CAJ | 학술논문

针对冷水机组运行能效模型结构复杂、受运行参数影响较大、机理建模困难等问题，本文建立了基于支持向量回归机的冷水机组运行能效预测模型，并采用粒子群优化算法对模型参数寻优，提高了模型的精度。论文以某商场中央空调离心式冷水机组为研究对象，随机选取396组运行数据对建立的模型进行训练和测试。结果表明，基于粒子群算法优化的冷水机组支持向量回归机模型较BP神经网络模型具有较高的预测精度，其相对误差基本上在3％以内。最后分别采集夏季和过渡季两日的运行数据验证模型的有效性，验证相对误差均在5％以内，因此，该模型能准确地反应冷水机组的运行能效，为冷水机组运行能效分析、故障诊断以及优化控制等提供参考依据。
침대랭수궤조운행능효모형결구복잡、수운행삼수영향교대、궤리건모곤난등문제，본문건립료기우지지향량회귀궤적랭수궤조운행능효예측모형，병채용입자군우화산법대모형삼수심우，제고료모형적정도。논문이모상장중앙공조리심식랭수궤조위연구대상，수궤선취396조운행수거대건립적모형진행훈련화측시。결과표명，기우입자군산법우화적랭수궤조지지향량회귀궤모형교BP신경망락모형구유교고적예측정도，기상대오차기본상재3％이내。최후분별채집하계화과도계량일적운행수거험증모형적유효성，험증상대오차균재5％이내，인차，해모형능준학지반응랭수궤조적운행능효，위랭수궤조운행능효분석、고장진단이급우화공제등제공삼고의거。
Since the difficulty of building mechanism model and the structure of COP model of chiller is complex, greatly affected by op-erating parameter, a COP prediction model of chiller is proposed based on Support Vector Regression, and the parameters are optimized by Particle Swarm Optimization algorithm. In this paper, 396 sets of operating data of chiller of a shopping mall are randomly selected to train and test this model. The results shows that the prediction accuracy of SVR model based on PSO optimization algorithm is higher than that of BP neural network and the relative error is within 3%. At last, operating data of two days in summer and transition season are randomly selected to verify the model. The relative error is within 5%. So this model can provide theoretical basis for the chiller energy efficiency a-nalysis, fault detection and diagnosis and optimizing control.