【摘要】：目前的家用空調(diào)通常只能控制室內(nèi)溫度,在低緯度夏季含濕量較大的地區(qū),僅控制室內(nèi)溫度同時室內(nèi)相對濕度過高會導(dǎo)致人體不適。理論上通過直膨式空調(diào)壓縮機(jī)和蒸發(fā)器風(fēng)機(jī)變頻可以實現(xiàn)室內(nèi)溫濕度同時控制,空調(diào)系統(tǒng)中的傳熱傳質(zhì)耦合是實現(xiàn)這一目標(biāo)的最大障礙。本文主要內(nèi)容即建立合適的直膨式空調(diào)系統(tǒng)模型進(jìn)行解耦,預(yù)測系統(tǒng)在不同工況下的制冷能力和除濕能力。本文對直膨式空調(diào)系統(tǒng)各部件建立了物理模型,組合起來的直膨式空調(diào)系統(tǒng)物理模型迭代次數(shù)過多,導(dǎo)致誤差相對較大,響應(yīng)時間長,不適合單獨用于開發(fā)面向控制的算法,故本文進(jìn)一步嘗試應(yīng)用經(jīng)驗建模的方法。另外本文對焓差法下的肋片效率公式進(jìn)行了簡化,得到了精度較高,運算量較小的經(jīng)驗公式。已經(jīng)有研究者采用當(dāng)下最常用的經(jīng)驗建模方法——人工神經(jīng)網(wǎng)絡(luò)模型對某一工況下直膨式空調(diào)系統(tǒng)Φs和Φl(顯熱冷量和潛熱冷量)進(jìn)行了預(yù)測,得到了不錯的預(yù)測結(jié)果。但經(jīng)驗?zāi)Ｐ偷娜秉c在于當(dāng)工況改變時,根據(jù)原工況下實驗數(shù)據(jù)建立的模型預(yù)測誤差較大。因此本文對工況漂移時人工神經(jīng)網(wǎng)絡(luò)模型的預(yù)測能力進(jìn)行了驗證,Φs和Φl平均誤差分別為1.3%和18.3%,預(yù)測結(jié)果誤差較大,說明人工神經(jīng)網(wǎng)絡(luò)模型也不適合單獨用于建立面向控制的直膨式空調(diào)模型。綜上,本文嘗試將物理模型和人工神經(jīng)網(wǎng)絡(luò)模型結(jié)合起來,綜合物理模型能解耦物理過程和人工神經(jīng)網(wǎng)絡(luò)模型快速準(zhǔn)確的優(yōu)點。直膨式空調(diào)系統(tǒng)傳熱傳質(zhì)耦合發(fā)生在與室內(nèi)空氣直接接觸的蒸發(fā)器上,因此本文對直膨式空調(diào)系統(tǒng)中的蒸發(fā)器建立了物理子模型,對壓縮機(jī)、冷凝器和電子膨脹閥建立了一個人工神經(jīng)網(wǎng)絡(luò)子模型。經(jīng)過細(xì)致分析后確定了將兩個子模型結(jié)合起來的輸入輸出,并分別對兩個子模型進(jìn)行了驗證,都得到了不錯的預(yù)測結(jié)果。最后驗證了混合模型在工況漂移時的預(yù)測精度,Φs和Φl平均誤差分別為5.8%和2.8%,預(yù)測效果遠(yuǎn)好于同樣條件下的人工神經(jīng)網(wǎng)絡(luò)模型,可見本文建立的面向控制的直膨式空調(diào)系統(tǒng)混合模型是有效的。在對包含耦合物理過程的系統(tǒng)進(jìn)行建模時可以參考本文對直膨式空調(diào)系統(tǒng)建立混合模型的過程,對物理過程存在強(qiáng)耦合的部件建立物理模型,對系統(tǒng)的其他部件整體建立一個經(jīng)驗?zāi)Ｐ汀?br/>[Abstract]:At present, domestic air conditioning can only control indoor temperature. In areas with high summer humidity in low latitudes, only indoor temperature control and indoor relative humidity are too high will lead to human discomfort. Theoretically, the indoor temperature and humidity can be controlled simultaneously by the frequency conversion of the compressor and the evaporator fan, and the coupling of heat and mass transfer in the air conditioning system is the biggest obstacle to achieve this goal. The main content of this paper is to establish a suitable model of direct expansion air conditioning system to decouple and predict the refrigeration capacity and dehumidification capacity of the system under different working conditions. In this paper, the physical model of the components of the direct expansion air conditioning system is established. The combined physical model of the direct expansion air conditioning system has too many iterations, resulting in a relatively large error and a long response time, so it is not suitable for the development of control-oriented algorithms alone. Therefore, this paper further tries to apply the method of empirical modeling. In addition, in this paper, the efficiency formula of rib plate under enthalpy difference method is simplified, and the empirical formula with higher precision and less operation is obtained. Some researchers have used the most commonly used empirical modeling method, artificial neural network (Ann), to predict 桅 s and 桅 l (sensible heat cooling and latent heat cooling) of direct expansion air conditioning system under a certain working condition, and good prediction results have been obtained. But the disadvantage of the empirical model is that the prediction error of the model established according to the experimental data under the original working condition is large when the working condition is changed. Therefore, this paper verifies the prediction ability of the artificial neural network model when the working condition drifts. The average errors of 桅 s and 桅 l are 1.3% and 18.3%, respectively. It is suggested that the artificial neural network model is not suitable for the establishment of a control-oriented direct expansion air conditioning model either. In summary, this paper attempts to combine the physical model with the artificial neural network model, which combines the advantages of the physical model and the artificial neural network model to decouple the physical process and the artificial neural network model quickly and accurately. The heat and mass transfer coupling of direct-expansion air conditioning system occurs on the evaporator in direct contact with indoor air, so the physical sub-model of evaporator in direct-expansion air conditioning system is established in this paper. An artificial neural network submodel was established for condenser and electronic expansion valve. After careful analysis, the input and output of the two sub-models are determined, and the two sub-models are verified, and good prediction results are obtained. Finally, the prediction accuracy of the mixed model is verified. The average errors of 桅 s and 桅 l are 5.8% and 2.8% respectively, which is much better than the artificial neural network model under the same conditions. Therefore, the control-oriented hybrid model of direct expansion air-conditioning system is effective. When modeling the system with coupled physical processes, we can refer to the process of establishing a hybrid model for the direct expansion air conditioning system in this paper, and establish a physical model for the components with strong coupling in the physical process. Establish an empirical model for other parts of the system as a whole.
【學(xué)位授予單位】：浙江大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：TB657.2

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 Minyoung Kim;Charles P.Gerba;Christopher Y.Choi;;Assessment of physically-based and data-driven models to predict microbial water quality in open channels[J];Journal of Environmental Sciences;2010年06期

2 ;STUDY ON THERMODYNAMIC MODEL OF A COMPRESSOR WITH ARTIFICIAL NEURAL NETWORKS[J];Chinese Journal of Mechanical Engineering(English Edition);1999年01期

，

本文編號：2338619