9 results about "Differential equation models" patented technology

This invention proposes a method for predicting the health status of lithium batteries based on multidimensional features and neural network constant differential equations. The method includes the following steps: S1, preprocessing the capacity data and charging stage operation data collected during lithium battery operation, and constructing features from historical health status data; S2, constructing multidimensional feature inputs for health status prediction based on the charging stage operation data; S3, inputting the multidimensional features into a gated recurrent unit network to fuse and encode the historical health status sequence and constant current charging stage features to obtain a potential feature representation characterizing the battery degradation state; S4, comparing the predicted health status value output by the neural network constant differential equation model with the corresponding actual health status value, calculating the prediction error, and evaluating the prediction accuracy. This application achieves high-precision prediction of lithium battery health status by integrating a multidimensional feature screening mechanism and a continuous-time state evolution modeling method.

The application discloses a kind of mouse brain ischemic lesion automatic quantification method and system based on deep learning, belong to medical image processing and artificial intelligence technical field.The method includes: standardization pretreatment is carried out to T1WI, T2WI, DTI and ASL and other multi-modal MRI data, and enhanced feature tensor is constructed;Lesion intelligent preliminary segmentation is carried out using BA-U-Net network guided by physical prior, and biophysical consistency loss is introduced;Boundary refining is carried out to segmentation result using graph convolution network;Based on LDDMM registration and neural ordinary differential equation model, dynamic evolution modeling is carried out to multi-time point lesion, and severity score is calculated;Finally, comprehensive pathophysiological feature vector is constructed, XGBoost model is used for prediction, and interactive report containing 3D visual model is generated by SHAP explanation.The system includes corresponding functional module.The application realizes high-precision, reproducible automatic quantification to fuzzy lesion, improves the robustness of pathological pattern recognition, and provides dynamic biomarker for drug evaluation.

This invention discloses a power system node inertia time constant assessment strategy based on continuous evolution feature learning. The steps include: First, collecting frequency response data of the power system during active power disturbances, extracting statistical and time-frequency features from the frequency response data, constructing a feature vector, and obtaining coded features through a multilayer perceptron encoding module; Second, gating and weighting the coded features through a hybrid expert module to form a fused feature vector, which is then input as a state variable into a neural frequent differential equation model. In the state space adaptively adjusted by the hybrid expert module, a numerical step-by-step solution algorithm is used to continuously evolve the state variable, obtaining a dynamic feature vector; Finally, the dynamic feature vector is input into a decoding module based on a multilayer perceptron, outputting the power system node inertia time constant assessment result. This invention achieves accurate assessment of the power system node inertia time constant through encoding, continuous evolution, and decoding processes.

A method for predicting drug pathways that modulate cellular phenotypes is provided. The method includes obtaining a logic-based differential equation model of a signaling network relating to a cellular phenotype, comprising nodes representing proteins or mRNAs and edges representing interactions. The method further includes identifying candidate compounds affecting the cellular phenotype, determining protein targets using a drug-target database, and mapping pathways from protein targets to network nodes by searching a protein interaction database and computing ranked pathways by minimizing a cost function based on edge weights. The method includes expanding the model by incorporating mapped pathways, simulating expanded model variants to predict compound effects on the cellular phenotype, and outputting predictions of how candidate compounds modulate the cellular phenotype through identified pathways.

Disclosed in the present invention are an adaptive conductivity measurement method and a system. The method comprises the following steps: 1) collecting in real time environmental temperature and illuminance data in a set period; 2) measuring the conductivity of a solution; 3) using a partial differential equation to simulate physical and chemical processes in the solution; and 4) calculating a conductivity compensation coefficient on the basis of the differential equation model obtained in step 3), and adjusting a conductivity measurement value in real time on the basis of the conductivity compensation coefficient. In the method of the present invention, temperature and illuminance changes are compensated on the basis of a partial differential equation model, so that measurement errors caused by environmental changes can be effectively eliminated, thereby improving the accuracy and reliability of conductivity measurement in experiments and industrial applications.

This invention discloses a method and system for identifying neutral line faults in a distribution network. The method includes: acquiring low-frequency channel data and high-frequency channel data; constructing a zero-sequence loop time-domain differential equation model based on the low-frequency channel data using a recursive least squares method with a forgetting factor, and identifying the pure resistance and inductance parameters of the neutral line in real time; extracting intrinsic mode components whose center frequencies fall within a preset frequency range using a variational mode decomposition algorithm based on the high-frequency channel data, and calculating their multi-scale permutation entropy as a micro-arc characteristic index; performing multi-mode comprehensive judgment based on the identified parameters and characteristic index; when the alarm criteria are met, constructing a first evidence body based on the pure resistance parameters and micro-arc characteristic index, and constructing a second evidence body based on the standard deviation calculated from the voltage data of the user-side smart meter; fusing the evidence using Dempster's synthesis rules, and outputting the final judgment result. This invention achieves a leap from post-disconnection alarm to pre-loosening warning, effectively reducing the false alarm rate.

The application discloses a lower limb venous thrombosis risk prediction method based on machine learning, which comprises the following steps: step one, constructing a three-axis thrombus evolution dynamic container space; step two, generating a pre-thrombus microstate orbit chain; step three, determining a thrombus formation critical mutation window; step four, introducing a causal intervention operator at the thrombus formation critical mutation window to construct a natural evolution path and a controlled path; step five, obtaining an orbit offset difference value; step six, calculating a dynamic orbit stability index; step seven, inputting the pre-thrombus microstate orbit chain and the dynamic orbit stability index into an improved PatchTST model, introducing a flow continuity adjustment operator into a reconfiguration module, and obtaining a corrected orbit stability index; and step eight, outputting a thrombus formation risk prediction result. The application realizes thrombus formation risk prediction by constructing a three-axis thrombus evolution dynamic container space and combining a neuro-architecture differential equation model and an improved PatchTST model.

This invention relates to the fields of engineering thermophysical calculations and industrial descaling technology. It provides a method for analyzing the residual evaporation of a water column under high-temperature conditions. The method includes: collecting operating parameters of the water column jetting operation and the thermophysical property parameters of the water; establishing a water column trajectory model, inputting the operating parameters and property parameters into the model to calculate the maximum duration required for the water column to reach a specified distance in the descaling pipeline under high-temperature conditions; constructing and using the integral transform method to derive and solve a two-dimensional partial differential equation model of heat conduction for the cross-section of the water column, obtaining the temperature distribution inside the water column at different times; and calculating the residual water content when the water column reaches the specified distance using the solved two-dimensional partial differential equation model of heat conduction for the cross-section of the water column based on the maximum duration, and outputting the analysis results. This invention can quickly and accurately predict the internal temperature field and residual evaporation of a high-temperature water column without conducting high-risk physical experiments.

This application relates to a multimodal artificial intelligence-based system for assessing the transformation process of hepatocellular carcinoma. The system includes: a data acquisition module for acquiring multi-phase image sequences, clinical indicator sequences, and treatment record sequences; an image registration module for performing non-rigid registration of multi-phase image data to obtain spatially registered image sequences; a lesion feature extraction module for extracting a dynamic lesion feature set; a multimodal feature fusion module for fusing the dynamic lesion feature set with clinical indicators to obtain an observation variable vector; an encoding module for encoding the treatment record sequence into a treatment mode vector; a trajectory prediction module for obtaining a continuous transformation state trajectory based on a neuronormal differential equation model; and an evaluation decoding module for obtaining the transformation process assessment results. This system can handle irregular follow-up intervals and dynamic changes in lesions, accurately characterizing the continuous evolution of tumor biological behavior.