加工过程的复合递归神经网络建模

针对BP神经网络的缺陷,在对角递归网络结构的基础上,提出了一种复合递归神经网络.BP算法收敛速度慢、产生局部极小点的原因之一是该算法采用了均方误差准则,为克服BP算法的不足,采用了一种广义熵方误差准则.把基于广义熵方误差准则的复合递归神经网络应用于加工过程的建模.仿真试验结果表明,复合递归神经网络建模具有比BP神经网络更快的收敛速度和更好的逼近效果.     

人工神经网络在机械加工中的应用

介绍神经网络技术在机械加工领域的应用现状,包括人工神经网络在工艺规程编制中的应用、在加工参数优化中的应用及在工况监测及预报中的应用。并对这项技术的应用作了进一步展望。     

磨料水射流切割质量的神经网络预测

磨料水射流切割质量影响因素较多,难以建立有效的理论模型,结合实验结果,建立磨料水射流切割质量的神经网络预测模型。结果表明,对于所给定切割参数,该模型能快速、准确、可靠地预测出切割质量。     

Parameter estimation for abrasive water jet machining process using neural networks

The abrasive water jet machining process, a material removal process, uses a high velocity jet of water and an abrasive particle mixture. The estimation of appropriate values of the process parameters is an essential step toward an effective process performance. This has led to the development of numerous mathematical and empirical models. However, the complexity of the process confines the use of these models for limited operating conditions; e.g., some of these models are valid for special material combinations while others are based on the selection of only the most critical variables such as pump pressure, traverse rate, abrasive mass flow rate and others that affect the process. Furthermore, these models may not be generalized to other operating conditions. In this respect, a neural network approach has been proposed in this paper. Two neural network approaches, backpropagation and radial basis function networks, are proposed. The results from these two neural network approaches are compared with that from the linear and non-linear regression models. The neural networks provide a better estimation of the parameters for the abrasive water jet machining process.     

Deburring of burrs in spring collets by abrasive flow machining

Micro burrs occurring inside the small and large diameters adversely affect the properties of products. Manual deburring of micro burrs in particular damages the processed surface and reduces production efficiency. In this study, spring collets made of chrome-molybdenum are used to test the deburring of the surface of collets including crossed micro grooves by abrasive flow machining. This revised version was published online in October 2004 with a correction to the issue number.     

利用摩擦学系统理论对磨粒流加工过程的分析

简介当前磨粒流加工技术的现状,得到了磨粒流和工件表面的摩擦以及导致的材料去除是该加工过程的核心问题.利用摩擦学系统理论,构建了针对磨粒流-工件表面的摩擦学系统框架,得到了系统的主要结构元素、元素的特性和相互之间的关系.归纳了磨粒流加工过程中材料去除的摩擦学机理,并讨论摩擦学机理与该摩擦学系统的关系,对从摩擦学角度认识磨...     

Determining dynamically active abrasive particles in the media used in centrifugal force assisted abrasive flow machining process

Abrasive flow machining (AFM) is a relatively new non-traditional process in which a semisolid media consisting of abrasive particles and a flexible polymer carrier is extruded through or across the component to be machine finished. This process is capable of providing excellent surface finishes on a wide range of simple as well as intricated shaped components. Low material removal rate happens to be one major limitation of this process, because during machining not all the abrasive particles participate in removing material from the work piece. Limited efforts have hitherto been directed towards improving the efficiency of the process so as to achieve higher material removal rates. An effort has been made towards the performance improvement of this process by applying centrifugal force on the abrasive media with the use of a rotating centrifugal force generating (CFG) rod introduced in the work piece passage. The modified process is termed as centrifugal force assisted abrasive flow machining (CFAAFM). This paper presents a mathematical model developed to calculate the number of dynamics active abrasive particles participating in the finishing operation in the AFM and CFAAFM process. The analysis of results show that there is great enhancement of number of dynamic active abrasive particles in CFAAFM as compared to the AFM process, which seems to be the contributing factor for the increase in material removal and % improvement in surface roughness for a given number of cycles in CFAAFM. The results of experiments conducted to validate the model show a close agreement between the analytical and experimental results.     

铣削加工误差的神经网络补偿技术

本文对影响铣削加工误差的信息进行分析,运用人工神经网络技术,介绍了在线误差补偿控制系统的建立原理,建立铣削加工误差信息的神经网络结构和误差补偿模型,并通过对样本的合理选择来提高补偿的能力,以提高加工精度。     

Position-oriented process monitoring in freeform abrasive machining

Process monitoring is necessary for the identification and avoidance of process disturbances that could cause poor surface integrity at selected machining parameters. In this paper, a position-oriented process monitoring strategy is introduced which enables determination of process characteristics for freeform abrasive machining. Through correlation of internal machine data of position and power during machining with laser displacement measurement, position-orientated maps of power and specific energy can be generated to enable an evaluation of the machining efficiency of the abrasive machining process. Measurement chains are described, and experimental results reveal that the measurement system provides a significant insight into the process by identifying regions of high power, depth of cut, engagement and specific energy on freeform parts.     

Modeling of TIG welding and abrasive flow machining processes using radial basis function networks

Input-output relationships of tungsten inert gas (TIG) welding and abrasive flow machining (AFM) processes were determined using radial basis function networks (RBFNs). A batch mode of training was adopted to implement the principle of back-propagation (BP) algorithm (which works based on a steepest descent algorithm) and a genetic algorithm (GA), separately. The performances of RBFN tuned by a BP algorithm and that trained by a GA were compared, on some test cases related to the above two manufacturing processes. The GA-optimized RBFN was found to perform slightly better than the BP-tuned RBFN. The back-propagation algorithm works based on the principle of a steepest descent method, whose solutions have the chance of getting stuck at the local minima, whereas the probability of the GA-solutions for being trapped at the local minima is less. However, their performances may depend on the nature of the deviation (error) function.     

基于人工神经网络的超声加工表面粗糙度预测

利用BP神经网络良好的非线性映射能力,建立了普通珩磨和超声珩磨条件下的磨削表面粗糙度预测模型,经过多次网络训练,得到了具有良好性能的BP神经网络。对超声珩磨加工钕铁硼材料表面粗糙度进行了预测,并取得了理想的预测结果。     

Prediction of surface roughness during abrasive flow machining

An analytical model is proposed to simulate and predict the surface roughness for different machining conditions in abrasive flow machining (AFM). The kinematic analysis is used to model the interaction between grain and workpiece. Fundamental AFM parameters, such as the grain size, grain concentration, active grain density, grain spacing, forces on the grain, initial topography, and initial surface finish (R _a value) of the workpiece are used to describe the grain-workpiece interaction. The AFM process is studied under a systematic variation of grain size, grain concentration and extrusion pressure with initial surface finish of the workpiece. Simulation results show that the proposed model gives results that are consistent with experimental results.     

A bicriterian flow shop scheduling using artificial neural network

This paper considers the sequencing of jobs that arrive in a flow shop in different combinations over time. Artificial neural network (ANN) uses its acquired sequencing knowledge in making the future sequencing decisions. The paper focuses on scheduling for a flow shop with ‘m’ machines and ‘n’ jobs. The authors have used the heuristics proposed by Campbell et al.(1970, A heuristic algorithm for n-jobs m-machines sequencing problem) to find a sequence and makespan (MS). Then a pair wise interchange of jobs is made to find the optimal MS and total flow time (TFT). The obtained sequence is used for giving training to the neural network and a matrix called neural network master matrix (NNMM) is constructed, which is the basic knowledge of the neurons obtained after training. From the matrix, interpretations are made to determine the optimum sequence for the jobs that arrive in the future over a period of time. The results obtained by the ANN are compared with a constructive heuristics and an improvement heuristics. The results show that the quality of the measure of performance is better when ANN approach is used than obtained by constructive or improvement heuristics. It is found that the system’s efficiency (i.e., obtaining the optimal MS and TFT) increases with increasing numbers of training exemplars.     

Application of a flexible structure artificial neural network on a servo-hydraulic rotary actuator

In this article the results of the application of a flexible structure artificial neural network for controlling the angular velocity of a servo-hydraulic rotary actuator are discussed. A mathematical model for the system is derived, and a flexible artificial neural network (ANN)-based controller with the feedback error learning method as a learning algorithm is applied to the system. The neural network-based controller has a feed-forward structure and three layers. The flexible bipolar sigmoid function was used as the activation function of the network. The simulation and experimental results show good performance of the developed method in learning the inverse dynamic of the system and controlling the angular velocity of the rotary hydro motor. The advantages of the developed method for servo-hydraulic actuators over other traditional approaches are discussed.     

Optimization of high-pressure die-casting process parameters using artificial neural network

High-pressure die casting is a versatile process for producing engineered metal parts. There are many attributes involved which contribute to the complexity of the process. It is essential for the engineers to optimize the process parameters and improve the surface quality. However, the process parameters are interdependent and in conflict in a complicated way, and optimization of the combination of processes is time-consuming. In this work, an evaluation system for the surface defect of casting has been established to quantify surface defects, and artificial neural network was introduced to generalize the correlation between surface defects and die-casting parameters, such as mold temperature, pouring temperature, and injection velocity. It was found that the trained network has great forecast ability. Furthermore, the trained neural network was employed as an objective function to optimize the processes. The optimal parameters were employed, and the castings with acceptable surface quality were achieved.     

Finishing effect of abrasive flow machining on micro slit fabricated by wire-EDM

This experimental research use the method of abrasive flow machining (AFM) to evaluate the characteristics of various levels of roughness and finishing of the complex shaped micro slits fabricated by wire electrical discharge machining (Wire-EDM). An investigative methodology based on the Taguchi experimental method for the micro slits of biomedicine was developed to determine the parameters of AFM, including abrasive particle size, concentration, extrusion pressure and machining time. The parameters that influenced the machining quality of the micro slits were also analyzed. Furthermore, in the shape precision of the micro slit fabricated by wire-EDM and subsequently fine-finished by AFM was also elucidated using a scanning electron microscope (SEM). The significant machining parameters and the optimal combinations of the machining parameters were identified by ANOVA (analysis of variation) and the S/N (-to-noise) ratio response graph. ANOVA was proposed to obtain the surface finishing and the shape precision in this study.     

模具结构化表面SAF精密光整技术

针对模具结构化表面难以采用传统抛光工具实现精密光整加工的问题,提出了一种基于软性磨粒流(SAF)的模具结构化表面无工具精密加工新技术.该技术通过约束模块与结构化表面组合构成特定形状的磨粒流流道,利用SAF在流道中的湍流流动使磨粒对结构化表面进行微力微量切削,进而实现光整加工.介绍了SAF加工技术原理、SAF流体力学特征...     

Optimisation of the electrical discharge machining process using a GA-based neural network

In this paper, the optimisation of the EDM process parameters from the rough cutting stage to the finish cutting stage has been reported. A trained neural network was used to establish the relationship between the process parameters and machining performance. Genetic algorithms with properly defined objective functions were then adapted to the neural network to determine the optimal process parameters. Examples with specifications intentionally assigned the same values as those recorded in the database or selected arbitrarily have been fed into the developed GA-based neural network in order to verify the optimisation ability throughout the machining process. Accordingly, the optimised results indicate that the GA-based neural network can be successfully used to generate optimal process parameters from the rough cutting stage to the finish cutting stage.