Muy_yy
发表于 2019-12-9 19:59:43
我这个是否需要采用加加速前馈来改善速度滞后的问题,还是采用微分来提高系统响应?
或者其他的建议。
PEN
发表于 2019-12-10 04:38:07
加速前馈会有所帮助,但阀门在300毫秒内无响应。 需要更好的阀门以达到主要要求的精度。
我注意到目标加速度是由正弦函数或5阶多项式生成的。
Acceleration feed forward will help but the valve does not respond for 300 milliseconds. A better valve is required to main required accuracy.
I noticed that the target acceleration is either generated by a sine function or a 5th order polynomial.
back2049
发表于 2019-12-10 22:42:52
PEN 发表于 2019-12-10 04:38 static/image/common/back.gif
加速前馈会有所帮助,但阀门在300毫秒内无响应。 需要更好的阀门以达到主要要求的精度。
我注意到目标加 ...
这个阀没有这么慢,一般30ms可以从0~100%,应该是他们液压或机械摩擦力,安装方式等造成的滞后问题
lygbeyondgod
发表于 2019-12-11 09:52:27
都是大神,迷迷糊糊的看不懂
小满哥
发表于 2020-3-14 23:57:01
您好,请问这个demo可以发出来供参考么?
fawei1007
发表于 2024-8-12 11:45:23
您好PEN:
我们的系统是MOOG760+ATLAS240H,也是设定系统是二阶的,控制器通过随机振动可以计算出伺服阀的静态增益、自然频率、阻尼。参数确定过程不知道您是否可以提供更详细的资料或说明。谢谢!
fawei1007
发表于 2024-8-12 12:35:51
您好PEN,我们的系统采用MOOG760C+ATLAS240,设定系统为一个二阶系统,控制器通过随机振动,可以识别伺服阀的频率、增益、阻尼,这个识别过程如何实现?能否给点指导,谢谢!
PEN
发表于 2024-8-16 01:11:26
必须收集数据。它应该具有控制输出和实际位置作为时间函数。时间增量应该是 1 毫秒,但多一点或少一点也可以。必须移动执行器,使其在收集数据的同时加速和减速,以查看执行器如何响应输出的变化。
然后你需要模型。我使用以微分方程形式编写的模型。必须估计四个参数。
1. 开环增益,其单位应为毫米/秒/% 控制输出。这可以通过使用 VCCM 方程来估计,我已经写过几次了,但最好是根据经验获得这个值。
2. 阻尼系数,无单位。这可以从过冲量中估计出来,但最好还是根据经验估计这个值。
3. 每秒弧度的固有频率
4. 死区时间,即压力波从阀门传到活塞所需的时间。阀门和活塞之间每 1.5 米的死区时间应约为 1 毫秒。如果阀门安装在气缸上,则死区时间通常很短,可以忽略不计。
接下来,您需要计算控制器闭环和开环增益的公式,这些公式与闭环极点位置和估计的开环增益、阻尼系数、固有频率和死区时间有关。
Data must be gathered.It should have the control output and actual position as a function of time.The time increments should be 1 millisecond but a little more or less often will do. The actuator must be moved so that it is accelerating and decelerating while collecting data to see how the actuator responds to changes in the output.
Then you need the model.I use the model written in a differential equation.There are four parameters that must be estimated.
1. An open loop gain which should have units of millimeters per second per % control output.This can be estimated by using the VCCM equation which I have written about a few times but it is best to get this value empirically.
2. A damping factor which is unitless.This can be estimated from the amount of overshoot but again it is best to estimate this value empirically.
3. The natural frequency in radians per second
4. The dead time which is the time it takes the pressure waves to go from the valve to the piston.The dead time should be about one millisecond for every 1.5 meters between the valve and the piston.If the valve is mounted on the cylinder then usually the dead time is so short it can be ignored.
Next you need formulas for calculating the controller closed loop and open loop gains as a function of the closed loop pole locations and the estimated open loop gain, damping factor, natural frequency and dead time.
fawei1007
发表于 2024-9-30 08:06:42
谢谢您的回复,估计我要实现它还需要再琢磨琢磨,再次感谢
erni
发表于 2024-10-21 12:56:21
游勇 发表于 2019-7-10 21:49
现时博世力士乐公司已提供完整的开环及闭环电液系统的仿真,内建比例方向阀、高频响阀及伺服阀,供我们验证 ...
这资料从哪里获取?