Neural network and genetic algorithm-based hybrid approach to expanded job-shop scheduling

The expanded job-shop scheduling problem (EJSSP) is a practical production scheduling problem with processing constraints that are more restrictive and a scheduling objective that is more general than those of the standard job-shop scheduling problem (JSSP). A hybrid approach involving neural networks and genetic algorithm (GA) is presented to solve the problem in this paper. The GA is used for optimization of sequence and a neural network (NN) is used for optimization of operation start times with a fixed sequence.

After detailed analysis of an expanded job shop, new types of neurons are defined to construct a constraint neural network (CNN). The neurons can represent processing restrictions and resolve constraint conflicts. CNN with a gradient search algorithm, gradient CNN in short, is applied to the optimization of operation start times with a fixed processing sequence. It is shown that CNN is a general framework representing scheduling problems and gradient CNN can work in parallel for optimization of operation start times of the expanded job shop.

Combining gradient CNN with a GA for sequence optimization, a hybrid approach is put forward. The approach has been tested by a large number of simulation cases and practical applications. It has been shown that the hybrid approach is powerful for complex EJSSP.     

Parametric optimization of some non-traditional machining processes using artificial bee colony algorithm

Selection of the optimal values of different operating parameters is of utmost importance for enhancing the performance of various non-traditional machining (NTM) processes. The performance measures (responses) of different NTM processes usually include metal removal rate, surface roughness, radial overcut, tool wear rate, heat affected zone, etc. In this paper, artificial bee colony (ABC) algorithm is employed to search out the optimal combinations of different operating parameters for three widely used NTM processes, i.e. electrochemical machining, electrochemical discharge machining and electrochemical micromachining processes. Both the single and multi-objective optimization problems for the considered NTM processes are solved using this algorithm. The results obtained while applying the ABC algorithm for parametric optimization of these three NTM processes are compared with those derived by the past researchers, which prove the applicability and suitability of the ABC algorithm in enhancing the performance measures of the considered NTM processes.     

A hybrid multi-objective approach based on the genetic algorithm and neural network to design an incremental cellular manufacturing system

One important issue related to the implementation of cellular manufacturing systems (CMSs) is to decide whether to convert an existing job shop into a CMS comprehensively in a single run, or in stages incrementally by forming cells one after the other, taking the advantage of the experiences of implementation. This paper presents a new multi-objective nonlinear programming model in a dynamic environment. Furthermore, a novel hybrid multi-objective approach based on the genetic algorithm and artificial neural network is proposed to solve the presented model. From the computational analyses, the proposed algorithm is found much more efficient than the fast non-dominated sorting genetic algorithm (NSGA-II) in generating Pareto optimal fronts.     

进化神经网络在倒立摆控制中的应用

倒立摆作为典型的非线性系统，伴随着多变量、快速运动和绝对不稳定的特征，难于建立精确的数学模型，这就使得对倒立摆的控制变得异常困难和复杂。智能控制理论则是解决此问题的一个有效途径，该文针对倒立摆控制的传统神经网络算法（即BP算法）的缺点，将遗传算法与神经网络结合起来，提出了倒立摆的进化神经网络控制方法。控制器在结构上采用神经网络，利用遗传算法优化神经网络的连接权值。实验研究表明，该控制器不仅具有良好的动态和稳态控制性能，而且对于干扰也具有很强的抑制能力。同时还具备结构简单，易于实现的优点。     

Driver fatigue detection using a genetic algorithm

Nowadays, many traffic accidents occur due to driver fatigue. Driver fatigue detection based on computer vision is one of the most hopeful applications of image recognition technology. There are several factors that reflect driver's fatigue. Many efforts have been made to develop fatigue monitoring, but most of them focus on only a single behavior, a feature of the eyes, or a head motion, or mouth motion, etc. When fatigue monitoring is implemented on a real model, it is difficult to predict the driver fatigue accurately or reliably based only on a single driver behavior. Additionally, the changes in a driver's performance are more complicated and not reliable. In this article, we represent a model that simulates a space in a real car. A web camera as a vision sensor is located to acquire video-images of the driver. Three typical characteristics of driver fatigue are involved, pupil shape, eye blinking frequency, and yawn frequency. As the influences of these characteristics on driver fatigue are quite different from each other, we propose a genetic algorithm (GA)-based neural network (NN) system to fuse these three parameters. We use the GA to determine the structure of the neural network system. Finally, simulation results show that the proposed fatigue monitoring system detects driver fatigue probability more exactly and robustly. This work was presented in part at the 11th International Symposium on Artificial Life and Robotics, Oita, Japan, January 23–25, 2006     

A neural–genetic (NN–GA) approach for optimising mechanisms having joints with clearance

In this paper, a dimensional synthesis method for a four-bar (4R) path generator mechanism having revolute joints with clearance is presented. Joint clearances are considered as virtual massless links. The proposed method uses a neural network (NN) to define the characteristics of joints with clearance with respect to the position of the input link, and a genetic algorithm (GA) to implement the optimization of link parameters using an appropriate objective function based on path and transmission angle errors. Training and testing data sets for network weights are obtained from mechanism simulation, and Grashof’s rule is used during the optimization process as constraint. The results show that the proposed method is very efficient for the purpose of modeling the joint variables and also adjusting the link dimensions to optimize planar mechanisms with clearances.     

Optimizing self-organizing overlay network using evolutionary approach

Self-organizing overlay networks are emerging as next generation networks capable of adapting to the needs of applications at runtime. Applications performance significantly depends on the structure and behaviors of the underlying self-organizing overlay networks. To achieve desired performance, not only the logical overlay topology but also the behaviors of nodes in this overlay network need to be optimized. Moreover, self-organizing overlay networks are extremely dynamic, unreliable and often large-scale. It is therefore important to design new optimizing approaches to meet these challenges. In this paper, we present an evolutionary optimization methodology for self-organizing overlay network. The optimizations of self-organizing overlay networks are modeled as dynamically evolutionary process, in which the nodes interact with each other, change their internal structures and alter their external links to improve the collective performance. To design appropriate fitness functions and rules that guides the direction of the evolution, overlay network can reach a stable state with desired global application performance eventually. Such a methodology leads to our distributed algorithms for proximity-based overlay topology maintenance and Peer-to-Peer living media streaming, in which every node in the overlay network rewires their behaviors and connectivity according to local available information and embedded rules. These algorithms are shown to perform well using simulations.     

补料分批发酵过程优化软件平台的设计

设计了基于DDE的VB、Exccl和Matlab的发酵过程优化软件平台，并将该软件平台应用到多粘菌素发酵过程pH值的寻优。实践表明该优化软件平台能够确定发酵过程被控参数的优化轨线，为优化控制提供一个主要目标。     

Classifying inventory using an artificial neural network approach

This paper presents artificial neural networks (ANNs) for ABC classification of stock keeping units (SKUs) in a pharmaceutical company. Two learning methods were utilized in the ANNs, namely back propagation (BP) and genetic algorithms (GA). The reliability of the models was tested by comparing their classification ability with two data sets (a hold-out sample and an external data set). Furthermore, the ANN models were compared with the multiple discriminate analysis (MDA) technique. The results showed that both ANN models had higher predictive accuracy than MDA. The results also indicate that there was no significant difference between the two learning methods used to develop the ANN.     

Decision tree based light weight intrusion detection using a wrapper approach

The objective of this paper is to construct a lightweight Intrusion Detection System (IDS) aimed at detecting anomalies in networks. The crucial part of building lightweight IDS depends on preprocessing of network data, identifying important features and in the design of efficient learning algorithm that classify normal and anomalous patterns. Therefore in this work, the design of IDS is investigated from these three perspectives. The goals of this paper are (i) removing redundant instances that causes the learning algorithm to be unbiased (ii) identifying suitable subset of features by employing a wrapper based feature selection algorithm (iii) realizing proposed IDS with neurotree to achieve better detection accuracy. The lightweight IDS has been developed by using a wrapper based feature selection algorithm that maximizes the specificity and sensitivity of the IDS as well as by employing a neural ensemble decision tree iterative procedure to evolve optimal features. An extensive experimental evaluation of the proposed approach with a family of six decision tree classifiers namely Decision Stump, C4.5, Naive Baye’s Tree, Random Forest, Random Tree and Representative Tree model to perform the detection of anomalous network pattern has been introduced.     

应用神经网络和遗传算法实现多变量离散系统优化

多变量系统是科学研究和实际应用中经常面对的系统,由于系统的应用环境各异,系统变量多,虽然可以通过试验获得系统的部分输入输出,但是系统的复杂性使得难以确定系统的数学模型,也无法实现对系统的优化。该文提出了一种针对多变量离散系统的建模和优化方法,用BP神经网络对系统的模拟能力建立离散系统的模型,同时为了寻找系统的最优解,结合遗传算法对建立的模型进行寻优,编写了神经网络建模和遗传算法的MATLAB程序,并将建模和优化算法在工艺优化中进行了应用。     

Segmentation of ultrasound images by using a hybrid neural network

A hybrid neural network is presented for the segmentation of ultrasound images.

Feature vectors are formed by the discrete cosine transform of pixel intensities in region of interest (ROI). The elements and the dimension of the feature vectors are determined by considering only two parameters: The amount of ignored coefficients, and the dimension of the ROI.

First-layer-nodes of the proposed hybrid network represent hyperspheres (HSs) in the feature space. Feature space is partitioned by intersecting these HSs to represent the distribution of classes. The locations and radii of the HSs are found by the genetic algorithms.

Restricted Coulomb energy (RCE) network, modified RCE network, multi-layer perceptron and the proposed hybrid neural network are examined comparatively for the segmentation of ultrasound images.     

An integrated neuro-genetic approach incorporating the Taguchi method for product design

Product design is a multidisciplinary activity that requires the integration of concurrent engineering approaches into a design process that secures competitive advantages in product quality. In concurrent engineering, the Taguchi method has demonstrated an efficient design approach for product quality improvement. However, the Taguchi method intuitively uses parameters and levels in measuring the optimum combination of design parameter values, which might not guarantee that the final solution is the most optimal. This work proposes an integrated procedure that involves neural network training and genetic algorithm simulation within the Taguchi quality design process to aid in searching for the optimum solution with more precise design parameter values for improving the product development. The concept of fractals in computer graphics is also considered in the generation of product form alternatives to demonstrate its application in product design. The stages in the general approach of the proposed procedures include: (1) use of the Taguchi experimental design procedure, (2) analysis of the neural network and genetic algorithm process, and (3) generation of design alternatives. An electric fan design is used as an example to describe the development and explore the applicability of the proposed procedures. The results indicate that the proposed procedures could enhance the efficiency of product design efforts by approximately 7.8%. It is also expected that the proposed design procedure will provide designers with a more effective approach to product development.     

Vibration control of load for rotary crane system using neural network with GA-based training

A neuro-controller for vibration control of load in a rotary crane system is proposed involving the rotation about the vertical axis only. As in a nonholonomic system, the vibration control method using a static continuous state feedback cannot stabilize the load swing. It is necessary to design a time-varying feedback controller or a discontinuous feedback controller. We propose a simple three-layered neural network as a controller (NC) with genetic algorithm-based (GA-based) training in order to control load swing suppression for the rotary crane system. The NC is trained by a real-coded GA, which substantially simplifies the design of the controller. It appeared that a control scheme with performance comparable to conventional methods can be obtained by a relatively simple approach. This work was presented in part at the 13th International Symposium on Artificial Life and Robotics, Oita, Japan, January 31–February 2, 2008     

Pruning backpropagation neural networks using modern stochastic optimisation techniques

Approaches combining genetic algorithms and neural networks have received a great deal of attention in recent years. As a result, much work has been reported in two major areas of neural network design: training and topology optimisation. This paper focuses on the key issues associated with the problem of pruning a multilayer perceptron using genetic algorithms and simulated annealing. The study presented considers a number of aspects associated with network training that may alter the behaviour of a stochastic topology optimiser. Enhancements are discussed that can improve topology searches. Simulation results for the two mentioned stochastic optimisation methods applied to non-linear system identification are presented and compared with a simple random search.     

神经网络、遗传算法在发酵优化软件平台设计中的应用

本课题设计了基于DDE的VB、Excel和Matlab的发酵过程优化软件平台，并将该软件平台应用到多粘菌素发酵过程PH值的寻优。实践表明：该优化软件平台能够确定发酵过程被控参数的优化轨线，为优化控制提供一个主要目标。     

用最优化方法解决BP神经网络训练问题

BP神经网络可以有效地对非线性系统进行逼近,但是传统的最速下降搜索方法存在收敛速度慢的问题。本文提出把BP神经网络转化为最优化问题,用一种共轭梯度算法代替最速下降法进行搜索迭代,极大地提高了收敛速度。     

Zigzag machining surface roughness modelling using evolutionary approach

Milling is one of the common machining methods that cannot be abandoned especially for machining of metallic materials. The cutters with appropriate cutting parameters remove material from the workpiece. Surface roughness has the major influence on both obtaining dimensional accuracy and quality of the product. A number of cutter path strategies are employed to obtain the required surface quality. Zigzag machining is one of the mostly appealing cutting processes. Modeling of surface roughness with traditional methods often results in inadequate solutions and can be very costly in terms of the efforts and the time spent. In this research Genetic Programming (GP) has employed to predict a surface roughness model based on the experimental data. The model has produced an accuracy of 86.43%. In order to compare GP performance, Artificial Neural Network (ANN) and Adaptive Neuro Fuzzy Inference System (ANFIS) techniques were utilized. It was seen that the surface roughness model produced by GP not only outperforms but also enables to produce more explicit models than of the other techniques. The effective parameters can easily be investigated based on the appearances in the model and they can be used in prediction of surface roughness in zigzag machining process.     

Neural network approach to acquiring free-gait motion of quadruped robots in obstacle avoidance

In obstacle avoidance by a legged mobile robot, it is not necessary to avoid all of the obstacles by turning only, because it can climb or stride over some of them, depending on the obstacle configuration and the state of the robot, unlike a wheel-type or a crawler-type robot. It is thought that mobility efficiency to a destination is improved by crawling over or striding over obstacles. Moreover, if robots have many legs, like 4-legged or 6-legged types, then the robot's movement range is affected by the order of the swing leg. In this article a neural network (NN) is used to determine the action of a quadruped robot in an obstacle-avoiding situation by using information about the destination, the obstacle configuration, and the robot's self-state. To acquire a free gait in static walking, the order of the swing leg is realized using an alternative NN whose inputs are the amount of movement and the robot's self-state. The design parameters of the former NN are adjusted by a genetic algorithm (GA) off-line. This work was presented in part at the 9th International Symposium on Artificial Life and Robotics, Oita, Japan, January 28–30, 2004