دانلود رایگان مقاله انگلیسی الگوریتم ژنتیک برای سنجش تاخیر در سیستم های اتوماسیون شبکه ای به همراه ترجمه فارسی
| عنوان فارسی مقاله: | الگوریتم ژنتیک برای سنجش تاخیر در سیستم های اتوماسیون شبکه ای |
| عنوان انگلیسی مقاله: | Genetic algorithms for delays evaluation in networked automation systems |
| رشته های مرتبط: | مهندسی کامپیوتر و فناوری اطلاعات، شبکه های کامپیوتری، هوش مصنوعی و مهندسی الگوریتم ها و محاسبات |
| فرمت مقالات رایگان | مقالات انگلیسی و ترجمه های فارسی رایگان با فرمت PDF میباشند |
| کیفیت ترجمه | کیفیت ترجمه این مقاله خوب میباشد |
| توضیحات | ترجمه این مقاله در سطح متوسط انجام شده است. |
| نشریه | الزویر – Elsevier |
| کد محصول | f392 |
|
مقاله انگلیسی رایگان (PDF) |
دانلود رایگان مقاله انگلیسی |
|
ترجمه فارسی رایگان (PDF) |
دانلود رایگان ترجمه مقاله |
|
خرید ترجمه با فرمت ورد |
خرید ترجمه مقاله با فرمت ورد |
| جستجوی ترجمه مقالات | جستجوی ترجمه مقالات |
|
بخشی از ترجمه فارسی مقاله: مقدمه |
|
بخشی از مقاله انگلیسی: 1. Introduction Fieldbus systems have been successfully introduced in industrial automation, ensuring real time requirements on one hand and devices safety on the other hand. Nowadays, the trend is also to use the same communication technology at different levels in the industrial organizations; management and automation. A solution that supports such a vertical integration has to be able to provide high throughputs in the upper level as well as small and accurate response times in the field level. The Ethernet solutions, which were initially developed to office networks, can be considered as such a new generation of fieldbuses. Currently, many automation producers and alliances developed their own industrial Ethernet standard (Neumann, 2007). Each solution with a specific protocol is best suited to a particular application. A Client/Server protocol like Modbus over Ethernet, even not adequate for strict real time applications like motion control, is a simple and a reasonable solution for many purposes in industrial control systems. Indeed, it is an application protocol (the 7th layer of the OSI model) that is completely compliant with the standard Ethernet. Therefore, vertical integration is easily achieved. Thus, high level functions like diagnosis and device management are easily implemented. Unfortunately, with such a protocol, no global medium access scheduling is available and different delays due to waiting for resources availability or synchronization are caused. So, the evaluation of its time performances like the response time is complex and the investigations that deal with this problem are rare. The existing methods are often studies of particular systems based on model checking (Greifeneder and Frey, 2007;Witsch et al., 2006; Ruel et al., 2009) that suffers from the classical state explosion problem. Another method is based on high level colored Petri nets simulation (Marsal et al., 2006; Zaitsev, 2004). This method is time driven and very onerous of time. Moreover, it does not provide a formal analysis or proof about its capacity to sweep the worst scenarios corresponding to the worst delays. Finally, an experimental method using a logical network analyzer dedicated to delays measuring is presented in Denis et al. (2007). Hence, the aim of the current paper is to propose an adequate method to assess upper bounds of end-to-end delays of switched packets in the context of Client-Sever automation systems. While a proof about the capacity of the method to assess the worst delays is provided, a genetic algorithm is developed to achieve it much faster. The remainder of this paper is organized as follows. Section 2 introduces the context of our investigation and the motivations to develop a method for end-to-end delays evaluation. Thereafter, two algorithms, to look for the worst delays, are developed in Section 3: an exhaustive algorithm in Section 3.1 and a genetic algorithm in Section 3.2. A case study is then considered to perform a comparison between them in Section 4. Finally, Section 5 addresses some concluding remarks. 2. Client server automation systems over switched Ethernet The studied automation architecture works according to Client/ Server protocol. It is constituted mainly of PLCs (programmable logic controllers), RIOMs (remote input output modules) and a switched Ethernet network that enables communication betweenall the components (Fig. 1). The PLCs (clients) send periodically requests to the RIOMs (servers) and wait for answers. When a RIOM receives a request, it puts it in a queue so as to process all the waiting requests according to a FIFO policy. A request is mainly to get information from the plant (e.g. is the maximal level of water reached?) or to provide orders (e.g. close the valve) or a combination of both. A major criterion of time performance of such a system is the response time Dr (Fig. 1). It is defined as the delay between the occurrence of an event in the plant (e.g. the maximum level is reached) and the arrival of the consequence, issued from a controller, on the controlled plant (closure of the valve). The response time can be defined intuitively as the reactivity delay of the automation system. As a matter of fact, the evaluation of the response time of these systems is tricky. Indeed, different delays due to non synchronization of the components, resources sharing and of course the intrinsic delays (processing) are to be considered. In Addad et al. (2010), we developed an analytic method to evaluate this response time. A formula giving an upper bound of the response time is obtained. Obviously, the delays we call the end-toend network delays (including only the delays experienced in the switches) are involved in the formula. A typical end-to-end delay is the time for a request to cross the switches from its generation by a PLC until its arrival to a RIOM. Therefore, to assess an upper bound of the response time, upper bounds of these end-to-end delays are needed. It could be thought then that existing methods like the well known network calculus (Cruz, 1991; Le Boudec and Thiran, 2004; Georges et al., 2005) or worst case methods (Fan et al., 2008; Lee and Lee, 2002) can be used for this purpose. Unfortunately, in the context of switched packets, combined to Client-Server paradigm, it is not so obvious. Indeed, the flows independency supposed by the previous methods is not verified. For instance, a RIOM does only answer a received request and therefore a request with its corresponding answer cannot exist at the same time in the system. Besides this impeding fact, the formula is very sensitive to the upper bounds of the end-to-end delays. Indeed, a small overestimation of these delays may lead to a huge overestimation of the response time upper bound. As a result, the quality of control in these automation systems are dramatically degraded since the control law synthesis is based mostly on the upper bound assessment (Addad and Amari, 2008). Thereby, an adequate method to evaluate the end-to-end delays in such systems is to be investigated. This is the objective of the next section. |