Papers by Keyword: PNN

Abstract: With the help of wide area information, a new fault identification algorithm of power grid based on PNN is proposed. This algorithm gives a definition of the line associated domain, the elements’ action information of the line associated domain gathered by line IEDs can form the feature vector into PNN classifier, and then the fault elements of power grid would be identified on PNN classifier. Through a large number of simulation experiments, it shows that the new fault identification algorithm of power grid based on PNN and wide area information has high accuracy and good fault tolerance.

Abstract: In order to realize functions of automatic monitoring, alarming and extinguishment on fire in crude oil storage tank, the paper came up with a fire detection model with multisensor, including selective module with fire feature combination, supervised training module and fire detection module. By regarding PNN as a classifier to carry out tests on effectiveness of the model, the conclusions that the model can reduce the influence of fire parameters’ fluctuation on detection results was drew. Moreover, an excellent fault-tolerant ability was possessed at the same time. Through some confirmatory experiments, The phenomenon was reached that two kinds of parameters in adoptive four parameters have no normal fire signal, but the model still can greatly distinguish from correct fire state.

Abstract: The principle of neural network’s PNN algorithm was introduced, Combining with the structure feature and work principle of the hydraulic pump, a fault diagnosis system based on PNN neural network was established. The feasibility of the system was proved through the identification, emulation and experimentation of hydraulic system’s fault patterns. The PNN control model was simulated using Matlab/Simulink toolbox. This model analyzed and studied the PNN network predictive diagnostic rate. Under different sample size and SPREAD, the simulation’s results show that this method has favorable identified capability of fault mode and favorable applicability to the hydraulic pump.

Abstract: Infant pain is a non-stationary made by infants in response to certain situations. This infant facial expression can be used to identify physical or psychology status of infant. The aim of this work is to compare the performance of features in infant pain classification. Fast Fourier Transform (FFT), and Singular value Decomposition (SVD) features are computed at different classifier. Two different case studies such as normal and pain are performed. Two different types of radial basis artificial neural networks namely, Probabilistic Neural Network (PNN) and General Regression Neural Network (GRNN) are used to classify the infant pain. The results emphasized that the proposed features and classification algorithms can be used to aid the medical professionals for diagnosing pathological status of infant pain.

Abstract: Multi-target tracking is one of the basic and difficult tasks in video analysis and understanding. This paper proposed an efficient tracking algorithm based on meanshift algorithm and PNN (Probability Neural Network) background model. Firstly, PNN detection results were used to initialize targets for meanshift tracking. Secondly, in the succeeding frames, every target was matched to detected regions before tracking. At last, only targets which couldnt match with new regions need tracking with meanshift tracking algorithm. Experimental results show that mean search steps for every target were dramatically reduced compare with original mean shift tracking algorithm.

Abstract: The artificial neural networks have received wide research efforts in fault diagnostics in recent years. This study proposes two types of feedforward neural networks (PNN and GRNN) for diagnosing the fault of the steam turbine. The eigenvectors of the vibration signals in steam turbine can be extracted by the time-domain analysis after the wavelet packet decomposition and reconstruction. Depending on these eigenvectors, we developed the fault diagnosis program with the PNN and GRNN approach for the steam turbine in Matlab, and diagnosed two common faults of steam turbine (mass unbalance and oil whirl). The diagnostic accuracy is up to 94.44%, and the diagnostic time is short. The results demonstrate that the diagnostic approach is able to identify the common faults of steam turbine quickly and efficiently.

Abstract: Movement whether it is actual or imaginary can produce different electroencephalogram (EEG) signals. How to extract features of signals and accurately classify them is a key to brain-computer interface(BCI) system. In the paper, BCI competition data downloaded from BCI website are used as study object, through time-domain analysis and frequency-domain analysis, according to the attribute of event-related synchronization (ERS) and event-related desynchronization (ERD) during imagery movement, energy difference of lead C3 and C4 are selected as features and wavelet package is used to extract them. Probabilistic neural networks (PNN) is used as classification method. Compared with other two calssification methods such as support vector method (SVM) and liner classifier, the classification accuracy rate of PNN reaches to 89.2% steadily and is higher than them. It is proved that the method provided in the paper are effective for identifying imaginary movements.

Abstract: In this paper, we analyze characteristics of two kinds of GA-Based neural networks. For large scale neural networks, it is necessary to optimize the initial network parameters. Using the global optimum ability of GA(Genetic Algorithm), we optimize the initial weights and biases of BPNN (Back-Propagation Neural Networks), which can avoid the local minimum. And we also optimize the spread coefficient of Gaussian Radial Basis Function of PNN (Probabilistic Neural Networks). Then the results in transformer fault diagnosis are compared. Experimental results based on Matlab show that the method of GA-Based greatly increases the reliability of diagnosis.

Abstract: In order to diagnose the health state of Aircraft effectively, a new method based on ARMA Model and probabilistic neural network(PNN) is proposed in this paper. First, an ARMA model is built using the original acoustic emission signal of aircraft crucial components, then use the autoregressive approximation theory to estimate model parameters, and order of the model is calculated according to Akaike Information Criterion(AIC). Use the autoregressive parameters to build feature vectors, then the probabilistic neural network is used to carry out the recognition of these feature vectors, and the health state of aircraft crucial components is effectively diagnosed. After the application on certain type of real aircraft, this method is proved to be capable of detecting the fatigue crack on crucial structural components. And we can conclude that the method is an effective way to carry out aircraft health diagnosis.

Abstract: For the probability neural network (PNN) algorithm is the non-surveillance's pattern taxonomic approach, the work load major problem, moreover the category number's selection will affect the cluster performance. How to optimize PNN enabled it to play a more effective role in the classified question, this paper proposed one use genetic algorithm optimization probability neural network method: introduction the auto-adapted mechanism genetic algorithm, to the probability neural network's parameter carries on the training, formed the supervised learning probability neural network based on the genetic algorithm, overcome the probability neural network existing algorithm flaw. Then introduces this model in the quality control, guaranteed that the production process is at the control state, achieves the quality control goal. Carries on the test through the simulation experiment to this algorithm, and with the probability neural network, the BP neural network carries on the comparative analysis, proved this method accuracy is high.