17 results about "Back propagation neural network" patented technology

The application relates to the technical field of tunnel engineering disaster prediction, and discloses a soft rock tunnel large deformation prediction method based on a particle swarm optimization neural network, which comprises the following steps: selecting geological construction characteristic parameters such as water content of surrounding rock and construction methods as input factors, and establishing a qualitative index and quantitative value mapping; based on orthogonal test design and numerical simulation, engineering sample data sets are constructed and normalized; subsequently, the initial connection weight and bias of an error back propagation neural network are globally iteratively optimized by using a particle swarm optimization algorithm, so that optimal initial values are obtained by taking network errors as fitness values; finally, data training is carried out to generate a mature model, and a predicted deformation value is output and combined with a double check mechanism to carry out engineering early warning. The application solves the problems that traditional neural networks are slow in convergence and prone to falling into local optimization due to random parameter initialization, and effectively improves the prediction accuracy of soft rock tunnel large deformation and the reliability of engineering early warning.

The application discloses a kind of mine roadway destruction damage state monitoring method, device, equipment and medium, it is related to mine shock evaluation technical field, comprising: determining different earthquake source release energy and the peak particle velocity of each roadway surface distance, based on the initial value of attenuation fitting coefficient determined by each roadway surface peak particle velocity, according to initial value determine roadway surface peak particle velocity threshold;Determine target support mechanics model by roadway surface peak particle velocity threshold and pre-constructed support mechanics model;Determine a plurality of support parameter vectors based on target support mechanics model, train initial back propagation neural network by each support parameter vector, damage degree and roadway surface peak particle velocity, obtain target back propagation neural network;Determine the damage degree of roadway based on target back propagation neural network, current support parameter vector and current roadway surface peak particle velocity.The accurate evaluation of mine roadway destruction damage is realized.

The application discloses a kind of steel reinforced concrete column impact damage prediction evaluation method based on machine learning, when steel reinforced concrete column suffers impact, multiple-source heterogeneous data are synchronously collected;The multiple-source heterogeneous data collected are processed, multi-dimensional damage features are extracted, and the features extracted are fused to construct a comprehensive feature vector;A back propagation neural network model based on particle swarm optimization algorithm is constructed and trained, the input of the model is the comprehensive feature vector, and the output is the comprehensive damage index of steel reinforced concrete column;For the damaged steel reinforced concrete column to be evaluated, obtain its comprehensive feature vector, and input it into the trained back propagation neural network model, obtain the predicted comprehensive damage index, to complete the quantitative evaluation of damage state.The present application precisely fuses multiple-source monitoring data, fully considers the characteristics of SRC composite structure, and realizes automatic intelligent diagnosis of impact damage prediction and evaluation by machine learning.

The application discloses a laser processing quality prediction method based on an improved starfish optimization algorithm and relates to the technical field of neural network models. The method comprises the following steps: first, improving the starfish optimization algorithm based on the Levy flight principle to obtain an improved starfish optimization algorithm; second, optimizing the weights and biases of a back propagation neural network based on the improved starfish algorithm, and inhibiting overfitting of the back propagation neural network based on an L1 regularization method to obtain a target back propagation neural network; third, acquiring process parameters of silicon carbide laser processing; and finally, inputting the process parameters into the pre-trained target back propagation neural network to obtain the processing quality parameters of the silicon carbide laser processing. The starfish optimization algorithm is improved by introducing the Levy flight principle, and high-precision prediction of the silicon carbide picosecond laser processing quality is realized.

The application belongs to the field of material nondestructive testing, and particularly relates to a P91 pipeline aging grade prediction method based on micro-magnetic characteristic parameters, which comprises the following steps: a two-dimensional calibration system covering microstructure characteristics and macroscopic mechanical properties is constructed, and multi-level standard samples are prepared; multi-dimensional micro-magnetic data of each level of samples under step-by-step tensile load is collected, and a full-dimensional data set containing aging grade, stress level and micro-magnetic parameters is constructed; a ReliefF feature screening algorithm is used to identify core characteristic parameters which are highly sensitive to aging and can effectively distinguish stress interference; a back propagation neural network prediction model is constructed with the core parameters as input and the calibration grade as output, and is applied to field detection. Through core parameter screening and multi-parameter model fusion, the application realizes effective decoupling of stress and aging influence, guarantees the integrity of the pipeline structure, and significantly improves the accuracy of quantitative prediction of the aging grade.

The application discloses a random load fatigue damage evaluation method based on a physical information neural network, and comprises the following steps: selecting input features: stress amplitude data of a random load collected for a time length, comprehensively considering time domain features and frequency domain features, considering the influence of average stress, selecting different mechanical properties of materials, time domain amplitude features of random loads and frequency domain features of random loads as input features, and output features being fatigue damage D; establishing a back propagation neural network: the number of input layer neurons is consistent with the input features, the number of output neurons is one, physical knowledge is embedded in the back propagation neural network architecture, and a loss function embedded with the physical knowledge is established; and training the back propagation neural network for subsequent fatigue damage prediction. The back propagation neural network method embedded with the physical knowledge is established, the influence of the load on the fatigue damage is captured, and the overfitting problem possibly occurring in a pure data driven method is avoided.

The present application relates to underground mining technical field, disclose a kind of roadway roof displacement field advance prediction method, equipment and medium, comprising: obtaining the measurement data of target roadway survey area, according to measurement data constructs the roadway numerical model of target roadway survey area;Stress simulation data is obtained by simulating and solving to roadway numerical model;Stress simulation data is used to train the model training of pre-constructed back propagation neural network, and displacement field prediction model is obtained;Displacement field prediction model is used to carry out displacement field prediction to the point data to be predicted, and displacement field prediction result is obtained.The present application can realize the nonlinear mapping from mining condition and geological parameter to displacement field, based on the inversion process of pure data driving, significantly improve the efficiency and adaptability of displacement field prediction.

The present application belongs to the technical field of flying robots, and particularly relates to a UAV safety control method based on fault risk learning, comprising the following steps: firstly, performing actuator fault modeling; then, designing a fixed-time fault observer to quickly and effectively estimate the fault; secondly, using conditional value at risk to quantitatively evaluate the fault risk to obtain the required label true value for learning, and learning the influence of the fault risk on the UAV position uncertainty based on a back propagation neural network; finally, designing an adaptive risk-oriented control compensation strategy to realize smooth adjustment and adaptive safety control of the control strategy. The method can improve the control response speed and trajectory tracking accuracy of the UAV under different fault levels, and is particularly suitable for task scenarios with high requirements for flight safety and reliability, such as inspection operations in complex environments, search and rescue and delivery tasks in disaster areas, and dangerous source monitoring and early warning.

The application relates to the technical field of sensor signal processing, and discloses a multi-gas sensor solving method and system resisting complex cross interference. The method comprises the following steps: calibrating a single-gas power law response model, fitting a sensitivity coefficient matrix and a power law index matrix; collecting mixed gas orthogonal experiment data and normalizing the mixed gas orthogonal experiment data to construct an original training set; mapping each sample to a response increment space, screening a sparse area sample set according to a near neighbor average distance; generating a logarithmic interval virtual sample in a concentration range of the sparse area, constructing a segmented weighted fitness function according to a local sparsity index, driving a sparrow search algorithm to optimize a back propagation neural network to complete weight optimization, and outputting each gas concentration solving result. The application improves the concentration solving precision of a sensor array under a low concentration complex cross interference working condition.

The application discloses a kind of based on particle swarm optimization algorithm optimization back propagation neural network real vehicle lithium ion battery fault early warning and diagnosis method and system, it is related to battery fault diagnosis technical field, including: from real vehicle data screening driving section data and carry out cleaning, extract temperature voltage outlying degree and statistical characteristics, when fault section data is missing, additional extraction key period feature before fault, after division obtains training set and test set, training set and battery state label are substituted into PSO-BP model and are trained, iteration optimization until model converges, and real vehicle lithium ion battery fault early warning diagnosis model is obtained, and the performance of model is evaluated using test set;Collect real vehicle lithium ion battery driving section characteristic parameter, input evaluation result qualified real vehicle lithium ion battery fault early warning diagnosis model carries out fault early warning and diagnosis.The application can realize early warning and accurate diagnosis using online data under complex working conditions of real vehicle, improve the security of lithium battery when real vehicle is running.

The application belongs to the technical field of satellite communication, and discloses a frequency offset estimation method, system, device and medium based on a clustering neural network. Firstly, a random access preamble root index selection principle is proposed, so that the relationship between the power delay spectrum generated by the receiving end preamble sequence and the carrier frequency offset is unique and is not affected by timing advance compensation error and multipath effect. Then, power delay spectrum dataset is generated by all available preambles under different frequency offsets, and for each root index, the sparsity and regularity of the power delay spectrum matrix are fully utilized to construct an efficient and lightweight comprehensive frequency offset estimation model based on the semi-supervised K-means algorithm optimized based on the initial value and the back propagation neural network optimized based on the sparse dimension reduction, which can be respectively used for estimating the integer part and the decimal part of the frequency offset. Through the optimization of the neural network parameters in the training process by using the dataset, the final frequency offset estimation model is obtained, and complete frequency offset estimation under each root index can be realized.

The application provides a kind of neural network-based turnout beam real-time vibration response monitoring method and system, by obtaining the original vibration response signal flow of turnout beam key section vibration sensing network;Original signal flow is reconstructed in time and space, mapped to the three-dimensional tensor space aligned with turnout beam physical coordinates, to obtain vibration response space-time distribution tensor;The space-time distribution tensor is input into the train load back propagation neural network, and the dynamic load transfer path mode and vibration energy attenuation topology are generated by load feature coding and time series propagation chain processing;Call beam body state evolution neural network to carry out multi-step state deduction, obtain beam body state evolution sequence and extract abnormal propagation path and structure weak link positioning;Accordingly, monitoring instruction stream is pushed to trackside monitoring and early warning terminal.The method realizes the accurate space-time modeling and future state deduction of turnout beam vibration response, effectively improves the early warning timeliness and positioning accuracy of turnout beam structure health monitoring.

Embodiments of the present application relate to the technical field of computer materials science, in particular to a method and system for determining high-temperature creep performance of molybdenum-based alloys and a computer storage medium. The method mainly comprises: determining the grain size change curve and the polycrystalline equivalent elastic modulus of the polycrystalline structure of the molybdenum-based alloys corresponding to different alloy compositions and service condition ranges according to the physical property parameters of the molybdenum-based alloys corresponding to different alloy compositions and service condition ranges; determining the high-temperature creep performance corresponding to the alloy composition and the service condition range of the molybdenum-based alloys according to the grain size change curve and the polycrystalline equivalent elastic modulus; and training a back propagation neural network model by using the high-temperature creep performance. The method enables the model to accurately reflect the mapping relationship between the alloy composition and the service condition range of the molybdenum-based alloys and the high-temperature creep performance thereof, and compared with traditional empirical models, the prediction accuracy and efficiency of the high-temperature creep performance of the molybdenum-based alloys are effectively improved.

The application relates to a high-precision humidity signal control method based on a temperature and humidity transmitter, a medium and equipment, which comprises the following steps: obtaining original humidity signals and original temperature signals collected by a temperature and humidity sensor, carrying out empirical mode decomposition denoising processing, and obtaining denoised temperature and humidity signals, wherein the empirical mode decomposition denoising processing is to decompose the nonlinear original humidity signals and the original temperature signals into intrinsic mode functions with different frequency characteristics to obtain the denoised temperature and humidity signals; inputting the denoised temperature and humidity signals into a target back propagation neural network to obtain a compensation temperature and humidity signal output by the target back propagation neural network after temperature compensation, wherein the target back propagation neural network is obtained by optimizing a back propagation neural network through a sparrow search algorithm; determining a control strategy for the temperature and humidity transmitter according to the compensation temperature and humidity signal and a pre-set temperature and humidity threshold value, and controlling the temperature and humidity transmitter according to the control strategy.

The present application discloses a precipitation prediction method, a precipitation prediction device, a storage medium and an electronic device, and relates to the field of data processing, which can improve the accuracy of precipitation prediction. The precipitation prediction method comprises the following steps: obtaining current precipitation observation data of a target prediction area and current environmental influence factors of the target prediction area; extracting target features from the current environmental influence factors based on a random forest; inputting the current precipitation observation data and the target features into a convolutional neural network model, a back propagation neural network model and a time convolution network model to obtain first prediction data, second prediction data and third prediction data output by the three models respectively; and determining persistent precipitation data of the target prediction area by combining the first prediction data, the second prediction data and the third prediction data.

The application provides a kind of based on unmanned aerial vehicle space positioning's beam tracking method and system, wherein the method includes constructing unmanned aerial vehicle ground millimeter wave communication system model and receiving spectral efficiency model;Coordinate system is constructed to establish the relationship between unmanned aerial vehicle attitude and transmission beam azimuth and elevation angle;Unmanned aerial vehicle attitude is estimated by extended Kalman filtering and back propagation estimation;Unmanned aerial vehicle trajectory is predicted;According to the trajectory prediction result of unmanned aerial vehicle, the extended Kalman filtering estimation result and the back propagation estimation result of unmanned aerial vehicle attitude are input into receiving spectral efficiency model, and the real spectral efficiency is obtained, to quantify the beam tracking effect with real spectral efficiency.The application estimates the attitude of unmanned aerial vehicle by combining back propagation neural network and extended Kalman filtering algorithm, and predicts the trajectory of unmanned aerial vehicle at base station, combines the estimation and prediction results, obtains spatial beam angle through coordinate conversion, forms analog beam forming vector, and realizes accurate beam tracking.