制冷与空调（四川）
製冷與空調（四川）
제랭여공조（사천）
REFRIGERATION & AIR-CONDITION
2015年
3期
345-349
,共5页
离心式冷水机组%运行能效模型%支持向量回归机
離心式冷水機組%運行能效模型%支持嚮量迴歸機
리심식랭수궤조%운행능효모형%지지향량회귀궤
centrifugal chiller%COP model%support vector regression
建立离心式冷水机组运行能效模型对其运行能效分析以及优化控制意义重大。离心式冷水机组结构复杂且其运行能效受多种因素的影响，机理建模困难，而支持向量回归机能够较好的解决非线性高维问题，因此提出了基于支持向量回归机的离心式冷水机组运行能效建模方法，提高了模型的精度。同时以某商场离心式冷水机组为例，对该方法进行验证，采用平均相对误差（MRE）和均方根误差（RMSE）对模型精度进行评价。结果表明，基于支持向量回归机的冷水机组模型MRE值较BP神经网络模型提高了37.93%，RMSE值较BP神经网络模型提高了28.81%，能准确的反应离心式冷水机组的运行能效。
建立離心式冷水機組運行能效模型對其運行能效分析以及優化控製意義重大。離心式冷水機組結構複雜且其運行能效受多種因素的影響，機理建模睏難，而支持嚮量迴歸機能夠較好的解決非線性高維問題，因此提齣瞭基于支持嚮量迴歸機的離心式冷水機組運行能效建模方法，提高瞭模型的精度。同時以某商場離心式冷水機組為例，對該方法進行驗證，採用平均相對誤差（MRE）和均方根誤差（RMSE）對模型精度進行評價。結果錶明，基于支持嚮量迴歸機的冷水機組模型MRE值較BP神經網絡模型提高瞭37.93%，RMSE值較BP神經網絡模型提高瞭28.81%，能準確的反應離心式冷水機組的運行能效。
건립리심식랭수궤조운행능효모형대기운행능효분석이급우화공제의의중대。리심식랭수궤조결구복잡차기운행능효수다충인소적영향，궤리건모곤난，이지지향량회귀궤능구교호적해결비선성고유문제，인차제출료기우지지향량회귀궤적리심식랭수궤조운행능효건모방법，제고료모형적정도。동시이모상장리심식랭수궤조위례，대해방법진행험증，채용평균상대오차（MRE）화균방근오차（RMSE）대모형정도진행평개。결과표명，기우지지향량회귀궤적랭수궤조모형MRE치교BP신경망락모형제고료37.93%，RMSE치교BP신경망락모형제고료28.81%，능준학적반응리심식랭수궤조적운행능효。
Establishment of the COP model of centrifugal chiller is of great significance to it's energy efficiency analysis andoptimal control. Since centrifugal chillers operation energy efficiency structure is complex, which is greatly affected by operating parameters, it is difficult to build the mechanism model. In this paper, a prediction model of centrifugal chillers operation energy efficiency was proposed based on Support Vector Regression, whose model parameters were optimized by Particle Swarm Optimization algorithm. This model was verified by centrifugal chillers equipped in a shopping mall, mean relative error (MRE) and root mean squared erro (RMSE) was adopted as the evaluation of the prediction accuracy. The results showed that the prediction accuracy MRE of SVR model based on PSO optimization algorithm was 37.93 % higher than that of BP neural network, and RMSE was28.81% higher than of BP neural network. This model can provide theorical basisfor the centrifugal chiller energy efficiency analysis, fault detection and diagnosis, and optimizing control.
