A fuzzy control system for decision-making about fungicide applications against grape downy mildew

A fuzzy control system (FCS) was developed to determine whether a fungicide application is needed to control Plasmopara viticola, the causal agent of downy mildew, in a vineyard. The FCS was conceived as an expert system to be used in connection with vite.net, which is a decision support system (DSS) for sustainable vineyard management. Using the information provided by the DSS, the FCS was able to reproduce the expert reasoning regarding the decision to apply a fungicide against P. viticola in a vineyard. The FCS uses the following information provided by the DSS as input variables: i) grapevine phenology; ii) risk of primary infection; iii) abundance of secondary sporangia; iv) risk of secondary infection; and v) residual protection provided by the last fungicide application. All possible combinations of these inputs are expressed as IF-THEN rules; fuzzification interface, inference engine, and defuzzification interface provide the FCS output as a label: ‘treatment’ or ‘no-treatment’. The FCS was tested by comparing the scheduling of copper fungicides against P. viticola in 18 organic vineyards of Italy as determined by a panel of five experts vs. the FCS. The FCS was able to reproduce the expert reasoning with an overall accuracy of 0.992. The probability that the FCS recommended a treatment given that the expert panel did was 0.878, and the probability that the FCS did not recommend a treatment given that the expert panel did not was 1. Once the FCS is incorporated into the DSS, it will help inexpert viticulturists in taking right decisions about downy mildew control.