation corresponding to the network traffic data set. According to the invention, the accuracy of network security situation assessment is improved.

The present invention relates to the technical field of medical record data risk assessment, and specifically to a preoperative anesthesia risk assessment method and system based on electronic medical record information. The electronic medical record data and anesthesia risk assessment level of patients undergoing preoperative anesthesia in a database are obtained, and an anesthesia risk assessment model is trained using the GBDT algorithm. During the training process, the impact of medical history and medical history combination on the anesthesia risk assessment level is considered, so as to optimize the loss function in each iteration, and improve the model training accuracy by evaluating the importance value of medical record data and adaptively adjusting the loss function, so as to enhance the model's sensitivity to medical history information. After the training is completed, the model can accurately predict the anesthesia risk assessment level, thereby assisting anesthesiologists in formulating safer and more reasonable anesthesia plans.

The present application relates to a kind of based on battery state's internal welding machine control method, device, equipment and medium, the method includes: according to the battery working parameter of internal welding machine, determine the SOC battery state of the battery, according to the task target and task target information of the internal welding machine, determine the load power demand of each power unit of the internal welding machine, according to each the load power demand and the SOC battery state, determine the target torque of the motor corresponding to each the power unit that meets the first constraint condition of pre-set, to control the internal welding machine based on each the target torque corresponding task target is executed.Through the method of the present application, based on the task target and task target information of internal welding machine, the load power demand of each power unit corresponding to internal welding machine when completing task target can be accurately determined, since each target torque determined meets the first constraint condition, then battery damage can be avoided.

The application relates to a liver cancer CT image segmentation method based on channel attention, a segmentation network for liver and tumor regions is constructed, residual modules are used to replace ordinary convolution layers on the basis of a U-Net encoding-decoding structure, the training speed and precision of the network are improved, an attention mechanism is added in the decoder stage, the importance and mutual dependence of different channels in a feature map are fully considered, feature maps output by each stage of the decoder are up-sampled to the size of an original image through a deep supervision module and are spliced with the feature map output in the last stage, feature loss is prevented, two same segmentation networks are used for segmenting a liver region of interest and segmenting tumor cells respectively. The method has good segmentation effect without significantly increasing the parameter amount; from the clinical point of view, liver and tumor regions in an abdominal CT image are accurately segmented, a good auxiliary role is provided for doctor diagnosis, and the diagnosis efficiency is improved.

The invention discloses a user identification method and device, equipment and a storage medium, and is applied to the technical field of artificial intelligence. The method comprises the following steps: acquiring user contribution feature data of a user to be identified; inputting the user contribution feature data into a pre-trained user identification model to perform high-quality user identification to obtain a user identification result; wherein the user identification model is obtained by pre-training a preset identification network model based on the contribution feature sample data and the contribution feature weight of the sample training object. According to the technical scheme of the embodiment, accurate identification of high-quality users in the financial field is realized.

The application discloses an electromagnetic compatibility fault prediction method based on a knowledge graph, relates to the technical fields of electromagnetic compatibility fault diagnosis and artificial intelligence, and constructs an electromagnetic compatibility fault knowledge graph through multi-source heterogeneous data, realizes the structured fusion of data, provides nodes and relations with physical meanings for a Bayesian network, makes the target Bayesian network have interpretability and be capable of tracing fault sources, adopts ALS-PSO to optimize the target Bayesian network, can effectively improve the training precision of the target Bayesian network, and thus can make the target Bayesian network after training accurately realize the prediction of electromagnetic compatibility faults.

The invention discloses a parallel training method and device of a global model, equipment, a medium and a product, and the method comprises the steps: carrying out the grouping of each calculation node, and obtaining a plurality of calculation groups; training the computing nodes in each computing group by adopting an asynchronous strategy so as to update the local model corresponding to each computing group; determining an outdated group in all the calculation groups, and adding the outdated group into an outdated group; when the number of outdated groups in the outdated group reaches a preset first threshold value, respectively calculating the training precision of each outdated group, calculating the training precision mean value of the outdated group, and recording the training precision mean value as a first training precision mean value; and updating the model parameters of the global model according to the local model parameters of the outdated group and other non-outdated groups with the training precision exceeding the first training precision mean value. By adopting the method, the parameter updating condition and updating means of the global model are improved, and the balance between the model training precision and the model updating interval is realized.

The present invention discloses a three-phase line current wide range measurement device and method based on TMR, the measurement device includes a TMR mutual inductor array and a data processing unit module; the mutual inductor array is composed of M rows and 3 columns of TMR mutual inductors, and each column of mutual inductors is correspondingly arranged vertically below the three-phase line, and each row of mutual inductors has a different range; the data processing unit module calculates the three-phase line current under the corresponding range according to the data collected by each row of mutual inductors; selects a row of mutual inductors that is not over-range and has the smallest error at the current moment, and obtains the corresponding three-phase line current as the initial current; selects a row of mutual inductors that is currently over-range and has the largest range, and uses the predicted current calculated by the current to correct the initial current, and obtains the final current of the three-phase line at the current moment. The present invention maintains high-precision measurement of the three-phase line current while greatly improving the range, and solves the problem of data mutation that may occur when switching between different ranges.

The present invention discloses a feature selection method, device, computer equipment and storage medium for image samples. The method comprises: obtaining a feature set including image sample features; performing redundant feature screening on the image sample features in the feature set; performing Schmidt orthogonalization processing on each redundant feature subset to obtain N orthogonalized feature subsets, and using a preset Wrapper method to perform feature screening on the N orthogonalized feature subsets, and forming the screened features into a first feature subset; calculating the correlation between each image sample feature in the feature set and a preset target category feature, and screening the image sample features related to the target category feature according to the correlation to form a second feature subset; performing a union operation on the non-redundant feature subset, the first feature subset and the second feature subset to obtain a new feature set. The technical solution of the present invention can improve the training precision and recognition accuracy of the image recognition model.

The invention provides a data set generation method, electronic equipment and a storage medium. The method comprises the following steps: determining a reference object belonging to the same category as a to-be-matched object in a virtual scene; under the condition that the to-be-matched object and the reference object are registered, obtaining a transformation matrix based on the object similarity between the to-be-matched object and the reference object; on the basis of the transformation matrix, updating the local coordinate system of the object to be matched into the local coordinate system of the reference object; based on the updated local coordinate system of the to-be-configured object, determining a posture of the to-be-configured object on a target object in the virtual scene; and obtaining a target picture corresponding to each to-be-matched object on the target object under the posture, and constructing a data set based on each target picture. According to the method, the construction efficiency of the data set and the precision of training the model by using the data set can be improved.

The invention relates to a model training method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps of receiving multiple groups of first model parameters fed back by multiple clients, determining the first model parameters as sample points to obtain a sample data set, determining distance inverse number density of the sample points for each sample point based on the distance between the sample point and adjacent sample points of the sample point, and determining the distance inverse number density of the sample points based on the distance inverse number density. The method comprises the following steps: determining a local density correction term of a sample point, clustering a sample data set based on the distance inverse number density and the local density correction term, determining the local density of the sample point, and based on the local density of each sample point, dividing clients with similar data distribution into the same client set to obtain a plurality of client sets, and scheduling the plurality of clients. Clients in the same client set share model parameters, and the model is jointly trained in a federal learning mode. By adopting the method, the model training precision can be improved.

The application relates to a large language model assisted domain self-adaption database query statement generation model training method and device, computer equipment and a storage medium. The method comprises the following steps: constructing a sample prompt word corresponding to a sample query question according to the sample query question and sample database mode information of a sample database corresponding to the sample query question; inputting the sample prompt word into a large language model to obtain a sample initial database query statement corresponding to the sample query question; performing pruning processing on the sample database mode information according to the sample initial database query statement to obtain pruned sample database mode information; and training a database query statement generation model by using the sample query question, sample format specification information corresponding to the sample query question and the pruned sample database mode information to obtain a trained database query statement generation model. By using the method, the training accuracy of the database query statement generation model can be improved.

The invention relates to a hierarchical federal learning method oriented to a heterogeneous environment, and belongs to the technical field of artificial intelligence. The method comprises the following steps: constructing a hierarchical federal learning framework in a cloud edge end scene; constructing a hierarchical federated learning algorithm capable of reducing model performance loss caused by heterogeneity between nodes; constructing a virtual data set used in the edge server enhancement algorithm and a data distribution mode of the edge server; under the strict mathematical hypothesis, the boundary of the average global gradient deviation is theoretically deduced to quantify the influence of the cloud edge aggregation interval, the number of participating devices and other important factors, and the theoretical convergence boundary of the hierarchical federated learning algorithm under the influence of the heterogeneous environment is analyzed.

The present invention relates to the technical field of thermal power plants, and in particular to a method and system for predicting the load of a thermal power plant. The method comprises the following steps: obtaining real-time operation data of a thermal power plant, including historical load data, unit operation parameter data, environmental parameter data, and power grid dispatching data; performing data missing processing based on a time series interpolation method on the real-time operation data, and detecting and correcting abnormal values ​​to obtain operation preprocessing data; performing multi-scale time feature extraction on the operation preprocessing data to obtain time feature data, wherein the multi-scale time feature includes hourly features, daily features, and weekly features; and constructing a multi-level load prediction model based on the time feature data. The present invention comprehensively covers the load features at different time levels such as hours, days, and weeks through multi-scale time feature extraction, providing richer input information for prediction.