Hongxia Pan

A Fault Diagnosis Method under Varying Rotate Speed Conditions Based on Auxiliary Particle Filter

Varying rotate speed can cause changes in a measured gearbox vibration signal. There is a need to develop a technique to provide accurate state indicator of gearbox under fluctuating rotate speed conditions. This paper presents an approach for gearbox fault detection under varying rotate speed condition based on auxiliary particle filter. Firstly, the model of vibration part which sensitive to the alternating rotate speed condition was established based on the relation of cosine signal three points sampling values. Then this part vibration signal was estimated based on auxiliary particle filter. Based on these the residual signal was obtained which lower sensitiveness to the alternating rotate rate condition. Thus the gearbox fault was detected by the residual signal statistic quantity kurtosis and amplitude of Fourier transform. Finally, the different work condition vibration signals of the laboratory gearbox under varying rotate speed condition were detected and signal processing was studied with those signals as examples. The results show that the proposed method is feasible and effective.

Singularity localization and characterization based on continuous wavelet transform for vibration signals with impacts

Vibration signals tested from mechanical components usually contain a great deal of information about the properties and status of the mechanical parts or even the whole mechanical system, which is worthy of further analysis. A method based on continuous wavelet transform to localize and characterize the singularities of vibration signals with impacts is introduced in this paper. Continuous wavelet transform was first performed to the signal, and then the ridges and ridge functions of the absolute value of the wavelet coefficients were extracted. By tracking the ridge lines from large-scale to small-scale, the singular points of the signal can be localized quite precisely, which is corresponding to the times when mechanical impacts occur. By performing a double logarithmic linear regression of the ridge functions, we got use the slopes and the intercepts of the regressed lines, which can be used as features to characterize the singularity of the impacts. This work provides a reliable and effective method for signal truncation and singular feature extraction, which will benefit the subsequent signal analysis works.

High-speed automaton fault detection and diagnosis based on multivariate multiscale entropy

An early method using multivariate multiscale entropy(MMSE) has been described for crack fault diagnosis in gun high-speed automaton. first of all, Impact vibration multichannel data were acquired in the gun shot tests in which different crack parts were used. secondly, MMSE algorithm is applied to extract fault features from multichannel data in different scales. finally, the high-speed automaton fault categories was identified. The result of experiments indicate that this method is able to diagnose effectively and accurately the gun automaton crack fault and is a potential method that is deserved to be introduced.

Method of automaton fault diagnosis based on narrow-band decomposition and fuzzy nearness degree

To diagnosis the fault of automaton using vibration signals, a method based on local narrow-band decomposition and frequency-domain fuzzy nearness degree was proposed. Three different faults were set on easy-to-crack locations of automaton, using electrospark wire-electrode cutting. Then two groups of piezoelectrical acceleration transducers were set on the front of the catridge receiver and the top of the machine guns tail respectively. Each transducer group detected the vibration of two directions. The author took one set of signals (five shoots in succession) from each working conditions database as a test sample. And another sample (three shoots in succession) from Fault 3 signal database was taken additionally as the reference sample. In order to diagnose the fault more easily, the local narrow-band decomposition was used to highlight fault features. Then the similarity measurement based on frequency-domain fuzzy nearness degree was used to compare similarities between the reference sample and each test sample. The results showed that the nearness degree of Fault 3 (five shoots in succession) was much lower than other test samples. So this method could distinguish different working conditions of automaton and reflect the severity of fault.

The NNES fault diagnosis technique of the naval gun systems

The naval gun weapon systems are the complex and large mechatronic systems which consist of mechanical system, electrical system and hydraulic system and so on. The systems have a wider working range, a worse working environment and a higher failure rate. Whether ship borne gun weapon systems work normally or not, they directly affect the performance indexes of the weapon systems, even do the entire naval war situation. In the paper, a fault diagnosis technology is introduced, which integrates neural network and expert system into a coherent whole. The theoretical analysis and the experimental results of the failure modes have been made. It was drawn a conclusion that NNES fault diagnosis technology could be extended to the integrated systems of the fault diagnosis of the naval gun systems.

Automaton fault diagnosis method of multi-parameters integration for feature extraction

High-speed automaton is core component of small caliber artillery, because of its poor working condition, the crack and wear of each component and its working reliability have gradually become the focus of fault monitoring and diagnosis. This traditional test method (mainly used in the field of weapon) not only needs a lot of cost and time, but also is vulnerable to many uncertain factors. Therefore, this paper uses modern test and analysis method collecting automaton vibration signal during shooting action and applying signal processing methods to extract features susceptible to fault, so as to identify the fault. Considering the high-speed automatic movement process and its nonlinear vibration signal, short time, transient, impact properties. In order to make fault information to be highlighted, firstly, according to the automaton movement patterns decomposition and time cycle diagram, the time-domain signal peak corresponds to collision between parts, and vibration signal corresponding to the motion of fault component is intercepted as the analysis object. Then, wavelet threshold de-noising method is used to preprocess the signal, making the open atresia impact obviously. Secondly, In order to comprehensively measure signal fractal characteristics, the override method is used to calculate the vibration signal generalized fractal dimension and draw the generalized fractal dimension spectrum, box dimension, information dimension, correlation dimension automaton as fault feature values are extracted. Then quantitative diagnosis index at the level of feature information integration --the index distance of three demensional characteristic parameters is proposed. In view of the fault feature parameters extracted under various conditions, We compute the average respectively, then obtain the four standard centers separately representing automaton four conditions in three dimensional space. In view of the vibration signals to be detected, according to the extracting three-dimensional characteristic parameters, we can find the corresponding characteristic index points in the three-dimensional space, respectively calculate these distances of between the characteristic index points and four standard centers, the index distance of three demensional characteristic parameters, and draw graphs of the index distance of three demensional characteristic parameters to identify fault conditions intuitively. Some identification errors are found in certain condition. In order to improve the deficiencies, we are determined to increase dimensions, increase fault characteristic parameters to identify conditions. So, singular spectrum entropy, power spectrum entropy, local wave spatial spectral entropy are extracted as quantitative features to describe the state changes of signal in time domain, frequency domain, time-frequency domain. Calculating the index distance of six-demensional characteristic parameters is suggested to identify conditions. Two graphs of the index distance of six-dimensional and three-dimensional characteristic parameters are drawed simultaneously to increase the comparative. Diagnosis results indicate that: the index distance of six-dimensional characteristic parameters can accuratly identify fault conditions of automaton, compared to the index distance of three-dimensional characteristic parameters. So, increasing the fault characteristic parameters and dimensions can improve the accuracy of fault identification. Also, multi-fractal theory and information entropy are sensitive to extract fault characteristic values. This paper provides a new idea for fault diagnosis of automaton.

Automaton fault diagnosis based on EEMD and Multi-component information entropy

For the detection and identification problems of common faults in the use of automaton, considering the nonlinear, short time, transient and impact properties of the automaton vibration response signal, a method of fault diagnosis based on Ensemble Empirical Mode Decomposition (EEMD) and Multi-component information entropy is presented in this paper. Firstly, EEMD decomposition method was used for automaton vibration signal and the first five Intrinsic Mode Function (IMF) components was selected to analyze with the help of correlation coefficients. Then the energy percentage of each component was extracted as the fault feature and its validity was verified. The singular entropy, which is a more sensitive feature, was selected in this paper to reflect the characteristics of the fault information. Phase space reconstruction was then performed for each IMF component. After that, using the mutual information method and false nearest neighbor method, the optimum delay time and embedding dimension were determined. Then singular entropy was calculated respectively, and this multi-component singular entropy was used as a feature that provides a quantitative description of the change of the signal state. Finally, the support vector machine (SVM) was used to recognize the fault pattern. The diagnostic results show that the method proposed can extract the fault feature and recognize the faults effectively, which can be used to solve the problem of the fault diagnosis for automaton. This work provides a new way for the fault diagnosis of automatic weapons and has great theoretical and practical significance for the fault diagnosis of high-speed automaton.

Modeling and simulating in MapleSim for a bench experimental system of machine gun

A one-dimensional lumped parameter model for a bench experimental system of machine gun was established and simulated in MapleSim environment. The critical problems encountered during modeling and their corresponding solutions were expounded. Some typical simulation results were presented and some other issues relative to this work were discussed. The simulation results show that this modeling method is effective and reliable, and the simulation in MapleSim is efficient, which can be extended to other applications.

Application of Mathematical Morphology in Diesel Vibration Signals

In this paper, according to the characteristics of the diesel engine vibration signal, the design for a class of adaptive generalized morphological filter is applied to the noise reduction of diesel engine vibration signal. After pre-processing the diesel engine vibration signal, then it is designed the shape, width and height of the structure elements. This paper, according to the characteristics of the diesel engine noise, chooses the semicircle structural elements and calculates the local maximum signal sequence and local minimum signal sequence so as to strike the height and width of the structural elements, then uses the gradient method to find the weight value adaptive. As a result, it will make noise reduction to achieve effectively, and the result has some superiorities compared with the traditional wavelet noise reduction.

Fault diagnosis of a high-speed automaton based on structure vibration response analysis

Vibration response is a particularly important signal to characterize the system state. This paper analyzes the reason of fault generated high speed machine, vibration response mechanism and its frequency characteristic. According to the measured vibration signals, done time and frequency domain features analysis, wavelet packet analysis and frequency domain energy analysis, put forward a kind of fault comprehensive diagnosis method with accurate and rapid identification characteristics, can adapt to the complex vibration response signal with interference and low signal to noise ratio.