Search

In recent years, deep-learning detection algorithms based on automatic feature extraction have become the focus of defect detection. However, limited by industrial field conditions, the insufficien...

Due to the complexity and variety of textures on Strip steel, it is very difficult to detect defects on rigid surfaces. This paper proposes a metal surface defect classification method based on an improved bat algorithm to optimize BP neural network. First, this paper uses the Local Binary Pattern(LBP) algorithm to extract features from six types of defect images including inclusion, patches, crazing, pitted, rolled-in, and scratches, and build a feature sample library with the extracted feature values. Then, the WG-BA-BP network is used to classify the defect images with different characteristics. The weighted experience factor added by the network can control the flight speed of the bat according to the number of iterations and the change of the fitness function. And the gamma distribution is added in the process of calculating loudness, which enhances the local searchability. The BP network optimized by this method has higher accuracy. Finally, to verify the effectiveness of the method, this article introduces the five evaluation indicators of accuracy, precision, sensitivity, specificity, and F1 value under the multi-class model. To prove that this algorithm is more feasible and effective compared with other swarm intelligence algorithms. The best prediction performance of WG-BA-BP is 0.010905, and the accuracy rate can reach 0.9737.

Alloy development on a commercial scale in recent years has become much more costly and as such has adopted a more iterative approach, making small refinements while keeping within known limits. Laboratory-based rapid alloy development is therefore a key in pushing new grade boundaries. This work shows the rapid alloy development (RAP) facilities at WMG, including capability, sizes and time required for each stage of the process, which have been mapped against the relevant industrial process. A case study of a dual-phase steel grade DP800 has been used for this work. It has been shown that through knowledge of the strengthening mechanism of the alloy (2nd phase distribution for DP800) is critical with respect to mechanical properties and that specific changes and optimisation to the laboratory production casting and subsequent reduction allows for the final product microstructure and mechanical properties to match the production product. This gives confidence in upscalability.

Thin gauge hot strip steel can be produced by thin slab continuous casting and rolling (TSCR) process without the aid of cold rolling. However, few attempts have been made to produce hot stamping steel by using TSCR technology. As given steel, the hot stamping steel produced by the TSCR process was used to study the influence of austenitizing parameters on the final microstructures and mechanical properties. A Gleeble3500 thermal simulator was employed to heat the samples to obtain different austenite microstructures. After quenching, the Vickers hardness examination along the thickness of the sample and tensile testing were conducted. The microstructure was observed using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The results showed that on the condition of relatively lower temperature and shorter holding time, the sample exhibited better mechanical properties at the cost of microstructure uniformity along the thickness, although higher austenitizing temperature enhanced the carbide dissolution which would increase the effective carbon content in the martensite matrix. Lower heating temperature led to small prior austenite grain size and small blocks which contributed greatly to the mechanical properties.

It is still a challenging task to detect the surface defects of strip steel due to its complex variations, including variable defect types, cluttered background, low contrast, and noise interference. The existing detection methods cannot effectively segment the defect objects from complex background and have poor real-time performance. To address these issues, we propose a novel saliency detection method based on Encoder–Decoder Residual network (EDRNet). In the encoder stage, we use a fully convolutional neural network to extract rich multilevel defect features and fuse the attention mechanism to accelerate the convergence of the model. Then in the decoder stage, we adopt the channels weighted block (CWB) and the residual decoder block (RDB) alternatively to integrate the spatial features of shallower layers and semantic features of deep layers and recover the predicted spatial saliency values step by step. Finally, we design the residual refinement structure with 1D filters (RRS_1D) to further optimize the coarse saliency map. Compared with the existing saliency detection methods, the deeply supervised EDRNet can accurately segment the complete defect objects with well-defined boundary and effectively filter out irrelevant background noise. The extensive experimental results prove that our method is consistently superior to the state-of-the-art methods with large margins and strong robustness, and the detection efficiency is at over 27 fps on a single GPU.

A shear-extrusion bonding (SEB) technology is developed for the bonding of metal plates with the background of endless hot rolling of strip steel. Two key process parameters, deformation ratio and deformation rate, are analyzed by experiments using Q345 steel from the perspective of microstructure and bonding strength. The main findings are: (i) With increasing deformation ratio, the width of bonding interface gradually decreased and the refined grains on both sides of the bonding interface are found. When the deformation ratio is below 75 %, the deformation rate has a significant influence on the closure of holes. The width of bonding interface is smaller at a low deformation rate. However, no obvious difference is found in terms of grain refinement degree. Nevertheless, when the deformation ratio reached 100 %, the grains on both sides of the bonding interface are finer at low deformation rate. When the deformation ratio increased to 150 %, an excellent bonding can be obtained. (ii) In terms of bonding strength, as deformation ratio increases, the plasticity of specimens tends to increase whether at low or high deformation rate. Interestingly, the plasticity property of processed specimens can precede the raw material when the deformation ratio reaches 125 % at low deformation rate. When the deformation ratio reaches 125 % at low deformation rate or 100 % at high deformation rate, the bonding strength can reach the tensile strength of the raw material.

The growing competition in the global steel market in the field of rolled products potentially implies an efficient, economical and environmentally friendly approach to the production of electrical steel. The following key elements of advanced strip steel casting are discussed: innovations in the use of cast strips for the production of the initial hot-rolled billet and cold-rolled steel from electrical anisotropic and isotropic steel, as well as amorphous steel strips for magnetic cores of electrical machines, which have been commercially applied in world practice.

This article discusses the solution to the urgent problem of increasing productivity and reducing the energy consumption of broad strip hot rolling mills (BSHRMs) while improving the quality of rolled products. The technological capabilities of spray cooling and the use of cooling rollers whose cavities are filled with aluminum or copper balls are analyzed. The roller-cooling method is shown to have a higher prospect of solving the formulated problem, because it features good observability of the operational state of equipment, is environmental friendly, allows to improve the quality of rolled products, reduces the need for expensive reagents for cleaning the cooler, reduces energy costs by at least 50%, and allows the use of heat released by hot billets.

This work presents the development of liquid-cooled industry-scale bipolar plates for improved water management in PEM Fuel Cells. The methods used for the design development are based on Computational Fluid Dynamics (CFD) modelling and simulation, and Neutron Radiography experiments to analyse liquid water distributions within the cell for different operating conditions. A novel 140 cm2 bipolar plate was designed and manufactured on 0.1 mm thick stainless steel using pre-coated strip steel. CFD modelling carried out for the novel design predicted a significant improvement in terms of cell performance, as well as a more uniform temperature distribution within the membrane. Liquid water distributions were later analysed by neutron radiography experiments, defining a set of different operating conditions (current density, stoichiometry, inlet gases dew point, and cell temperature). Electrochemical and neutron radiography results are presented for all cases and the influence of the operating conditions is discussed. Liquid water distributions within the cell are also analysed and compared against the CFD model results obtained. The influence of the gas flow configuration (reactant gases and cooling water) is clearly observable in the results.

Impurity elements of Cu, Ni, Sn and Cr have been added to a model low-carbon strip steel to mimic the levels that could arise due to increased scrap recycling during steel production. The microstructure and tensile properties of the base composition and impurity added compositions after thermomechanical processing, consisting of hot rolling and simulated coil cooling, have been investigated. The addition of impurities results in a lower bainite fraction, a smaller effective grain size and the precipitation of Cu-rich particles. Meanwhile, higher strength and lower ductility are found in the steel with impurities.

Physical vapour deposition (PVD) of zinc alloy coatings was investigated as a potential substitute process for commercially available hot dip galvanising (HDG) of strip steel. Therefore, zinc alloy coatings deposited by PVD were systematically compared with traditional sacrificial HDG zinc alloy coatings, in terms of bare metal corrosion resistance and resistance to corrosion-driven delamination of an organic overcoat, to establish the effects of magnesium content, microstructure and surface treatment. The effectiveness of modern corrosion inhibitor pigments, of known volume fraction, on HDG and PVD zinc coatings was also explored. All PVD coatings and commercially available HDG coatings were characterised using microscopy techniques and x-ray diffraction to identify the microstructure and phases present as a function of magnesium content. It was confirmed that the PVD coatings were significantly thinner than the HDG coatings. The pure zinc PVD coating was comprised of hexagonal microplates, whereas the HDG counterpart contained grains 5-10 time larger. The PVD coating containing 4 wt% magnesium exhibited a discrete structure, a binary system of zinc-rich and Mg2Zn11-rich phases, much finer than the HDG Zn-Mg-Al (ZMA) coating. The PVD coatings containing 10 wt% and 20 wt% magnesium were studied using transmission electron microscopy as they possessed nanostructures containing Mg2Zn11 and MgZn2 phases respectively. Open circuit potential (OCP) measurements in chloride-containing solution established that an increase in magnesium content in PVD coating resulted in a decrease in the initial immersion open circuit potential. Additionally, increased magnesium content in the PVD layers also correlated with an increase in corrosion resistance, as made evident by reduced Ecorr and Icorr values during potentiodynamic studies. Electrochemical impedance spectroscopy (EIS) comparative studies suggested an improvement in corrosion resistance exhibited by PVD0 compared to HDG, both zinc-only coatings, attributed to the finer and more compact surface morphology. Bare metal corrosion response for all coatings was studied using a novel augmentation of the scanning vibrating electrode technique (SVET), known as SVET-TLI (time-lapse imaging). The combination of electrochemical mapping and photographic imagery revealed a potential optimum magnesium content within the PVD coatings. PVD4 exhibited the lowest anodic current density over a 24 hour study compared to the HDG, ZMA and other PVD coatings. Furthermore, the characteristic black staining attributed to magnesium corrosion was observed on the magnesium-containing PVD coatings. However, on the PVD Zn-Mg coatings the staining was observed in the regions established as net cathodes, which is contrary to association of staining with magnesium dissolution which takes place in local anodes. Using the scanning vibrating kelvin probe (SKP) method, PVD4 was identified as the optimum magnesium composition as it was found to be resistant of both corrosion-driven cathodic delamination and anodic undermining. Cathodic delamination was observed on the zinc-only coatings, PVD0 and HDG, as well as PVD10 (although at a much slower rate). ZMA and PVD20, both MgZn2-containing systems, showed resistance to cathodic delamination and evidence of anodic undermining. Exploring several modern inhibitive pigments incorporated in the organic overcoat allowed the identification of a commercial pigment “PAM” to provide the greatest improvement in delamination resistance for the zinc-only metallic coatings.

Hot-rolled strip steel is widely used in automotive manufacturing, chemical and home appliance industries, and its surface quality has a great impact on the quality of the final product. In the manufacturing process of strip steel, due to the rolling process and many other reasons, the surface of hot rolled strip steel will inevitably produce slag, scratches and other surface defects. These defects not only affect the quality of the product, but may even lead to broken strips in the subsequent process, seriously affecting the continuation of production. Therefore, it is important to study the surface defects of strip steel and identify the types of defects in strip steel. In this paper, a scheme based on ResNet50 with the addition of FcaNet and Convolutional Block Attention Module (CBAM) is proposed for strip defect classification and validated on the X-SDD strip defect dataset. Our solution achieves a classification accuracy of 94.11%, higher than more than a dozen other compared deep learning models. Moreover, to adress the problem of low accuracy of the algorithm in classifying individual defects, we use ensemble learning to optimize. By integrating the original solution with VGG16 and SqueezeNet, the recognition rate of oxide scale of plate system defects improved by 21.05 percentage points, and the overall defect classification accuracy improved to 94.85%.

The blank is prone to wrinkles and spring-back in multi-point forming, seriously affecting the quality and forming accuracy of the formed part. To improve this state, a sandwich structure constructed from a strip steel pad and blank was presented. The deformation behaviour and forming accuracy using three forming processes, namely, without a cushion, using a polyurethane elastic pad and using a strip steel pad, were analysed by numerical simulation and experimentation. The results showed that the use of the strip steel pad in multi-point forming can effectively increase the deformation resistance of the sheet bending; make the material evenly thinned during forming; suppress defects such as wrinkles, spring-back, dimples and straight edges; and improve the forming accuracy.

Over many decades, the traditional route for material product developments, especially in the steel industry has been the laboratory VIM cast route at scale of 25 to 60kg, followed by through-processing of steel ingots involving hot rolling and cooling as well as further downstream processes to simulate finished cold annealed rolled and coated products. This traditional route has so far delivered value for optimising current grades and process routes as well as developing new products prior to production implementation. However, in order to accelerate process and grade developments even smaller scale and faster laboratory synthesis and processing is desired. The AccMet project [1] developed strategies for new alloy development [2,3] and this needs to be further developed to account for the complex processing route for strip steel production. Strategies combining small scale laboratory alloy processing routes, together with mechanical/thermal testing and modelling are being developed, ranging from 20-30g to 4.5 kg [4-8]. This paper summarises current Rapid Alloy Prototyping (RAP) approaches and rationale developed under a new UK Engineering and Physical Sciences Research Council (EPSRC) Prosperity project between Tata Steel and the Universities of Swansea and Warwick (WMG). Specific attention is paid to the overall experimental methodology as well as benefits (throughput) of small-scale manufacturing and testing, the generation of representative microstructures for a range of strip grades as well as ways of integrating new concepts which bridge the physical length scale. A range of experimental facilities (20-40g) based on a powder route and induction melting (IM)/heat treatments is being developed to provide material for hot/cold rolling/annealing prior to mechanical testing. Modelling and testing to account for mechanical test specimen size effects for small scale RAP samples is being carried out to ensure consistent mechanical properties are obtained. This small-scale RAP is also being complemented with an intermediate material route operating between 200g and 4.5kg using centrifugal casting and small size ingot vacuum induction melting respectively to provide additional material and throughput sitting alongside the more traditional pilot-scale 25-30kg route. Finally, the 25-30kg standard route is being reviewed to provide a bridge to the laboratory routes through various innovative concepts. This paper concludes with a review of future activities and challenges for effective development and implementation of a range of small scale experimental and pilot manufacturing lines.

Surface quality is the most important index to improve the overall quality of strip steel. In order to implement the fault location on the hot-rolling line with surface defects of strip steel, a fault tracing model based on information fusion of historical production cases and process data is proposed. For historical cases, the model determines the defect cause labels through text similarity calculation, and fuzzy semantic inference is used to obtain the probability distribution of defect causes on this basis; for the process data, the model uses L1 regularization method for feature selection, and XGBoost integration method is used to train the correlation model between process data and defects to determine the contribution of each feature in the data source. Finally, based on the D-S evidence theory, different rules are set to merge the two judgments to determine the probability of each source of failure on the hot-rolling production line. The model is applied to the real production environment of iron and steel enterprises, and it is verified that the proposed method can effectively assist experts in decision-making, which greatly improves the efficiency of tracing the source of faults on the hot-rolling production line.

Mechanical properties are important indexes to evaluate the quality of hot rolling strips. It is a research hotspot in the field of hot rolling that realizing timely and accurate soft sensing of mechanical properties. Traditional soft sensing methods have poor performance in the application of strong nonlinearity and multiple working conditions. Moreover, the utilization rate of data is relatively low, which limit the improvement of prediction accuracy. To solve the problems above, a just-in-time learning (JITL) based multi-block weighted semisupervised Gaussian mixture regression (JMWSSGMR) soft sensor is proposed in the paper. There are two stages in the soft sensor: off-line variable blocking and on-line local modeling. In the off-line phase, process variables are divided into different sub-blocks by partial least square (PLS) according to distinct principal component directions. In each sub-block, original variables with high contribution rate are retained. In the on-line phase, optimized Mahalanobis distance is constructed to select the most similar historical samples to the query sample. Next, various real-time semisupervised sub-models are built to estimate the output of the query sample. Finally, predicted values of sub-models are fused and ultimate prediction of mechanical properties is obtained. Case studies are carried out on a numerical example and a hot rolling process. The feasibility and effectiveness of proposed soft sensor are verified by the predicted results.

A quality-related process monitoring method based on parametric t -distributed stochastic neighbour embedding (P- t -SNE) and modified minimum error minimax probability machine (MMEMPM) is proposed for a variety of hot rolling processes. This work pays close attention to the connections between different types of strip steel products instead of conventional multimode methods, which could be useful in process modelling and monitoring. First, a parametric version of t -SNE is developed for streaming data in the hot rolling process. Then, a new space separation method based on parametric t -SNE and quality variables is implemented to extract internal shared information and external unique information among different types for quality-related fault detection. After that, a variety identification method is proposed to identify the online data in the hot rolling process. Finally, the performance of proposed quality-related process monitoring method is examined through a real hot rolling process. The efficiency and feasibility are demonstrated.

Automatic defect detection on strip steel surfaces is a challenging task in computer vision, owing to miscellaneous patterns of defects, disturbance of pseudodefects, and random arrangement of gray-level in background. In this paper, a novel template establishment is presented. Further, a simple guidance template-based algorithm for strip steel surface defect detection is proposed. First, a large number of defect-free images are collected to obtain the statistical characteristic of normal textures. Second, for each given test image, the initial template is built according to the statistical characteristic and the size of test image. Then, a sorting operation is applied to the given test image. Further, by updating the initial template, a unique guidance template is generated based on specific intensity distribution of the sorted test image. So far, the background of each test image is approximately reconstructed in the guidance template. Finally, based on pixel-wise detection, the defects can be located accurately by subtraction operation between the guidance template and sorted test image, reverse sorting operation, and adaptive threshold determination. Experimental results show that the proposed method is both efficient and effective. It achieves a better average detection rate of 96.2% on a data set including 1500 test images.

A novel self-excited oscillating mixer is developed for the descaling of strip steel, which synthesizes the post-mixed abrasive water jet and self-excited oscillation. The realizable k-ɛ model is selected to investigate the effects of different abrasive entrance, inlet pressure, mixing chamber diameter, and length on the jet characteristics. Meanwhile, the effects of different inlet pressure and target distance on the outlet velocities for two kinds of mixers (including the developed mixer and conventional post-mixing mixer) with or without abrasive jet are investigated through simulation. Experiment as well as simulation results exhibited: (1) The oblique abrasive entrance can accelerate the mixture of water and abrasive due to its larger turbulent kinetic energy, and its outlet velocity is larger than that of radial and axial abrasive entrances. (2) For the developed mixer, the outlet velocity is preferable when the mixing chamber diameter is about 40–50 mm and the mixing chamber length is 20 mm. (3) The descaling efficiency of the developed mixer is superior to that of conventional post-mixing mixer. The research will lay foundation to optimize the structure of self-excited oscillating mixer for the descaling of rusty steel.

The surface defects’ region of strip steel is small, and has various defect types and, complex gray structures. There tend to be a large number of false defects and edge light interference, which lead traditional machine vision algorithms to be unable to detect defects for various types of strip steel. Image detection techniques based on deep learning require a large number of images to train a network. However, for a dataset with few samples with category imbalanced defects, common deep learning neural network training tasks cannot be carried out. Based on rapid image preprocessing algorithms (improved gray projection algorithm, ROI image augmentation algorithm) and transfer learning theory, this paper proposes a set of processes for complete strip steel defect detection. These methods achieved surface rapid screening, defect feature extraction, sample dataset’s category balance, data augmentation, defect detection, and classification. Through verification of the mixed dataset, composed of the NEU surface dataset and dataset in this paper, the recognition accuracy of the improved VGG19 network in this paper reached 97.8%. The improved VGG19 network performs slightly better than the baseline VGG19 in six types of defects, but the improved VGG19 performs significantly better in the surface seams defects. The convergence speed and accuracy of the improved VGG19 network were taken into account, and the detection rate was greatly improved with few samples and imbalanced datasets. This paper also has practical value in terms of extending its method of strip steel defect detection to other products.

Strip flatness is an important indicator of steel quality, and its detection technology has always been the focus of research at home and abroad. This paper proposes a non-contact online real-time measurement scheme for hot rolled strip steel. Our research is based on the laser triangulation method, which transforms the flatness detection into the strip surface height detection. In order to prevent errors caused by jitter during strip transportation, we have added a three-point measurement method. The flatness measurement system has carried out the system hardware design with FPGA and linear CCD as the core and the system software design with the laser spot positioning as the core. The basic technology realization process is based on FPGA as the main control and processing chip, collecting the laser reflection light signal of the measured strip steel surface through the optical system through the linear array CCD to obtain the strip surface height information. The collected signals are subjected to photoelectric conversion, filtering and amplification, and A/D conversion, and are transmitted through a network port based on UDP protocol. Finally, the spot center is positioned by comparing the selected threshold method to obtain important data for calculating the flatness information. In this study, a semiconductor laser with a wavelength of 450nm, a TCD1304AP linear CCD and an FPGA chip of the XC7A35T model produced by Xilinx were selected. In the range of 4 mm above and below the reference position, the measurement accuracy of the system can reach 10μm, which meets the measurement requirements of the measured strip elongation accuracy of 1% (when the elongation is 10~200 I) or 1 I (when the elongation is less than 1 I).