水產(chǎn)養(yǎng)殖系統(tǒng)外文翻譯--專家系統(tǒng)在水產(chǎn)養(yǎng)殖監(jiān)測(cè)和控制方面中的應(yīng)用（英文）

1、Application of an expert system to monitoring and control in aquaculture D D Harris, Feng Zhang* and P H Sydenham When any measurement task is approached from the point of view of the knowledge which is needed, it is f

2、ound that this knowledge often comprises a range of data, interpreted in terms of a set of rules. In the current explanatory application, measurement and control of water quality is fundamental to the breeding of heal

3、thy fish in aquaculture installations. It is the rule base, fed with real-time measured data, which is of primary importance in ensuring the health of the fish. Computer based expert systems simplify the applica- tion

4、 of rules and heuristics to real-time monitoring and control. This paper discusses the development of a computer-based, distributed monitoring and control system built around the “Crystal“ expert system. The system e

5、nables both heuristic and real-time varying knowledge to be integrated into the rule base on which control decisions are automatically made. An important feature is that, by directly accessing the expert system's r

6、ule builder, the application domain user can change the configuration and operating rules of the system using only low-level computing skills. By also building 'sensor knowledge“ into the expert system, the senso

7、r design and operating requirements are simplified, allowing the inexpert user to specify the needed sensors, or build them to drawings supplied by the system. Keywords: expert systems, monitoring and control, knowled

8、ge base, user interaction The Measurement and Instrumentation Systems Centre (MISC) is a research centre of the University of South Australia with the goal of developing better methods and equipment for measurement. T

9、he under- lying basis of all programmes is the use of knowledge- Measurement and Instrumentation Systems Centre, University of South Australia, PO Box 1, Ingle Farm, SA 5098, Australia *Shandong Academy of Agricultur

10、al Sciences, Jinan, Shandong, China Paper received 6 February 1990. Revised paper received 30 May 1991 based methodology. The Centre has recently under- taken a project with the South Australian Department of Fisher

11、ies to apply this expertise to the monitor- ing and control of aquaculture installations, with par- ticular application to the breeding of marine fish species. BACKGROUND In order to appreciate the knowledge-based met

12、hodol- ogy represented here, it is necessary to develop a basic understanding of the application. Current aquaculture practice Throughout the world, fish are farmed in large quan- tities in installations ranging from

13、open sea pens through to controlled ponds on land. Both freshwater and marine species are farmed, including fin fish, crustaceans and molluscs. However a number of common threads emerge in the literature on the subj

14、ect: ? The capital costs of aquaculture installations are often high. ? The viability of aquaculture operations is very sen- sitive to labour and operating costs. ? The methods of conventional industrial process and cont

15、rol engineering are not simple to apply because of the differences between typical industrial 'pro- cesses' and aquaculture situations. ? Knowledge of the physiology and behaviour of fish, especially in intensi

16、ve farming situations, is generally limited and localized. ? The slow timescale of fish growth leads to difficulties in experimentation and in modelling the growth process and hence in its monitoring and controP. ? Fis

17、h are very sensitive to some environmental parameters and lack of adequate water quality con- trol can be catastrophic in a very short time. ? There is a strong tendency towards the use of computers because of their pot

18、ential for handling Vol 4 No 3 September 1991 0950-7051/91/030165-07 © 1991 Butterworth-Heinemann Ltd 165 range of instruments available, at all levels of price and quality, but problems may still be identified:

19、first, while there is copious technical sales literature on the capa- bility of instruments from specific suppliers, there is little expertise or help available to guide users in the selection and specification of th

20、e instruments which are needed to provide the correct knowledge required for a specific installation. Second, as reported by Cobb 9, there is a lack of standardization in communication systems. There is no standard wa

21、y to connect instruments into a complete monitoring network, especially if instruments from a range of suppliers are used. This problem is much wider than aquaculture and standards will soon emerge for manufacturing

22、industry. How applicable these will be to aquaculture remains to be seen. Third, the cost of instruments is often high, and the cost and ready availability of instrument consumables and spare parts is also a problem. K

23、NOWLEDGE-BASED APPROACH Key concepts The fish farmer requires knowledge from the environ- ment in which the fish are living and this knowledge can be applied either by the farmer or automatically by the system to modi

24、fy and control the environment. In addressing the problems outlined above, we have used a number of concepts considering the aquaculture operation as a 'knowledge' system. These key concepts include rule-base

25、d sensing, rules, user accessibility and sensor design knowledge, as discussed below. Rule-based sensing Knowledge is always context-dependent. The signifi- cance of a particular sensor reading is usually depend- ent

26、upon other factors, and its interrelation with other sensor readings. Context, together with the structure, or 'rules' of the task, provide a model of the measured system. As an expert system provides a conven

27、ient method for manipulation of such a rule-based model, it would appear to have a place in monitoring and control. However, applications appear to be rare and more often serve as an operator guide than as a full cont

28、rol system. A typical example is described by McNamara 1°. A sensor which measures the temperature of the water in an aquaculture pond provides simply a number. Leaving aside considerations of just what the numbe

29、r represents (it is normally simply an electronic signal), just what is the significance of this reading in terms of the health of the fish in the pond? Other factors such as fish species, dissolved oxygen content an

30、d time of day are all relevant. It is important to note that, contrary to established practice, the precision of the reading is often of less importance than its inter- relation with other influencing factors in arriv

31、ing at knowledge about the health of the fish. Barber u makes the point that knowledge can be thought of as a transforming operator, mapping infor- mation (evidence) to information (conclusions). The factors relating

32、to fish health represent a model which holds the knowledge required to allow the measure- Information in “-Proces s~~ Model of being [ process Action t ,-q tr°l~ 1 Knowledge~ i~ ] ~' L ..Y Actua

33、tors Figure 1. Knowledge mapping sensed data to action in a measurement and control system ments to be transformed into action. This concept is illustrated in Figure 1. A person setting up an instrumentation system usu

34、- ally does so within a model, or rule structure, of the process but most often implicitly and without being aware of the fact. Nevertheless, while it is relatively easy to obtain highly detailed, technical informatio

35、n on the measuring instruments it is difficult to obtain a succinct statement of the operating rules of a fish farm. The former have tended to be given more importance than the latter due to their better definition a

36、s numbers. Houvenaghe112 describes the development of such a model and highlights some practical difficulties in doing SO. In this work the monitoring and control system has been approached as a rule-based decision-mak

37、ing system rather than as a collection of measurements which must then be ordered into a decision tree. Expert systems software provides a good base for this kind of development. Decisions are clearly made by humans

38、 by combining measured data with heuristic knowledge. This system emulates that process as an information gathering and processing system. Rules There are essentially three types of rules which must be applied to the

39、monitoring system: ? Specific, technical rules which do not change and which carry expertise which the fish farmer does not have. An example is the method of interrogation by the controller of the sensors in the networ

40、k. While these are implemented as 'rules' within the expert system they are essentially hard coded through an interface function. ? More general fish breeding rules which reflect the current (highly changing)

41、state of knowledge of the subject. An example of such a rule from this test installation is that abalone (molluscs) grow best in a temperature range of 16.5 to 18.5°C. It is worth noting that this rule was not k