新西兰跪求高人翻译此文

在新西兰

I. INTRODUCTION
Agricultural greenhouses are used to increase plant quality and productivity by
controlling temperature, and occasionally humidity, of the internal air. Research
greenhouses, on the other hand, are designated to control light intensity and CO2
concentration besides temperature and humidity in order to investigate their effects
on plant dynamics and growth quality [1-31.
Since the 1980s a number of researches have been devoted to integration of
microcomputer based control systems to operate greenhouses in order to improve
plant quality and growth that also lead to economic savings by efficient control of
temperature and humidity [4-9]. Saffell and Marshall controlled the temperature of a
greenhouse using a PDP-I1 computer [7], and Matthews and Saffell then extended
their work successfully to control the humidity [8]. However, their control algorithm
did not incorporate external humidity and they did not report results on simultaneous
control of the coupled variables of humidity and temperature of the resulting MIMO
process which can cause control and experimental problems.
In their work, Davis and Hooper [10] showed that robust greenhouse temperature
control can be achieved with the addition of heating pipe temperatures to proportional
and integral feedback control of the measured internal air temperatures.
Recent developments in the fields of artificial neural network theory, fuzzy control
technique, H,, synthesis and expert systems provide significant research and application
potentials in air conditioning research in general, in greenhouse control in
particular, because these are more flexible methods than the classical three term
controllers to deal with nonlinear multivariable processes [11-13].
607
608 O.S. TI]RKAY
The objective of this research is to develop the core of an expert system to control
the coupled variables of temperature and humidity of a glasshouse based on
empirically derived heuristic rules of the form
IF [inside and/or outside glasshouse[ condition
THEN [heater, fan, humidifier] actuation
To this end, the shallow knowledge of a process operator is used to formulate the
core of an expert control algorithm that is experimentally implemented in this work.
On the other hand. the presented algorithm can be improved with the deep
knowledge of specialists on plant dynamics, on the specific actuators and on the
objectives of the whole process [14. 15].
This research is a fusion of thermo-mechanical, electrical and computer control
applications which constitutes a typical application of modern mechatronics engineering.
In addition, this work enables the experimental validation of a humidity
measurement technique based on the regression analysis of the standard psychrometric
chart first proposed by Jepson [16].
2. EXPERIMENTAL SYSTEM DESIGN
2. !. Description of the system
The microcomputer based expert control system is shown in Fig. 1. The laboratory
scale glasshouse of dimensions 1.0 m x 0.5 m x 0.4 m is heated underneath by three
resistor heaters of 390 W total power.
The humidification process is realized by pumping water through a pipe with holes
onto soil of the glasshouse for a specified short period. This method is cheap but it
has the disadvantage of being a slow humidification technique compared to others. A
mains-driven ventilator of 0.2m~s -~ flow rate is used to control cooling and
dehumidification of the internal air. The effectiveness of cooling and dehumidification
mechanisms are important for glasshouse control and they can cause problems as was
the ease in this work and th~,t reported in [8].
r•R ~ T xw. ~.
Measurin8 ~ 0
Electronics~ T xd _[~
;~. ~"'~ Psychrometers
e lays ~ ' , ~ T w
z I:1 F~ _- ~. .- _TII~, I l
Itovl_ F l
I , - .... ,
Dimensions 1.0m x 0.5m x .4m
Fig. |. Microcomputer based glasshouse control system.
Expert system to control humidity and temperature 609
Since the process is a slow dynamics system an 8 MHz XT-type PC is used. A 12-bit
data acquisition and control board with 8 AD and 3 DA channels is mounted into the
microcomputer. Four of the AD channels of 5 V input each are utilized to digitize the
measured internal and external, dry-bulb and wet-bulb temperatures. The measured
temperatures are processed on-line to estimate the internal and external relative
humidities using Jepson's method as explained in the next section. The three
10 V-DA output ports have been used to actuate the relays controlling the fan, the
humidifier pump and the resistor heater according to the decision to be taken by the
expert control algorithm.
2.2. Temperature measurement
The temperature signals are measured using thermistors that are more accurate
than thermocouples but suitable only for low temperature ranges. The outputs of the
thermistors are conditioned through an in-house-made electronic circuit including four
channels. Each channel is calibrated to a gain of 0.012V°C -t in the range of
0-100 °C. Thus, the resolution of temperature measurement becomes equal to
Tr~ - - 1- °C x 5 V - 0.1 oc unit -x.
0.012 V 12-"
2.3. Humidity measurement
The internal and external humidities could have been measured using humidity
transmitters. However, this research permitted the experimental validation of a
humidity measurement technique first proposed by Jepson [16l and described in the
next paragraph.
The dry-bulb temperature, T, is simply that indicated by an ordinary bare
transducer such as a thermometer or a thermistor. The wet-bulb temperature, Tw, is
that indicated by the thermistor bulb covered with a wet absorbent wick and exposed
to an unsaturated air-water vapor mixture moving between 2.5 and 5 m s -t. [17]. To
this end, a permanent magnet D.C. motor driven fan has air continuously blown onto
the thermistor bulb Covered with an absorbent wick embedded in distilled water
(Fig. 1).
For air-water vapor mixtures the wet-bulb temperature and the adiabatic saturation
temperature, T*, differ by only a few tenths of a degree at atmospheric pressure.
Thus, T* can be substituted by T,, in psychrometric calculations. Corresponding to a
given combination of a dry-bulb temperature T and a wet-bulb temperature Tw, the
nonlinear psychrometric relations specify the value of relative humidity • as shown on
the schematic psychrometric chart of Fig. 2.
The conv.entional methods of humidity measurement rely upon the use of a large
matrix table of T, T,, and ~0 calculated from the nonlinear psychrometric relations, or
upon an interative procedure of these relations [18]. Jepson used the least squares
regression analysis to obtain a set of linear equations approximating the standard
psychrometrie chart curves under an atmospheric pressure assumption. The resulting
linear equation
T,, = b(#,) T + a(,t,), (1)
610 O.S. TURKAY
Tw ~ T* I ~=0.9~
I*Ci
I "~ = 0.3 : 30 T [ °C] 80
Fig. 2. Relations of T. T* and q) of schematic psychrometric chart.
where (^) denotes "'estimate", together with the regression coefficients bop ) and a(<p)
of Table 1, establishes the linearized relationship between the variables T, T,+ and ~p.
Thus, Eqn (1) and the entries of Table 1 can be used to estimate one of the variables
with sufficient accuracy for most engineering applications knowing the other two [16].
In this work. the measured temperatures T and T~, are used to estimate the humidity.
Since the regression coefficients in Table 1 are given for q~ = 30, 40, 50, etc., a simple
interpolation is used to compute the intermediate regression coefficients corresponding
to each integer relative humidity. The interpolation is iterated within a subroutine
until the condition
If T,+ (measured) - 7"+ [from Eqn (i)] < 0.2 °C
Then Output the calculated relative humidity as tile measured one.
3. EXPERT CONTROL ALGORITHM
The expert system to control the temperature and humidity of the glasshouse is
shown in Fig. 3. The desired ranges of internal relative humidity and temperature of
the glasshouse are specified as inputs to the computer code with their minimum and
maximum values. The measured internal states together with the measured external
relative humidity are fed back to the rule-based expert control algorithm consisting of
three levels of hierarchy.
In commercial greenhouse applications the internal temperature is usually greater
than the outside one. Hence. the external temperature is not included within the
decision tree. However. its inclusion is straightforward and this could be done by
incorporating one more level of hierarchy.
Table 1. Regression coefficients of linearizcd psychrometric relation
&cent;~ h(~) [°F/~FI a(~) I°FI Corr. coeff.
0.3 0.750256 - I. 15645{] 0.999897
0.4 O. 18112{)2 -2.491510 0.999945
0.5 0.858568 -2.911680 0.999973
O. 6 O. 896106 - 2.69237{) O. 999993
O. 7 0.928161 - 2. 6967311 I.(~)<M)IM]
0.8 0.955fM6 - 1.561950 I .{)00000
0.9 0.978813 -0.805760 1 .[k')O000
1.0 1 0 1
Expert system to control humidity and temperature 611
Rel'. Inputs [ Measurement1s
r,-r T P!
H0=Tmin Hl=Hmax H : Internal Rel. Hum]
Fix :External ReI.Hum]
IF
ACTUATE
I^ew,t., [Q:~ter V: F~n H: Humidir~ I ~, 1
I°'.-"-- I
Fig. 3. Decision tree of expert control algorithm.
In the first level of hierarchical control, the measured internal temperature T is
compared with the reference range of T,,i,-Tm~x. Assume, for example, that T > Tm~
at a given sampling time. Then the algorithm goes to the second level of hierarchy to
compare the measured relative humidity H with the specified H,,~,-Hm~, range.
Assuming that H > H,,~x, the algorithm based on the shallow knowledge of an expert
operator takes the decisions to put the heater off, the fan on and the humidifier off.
Thus, the temperature and the relative humidity of the glasshouse decrease until they
fall within the specified ranges.
4. RESULTS AND DISCUSSION
The experiments were conducted in March during a week period when the
temperature and humidity of the laboratory were around 14 °C and 70% relative
humidity, respectively. The experimental temperature measurement was checked
using a thermometer of 0.1 °C accuracy. The relative humidity measurement was
calibrated using a commercial psychrometer. Since the process dynamics is slow the
sampling period was chosen as 3 s.
First, the heating dynamics of the glasshouse was investigated. To this end, at
initial internal conditions of 24.1 °C and 75% RH, the desired range of temperature,
[Tmi, = 25.9°C, Tm~ = 26.00C], and the uncontrolled range of relative humidity,
[RHm,, = 3, RHm~ = 99], were specified as inputs. This way the temperature was
controlled alone without humidity control. In Fig. 4, it is seen that the internal
temperature reaches its reference range in 4 min after a time delay of approximately
1 min. Since bang-bang control is used, the recorded signal shows a limit cycle
612 O.S. TORKAY
27
26.5
26
T 25.5
['C] 25
24.5
24
Tmax=26.0 RHmax=99]
Tmin=25.9 R Hmin=30 i 78
' 76
0 1 2 3 4 5 6 7 8 9
Time [ Min. ]
Fig. 4. Heating dynamics of the glasshouse without control of humidity.
oscillation of approximately 45 s period. The temperature remained within the actual
range of [T = 25.8 °C. T = 26.2 °C] with an error of 0.1 °C which is due to the overall
thermal capacitance of the glasshouse. The achieved control is excellent for a
glasshouse application. It is noted that while the temperature increases, the uncontrolled
relative humidity decreases and finally reaches a steady-state value of 69%.
A similar experiment was conducted to observe the dehumidification dynamics. At
internal initial conditions of 79% RH and 19.2 °C, the reference range of relative
humidity [RHm,, = 67, RH,,a~ = 69] and the uncontrolled range of temperature
[Tmi, = 10°C, Tm,,x =30°C] were specified as the inputs. Figure 5 shows that the
dehumidification process is very slow. The reference range is reached in about
10 rain. In fact, the humidification and dehumidification processes of the experimental
set-up had very slow response times due to the filirness of the physical devices used
for these purposes.
The relative humidity measurement method proposed by Jcpson is experimentally
verified also in Fig. 5. It is seen that the rchttive humidity is measured with a
T
[ "C]
24
23
22
21
20
19
18
17
rmax=ao RHmax=69
~ Tmin=10 RHmin=67
PSV ~
RH
" RH Range
i
0 1 2 3 4 5 6 7 8 9
Time [ MIn. ]
85
80
75 %
70 RH
65
Fig. 5. Dehumidification dynamics of the glasshouse without control of temperature.
Expert system to control humidity and temperature 613
resolution of 1%. The accuracy of this method is significantly dependent upon the
correctness of the measurement of the wet-bulb temperature. However, this drawback
is also true when a psychrometric matrix table or iterative calculations of psychrometric
relations are used for humidity measurement. Hence, Jepson's method is a
good alternative technique compared to the other two methods due to its memory
storage advantage and possible computational speed advantage.
The simultaneous control of temperature and humidity are shown in Figs 6-8. In
these cases, the process becomes a MIMO control system.
At initial conditions of [13.5 °C, 79% RH], a step change increase of [Tmm =
15.6 °C, T,,~., = 15.8 °C] and a step change decrease of [RHmi. = 72, RHr, a, = 74] for
the desired ranges of temperature and relative humidity were specified as the inputs,
respectively. The resulting curves of temperature and relative humidity together with
16
15.5
15
T 14.5
[ °C] 14
13.5
13
on
off
Tmax= 15.8 RHmax=74 ]
Tm n = 15.6 RHmin = 72
[
I ,.,,~ T Range ~ [80
~ 76%
on ~._ te~-F-an-~ RH Flange }
%ff ÷ " --~_~__~-÷L_,__÷._r~.__ ; ;t 70
0 2 4 6 8 10 12 14 16 18
Tlme [ Min. ]
Fig. 6. Simultaneous control of increasing the temperature and decreasing the relative humidity.
Tmax=18.2 RHmax=77 1
Tin!n=18.0 RHmin=75 ] 78
20 I ~ RH Ra~nge
19.5 t ""~'~.~
19 t . "~~, ~..~~ %,%IU- ~_~ [~ - 7 4 %
18.5 .~-=~C(~=j T Range . . . . . . . ~"-~'-~;,-~t RH
["C] 18 ~ - ~ C . . . . " ~'v~.70
17 ~f ~ ÷: ÷ 66
0 1.5 3.0 4.5 6.0 7.5 9.0 10.5
Time [ Min. ]
Fig. 7. Simultaneous control of decreasing the tcmlrmrature and increasing the relative humidity.
614 O.S. TORKAY
27
26
25
T 24
[°C] 23
22
21
2O
!Tmax=24.0 RHmax=72 t
!Train=23.8 RHmin=70 I
! 76 r - - - - "
RH . . . . - -~/~- ~-~ 72
T Range =- ~-; ' P ~ ;
. . . . , - . -_~-.-~-4~' 70 %
,~ _ _.-/..n,,'i,.' ~ Heater • Fan i I 66
0 1 2 3 4 5 6 7 8 9 10
Time [ Min. ]
Fig. 8. Simultaneous control of increasing both the temperature and the relative humidity.
the control actions of the heater and the fan are depicted in Fig. 6. The actual
temperature and relative humidity reach their reference ranges in 6 and 10 min,
respectively. After 14 min the internal temperature remains within the desired range,
thus the heater becomes off. On the other hand, the fan switches on and off to
maintain the relative humidity within the reference range.
The results of a similar experiment conducted by decreasing the temperature and
increasing the relative humidity from the initial values of [19.8°C, 72% RH] are
displayed in Fig. 7. The desired ranges of temperature and relative humidity,
[T,,i, = 18.0°C, T,,~ = 18.2 °C] and [RI-I,,,,~ = 75. RH ...... = 77] are reached successfully.
Note that the cooling mechanism of the glasshouse is inefficient and this renders
the implementation of the expert control algorithm a more difficult task.
Figure 8 displays the results obtained by increasing both the temperature and the
relative humidity of the glasshouse from the initial values of [21.0 °C, 64% RH] to the
reference ranges of [7",,,,, = 23.8 °C, T ...... = 24.0 °C] and [RH,,,, = 70, RHm,x = 72]
which are also achieved successfully in approximately 10 min.
With the experimental facilities available it was not possible, however, to decrease
the temperature and the relative humidity simultaneously. This was due to the
ineffectiveness of the dehumidification process and to the lack of an effective cooling
device which are both realized by replacing the wet air inside the glasshouse with the
dry air from outside. A similar ambiguity is reported by Matthews and Saffell [8].
5. CONCLUSIONS
The temperature and relative humidity of a laboratory scale greenhouse have been
controlled using a rule-based expert system. The expert control decision tree has been
devised upon the shallow knowledge of a process operator. The experimental results
demonstrated that the expert algorithm works successfully for SISO or MIMO control
of the coupled variables of temperature and relative humidity of the glasshouse.
However, it was not possible to decrease temperature and humidity simultaneously.
This was mainly due to the ineffectiveness of the ventilator to cool the internal air
Expert system to control humidity and temperature 615
and to decrease the humidity of the glasshouse. However, this does not shade the
validity of the expert control algorithm which can be improved further with the deep
knowledge of experts for specific applications.
This experimental work has further validated the humidity measurement technique
proposed by Jepson. The described method is applicable with sufficient accuracy for
most engineering implementations as long as the wet-bulb temperature is measured
accurately. This, however, is a necessity for other conventional methods such as
iterative method of psychrometric relations or using psychrometric matrix tables.
Thus, Jepson's method is assessed to be a good alternative to these methods.
Acknowledgements--The support of the Research Fund of Bogazifi..University is acknowledged. The author
would like also to thank Bogaziqi University Alumni Society (BUMED) for their financial contribution.
Special thanks are due to S. Giilener and G. Aysun who initiated this research as their graduation project
and to Assoc. Prof. V. Kalenderoglu for his advice.
REFERENCES
I. Morimoto T., Fukuyama T. and Hashimoto Y., Growth diagnosis and optimal control of tomato plant
cultivated in hydroponics. Environ. Control Biol. 27(4), 137-143 (1989).
2. Hanan J. J., Coker F. A. and Goldsberry K. L., A climate control system for greenhouse research.
ltortscience 22(5). 704-708 (1987).
3. Morimoto T., Hashimoto Y. and Fukuyama T., Identification and control of hydroponic system in
greenhouse. Proc. 7th IFAC Symposiunz on Identification and System Parameter Estimation 2,
1677-1681. Pergamon Press, Oxford (1985).
4. Willits D. H.. Karnoski T. K. and McClurc W. F.. A microprocessor-based control system for
greenhouse research: part i hardware. Trans. Am. Soc. Agric. Engng 80. 688-692 (19811).
5. Willits D. H.. Karnoski T. K. and Wiser E., H.. A microprtv,:essor-bascd control system for greenhouse
research: part ii software. Trans. Am. Soc. Agric. Engng gll. 693-698 ([t)8l)).
6. Monteith J. L., Marshall B., Saffell R. A., Clarke D.. Gallagher J. N.. Gregory P. J.. Ong C. K.,
Squire G. R, and Terry A.. Environmental control of a glasshouse suite for crop physiology. J. Exp.
Biol. 34, 3(19-321 (1983).
7. Saffcll R. A. and Marshall B.. Computer control of air temperature in a glasshouse. J. Agric. Engng
Res. 28. 469-477 (1983).
8. Matthews R. B. and Saffell R. A.. Computer control of humidity in experimental glasshouse. J. Agric.
Engng Res. 33,213-221 (1986).
t,~. Sangcr T. R., Measurement and control of temperature, humidity and carbon dioxide concentr:ttion in
greenhouses. R. Agric. Soc. Engl. Monograph Series No. 4, Farm Electron. Comp.. pp. 269-287
( 19851.
10. Davis P. F. and Hoopcr A. W., Improvement of greenhouse heating control, lEE Proc.-D 138(3),
249-255 ( 1991 ).
11. Guesalaga A. R. and Krophollcr H. W., Improved temperature and humidity control using !1+
synthesis, lEE Proe..D 137(6)+ 374-3811 (199111.
12. Morimoto T., Fukuyama T. and Hashimoto Y., Identification of physiological dynamics in hydroponics.
Proc. 8th IFAC Symposium on hlentifieation atul System Panoneter Estimation 3, 1736-1741. Pegamon
Press, Oxford (1'-)88).
13. Payne E. C. and MeArthur R. C.. Developing Expert Systems. John Wiley, New York (1991)).
14. Price C. J. and Lee M. H., Applications of deep knowledge. Artifcial h+tell. Engng 3(1), 12-17 (1988).
15. Atkinson R. M., Montakhab M. R., Pillay K. D.. Woolons D. J.. Hogan P. A., Burrows C. R. and
Edge K. A.. Automated fault analysis for hydraulic systems. Part 1: fundamentals. Proc. inst. Mech.
Engrs part I-J. Syst. Contr. 206(14). 207-214 (1992).
16. Jepson S. C., Approximating wet bulb, dew point temperatures. ASHRAE Jnl 30(1), 26-28 (1988).
17. Jorgensen R., Fan Engineering, 8th edn. Buffalo Forge Co. (19831.
18. Sonntag R. E. and Van Wylen G. J.. huroduction to Thermodynamics: Classk'al and Statistical,
4th edn. John Wiley. New York (

评论
我靠。。。 不带你这样的

评论
农业温室大棚用于提高工厂的质量和生产力
的温度控制，并偶尔湿度，内部空气。研究％S（C％^ [/ E2 O-[* H，K
温室，在另一方面，被指定给控制光强度和CO28 R＃H. A，] 1 f
除了温度和湿度的浓度，以探讨其影响L. \ 5 |＃H-F / F：A
植物动力和增长的质量[1-31。
自20世纪80年代的研究一直致力于整合
微机控制系统，以经营大棚，以提高
植物的品质和成长，也导致经济上的节约有效的的控制of9克'SC）R。 C％_＃R
温度和湿度的[4-9]。 saffell和Marshall控制的温度的
温室用PDP-I1电脑[7]，和马修斯和Saffell的再延伸
他们的工作成功地控制湿度[8]。然而，他们的控制算法|！ O！ M5 O6 L4（P＃J4 d月％d
不包括外部湿度和他们没有报告结果同时
的湿度和温度下将所得的MIMO控制耦合的变量
过程，这可能会导致对照组和实验的问题。，_，T％X E2 W！吨
在工作​​中，戴维斯和Hooper [10]的研究表明，强大的温室温度
控制可以实现与加热管温度成正比，Z4}％％P3} 0美元的P6 K表
所测得的内部空气temperatures.0吨和积分反馈控制的＃@ - `/ X-～％步u7 V-}
人工神经网络理论，模糊控制等领域的最新发展。 Z + Z'G（N'U
技术，H，合成和专家系统提供了大量的研究和应用M5 N C！ T-R
一般空调研究的电位，在温室控制在
特别是，因为这是更灵活的方法比传统的三个学期
控制器处理非线性多变量过程[11～13]。
607）I8 Z E + X“[2 X8 N9～，P $我：Z
608 O.S. TI] RKAY7 S1 S。 R + U“K＃G，X，M * C'P
本研究的目的是开发专家系统的核心控制+ J6 W：Q9 C / S9é？
的温度和湿度的温室基于已有详细的耦合的变量^）U，Z等，x：～ - g7的_）|
根据经验得出的启发式规则的形式* T，R＆N2 G9 B-J-Q-| / Y（N％？
IF [内部和/或外部温室条件？8 T7（％）{）T，S2 Z2ĝ！ {
然后加热器，风扇，加湿器驱动
为此，肤浅的知识用于制订的过程操作
专家在这项工作中实现的控制算法，实验的核心。
另一方面。所提出的算法，可以提高与深
知识的具体执行机构和专家对植物的动态，
目标的全过程[14。 15]。
这项研究是一个融合的热 - 机械，电气和计算机控制
应用程序构成的一个典型应用现代机电一体化工程。
此外，此工作使实验验证的湿度
测量技术的基础上的回归分析，标准温湿-D'J6 R（D6 N％E; V
排行榜首次提出了杰普森[16]。
2。实验系统设计
2。 ！系统的描述E * H7 C7 Z5 U1我+？9个“^ 3米。`'U
微型计算机基于专家控制系统示于图中。 1。实验室
尺寸为1.0米×0.5米×0.4米规模的温室加热下的3％L $ Q I $ U * ^
电阻加热器的总功率390 W。
加湿过程中，实现了通过管道与孔抽水-_＆[：M6 X％～！ J /小时。我
到指定的短时期的温室土壤。这种方法很便宜，但IT0 L9 O + X4 F3 N6 V的（J，Q
具有的缺点是一个缓慢的加湿技术相比，给他人。 A7 O3 A（Z $升＃Y9Ç＃V * P8 G：N6 F + [
电源驱动通风机0.2米~~ - ～流速是用来控制冷却和“五{1 + F +诉NJ＃v4的A6 5233 P
的内部的空气的除湿。制冷和除湿的效果（H0 H1 F4 M6 G＆M0 K表
机制是重要的温室控制，它们可能会导致问题，因为是
在这项工作中日～轻松，T报道[8]。
R•R～T XW。 ～。 F％V0 V1 J＆T; K4 {6小时
Measurin8～0！ K：{＃S-C“V＆G”U
电子～T XD _ [～9 E * P3 E1Ÿ* O）Y）X5 C'E1 \
;～。 ～“'～干湿球温度计+ N9Ñ！X＃@ +？1 B，W
Ë奠定了～，～T W5 N！摹＃W C-S
Z我：F～_-～。 - _TII～，I L
Itovl_ F L＃C2 F1 D：P！ T：L5 E $ R.Ĥ！ [6 V
我 - .... ，
尺寸1.0米x 0.5米x .4米，\; L4 O（O＆Ž。V
图。 |。微机温室控制系统。
专家系统来控制湿度和温度609'C7＃？5 I $ E8Ĥ
由于过程是一个缓慢的动态系统的一个8 MHz的XT型PC被使用。一个12位
8 AD和DA通道的数据采集和控制电路板被安装到
微机。四的AD通道的5 V输入，利用数字化the5 R3 E4 Q / S7 T2 ['K / Y，G $ \
测得的内部和外部，干球和湿球温度。测得的
处理温度上线，估计内部和外部的相对（N! _1 M]％s'的G1 R
湿度杰普森的方法在下一节中解释。三
10 V-DA输出端口已被用来致动继电器控制风扇，信息第1e; J / O7 v1的O0 J5中号
加湿器泵和电阻加热器，根据该决定将要采取的
专家控制算法。* X7 U4 C-D.ü$ V
2.2。温度测量
使用更准确的热敏电阻测量的温度信号
比热电偶，但只适用于低温度范围内。输出* ^ 3 D $ ^'P1 [3 K1 _
热敏电阻，通过内部电子电路包括四个条件
渠道。每个通道校准增益为0.012V°C-T的的范围OF0Ĵ＆U（I：U！Ĵ％Z $ C
0-100℃因此，温度测量分辨率等于U。 U'^ 9小时。 S6 J！ `，P
风帆 - 1 - °C×5 V - 0.1 OC单位-X / D＃T $ J＃B7 D2 Z + [
0.012 V 12 - “
2.3。湿度测量* Q9 A7 B）+ F'I / U4 Q）|
的内部和外部的湿度可能已被使用湿度测量！ Q（_ F1 R0的I7 L0 H2 Q
发射器。然而，这项研究允许实验验证A6 @ 2 Y！ A：J。 A. {＃X5 F“^
杰普森[16升首次提出并描述了湿度测量技术
下一段。
干球温度，T，很简单，就是由一个普通的裸* M8 Z2 O3 C9 \ 4 P
传感器，如一个温度计或热敏电阻。的湿球温度，Tw的，IS4 R-U Y2 v5的G1小号+ @！ K，W
表明由热敏电阻器灯泡用湿的毛细作用吸收件覆盖和暴露
不饱和的空气 - 水蒸汽混合物的2.5和5毫秒-叔之间移动。 [17]。 ％X％@，Z，Q3 R'E-C2 V2 C P
为此，永磁直流马达驱动的风扇空气不断吹向
热敏电阻灯泡嵌入在蒸馏水中，毛细作用吸收件涵盖
（图1）。
对于空气 - 水蒸汽的混合物的湿球温度和绝热饱和
温度，T *，与仅通过在大气压力下的程度十分之几。
因此，T *可以为T时，被取代的，在空气湿度计算。相当于一个：M“T2 N7 U＃Y3 Q * E / R +ü* C
给定组合的干球温度T和湿球温度Tw，
非线性温湿关系的指定值的相对湿度％J＆^ Y'N（| 0 F * B T1 {
空气湿度图的概略图。 2。+ H + T * W％{3 O2 D9 p
的conv.entional湿度测量方法依赖于使用了大量的O * Y; G5 X4 G“T：L，V4～
T，T，矩阵表，并计算出的非线性温湿关系，或％Q“L +？）C：L + R3 T！Ø
后，对这些关系的一个交互程序[18]。赫普松用最小二乘
回归分析，以获得一组线性方程组的近似标准的“{1 _3} 6 \。 B * L6`4×
psychrometrie图表曲线在大气压下的假设。由此产生的
线性equation0 S2}'E D）小时。 W + P6Ø'K
T，B（，）T + A（，T），（1）K-R＃P $ T7e
610 O.S. TURKAY7 Q * C. L6 U7 R7 N6 N0？
TW～T * I​​ = 0.9～9 F：U-Z的“12米
I *次/ D！ （@ * B0 T“的M-？
I“～= 0.3：30 T [°C] 80
图。 2。 T. T *和q），原理图焓湿图的关系。
（^）表示“估计”，一起回归系数BOP）和（<P）
表1中，建立变量T，T，+和～P之间的线性关系。
因此，等式（1）和表1中的条目可以被用于估计的变量之一
以足够的精度对大多数工程应用知道其他两个[16]。％{/ k的：P + _ + n）O的;！ Ÿ
在这项工作中。测得的温度T和T～，使用估计的湿度。
由于回归系数在表1中给出对于q = 30，40，50等，一个简单的+ k2的g1的A4 C3 H4 8 R
使用内插来计算的中间对应的回归系数
到每个整数相对湿度。在子程序的插值迭代
直到满足条件
如果T +（测量值） - 7“+ [从方程式（ⅰ）] <0.2°C * {”@！ [0 B E“T-P
然后输出测量计算出的相对湿度瓦one.1 8 X1 G2 q“？1 6 O + N * Y
3。专家控制算法和Z; Q0～＃Q（～* D /小时
专家系统来控制温度和湿度的温室IS9 V'Ž[％L7Ÿ！ `9 L
示于图。 3。所需的范围的-S内部的相对湿度和温度。 A4 M，J / B＆A6酷睿i3Ÿ
温室被指定为输入到计算机代码与它们的最小和
最大值。测得的内部状态与测得的外部，Y，P，]“S”M4 I-一L5一
相对湿度反馈到基于规则的专家控制算法由$ @'W8 J8 V，X＃L8 U（} 2 H
三个层次的hierarchy.8 T. Y“J”U8 W1ü。 {
在商业的温室应用的内部温度通常大于
比外面的一个。因此。外部温度是不包含在
决策树。然而。将其列入非常简单，能够做到这一点。 V0 \！ U1 J2 D9 G-[+ F
一个多层次的层次。
表1中。回归系数的linearizcd温湿关系！ `1} - U＃E！ F-W（^：J
Ⅲ～H（～）[°F / FI A（～）I度FI科尔。 COEFF。
0.3 0.750256 - I. 15645 {0.999897 + U * H8 |“N，W，V（U＆P1 P9 T7e
0.4 O. 18112）2 -2.491510 0.999945
0.5 0.858568 -2.911680 0.999973“D + V：～8 T：N！C
O. 6 O. 896106 - 2.69237 {）O. 999993％A“D0 O * {$ L％M + T4 P-M，O，C
O. 7 0.928161 - 2。 6967311一（～）<M）IM] | * @ $ M6 C / Q）v
0.8 0.955fM6 - 1.561950。）00000 - B％M）C9 _。 I. @ 8 A1ŋ
0.9 0.978813 -0.805760 [K'）O000
1.0 1 0 1
专家系统来控制湿度和温度611 * Q＃]'}。 G％|＃C。 `！ U0Ň
相对“。输入[Measurement1s）S1 G＆？5 _9 R-E，V; _ $ F
R，-R T P！ R，R'H B0 W：B2 |
H0 = TMIN HL =扬程H：内部版本。坎。 J6 A，`3 N9 T6ü
修复：外部ReI.Hum] 7～0 N9 H（B％P！x $ W1 L5 N $ N5 ^
IF
开动
我EW，T，[Q：TER V：F～N H：Humidir～我～，U，Y＃B0 Z'F＃F）N $ J. @ 5℃
我度。“ - ” - I“\ + M'L8 |）L $ S％Z0Ÿ
图。 3。专家控制算法的决策树。
在第一级的分级控制，测得的内部温度T为
与参考范围T，我，相比的Tm～x的假设，例如，T>的Tm～
在一个给定的取样时间。然后算法进入到第二级别层次，Q2 x！ B：Q0 C. X：N（B7 U6`7 \
比较指定的H，测得的相对湿度H，～，嗯～，E＃M＃range.0“E5 A＆C＆P. P＃J”Z2 Z＆]
假设H> H，～X，该算法的基础上浅薄的知识的专家＃]，M5Ĥ＃@ 9}; v
运营商需要决定把加热器，风扇，加湿器off.5`+ Z'I：K3 M D * S1 K'V8 W
因此，温度和相对湿度的温室减少，直到它们
属于内的指定ranges.6 I0 T1 W $ E4 Z5 B + M
4。结果与讨论
3月进行的实验是在一个星期期间：{3 T1 E，C7中号！ _ $ Q $ X
约14°C和70％相对温度和湿度的实验室
湿度，分别。实验温度测定检查
用温度计0.1°C精确度。相对湿度测量
校准使用的是商业干湿表。由于过程的动态特性是缓慢的
％E＃高压，D％R＃U6 b，U7 f选择采样周期为3秒。
首先，玻璃温室加热动力学进行了研究。为此，AT3]＆S $我。 C / I4 N2我Y0 F
24.1℃和75％RH环境下，在所需的温度范围内，初始的内部条件。瓦特* J $ F。 B6 B4}！ \“Z：H”I J
[中Tmi = 25.9°C，铥～= 26.00C]，不受控制的相对湿度范围内，
[RHM = 3，RHM～99]，被指定为输入。这样的温度为
单独控制而不湿​​度控制。另外，在图4，可以看出，内部
温度达到参考值范围在4分钟后的时间延迟approximately2 A / C'O'EH％^＆D＆W
1分钟。由于采用Bang-Bang控制，所记录的信号显示了一个极限环$ {6} 8 D）D * W5 |
612 O.S. TORKAY-N“S2 Q ^ 4 V！|，\
27
26.5
26
Ţ25.5
[C] 25
24.5
24
最高温度= 26.0 RHMAX = 99] 3 D5 B3 q * S：J：_2 _＆G9克
TMIN = 25.9 R HMIN = 30我78 - T（J'X）X8 ^！ P
“76”T，L“C M / N
0 1 2 3 4 5 6 7 8 9 10“S-C6 S8 ^ 8 E $～，A0 T（};
时间[min。 “”P，R（E! V“？5 P
图。 4。暖气不控制湿度的温室动态。
约45秒内振荡。温度保持在实际的
范围[T = 25.8°C。 T = 26.2°C误差为0.1°C，这是由于overall3 W）B8 E1 O'R7？
温室的热容。实现了控制非常适合在
温室的应用程序。值得注意的是，虽然温度的升高，不受控制的
相对湿度降低，最终达到一个稳态值的69％。
一个类似的实验进行观察除湿动力学。在
相对湿度79％和19.2°C，参考范围的相对的内部初始条件
湿度[RHM = 67，RH，A～= 69]和温度4 W + R5 Z1 D4 N9不受控制的范围，我
[TMI，= 10°C，TM，X = 30°C]被指定为输入。图5示出了该
除湿过程是很慢的。的参考值范围在大约达到
10雨。事实上，加湿和除湿过程的实验
设置有非常缓慢的响应时间，由于所使用的物理设备的filirness
用于这些目的。
的相对湿度的测量方法提出的Jcpson是实验
验证还图。 5。它被看作，该rchttive湿度是衡量一个
T，K;升。 Z * ^'U
[C]。S4 M + E / J，K，L）X3 U / L8 @
24 + G6 \ 4 S / D％Y，K $ [è
23
22
218我“F”C-Z＃L2] 9 V / O，克
201 J）米。 S / C！ V / @
19。 R'T + G M＃N）K8}＆B
18＆Z-Y3 M7 L8 G％Y * R $ T9 f
17 + E7`（I7 A9 B0～4 f）P
RMAX =和AO RHMAX = 69
～TMIN = 10 RHmin = 67，X + L）G7 T-C6`8 M8 V
PSV～+ F. A，U）N＆D; T3 H0 Q-x
RH3 R'O8 F，L1 F`％？ Ÿ
“RH范围（L / N4 ^ 8 J $ L”～，} E'T0 V（N％J
我
0 1 2 3 4 5 6 7 8 99 N＆Z5 A4 Q2 P + O2ŕ
时间[min。 * J`（V; T6 H2 S
85
80元酷睿i3～9 H8 D-​​I（S-A，F＆A
75％G1 P＆S'P'Y
70 RH $ G+ C％W4 T，T / N'|
65
图。 5。除湿动态无控制的温室温度。
专家系统来控制湿度和温度613
分辨率为1％。这种方法的精度显着依赖于
的湿球温度的测量的正确性。然而，这drawback5 A“R＆\）D; \ P /`
时，也是如此的湿度矩阵表或迭代计算的温湿
关系用于湿度测量。因此，赫普松的方法是一个
很好的替代技术相比，由于其存储器中的其他两种方法
存储的优势，可以计算速度上的优势。
在图6-8所示的温度和湿度的同时控制。在
这些情况下，该过程成为一个MIMO控制系统。
在初始条件的[13.5℃下，相对湿度79％]，一个阶跃变化增加[TMM = 3 ^ 9×-P *？ - U9的F / U）ç“×
15.6°C，T，～，= 15.8°C]与阶跃变化[RHmi减少。 = 72，RHR中，a，= 74]
所需的温度和相对湿度范围被指定为输入，S，F7 F（J3 S8ü
分别。将得到的曲线的温度和相对湿度，连同
16）E：H＃^％E1e
15.5
156 M）`！ （～5 J P5 K +]
Ţ14.5
[°C] 14
13.5
13
上
off4 O3 M！ Q4我U1 G-U8 C2中号
Tmax为15.8 RHMAX = 74] / N * U-_％Z＃A / A / J3 {
TM N = 15.6 RHmin = 72
[3 U“P”R：A + J-E _4 S5 V
我，，，～T范围内[80：`％W＃A1 D，E：]
～76％* @：D，J5一个V！ ＃\
_德上～～-F--～RH法兰}
％FF÷“ - ～_～__～ - ÷L_ __÷。_r。__ T 70
0 2 4 6 8 10 12 14 16 18（C“T8 J6 @”B6} / | 8 Z8 |
Tlme最小。 6 E，A $ C3 M：I，？
图。 6。增加温度，降低相对湿度。湖的同时控制。 B / O，Y＆？
最高温度= 18.2 RHMAX = 77 19 I0 S＃M5 C T1 Z8 P＆e
天！N = 18.0 RHmin = 75] 784 A; [＆\ 6 V“S”O7Ĵ* M
20 I～RH RA～NGE B5 J：％E4 v）克'F，V * G“W
19.5吨“，”～“～～
19吨。 “～～，～..～～％％IU-～_～～ - 7 4％5 E1 Y7～$ L％G）F，N C5一
18.5～ - =～C（= J T范围.......～“ - ～' - ， - ～T RH：K1 X8} * N，E B3米
[C] 18～ - ～C ......“～'V～0.70
17～F～÷÷66 $ 2 U4eR8。 Z9？
0 1.5 3.0 4.5 6.0 7.5 9.0 10.5
时间[min。 ]
图。 7。同时控制降低的tcmlrmrature和增加的相对湿度。
614 O.S. TORKAY * G $ S / A＆Y。 B $ E（_'F $ A. C：
27，U \。 ]）Q8的D + \ 7 _1 Z9 P“瓦特
26
25＆J＃\＃A + J。 P，L：J $ @ * J
T 24
[°C] 23
22
21
2O P％D-J * I8 P，_7 q
Tmax为24.0 RHMAX =72吨
火车= 23.8 RHmin = 70 I
！ 76 R ----“
RH。 。 。 。 - ～/～ - ～ - ～72
T范围= - ～ - 'P + _5 L6Ÿ_8 O'Z＃P（I
。 。 。 。 ， - 。 -_～ - 。 - ～-4～“70％15 T6的n E＃Z2 K表！ Z8 q
～__ .-/..n，“我。” ～加热器球迷，我我66'B8 F'W-U9]，C
0 1 2 3 4 5 6 7 8 9 10，R（`_ * V9 L：J5}
时间[min。 ]
图。 8。增加的温度和相对湿度的同时控制。
在图所示的加热器和风扇的控制动作。 6。实际的H＆，I2 H7Ĵ。 M $ R（_9 R）Q *] 3]
温度和相对湿度达到他们的参考范围在6和10分钟，9 k图片！ P0 U /平方米F＆？
分别。 14分钟后，将内部温度保持在所需范围内，。 Q $ Y / L; ^ 8 A1小号
因此，加热器成为关闭。另一方面，风扇和关闭，以切换。 L-Z！ ＃C G，R（^
相对湿度保持在参考范围内的。
的结果进行了类似的实验，，降低温度AND4 B / X-A，R {4 S'Z9 QK
[19.8℃，72％RH]＆～“W.克”E5ı：V“k8的V从初始值增加的相对湿度
在图中显示。 7。所需的范围的温度和相对湿度，3 j'的Q）S（％）} / d8上B2 R6瓦特
[T，I，= 18.0°C，T，～= 18.2°C]和[RI-I，，，，～= 75。 RH ...... = 77]，达到成功。
需要注意的是玻璃温室的冷却机构是低效的，这renders3 B：K j的，G2的f /`
专家控制算法的实现，更艰巨的任务。+ T-V，6个Y％W）T
图8显示了得到的结果通过增加的温度和
相对湿度的温室从初始值的21.0℃，相对湿度64％]，+ R-i3的吨f8热键小号
[7参考范围“，，，，，= 23.8°C，T ...... = 24.0°C]和[RH，，，，= 70，RHM，X = 72]
也成功地实现，在大约10分钟内。 - e$ C 7 X＃N1 q，我_
与实验设施，这是不可能的，但是，为了减小
的温度和相对湿度的同时。这是由于F / $]，_3 U＆Y-] $ J“M”10我
无效的除湿过程中，缺乏有效的冷却
装置都通过更换玻璃温室内的湿空气与实现
来自外部的干燥空气。类似的模糊报道马修斯和Saffell的[8]。
5。结论
在实验室规模的温室的温度和相对湿度的已被“G6 S6 S（I + {+】j0 @）电子
控制使用基于规则的专家系统。专家控制决策树
设计的过程操作后，浅薄的知识。实验结果3 ^ 5 U $ R（Z“Y6 I $ Y7 O（C1中号
表明，专家SISO和MIMO控制算法成功的
耦合的变量的玻璃温室的温度和相对湿度。
然而，这是不可能的，同时降低温度和湿度。
这主要是由于的无效呼吸机来冷却内部air5 P）B！ W（H：V;～
专家系统来控制湿度和温度6153] 7 K。 {％Y2 V6 W8 Q5 ^
降低湿度的温室。然而，这并不遮蔽
专家控制算法的有效性，可以进一步提高与深
知识的专家为特定的应用程序。
这个实验性的工作，进一步验证了湿度测量technique8 D $ X＆X：A“H + B，Q＆F
提出的赫普松。所描述的方法是适用的具有足够的精度的
大多数工程实现长的湿球温度测量％P％X（R! ['N4 V＆I％F;ü％R
准确。然而，这是一个必要的其他常规方法，例如$ n为@ 2 v1的L * r3的k6的j $的S，Y
温湿关系或迭代法，使用温湿矩阵表。
因此，杰普森的方法来评估这些方法是一个很好的选择。＃W-E2 C / E。 Ç
致谢 - 的Bogazifi研究基金的支持。大学是公认的。作者
也想感谢Bogaziqi大学校友协会（BUMED“）的财政贡献。，H！ T6 {2 V，Z，T
特别感谢由于S. Giilener和G. Aysun的是谁发起这项研究作为他们的毕业设计

评论
......

评论

牛人 连火星乱码都一并翻译了  哈哈~~

评论

哇哈哈哈 翻译必须得彻底！

评论
楼上的不会是直接google翻译贴上来吧

评论

一看就是了.。。。

评论
与其这样倒不如去找找这书有没有中文版的买本得了。。。