The chips feature high dispersion, high transparency, high gloss, and bright color, without smell or dust, and meanwhile maintain stable performance, safety, and environmental protection, which generate great benefits in terms of storage and transportation. With its excellent compatibility and dispersibility, Keytec Pre-Dispersed Pigment Chips allow customers to prepare their ideal colorants in a short time. The relatively high pigment content generally ranges from 30% to 80% (which depends on system types), while the resin content can reach 20% to 70%. CAB Predispersed Pigment Chips, Pre-dispersed Pigment Material, Pre-dispersed Pigment Paste Guangzhou Quanxu Technology Co Ltd , https://www.silicamattingagent.com
The key to gear fault diagnosis is how to extract valid eigenvalues ​​from the acquired signals to characterize their fault conditions. The traditional feature extraction method often has redundant information between the feature quantities and erroneous interference information, which often reduces the efficiency of diagnosis and even misleads the diagnosis results. In this paper, a gear fault diagnosis method based on rough set neural network is proposed. Firstly, the rough set is used to reduce the attribute of the extracted gear feature set, and then the reduced feature set is taken as input, and BP neural network is used for modeling. Effectively improve the efficiency and accuracy of gear fault diagnosis.
Rough Set Neural Network Structure and Principle: The concept of rough set based feature attribute reduction rough set was first proposed by Z.Pawlak, which has a wide range of applications in the knowledge mining of inaccurate, uncertain and incomplete systems. The main idea is to derive the decision-making or classification rules of the problem through the reduction of the attributes under the premise of keeping the classification ability of the information system unchanged. In the data preprocessing of high-dimensional feature quantity, the rough set can be used to reduce the attribute of high-dimensional data and eliminate the redundant information in the feature data, so as to achieve the dimensionality reduction of high-dimensional feature quantity. Get a more accurate feature quantity. The use of rough sets for feature attribute reduction can be divided into the following steps:
1) Defining the domain: Let 21nxU be a non-empty set of objects to be studied, called the domain. Define fVDCUS as an information system, where DCP is a set of attributes; C is a conditional attribute, D is a decision attribute; pUV∈=α, which is a set of values ​​of the attribute P∈α; VPUf→×: is an information function.
2) Discretization of feature attributes: The dispersion of feature attributes is to interpolate the attribute sets of all objects in the universe in a certain way to obtain a series of discretization intervals, thereby dividing the original attribute sets into different discrete intervals and reducing The size of the attribute set value field.
3) Reduction of knowledge: Knowledge reduction is to obtain the minimum knowledge coverage of the information system S by calculating the discrimination matrix and the resolution function, and to remove unnecessary attributes to achieve the simplification of the system.
4) Generation of logical rules: According to the reduced simplified form, the knowledge kernel of the information system is determined, and the rule set of all decision classes is obtained.
How to extract valid eigenvalues ​​from the acquired signals