24results about How to "Good prediction accuracy" patented technology

The invention discloses an automobile part intelligent production control and quality early warning system based on an AI model. Comprising a data acquisition module, a production process monitoring module, an AI analysis engine module, an adaptive control module, a quality early warning module, an equipment collaborative scheduling module, a data governance and model training module, a user permission and audit module and a remote operation and maintenance and knowledge precipitation module. According to the system and the method, an artificial intelligence algorithm and an industrial automation control technology are fused, and deep perception, dynamic regulation and control and defect pre-recognition of a production process are realized. The prediction precision of the method is superior to that of a traditional statistical process control method in a plurality of actual production scenes.

The invention discloses a CBN grinding wheel non-contact type grinding high-temperature alloy taper hole abrasion prediction method, and relates to the technical field of state monitoring and intelligent prediction in the precise grinding machining process. The method mainly comprises the following steps: calibrating a dynamic contact arc acoustic propagation model by using a system identification method to obtain dynamic geometric parameters, and constructing acoustic-vibration geometric coupling characteristics in combination with multi-mode signals in a grinding process; correcting the acoustic-vibration geometric coupling features by using a maximum mean difference algorithm and an adaptive window algorithm, training a wear prediction model by using the corrected acoustic-vibration geometric coupling features, and performing regression analysis on the corrected acoustic-vibration geometric coupling features by using the trained wear prediction model to obtain a prediction result. According to the grinding wheel abrasion prediction method for high-temperature alloy taper hole grinding, the adaptability, robustness and sensitivity of abrasion state characterization under complex variable working conditions can be improved.

This invention relates to the field of medical data-driven dose prediction technology, and discloses a method, system, device, and storage medium for individualized dose prediction of hypothyroidism based on medical data. The method includes: constructing a standardized feature vector based on the child's weight, age in days, corrected age in months, current L-T4 dose, TSH value, FT4 value, previous TSH value, TSH rate of change, feeding method, month of consultation, etiology of hypothyroidism, comorbidity status, previous dose adjustment magnitude, and age at which TSH first reached target levels; extracting TSH dynamic trajectory features from the child's TSH time-series data from previous follow-ups; obtaining a basic recommended dose using a gradient boosting decision tree model constructed with counterfactual filtering training data; and correcting the basic recommended dose to obtain an individualized recommended dose. This method improves the prediction accuracy of the gradient boosting decision tree model and allows the individualized recommended dose to simultaneously take into account multiple clinical confounding factors.

An alloy performance prediction method fusing vision-language-process multi-modal data comprises the steps that a multi-modal data set of an alloy material is constructed, and the multi-modal data set comprises three kinds of modal input including process parameters such as temperature and time of solid solution and aging treatment, SEM image visual information and SEM image description text information and corresponding alloy mechanical property true values; training a visual encoder ResNet50 model through comparative learning and training a language encoder BERT model through mask language modeling to obtain an SEM image and vector codes of language description of the SEM image; splicing and fusing process parameters such as temperature and time of solid solution and aging treatment and the vector codes obtained in the second step, and training a random forest regression device according to corresponding alloy mechanical properties; and the random forest regression device obtained through training is used for alloy performance prediction. According to the method, the structured process data, the unstructured SEM image data and the derived text description data are subjected to collaborative fusion and joint modeling for the first time, complementarity among different modal data is fully utilized, and more comprehensive and more three-dimensional digital representation of the alloy state is constructed; the multi-modal fusion framework and the feature learning mechanism are suitable for wide material systems. Meanwhile, the adopted'depth representation + random forest 'hybrid model has the advantage of high training efficiency while ensuring high prediction precision.

The application discloses a kind of associated multi-intersection traffic flow prediction methods.The steps of the present application are as follows:1, collect the traffic flow data of associated intersections, and divide the intersection traffic data into training set, validation set and test set after preprocessing;2, use CNN to extract spatial features from the input intersection traffic data;3, use Transformer to extract time features by inputting spatial features;4, after the Decoder layer is fully executed, the three time window data are finally input into three vectors, and the three vectors are stacked and input into the average pooling layer;5, set the model parameters;6, train the model until the maximum training period, and use the final model to predict the traffic flow of associated multi-intersections.The application uses CNN and Transformer to extract the spatial and temporal features of associated multi-intersections.Learning time encoding is used to embed the position encoding of Transformer, which injects position information and time information into the model together, helping the model to better learn the time features of traffic volume.

The invention relates to the technical field of artificial intelligence, big data mining and complex time sequence analysis, in particular to an AI quantization time sequence feature adaptive extraction algorithm system, which comprises a space-time multi-dimensional adaptive purification module used for optimizing a data signal-to-noise ratio through a volatility ratio and a time period mask; the microstructure sensing feature construction module is used for constructing a relative feature space and extracting fractal and entropy features; the dynamic asymmetric volatility label generation module is used for generating a balanced sample label based on the average real wave amplitude; the double-path parallel integration representation learning module is used for performing prediction through Bagging and Boosting path fusion; and the closed-loop feedback optimization module is used for feeding back and optimizing system parameters according to the simulation objective function. The method comprises corresponding steps. According to the method, the problems that the features lack state perception, sample construction destroys physical significance and the like are solved, and high-robustness and self-adaptive evolution time sequence feature extraction and trend identification are realized.

The present application relates to a ship heave motion prediction method and system based on a lightweight context-aware network, belonging to the field of ship and ocean engineering motion control technology, including the following steps: signal acquisition and preprocessing, heave motion multi-step prediction based on LCGNet network, model training and optimization, quantization and compression of the trained LCGNet model, and deployment on a shipborne edge computing device; during system operation, historical heave data is collected in real time and input into the model, and step S2 is executed in a loop to realize continuous online prediction of future heave motion; through innovative lightweight network structure design, high-precision, multi-step heave motion real-time prediction is realized at extremely low parameter and calculation cost to meet the engineering deployment requirements of the shipborne active heave compensation system.

This invention proposes a method and system for fault early warning and prediction of rotating equipment in thermal power plants based on neural networks. The method includes: acquiring multi-source heterogeneous operating data of rotating equipment in thermal power plants, including vibration signals, temperature signals, rotational speed signals, oil quality signals, and current signals; performing denoising and standardization processing on the operating data to construct a multi-dimensional feature vector containing time-domain, frequency-domain, and time-frequency-domain features; based on the multi-dimensional feature vector, using a convolutional neural network to extract spatial correlation features, modeling long-term dependencies through a bidirectional long short-term memory network, and dynamically strengthening the weight allocation of fault-sensitive features using an attention mechanism to generate a fault identification model; and using the fault identification model to analyze the preprocessed feature data in real time, outputting fault type, remaining life prediction results, and risk level classification, wherein the risk level classification is generated based on a joint decision of fault severity and remaining life.

The invention discloses a GPU (Graphic Processing Unit) chip temperature sensing thread bundle management method, relates to the technical field of GPU chip application, and solves the problems of low efficiency of a thermal management strategy and the like caused by the fact that a thermal prediction method in the prior art cannot cover hot spots of a whole chip, is high in modeling complexity and poor in real-time performance. The thermal prediction model is used for predicting the hot spot temperature of the chip in the running process of the graphics processor; the TAWS scheduling module is constructed based on a two-stage Warp scheduling strategy and is used for comparing the hot spot temperature with a preset maximum temperature threshold value and selectively executing a normal scheduling mode or a temperature control scheduling mode; the temperature of the chip is regulated and controlled; and when the duration of the high-temperature state exceeds the preset maximum allowable duration, the TAWS scheduling module triggers a scheduling freezing mechanism to realize safety control of the chip temperature. According to the method, cross-architecture high-precision generalization, a low-overhead instruction-level thermal sensing scheduling mechanism, refined analysis of instruction-level thermal characteristics and the like are realized.

The invention discloses a lithium ion battery health state estimation method based on physical information Transform. The method comprises the following steps: acquiring voltage and current time sequence data of a lithium ion battery under a pulse charging condition; feature parameters are extracted from the time series data, and a training data set containing the feature parameters and corresponding health state labels is constructed; and constructing a physical information neural network model based on a Transform architecture, inputting the training data set into the physical information neural network model for training, processing pulse charging data of the lithium ion battery to be tested by using the trained physical information neural network model, and outputting a battery health state estimation result. According to the method, a framework which is high in principle and can be popularized is established, and an important contribution is made for constructing a battery health prediction model which is higher in interpretability, better in data efficiency and higher in physical credibility.

The invention relates to the technical field of material informatics, in particular to an aluminum alloy mechanical property prediction method based on deep learning, which comprises the following steps: collecting an original data set containing alloy components, process parameters and performance actual measurement data, and cleaning, preprocessing and dividing the original data set; constructing a fusion neural network prediction model, wherein the fusion neural network prediction model integrates a Mama module to extract process sequence features, a KAN network module to extract a component nonlinear relation and a Transform encoder module to perform global feature fusion; training the model by using the training set, and performing hyper-parameter tuning and monitoring by using the verification set; evaluating model precision and engineering applicability through a test set and an external experimental sample; and finally, inputting alloy components to be predicted and process parameters into the trained model, and synchronously outputting predicted values of the yield strength, the tensile strength and the ductility. The method provides an efficient and accurate prediction tool for aluminum alloy design and optimization.

The invention provides a traffic network flow prediction method and system based on a pre-training space-time diagram neural network model. The method comprises the following steps: in a pre-training stage, training an encoder and a decoder by using a long-term space-time sequence, and freezing parameters of the trained encoder; in a fine tuning stage, a long-term space-time sequence is coded based on a frozen encoder, features of a short-term space-time sequence are extracted by using a space-time diagram neural network STGNN, then long-term feature representation and short-term feature representation are fused by using a multi-layer perceptron, and after a fusion result passes through a linear layer, a future traffic flow sequence is obtained. According to the invention, accurate short-time traffic flow prediction can be realized.

This invention belongs to the field of thermal energy storage control technology, specifically involving a thermal energy storage control method and control system based on flue gas parameter fluctuations. First, flue gas parameters from the boiler tail end are collected and preprocessed. Then, the parameters are decomposed and fluctuation characteristics are extracted using the EEMD algorithm. An LSTM-BP fusion model is used to predict the fluctuation range and trend. Next, a multi-objective optimization function is constructed based on relevant parameters, and the dynamic operating condition baseline value is obtained by solving it. Then, based on a fuzzy adaptive PID algorithm, the thermal energy storage device is adjusted by combining the deviation and fluctuation frequency. Finally, the actual fluctuations are monitored, and the model and optimization function are corrected to form a closed-loop control. This invention can accurately capture the fluctuation characteristics of flue gas parameters, achieve accurate prediction of fluctuation trends and dynamic adaptation to the operating condition baseline value, improve the accuracy and stability of thermal energy storage control, solve the pain points of existing methods such as adjustment lag and insufficient accuracy, and improve the efficiency of flue gas waste heat recovery.

The invention discloses a front fracturing well closing flowback numerical simulation method considering reservoir damage, and belongs to the technical field of petroleum development reservoir transformation. The method comprises the steps that initial parameters of a target oil reservoir are obtained, fracturing fracture propagation simulation is conducted, and an embedded discrete fracture grid is constructed based on the fracture form obtained through simulation; determining a function relationship between the reservoir damage coefficient and the water saturation according to an experimental result; establishing a reservoir parameter dynamic calculation model; constructing a mass conservation equation and carrying out differential discretization to obtain a numerical discretization solving model; and performing iterative solution on the model, dynamically updating reservoir physical property parameters at each time step, and finally obtaining a pressure field, a saturation field and crude oil yield prediction in the whole fracturing process. The problem that reservoir damage caused by fracturing fluid invasion cannot be dynamically reflected through traditional numerical simulation is solved, accurate simulation of the full-period process of preposed fracturing, well closing and flowback is achieved, and a reliable tool is provided for fracturing optimization and productivity prediction.

This invention discloses a comprehensive energy load forecasting method based on a multi-scale graph conditional state-space model. The method includes constructing a comprehensive energy load forecasting dataset for a park, performing data preprocessing, and then using Pearson correlation analysis to select highly correlated features; dividing the dataset into training, validation, and test sets, and standardizing the data; constructing a joint prediction model based on dynamic graph learning, a multi-scale graph conditional state-space model, and a three-dimensional attention mechanism; training the joint prediction model using the training set; adjusting hyperparameters and selecting the optimal model using the validation set; inputting the test set into the trained model, and outputting the predicted electricity, cooling, and heating loads; restoring the actual predicted values ​​through inverse normalization; and evaluating the model performance using multiple indicators. This invention ensures the real-time requirement of the forecast and is suitable for online application scenarios in park energy dispatching.

The application discloses a kind of composite material mixed connection performance modularization quick prediction method, comprising the following steps: S1. the circuit model of composite material connecting structure is constructed, and the circuit model is established based on the mechanical equivalent principle of spring stiffness method;S2. the displacement response signal and load information of connecting structure under the action of load are acquired by simulating composite material connecting structure using the circuit model;S3. based on the displacement response signal, whether the structure is damaged and damage type is judged in combination with corresponding damage criterion;S4. the circuit model is adjusted according to the judgment result;S5. based on the circuit model after adjustment, the complete load-displacement curve of the composite material connecting structure is output in combination with the load information predicted.The application can efficiently predict the nonlinear mechanical response of bolt connection, glue joint connection and glue screw mixed connection structure before damage occurs, and accurately simulate the whole process behavior of stiffness degradation and load redistribution after damage occurs.

The application relates to a cutting tool wear prediction method based on an MDRSNet, which comprises the following steps: collecting a cutting force signal and a moment signal of a cutting tool in a milling process; inputting the cutting force signal and the moment signal into a preset MDRSNet tool wear prediction model to output a predicted wear amount of the cutting tool, wherein the MDRSNet tool wear prediction model is obtained by training based on a tool wear dataset, the tool wear dataset comprises cutting force signals and moment signals of X, Y and Z three axes of different cutting tools in the milling process and a wear amount of a tool flank surface of the cutting tool after each tool feed, and the MDRSNet tool wear prediction model is constructed by introducing a multi-branch structure into a deep residual shrinkage network (DRSN). The application enhances the expression ability and learning ability of the model by combining the residual shrinkage unit with the multi-branch structure, so that the model is light-weighted while the excellent prediction precision of the model is ensured.

The application discloses a soft-hard interbedded rock mechanical parameter prediction method and system based on a residual attention network, relates to the technical field of rock mechanical parameter prediction, and has the advantages that the traditional neural network is prone to gradient disappearance when processing a deep network, which influences the training effect; the existing method lacks an attention mechanism and cannot effectively identify and strengthen key features; and the modeling capability for interlayer interaction is insufficient; the application provides a soft-hard interbedded rock mechanical parameter prediction method based on a residual attention network, which comprises the following steps: obtaining structure parameters and target mechanical parameters of a soft-hard interbedded rock sample; converting the structure parameters into an enhanced feature vector; constructing a residual attention network model; inputting the enhanced feature vector into the residual attention network model; and outputting a mechanical parameter prediction result and reliability evaluation information.

The invention relates to the technical field of thermal protection health management of high-speed aircrafts, and particularly discloses a thermal environment rapid prediction method and system based on pre-training. The method comprises the following steps: establishing a high-fidelity engineering algorithm by using a modified Newton theory and a Fay-Riddle formula, and generating a source domain heat flow data set; constructing a deep convolutional neural network composed of a geometric feature extraction module, a working condition feature extraction module and an overall heat flow calculation module; a pre-training-fine tuning framework is adopted, a deep convolutional neural network is trained through multi-working-condition source domain data, a pre-training model is obtained, and freezing-fine tuning is carried out on a small amount of sensor measuring point data of a target working condition; further fusing local sensor data into the network to form a sensor data fusion network; and finally, rapid prediction of the thermal environment is completed within millisecond-level time. According to the method, the prediction precision is remarkably improved, and the application requirement under the condition of finite computing power is met.

The application relates to the technical field of tailing resource utilization, and discloses a tailing sand particle grading dynamic optimization processing method and system. The method comprises the following steps: collecting particle size data at a detection node of a conveying pipeline and dividing the particle size data into particle size groups; calculating dynamic weight coefficients according to data stability coefficients and flow influence coefficients and performing weighted fusion; predicting a grading deviation value through a long short-term memory neural network to trigger an optimization decision to obtain a target grading ratio; calculating accurate adjustment amounts of each bin to generate a batching control instruction; performing closed-loop adjustment and secondary correction to obtain a target grading product. The application solves the problems of tailing sand particle grading optimization response lag, insufficient control precision, difficulty in multi-target cooperation and lack of self-adaptive capability in the prior art, and improves the real-time performance and accuracy of grading control.

The invention discloses a marine route planning method considering uncertain factors, and belongs to the field of marine traffic safety management, and the method comprises the steps: S1, carrying out the preprocessing and feature fusion of multi-source data; s2, carrying out the uncertainty modeling of the marine route, and quantifying the probability distribution of uncertainty parameters; s3, predicting a candidate route trajectory based on PCA hierarchical attention; s4, in combination with the probability distribution of the uncertainty parameters and the predicted candidate route trajectory, calculating the operability super probability of each segment of the route, and generating a probability operability index; and S5, by adopting an approximate MinSumA algorithm, outputting an optimal route by taking the integration of the minimum total navigation time and the maximum probability operability index as targets. By the adoption of the sea route planning method considering the uncertainty factors, a data-modeling-prediction-evaluation-optimization closed loop is constructed, the problems that a traditional method ignores uncertainty, is high in optimization complexity and poor in real-time performance are solved, and safety-efficiency-real-time performance collaborative optimization is achieved.

The utility model discloses an automobile auxiliary part with scale belongs to automobile manufacturing technical field, including automobile windshield glass, rear -view mirror and engine bonnet, wherein windshield glass and rear -view mirror are double -layer glass interlayer structure, and its interlayer interlayer bottom part is printed respectively the metric scale ruler of adaptation driver visual range, engine bonnet is printed along the length direction with the metric scale ruler of driver visual path matching. The utility model provides accurate visual reference for the driver, can promote trajectory pre -judge accuracy, and simple structure, high reliability, is applicable to various vehicle types, can effectively reduce the driving difficulty and accident rate.

The application discloses a kind of high-precision prediction method of friction coefficient based on ceramic coating, it is related to ceramic coating friction coefficient prediction technical field, comprising the following steps: step one: obtaining the experimental data and literature auxiliary data of APS spraying ceramic coating;Step two: complete field alignment, training / test set division, missing value processing and training set enhancement;Step three: construct five-level physical feature chain, convert original variable into descriptor with tribological physical meaning;Step four: descriptor input machine learning model for training and prediction;Step five: robustness verification and explainability analysis are carried out, the method constructs five-level physical feature chain, and further maps original material parameter and working condition parameter into physical descriptor related to contact stiffness, pore weakening, bearing capacity and lubrication response, so that model input no longer stays in surface experimental variable, but can more fully characterize tribological mechanism.

The present application relates to the technical field of lithium ion battery, and provides a GRU-EC-GAPSO lithium ion battery SOH prediction method, which comprises the following steps: extracting a health factor from voltage and time data in a local SOC interval during charging and discharging of a lithium ion battery, inputting the GRU model to perform SOH prediction, calculating an error sequence of a predicted value and an actual value, training an EC model by using the error sequence, and obtaining a GRU-EC model; using a GAPSO algorithm to optimize parameters of the GRU-EC model, and obtaining a GRU-EC-GAPSO model; and applying the trained GRU-EC-GAPSO model to different types of lithium ion battery data to perform SOH prediction. The present application has high prediction accuracy and good robustness in lithium ion battery SOH prediction.