Influence Factors and Compensation of Pneumatic Cylinder Crawling in Low-Speed Motion
Abstract：Through the establishment of aerodynamic model and the joint simulation，pneumatic cylinder crawling at low speed and the source of location error were discussed. Air source pressure，mass of load，and friction difference influence on pneumatic cylinder crawling were analyzed. System is controlled by single neuron PID and adding vibrating signal instead of traditional PID. The results show that big air source pressure，small mass of load，and small difference of static and dynamic friction can help prevent the crawling. And system crawling was solved using single neuron and vibrating signals. It is confirmed that positioning accuracy was improved from ±0.61mm to ±0.25mm by simulation. It provided a more ideal control strategy for pneumatic cylinder smooth running at low speed.
1 Introduction
As the most common actuator in pneumatic system, cylinder is widely used in automatic production. In many industrial applications, it is often necessary to drive at low speed. However, when the cylinder is moving at low speed, due to the influence of cylinder friction resistance and gas compressibility, it often appears the alternating movement of “stop and go” or “fast and slow”, which is called “crawling phenomenon” [1]. The crawling of the cylinder will not only produce a lot of noise and seriously affect the environment, but also lead to the instability of the system and reduce the accuracy of the system.
Literature [2] studies the effect of adding nano particle lubricant to the cylinder lubricant on improving the friction of the cylinder, which greatly reduces the creeping phenomenon of the cylinder. In reference [3], a flutter signal composed of a series of low amplitude pulse signals is superimposed on the controller to compensate the friction of the cylinder; in reference [4], a new type of low friction cylinder is designed, which largely eliminates the problem of low-speed crawling of the cylinder. In order to study the influence trend of the main factors of cylinder crawling and eliminate its influence on the motion state of the pneumatic system, the rodless cylinder system is modeled and co simulated. The single neuron control and superimposed flutter signal control method are adopted to achieve the purpose of solving the crawling phenomenon stably.
Analysis on mechanism and influencing factors of creeping phenomenon of 2 cylinder
2.1 cylinder creeping mechanism
The air source fills a cavity of the air cylinder, and the air cylinder moves under the action of the gas. Due to the influence of the static friction on the inner wall of the piston, when the pressure in the intake chamber reaches a certain value, the thrust overcomes the static friction and the cylinder moves. At this time, the static friction changes into a smaller dynamic friction, while the pressure in the cylinder decreases due to the expansion of the volume, resulting in the pressure less than the dynamic friction, that is, the thrust is less than the resistance, and the cylinder stops gradually; with the air supply, when the pressure in the intake chamber increases again After that, the piston can be pushed to move again, and the cycle of the process forms the crawling of the cylinder.
2.2 influencing factors of cylinder creep
By analyzing the characteristics of the cylinder structure and pneumatic system, it can be concluded that the crawling phenomenon is related to the compressibility of the gas, the main components of the system and the friction of the cylinder. The main factors affecting the crawling of the cylinder are summed up as: air pressure, load, friction characteristics.
- Modeling and Simulation of pneumatic system
3.1 pneumatic system model
Taking pneumatic proportional valve controlled rodless cylinder as an example, the model is established. For this kind of nonlinear system, which is difficult to establish accurate model, the advantages of AMESim in graphical modeling are used to establish aerodynamic simulation model. AMESim provides pneumatic library and PCD. Users can use the existing components in the standard library or the combination of component sub models of PCD library to build the required model according to their own simulation purposes [6].
According to the PCD library, a rodless cylinder and a three position five way proportional valve are constructed to connect the pneumatic circuit, as shown in Figure 1. The pressure is provided by the air source, and the gas flow is sent to a cavity of the rodless cylinder through the proportional directional valve. Under the control of the control system, the proportional valve adjusts the opening of the valve port to control the flow and drive the cylinder. A displacement sensor is set at the cylinder, and the position of the cylinder is fed back to the control system through the joint port, which is compared with the target displacement and adjusted in real time to form a closed-loop system The system realizes the cylinder positioning and track tracking.
The parameters of the sub model are as follows: (1) rodless cylinder: piston diameter 25 mm; piston mass 0.37 kg; stroke 450 mm; viscous friction coefficient C 90 NS / m. (2) Proportional directional control valve: the displacement of valve core is 2.5 mm; the fit between valve core and valve sleeve is negative opening, the covering amount is – 0.1 mm at zero displacement; the mass of valve core is 0.021 kg; the diameter of valve core piston is 11 mm, and the diameter of piston rod is 7 mm.
3.2 co simulation
Matlab / Simulink is a well-known integrated environment in the field of dynamic system simulation; AMESim has a more accurate model, which avoids the error of mathematical modeling, and co simulation using the advantages of the two softwares is helpful to better understand the system performance. The loop model of pneumatic system is established in AMESim, and the control system model is established in Simulink. The dynamic characteristics of the two models are simulated and analyzed jointly [7]. The input displacement signal is slope signal, unit: mm, initial value is 20, slope is 10; the adjusted PID parameters are 50, 50, 6, gain K is 50. In the simulation, under the working conditions of 450 mm stroke and 10 mm / s speed, the PID control of pneumatic system is carried out, and the influence of air source pressure, load mass, dynamic and static friction difference on cylinder crawling is analyzed.
(1) The influence of air source pressure. When the load mass is 2kg and the difference between dynamic and static friction is 5N, the gas source pressure is 0.6MPa, 0.4MPa and 0.2MPa respectively. The simulation results are shown in Fig. 2. The effect of air pressure on creeping is not different, and the creeping is more obvious when the air pressure is low. With the increase of pressure, the piston sealing ring clings to the inner wall of the cylinder, resulting in the increase of friction; at the same time, the increase of pressure increases the rigidity of the system composed of the cylinder, gas pipe and valve, which is conducive to preventing the system from crawling. Under the action of double factors, crawling is not sensitive to gas source pressure.
(2) Load quality. When the air source pressure is 0.6MPa and the difference between dynamic and static friction is 5N, the load mass is 2kg, 5kg and 10kg respectively. The simulation results are shown in Fig. 3. With the increase of load mass, that is, the normal pressure of cylinder increases, the friction of cylinder increases, so the creepage will be more and more serious.
(3) The influence of dynamic and static friction difference. When the air source pressure is 0.6MPa and the load mass is 2kg, the dynamic friction of the cylinder is set as 25N, and the static friction is set as 30n (the difference is 5N), 40n (the difference is 15N) and 50N (the difference is 25N). The simulation results are shown in Figure 4. The larger the difference between the dynamic and static friction, the more serious the creep, and this factor has the most obvious effect. Therefore, the maximum static friction of the cylinder is an important feature of the cylinder friction, and the difference between the dynamic friction and the static friction is the main reason for the creep.
Between Static and Dynamic Friction
4 single neuron PID control
4.1 single neuron PID
To achieve good control effect, it is necessary to design a controller with excellent performance. Artificial neural network is an important branch of intelligent control method, which is proposed through the simulation of human brain function. It is a mapping relationship from multiple input to single output [8]. Combining single neuron with traditional PID controller, complex system control and time-varying system control can be realized, which is called single neuron adaptive PID controller [9]. According to the nonlinear characteristics of pneumatic system, the single neuron PID control can achieve good control.
4.2 single neuron control simulation
In order to ensure the convergence of the learning algorithm and the robustness of the control, the supervised Hebb learning rule is used for normalization [10]. In the Simulink block diagram, S-function is introduced as the carrier of single neuron algorithm code. According to the fixed program format, the single neuron algorithm is written in C language and embedded into S-function to run. Although the M-file s function of MATLAB is very easy to write and understand, c-MEX s function not only runs faster than M-file, but also can generate independent simulation program.
and Conventional PID Control
The parameters of PID controller replaced by single neuron control are set as follows: the values of learning rate η P, η I and η D are 80000, 400 and 600 respectively, the initial values of weighted coefficient W1 (0), W2 (0) and W3 (0) are all 0.3, and the neuron proportion coefficient K The simulation results are shown in Fig. 5 under the working conditions of gas source pressure of 0.6MPa, load mass of 2kg, dynamic and static friction difference of 5N, working stroke of 450mm and speed of about 10mm / s. The curves in the figure are target curve, PID control curve and single neuron control curve.
The traditional PID control has obvious crawling phenomenon, while the single neuron PID control can improve the positioning accuracy and crawling by adjusting the weighting coefficients W1, W2 and W3 adaptively. It can also be seen from the figure that although the single neuron control improves the system accuracy, it does not completely solve the crawling phenomenon.
5 superimposed flutter signal control
5.1 advantages of flutter signal compensation
Flutter control is to superimpose a high frequency and unequal amplitude signal on the control signal. Because the difference between the dynamic and static friction is the most fundamental link of the friction effect on the system, from the perspective of mechanics, the high-frequency flutter signal is superimposed on the control signal, so that the valve has enough energy to eliminate the static friction, which can achieve the state of quasi dynamic and non dynamic, so as to ease the crawling. From the point of view of aerodynamics, when there is no flutter signal, the displacement of the low-speed cylinder is fed back to the controller, and its crawling phenomenon is reflected in that the control signal is an oscillation signal. The superimposed high-frequency flutter signal can make the control signal have a higher frequency, so as to overcome the oscillation of the control signal, solve the crawling problem, and change the characteristics of the system in essence. The flutter signal compensation is combined with single neuron control to form a robust controller, which improves the control ability of nonlinear system.
5.2 adding method of flutter signal
The system block diagram after adding flutter signal is shown in Fig. 6 (a).
In the figure: ① – the input displacement of the system, the desired output displacement; ② – the difference between the input and output displacement of the system; ③ – the signal output by the controller and superimposed with the flutter signal to form a signal; ⑤ control by comparing with the sample valve; ⑥ – the proportional valve transforms the spool displacement according to the received signal to supply air to the cylinder; ⑦ – the actual output displacement; Ⅷ – the actual displacement feedback.
The control strategy is implemented in Simulink, and the superimposed signal is introduced in AMESim. The flutter signal is superimposed on the single neuron control quantity, which acts on the proportional valve together, and is superimposed on the basis of Figure 1, as shown in Figure 6 (b). The flutter signal is a sinusoidal signal, and the optimal amplitude and frequency of the sinusoidal signal, namely 400sin35t, are obtained by trial and error method.
5.3 simulation results after compensation
Under the working conditions of air pressure of 0.6MPa, load mass of 2kg, dynamic and static friction difference of 5N, working stroke of 450mm and speed of about 10mm / s, the simulation results of cylinder displacement superimposed with flutter signal are shown in Figure 7. Compared with the curve in Figure 5, the performance of the system at this time has changed significantly compared with that without adding flutter signal. The influence of the difference between dynamic and static friction on the system no longer exists, and the static friction in the system is converted into sliding friction under the effect of flutter signal. The response speed and accuracy of single neuron control are improved, and the crawling phenomenon of cylinder at low speed is overcome. Although the crawling phenomenon disappears after adding flutter signal to traditional PID, the response speed and positioning accuracy are reduced. At the same time, the positioning accuracy of traditional PID is ± 0.61mm, while the positioning accuracy is ± 0.25mm after using single neuron PID and flutter signal control compensation.
6 experimental verification
Combined with the simulation results, using Ni pci-6251 data acquisition card and LabVIEW software, the low-speed motion of FESTO rodless cylinder is experimentally studied. Under the working conditions of air source pressure of 0.6MPa, no load, working stroke of 450mm and speed of 25mm / s, the results are shown in Figure 8. Based on the single neuron control, the flutter signal is superimposed, which is obviously better than the traditional PID control, and is consistent with the simulation results.
7 conclusion
(1) The co simulation of AMESim and Simulink makes the motion process of the cylinder close to the real object, and the analysis results are more reliable. Through the co simulation analysis of the influence factors of cylinder crawling, large air pressure, small load mass and small dynamic and static friction difference are conducive to prevent the system from crawling; (2) for the phenomenon of cylinder crawling, the method of linear superposition of single neuron PID control signal and flutter signal is used to compensate, which solves the problem of cylinder crawling and improves the positioning accuracy of the system from ± 0.61 mm 25 mm, which lays a foundation for the application of low-speed cylinder in various industries.
Reference
中图分类号：TH16；TH138 文献标识码：A 文章编号：1001-3997（2018）05-0086-03 刘静，朱志松，陈凯聪（南通大学机械工程学院，江苏南通226019）
0 thoughts on “Influence Factors and Compensation of Pneumatic Cylinder Crawling in Low-Speed Motion”