Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

169results about How to "Improve learning ability" patented technology

Early warning and cooling system for thermal runaway of battery

The invention relates to the technical field of batteries, and discloses a battery thermal runaway early warning and cooling system which comprises a multi-parameter monitoring module, a data processing and early warning model and a cooling system control module. Wherein the multi-parameter monitoring module is used for acquiring the internal resistance value of each battery monomer in real time in a battery charging process, calculating the internal resistance range, the internal resistance standard deviation and the variable coefficient according to the internal resistance value at a sampling time point, monitoring the working environment gas of the battery through a mass spectrometer, and acquiring the real-time mass spectrum monitoring data of the gas in real time; release characteristics, including hydrogen, hydrocarbon compounds and the like, of escaped gas are quantitatively analyzed on the basis of an equivalent characteristic spectrum method. According to the battery thermal runaway early warning and cooling system, various key parameters in the battery charging process are comprehensively monitored, early warning of the battery thermal runaway risk is achieved, timely intervention is conducted in combination with the cooling system, the thermal runaway risk is reduced, and vehicle and personal safety is guaranteed.
Owner:杨之星

Multi-source heterogeneous big data fusion and reasoning method and system based on knowledge graph

The invention relates to the technical field of big data analysis and artificial intelligence, in particular to a multi-source heterogeneous big data fusion and reasoning method and system based on a knowledge graph, and the method comprises the following steps: constructing mapping interfaces of various data sources, collecting and storing entities in the data sources, and constructing nodes in the knowledge graph; performing representation learning on a graph structure of the knowledge graph based on a graph neural network, extracting an embedded vector of a node, and extracting a fusion node through the embedded vector and a mapping interface; performing knowledge completion on the fusion nodes by using a neural network based on symbolic logic; generating candidate reasoning paths by adopting a depth-first search algorithm based on the knowledge graph; reasoning is carried out based on an attention mechanism and path similarity, candidate reasoning paths are screened and sorted to obtain an optimal reasoning path, a graph neural network is adopted to carry out representation learning of a knowledge graph, symbolic logic and the neural network are combined, the interpretability of the reasoning process is guaranteed, and the learning ability is high.
Owner:TIBET UNIVERSITY FOR NATIONALITIES

KASP primer intelligent typing evaluation method and system TAL-SRX based on multi-model fusion

ActiveCN119132403Aavoid defectsImprove functional propertiesEnsemble learningBiostatisticsRobustificationTyping
The invention relates to the technical field of artificial intelligence, and particularly provides a KASP primer intelligent typing evaluation method and system TAL-SRX based on multi-model fusion. The method comprises the following steps: firstly, optimizing the performance of a basic model and a meta model by using RS, PSO and BO algorithms, constructing a Stacking integrated learning framework to integrate the advantages of an XGBoost learner, and enhancing the stability of the model by adopting five-fold cross validation; then, combining ANN and LSTM to construct a hybrid neural network, capturing a nonlinear relationship and extracting complex features, and meanwhile, introducing a Transform algorithm to capture a global dependency relationship in a high-dimensional feature space; and finally, fusing the two machine learning algorithms through a soft voting integration strategy, and outputting a KASP mark typing effect score. According to the method, the KASP marker with an excellent typing effect can be accurately screened, relatively high robustness and good and reliable evaluation performance are achieved, and technical support is provided for work such as molecular marker-assisted breeding.
Owner:SANYA NATIONAL INSTITUTE OF SOUTHERN BREEDING CHINESE ACADEMY OF AGRICULTURAL SCIENCES +1

Enterprise multi-type data labeling method and system based on feature engineering

InactiveCN119128612AEfficient labelingImprove applicabilityData setCharacteristic space
The invention belongs to the technical field of data processing, and discloses an enterprise multi-type data labeling method and system based on feature engineering, and the method comprises the steps: collecting initial data of an enterprise internal information system, and dividing the initial data to obtain a text data set, a table data set and an image data set; feature vectors of different types of data are fused after features are extracted and screened, and a multi-dimensional feature space is constructed; constructing an annotation model according to the feature information of the feature space, constructing similar sample pairs as positive sample pairs and irrelevant sample pairs as negative sample pairs for unannotated new data, and comparing, learning and optimizing the similarity between the similar sample pairs; and carrying out label inference based on the feature vectors obtained by comparative learning, and generating a labeling result. According to the method, efficient and automatic labeling of various types of data can be achieved, the method is suitable for various heterogeneous data such as texts, tables and images, and high labeling precision and expansibility are achieved.
Owner:江西省通信产业服务有限公司

Pavement detection method based on artificial intelligence and road inspection robot

The invention discloses a road surface detection method based on artificial intelligence and a road inspection robot, and the method employs a self-designed RCDIT network model, proposes an implicit expression method and a multi-head cross attention mechanism in the network model, guarantees the expression integrity and continuity of image blocks, and improves the detection precision. Complex interaction of image blocks and position information can be effectively processed, and the integration capability and information relevance of the model to multi-modal data are enhanced. In addition, the introduced position feedforward network further improves the abstract ability of feature representation, and is helpful for more effectively processing complex features and improving the expressive force of the model. And the RCDIT network model is combined with the road detection robot, so that the crack defects of the road surface and the road can be accurately and quickly identified and segmented, and the reliability and the accuracy of detection are greatly improved. The road detection robot can work in a 24 / 7 all-weather mode and is not limited by time and environment, manpower resources are effectively saved, and continuous monitoring on the road surface can be achieved.
Owner:EAST CHINA JIAOTONG UNIVERSITY

Photovoltaic construction management system

The invention relates to the related field of photovoltaic construction, in particular to a photovoltaic construction management system, which comprehensively analyzes video data and environmental sensor data, improves the accuracy and reliability of the monitoring effect, and can effectively reduce the blind area and false alarm rate of a single data source through the fusion of multi-modal data. According to the invention, a genetic algorithm and a decision tree algorithm are integrated, so that the system has strong learning ability, and the system can be continuously self-optimized according to historical data and operation logic to gradually adapt to different construction environments and working conditions; through regular updating and model training, the system can continuously improve the monitoring precision and the safety management level, and it is ensured that an efficient and stable operation state is always kept in long-term use; an analysis result is displayed through a visual platform, accurate defect positions and severity degree marks are provided, a detailed detection report can be automatically generated, long-term monitoring and data comparison are supported, and powerful technical support is provided for construction quality guarantee.
Owner:JIANGSU SAILAFU ELECTRIC POWER DEV CO LTD

Source code vulnerability detection method and system based on adaptive graph neural network

The invention relates to the technical field of software vulnerability detection, in particular to a source code vulnerability detection method and system based on an adaptive graph neural network. Extracting a code attribute graph from the normalized code by using an analyzer; converting the code attribute graph into a learnable vector representation by using CodeBERT; calculating a centrality index of each node in the code attribute graph, and splicing the centrality index with a node feature vector; learning global features and local features of the graph in combination with an adaptive attention convolution module and a GCN module; and fusing features learned by the adaptive attention convolution module and the GCN module, inputting the fused features into a full connection layer, and outputting a final vulnerability detection result. According to the method, firstly, based on a centrality method, structural representation of CPG on a function is enhanced, and CPG features are enhanced; and meanwhile, a CPG-based adaptive attention graph neural network is designed and realized, so that the accuracy and efficiency of software vulnerability detection are improved while the CPG feature analysis capability is enhanced.
Owner:Chinese People's Liberation Army Cyberspace Force Information Engineering University

Method and system for predicting irreversible loss of centrifugal air compressor of fuel cell

The invention belongs to the technical field of fuel cells, and particularly relates to a fuel cell centrifugal air compressor irreversible loss prediction method and system, and the method comprises the steps: collecting performance data based on a fuel cell centrifugal air compressor comprehensive performance test platform, and building and verifying a multi-physics coupling CFD simulation model; defining an irreversible loss type and related structure parameters, constructing an integrated CFD simulation framework, generating a parameter combination table through an orthogonal test, and performing simulation calculation; an RL-GHEM method is provided, the weight and combination strategy of a base model are optimized, linear and nonlinear combination modes are dynamically switched, and irreversible loss prediction models under different working conditions are constructed. The hybrid integrated model based on reinforcement learning guidance is provided, the model combination strategy is dynamically optimized through the reward mechanism, the base model with the optimal performance is optimized, the prediction precision, stability and robustness are improved, precise prediction of the irreversible loss of the air compressor is achieved, and the problem that in the prior art, the prediction precision is insufficient under variable working conditions is solved.
Owner:DALIAN UNIV OF TECH

Text recognition method based on quantum transfer learning

The invention provides a text classification method based on quantum transfer learning, and belongs to the field of fusion of quantum computing and artificial intelligence. In consideration of complex feature space and nonlinear characteristics in text classification, quantum circuit design, a self-attention mechanism and a quantum transfer learning process are optimized, so that the learning efficiency and classification accuracy of the model are improved. The high-dimensional feature representation and a complex nonlinear decision process are converted into a quantum self-attention network structure, an optimal strategy of text classification is obtained by utilizing a complex quantum kernel self-attention network (CQKSAN) model, and a quantum superposition and entanglement mechanism is introduced to enhance the expression ability of the model. A simulation result shows that compared with a traditional deep learning algorithm and other mixed quantum-classical algorithms, the algorithm provided by the invention has higher classification precision and learning efficiency. The method not only shows the potential of a mixed quantum-classical method, but also provides a basis for further research of a future full-quantum learning system.
Owner:SHANGHAI CHUNKUI INFORMATION TECH CO LTD

Electric vehicle charging load prediction method

The invention discloses an electric vehicle charging load prediction method, which comprises the following steps: firstly, screening variables with high correlation through a maximum information coefficient, selecting features with high correlation as model input, and taking a charging load as output; reconstructing the data training set by adopting cross validation; a charging load integration model is constructed, the integration model is divided into a base learning device and a meta learning device, the base learning device comprises a DCformer model, a BiTCN model and an artificial neural network ANN model, and a lightweight gradient elevator LightGBM is used as the meta learning device; the parameters of the LightGBM model are optimized by adopting an improved star graffiti optimization algorithm INOA; an error sequence is obtained after a preliminary prediction value is obtained through model prediction, and a final prediction value is obtained by taking LightGBM as an error prediction model. According to the method, data features are effectively extracted by integrating a plurality of single models, the prediction accuracy is improved, and compared with a traditional model, the stability is better, and the generalization ability is higher.
Owner:HUAIYIN INSTITUTE OF TECHNOLOGY

Automobile battery fault prediction method and device

The invention provides an automobile battery fault prediction method and device. The method comprises the following steps: collecting time sequence voltage sampling data and battery state sampling data; performing characteristic decomposition and characteristic selection on the time sequence voltage sampling data to obtain strong characterization characteristic data; constructing a feature data set according to the strong representation feature data and the battery state sampling data; constructing an original fault detection classification model based on a long short-term memory network; training the original fault detection classification model according to the feature data set to obtain a trained fault diagnosis classification model; acquiring a real-time battery voltage value of the to-be-diagnosed vehicle; predicting whether the battery state has a fault according to the real-time battery voltage value and a fault diagnosis classification model; and if yes, outputting battery fault early warning information. According to the method and the device, the battery fault can be quickly and accurately predicted in real time based on the battery voltage data, the dependence on the sensor precision and the environmental change is reduced, and the reliability and the stability of fault detection are improved.
Owner:GAC AION NEW ENERGY AUTOMOBILE CO LTD

Multi-modal sentiment analysis method and system based on pre-training and online learning

The invention relates to the technical field of multi-modal sentiment analysis, in particular to a multi-modal sentiment analysis method and system based on pre-training and online learning. The method comprises the following steps: acquiring sentiment analysis multi-modal data, constructing a sentiment analysis model, namely constructing a multi-head attention mechanism to respectively encode text, image and audio modals, and splicing and fusing each encoded modal feature; mining an interaction relationship among all modes in the fusion features by using a feedforward neural network and a multi-head attention mechanism; evaluating the contribution degree of each mode in the joint features to the sentiment analysis task by constructing a full-connection neural network; according to the sentiment analysis method based on the multi-head attention mechanism, the multi-head attention mechanism is constructed to carry out coding processing on each mode, the coded mode features are spliced and fused, and therefore the sentiment analysis accuracy is remarkably improved. The multi-head attention mechanism based sentiment analysis method based on the multi-head attention mechanism has the advantages that the multi-head attention mechanism is constructed to carry out coding processing on each mode, and the coded mode features are spliced and fused.
Owner:YANTAI UNIV +1

Graph convolutional neural network and reinforcement learning fused highland cooperative control method

The invention provides a highland cooperative control method fusing a graph convolutional neural network and reinforcement learning, through controlling urban road traffic signal lamps, an expressway ramp signal machine and a variable speed limit unit, expressway traffic congestion is relieved, the urban traffic passing efficiency is improved, road driving vehicles are controlled by means of the signal lamps and the speed limit unit, and the traffic efficiency is improved. Traffic flow scheduling optimization in a global road network is achieved; the whole structure adopts a distributed cooperative control framework, is controlled by a decision-making agent, and performs autonomous decision-making and adjustment according to real-time traffic conditions, so that the adaptability and robustness of the system are enhanced, and the defects of large data calculation amount, difficulty in processing, control strategy lag and the like caused by traditional centralized control are effectively avoided; the graph convolutional neural network model is constructed to aggregate the feature information of the peripheral decision intelligent agent, and the control weight is dynamically allocated, so that the information sharing capability is enhanced, the suboptimal decision caused by lack of comprehensive perception of the surrounding environment is prevented, and the cooperation efficiency of the whole system is improved.
Owner:NANJING UNIV OF INFORMATION SCI & TECH

Sewage treatment fault diagnosis method and system based on data analysis

The invention discloses a sewage treatment fault diagnosis method and system based on data analysis, and the method comprises the steps: continuously optimizing a diagnosis strategy in the input and training process of a large amount of data through introducing a fault diagnosis model, and improving the accurate recognition capability of a fault under a complex working condition; therefore, the system not only has strong learning ability, but also can perform self-optimization through a reward mechanism, thereby reducing the occurrence rate of false report and missing report. An adaptive weight adjustment mechanism is introduced, so that the model can automatically adjust weight distribution and increase response to a new fault when the system has a novel fault or few faults; by dynamically adjusting the weight vector, the diagnosis model of the invention has strong self-learning ability and can continuously adapt to new fault modes and working condition changes, and the weight adjustment mechanism greatly improves the generalization ability and adaptability of the model, and especially has significant advantages in the face of fault scenes with strong uncertainty.
Owner:南京云涧环境科技有限公司

Online monitoring method and monitoring system for grounding defects of outer metal sheath of high-voltage cable

The invention discloses a high-voltage cable outer metal sheath grounding defect on-line monitoring method and monitoring system, which can work under the condition that a cable is not powered off, realizes on-line monitoring, and realizes on-line monitoring through the cooperation of a current sensor, a voltage sensor, a single-chip microcomputer and an upper computer. And the equipment can quickly detect when a fault occurs and is convenient for maintenance workers to timely eliminate obstacles, so that the risk of major electric power accidents is reduced. The algorithm is similar to an improved neural network algorithm and is used for diagnosing the defects of the outer metal sheath of the cable, the weight is obtained through training learning, and finally the defects of the outer metal sheath of the high-voltage cable are determined. The protection content covers each component of the technical scheme, and comprises a monitoring method, a device structure, a sensor technology, a data processing algorithm, a neural network learning algorithm and the like.
Owner:SHANGHAI GOALSEE ELECTRIC TECH CO LTD

Event-triggered single-link mechanical arm control method and system under cloud side-end cooperation

The invention discloses a single-connecting-rod mechanical arm control method and system triggered by events under cloud side end cooperation. The terminal collects operation data of the single-connecting-rod mechanical arm and sends the operation data to the edge end for preprocessing; the edge end inputs the data into the model-free controller or the trigger holder according to a preset event trigger rule to obtain a control instruction; the terminal controls the single-connecting-rod mechanical arm to move according to the control instruction, collects operation data and sends the operation data to the edge end; the cloud end selects to optimize and update parameters of the model-free controller according to the operation data fed back by the edge end and a preset event triggering rule; the control method and system are suitable for the single-connecting-rod mechanical arm, the modeling difficulty of the single-connecting-rod mechanical arm is reduced, the control precision is improved, the calculation efficiency is improved, system resources are saved, and the control method and system have the application prospect of being popularized to other single-input and single-output systems.
Owner:CHINA RAILWAY DESIGN GRP CO LTD

Satellite initial orbit determination method of embedded physical knowledge neural network

The invention relates to a method for determining an initial orbit of a satellite with an embedded physical knowledge neural network, which is used for determining an initial orbit of a short arc segment under an optical observation condition and providing support for subsequent precise orbit determination. The network comprises an encoder-decoder structure, an encoder predicts the motion state of a satellite in an orbit according to observation information, the decoder converts the predicted satellite motion state into an observation value, the prediction accuracy of the encoder is evaluated, physical information of satellite orbit dynamics is introduced into a loss function, and the prediction accuracy of the satellite orbit is improved. And satellite orbit determination based on the neural network is realized. The input of the encoder is the observation angle quantity (namely azimuth angle and pitch angle) of the satellite relative to an observer, the observation time and the position information (namely longitude, latitude and altitude) of the observer; the output of the encoder, namely the input of the decoder, is the absolute motion state quantity (namely the position and the speed of the satellite in a geocentric inertial coordinate system) of the satellite in the orbit and the position information of an observer, and the output of the decoder is the observed quantity of the satellite relative to the observer. When the network is trained, firstly, a decoder is trained by taking a mean square error between a predicted observation value and a real observation value as a loss function, and a transformation relation from a satellite motion state to the observation value is learned; fixing the structure and parameters of a decoder, training the whole encoder-decoder network, calculating the derivative of the predicted satellite absolute state quantity relative to the observation time, and calculating the physical information loss of a speed item and an acceleration item; adding the physical information item loss and the mean square error of the observed quantity, constructing a total loss function, and optimizing network parameters through error back propagation; in the reasoning stage, an encoder network is used for reasoning, and the absolute motion state quantity, predicted by an encoder, of the satellite is converted into an orbit element form, namely, the initial orbit is determined.
Owner:BEIHANG UNIV

State variable guided support vector machine elastoplastic constitutive relation simulation method

The invention discloses a state variable guided support vector machine elastoplastic constitutive relationship simulation method. The method comprises the following steps: 1, collecting elastoplastic constitutive model data sets of reinforced concrete columns of different failure types and different building materials; 2, dividing the hysteretic curve into a training set and a test set according to the time of the hysteretic curve; 3, mapping input variable displacement or strain to a state space by utilizing an operator for correcting the three-parameter hysteresis model to form a state variable capable of considering load path correlation; 4, based on an LS-SVM algorithm, establishing a nonlinear single-valued function relationship between the state variable and the force; 5, verifying the established state variable guided support vector machine elastoplastic constitutive relation simulation method, and comparing with the existing method; according to the method, a state function of a corrected three-parameter hysteresis model is utilized, a deformation input variable is mapped to a state space to form a state variable, and a single-valued function between force and the state variable can represent a complex multi-valued hysteresis relation between the force and deformation.
Owner:CHINA THREE GORGES UNIV

Infrared temperature image enhancement method and system based on image processing

The invention provides an infrared temperature image enhancement method and system based on image processing, and the method comprises the steps: obtaining an infrared temperature image, extracting data of different frequencies and different scales in the infrared temperature image through a multi-scale image processing technology, carrying out the normalization processing of the data, and constructing a neural network model; performing image enhancement on multiple scales, inputting the preprocessed infrared temperature images of multiple scales into a neural network model, designing a Q-learning algorithm, and learning and updating Q values of the infrared temperature images under different enhancement strategies; selecting and executing an optimal enhancement strategy to train the neural network model, and optimizing a loss function of the neural network model; the enhancement results on multiple scales are fused and then post-processed, and the visual effect of the image is optimized; and outputting the enhanced infrared temperature image. According to the method, the Q-learning algorithm is used for designing the reward function, the loss function generated by recognition is continuously optimized and enhanced, and the recognition precision of image processing is improved.
Owner:GUANGZHOU POWER SUPPLY BUREAU GUANGDONG POWER GRID CO LTD

Electrocardiogram abnormal degree detection method based on artificial intelligence

The invention discloses an electrocardiogram abnormal degree detection method based on artificial intelligence. The method comprises the following steps: removing low-frequency noise from original electrocardiogram data; constructing an electrocardiogram time sequence anomaly detection model according to the electrocardiogram signal after the low-frequency noise is removed; decomposing the electrocardiogram signal without the low-frequency noise into different low-frequency components and high-frequency components; performing threshold processing on the high-frequency component, and reconstructing an electrocardiogram signal; performing feature extraction on the electrocardiogram signal data subjected to noise interference removal to obtain an electrocardiogram feature sequence; inputting the electrocardiogram feature sequence into an electrocardiogram time sequence anomaly detection model for model parameter estimation, and outputting model parameters; and predicting a predicted value of the electrocardiogram feature sequence according to the estimated model parameters to obtain an electrocardiogram anomaly score. According to the method, after the electrocardiogram time sequence anomaly detection model is automatically constructed through the intelligent engine according to the electrocardiogram signal, the model parameters are adjusted, the accuracy and efficiency of electrocardiogram anomaly detection are improved, and the generalization ability is enhanced.
Owner:MEDEX (BEIJING) TECH LTD CORP

Two-stage encrypted traffic classification method for programmable switch

The invention discloses a two-stage encrypted traffic classification method for a programmable switch, which relates to the technical field of computer networks and comprises the following steps: a switch data plane extracts semantic features and statistical feature sequences of input traffic and preprocesses the semantic features and the statistical feature sequences; inputting the semantic features into a matching-action pipeline, matching the semantic features with a convolutional neural network model classification table issued by a server, and if the semantic features are matched with the convolutional neural network model classification table, obtaining a flow category based on the semantic features; if not, inputting the statistical feature sequence into a matching-action pipeline, matching the statistical feature sequence with a long-short-term memory model classification table to obtain a matching result of each statistical feature, and performing bitwise AND on the matching result to obtain a flow category based on the statistical features; executing a traffic decision action based on the traffic category; the encrypted traffic classification method aims at designing a network model for realizing encrypted traffic classification and deploying the network model to a switch, and a low-delay, low-load, high-precision and high-availability encrypted traffic classification scheme running in the switch is realized.
Owner:UNIV OF SCI & TECH OF CHINA

Power grid fault plan entity identification method, device, equipment, medium and product

The invention discloses a power grid fault plan entity recognition method and device, equipment, a medium and a product, and relates to the technical field of entity recognition, and the method comprises the steps: enabling a preset to-be-recognized entity type, a target power grid fault plan text set and a to-be-labeled power grid fault plan text to form a first prompt word, and inputting the first prompt word into a large language model, obtaining a first entity set and the confidence of each entity in the first entity set; forming a second cue word by a preset to-be-identified entity type, related knowledge of each entity in the target entity set and the target entity set, and inputting the second cue word into the large language model to obtain a second entity set; and determining the entities with the confidence coefficient greater than or equal to a set threshold value in the first entity set and the entities in the second entity set as final entity results. According to the method and the device, the problems that a traditional deep learning model is poor in performance under the condition of small samples and the generalization of extraction of different types of power grid plans is poor can be solved; and the accuracy and the reliability in power grid fault plan entity identification are improved.
Owner:NANJING UNIV OF AERONAUTICS & ASTRONAUTICS

Neural network weight side channel recovery method based on deep learning

The invention discloses a neural network weight side channel recovery method based on deep learning, relates to the technical field of artificial intelligence and security, and solves the problems that in the prior art, power consumption traces cannot be perfectly synchronized with a data processing process, and errors may be introduced in acquisition and processing of each power consumption trace. The method comprises the steps of collecting power consumption traces and constructing a data set, and then determining a leakage interval to perform dimension reduction on the power consumption traces in the data set; then selecting a classifier, constructing the capability from side channel information to a correct weight model, and finally recovering the weight of a corresponding neuron in the MLP according to the trained model; the weight of each neuron in the MLP is deduced through a small number of power consumption traces while the high accuracy is kept, and the side channel analysis efficiency is greatly improved.
Owner:XIDIAN UNIV

Multi-unmanned aerial vehicle trajectory optimization method for user mobile communication in heterogeneous network

The invention discloses a multi-unmanned aerial vehicle (UAV) trajectory optimization method for user mobile communication in a heterogeneous network, and aims to improve the security of a considered communication system and suppress the eavesdropping capability of an air mobile eavesdropper by transmitting an interference signal from a friend UAV. The invention provides an optimization problem of a plurality of unmanned aerial vehicles under three-dimensional flight path flight conditions, ground mobile user scheduling and limited energy consumption constraints, and aims to maximize the total fair secrecy rate of ground mobile users. According to the invention, a multi-agent 3D flight path and power control algorithm is designed by using DRL, and the algorithm has a centralized training distributed execution framework. And a heterogeneous MADDPG is adopted to carry out reasonable 3D trajectory design and power control on the communication unmanned aerial vehicle and the interference unmanned aerial vehicle, so that the communication unmanned aerial vehicle and the interference unmanned aerial vehicle are used as different types of intelligent agents to independently interact with the environment. According to the method, due to the diversity of action variables between the intelligent agents, different types of intelligent agents need to utilize unique information of the intelligent agents and depend on the speciality of other intelligent agents to collaborate to find an effective strategy. The total fair secrecy rate in a communication system is maximized by jointly optimizing the 3D trajectories and transmitting powers of the communication unmanned aerial vehicle and the interference unmanned aerial vehicle.
Owner:CHONGQING UNIV OF POSTS & TELECOMM

Blood glucose prediction method based on LSTM encoder-decoder architecture

A blood glucose prediction method based on an LSTM encoder-decoder architecture comprises the following steps: performing smoothing, noise reduction, window sliding and other processing on acquired blood glucose data of a diabetic patient; a blood glucose prediction model based on an LSTM encoder-decoder architecture is constructed, an encoder is responsible for converting an input sequence into a hidden state vector, and a decoder performs prediction output by using the hidden state vector; in order to enhance neuron memory and reduce accumulation of error propagation, in an encoder-decoder structure, an attention mechanism based on a learning similar mode is introduced, and a DTW-based chain address search algorithm is constructed to find a similar window sequence; the attention feature and the hidden state vector are jointly input into the decoder, and the output of the decoder is mapped to the vector conforming to the output dimension through the projection layer, so that a more accurate blood glucose prediction result is obtained. The method has strong learning ability, and can effectively improve the blood glucose concentration prediction precision through experimental authentication.
Owner:NANJING UNIV OF POSTS & TELECOMM +1

Classroom intelligent terminal integrated with micro host, network access and audio acquisition

The invention relates to the technical field of classroom intelligent terminals, in particular to a classroom intelligent terminal integrating a micro host, network access and audio acquisition. According to the technical scheme, the classroom intelligent terminal integrating a micro host, network access and audio acquisition is provided. Comprising a main control unit, an acoustic array device, a network interface device, a deep learning audio processing unit, a multi-mode interaction device, an adaptive network technology component, an edge computing unit, a virtual reality and augmented reality device, a block chain module and a voice synthesis and emotion recognition module. The system provides more advanced, intelligent and interactive rich education experience, is expected to improve the learning effect and teaching efficiency, has potential revolutionary influence in the education field, and provides a more flexible, personalized and efficient learning environment.
Owner:NANJING JINSHENG YURUN SCI & EDUCATION EQUIP CO LTD

Equipment fan control method based on reinforcement learning, equipment and medium

The invention discloses an equipment fan control method and equipment based on reinforcement learning and a medium, and relates to the technical field of fan control. The method comprises the steps that an action space is constructed according to operation data of a fan, a state space is constructed according to operation data of target equipment, and the operation data of the target equipment comprises working temperature; obtaining fan power, calculating rewards according to the fan power and the working temperature, and constructing a strategy network and a value network according to the state space; obtaining the current sensor reading and the rotating speed of the fan, and carrying out online training on the strategy network and the value network; after training is completed, sensor readings are collected, and fan actions with the maximum probability are executed. By means of the method, the total power consumption of the fan meeting the temperature constraint condition is minimum while it is guaranteed that the equipment temperature is controlled within the ideal range.
Owner:INSPUR NETWORK TECH (SHANDONG) CO LTD

Ultrasonic examination report auxiliary generation method, electronic device and storage medium

The present application relates to the field of ultrasound examination report generation, and specifically to an ultrasound examination report auxiliary generation method, electronic device and storage medium, the method comprising: collecting ultrasound examination reports, obtaining structural data and diagnostic data; determining the data universality of ultrasound examination reports, obtaining the information contribution of ultrasound examination reports; dividing ultrasound examination reports into examination report clusters, obtaining the data quality excellence of examination report clusters; determining a training data set based on the data quality excellence of all examination report clusters to train a neural network model, obtaining an ultrasound examination report auxiliary generation model, and obtaining an ultrasound examination report auxiliary generation result based on the ultrasound examination report auxiliary generation model. The present application solves the problem of information redundancy corresponding to ultrasound examination reports, which leads to unreasonable auxiliary generation diagnostic suggestions of ultrasound examination report auxiliary generation models.
Owner:HANGZHOU QUANXIAN MEDICAL TECH CO LTD

Fatigue detection network for extracting blood flow motion features based on face video

The invention provides a fatigue detection network for extracting blood flow motion features based on a face video. The network is composed of a face video preprocessing module, a feature extraction module, a feature constraint processing module and a classifier. The objective of the invention is to solve the problems of strong invasion of a biochemical index method, poor real-time performance of a physiological signal method, weak reliability of a behavior characterization method and the like in an existing fatigue detection method.
Owner:NANJING UNIV

Dose superposition method and device based on depth deformation registration, equipment and medium

The invention discloses a dose superposition method and device based on depth deformation registration, equipment and a medium, and the method comprises the steps: obtaining a multi-modal data set, and preprocessing image data; the multi-modal data set comprises a reference CT image, a floating CT image, a reference dose image corresponding to the reference CT image and a floating dose image corresponding to the floating CT image; constructing a floating rigid transformation CT image according to the preprocessed multi-modal data set, importing a computed tomography image sequence of the reference CT image and the floating CT image, and importing a three-dimensional dose image of the reference dose image and the floating dose image, the deep deformation registration model has strong feature extraction and learning capabilities, and can learn a complex deformation rule from a large amount of training data, thereby realizing accurate registration of complex deformation.
Owner:SHENYING MEDICAL TECH (SHENZHEN) CO LTD +1