Search

The safety and reliability of chemical processes are of paramount importance. However, due to the complicated heat, mass, and momentum transfer processes coupled with chemical reactions, and the interrelated effects among various equipment units, it is difficult to accurately detect and identify faults occurring in the chemical system. This study proposes an innovative fault diagnosis framework specifically designed for chemical systems. A method based on orthogonal self-attention convolutional autoencoder (OSCAE) is constructed for fault detection. The convolutional autoencoder is combined with an orthogonal attention mechanism to extract time-series characteristic information of the operational state, thereby achieving precise detection of fault conditions. Additionally, a convolutional neural network (CNN) model uses both raw signals and data reconstructed by OSCAE to identify different faults, which enhances the features of the input and enables highly accurate fault identification. The effectiveness and robustness of the propose method is validated with simulation data of our self-established chemical distillation system and the publicly available Tennessee Eastman process system. The diagnosis efficacy in fault detection and identification is validated by applying fault detection metrics, high-dimensional feature visualizations, and confusion matrix analyses across two case studies. This paper not only provides an effective diagnostic framework for chemical systems but also offers a usable benchmark dataset for chemical distillation systems to the academic community, with the hope that the research content can enhance the stability and reliability of chemical processes.

Particle detection is a key procedure in particle field characterization with digital holography. Due to various background noises, spurious small particles might be generated and real small particles might be lost during particle detection. Therefore, accurate small particle detection remains a challenge in the research of energy and combustion. A deep learning method based on modified fully convolutional networks is proposed to detect small opaque particles (e.g., coal particles) on extended focus images. The model is tested by several experiments and proved to have good small particle detection accuracy.

Comprehension on the primary fragmentation of coal particles is important in understanding and optimizing practical coal combustion and creating credible combustion models. High-speed digital inline holography (DIH) at 6000 Hz is employed to investigate the primary fragmentation of China Ximeng lignite (90–154 μm) during coal devolatilization. The particle morphologies, 3D trajectories, and size evolutions of the parent coals as well as the corresponding fragments are captured by time-resolved holographic visualization. Three fragmentation modes, fragmentation at the particle center, exfoliation/fragmentation at the outer zone, and a hybrid fragmentation mode of both, are observed in this study. The particle morphologies show distinct appearances under the three different fragmentation modes. It is found that the pressure resulted from the volatile products inside particle under the mode of fragmentation at the particle center will accelerate or decelerate the fragments. Statistical result on particle sizes on the recorded 26 coal particles undergoing fragmentation indicates that small particles formed because of coal fragmentation, which occupy 17.1% in the range of 26–40 μm. It is also concluded that the flame temperature and particle residence time during coal devolatilization have an obvious influence on the possibility of coal fragmentation in the operating condition and target research region. Digital inline holography is demonstrated to be capable of in situ three-dimensional (3D) measurement of particle fragmentation during coal devolatilization.

A long working distance digital holographic system is developed to study burning coal particles, which are less than 200 μm in diameter, in a methane-air flame. A general model for the lensed digital holography system is introduced first. An optimal lateral resolution of 3.9 μm is achieved while the object is located at a safe distance (>15 cm) from the optical apparatus. The evolutions of particle morphology and devolatilization products are observed. Sizes and three-dimensional velocities of particles are measured and compared at the burner outlet, 10 and 24 cm above the burner. At the ignition, a condensed phase cloud is observed surrounding the coal particle due to the ejection of devolatilization products. Then the volatile cloud accumulates into stringy tails after detaching from the parent particle. The surface of the coal particle has rough morphology due to formation of porous structure after the reaction. Statistical size distributions at different heights above the burner show a 35% increase of peak size after combustion, and the fraction of the larger particles also significantly increased. Particles spread in the radial direction; meanwhile, particle velocities along the axial direction decrease with the height because of the drag force and gravity.

Dynamic stringy objects such as liquid rims and ligaments are frequently observed in important applications such as the multiphase breakup of fuel droplets. We develop a new method based on digital in-line holography to automatically measure complicated stringy objects. A static spring mounted on a rotator is measured to validate the effectiveness and accuracy of the method. The sections are extracted along the skeleton of the spring in a depth-of-field extended image and then sized and located as individual particles using a hybrid method. The surface points of sections are stitched together to visualize the entire spring. Local thickness errors smaller than 5.3%, and z errors smaller than 230 μm are achieved. This method is applied to characterize the spatial-temporal features of the liquid rim formed in the bag-type regime of the aerodynamic breakup of a falling drop. The evolution of the rim/ligament structures is continuously captured in seven frames, lasting in 1.58 ms. This Letter extends the application of digital holography as an effective 3D diagnostic tool.

Many studies focus on feature extraction and selection of gas sensor arrays for gas identification. In this work, we intended to find a feature subset obtained by selecting the most important features for simultaneously improving component and concentration detection performance of a gas sensor array to three harmful VOCs (toluene, methanol, and ethanol) and their mixtures. First, 30 features were extracted from 6 sensors’ responses to form a multi-feature set. Then, two feature selection methods based on Wilks’ Λ-statistic and random forest were employed to obtain the best feature combination. Seven out of 30 features were finally selected to form the optimal feature set. The gas identification accuracy is 94.3%, and the concentration estimation error is 0.79 ppm (RMSE). Through feature selection, not only qualitative and quantitative analyses performance of VOCs mixtures are significantly improved, but also system complexity (6 to 4 sensors) and computation cost (by about 15%) are effectively reduced.

The effects of the individual scattering process on the formations of both the particle hologram and its corresponding reconstructed three-dimensional particle image are investigated using the Debye series. A particle hologram model using the Debye series decomposes the object wave into different scattering modes and thus permits evaluating the effects of the individual scattering process [diffraction, reflection, transmission, refractions with (p−1) internal reflections] on the particle holography quantitatively. In the Gabor inline holography of a transparent droplet, the transmission light causes small discrepancies between the hologram fringes of an opaque particle (diffraction) and a transparent particle near the zero point of the Bessel-like modulation function, eventually giving rise to the glory spot in the center of the reconstructed dark particle image. For off-axis holography, this paper reveals the effects of reflection, particularly total reflection by bubbles, transmission, and refractions with (p−1) internal reflections of the scattered light on the formation and the reconstructed glory spot images of typical forward and backward off-axis holography.

The 3D measurement of the particles in a gas–solid pipe flow is of great interest, but remains challenging due to curved pipe walls in various engineering applications. Because of the astigmatism induced by the pipe, concentric ellipse fringes in the hologram of spherical particles are observed in the experiments. With a theoretical analysis of the particle holography by an ABCD matrix, the in-focus particle image can be reconstructed by the modified convolution method and fractional Fourier transform. Thereafter, the particle size, 3D position, and velocity are simultaneously measured by digital holographic particle tracking velocimetry (DHPTV). The successful application of DHPTV to the particle size and 3D velocity measurement in a glass pipe’s flow can facilitate its 3D diagnostics.

A digital holographic particle tracking velocimetry system is applied to quantitatively study the drop atomization induced by capillary waves, and the breakup caused by increased sound pressure levels. A wavelet-based algorithm is used for particle detection and autofocusing with a wide size range of 20 μm–2 mm. To eliminate the influence of large particles on small particles, a two-step detection method is adopted. Large drops are first characterized and simulated by a diffraction-based model. Then the contributions of the drops are subtracted from the original hologram followed by the detection of small droplets. Finally, the velocity and size distribution of the secondary droplets are obtained from the experimental holograms. The results demonstrate the validity of the digital in-line holographic technique for the atomization and breakup study of acoustically levitated drops.

We propose a method for simultaneous 3D temperature and velocity measurement of a micro-flow field. The 3D temperature field is characterized with two-color laser-induced fluorescence particles which are tracked with micro-digital holographic particle tracking velocimetry. A diffraction-based model is applied to analyze defocused particles to determine the intensity ratio of two fluorescent dyes on the particle. The model is validated with experimental images. As the result shows that the intensity ratio nearly remains unchanged with respect to depth positions, defocused particles can be used as 3D temperature sensors. Numerical work is carried out to check the method, and 3D temperature and velocity field in a 120??μm×120??μm×80??μm test volume are retrieved.