To increase the reliability of cycloidal wheel grinding machines, reduce the failure rate of machine tools, and shorten maintenance times, a reliability modeling method for small-sample fault data is proposed based on the Bootstrap-Bayes method. The mean time between failures (MTBF) of a machine tool generally conforms to a Weibull distribution. Based on the historical fault information and similar fault information, the distribution function for the mean time between failures of a machine tool is determined. The fault information is expanded by the self-help method, and the two-parameter distribution interval of the distribution function is calculated by the Bayesian formula. An example reliability calculation for a cycloid gear grinding machine is given, including the failure analysis and maintenance methods of the gear grinding machine. This approach can also be used to simulate and analyze the reliability of other computer numerical control (CNC) machine tools.

Tool wear leads to a reduction in dimensional accuracy and surface quality, as well as unexpected sudden tool failure. A broken tool can cause irreparable damage to an expensive workpiece, resulting in increased operating costs and production delays. Since the mechanical strength of small-diameter drills is inadequate for the load and prone to breakage, tool condition monitoring and diagnosis is important to prevent sudden tool breakage, increase productivity, and promote automation in machining process. The present work is aimed to investigate a tool condition monitoring method based on the analysis of acoustic emission (AE) signals emitted during small-hole drilling. We propose DDM (Deep feature Distribution Modeling), a method for image-level anomaly detection and anomaly segmentation in time-series signal analysis. The peck drilling experiments on SKD61 steels were performed with high-speed steel (HSS) drills. The continuous wavelet transform (CWT) was applied to generate time-frequency (TF) image of the AE signals during the drilling process. The TF images were quantified as anomaly scores using the DDM, which establishes normality by fitting a multivariate Gaussian (MVG) to pre-trained deep features. The anomaly detection capability of the DDM and the convolutional autoencoder (CAE) was compared using dummy data for validation. The digital microscope was employed to measure tool wear. Chip morphology was also observed by the laser microscopy. As the tool wear progressed, the anomaly score increased or decreased, with several sharp increases observed between holes 3805 and 3869 just prior to tool failure. An increase in the width of the shear layer spacing of the chips was also observed just prior to failure. Changes in the anomaly score associated with tool wear were more clearly identified by creating anomaly maps. The present investigation shows that waveform processing of AE signals using the CWT and anomaly detection based on the DDM are efficient methods for tool condition monitoring. Our proposed approach makes it possible to visualize the differences in anomaly states using a more subdivided layer context by generating multiple anomaly maps with deep feature vectors obtained from multiple layers.

In end milling, tool deflection can often be a main cause of machining error. Therefore, in our previous study, a method to simulate the machining error due to elastic deformation of the tool system was proposed because simulation is an effective tool to optimize cutting conditions and machining accuracy. However, the large computational time needed for machining error prediction is unacceptable for practical use. Thus, this study proposes a fast simulation of the machining error for end milling based on our conventional method. To accelerate the simulation and reduce the computational time, the multiple analysis steps that are sequentially calculated in our conventional method are parallelly processed using a graphics processing unit because each step can be treated as an independent phenomenon. In addition, to improve the efficiency of geometric calculation, the workpiece and tool are respectively represented by a dexel model and polygon model that correspond to the swept shape of cutting edges. To verify the validity and effectiveness of the proposed method, a machining error simulation was performed based on both our conventional and proposed methods. It was confirmed that the simulation results based on both methods are in good agreement, and the proposed method can reduce the calculation time significantly.

Efficient aerodynamic shape optimizations are realized in this research utilizing dimension reduction technologies. Several dimension reduction methods such as proper orthogonal decomposition, independent component analysis, kernel principal component regression and deep auto encoder, are investigated to reduce the dimensionality of design variables space. The number of design variables can be efficiently reduced by the proposed approach while obtained optimization results are comparable with that of a conventional optimization approach. The effect of each dominant mode is clarified in this study. A variable fidelity method is introduced by adopting a low-fidelity performance evaluation in the pre-process of the dimension reduction. By introducing the variable fidelity method, a multi objective aerodynamic shape optimization problem can be efficiently solved. Furthermore, design knowledge with respect to the tradeoff relationship between objective functions can be obtained from the results of dimension reduction.

According to the theory of gearing, the mathematical model of a recess action (RA) worm gear with double-depth teeth has been developed in the previous study. In this study, tooth undercutting of the aforementioned worm gear is investigated by applying the theory of gearing, tooth surface equations and the developed computer simulation programs. Limit curves of tooth undercutting can thus be plotted, and the minimum integer numbers of teeth for tooth non-undercutting of the full RA, semi RA and standard proportional tooth worm gears are studied. Besides, tooth undercutting lines of the full RA, semi RA and standard proportional tooth worm gears with double-depth teeth are also investigated.

Topology optimization (TO), which is a design optimization technique that does not require design parameters, has been attracting attention. TO has a high degree of freedom and can obtain a novel design shape suitable for desired purposes. However, the characteristics of the obtained design shape may be greatly damaged due to uncertainty such as variations in material properties. Also, when multi-objective optimization for TO is considered, an increase in the calculation load becomes a major problem. In this study, as a method for efficiently conducting multi-objective robust TO, multi-objective robust solutions can be obtained by using a single-objective robust covariance matrix adaptation evolution strategy and the hybrid method of robust genetic algorithm and multi-objective robust design optimization. The proposed method was applied to design problems of magnetic shields, and obtaining a wide range Pareto front with robustness that cannot be obtained by conventional methods is possible.

The applications of wear resistant material coatings or claddings by directed energy deposition (DED) are rapidly increasing in industrial engineering fields. DED allows us the use of non-expensive substrate materials and add functional materials such as Co-based superalloys on them. However, depending on the deposition parameters, cracks can easily occur on the workpiece. The determination of the appropriate deposition parameters is crucial because it is still difficult to predict and to control cracking. This paper discusses the effect of substrate preheating by an induction heater on the crack generation for DED of Co-based superalloys. An induction heater was adopted for its high efficiency, low waste heat and ability to quickly induce a high temperature in the workpiece directly. The material used in this study is Stellite® 1, which is widely used for die cut rolls or screw shafts. To investigate the effect of the preheating, experiments were conducted by changing the process parameters: preheating temperature and laser power. Sufficient hardness of 642 HV was obtained without cracking when the preheating temperature was 500 ℃ and the laser power was 1750 W. Although cracking did not occur when the preheating temperature or the laser power was higher, the hardness decreased. There is a trade-off between hardness and crack occurrence. This trade-off is considered to be influenced by thermal stress and the precipitation amount of chromium carbides, which are hard and brittle materials. The amount of chromium carbide precipitation is affected by the cooling rate of the deposited material. In addition, a higher cooling rate may result in greater thermal stress.

Although a wide range of machine failure diagnosis methods has been studied for bearings, gears, propellers, etc., there have been few studies on failure diagnosis methods for press machine dies. In this study, we develop a method to diagnose press die failure using a single microphone, which is easy to install, inexpensive, and robust. Because a damped vibration waveform is repeatedly observed in the processing noise of a press, general frequency analysis methods such as the FFT have a problem in that their spectra are blurred and only large amplitude vibrations can be detected from a frequency band comprising multiple vibrations. Therefore, we developed a method to perform dynamic mode decomposition by mapping repetitive vibration waveforms into a two-dimensional array. Based on the obtained eigenmodes and time characteristics, the main vibration components were identified using sparsity and extracted as failure diagnosis indicator values. It was found that the proposed method separates two vibrations that appear as a single peak in the frequency spectrum by the FFT. When this method was applied to the processing noise of abnormal and normal dies, a vibration component at approximately 200 Hz appeared only for an abnormal die. From the eigenvalue analysis of the material using the finite element method, it was realized that a 200-Hz horizontal mode was excited by the force in the direction orthogonal to the pressing direction owing to the die abnormality, resulting in the generation of a 200-Hz abnormal sound. Therefore, by monitoring the sound at the natural frequency of the material’s horizontal mode using the proposed method, die failure can be diagnosed. The proposed method can be applied to vibration analysis of structures and failure diagnosis of bearings, gears, etc., because the frequency, amplitude, and damping ratio of major vibrations can be obtained simultaneously.

Topological modification of gear surface is used to achieve better meshing transmission performance and accuracy. However, in the traditional gear modification grinding process, the topological modified tooth surface is usually simplified to the control of profile modification and lead modification, which is difficult to achieve the coincidence of machining and design. To solve this problem, a flexible topology flank modification method based on polynomial interpolation function is proposed in this paper. Based on the gear meshing principle and polynomial interpolation technology, the method realizes a topological modification of the gear by controlling multiple axes’s position on the machine tool. Firstly, a gear grinding model of worm grinding wheel with controllable grinding precision is established. Then the axial, radial and tangential motions of worm grinding wheel is expressed as fifth order polynomials, and the polynomial coefficients is optimized by particle swarm optimization algorithm. Finally, numerical simulation was used to compare the proposed method with the sensitivity matrix method, and the results showed that the proposed method had better optimization effect. The new flexible topological modification method can realize the topological modification machining by controlling the motion of each axis of the tool, and the problem that the topological modification machining does not match the design is solved.