سیستم مدل دانش برای مدیریت تولید گندم

A knowledge model with temporal and spatial characteristics for the quantitative design of a cultural pattern in wheat production, using systems analysis and dynamic modeling techniques, was developed for wheat management, as a decision-making tool in digital farming. The fundamental relationships and algorithms of wheat growth indices and management criteria to cultivars, ecological environments, and production levels were derived from the existing literature and research data to establish a knowledge model system for quantitative wheat management using Visual C++. The system designed a cultural management plan for general management guidelines and crop regulation indices for time-course control criteria during the wheat-growing period. The cultural management plan module included submodels to determine target grain yield and quality, cultivar choice, sowing date, population density, sowing rate, fertilization strategy, and water management, whereas the crop regulation indices module included submodels for suitable development stages, dynamic growth indices, source-sink indices, and nutrient indices. Evaluation of the knowledge model by design studies on the basis of data sets of different eco-sites, cultivars, and soil types indicated a favorable performance of the model system in recommending growth indices and management criteria under diverse conditions. Practical application of the knowledge model system in comparative field experiments produced yield gains of 2.4% to 16.5%. Thus, the presented knowledge model system overcame some of the difficulties of the traditional wheat management patterns and expert systems, and laid a foundation for facilitating the digitization of wheat management.

Development of expert systems and decision support systems has provided new tools for crop management in modern farming (Jones, 1989; Plant and Stone, 1991; Bouman et ul., 1996; Cao, 2000). During the past 20 years, many expert systems for agricultural production management have been developed using knowledge engineering and artificial intelligence (Lemmon, 1986; Jones, 1989; Goodell et al., 1990; Chai et al., 1994; Zhao et al., 1997; Xiong et al., 1999). Wide application of these expert systems to agricultural decision-making has generated social, ecological, and economic benefits (Mc- Kinion et al., 1989; Ma, 2000). The performance of the expert systems in management decision-making relies heavily on the capacity of knowledge rules as a core base of the expert system. In traditional expert systems, the knowledge rules often contain a large number of qualitative and semi-quantitative expert experiences and empirical parameters with site- and time-specific characteristics (Goodell et al., 1990;Plant and Stone, 1991; Chai et al., 1994; Zhao et al., 1997; Xiong et al., 1999). Because of this basic feature, the expert systems are well adapted to local environment and production conditions, but when extrapolated to wider circumstances they perform poorly. This shortcoming has limited the adaptation and accuracy of the traditional expert systems under diverse environmental conditions (Zhu, 2003). Meanwhile, crop growth simulation models, such as CERES-Wheat (Ritchie and Godwin, 1988) and AFRCLVHET (Porter, 1993) for wheat, and CERES-Rice (Singh et al., 1993) and ORYZA2000 (Bouman et al., 2001) for rice, have achieved notable success in knowledge integration and system prediction and have provided a dynamic decision-making tool to support evaluation of varied management strategies in crop production. Through system analysis and mathematical modeling, crop simulation models can quantitatively describe and dynamically predict the processes of crop growth and development in response to various environment and production factors (Bouman et al., 1996; Sinclair and Seligman. 1996: Cao and r\.loss, 1997; Cao and Luo, 2003). These simulation models can only predict the possible outcome of systems operation under a given set of conditions, but cannot directly provide optimum decisions on crop management (Cao and Luo, 2003). In other words, the simulation models seem to be complementary to the expert systems in terms of the prediction function us. decision-making function Thus, if the system modeling methodology was used to integrate and quantify the knowledge base in crop management, this would help to develop a quantitative management model or a modeled expert system for dynamic decision-making in crop production, and would replace the knowledge rules and inference engine in the traditional expert systems with a quantitative niodel. This approach would overcome the difficulties of traditional cultural patterns and expert systems, such as specific site limitation, a massive knowledge base, and low quantification, and lay a foundation for facilitating dynamic decision-making on crop management and for further constructing a new digital farming system by incorporating other subsystems (Liang et al., 2003). Thus the present study aimed to quantify knowledge system for crop production management in wheat, using systems analysis and dynamic modeling. The objectives were to develop a knowledge model with dynamic characteristics for quantitative design of management plans and regulation indices in winter wheat by extracting the general relationships and algorithms of management techniques and growth dynamics to cultivars, ecological environments, and production levels, and then to establish a computerized knowledge model system or a modeled expert system to support winter wheat management with the programming language Visual C++

The present study established a quantitative knowledge model system, with temporal and spatial characteristics, to support dynamic decision-making on cultural management in winter wheat. The knowledge base in wheat management was expressed as a quantitative knowledge model containing the fundamental relationships of wheat cultural techniques and growth indices to cultivars, ecological environments, production levels, and target yields. Testing of the knowledge model system indicated favorable performance of the model system in decision-making and wide applicability. For instance, suitable plant densities and sowing rates were in agreement with the current cultural guidelines in the Tai’an area (Agricultural Department of Shandong Province, 1990; Shandong Agricultural University, 1991; Yu et al., 1999); total nitrogen rates withdifferent soil fertility levels, production levels, and wheat cultivars at two eco-sites were consistent with t,he current. recommendation schemes on the total nitrogen rates (Guo et al., 1994; Wang et al., 2000); the stem numbers recommended for two eco-sites, different yield targets, and wheat cultivars were consistent with practical cultural patterns for current wheat production systems (Agricultural Department of Shandong Province. 1990; Shandong Agricultural University, 1991; Nanjing Agricultural University, 1991; Guo et al., 1994; Diao, 1997; Yu et al., 1999; Ling, 2000). Nevertheless, greater gains could be expected from practical implementation of the system under lower production levels. The results indicated that the knowledge model system was feasible for quantitative decision-making in a wheat management system. This helped to open a new area of digital farming management in modern crop production and information agriculture. The present knowledge model system can be applied in several ways, such as being an independent decision-making package for optimum management, coupled with a crop simulation model for intelligent decision support, or integrated with 3s information technologies applied to agriculture: remote sensing (RS) ~ global positioning system (GPS), and geographical information system (GIS) for precision farming. All these potential uses of the knowledge model system have achieved preliminary success in the laboratory (Zhu et al., 2003; Chen et al., 2004) and remain to be further investigated. The present knowledge model system for quantifying cultural patterns in wheat production integrated the decision-making function of expert systems and the dynamic prediction function of simulation models for constructing a modeled expert system in modern crop production. The knowledge model can be considered as a set of algorithms with temporal and spatial characteristics, for the quantitative design of a cultural management plan for general management guidelines and design of dynamic crop regulation indices for time-course control criteria during a crop growing period. However, it is a modeled expert system for quantitative design, for optimum decision-making on crop management (Cao and Luo, 2003), rather than a simulation model for real-time prediction of crop growth status on a “what-if” cycle from a given management scheme (Bouman et al., 1996; Sinclair and Seligman, 1996; Cao and MOSS, 1997). In principle, the crop management plan designed or generated by the knowledge model system can be used as a cultural management input, for the operation of a crop simulation model, and the dynamic crop growth indices from the knowledge model system can be used as ideal growth curves for guiding crop system regulation (Cao and Luo, 2003; Zhu, 2003). This also implies that the knowledge model system should be evaluated as a type of expert system with emphasis on practical application for dynamic and optimum decision-making in crop management, r’ather than as the simulation model with emphasis on the goodness of fit. (poor, fair, or excellent) between observation and simulation for any specific conditions. However, as the present work is a new effort on creating a quantitative knowledge model or modeled expert system, its definition and methodology are not mature yet and remain to be improved and refined during future research and practice. As compared with t,he traditional expert systems (Goodell et al., 1990; Plant and Stone, 1991; Chai et al.. 1994: Zhao et al., 1997; Xiong et al.? 1999) for crop management, the knowledge system seemed to have three unique charact.erist,ics: favorable applicability, high proficiency, and strong practicability. The dynamic relationships of wheat growth characteristics and management techniques to geographic and seasonal environments were quantified and integrated into a fundamental or general knowledge model, which could essentially be applied for decision-making in wheat production under diverse conditions, including different locations. soil fertilities, cultivars, product,ion levels, and targets of grain yield and quality. This overcame the difficulties of site-specific patterns with traditional crop cultivation schemes and expert systems and introduced wide applicability. Also. the simplified funct’ions and dynamic models in this study replaced the massive knowledge rules in ;t traditional cxpcrt syst,eiii. This rlfwal.rcl tlir rIuiiriva1 description of a kriowlcdgc: base. in previous wheat cultivation schemes and expert systems to the level of an explanatory quantitative model system. Strong practicability was achieved through convenient operation of the current knowledge modelsystem. Users under varied circumstances only need the input data of local climate, cultivars, soils, and production conditions, without having to add and modify any knowledge rules as in a traditional expert system. During the past decades, Chinese crop scientists have achieved great progress and accumulated some unique data in the areas of wheat growth, cultural theory, and management technology, such as in Guo et al. (1994), Yu et al. (1999) and Ling (2000). Yet most of the Chinese data are published only in domestic journals, and thus not well exposed to the international community. The present knowledge model system tried to summarize and quantify some unique databases from Chinese sources, whereas a special effort was made to incorporate relevant international literature. This work should be helpful for integrating the dynamic inter-relationships between eco-environments, cultural techniques, and growth and yield formation characteristics in farm crops, and for quantitative decision-making in a digital farming system (Liang et al., 2003). The present knowledge model system for wheat management, as a first try in agronomic crops, was assumed to be able to provide a general framework for development of a digital knowledge system with other agronomic crops and in other countries. In spite of that, more comparative field experiments should be conducted with the model system-designed patterns and traditional cultural patterns under different eco-sites and production levels for wide testing of the present knowledge model system. This process would continuously improve the reliability and applicability of the model system for recommending cultural patterns in future digital farming. Besides, the model system-designed fertilization strategy was supposed to be a complete recommendation package including basal and top dressing fertilizer. Among these fertilization schemes, the designed top dressing plan was used as a preliminary guideline and would be further adjusted according to actual growth status at a given stage under specific conditions. Therefore, additional studies should be undertaken to integrate the top dressing plan with real-time growth monitoring and diagnosis for precisely regulating in-season fertilizer application during crop growth.