【作者】 黄超；

【导师】 周新民；

【作者基本信息】 武汉理工大学 ， 控制科学与工程， 2014， 硕士

【摘要】 本文主要针对该码头门座起重机频繁出现的过载现象，对天津港焦炭码头30吨门座起重机控制系统进行改进。通过现场多次观察整个物料抓取过程，发现过载现象是由于操作人员仅凭经验来判断闭斗的时机而导致抓取过多的货物。当起重量超过额定负载时，需将物料全部放下，重新进行抓取。这样重复操作不仅浪费时间，影响整个港口的装卸效率，而且浪费能源。若继续超载运行，会损害机械设备甚至出现安全隐患。为了避免这种现象的发生，本文通过方案比较，最终采用研华USB4711-A数据采集卡对整个作业过程的起升电机以及开闭电机的电流和力矩进行采集，并为数据采集卡设计了信号预处理电路。基于硬件设计的基础，通过编写上位机程序以及数据库技术实现数据实时显示和存储。通过对闭斗过程的数据分析和现场的观察，整个抓取过程分为三个步骤：首先，打开抓斗置于物料之上进行闭斗操作；然后，在抓斗处于半闭合状态时起升抓斗，漏出部分物料；最后，完成整个闭斗过程。通过对整个抓取过程的观察，选取第一步中开闭电机的电流最大值、最小值、平均值以及第二步中起升电机的电流尖峰值作为样本输入，将最后的实际起重量作为样本输出，代入建立好的BP神经网络模型。通过对网络的训练找到特征数据和最终起重量之间的映射关系，能将预测的结果和实际起重量之间的误差控制在令人满意的范围内。在此基础上，以ATMEGA128为核心控制芯片设计一个干预控制系统。通过对整个抓取过程的数据采集，该系统能利用程序筛选出所需要的样本输入数据，并且代入已经训练好的神经网络数学模型，在闭斗之前预测出本次起重量并判断其是否超出额定值，然后再对控制过程加以干预。即在抓取过程中就预判出最终的起重量从而利用程序来控制电机代替人来调整这个过程。这种改进方案的实现，必须建立在大量的数据采集的基础上，所需完成的工作包括数据采集系统的设计、数学建模和控制改进系统的设计。通过本文所设计的系统，理论上能极大程度上的避免门机过载情况的出现，减少了机械设备的损耗，提高港口装卸效率，同时亦能达到节能减排的目的。更多还原

【Abstract】 This paper aims at the situation of overload that frequently happened in gantrycranes in Tianjin port, present the improved scheme to avoid the overload situation.By means of repeated observation at the spot, the cause of the overload was found out.The result reveals that this was caused by the way of operation that the operatorsjudged the timing to close the grab just by their own experience. When the overloadsituation occurred, the operator should lay down the materials and repeat theoperation again. These repeated operations not only waste time and energy, but alsoreduce the efficiency of loading and unloading.To prevent this situation from happening, through the scheme comparison，thisdesign adopt Advantech USB-4711A data acquisition card to read the circuit andtorque signal data of lifting motor and open-close motor during the whole operationprocess, it also designed a signal preprocessing circuit. Moreover, with the uppercomputer program and database technology, it achieves the goal that the real-timedisplay and storage of the signal data. Through the analysis of whole data andobservation of the fetching process, we can divided the whole process into three steps:first of all, open the grab and set it on the materials and close it; then, lift the grabwhen it is half closed so that the materials will leak out; finally, close the grabcompletely. Select maximum current、minimum current、average current of closemotor in first step and peak current of lifting motor in second step as the inputsamples, set the actual lifting weight as the output samples, put them into establishedBP neural network model, through training, we will find out the mapping relationbetween signal data and final lifting weight, and the error between predictive weightand actual weight is satisfactory. On this basis, it is necessary to design an assistantcontrol system which set ATMEGA128as its core control chip. Through the dataaquisition of whole grab process, this system can select input data by program and putthem into the neural network to predict the final lifting weight before the operatorclose the grab. This system can interfere the original control system in case that itforecasts that the gantry crane will be overload. Finally, it can avoid overload totallyby programme rather than operators. The achievement of this modified plan must be based on a large amount of data collection, the required work should include dataacquisition system, data mining model and improved control system.Theoretically, the system designed in this paper, to a great extent, can avoidoverload situation from happening, increase the efficiency and reduce the mechanicalloss. Meanwhile, it can save energy and reduce emission.更多还原