一项关于PLC的新技术
被精炼的控制模型。为了使可允许的装置的任何用控制程序的新模版来随时处理事件,我们分离了Fig的最初模型的环10。二环1扫描只适合表盘的结束条件处理并且执行相应控制碎片PB0(i)。而且,在第2环被进入之前,任何条件都可以确定没有给出错误值。扫描剩下的相应控制碎片PB1(i)是被执行过程的表盘的有效前提。环的这新版本全部安全系统包含考试监控,将停止控制器时间超过最大时间限制。这将引起被结合的表盘和控制模型结束。
像上述原因描述一幅控制过程的转移结构的指数的巨型图片的那样非常精炼。对控制全球系统参数减少的这个现象的介绍指定最大的数量事件在分开使成为可能的可能推迟的点时。如果没有使用这个参数,状态空间在几乎全部实验里超过我们的资源,不能 找到最佳的时间表。
寻找我们限制随着与最初表盘有关装满1到7批的有趣原因的我们自己的最佳时刻表。对我们的最初实验来说我们修理最大时间限制到5000次单位。对于每个初始装入来说,我们需要2或者3运行确定最大的批的数量,相应的产品可能被在非常短的时间(按照系统时间次序)生产。事实表明全部相应的产品包含着迅速生产的模式和固定的使用期限。相应的产品都不需要一个削减大于2的价值。最好的尺寸与最短的时期有一定的关系,具体解释如下:
装载:表明表盘是按批的数量装置,
最大时间:表明相应产品的路径的持续(用模拟时间单位),
批:批的数量在那路径里生产,
区域:国家的数量访问生产路径,
步骤:原子步骤的数量执行,
时期:及时的定期行为的时期单位。
A New technology about PLC
The refined control model. To allow the new model of the control program to enable any subset of the permissible plant process events at any time, we split the loop of the original model of Fig.10. The first of the two loops scans only for termination conditions of plant processes and executes the corresponding control fragments PB0 (i). Subsequently, any of the conditions can be set no deterministically to the value false before the second loop is entered. this loop scans the remaining valid preconditions of the plant processes of which the corresponding control fragments PB1(i) are then executed. All guards in both loops of the new version contain tests to monitor the progress time and will stop control if time exceeds maxim. This will cause the combined plant and control model to terminate.
The refined as described above cause an exponential blow-up of the branching structure of the control process, as maximal sets of non-conflicting events are replaced by a choice between all their subsets are reached in this way. to control this phenomenon a global system parameter cuts is introduced that specifies the maximal number of events per branching point that may be postponed when enable. Without the use of this parameter the state space exceeded our resources in practically all experiments.
Finding optimal schedules. Looking for optimal schedules we restricted ourselves to the interesting causes involving initial plant loads of one through seven batches. For our initial experiments we fixed maxim to be 5000 time units (5000s). For each initial load we need two or three runs to determine the maximal number of batches for which counterexamples could be produced in a very short time (in the order of seconds system time). It turned out that all counterexamples produced contained scheduled that rapidly (i.e. within 700 time units) converged to a repeating pattern with a fixed duration. None of the counterexamples required a cuts value greater than 2.
The best measurements in terms of shortest periods detected, the interpretation of the columns is as follows:
Load: indicates the number of batches with which the plant is initializes,
Biggest time: indicates the duration of the counterexample traces,
Batches: the number of batches produced in that trace,
States: the number of states visited to produce the trace,
Steps: the number of atomic steps executed,
Period: period of the periodic behavior in time units.
The Appeal model