Huimin Zhao

Mechanical fault diagnose of diesel engine based on bispectrum and Support Vector Machines

The vibrant signal of diesel engine is analyzed by the method of bispectrum, and bispectral feature planes are searched along diagonal and parallel lines of diagonal at certain step in the bispectral modulus field, and the mean breadth value in bispectral feature planes are calculated as signal feature which is capable of describing the fault. The Support Vector Machines is used to diagnose the fault successfully by importing signal features as training samples. The experiment results show that, the noise in the vibrant signal of diesel engine can be eliminated by bispectrum and the signal feature can be extracted effectively; The signal features are perfectly described by feature planes and exist not only on diagonal and but also in the field besides the diagonal; Support Vector Machines can study effectively and diagnose successfully with limited fault samples.

A Method Combining Order Tracking and Fuzzy C-Means for Diesel Engine Fault Detection and Isolation

Diesel engine works under variable speed conditions; fault symptoms are clearer in the angular/order domains than in the common time/frequency ones. In this paper, firstly, the acceleration signal of diesel engine is resampled by order tracking, in which the rotating speed is computed in every working cycle, and the order tracking spectrum is created in each interval’s speed; then different order band accumulated energy is computed as feature vector. After standardizing these features, the fuzzy c-means (FCM) is introduced to use them as input vector; the optimized classified matrix and clustering centers can be obtained using FCM iteration method; then the fault can be detected by calculating the approach degree between the unknown samples and the known ones. To validate the method, some experiments have been performed; the results show that the signal can be reconstructed, and the features of order band accumulated energy can reflect the information of different wear conditions in crank-shaft bearing; then the fault can be detected accurately. The method of nonentire work cycle is also introduced as a comparison with our method; the result shows our method has more accuracy classification.

Fault detection in an engine by fusing information from multivibration sensors

Fault detection based on the vibration signal of an engine is an effective non-disassembly method for engine diagnosis because a vibration signal includes a lot of information about the condition of the engine. To obtain multi-information for this article, three vibration sensors were placed at different test points to collect vibration information about the engine operating process. A method combining support vector data description and Dempster–Shafer evidence theory was developed for engine fault detection, where support vector data description is used to recognize the data from a single sensor and Dempster–Shafer evidence theory is used to classify the information from the three vibration sensors in detail. The experimental results show that the fault detection accuracy using three sensors is higher than using a single sensor. The multi-complementary sensor information can be adopted in the proposed method, which will increase the reliability of fault detection and reduce uncertainty in the recognition of a fault.

An Approach on Fault Detection in Diesel Engine by Using Symmetrical Polar Coordinates and Image Recognition

Vibration technique provides useful information in fault detection of diesel engine, bringing significant cost benefits to diesel engine condition monitoring. Usually, time-frequency calculation on vibration signal is so complex that it is difficult to achieve online fault detection. In this paper, a method of fault detection in diesel engine is developed based on symmetrical polar coordinates and image recognition. In this method, time-domain waveform of vibration signal is transformed into snowflake-shaped in mirror symmetry pattern without time-frequency analysis. By the comparison of the geometric features of the snowflake images from different wear conditions of crankshaft bearing in diesel engines, we use centroid position and direction angle of the petal in snowflake image as features to detect the fault. Then, fuzzy c-means (FCM) are used to detect the conditions of the engine according to these features. In order to validate the methods, some experiments have been performed, the experimental results show that the centroid position and direction angle of the petal in snowflake image can reflect the information of different wear conditions in crankshaft bearing, and the fault of crankshaft bearing can be detected accurately. Hence, the method can work as fault detection in diesel engine, which is simple and effective, compared with time-frequency calculation method.

Unstable engine vibration signal analysis using cyclostationarity and support vector machine theory

According to the characteristics of unstable vibration signals, this paper proposes a combined approach to detect engine crank bearing mechanical faults by using cyclostationarity and support vector machine theory. The unstable vibration signals of engine accelerating process are analyzed by cyclostationarity theory. The fault diagnostic rules are generated by combining signal acquiring process and extracted fault features. And support vector machine is then trained. The result shows that the feature extraction is effectively realized by using cyclostationarity theory. Second order cyclical frequency bands of characteristic can be found corresponding to specific cyclical frequency. The support vector machine is superior to neural network because of the high classification precision and strong generalization ability for small samples. The diagnostic precision can be improved by means of optimizing parameters greatly.

Crankshaft Bearings Fault Diagnosis Based on SVD and Bispectrum

Singular value decomposition (SVD) can realize denoising without relying on spectral characteristics. It is more useful for small scale denoising. Bispectrum can effectively inhibit the interference of non Gaussian noise, which makes the signal feature extraction convenient. The two methods are combined in this research. In the beginning, the vibration signals of engine crankshaft bearings go through SVD-based denoising, and then the high-order spectral theory is adopted to get the bispectrum of the signals after denoising. In the end, the frequency band of the fault crankshaft bearings signal is extracted by searching the whole 2-D frequency field, and favorable diagnosing result is obtained.

Extraction of transmission bearing fault characters based on EMD and fractal theory

The vibration signal of transmission bearing is decomposed by empirical mode decomposition, and fractal dimensions of decomposed intrinsic mode functions are calculated to extract fault characters of transmission bearings of different conditions. The results show that empirical mode decomposition method is able to separate signals of different frequency bands effectively, and the fractal dimension of specific IMF component is able to reflect the technology state of transmission bearing sensitively, which can be selected as the characteristic parameters to diagnose transmission bearing fault. The combination of empirical mode decomposition and fractal dimension is an effective method of transmission bearings fault character extraction.