Smart Science Vol. 2, No. 3, pp. 132-138(2014)
http://dx.doi.org/10.6493/SmartSci.2014.233
Precision Force Control for an Electro-Hydraulic Press Machine Hong-Ming Chen1,*, Guo-Wei Yang1 1Department
and Chong-Cyuan Liao1
of Electronic Engineering, Chienkuo Technology University, Changhua City, Taiwan, ROC * Corresponding Author / E-mail:
[email protected]
KEYWORDS : Electro-hydraulic servo press system, Force control, Relief valve, Flow servo valve, Composite control
This thesis is primarily intended to design a PC-based control system to control the force of an electro-hydraulic servo press system for implementing precision force control. The main feature is to develop a composite control by using the relief valve and the flow servo valve. Using feedback from a force sensor, a fuzzy controller was designed with LabVIEW software as the system control core for achieving a precision force control for the hydraulic cylinder on its travel and output. The weakness of hydraulic systems is that hydraulic oil is compressible and prone to leaking, and its characteristics can vary with oil temperature, thus making it difficult for a general linear controller to achieve accurate control. Therefore, a fuzzy controller was designed with LabVIEW along with a NI-PCI_6221 interface card and a load cell to control the servo valve flow and the relief valve to control the pressure source. The testing results indicate that accurate force control output of an electro-hydraulic servo press system can be obtained. Manuscript received: March 13, 2014 / Accepted: March 31, 2014
1. Foreword
pressure was regulated by a spring pressure regulator which was adjusted manually, thus making it impossible to slightly adjust the system structure. Therefore, achieving precision control was highly limited and dynamic adjustment was inflexible. However, due to the recent maturing of semi-conductor development, various kinds of precision sensors and servo valves have come into existence. Subsequently, hydraulic cylinder pressure can be measured by a pressure sensor, and the pressure output of the hydraulic cylinder can be controlled using a relief valve. Alternatively, the force output is measured with a load cell, and the movement direction of the hydraulic cylinder and the oil flow is controlled using a servo valve. It is therefore no longer difficult to have precision positioning and accurate output force control [1, 2, 5]. However, in a hydraulic system, hydraulic oil is compressible. Friction can develop between the cylinder wall and the piston, causing leakage problems [6]. In addition, the system characteristics may also vary with oil temperature. The servo proportional valve also has a dead-zone and hysteresis problems [7, 8]. The above properties may consequently have a tremendous effect on the control performance of the hydraulic servo system. It is therefore impossible for a general linear controller to achieve effective control. Therefore, many scholars have proposed using non-linear control principles to achieve the control, of which a fuzzy controller [3, 4, 9] is often used. The main feature of a fuzzy controller is that it does not need to obtain an accurate system mode with no need to establish a complex system equation. On the contrary, fuzzy controllers are designed by using rules based on the control principles of experts or
In a general mechanical system, a motor used for replacing human force drives a mechanical device to finish many tasks. Depending on the power transmission method, a mechanical system can possess different characteristics. For instance, take systems using a belt or gears as a power transmission medium. Although the system has accurate positioning characteristics, its power output is restricted by the motor properties and the complicated mechanical structure, thus failing to meet the requirements of high power from a small size. A hydraulic system uses hydraulic oil as the power transmission medium, which is ideal for small size with high power output and high response scenarios. The hydraulic system was gradually developed from the hydrostatic transmission theory was first proposed by Blasé Pascal (1623–1662), a French mathematician and physicist, in the 17th century. The first hydraulic machine using water as the working medium was built by English engineer Joseph Braman (1749–1814) at the end of 18th century and had industrial applications thereafter. In 1905, oil replaced water as the power transmission medium, which significantly increased the performance of hydraulics. In 1925, F. Vickers proposed the pressure-balance type vane pump theory to build a more stable foundation for hydraulics. Hydraulic systems are still widely used today in industrial fields. Restricted by their component properties, early hydraulic driving systems could only control the movement of hydraulic cylinders using an opening or closing control method. Hydraulic cylinder
Smart Science 132
Smart Science Vol. 2, No. 3, pp. 132-138(2014)
2. Electro-hydraulic servo force control system
experienced people. Its control effects are thus not inferior to the performance of other advanced controllers. In this study, force output was controlled with PC-based real-time control. The control system consists of the controller designed with the graphic control software LabVIEW which is developed by National Instrument Corp. With feedback from the NI-PCI_6221 interface card and the force sensor, the fluid flow is controlled by the servo valve while the pressure source is controlled by a relief valve, thus completing precision force control of the electro-hydraulic servo press system. A fuzzy controller is used in this system and was designed with assistance of the Fuzzy toolkit in LabVIEW. Using a real hydraulic servo press system platform for testing the control principles, generally good control effects were observed with regard single step, repeated pressure-load, sinusoid trace of force, and multistep force increase and decrease response, response speed and steadystate error.
The architecture of the electro-hydraulic servo press system is shown in Fig. 1. The hydraulic system used in this study is a verticaltype press. The hydraulic pump is driven by an AC motor, and the pressure of the hydraulic cylinder is adjusted by a relief valve. The moving direction of the hydraulic cylinder and the hydraulic flow volume are controlled by the servo valve to achieve positioning of the hydraulic cylinder and force control. The hydraulic servo system consists of a servo valve, a relief valve, a hydraulic cylinder, an optical ruler, a force sensor, a spring load, a PC and a data-acquisition card. The hydraulic cylinder moves with a reciprocal stroke travel of 20 cm.
Fig. 1 Architecture of the electro-hydraulic servo press system
3. Description of the electro-hydraulic force servo system In the hydraulic servo system, hydraulic oil is supplied from the pump driven by the motor to the hydraulic cylinder as a power source. The control system used to design the fuzzy controller is PC-based as it offers precision force control to the hydraulic cylinder. Its main control method is through the hydraulic cylinder pressure, which is measured by the pressure sensor, and the output force of the hydraulic cylinder, which is measured by the force sensor. With the PCI-6621 interface card, the pressure and force signals are captured and sent to the computer where the control level is calculated by LabVIEW as subject to the control principles of the controller. The control level is then output via the interface card to control the relief valve for initial force control of the hydraulic cylinder pressure. At the same time, the servo valve is controlled by the output of the interface card to finetune the flow, thus achieving composite control and an accurate force control effect.
The hydraulic cylinder servo position system is shown in Fig. 2 [6-8]:
Fig. 2 Architecture of the hydraulic cylinder servo position system
Smart Science 133
Smart Science Vol. 2, No. 3, pp. 132-138(2014)
The hydraulic cylinder servo position system mainly consists of the servo valve and the hydraulic cylinder. The main purpose of control is to let the fluid flow rate of the hydraulic cylinder reach the expected value Q L as soon as possible. The relationship between the servo valve spool position x v and the load flow rate Q L at the orifice can be determined from the orifice law [6] as follows:
Q L k q x v k c PL
be outlined specifically for this servo system in order to provide precision force servo control.
4. Principles and design of a fuzzy controller
(1)
Where, k q is expressed in equation (2):
k q k v cd w
1
PS sgn( xv ) PL
(2)
and PS is the oil supply pressure, Cd is the fluid emission coefficient, w is the servo spool area gradient, xv is the servo spool displacement, is the oil density, PL is the pressure variance between two sides of the hydraulic cylinder, kv is the servo valve gain, and k c is the gain constant of the pressure variance. Since the moving speed of a normal hydraulic cylinder is within the bandwidth of the servo valve, the servo valve can be assumed to be under the zero-order system, the direct proportional relationship [7] of the servo spool displacement x v and the control input u can be expressed as follows in (3):
xv k a u
(3)
The basic structure of a fuzzy controller can be divided into four parts [3, 4]: the fuzzification interface, the If-Then rule base, the fuzzy inference engine and the defuzzification interface as shown in Fig. 3. First, the input error and the error varying rate is fuzzificated to allow the input value of the error or the error varying rate to fall within the range set by the fuzzy inference. Therefore, at the input end, the gain adjustment parameters K1 and K2 are added to allow the output to fall within the control range. At the output end, a parameter K3 is also added to adjust the output control. The membership functions of the input error and the error variance for the fuzzy controller are shown in Fig. 4 and Fig. 5 and the membership function of the fuzzy output is shown in Fig. 6. In addition, the rule base design is completed with the If-Then expressions of the corresponding relationship between the system control requirements and expert experience. For the fuzzy inference method, this paper uses the common Min-Min-Max fuzzy inference of Mamdani. Lastly, the fuzzy value derived from the fuzzy inference is converted into a clear value for output through defuzzification to control the system. There are many means of defuzzification, but this paper uses center of gravity defuzzification [3, 4, 9] to determine the inference results. The inference rule base of the fuzzy controller is shown in Fig. 7.
ka is the amplified gain of the servo valve and u is the control input of the servo valve. In consideration of the internal leaking and compressibility of oil in the hydraulic system, from the flow equation [6] of the hydraulic cylinder the following equation is arrived at:
QL Ax
Vt PL Ct PL 4e
u
(4) Fig. 3 Architecture of a fuzzy controller
QL is the total oil flow volume to the hydraulic cylinder from the orifice of the servo valve, A is the area of the effective cross-section for the pressure of the hydraulic cylinder, Vt is the effective volume of the hydraulic system, e is the oil compressibility module, and C t is the total oil leaking coefficient of the hydraulic cylinder. If we rearrange equations (2) to (4), we can obtain the system mode as follows:
4 e 4 F (t ) A2 x Ct K c e F (t ) Vt Vt AK q K a
4 e u Vt
Fig. 4 Error membership function of a fuzzy controller (5)
From the equation (5) we see that the hydraulic cylinder force servo system is a parameter-variable non-linear system. An accurate control effect is impossible to achieve with a general opening or closing control method. In the next chapter, a non-linear fuzzy controller will
Fig. 5 Error variance membership function of a fuzzy controller
Smart Science 134
Smart Science Vol. 2, No. 3, pp. 132-138(2014)
Fig. 6 Output membership function of a fuzzy controller
Fig. 9 Human-machine interface control panel
Fig. 7 Inference rule base of a fuzzy controller
5. Experimental results and discussion Fig. 10 Characteristic curve of the relief valve Fig. 8 shows the real hydraulic servo press system used by this study. This thesis is primarily intended to design a fuzzy controller and apply it to the electro-hydraulic servo platform. First, the LabVIEW software was used as the system control core to design the human-machine interface control panel, as shown in Fig. 9, and the fuzzy controller in order to achieve precision force control of the hydraulic cylinder. The relationship between the input and output of the relief valve is shown in Fig. 10, from which an obvious hysteresis can be seen. This would make it difficult to achieve accurate force control.
To achieve precision force control, this study used a mixed force output control structure to achieve precision control. This mixed force output control structure is mainly uses the relief valve to control the hydraulic pressure of the hydraulic cylinder. In the meantime, the servo valve is also used to achieve an accurate flow. Finally, accurate force output control can be achieved. Fig. 11 is the block diagram of this force servo control system.
Fig. 11 Architecture of mixed force output control To investigate the performance strengths and weaknesses of this experimental platform as applied to various control structures, performance was compared with a traditional PID controller as below. In this study, parameters of the traditional PID controller are determined by using the ZN rule. The values are then slightly adjusted following a trial and error method. The parameter values are as follows:
Fig. 8 Picture of the hydraulic press system
Smart Science 135
Smart Science Vol. 2, No. 3, pp. 132-138(2014)
Pressure relief valve:
Kp=0.015
Ki=0.00237
Kd=0.00023
Servo valve:
Kp=0.023
Ki=0.01115
Kd=0.00985
However, the membership function of this fuzzy controller is designed specifically for the system's characteristics with consideration of the minimum resolution of the sensor as fine-tuned with the experimental data. The rule base was established according to empirical rules. The parameters of K1, K2 and K3 are expressed as follows: Servo valve:
K1=1.25
K2=2.75
K3=0.00425 Fig. 13 Multi-step increase error response
This experiment compared the controller performance by single step response, multi-step increase response, multi-step decrease response, sinusoid response, and repeated force pressure-load response of the force output. The red curve shown in each diagram is the PID response curve, the green curve is the fuzzy response curve, and the black curve is the setting amount. First, the multi-step force increase response is shown in Fig. 12. The initial force was set at 500 kgf with an increase of 100kgf for 10 seconds each until 1000 kgf was reached at which it stopped. The multi-step increase error response is shown in Fig. 13, from which we can see the error is about ±0.5 kgf. The multi-step force decrease response curve is shown in Fig. 14. The initial force was set at 950 kgf with a decrease of 100kgf for 10 seconds each until 550 kgf was reached at which it stopped. The multi-step decrease error response is shown in Fig. 15, from which we see the error is also about ±0.5 kgf. The force sinusoid trace response is shown in Fig. 16, and the force sinusoid trace error response is shown in Fig. 17, from which we see the trace error is about ±4 kgf. The force single step response is shown in Fig. 18, and the force single step error response is shown in Fig. 19. Its error is also about ±0.5 kgf. In addition, the repeated force pressure-load response is shown in Fig. 20, and the repeated force pressure-load error response is shown in Fig. 21. The repeated pressure-load movement flow was set to the origin position for the hydraulic cylinder when it started. When it reached the set cycle, to prevent the hydraulic cylinder and the spring load from excessive colliding, the hydraulic cylinder was set to descend slowly. The control pressure did not increase until the hydraulic cylinder touched the spring load. This is done cyclically.
Fig. 14 Multi-step decrease response
Fig. 15 Multi-step decrease error response
Fig. 16 Force sinusoid trace response
Fig. 12 Multi-step increase response
Smart Science 136
Smart Science Vol. 2, No. 3, pp. 132-138(2014)
Fig. 17 Force sinusoid trace error response
Fig. 21 Repeated force pressure-load error response The variance between each controller can be observed from the series of response curve diagrams. This thesis compares the maximum overshoot and the steady-state error by arranging the curves as shown in Table 1 & 2. For the maximum overshoot, it is apparent that the performance of the fuzzy controller is better than that of the PID controller. Table 1 Comparison of Maximum Overshoot (kgf)
Fig. 18 Single step response
Single step response
Multistep increase
Multistep decrease
Sinusoid force response
PID
12.25
11.8
5.5
9
Repeated pressureload response 12.8
Fuzzy
0.5
5.8
4.8
9.5
10.4
Table 2: Comparison of Steady-State Error (kgf) Single step response
Multistep increase
Multistep decrease
Sinusoid force response
PID
