Reusable steel member service life evaluation method and lease management method
A technology for life assessment and steel components, applied in neural learning methods, sales/lease transactions, neural architectures, etc., to achieve accurate management methods and save resource costs
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0034] The neural network model commonly used is the BP (Back Propagation) neural network, which includes an input layer, a hidden layer, and an output layer, and uses an error back propagation algorithm. The weights and thresholds are updated by calculating the error between the output value and the true value. However, the initial weights and thresholds of the BP neural network are randomly obtained, which may easily lead to training into local optimization or prolong the optimization time, such as figure 1 As shown, the present embodiment preferably adopts a convolutional neural network (CNN) with more hidden layers for training to solve the problem that the initial weights and thresholds of the BP (Back Propagation) neural network are randomly obtained, causing the training to fall into local optimization or prolong A matter of optimizing time. .
[0035] Combine below Figure 4 The method of the present invention is further introduced:
[0036] S1. The step of establi...
Embodiment 2
[0042] The main difference between this embodiment and Embodiment 1 is that the prediction model is a support vector regression machine (SupportVactor Regerssion, SVR).
[0043] like figure 2 As shown, in the training method of the SVR model, the independent variables are the multi-dimensional comprehensive parameters and states of the training set, which are normalized into multi-dimensional feature vectors through data preprocessing. The dependent variable is the prediction result output by the SVR model, which is the possible bearing strength of the material (that is, the predicted value of the mechanical data). The SVR penalty factor and parameters are calculated by the genetic algorithm to create the SVR regression model, and then use the created model to predict the test set, output the predicted value of the test set, and compare and verify it with the measured results of the test set. The SVR regression model after verification can accurately predict the mechanical c...
Embodiment 3
[0045] like image 3 As shown, this embodiment is based on the aforementioned life expectancy method, combined with the results of mechanical sampling inspection, appearance inspection, size inspection and non-destructive testing of materials, to make a comprehensive quality evaluation of building materials, and to establish building material data files to achieve scientific, safe and convenient leasing management. Specifically, in the warehousing system, an identity is assigned to each steel component, and the trained prediction model is used in the storage and / or output process of the steel component to perform mechanical data detection and life evaluation, and the detection and evaluation results are used as the corresponding steel components. The quality data of the component is stored in the database together with the corresponding identity mark, and the quality management file of the steel component is established.
[0046] To sum up, the present invention obtains the c...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More