PERTINENCE CURVES IN FUZZY MODELING OF THE PRODUCTIVE RESPONSES OF BROILERS

ABSTRACT The selection of the type of fuzzy systems pertinence curve allows a better representation of the mathematical model and a smaller simulation error. We aimed to study the effect of pertinence curves in fuzzy modeling of broiler performance, created in different production systems. For the development and testing of fuzzy models, three commercial aviaries (conventional, tunnel with negative pressure, and dark house) were evaluated over one year, totaling six lots per system. For the development of the model, the input variables were enthalpy in each rearing phase (initial: phases 1, 2, and 3; growth: phase 4; and final: phase 5) and the output variables were feed intake (FE), weight gain (GP), feed conversion (FE), and the productive efficiency index (PEI). Triangular, trapezoidal, and Gaussian pertinence curves were combined and applied to represent the input and output fuzzy sets, totaling nine fuzzy models for each output variable. The combinations of pertinence curves provided adequate responses for the prediction of AL, GP, RC, and PEI. However, the selection of the types of curves should be studied on a case-by-case basis, so that the smallest possible simulation errors are obtained.摘要 选取适配的模糊系统（fuzzy systems）隶属度曲线（pertinence curve），可更精准地构建数学模型并降低模拟误差。本研究旨在探究不同生产系统中，隶属度曲线对肉鸡生产性能模糊建模的影响。为构建与测试模糊模型，本研究对3类商品肉鸡舍（常规舍、负压通风隧道式鸡舍、密闭遮光鸡舍）开展了为期1年的监测，每个养殖系统共计6批肉鸡。模型构建所用的输入变量为各饲养阶段的焓值：初始阶段（第1、2、3期）、生长阶段（第4期）及终末阶段（第5期）；输出变量分别为采食量（FE）、增重（GP）、饲料转化率（FE）与生产效率指数（PEI）。将三角形、梯形与高斯型隶属度曲线进行组合，用于表征输入与输出模糊集，最终针对每个输出变量共构建9种模糊模型。上述隶属度曲线组合可为采食量（AL, feed intake）、增重（GP）、饲料转化率（RC, feed conversion）与生产效率指数（PEI）的预测提供适配的响应结果。不过，隶属度曲线的选取仍需结合具体案例开展针对性研究，以尽可能获取最小的模拟误差。