Search

In situ monitoring and accurate detecting of welding quality have been one of the common challenges of automatic welding process. This paper contributes an intelligent decision-making framework for the weld penetration prediction from the keyhole dynamic behavior under time-varying VPPAW pools. Initially, a series of dynamic experiments under different welding conditions were conducted to acquire the backside images of keyhole and corresponding backside bead width. Then, the geometry appearance of keyhole was described by the supervised descent method (SDM)–based image processing algorithm. Subsequently, the internal correlation between the keyhole characteristics and the backside width was further derived to help understand the nonlinear and time-varying VPPAW process. Finally, a novel dynamic model based on an online sequential extreme learning machine (OS-ELM) was designed to predict the weld penetration as measured by the backside bead width in real time. Extensive experiment results further verify and validate that the proposed dynamic OS-ELM model is significantly better than other state-of-the-art algorithms in terms of predicting accuracy, efficiency, and robustness. [ABSTRACT FROM AUTHOR]

The intelligent monitoring of the grinding wheel wear state has the potential to enhance several key aspects of grinding operations, including wheel utilization, wheel dressing, grinding efficiency, grinding quality and so on. In this paper, it is proposed as an acoustic emission signal–based monitoring method of electroplated diamond grinding wheel wear state for C/SiC composite material groove grinding. Firstly, the full-life wear experiment of electroplated grinding wheel grinding C/SiC composites was carried out, and the connection between the acoustic emission signal and the wear state of the grinding wheel was established by frequency domain and time–frequency domain characteristics. Secondly, the time domain, frequency domain and time–frequency domain features of the signals in the stable grinding stage of C/SiC composites were extracted by wavelet packet method. Finally, based on the extracted features, the Extreme Learning Machine (ELM) was optimized by Mayfly Algorithm (MA) to realize online monitoring and intelligent recognition of grinding wheel wear. The results show that the sample classification accuracy of this method is 96.67%, which can effectively identify the different states of grinding wheel wear. [ABSTRACT FROM AUTHOR]

This paper considers the single-machine scheduling problems with learning effect and deteriorating jobs. Release time of jobs is a positive and strictly decreasing function about resource consumption. The optimal algorithm is presented for the problem to minimize the makespan with the total resource consumption constraints. Then, for a given sequence, we propose an optimal solution for the problem to minimize the total resource consumption with the makespan constraints. [ABSTRACT FROM AUTHOR]

In this paper, we consider single-machine scheduling problem with controllable processing times and learning effect, i.e., processing times of jobs are controllable variables with linear costs and also are defined as functions of positions in a schedule. We concentrate on two goals separately, namely minimizing a cost function containing makespan, total completion time, total absolute differences in completion times, and total compression cost and minimizing a cost function containing makespan, total waiting time, total absolute differences in waiting times, and total compression cost. The problem is modeled as an assignment problem and thus can be solved with the well-known algorithms. [ABSTRACT FROM AUTHOR]

The paper presents an algorithm for integrated scheduling, dispatching, and conflict-free routing of jobs and AGVs in FMS environment using a hybrid genetic algorithm. The algorithm generates an integrated schedule and detail routing paths while optimizing makespan, AGV travel time, and penalty cost due to jobs tardiness and delay as a result of conflict avoidance. The multi-objective fitness function use adaptive weight approach to assign weights to each objective for every generation based on objective improvement performance. Fuzzy expert system is used to control genetic operators using the overall population performance improvements of the last two previous generations. Computational experiments was conducted on the developed algorithm coded in Matlab to test the effectiveness of the algorithm. Integrated scheduling of jobs in FMS which are in synchrony with AGV dispatching, scheduling, and routing proved to ensure the feasibility and effectiveness of all the solutions of the integrated constituent elements. [ABSTRACT FROM AUTHOR]

In scheduling problems with learning effects, most research assumes that processing times are deterministic. This paper studies a single-machine scheduling problem with a position-based learning effect and fuzzy processing times where the objective is to minimize the makespan. The position-based learning effect of a job is assumed to be a function of its position. The processing times are considered to be triangular fuzzy numbers. Two different polynomial-time algorithms are developed for the problem. The first solution methodology is based on the fuzzy chance-constrained programming, whereas the second is based on a method to rank fuzzy numbers. Computational experiments are then conducted in order to evaluate the performance of the algorithms. [ABSTRACT FROM AUTHOR]

The directed energy deposition (DED) process is attracting significant attention in high value-added industries, such as automobiles and aviation, because the process can freely manufacture components with complex shapes and directly stack them on metal substrates. However, it has a problem of degradation in reliability and poor reproducibility due to the influence of various parameters present in the process, and various defects are likely to occur inside and outside the product. To solve this problem, a proper data-driven prognostics and health management (PHM) approach is required. Therefore, this study proposes a multi-defect diagnosis algorithm for the DED process based on in situ melt-pool monitoring. First, the DED process monitoring testbed using a CCD camera and a pyrometer was established. The image pre-processing algorithms are developed for the effective extraction of region-of-interest (ROI) areas of the melt-pool and for effective quantification of internal defects, such as pores. Then, critical features of the melt-pool that are closely related to various defects—melting balls, low pores, and high pores—are extracted. Finally, the multi-defect diagnosis algorithm combining several binary classification models is developed, and it is demonstrated that support vector machine (SVM) showed the best performance, with an average accuracy of 92.7%. [ABSTRACT FROM AUTHOR]

In the last years, the use of metamodel-based simulation optimization techniques to solve industrial problems stood out as a promising research field, mainly due to the advance of machine learning techniques. The number of metamodeling studies has grown considerably in recent years, but many academics and practitioners still have doubts about which metamodels to choose for their projects. In this way, some studies have compared the effectiveness of metamodeling algorithms. However, they have just analyzed the performance of one or more metrics separately; i.e., they did not analyze the overall efficiency of these metamodels. Basing the metamodels' choice only on one or more metrics empirically might generate biases, causing distortions in decision-making. Therefore, we propose using the multi-criteria data envelopment analysis (MCDEA) model to systematically compare some of the main machine learning algorithms (support vector machine, artificial neural network, gradient-boosted trees, random forest, and Gaussian process). To evaluate the proposed approach, we developed discrete-event simulation models of three real case studies to obtain their input and output data. Moreover, we used machine learning algorithms to train and optimize the metamodels and, finally, new-MCDEA was adopted to compare the metamodels' efficiency considering the associated error, fitting, training and prediction times, and response, among other metrics. Different from traditional comparison approaches, where different algorithms could be chosen depending on the decision-maker bias, the proposed work allowed a good balance between all metrics, and for all cases, the metamodels based on gradient-boosted trees were considered the most efficient. [ABSTRACT FROM AUTHOR]

Micro-plasma transfer arc metal additive manufacturing (μ-PTAMAM) process is a unique direct energy deposition (DED) type additive manufacturing (AM) process capable of manufacturing three-dimensional metallic components through simultaneous use of wire and powder forms of AM material. DED process yields an uneven surface on the manufactured components which necessitates their further post-processing to attain the required surface finish, dimensional accuracy, and geometrical accuracy. This research focuses on using machine learning algorithm, namely, K-nearest neighbors (KNN), to predict the surface roughness. The data of surface roughness for training KNN algorithm was generated by depositing multi-layer single-track depositions yielding wall-like structures using Stellite-6 as AM material in powder and wire forms. It was found that surface roughness increases with an increase in power supply to micro-plasma and AM material feed rate whereas it decreases with an increase in traverse speed of the deposition head for both powder and wire of the AM material. Surface roughness of the walls for powder form of the AM material is smaller (i.e., 118 to 149 μm) than that obtained by the wire form (i.e., 195 to 227 μm). Surface roughness prediction error for KNN algorithm is found to be from − 6.2 to 2.8% for powder form and − 5.8 to 2.3% for wire form of the AM material proving capability of KNN algorithm for accurate prediction of surface roughness produced by μ-PTAMAM process. Since the prediction error depends on number of data set used to train KNN algorithm, therefore, it can be further reduced by increasing number of training data set. [ABSTRACT FROM AUTHOR]

Piezoelectric ceramics (PZT) actuator has been widely used in flexure-guided micro/nanopositioning stage because of their high resolution. However, it is quite hard to achieve high-rate precision positioning control because of the complex hysteresis nonlinearity effect of PZT actuator. Thus, an online RELM algorithm with forgetting property (FReOSELM) is proposed to handle this issue. Firstly, we adopt regularized extreme learning machine (RELM) to build an intelligent hysteresis model. The training of the algorithm is completed only in one step, which avoids the shortcomings of the traditional hysteresis model based on artificial neural network (ANN) that slow training speed and easy to fall into the local minimum. Then, based on the regularized online sequential extreme learning machine (ReOS-ELM), an online RELM algorithm with forgetting property (FReOS-ELM) is designed, which can avoid the computational load of ReOSELM in the process of adding new data for learning online. In the experiment, a real-time voltage signal with varying frequencies and amplitudes is adopted, and the output displacement data of the micro/nanopositioning stage is also acquired and analyzed. The experimental results show that the RELM-based hysteresis modeling algorithm has higher efficiency and more stable learning ability and generalization ability than the traditional neural network. In the aspect of online modeling, FReOS-ELM hysteresis modeling can achieve a better result than ReOS-ELM. [ABSTRACT FROM AUTHOR]

A robot grasp synthesis algorithm for automated disassembly is presented. The goal is to select grasping points in each part to be disassembled so that a previously planned disassembly sequence can be performed holding the parts firmly and avoiding collisions. The algorithm is structured in five steps in order to make it general enough to cope with different robot grippers and different geometrical data (2D or 3D). The system is learning based, and behaviour rules are automatically extracted from grasping examples given by the user, using mainly decision trees and nearest neighbour techniques. Some simulation experiments have been carried out and results with a two fingered robot gripper are presented. [ABSTRACT FROM AUTHOR]