Wang Keqi

Classification of wood surface texture based on Gauss-MRF model

The basal theory of Gauss-MRF is expounded and 2–5 order Gauss-MRF models are established. Parameters of the 2–5 order Gauss-MRF models for 300 wood samples’ surface texture are also estimated by using LMS. The data analysis shows that: 1) different texture parameters have a clear scattered distribution, 2) the main direction of texture is the direction represented by the maximum parameter of Gauss-MRF parameters, and 3) for those samples having the same main direction, the finer the texture is, the greater the corresponding parameter is, and the smaller the other parameters are; and the higher the order of Gauss-MRF is, the more clearly the texture is described. On the condition of the second order Gauss-MRF model, parameter B1, B2 of tangential texture are smaller than that of radial texture, while B3 and B4 of tangential texture are greater than that of radial texture. According to the value of separated criterion, the parameter of the fifth order Gauss-MRF is used as feature vector for Hamming neural network classification. As a result, the ratio of correctness reaches 88%.

An Improved Ant Colony Algorithm Combined with Particle Swarm Optimization Algorithm for Multi-objective Flexible Job Shop Scheduling Problem

In this paper an improved ant colony algorithm is presented and an algorithm in combination with particle swarm optimization algorithm and the improved ant colony algorithm for multi-objective flexible job shop scheduling problem are employed. The algorithm proposed in this paper includes two parts. The first part makes use of the fast convergence of PSO to search the particles optimum position and make it as the start position of ants. The second part makes use of the merit of positive feedback and structure of solution set proposed by our improved ACA to search the global optimum scheduling. The algorithm we presented is validated by practical instances. The results obtained have shown the proposed approach is feasible and effective for the multi-objective flexible job shop scheduling problem.

Fault Diagnosis Method of HV Circuit Breaker Based on Wavelets Neural Network

Fault diagnosis of HV circuit break has been investigated extensively as an important device in the field of power system. In view of the shortcoming of the traditional neural network, such as the slow convergence rate and the lo- cal minimum easy to form, fault diagnosis method of HV circuit breaker is proposed to remedy the defects of traditional neural network based on wavelets neural. This method adopts wavelets function rather than the hidden nodes of traditional neural network, which is propitious to conducive to achieve a rapid convergence of online learning. This work firstly dis- cussed the principles of fault diagnosis method in detail, and then compared diagnosis effect using wavelets function with that of the traditional neural network. The results show that the training speed and classification effect of wavelets neural network are superior obviously to those of traditional neural network. Wavelets neural network based on vibration signals is more suitable in application to the fault diagnosis of HV circuit breakers.

A New Fault Diagnosis Method for High Voltage Circuit Breakers Based on Wavelet Packet and Radical Basis Function Neural Network

A new method that researching fault diagnosis of high-voltage (HV) circuit breaker (CB) is proposed. The method combines Wavelet Packet (WP) with Radical Basis Function (RBF) Neural Network (NN). Firstly, by applying the theory of WP decomposition and reconstruction, the mechanical vibration signal of CB was decomposed into different frequency bands, and the coefficients are reconstructed in the corresponding node. After that, the feature vector was ex- tracted by equal-energy segment entropy from reconstructed signals. Finally, fault diagnosis has been realized through the classification of feature parameters combined with RBF neural network. The experiment outputs show that the method can be applied in diagnosis.

A method of classified HV circuit breaker fault signal based on EEMD and BP neural network

During the process of high voltage circuit breaker run, the change of the vibration signal reflects the mechanical state of the circuit breaker, an efficient method of extracting vibration signal features is directly related to the accuracy and practicability of fault diagnosis. In the paper, a status feature extraction based on overall empirical mode decomposition (ensemble empirical mode decomposition EEMD) and correlation dimension has been presented. Firstly, the original non-stationary vibration signals are broken down to a plurality of stationary intrinsic mode function (IMF); Secondly, using GP algorithm to calculate the correlated dimensions of first four IMF as a high voltage circuit breaker vibration signals feature vectors. Finally, constructing BP (back propagation) neural network to classify the feature vectors. Through testing different fault vibration signals of circuit breaker, it showed that the method can accurately diagnose all kinds of circuit breaker fault state and provide a new thinking way about fault diagnosis.

Feature extraction of HV circuit breaker based on ensemble empirical mode decomposition and correlation dimension

During the operation process of the high voltage circuit breaker, the changes of vibration signals reflect the machinery states of the circuit breaker. The extraction of the vibration signal feature will directly influence the accuracy and practicability of fault diagnosis. This paper presents an extraction method based on ensemble empirical mode decomposition (EEMD) and correlation dimension. Firstly, original vibration signals are decomposed into a finite number of stationary intrinsic mode functions (IMFs). Secondly, calculate the correlation dimension of the top four IMF by the G-P algorithm. At last, calculate the root mean square value of the four correlation dimensions to be as the characteristic of vibration signal for HV circuit breaker. Practical examples show that the characteristics through the extraction approach put forward can represent effectively fault patterns of HV circuit breaker.

Study on nonlinear compensation of eddy current sensor based on support vector machine

The input/output characteristic curve of eddy current sensor is nonlinear. In order to ensure the linear relationship between the input and output, the compensation of eddy current sensor is very necessary. According to the non-linear problem caused by temperature and circuit interference during the detection process of the eddy current sensor, a nonlinear compensation method is proposed based on the support vector machine theory. The nonlinear characteristics of measured values of sensor are analyzed, and the support vector machine (SVM) inverse model is established while the displacement parameters as the output and voltage parameter as input. The effectiveness of SVM inverse model is verified by the simulation. Compared with the RBF neural network model, the SVM model is efficiency. The average absolute error of the model prediction is 0.0932mm, so it has good linearity, and nonlinear compensation of eddy current sensor is realized.

Rheological properties of low consistency TMP from thinning wood of Chinese fir

The rheological behavior of low consistency thermomechanical pulp of Chinese fir harvested by intermediate thinning was analyzed. The results show that the apparent viscosity of pulp changed along with the beating degree, pulp consistency and shearing velocity. With the increasing of pulp consistency, the apparent viscosity of pulp increased gradually. Beating degree of pulp had an effect on micro-structure of pulp. The apparent viscosity of pulp declined as beating degree of pulp increased, and the apparent viscosity of pulp fell along with the shearing velocity increasing. Based on the results, the rheological models are set up. The models showed that the fluid types of the low consistency pulp could be described as pseudoplastics fluids (non-Newtonian fluids).

Feature Extraction of defects on wood surface

The computer image processing technology was used to accomplish the feature extraction of defect images on wood surface. By calculation of gray values of defects. three feature data which are useful to identify the defects have been achieved. The experiment indicates that this way is effective to the automation recognition of the defects on wood surface.

A machine vision system for inspecting wood surface defects by using neural network

With the development of wood industry, the processing of wood products become more significant. This paper discusses the development of machine vision system used to inspect and classify the various types of defects of wood surface. The surface defects means the variations of colour and texture. The machine vision system is to detect undesirable “defects” that can appear on the surface of rough wood lumber. A neural network was used within the Blackboard framework for a labeling verification step of the high-level recognition module of vision system. The system has been successfully tested on a number of boards from several different species.