69results about How to "Guaranteed interpretability" patented technology

The invention provides a penetration test agent system driven by a large language model and a test method, and the system comprises a memory unit which is composed of a penetration test knowledge base, comprises data elements including vulnerability information, a penetration mode and tool information, and employs a knowledge graph to establish a vulnerability knowledge representation model; the intelligent unit is realized by a large model and is used for generating a tool calling instruction, a penetration test load and a penetration test strategy based on the memory unit; and the planning unit is used for splitting the penetration test task into a task planning state machine consisting of controllable sub-tasks so as to guide the large language model to execute a specific task. Through introduction of the large model, the knowledge base and the task planning unit, the automation and intelligence level of the penetration test task is improved; the integrity of knowledge retrieval is improved, and the interpretability and consistency of large model output are ensured; for a complex scene of a penetration test process, a highly customized proxy behavior is realized, and a penetration test task execution process is guaranteed.

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.

The invention discloses a large-scale customized quality prediction method based on a heterogeneous causal graph attention network. The quality prediction method is designed for the characteristics of modular production, customized assembly and the like of a large-scale customized production mode and the problems of multiple scales and data mixing in data. The method comprises the following steps: firstly, providing a PCAM causal discovery algorithm, constructing a causal skeleton by utilizing mechanism and data combination, and obtaining a reliable causal graph structure based on an additive noise model and by introducing a scoring mechanism; then designing a heterogeneous causal graph attention network HCGAT for a multi-scale problem based on a causal graph structure, mapping node features of different dimensions to the same dimension by using different weight parameter matrixes, and training through the graph attention network; and finally, aiming at the characteristics of large-scale customized production of customized products, a quality prediction framework suitable for large-scale customized production is provided, customized codes are designed through customized assembly information, and a filter and a judger are constructed based on the customized codes to classify and process data. And performing quality prediction on different types of product data by using HCGAT as a sub-model.

The invention belongs to the field of computer vision, and particularly relates to a histopathology full-slice image analysis system and method based on a double-branch cooperative training deep network model, and the method comprises the steps: obtaining a histopathology full-slice image data set with an image level label, and carrying out the segmentation preprocessing; block image coding based on a pathology basic large model; establishing a double-branch cooperative training deep network model based on the slice map; using a histopathology full-slice sample data set with an image level label to train a network model in an off-line manner; and predicting the histopathology full-slice image by using the trained network model, and outputting an analysis result of the full-slice image. According to the method, the correlation between the full-slice images is modeled by introducing the slice images to assist the deep network model, the defect that the correlation between the full slices is neglected in an existing histopathology full-slice image analysis method can be made up, and the histopathology full-slice analysis precision of the deep network model is improved.

The invention discloses a thermal power generating unit coordination control data and mechanism hybrid modeling method and device, computer equipment and a computer readable storage medium, and relates to the technical field of thermal energy engineering and power system modeling and control. Establishing a high-precision dynamic mechanism model of the working medium flow and heat transfer process of the boiler-steam turbine system; then, constructing an error correction model fusing the long-short-term memory network, the convolutional neural network and residual connection; and finally, fusing the mechanism model and the error correction model through parallel computing to form a dynamic hybrid model with physical interpretability and high precision. Through the hybrid modeling strategy, the prediction accuracy of key parameters such as the main steam pressure, the separator outlet steam enthalpy value and the unit load in the wide load range, especially under the dry-state operation working condition is remarkably improved, and a reliable model basis is provided for designing an advanced unit coordination control system.

The invention discloses a PPG blood pressure estimation method and system combining deep learning and clinical priori knowledge. The PPG blood pressure estimation method combining deep learning and clinical priori knowledge comprises the following steps: firstly, acquiring a pulse wave signal through a signal acquisition module; preprocessing the pulse wave signal, and respectively outputting a first branch pulse wave signal and a second branch pulse wave signal; performing dimension conversion on the first branch pulse wave signal, extracting multi-dimensional features through a deep learning network, and extracting clinical priori knowledge features in the second branch pulse wave signal; and finally, constructing a blood pressure estimation model of a corresponding time period according to the multi-dimensional features and the clinical priori knowledge features, and predicting and outputting a blood pressure value of the corresponding time period through the blood pressure estimation model, so that the correlation between the extracted PPG features and BP and the blood pressure measurement precision are improved by combining deep learning and double-branch feature extraction of clinical priori knowledge. The effectiveness and interpretability of the blood pressure estimation method are ensured, and the credibility of the estimated blood pressure value is improved.

The invention discloses an attention neural network-based personnel ability automatic evaluation method and system. The method comprises the following steps of: dividing an ability evaluation data set # imgabs0 # into a training set # imgabs1 #, a verification set # imgabs2 # and a test set # imgabs3 #; setting a plurality of different neural network configurations to form a configuration set # imgabs4 #; the obtained hyper-parameter # imgabs5 # is selected, and a neural network model # imgabs6 # is initialized; using a training set # imgabs7 # to train a neural network model # imgabs8 # obtained through initialization; after a # imgabs9 # round is carried out, the neural network model trained in the # imgabs10 # round is recorded as # imgabs11 #, a verification set # imgabs12 # is used for evaluating the trained model, and the neural network model and the evaluation result after each round of training are stored; the model obtaining the highest evaluation index on the verification set in N rounds is selected to be tested and stored on a test set # imgabs13 #; and until all hyper-parameters are used for initializing the neural network models and tested by a test set # imgabs14 #, deploying the neural network model with the highest model evaluation result for personnel ability grade evaluation work. And the requirement on multi-dimensional personnel ability portraits in a personnel evaluation scene is met.

The invention discloses a multi-modal contrast learning fault diagnosis method for a small sample scene, and the method comprises the following steps: carrying out the enhancement of a one-dimensional signal and two-dimensional image fused fault data set through physical simulation for the small sample scene with scarce industrial fault data, and constructing a positive and negative sample pair through a plurality of data enhancement strategies; based on heterogeneous multi-modal fault data, designing a double-flow encoder architecture of a time sequence branch and an image branch, extracting depth features and mapping the depth features to a unified feature space through a projection head; performing supervised contrast learning pre-training based on intra-modal and inter-modal dual contrast loss; supervision fine tuning is carried out based on multiple loss functions such as physical guidance, so that accurate diagnosis of equipment faults is realized in a small sample scene. According to the fault diagnosis method under the unbalanced sample and limited labeling conditions, the problem that a traditional data driving model depends on large-scale labeling samples is effectively relieved through supervised comparative learning and cross-modal information alignment.

The invention provides a clothing pattern intelligent generation method and system based on multi-modal visual identification, and relates to the technical field of visual identification. The method comprises the following steps: firstly, acquiring clothing image data and text description data, performing unified preprocessing, obtaining a style category label, an appearance attribute label and a local contour key point set through multi-modal visual identification, generating structure semantic elements and geometric anchor points by utilizing a semantic-to-structure mapping knowledge base, and constructing a plate type structure diagram; further analyzing the structural constraint and executing constraint solution to generate a candidate pattern; and in combination with manufacturability verification and constraint correction, closed-loop optimization is formed, a target plate type file meeting the structure and manufacturing requirements is output, and effective connection from semantic recognition to plate type structure generation is realized.

The application provides a spoken-to-written conversion method and device based on reinforcement learning, equipment and medium, wherein the method comprises: obtaining a spoken text; inputting the spoken text into a conversion model to obtain a written text output by the conversion model; the conversion model is obtained by reinforcement learning, taking the editing operation of each word in the sample spoken text as an action, and taking the semantic consistency between the sample written text obtained by performing the editing operation and the sample spoken text and / or the written degree of the sample written text as a reward. The method, device, equipment and medium provided by the application break the limitation of insufficient labeled data in the process of reinforcement learning, and the semantic consistency and written degree give high-level and interpretable rewards. The conversion model obtained by application of the conversion model ensures the reliability and interpretability of the conversion from spoken text to written text.

The present invention relates to the technical field of code similarity detection, and in particular to a code similarity detection method, device and medium based on graph neural network, the method comprising the following steps: calling a control flow graph conversion module, converting a code to be detected into a control flow graph, encoding the control flow graph for topological structure, and obtaining a node feature vector; constructing a graph edit distance calculation model; using a model loss function to train the graph edit distance calculation model, calling the trained graph edit distance calculation model to simultaneously predict the graph edit distance and node matching confidence matrix between two control flow graphs, wherein the graph edit distance is used as a quantitative indicator of code similarity, and the node matching confidence matrix can generate a graph edit path. The present invention uses a control flow graph to capture the logic and semantics of the code, and uses a deep learning model based on a graph neural network to analyze the control flow graph, thereby achieving the purpose of detecting code similarity.

A dynamic statistics and result analysis system for watershed river and lake shoreline development and utilization comprises a remote sensing data acquisition module, a data preprocessing and classification module, a database construction module, a change trend analysis module, a river and lake directory and association retrieval module and an interaction management module. The data preprocessing and classification module is used for data preprocessing and statistical classification, the database construction unit is used for constructing a system database, the variation trend analysis module is used for variation trend analysis and statistical report generation, and the river and lake directory and association retrieval module is used for directory maintenance and result association. According to the dynamic statistics and result analysis system for watershed river and lake shoreline development and utilization, a remote sensing data statistical classification algorithm based on a neural network is proposed to classify remote sensing data; and a river-lake shoreline development and utilization change trend analysis algorithm based on deep learning is proposed to analyze the change trend.

The invention discloses a power transmission line forest fire detection method based on a KAN-YOLO network, and relates to the technical field of intelligent power grid operation and maintenance. The detection method comprises the following steps: constructing a power transmission line forest fire detection model based on a KAN-YOLO network, and inputting a power transmission line image into the model to obtain a forest fire event reasoning result; the model construction method comprises the following steps: acquiring forest fire accident video data, segmenting and classifying the data to obtain an image data set for network training, and dividing the image data set into a training set, a verification set and a test set; preprocessing the data set; constructing a power transmission line forest fire detection model based on a KAN-YOLO network; training the model by using the training set, and checking a training result by using the verification set to obtain an intermediate training model; and performing performance evaluation on the intermediate training model by using the test set, and selecting the model with the optimal performance. Compared with a traditional YOLO model, the model provided by the invention greatly improves the recognition precision of forest fire detection, reduces the false detection rate, improves the efficiency of power transmission line maintenance, operation and maintenance, and is high in practicability.

The invention relates to the technical field of robot performance evaluation, in particular to a robot acceleration performance evaluation method and device, and the method comprises the steps: generating an acceleration mapping matrix of a robot according to a pre-constructed robot dynamic model; performing singular value decomposition on the acceleration mapping matrix to obtain at least one singular value representing the geometric features of the dynamic operable ellipsoid of the robot; and calculating at least one composite acceleration performance index of the robot according to the at least one singular value, and generating an acceleration performance evaluation result of the robot according to the at least one composite acceleration performance index. Therefore, the problem that the overall dynamic performance of the robot under the actual complex working condition cannot be accurately represented due to the fact that the evaluation result is easy to have one-sidedness due to the fact that a single acceleration related index is usually adopted as the evaluation basis in the related technology is solved.

The invention discloses a gastrointestinal symptom identification method and a sensing system, and relates to the technical field of health informatics and medical information processing, and the method comprises the steps: collecting excitation response and abdominal sound to calculate an acoustic transfer function, and employing a main formant frequency to invert a quality loading index representing a condensation water film effect; constructing an equalization gain based on the transfer function to carry out spectral shape correction on the abdominal sound time-frequency spectrum; constructing a symptom evidence vector and performing adaptive weighted scaling on the symptom evidence vector by using a quality loading index; and finally calculating the distance between the weighted vector and the symptom mode prototype library and mapping the distance into a prompt probability. According to the method, the problems of acoustic link drift and signal distortion caused by microenvironment humidity change in long-time monitoring are solved, and the accuracy and robustness of symptom identification are improved.

The application discloses a carbon footprint data quality detection method and device and a storage medium, and belongs to the technical field of data quality detection. The method comprises the following steps: acquiring to-be-detected data, searching for at least one field knowledge information associated with the to-be-detected data in a knowledge database, calling a data quality discrimination model, identifying data characteristic information of the to-be-detected data, determining data quality information of the to-be-detected data according to the data characteristic information, combining the to-be-detected data, the field knowledge information and the data quality information, generating enhanced prompt information, inputting the enhanced prompt information into a large language model, and obtaining a quality detection result of the to-be-detected data generated by the large language model based on the enhanced prompt information. According to the cooperative detection mechanism of the fusion of multi-source information and the artificial intelligence discrimination model, the accuracy and reliability of the multi-modal and heterogeneous carbon footprint data quality control are significantly improved.

The application provides an optical element surface simulation method based on a physical information neural network, relates to the technical field of element design and simulation, and comprises the following steps: constructing a feedforward neural network; generating a deformation control equation model library; constructing a three-dimensional simulation database; forming a machine learning parameter library; training a PINN deformation simulation model in combination with the deformation control equation model library, the three-dimensional simulation database and the machine learning parameter library; screening out a target PINN deformation simulation model meeting preset conditions; inputting design parameters, support structure parameters and given external load parameters of different combinations into the target PINN model to obtain key performance prediction results of the optical element; the application can not only quickly evaluate the key performance of the element under various parameter combinations, shorten the design iteration cycle and reduce the trial and error cost, but also provides direct data support for support structure optimization design, and effectively solves the bottleneck problem of the traditional method in restricting the research and development efficiency of a high-precision optical system.

The application relates to an intelligent question and answer method and system based on a domain knowledge graph, and belongs to the technical field of computer natural language processing. The method first constructs a basic database, inputs a user question, uses a rule-based template matching method to process the question, and then processes the question. The question and answer system constructs a corresponding Cypher query statement based on question and answer information to query the answer in the database, and further processes the answer, and returns the answer according to the question type. The system comprises a template module and a deep module. The application provides good support for training of a knowledge graph-based question and answer model, saves sub-task time, improves the correctness of the sub-task, eliminates the influence of professional field knowledge on relationship recognition, strengthens the bidirectional communication and modeling between the knowledge base and the question, improves the accuracy and response speed of the system, reduces the maintenance and updating cost, guarantees the accuracy of the question and answer model, and guarantees the interpretability of the model.

The invention relates to the technical field of engine state monitoring, and provides an engine oil water pollution prediction method. The method comprises the steps that a condensation sub-model is obtained, and the condensation sub-model is established according to the engine blow-by mass flow and the blow-by temperature history and used for calculating the mixing amount of liquid water entering engine oil; an evaporation sub-model is obtained, and the evaporation sub-model is established according to the crankcase ventilation flow and the engine oil temperature and used for calculating the discharge amount of water vapor from the engine oil; and predicting the moisture content in the engine oil according to the condensation sub-model and the evaporation sub-model to obtain a predicted value of the moisture content. The problem that cost and precision are difficult to consider in the prior art is solved, and a low-cost and high-precision engine oil water pollution prediction scheme is realized.

The application relates to a Transformer-based power multi-modal full-element sample fusion labeling method and system, which comprises the following steps: S1: obtaining multi-modal original data of a production operation site, and obtaining a preprocessed multi-modal data set through preprocessing; S2: performing multi-modal feature coding and Transformer representation modeling according to the preprocessed multi-modal data, and obtaining a fusion feature representation through cross-modal feature fusion and information completion; S3: acquiring a multi-modal semantic alignment label set through semantic consistency detection and a semantic mapping mechanism according to the fusion feature representation and a corresponding candidate label set; S4: constructing a label system structure tree according to the multi-modal semantic alignment label set; and S5: constructing an automatic labeling module to generate preliminary labeling based on the fusion feature representation and the label system structure tree, and directly inferring labels based on similar samples and category probabilities. The application realizes efficient automatic label generation and sample dynamic classification.

The application provides a pre-stack slice energy adjustment method, device, equipment and storage medium, and belongs to the technical field of seismic data processing. The method comprises the following steps: forming a first data body before energy adjustment based on the seismic data migration of multiple work areas; obtaining a second data body after gain adjustment is performed on the first data body; extracting sample points in the first data body to obtain first samples, and extracting sample points in the second data body to obtain second samples; extracting envelopes of the first samples and the second samples; for each seismic trace, taking the ratio of the corresponding values of the first envelope surface and the second envelope surface of the seismic trace as the initial energy adjustment coefficient of the seismic trace, and taking the initial energy adjustment coefficients of all seismic traces to form an initial energy adjustment coefficient body; and sequentially performing time and space direction smoothing on the initial energy adjustment coefficient body to form a target energy adjustment coefficient body. The application realizes spatial direction energy unification, eliminates the structural false image of the splicing area, and maintains the time direction energy characteristics.

This invention belongs to the field of intelligent perception and simulation training technology for maritime navigation. It discloses a method and system for realistic training and evaluation of ship collision avoidance rules at sea. By constructing a hierarchical question bank, it generates scenario assembly lists and uniformly schedules ship motion simulation, 3D visual scenes, radar / AIS fusion display, and voice communication interaction. During training, it collects control inputs, motion states, target information, and communication records to form a traceable chain of evidence, and aligns sampling based on evaluation anchor points. An automatic scoring model is established based on result and process indicators, outputting an evaluation report with sub-scores, evidence location, a list of violations, and retraining suggestions. It is suitable for teaching, retraining, and assessment, improving the efficiency and achievement rate of collision avoidance decision-making training.

The present invention belongs to the technical field of neural network optimization, and discloses a two-stage training deep convex neural network optimization method and system, which adopts a deep convex neural network to approximate MPC. Compared with other neural networks, its input has a certain convexity relative to the fitting function, thereby ensuring the rapidity and interpretability in the solution process. The present invention proposes two-stage training and F_shap auxiliary data screening process to carry out neural network training. In the offline stage, the system proposes an expanded data set collection method and a swimming data set enhancement method, which can effectively collect data with different probabilities in different control tasks. Adding a safety filtering component during the use of the online stage can greatly ensure the output security of the trained neural network, and at the same time, collect data in the online stage and fine-tune the neural network to make it have better self-evolution.

The application discloses a generator set vibration trend early warning method based on a deep timing model, comprising the following steps: collecting generator set operation data and performing data preprocessing; constructing an enhanced Neural RDE model; running a conditional rough driving construction unit to generate a logarithmic signature feature set; running a frequency band gate vector field unit to generate a gate coefficient set and a hidden state sequence; running a physical consistency constraint unit to calculate and generate a constraint signal sequence; based on the hidden state sequence, the logarithmic signature feature set and the constraint signal sequence, calculating and generating a trend index set and a risk index set; calculating a joint loss function and updating model parameters; inputting a real-time collected continuous data stream and combining the model parameters and a threshold set to generate a vibration trend early warning result and an early warning lead time. The application significantly improves the vibration trend prediction accuracy, prolongs the early warning lead time and enhances the operation reliability under complex working conditions.

The invention relates to a reinforcement learning-based variable configuration aircraft control law intelligent optimization method, and belongs to the technical field of aircraft attitude control. The method specifically comprises the steps that an aerodynamic parameter neural network model is established, model input, model output and a loss function are defined, and real-time aerodynamic parameters are obtained according to in-flight state quantity identification; establishing a basic control law, and designing according to a pole assignment method to obtain standard control parameters; and a control parameter increment model is established, a control parameter increment is calculated according to the real-time aerodynamic parameters and the flight state obtained through identification, and the standard control parameters, the control parameter increment and the basic control law jointly form an optimized variable configuration aircraft control law. The method has the advantages of global intelligence, strong adaptability and explainable control law main body, and can be used for improving the adaptability of attitude control of the aircraft in an extremely wide range.

The present application relates to dilution refrigerator temperature calibration technical field, and disclose a kind of dilution refrigerator thermometer in situ calibration system based on natural cooling dynamics, comprising: according to the dynamics equation of cooling power and temperature Curvature is calculated, according to the temperature corresponding to thermometer resistance value, the temperature is substituted into temperature curvature, and the calibration function of resistance value and temperature is obtained;Identify the multivalued region of temperature, and evolve multiple branch functions in multivalued region in calibration function, calculate the deviation measure of each branch function and calibration function;In multivalued region, the support degree of each branch function to calibration function is calculated, and real-time data is obtained by adjusting calculation load, to minimize the number of experiments to verify calibration function;According to the branch verification result of multivalued region, dynamically adjust the verification strategy of non-multivalued region, under the premise of not relaxing calibration accuracy, through dynamics branch and load guide, significantly reduce the repeated experiment cost of dilution refrigerator cryogenic calibration.

The invention discloses an HITL method and system for unloading a large model reasoning task in a heterogeneous GPU computing power cloud. The method comprises the following steps: constructing a system model of the heterogeneous GPU computing power cloud; an HITL system architecture based on multiple agents is designed; human expert feedback is introduced through an HITL method, and recognition of a large model on a user intention is corrected; human expert feedback is introduced through an HITL method, and recognition of a large model on a user intention is corrected; the multi-agent cooperation module maps the intention of a user to a utility function through interaction among a plurality of agents; an optimized cue word method is used, the thinking process of a large model is decomposed, and the intention recognition accuracy is improved in a chain reasoning mode. According to the method, dynamic scheduling of computing power resource management and large model reasoning tasks can be optimized, and the service quality and the user experience in the heterogeneous GPU computing power cloud environment are improved.

The application discloses a kind of fusion driving and cascading error attribution hydroelectric generating set multi-subsystem parameter identification method, comprising: construct the high-fidelity simulation model of four cascading subsystems of local control unit LCU, speed regulator, electro-hydraulic servo system, hydroelectric unit;Design dynamic fusion coefficient with iteration attenuation, measured signal and model simulation signal are fused as the input of each subsystem, realize the transition from data-driven to model consistency-driven;Each subsystem output error is decomposed into upstream propagation error and its intrinsic error, and the optimization objective function with intrinsic error as the core is reconstructed;Improved particle swarm optimization algorithm, based on intrinsic error dynamic distribution search weight and introduce propagation error suppression factor, realize multi-subsystem parameter collaborative optimization.The application can effectively suppress cascading error accumulation, improve the physical interpretability and identification accuracy of parameter, and is suitable for hydroelectric generating set digital twin and online control scene.

This invention relates to the field of natural language-driven clustering technology, specifically to an e-commerce customer classification method based on end-to-end goal-driven clustering, comprising: S1: acquiring business goals and raw consumption behavior data of all e-commerce customers; S2: standardizing the raw consumption behavior data of all e-commerce customers to generate an e-commerce customer feature matrix; S3: inputting the business goals and the e-commerce customer feature matrix into a large language model, and generating importance scores corresponding to each consumption behavior feature through the independent query mechanism of the large language model for each consumption behavior feature; S4: performing clustering based on proximity and volume clustering algorithms, combined with the importance scores of each consumption behavior feature, to generate cluster labels for each e-commerce customer; S5: distinguishing the importance of e-commerce customers based on their cluster labels and formulating corresponding service strategies. This invention improves the objectivity and goal orientation of e-commerce customer classification.

The invention discloses a highly compact human body long-time motion prediction method based on a full MLP architecture, and the method comprises the steps: obtaining original time sequence human body skeleton data, carrying out the preprocessing, converting the original time sequence human body skeleton data from a time domain to a frequency domain, and then mapping the original time sequence human body skeleton data to a high-dimensional space, so as to form a high-dimensional space feature; establishing a full MLP architecture model, wherein the full MLP architecture model comprises N dynamic spatio-temporal feature sensing modules which are connected in sequence; inputting the high-dimensional spatial features into a full MLP architecture model to obtain first extraction features; the first extraction feature is recovered from a high-dimensional space to a three-dimensional space through layer normalization and a first linear layer in sequence; converting the recovered first extraction feature from a frequency domain to a time domain to form a second extraction feature; and inputting the second extraction feature into a multi-period joint prediction module to obtain a final prediction result. And the parameter quantity of the model is greatly reduced through lightweight design, so that the calculation complexity and resource consumption are reduced, the prediction accuracy and reliability are improved, and the universality is good.