interface the system to pc by RS232 to serial communication
port. Then we use the Instrument Control Toolbox of the
MATLAB SIMULINK for data acquisition and plotting the
response of the system. We plot graph of LEVEL IN THE
TANK VS TIME. Then we use this data for the system
identification. We use IDENT command of the MATLAB for
Most PID controllers nowadays are digital. For digital
implementation, discrete time domain is considered. Here, the
value of sampling time is of utmost importance.  
The velocity algorithm is based on splitting the calculation of
the control value into two steps:
1. First the incremental control value Δu(tk) is calculated.
2. Then the total or absolute control value is calculated with
     u(tk) = u(tk−1) + Δu(tk−1)                                            
Thus, the final discrete PID velocity algorithm is: As opposed to the fixed control reference used in the
positional algorithm, here, the calculation of current control
uses the previous control value as reference. In essence, the
control is calculated as a change, hence the term ‘velocity
form’.
III.  COMPLETE IMC TUNED PID RESULT
A.  Set Point  Tracking
A MATLAB based GUI is developed for interfacing the
practical system with the software. The set point tracking is
done through the MATLAB based GUI and load variation
given manually to the system. As shown in the Figure 9.  the
MATLAB based GUI is used to interface with the system.
Here the range is 0 to 300 mm has been set for plotting the
response of the system.
B.   Disturbance Rejection
 Here as shown in the Figure 10 when the load variation is
applied system is settle down quickly to the set point value of
14.3 cm . Disturbance is applied to the system manually by
opening another drain valve of the system for 1 minute.  The
second valve is closed then after however the  main drain
valve is still remaining in open condition. As shown in the
figure 10 system is able to reject the disturbance also without
any overshot condition.    
C.  Load variation   
In first part of the figure 10 as shown,  the main drain valve is
slightly closed during the normal operation. The process
value is suddenly reached to the 150mm. but again it came
back to the normal set point value that is 143mm..          
IV.  CONCLUSION
We have successfully implemented IMC based level control
system starting from identification of system to be
implementation suitable control algorithms in
microcontroller. We have successfully tuned the system using
matlab simulation. The comparison of system responses
shows that a system using a IMC based digital PID controller
with tuning parameters derived from auto-tuning shows a
much better system response than conventional PID
controller. The system response has fewer oscillations, less
settling time and rise time and there is no steady state offset
as compared to open loop system and system with
proportional controller which shows that the tuning
parameters derived were accurate. Due to DC pump is used in
the system this system  does not required any kind of
pneumatic supply and I/P converter  for operating pneumatic
pump which make the system least bulkiest and cost also
reduces.   
Starting from the development of DC pump based  liquid PID控制器仿真英文文献和中文翻译(3):http://www.751com.cn/fanyi/lunwen_7331.html