【作者】 杜永良；

【导师】 杨公源；

【作者基本信息】 天津工业大学 ， 模式识别与智能系统， 2005， 硕士

【摘要】 复合肥是提高肥料利用率的有效途经,是化肥工业发展的趋势。复合肥适应作物多种营养成分的需要,且便于使用。专家指出,21世纪的中国化肥工业,为确保粮食供应,将不能再单纯依靠扩大生产能力,而是要通过技术更新、产品开发,生产高效、节能、复合化的肥料。将自动控制系统应用于复合肥生产线,不仅可以提高复合肥生产的产量,同时也提高了复合肥料的质量,且节省人力、物力。 复合肥生产中的控制对象多为流量控制,流量控制为具有纯滞后的一阶惯性。纯滞后对象都难以控制,比较有代表性的纯滞后控制算法为大林算法,但是大林算法要求准确确定被控对象的特性,我们知道工业控制对象中具体对象的准确数学模型是很难建立的,但是在不能够准确知道被控对象的数学模型及相关参数时,用大林算法控制效果极不理想。在不确定对象数学模型的情况下,神经网络控制算法可以达到比较理想的效果,所以对于这种情况,控制系统最终加入了小脑模型神经网络(CMAC)控制算法.同时由于混合反应等原因,一些控制对象受到的干扰信号比较大,采用普通的控制算法系统振荡特别严重,尤其是控制器的输出,因而造成执行机构动作频繁,磨损严重,且不能达到较好的控制效果。为了减小这种现象,不但对所有采集信号进行了滑动平均滤波,而且在混酸、喷浆流量控制器设计中加入了低通滤波环节。并结合PID、CMAC、砰砰等控制算法的优点,采用了多模态控制,达到了较好的效果。在算法的实现过程中,考虑到实现的难易程度等问题,采用了上位机、下位机相结合的控制策略,算法比较复杂的CMAC+P算法在上位机上实现,采用VB编程,并通过DDE协议与监控系统软件进行数据交换。下位机接收上位机的数据,并结合自己的控制算法,根据一定的规则,将其合成为多模态控制,达到了较好的效果。监控系统采用组态王6.5进行组态,主要起参数调整、数据监视功能,是VB应用程序与下位机交换数据的媒介,也是操作者与人机对话的窗口。更多还原

【Abstract】 Compound fertilizer is an effective way in improving the utilized efficiency of the fertilizer, a new trend in the fertilizer industry development. It suits the needs of various nourishments of the farm crop, and easy to use. For ensures the food supplies, the 21 centuries’ China chemical fertilizer industry, the expert points out, couldn’t only depend on expanding production capacity any more, but should through renewing skills, production research and development, and bring out efficiently, save energy, complex fertilizer. Using the automation technical in compound fertilizer production, which can raise the output of production, but also improve its quality, conserve on manpower and material resources. Most of the operating objects of the compound fertilizer are the flow of liquids, which are first order lag. All objects are difficult to be controlled, the typical algorithm is Dahlin control algorithm, but it is need to know accurate mathematical model and peculiarity of the controlled object. It is well know that to establish the accurate model of the industry object is not an easy thing. But Dahlin control algorithm disadvantages in unknown it. We know Artificial Neural Network is especially excellent in this field. So, at this condition, we added Cerebellum Model Articulation Controller to algorithmic aggregation of the controller. At the same time, for the reaction caused by amalgamate and some other reasons, some of signal of the object fall under serious disturbance, if we use the normal algorithm, it would be unsteadiness, especially the output of the Controller, which will caused the machinery swing all the time, and the machinery will serious scathed. Even so, the objects still can’t work in stable ways. To diminish this phenomenon, we not only carry out leach disturbance by slide smoothing algorithm, But also added Low pass Filter to mix acid and spout oar. In addition, we incorporated the advantage of the PID, CMAC and Bang-Bang control algorithm etc, which is so-called Multi-model control. In the process of achieving the operating algorithmic, we apply a complex way of PLC and the IPC in view of the convenience of excogitating the program of the controller. The involved section such as CMAC+P is designed in IPC, which using VB program language. The VB application exchange data with the King view system, which by means of DDE communication protocol. The PLC receiving those parameters and incorporate them with its own data according a certain way. In this way, the controlled system runs steadily. The supervised and control system which function is mainly in adjusting parameter and display of the data is programmed by King view 6.5. It is the medium for the PLC to exchange the data with VB application, and also the human mechanic interface.更多还原