Refining device pipe network integrity management level intelligent evaluation method

Abstract

The invention relates to a refining device pipe network integrity management level intelligent evaluation method. The method comprises the following steps of 1, construction of a refinery device pipe network integrity management system, 2, determination of a refinery device pipe network integrity management evaluation index system, and 3, design of an intelligent algorithm. The fuzzy ultra-wavelet neural network optimized by using the constructed intelligent algorithm can determine the refinery device pipe network integrity management level, and has high evaluation precision and evaluation efficiency. According to the model, the intelligence of the refinery device pipe network integrity management level evaluation can be improved, and the actual condition of the refinery device pipe network integrity management level can be accurately obtained, so that favorable theoretical support is provided for perfecting a refinery device pipe network integrity management system.

Description

technical field [0001] The invention relates to a method for assessing the integrity management level of a pipe network, in particular to a fuzzy ultrawavelet neural network evaluation method optimized by an intelligent algorithm for the integrity management level of a refinery plant pipe network. Background technique [0002] The integrity management of the pipeline network has received extensive attention. A complete integrity management system can effectively reduce the incidence of accidents, improve the service life of the pipeline network, and reasonably allocate the maintenance costs of the pipeline network. The research on pipe network integrity management in my country has made considerable progress, and the constructed pipe network integrity management system has been industrialized. The pipeline network of refining and chemical equipment plays a very important role in the process of national economic development, and the development of the safety management model ...

AI Technical Summary

Problems solved by technology

At present, the main method of pipeline network integrity management evaluation is neural network, commonly used is BP neural network and deep neural network, BP neural network has the characteristics of simple structure and strong self-learning ability, but it also has poor generalization ability and easy to fall into The pitfalls of local minima

A deep neural network is a perceptron composed of multiple hidden layers, which can effectively map complex nonlinear relationships, but the increase in the number of neurons in the hidden layer of a deep neural network will lead to a more complex model, resulting in a large amount of calculation increase, the calculation efficiency is not high, and it is easy to cause the defects of gradient disappearance and low convergence efficiency

Wavelet neural network is a feed-forward neural network with strong learning ability, but it also has the following defects. Multi-dimensional variable input will lead to a huge neural network structure and low convergence efficiency

The free parameters of the neural network include weights and superwavelet parameters. The training process of the network is relatively complicated. The gradient descent algorithm is a commonly used network learning algorithm. During the training process, the network convergence efficiency is low. In order to meet the preset evaluation accuracy, More hidden layer neural units and longer training time are required, and the gradient descent algorithm has few adjustable parameters, which is difficult to control and prone to numerical oscillation

Images