To achieve Condition-Based Maintenance through IoT, it is very important to catch the symptom of faults. As a method of grasping the operating states with faults, it is conceivable to collect operation data by embedding the damaged part into the actual machine. If the equipment is large, the lifetime is long, or the equipment is expensive, it is difficult to accumulate data. Hence, it is important to carry out analytical investigations in order to gain some insight into diagnostics. In this research, we focused on rotating machinery. We created the rotating machinery library by the equivalent method as the transfer matrix method in Modelica. We described a modeling method for rotating machinery in Modelica. To validate our models, we compared both Modelica simulation and experiment with a rotor kit as a test case.
This study concerns the nonlinear piezoelectric impedance modulation method which utilizes the fluctuation of the electromechanical coupled admittance of a piezoelectric element attached on the structure, which is synchronized with the low-frequency vibration of the structure (i.e., pump vibration) due to the strong nonlinearity of contact-type damages. Since the damage index adopted in this method so far assumes a constant pump vibration, it is necessary to find a novel damage index which is independent of the pump amplitude. To this end, the previous study found that the instantaneous admittance curve on the complex plane drastically changed its shape and the location as well. In this study, a SDOF model considering the asymmetric stiffness and the localized damping at the contact surface is proposed. An illustrative analysis of this model shows that the admittance curve drifts on the complex plane with the increase of the pump amplitude.
The author proposed a new identification method of linear single-degree-of-freedom system using Gaussian random vibration response in a previous paper. However, there is a problem that conventional study including the expansion to asynchronous data. In this paper, an expansion to asynchronous data is conducted. At first, propose identification method is expanded to the asynchronous data using numerical differential of a displacement signal. The correlation between the obtained velocity signal and measured displacement signal is not observed, therefore, the independently between two probabilistic variables is confirmed. Moreover, the numerical simulation of identification is conducted. Especially, the identification experiment is conducted in terms of the number of samples and the damping coefficient. In term of dependency of the number of samples, our proposed method is represented the higher estimation accuracy than AR method and Half Power method. In addition, in term of dependency of the damping coefficient, our proposed method is represented the higher estimation accuracy than AR method and Half Power method, too.
An author proposed a new identification method of linear 1-dof system using Gaussian random vibration response in a previous study. However, there is a problem that conventional research containing the expansion to recursive algorithm. In this paper, an expansion to recursive algorithm is conducted. At first, batch algorithm is described. Moreover, based on the above description, recursive algorithm is derived. The algorithm is obtained by expansion of estimation formula of batch algorithm. The numerical identification experiment is conducted in order to confirm the estimation operation of our proposed method. The convergences of estimation values of unknown parameters are observed in case of 1000 samples over. After the convergence of recursive algorithm, the estimation values were good agreement with the true values. Moreover, the dependency of damping coefficient is considered in order to verify the applicability to the large damping system. As the result, the estimation values are good agreement with the true values in the large damping region. Especially, the estimation values of spring constant took the constant values against increase the damping coefficient. Moreover, the estimation values of ratio between diffusion coefficient and the damping coefficient is represented the gradually decreased against increase the damping coefficient.
Geographic Information System (GIS) is expected in order to plan of the optimal inspection path in water pipe network. The problem of inspection path optimization is formulated by Travel Salesman Problem (TSP). In generally, TSP is contained the constraint in order to obtain the proper inspection path. Moreover, the constraint is multiplied Lagrange multiplier. However, there isn’t the effective method in order to obtain the Lagrange multiplier. In this paper, in order to give a guide to the method of determining the Lagrange multiplier for TSP using the annealing optimization. Especially, we propose a method for determining the Lagrange multiplier based on the Hamiltonian decomposition focus on the appropriateness of the solution. First, the appropriateness of the solution is considered. Moreover, the determination method of Lagrange multiplier that focuses on large/small relation of the decomposition Hamiltonian is suggested. Moreover, Lagrange multiplier and the lowest eigenenergy dependence of the whole system are calculated. Lagrange multiplier dependence of the appearance probability of the annealing solution is investigated by numerical calculation.
The authors proposed the identification method of nonlinear spring coefficient using Auto-Regressive time series analysis and Krilov-Bogoliubov-Metropolsky method (KBM method) in a previous paper. Especially, the conventional works had independently deal with the case of second order nonlinearity and the case of third order nonlinearity. However, there is a problem that conventional work didn’t consider the case of second + third order nonlinearity. In this paper, the system identification method in asymmetric nonlinear system included second order and third-order nonlinearity is investigated. At first, formulation of identification problem is conducted. The identification problem is described using the KBM method. Regression problem is formulated using the difference between instantaneous natural angular frequency and linear natural angular frequency. Identification experiment is conducted using numerical investigation based on 4th Runge-Kutta method. Here, second-order nonlinear coefficient is considered in two cases, positive case and negative case, respectively. In both cases, the estimation values of our proposed method are good agreement with the true values.
The author proposed the pipe diagnosis technology using the eigen frequency of the in-plane bending vibration mode in a previous paper. However, there is a problem that conventional study including un-known factor about coupled behavior of a lots cylindrical shell. In this paper, dependency of the number of connections in eigen frequency of the coupled cylindrical shells is considered. Especially, following cases are considered, bi-coupled, tri-coupled, quad-coupled, sex-coupled, oct-coupled, respectively. At first, the vibration experiment is introduced using thin cylindrical shells. Each shell is connected by specified boundary condition that is composed by aluminum tape. Moreover, the frequency response functions are measured using by the exciter and the acceleration sensor. As the result, resonance peak is obtained near the frequency bands of two-dimension ring theory, in case of oct-coupled cylindrical shell. Moreover, the eigen frequencies are good agreement with the theoretical values based on the two-dimensional ring and three-dimensional shell, respectively. In addition, the resonance peak of two-dimension ring is clear obtained by the increasing the number of connections.
The purpose of this study is to propose a new detection method of damage which generates in large structures. The detection method using natural vibration is studied by a lot of researchers, but the environmental disturbance is used to excite natural vibration for inspection of large structure because of difficulty in generating the natural vibration of large structures. In this paper, the concept of the novel detection method using the vibration characteristics of local standing waves. In the case of a large structure of which natural vibration cannot be excited, the local standing wave which is can be observed. The vibration characteristics of the local standing wave may be sensitive to damage which is generated at the location between excitation points. The purpose of this paper is the characteristic investigation of local standing waves. Firstly, the confirmation of local standing wave and measurement of that vibration characteristics is done by experiment using an ideal large beam structure which means the reflection at boundary condition is not generated. The ideal beam structure is made by the local setting of damping material. Thus, the vibration characteristic shift caused by mass variation is confirmed by experiment. As the result, it is possible to detect of damage by which frequency is changed caused by damage.
Reciprocating compressors are used in petrochemical plants, and it has been reported that unexpected damage occurs due to long-term continuous operation, resulting in production loss. When monitoring the state during operation, it is necessary to understand the characteristic frequency generated during operation and the vibration mode. In this study, we experimentally investigated the relationship between the natural vibration characteristics of the system and the vibration mode during operation. In the experiment, we performed a static impact test by changing the crank angle and a dynamic frequency analysis during operation, and compared and examined them. As a result, it was found that the resonance frequency that appears in the operation appears lower than the natural frequency that appears in the static impact test. As for the vibration modes, it was found that vibration modes related to the first order of bending, the first order of bending, and the first order of bending are observed in the vibration modes that appear during operation.
Aiming at the diagnosis of AC motor fault, a recognition method based on symmetrized dot pattern (SDP) images is proposed. The method converts the vibration signal into a visualization image, which better highlights the difference between faults, facilitates the extraction of fault characteristics, and improves the correlation coefficient for fault diagnosis. In this paper, a SDP method is introduced to analyze the vibration signals of the AC motor under normal and four fault states, and the fault diagnosis is realized by the correlation coefficient between the SDP images. The experimental results show that the proposed method can effectively extract the fault characteristics and greatly improve the recognition rate, thus achieving accurate diagnosis of AC motor faults.
Condition-based maintenance(CBM) is maintenance performed on the basis of the condition of the equipment. The transition to CBM being considered, it is necessary to make appropriate decision, such as judgment of normality or abnormality. The threshold approach has generally been used for the judgment, but it provides ambiguous decisions due to individual variability of the equipment. Diagnosis using Data Analysis is a promising method to detect variations of the condition of rotating machinery. In this method, it could be possible to make a judgment of the presence or absence of abnormality without criteria. We apply one of Data Analysis methods called “Topological Data Analysis” to vibration diagnosis of a pump as an example, and show that it classifies the exceptional data as abnormality.
Lubricating oil is required to replace before mechanical failure and the degradation. In automobile, replacement time of engine oil is determined by mileage or operating period. Practically, both degradation and replacement time depend on driving conditions of automobile. Therefore, we have to change it based on driving conditions of automobile. It is effective to diagnose oil degradation using condition monitoring. In this study, we focus on studying the relationship between electrical parameters and indicator of gasoline engine oil used in actual condition. Moreover, we conducted chemical analysis by using ΔERGB and FT-IR to investigate the relationship between electrical parameters and fuel consumption rate. We found that reactance, fuel consumption rate, ΔERGB and FT-IR peak area increased at short driving distance while antioxidant decreased. As antioxidant continued decreasing, reactance, fuel consumption rate, ΔERGB and 1800 - 1650 cm-1 peak area were stable, but 1640 - 1620 cm-1 peak area increased. At long driving distance after antioxidant ran out, reactance, ΔERGB and FT-IR peak area increased again, and fuel consumption rate decreased.
To diagnose the fault of low speed rolling bearing, adaptive feature extraction based on improved Teager energy operator(ITEO), Auto-encoder (AE) and principal component analysis (PCA) is proposed in this paper. The method can extract the useful fault information from the vibration signals and diagnose the faults quickly and accurately. Firstly, the ITEO is adopted to signal denoising and enhanced the transient shock in fault vibration signal. Secondly, auto-encoder is used to adaptive extract the fault information from the enhanced signal. Finally, the diagosis model based on PCA is designed to realize the fault diagnosis. The experiments are performed and the result show that the proposed method is effective in low speed rolling bearing fault diagnosis.
Of the life cycle costs of wind turbines, O & M costs, especially maintenance costs, account for a large proportion and are required to be reduced. In addition, with the increase in the size of wind turbines and the increase in offshore wind turbines, cost reduction of the lubricating oil replacement for the gearbox has become a major issue. Based on the change in color associated with the use of gearbox lubricant, we have developed a technology that extends and optimizes the exchange cycle. For the diagnosis, a low-cost and easy-to-use oil sensor was adopted, and remote monitoring was demonstrated.
Problems of important low speed rotating equipment often occur in plants. Generally, a vibration method is the most commonly used diagnostic method for rotating equipment, but the vibration method cannot be used for low speed rotating equipment. It is said that AE method is the most effective method for low speed rotating equipment, but the signal level depends on surface irregularities of measurement points. Tn this paper research is conducted to suggest a new diagnosis method of “Airborne ultrasound method”.
As a non-stationary signal processing method, the empirical frequency slice wavelet transform (EFSWT) can decompose the signal into a series of frequency slice components (FSCs). The number of FSCs is determined by observation frequency. The observation frequency obtained by the envelope extremum method based on the order statistics filter (OSF) is subject to the trend of spectral fluctuations, so that different frequency components in the signal can be distinguished. The frequency slicing function is used to reconstruct FSCs with better filtering characteristics than finite impulse response filters. Spectral negentropy (SNE) is used to screen the components of the FSCs that contain periodic shock information. The simulation signal and experimental signal verified that the proposed method can be effectively applied to rolling bearing fault diagnosis.
In recent years, the deterioration of infrastructure facilities such as bridges has become a problem. Precautionary measures such as visual inspections and repairs by humans are in place as countermeasures for aging, but there are problems with cost and safety in such inspections. If inspection by robots becomes possible, both will be improved, which will greatly contribute to the maintenance of infrastructure facilities. In this paper, we propose a position specifying system for a robot supporting bridge inspection. In this study, the robot moves under the bridge to support inspection of the bridge. Since the robot must travel on the steel structures at the bottom of the bridge, strong permanent magnets need to be installed on the edge parts in the wheels. Also, it is not possible to receive the coordinate information from the satellite as it travels at the bottom of the bridge. Therefore, a 1-dimensional low-cost LiDAR sensor and a stepping motor are used to implement as a 2-dimensional LiDAR sensor which own use in a 2-dimensional space. The 2-dimensional LiDAR, which rotates 360 degrees, specifies the position of the current robot by obtaining distance and direction data from fixed feature points. The position specification through the 2-dimensional LiDAR are examined by the experiment.
In recent years, aging of bridges has become a problem, and the demand for robots that can inspect bridges on behalf of current bridge inspectors has been increasing. In our study, we developed robots that can move on walls and ceilings by the magnetic wheels attracting the bridge surface. However, in bridge inspection by robots, it is required to carry heavy inspection devices and move the bridge structure freely. Therefore, we have developed robots with high loading capacity to carry heavy inspection devices, which should travel on the three-dimensional complex paths.
From a viewpoint of safe operation and the improvement in an operating ratio of equipment, Efficient maintenance engineering and inspection work is needed. Therefore, we developed a patrol system using 360-degree camera and self-localization technology. A 360-degree camera which can photo the image of all the directions was used for this system. It can record the spherical image of an inspection worker, and enables the inspection image of the direction to need. Moreover, early detection of change of an on-site situation was made possible by difference processing with the past spherical image. In this paper, the prototype under development is reported.
To prevent lighting poles from breakage due to rust and corrosion, nondestructive inspection is necessary. However, defects inside the poles cannot be detected by the visual test. Then the other method, such as ultrasonic test, must be used. In this study, we applied Lamb wave to the flaw detection of the lighting pole． The Lamb wave is suitable for the long-range inspection, because attenuation of the guided wave is smaller compared with the bulk waves. Using electromagnetic acoustic transducers, which do not require surface treatment of the inspection object, flaw detection is performed on a sample prepared from a used lighting pole, and attenuation and diffusion of the Lamb wave are investigated. It was found that total attenuation of the flaw echo is product of those in the forward path and the return path, which means that the drilled hole acts as a point source. We also revealed that the circumferential scan at intervals of 8mm is enough for the transducers used in this study.