Stored malting barley: management of quality using an expert system ... Today, the cereal processing channels are mainly oriented towards the satisfaction of food .... deleterious effects of fungi contamination are: i/ - a decrease in germination ...
Stored malting barley: management of quality using an expert system Reims (France), May 28-29, 2001 Ed. INRA, Paris 2004 (Les Colloques n°101)
Predicting quality changes in stored grain ecosystems: feasibility of integrated quality management by an Expert System F. FLEURAT-LESSARD1, D.R. WILKIN2, A. NDIAYE1 1
2
INRA, U.R. Biologie et Technologie Après Récolte (UBTAR), 71, avenue Edouard Bourleaux, B.P. 81, F-33883 Villenave d'Ornon Cedex, France Department of Environmental Science & Technology, Imperial College London, Silwood Park Campus, Buckhurst Road, Ascot, Berks SL5 7PY, UK
ABSTRACT Long term storage of cereal grain prior supply to the processing industries is a critical period for quality preservation. During storage, the interactions between abiotic and biotic variables within the grain bulk can lead to deterioration in quality with limited means of prevention or control. Among the cereals for human consumption, malting barley is considered as the most difficult to maintain quality during long-term storage. The germinative energy may be easily impaired by uncontrolled poor storage conditions. Nevertheless, in theory the grain store manager has at his disposal several means of intervention and types of equipment available to prevent quality losses. Yet, information and advice for optimal combination of these management tools are often lacking at the storekeeper level. This situation often leads to difficulties in preventing quality changes of malting barley in long term storage (more than 6 months) and therefore, as a consequence, supplies of premium grade malting barley after long-term storage.
12 Artificial intelligence tools have become available for the transfer of technical or scientific expertise to the practical level. These “expert systems” are especially suitable for modelling and control of the ecosystem changes and are potentially transposable to the management of the changes in quality characteristics of stored grain bulks. This approach was tentatively applied to the quality management of stored malting barley by getting together the main European Experts on the management of stored grain. Using an existing knowledge-base system based on modular architecture designed in France prior to this Project, this approach was applied to the problem of malting barley quality with the support of the European Commission ("QualiGrain" European Project, ref. FAIR97-CT3648). The Project started in February 1998 and lasted 40 months. RESUME La conservation de longue durée des céréales avant la première transformation est une période délicate pour la préservation de la qualité. Au cours du stockage, les interactions entre variables abiotiques et bio-agresseurs peuvent conduire à une détérioration, difficile à contrecarrer ou à prévenir. Parmi les différentes céréales alimentaires, l'orge de brasserie est considérée comme la céréale la plus difficile à préserver de l'altération en cours de conservation de longue durée. L'énergie germinative est très sensible aux conditions de stockage défavorables et non maîtrisées. Le responsable des stocks de grain dispose de moyens d'intervention et d'équipements qui permettent théoriquement de prévenir les pertes de qualité pendant la conservation. Mais les sources d'information et de conseil pour l'utilisation de ces moyens techniques au moment propice, ainsi que l'aide à la planification des opérations de stockage appropriées sont souvent indisponibles pour les responsables d’organismes stockeurs. Il s'ensuit des pertes économiques liées à la difficulté d'assurer le maintien de la qualité commerciale de l'orge de brasserie sur les longues périodes de stockage (plus de 6 mois), ainsi qu'une absence de garantie d'approvisionnement des malteries avec de l’orge aux spécifications requises, à certaines périodes de l'année. Les progrès récents de l'I.A. ont favorisé la création de systèmes informatisés à base de connaissance, adaptés au raisonnement sur les écosystèmes complexes et évolutifs à la manière des experts humains. Ces systèmes experts sont potentiellement transposables à la résolution des problèmes posés par la préservation de la qualité des stocks de grain. Ils sont capables d’apporter au responsables de stock une aide à la prévision et la prévention des risques pour la qualité des céréales stockées. Cette approche du management général de la qualité des stocks de grain par un système expert avait été étudiée en France, préalablement au présent projet. La transposition de
13 l’architecture de ce premier système expert à base de connaissance à l’orge de brasserie a été le point de départ du projet de recherche Européen "QualiGrain » soutenu par la Commission Européenne. Le projet a débuté en février 1998 et les études ont duré 40 mois. INTRODUCTION - BACKGROUND Today, the cereal processing channels are mainly oriented towards the satisfaction of food industry requirements and have to take into account the implementation of highly sophisticated processing technologies. Storage is at the crossroads of the grain production and the uses (fig. 1). MOLECULAR BIOLOGY OF FUNCTIONAL MACROMOLECULES
GENETIC IMPROVEMENT T SOIL QUALITY
IDENTIFICATION OF TECHNOLOGICAL AND NUTRITIONAL QUALITIES
SEEDS
CLIMATIC INFLUENCES
ELABORATION OF RESERVE (ORGANS) AND MATURATION
CULTIVATION
PLANT PHYSIOLOGY PATHOLOGY
HARVEST
STORAGE AND PRESERVATION
MILLING BRAN
GRINDING STARCH MAKING GLUTEN MAKING
FLOUR
PHYSICAL AND BIOLOGICAL PROCESS ENGINEERING
ACIDS
STARCH
GLUTEN BYGLUCOSE PRODUCT BREAD OF NOODLES STARCH FRUCTOSE SNACKS FERMENTATION
PROTEINS
ETHANOL
ANIMAL FEED
LYSIN
Figure 1. Grain storage (wheat model) is at the crossroads between the production and processing industries, with a primary aim of preservation of the intrinsic quality characteristics.
14 sieving, cleaning
drying
fumigation cooling aeration
Grain treatment for safe storage
% dockage ripening state temperature % breakage during moisture content the harvest Grain initial condition
effects
grain variety and type
qualitative variables
insecticide treatment
Storage time period
intervention means (physical or chemical measures )
storage structure type external Climatic conditions temperature at silo location relative humidity
atmospheric composition
environmental variables
biotic
moulds bacteria birds and rodents
Evolution of variables during storage
variables
Organisms involved in grain deterioration insects and mites
Figure 2. Causal relationships network in the stored grain ecosystem integrating all the causes of quality characteristics changes during storage (Panneton et al., 2001) In this central position along the processing chain, the storage step has to assume two main roles: i/ - the uninterrupted supply of processing industries with raw cereal grain with a specified quality grade, and ii/ - ensuring the preservation of original properties of grain set at the harvest. Ensuring quality preservation is the major objective of any grain store manager and maintaining quality attributes unchanged during storage requires a great level of knowledge. The stored grain bulk can be considered as an ecosystem (Jayas, 1995) with biotic and abiotic variables linked each other by complex relationships (fig. 2 and Panneton et al., 2001). To understand the relationships between all significant variables that determine grain quality is essential for the store manager who faces several complicated tasks. Managers needs to predict future condition of the grain, then select and apply the management actions available to ensure that the quality will fulfil quality requirements of the end-users, and finally to record their management interventions to meet modern quality assurance conditions (Wilkin and Mumford, 1994). The store manager who is in charge of quality maintenance has generally a good practical knowledge but he cannot be competent in all aspects of overall quality management (fig. 3).
15 Relative humidity Energy loss CO2 O2
birds
Displacement of pest populations (behaviour)
birds GRAIN (kernels) microflora Insects
Reserve of pest populations
Crosscontamination by fungi (interaction between species)
Mites
Changes in grain characteristics
rodents
Naturals ennemies
Multiplication process
Figure 3. Main issues encountered by a grain storekeeper in charge of the preservation of the quality of stored grain against the main deterioration agents. STORED GRAIN ECOSYSTEM MANAGEMENT Components of quality and their complex interactions An important dilemma for the storekeeper relates to the clear definition of the grain quality. Generally, his definition of quality is different from the definition of the industrial users; and among these users, the maltsters have specific ideas about quality, which are different from those of the brewers. In addition, the final end-user, the consumer may look for other specific characteristics (e.g. health allegation). In order to satisfy all these different requirements covered by the same word of "Quality", it is possible to divide quality into a number of main components, more or less directly in relation to specific requirements of the end-users and also of the storekeeper. In the model for malting barley for instance, the quality of grain can be built up from four main components: i/ - intrinsic initial condition; ii/ - sanitary and safety condition; iii/ - cleanliness and freedom from impurities; iv/ - suitability for processing (fig. 4).
16
MALTING BARLEY QUALITY
INITIAL CONDITION
Physicochemical Temperature
Moisture content
Sanitary Safety External aspect
Insects Moulds Insecticide residues
Heavy metal traces
Impurities Foreign matters Broken grain Sunflower seeds Other cereals
Quality for MALT processing Germinative energy Protein content Tetrazolium test («vitascope») Variety Varietal purity Kernel size
Figure 4. Logical decomposition of the commercial quality of malting barley grain into its major "technical" components. The intrinsic quality is the sum of initial condition and biochemical composition: protein, starch, lipids, minerals, cellulosic matter, etc. Even if the composition of these constituents is under genetic control in a given variety, the environmental impact during cropping can affect seriously the potential of each variety and consequently, its processing performance. Although the basic knowledge of the relationships between grain composition and quality attributes for a specific end-use are well developed, it is still difficult to relate compositional and biochemical parameters to processing needs, except at an empirical level. The interactions between the content and the properties of each grain constituent is largely unpredictable by the store manager. The extrinsic component of quality is linked to the action of the deteriorative
17 forces that occur during the storage period and these may induce changes in the initial sanitary and safety condition, and secondarily in nutritional potential. Impact and evolution of grain "deteriorative forces" Microorganisms are one of the major causes of deterioration in stored grain. The major deleterious effects of fungi contamination are: i/ - a decrease in germination energy of seeds (especially important for malting barley); ii/ - changes in the appearance of contaminated kernels; iii/ - induction of heating and mustiness in grain; iv/ - biochemical changes, especially with lipid components; and, last but not least, v/ - the production of mycotoxins by some fungal species that can grow and develop at relatively low moisture content (xerotolerant fungi). The mycotoxins are secondary metabolites produced by only a limited number of species. However, even a potentially toxigenic species only produces harmful mycotoxins if there is a genetic predisposition in that strain and there are favourable environmental conditions (Cahagnier, 1998). The tolerance limits for mycotoxins of the storage fungi are very low for the most dangerous molecules such as ochratoxin A and Aflatoxin B1. Nevertheless, these “storage mycotoxins” are only produced during growth of the mycelium and a store manager using good storage practices can prevent any fungal growth on stored grain. Fungi are highly dependant on the activity of water in stored grain for growth and the control and monitoring of moisture, especially at the interface of the grain bulk and the external atmosphere, is of prime importance to prevent mould development. Insects and mites are a hazard to stored grain for various reasons: i/ - The primary feeders develop inside the kernel where they cannot be detected by direct observation; ii/ - the adult stage of all grain-feeding species may migrate over long distances when temperature is above the lower threshold for activity (in the range 12 to 15°C for the most common species); iii/ - when temperature is high, the multiplication rate of the most dangerous species can be as high as x50 in 4 weeks; iv/ - most stored grain beetles reproduce continuously under favourable environmental conditions (the key limiting factor is temperature that enables reproduction and development). The capacity to disperse is prime reason for the success of grain feeding insects to colonise the bulk grain ecosystem (mainly from colonies resident inside the store itself, on handling equipment or in aeration systems on the storage sites). The grain market requirement is officially zero tolerance for live insects. In practice, this limit for infestation can only be applied within the limits of population densities that are detectable by conventional means of grain sampling. This is seldom below one insect per kg of grain and mostly far greater. At the limit of one insect per kg, any hidden infestation that is not detected by conventional means can reach 10 juvenile instars
18 (eggs, larvæ, pupae and pre-emergent adults) per kg. Even when temperature has decreased below the developmental threshold, insects at all developmental stages can survive for long periods without feeding. In these conditions, the natural decrease in bulk grain temperature that occurs during winter in the temperate countries has a limited effect on the survival of the hidden infestation. Development and multiplication start again when grain temperatures increase again to the developmental threshold (in April or May). To date, one of the major problems the store manager faces is the inability of existing technology to detect an infestation at a low density (less than one insect per kg). Consequently, the major method of control of insect pests in stored grain in Europe is the prophylactic use of persistent insecticides leaving residues on the kernels or a fire brigade approach of fumigating after pests have reached a high level. Despite the widespread use of residual pesticides, the effective period of protection and the fate of residues in storage was poorly defined when the QualiGrain Project started in 1998. This situation has completely changed today (Fleurat-Lessard et al., 1998; Fleurat-Lessard, 2001). Intervention tools available to prevent deterioration of stored grain quality The store manager has a range of equipment, tools, materials or techniques that enable him to modify the grain condition when there is a risk of deterioration. Consequently, he has to decide which method to select and apply if deterioration seems likely during storage.
Cooling aeration seed cleaning Controlled atmosphere Residual insecticides Fumigants inert dust
Disinfestation of empty bins Preventative maintenance Chilling ventilation Drying Granary cleaning Grain turning & moving
Figure 5. Storage interventions available and related equipment present in grain stores for the management, control and stabilisation of the changes in quality characteristics.
19 Every storage operation that can be used as a "preventive intervention" when biodeterioration risks are predicted is targeted at the different causes of such biodeterioration (fig. 5): i/ - Drying reduces abnormally high moisture contents and limits the risk of storage fungi spoilage; ii/ - Cooling aeration to decrease grain temperature after the harvest and limit insect development; iii/ - Refrigerated aeration for a quick decrease of temperature in warm climatic locations in order to limit the rate of increase of insects under warm climates; iv/ - Insecticide treatment for the control of insect populations during long-term storage; v/ - Fumigation for the immediate and total killing of all developmental stages of insects; vi/ - Cleaning to decrease the level of impurities that favour biodeterioration (moulds and insects or mites); vii/ - Turning over a grain bulk consisting of moving the grain from one bin to another in order to aerate the grain and mix the different layers for greater homogeneity; viii/ - Airtight storage, under controlled atmosphere or not, provides protection against insect attack and limit the growth of fungal flora; ix/ - Heating for a limited period of time in order to break the dormancy of seeds (malting barley). The choice of appropriate storage operations by the grain store manager is related to the maintenance of a specific component of overall quality. For instance, for malting barley, the main deterioration risk is the decrease of seed viability (equivalent to germinative energy). Mould spoilage and excessively hot storage are the most important causes of deterioration of malting barley germination capacity, consequently, the risk is reduced by drying grain to a moisture content below the limit for mould growth and cooling the grain. Selecting the most appropriate storage operations to retain the maximum quality components is generally a great dilemma for the storekeeper. There are many constraints on the storekeeper who has to take important decisions without the access to specific advice. In a number of situations this can lead to inappropriate or wrong decisions resulting in downgrading of quality and a loss of financial returns. It is easy to turn premium grade cereal into animal feed with inappropriate storage practices. Much of the knowledge required by the grain store manager is fragmented and applied in different ways with varying success (Ndiaye et al., 1997). The use of a computerised knowledge-
20 based decision support system should improve the preservation of stored grain. This preventive approach to stored grain management was at the heart of QualiGrain European Project. A MODERN APPROACH TO GRAIN QUALITY MANAGEMENT Philosophy of a decision support system based on experts knowledge Grain store managers face several tasks that can be made easier or more reliable with the help of computerised decision support systems (Wilkin et al., 1991). Managers need to assess initial quality of a delivery, to load bins with lots of the same grade, to predict future conditions in the store and, then select storage operations to ensure that grain will meet their marketing objectives. Finally they must record their management actions to ensure "traceability". All these tasks must be accomplished at minimum cost. The use of expert systems (ES) for problem solving and decision-making has led to the practical development of integrated expert systems linking management models with ecosystem biodeterioration models. These procedures provide new ways of solving classic problems in ecosystem evolution analysis. Management of the complex biodeterioration processes that can occur within the bulk grain ecosystem provides a very suitable subject for the application of decision support systems (Wilkin and Mumford, 1994; Mann et al., 1997). In response to the increasing requirements for specific qualities of the grain end-users, computer-assisted management is one of the best ways to promote the correct use of up-to-date technologies or even new strategies to overcome the difficulties and issues encountered by the store-keepers (Knight et al., 1999). The storekeeper has to store each grain lot for the highest profit-earning end-use and his choice of a given storage strategy will influence the nature of the potential end-use that can be expected. This initial assumption of potential quality and end-uses is at the basis of the qualitative reasoning approach that has been developed for stored-grain quality management (Ndiaye, 1994 ; Ndiaye et al., 1997). The selected approach uses a modifiable knowledge base and logical inference procedures with a wide scope (Ndiaye and Fleurat-Lessard, 1998). It is based on a four step process: i/ - initial grading at grain batch delivery, taking into account the potential highest grade for optimal end-use and the deterioration risks; ii/ - defining the optimal technical route, i.e. the chronology of the storage operations that have to be implemented and advice for appropriate "treatment" to be applied and the period of application; iii/ - regular monitoring of the quality features and corresponding parameters allowing the early detection of anomalies; iv/ - replanning the storage route when a risk for quality change is increasing (Ndiaye et al., 1997). This
21 is the first attempt to apply the approach of global management of grain quality using a computerised knowledge-based system. Main advantages of expert system management of stored grain quality The building and development of an ES based on knowledge for the management of quality changes and grain deterioration risks represent a new field for artificial intelligence application (Ndiaye and Fleurat-Lessard, 1998). The system that has been built up during the current Project was directly descended from a first knowledge-base system with a modular architecture formed for global quality management of stored cereal grain (Ndiaye, 1994 ; Ndiaye and Fleurat-Lessard, 1994). Thus, the first research works aimed at the development of basic data and models required for the comprehensive use of an ES for total quality management of stored grain. It was conceived with a modular architecture in order to support the decisions of the storekeepers throughout Europe. The system enabled the calculation of the optimal storage technical route increasing the safe storage period. An additional advantage of increased quality at delivery to the grain store was also expected (fig. 6). It may also provide access to new markets
Level high 5
4 3 2 low 1 0 -1
quality grade of each grain delivery
D.S.S. new management
quality grade
money receipts
4
total = 20 + 12 + 16 + 12 + (2x0) = 60
3 (enabling full grading)
2
5 4
1 0
actual situation
8
12
2
delivered quantity
quality grade 4
money receipts
3
total = (29 x 1.5) + (2x0) = 43.5
human decision 2 (accurate grading 1 not possible) 0
29
2
delivered quantity
Figure 6. Economic advantages of full grading of each grain delivery at the grain storage company
22 where very specific qualities were required. The ES should become a tool to move grain store managers into line with the new philosophy of preventative management and control of quality changes. The building of an effective EU-wide decision support system for maintaining the quality of malting barley to process into good malt and beer, was proposed with a very similar approach based on the principle of prevention of quality changes. RATIONALE ABOUT QUALITY ISSUES IN STORAGE AND PRESERVATION OF MALTING BARLEY To date, almost 20 % of malting barley production within the European Union does not meet the commercial standards of malting and brewing industries, especially after long term storage of grain (through into spring and early summer). This downgraded barley represents a great economic loss both for cereal producers and for the storage and handling companies. Much of this loss may be associated with the lack of a common tool providing technically sound support and advice for the safe management of stored grain quality. This may lead to difficulties in the supply to the maltsters or the marketing channels with malting barley with the required quality characteristics. The malting barley industry is of prime importance for the economy and for foreign exchange of the European Union. The future developments depend on the access by any storekeeper to modern tools for overall quality management and control from the harvest to the first processing into malt. The delivery of a high level of expertise on how to solve storage issues at the level of the storekeeper, is now possible through artificial intelligence tools such as expert systems. The European Project "QualiGrain" dealt with the building up of such a decision support system based on exhaustive knowledge afforded by the best experts, scientists and engineers in grain storage practices. The QualiGrain project started in February 1998. METHODS AND ORGANISATION OF THE QUALIGRAIN PROJECT Preliminary considerations An expert system generally is composed from three parts : 1 – A knowledge base including all the existing knowledge about the explored domain in order to compute this knowledge and afford the same level of expertise as recognised experts. This knowledge is available for computation either under the form of facts or as rules in a static configuration.
23 2 – A cognitive system which is the inference engine of the expert system and its dynamic part. It includes reasoning mechanisms, a knowledge acquisition function, and a module for the explanation of its reasoning mechanisms that are automatically carried out. 3 – A man-machine interface allowing the dialog and information exchange between the user and the system. The first expert system (ES) dedicated to the long-term management of the quality of stored cereal grain was conceived at the INRA Stored Product Protection Laboratory (now renamed UBTAR, Unité de Biologie et Technologie Après Récolte) in the years before the delivery of the "QualiGrain Project" proposal to the ECC in 1997 (Ndiaye, 1994 ; Ndiaye and Fleurat-Lessard, 1994; Ndiaye et al., 1997). The originality of the methodology consisted of a knowledge representation that does not need a direct description of the causal relationships between explanatory and observed variables with a definitive link to grain quality criteria changes or biodeterioration process intensity. The first ES prototype was built with a modular architecture enabling the introduction of changes in any of the three parts cited above without influence on the two others. With this partitioning, the knowledge base can be modified without modification of the cognitive system (the inference engine). This enabled the use of the same cognitive system structure for different cereal species with minimal changes and programming. The cognitive system includes reasoning mechanisms and an explanation output detailing the logical sequences carried out by the ES to the user. The man-machine interface (graphical user interface and expert knowledge entering interface) was built in order to take into account the storage equipment and the situation of any user. For the implementation of « QualiGrain » European Project dealing with stored malting barley quality management, each Partner’s research team contributed expertise to one of the multiple topics involved in the "ecophysiology" of stored-grain ecosystems (each grain bin should be considered as an individual ecosystem with its own dynamics and changes in time and in location). Each Partner was in charge of different aspects of the modelling of the physical, chemical and biological changes occurring in a malting barley grain bin with storage time and grain condition. The main variables used in quantitative models should be easily measurable or assessed: grain moisture content or water activity, grain temperature, post harvest storage time, barley variety, and some other characteristics easily observed during quality assessment of grain lots on delivery to the silos. The cognitive engineer that was in charge of the programming of the different parts of the ES developed a logical combination of facts and rules and a dynamic reasoning system simulating the experts reasoning skills on this particular problem. After a
24 comparison of the logical reasoning produced by the expert system with the human experts evaluation on the same situations, the reasoning mechanisms could be validated before their integration into the cognitive system of the ES. Building up the Expert System prototype The general objective of this research project concerned the development of a computerised decision support system (DSS) for the management and control of stored malting barley quality. Four complementary aspects of this scientific and technological challenge have been considered in depth (fig. 7):
Temperature treatments: heat-treatment & cooling
Quality change germination & dormancy
Models compilation
Shell for the ES
Impact of pest dynamics: insects, moulds & mites Insecticide residues Hermetic storage New quality tests
Understanding and modelling quality changes during grain storage
Inference engine Knowledge representation User interface
Acoustical detection Building and programming the ES for quality managemement
Economical optimisations
Validation by human experts
Refining the ES
Validation by the users Validation and adaptation of the ES
Preparation of results for edition & diffusion Promotion of the ES
Figure 7 - Co-ordination of actions and tasks of QualiGrain Project •
At first, the implementation of an exhaustive knowledge base was addressed in
modelling biochemical, microbiological, entomological and ecophysiological models of quality changes during the storage period under the influence of abiotic or biotic variables. It was planned that the base would be completed by experimental work to fill the main gaps identified as the modelling proceeded. This knowledge base enabled the prediction of events and the impact of management inputs such as drying or cooling grain in order to completely inhibit mould growth and prevent consecutive spoilage. •
Second, the programming of the dynamic part of the DSS was undertaken based
on the previous knowledge base content and combining the expertise of scientists and
25 engineers (including their know-how) (Ndiaye, 2001). The expected functions of the DSS were included in the DSS software: i/ - the initial assessment of the quality level of a delivery and its “storage aptitude” (storability), at the delivery of a malting barley load to the silo; ii/ - The proposal of one or several technical strategies for enhancing the safe storage period duration; iii/ - The planning of the use of available equipment at the grain store location in order to minimize the quality changes during long periods of storage; iv/ - The monitoring with remote equipment of indexes and criteria related to malting barley quality parameters changes. The economic costs associated with the choice of a storage strategy were also part of the DSS software. •
Third, the validation of the DSS prototype under laboratory, pilot scale and
commercial scale was conducted in order to check the performance of the DSS compared to the expertise of human experts and to the skills of grain storekeepers or maltsters. •
Fourth, the promotion of the validated DSS to potential beneficiaries through
meetings and demonstration workshops was scheduled at the end of Project works. Deliverables of the Project presented at the final symposium and workshop The first part of the project was devoted to the modelling of the impact of temperature on germination and dormancy upon different varieties of malting barley (see Chapters 3 to 9). The effects of heat treatment (mainly applied during drying) and of cooling have been modelled for predictive assessment of seed viability changes after the treatment and then on a long-term period of storage under controlled conditions of moisture content and temperature (see chapters 3 and 4). Special attention was paid to dormancy break either after a heat shock or after a period of storage in constant conditions (see chapter 5). The impact of storage moulds and insect infestation on malting barley quality for malt processing was demonstrated (see chapters 6 and 7). The decay of residual insecticides and the decrease of their biological efficacy in a large range of temperatures (between 10°C and 35°C) and moisture contents were modelled for organophosphate insectidcides (see chapter 8). The impact of fumigation or long-term storage of malting barley under controlled atmospheres was also highlighted (see chapter 9). The evaluation of available rapid tests for mould spoilage or mycotoxin contamination assessment was carried out (see chapter 10). The correlation of the results of rapid germination tests (one day test) with Carlsberg’s germination index (GI) was also achieved (see chapter 11). Important experimental work was undertaken on a promising new tool for detection of insect infestation at very low population density levels and even of hidden infestations of primary feeders (weevils, lesser grain borer, angoumois grain moth) (see chapter 12).
26 The second part of the Project consisted of the development of the DSS prototype, i.e. programming its different parts: the knowledge base, the inference engine and the graphical user interface (see chapter 13). The DSS software was completed with the addition of an economic cost assessment module and by an exhaustive encyclopædia on malting barley good storage practices giving advice for the prevention or solving storage issues (see chapters 14 and 15). This encyclopædia took into account the quality standards for malting barley grain and malt of the main European countries with important malt and beer production or consumption: Belgium, Denmark, France, Germany, Portugal, Spain and United Kingdom. During the third phase of the project, different levels of validation of the DSS prototype were performed. A first validation by the experts who afforded their knowledge and expertise for the building of the knowledge base and logic-reasoning rules of the cognitive system was carried out and finally allowed to refine the prediction of optimal storage strategy. Two pilot scale storage experiments were achieved in Denmark and in UK in order to detect the differences between DSS-driven management and observed quality changes (see chapters 16 and 17). Finally, a test of the prototype in its final configuration was made in a large commercial grain store (see chapter 18). At the final restitution meeting (where all the results obtained during this Project were presented), a demonstration workshop was organised and the reactions of the users were recorded (see chapter 19). From all these results, the future of the use of such a decision support system either by grain store or malting plant managers was evoked as a conclusion of this “restitution symposium” of the results obtained in the QualiGrain EU Project (see chapter 20). REFERENCES CAHAGNIER B., 1998. Moisissures des aliments peu hydratés. Lavoisier Tec & Doc, Paris (France), 226 p. FLEURAT-LESSARD F., VIDAL M.-L., BUDZINSKI H., 1998. Modelling biological efficacy decrease and rate of degradation of chlorpyrifos-methyl on wheat stored under controlled conditions. J. Stored Prod. Res., 34:341-354. FLEURAT-LESSARD F., 2002. Qualitative reasoning and integrated management of the quality of stored grain: a promising new approach. J. Stored Prod. Res., 38:191-218. FLEURAT-LESSARD F., 2001. Devenir des résidus d'insecticides au cours du stockage et à la transformation: des risques à gérer avec précision. Industries des Céréales, 121:5-17. JAYAS D., 1995. Mathematical modeling of heat, moisture, and gas transfer in stored-grain ecosystems. In: D.S. Jayas, N.D.G. White, W.E. Muir (Ed.) Stored-Grain Ecosystems. Marcel Dekker, Inc., New York, USA, pp. 527-567. KNIGHT J.D., ARMITAGE D.M., WILKIN D.R., 1999. Evolution of a decision support system with changing markets. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Ed.),
27 Proc. 7th Int. Working Conf. on Stored Product Protection. Beijing, P.R. China, 14-19 October 1998, II, pp. 1914-1918. MANN D.D., JAYAS D.S., WHITE N.D.G., MUIR W.E., EVANS M.S., 1997. A grain storage information system for Canadian farmers and grain storage managers. Can. Agric. Eng., 39:49-56. NDIAYE A., 1994. Le (futur) système de pilotage des stocks de grain par un système expert informatisé. In: Comptes-rendus de la journée technique du 02-02-1994 : Les progrès dans la préservtion des stocks de graines (et leur passage dans le pratique). GLCG Association édit., La Rochelle, pp. 7992. NDIAYE A., FLEURAT-LESSARD F., 1994. Research on an expert system for appropriate management of the quality of stored grain for food and feed processing. In: Xie Guifang and Ma Zongdeng (Ed.) Proceedings "94 International Symposium and Exhibition on new approaches in the production of foodstuffs and intermediate products from cereal grains and oilseeds. Beijing, 16-19 November 1994, pp. 537-540. NDIAYE A., FLEURAT-LESSARD F., 1998. Diagnosis and grading of wheat initial quality by a computerised decision support system. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Ed.), Proc. 7th Int. Working Conf. on Stored Product Protection. Beijing, P.R. China, 14-19 October 1998, II, pp. 1923-1934. NDIAYE A., PERON L., FLEURAT-LESSARD F., 1997. Diagnosis and grading of grain initial quality. In: Munack A., Tantau H.-J. (Ed.) Proc. IFAC Conf. on Mathematical and Control Applications in Agriculture and Horticulture. Hannover, Germany, 28 Sept.-2 Oct. 1997, pp. 219-224. NDIAYE A., 2001. Registered Copyright "QualiS Expert System Software ": 2001-07-20 PANNETON B., VINCENT C., FLEURAT-LESSARD F., 2001. Plant protection and physical control methods – The need to protect crop plants. In: Vincent C., Panneton B., Fleurat-Lessard F. (Ed.) Physical control methods in plant protection. Springer-INRA Editions, Berlin, Germany, pp. 9-32. WILKIN D.R., MUMFORD J.D., 1994. Decision support system for integrated management of stored commodities. In: Highley E., Wright E.J., Banks H.J., Champ B.R. (Ed.) Proc. 6th Int. Working Conf. on Stored Product Protection. Canberra, Australia, 17-23 April 1994, II, pp. 879-883. WILKIN D.R., MUMFORD J.D., NORTON G.A., 1991. An expert system for stored grain management. In: Fleurat-Lessard F., Ducom P. (Ed.) Proc. 5th Int. Working Conf. on Stored Product Protection. Bordeaux, France, 9-14 Sept. 1990, III, pp. 2039-2046.
