Research On The Spray Evaporation And Its Soft Sensing Technology For The Vacuum Oil Purifier

To improve the performance of the oil, vacuum oil purifier is employed to remove water contamination from oil. So the purification efficiency directly relates to the water evaporation capacity. Oil spray atomization can enlarge the evaporation square so as to promote evaporation capacity. By analyzing the spray demand, taking into account either running condition or atomization size, the spiral cone nozzle and the pressure swirl centrifugal nozzle were presented for different spray places. Based on above spray technologies, a circulation spray structure was proposed. And a vacuum oil purifier was developed for refrigerator oil purification in air conditioner compressor product line.Moreover, vacuum spray evaporation is a very complicated system process. So the purification effect for a small quantity of water (i.e. moisture) content is hard to be calculated from a mechanism model. However, it is a vital quality index for refrigerator oil. Usually it is tested offline in laboratory which makes the local quality control of oil purification to be more difficult. Fortunately, soft sensing technology can be applied for its real time estimation. The soft sensing technology, quality control and integration scheduling were all applied for the purifier. So the qualified refrigerator oil production efficiency and the support capacity for other working procedures were improved to fulfill the production demand. The main contents are illustrated as follows.①Aiming at the anti-jam and lower feeding pressure at oil inlet, a spiral cone nozzle with lower cost atomization was presented. It was designed as internal and external angle-tapered hollow conic helicoids. Its beginning and end was optimized to make the stress distribution more uniformly and to avoid atomization overlapping. It allowed bigger size particles traversal without jam. Then the space geometry equations of the spray surface were formulated by mapping the dynamic from static coordinates. Sequentially the volute trace geometry model of the atomization plane and the outlet square was calculated. A design case was studied and examined in the spray test-bed developed ourselves, which grounded the optimization of the spiral cone nozzle.②According to the fine atomization demand of circulation spray point, a pressure swirl centrifugal nozzle was employed for better spray effect. The computational fluid dynamics method was used for flow field analysis. Though geometry modeling the nozzle, k-εmathematic model was proposed to simulate the impact of internal swirl distributor, running condition and geometry dimension to atomization. In this way, the optimized structure and running parameters can be selected for the nozzle.③Based on above nozzles, a circulation evaporation vacuum spray structure was presented, which achieved double-effect vacuum evaporation in one circulation so as to improve the water removal ability. Combining with other parts, the hardware of a vacuum oil purifier and its basic configuration control system was developed. Then an evaporation experiment was carried out to set up the soft sensor parameter system of the primary and secondary variables in the process.④To estimate the moisture content at oil outlet, a soft sensor model based on least squares support vector machine (LSSVM) was presented. The Rough Sets Theory was used as pre-treatment system for data reduction. Then LSSVM was used for modeling accurately. In this way, both sides'advantages were integrated for attributes reduction and model generalization. The method was applied for moisture content in purified oil soft sensing. The input dimensions and training samples were all compressed and the impact of wrong values to the model was eliminated. So the delivered model has the characteristics of faster modeling speed and higher estimation precision.⑤A dynamic soft sensor modeling method based on distributed LSSVM was presented. Taking into consideration the impact of dynamic transition to the soft sensing, the input vector was divided into blocks in accordance with the transition period. And the calculation space was divided into multi-subspace for decomposition and synthesis mapping. According to the dynamic impact of intermittent oil feeding to the vacuum degree, the proposed method was applied for feeding oil moisture content estimation. Combining with the purified oil static sensor model, the practicability was improved. Then the multilayer error transfer of the synthesis sensor model is analyzed. The result shows that the delivered model has the better error transfer control ability in general.⑥According to the refrigerator oil purification demand in the line, the oil purification quality margin control was applied based on soft sensing technology. In addition, an optimized scheduling algorithm was presented for integration scheduling with the oil filling machines in the line. Integrating the basic configuration control with soft sensing technology, quality control and integration scheduling, the delivered control system improves the production capacity and running effect of the vacuum oil purifier.