对于交变负载（就是负载力随着运动位置的不同，力是变化的，不但力的大小变化，而且方向也变化）所以用VCMM方程计算的液压系统开环增益就是变化很大，那么在这个过程中控制器的增益又应该如何变化？怎样实现的？
谢谢，利用VCMM方程，难点在于分析和确定负载力，准确估算粘性力，动静摩擦力，弹性力并不容易，还有交变负载的惯性力，这些力的最大值并非同一时刻出现，不能简单叠加。还是要进行负载分析。
对于交变负载（就是负载力随着运动位置的不同，力是变化的，不但力的大小变化，而且方向也变化）所以用VCMM方程计算的液压系统开环增益就是变化很大，那么在这个过程中控制器的增益又应该如何变化？怎样实现的？
“还是要进行负载分析。”
是的，你现在不加载分析吗？
我没说这很容易。
如果机械设计师有任何好处，他们应该提供帮助。
“所以用VCMM方程计算的液压系统开环增益就是变化很大”
是的，某些应用很难，但可以补偿开环增益的变化。
“那么在这个过程中控制器的增益又应该如何变化？怎样实现的？”
RMC可以每毫秒改变一次控制器增益。这是使用RMCTools中的用户程序和曲线工具完成的。曲线工具对于补偿非线性系统非常有用。有时，如果由于多个变量导致开环增益发生变化，则需要一个等式。
这是一个优化困难系统的学生。不需要计算器。 RMCTools就是必要的。学生使用我们的“自动调整”功能来计算20,40,60,80,100度的增益。将增益作为角度的函数输入曲线工具。每毫秒角度用于索引曲线以获得该角度的最佳控制器增益。
https://deltamotion.com/peter/Videos/Non-Linear-Lab_Medium.mp4
我希望尽快制作出更好的视频。
但是，液压和机械设计必须使这成为可能。我们设计这个系统很困难，但可以进行优化。
"还是要进行负载分析。"
Yes, don't you do load analysis now?
I didn't say it is easy.
The mechanical designers should help with the calculations.
"所以用VCMM方程计算的液压系统开环增益就是变化很大"
Yes, some applications are difficult but it is possible to compensate for changes in the open loop gain.
"那么在这个过程中控制器的增益又应该如何变化？怎样实现的？"
The RMC can change controller gains every millisecond.This is done using a user program and the curve tool in RMCTools. The curve tool is very useful for compensating for non-linear systems. Sometime an equation is required if the openloop gain changes due to more than one variable.
This is a student optimizing a difficult system.No calculator is required.RMCTools is all that is necessary.The student uses our "AUTO TUNING" feature to compute the gains at 20, 40, 60, 80, 100 degrees. The gains are entered into the curve tool as a function of angle.Every millisecond the angle is used to index into the curve to get the best controller gains for that angle.
https://deltamotion.com/peter/Videos/Non-Linear-Lab_Medium.mp4
I hope to make a better video soon.
However, the hydraulics and mechanical design must make this possible.We designed this system to be difficult but possible to optimize.
RMC可以每毫秒改变一次控制器增益。假如视频中运动循环周期是2.5秒，那么控制器可以改变2500次增益，但限制瓶颈是伺服阀的响应，假设伺服阀是50HZ，那么这个系统最多可以设置50组不同增益，我这样理解对吗？
实际控制这样的系统是需要20,40,60,80,100度的5组增益就够了，还是需要50组增益？
关于"AUTO TUNING"
我猜测：将VCMM方程式作为算法，植入控制器固件，通过检测和计算每个不同时刻的速度V，用v%作为不同增益，是这样实现“AUTO TUNINGgains”，我这样理解对吗？ 本帖最后由 PEN 于 2019-6-5 12:47 编辑
"RMC可以每毫秒改变一次控制器增益。假如视频中运动循环周期是2.5秒，那么控制器可以改变2500次增益，但限制瓶颈是伺服阀的响应，假设伺服阀是50HZ，那么这个系统最多可以设置50组不同增益，我这样理解对吗？
实际控制这样的系统是需要20,40,60,80,100度的5组增益就够了，还是需要50组增益？
使用曲线修改增益。
X轴是以度为单位的角度。
Y轴是增益。
每毫秒当前角度用于索引曲线以获得增益。 作为角度的函数，增益平滑地变化。 阀门平稳打开和关闭没有问题。
我刚刚将这条曲线作为插图。
实际上，必须有8条曲线作为角度的函数。
1.气缸延伸
2.比例收益
3.衍生收益
4.二阶导数增益
5.积分器增益
6.速度前馈
7.加速前馈
8.挺举前馈
The gains are modified using curves.
The X axis is the angle in degrees.
The Y axis is the gain.
Every millisecond the current angle is used to index into the curve to get the gain.The gains change smoothly as a function of the angle.The valve has no problem opening and closing smoothly.
I just made this curve up for illustration.
In reality there must be 8 curves as a function of angle.
1. Cylinder extension
2. Proportional gain
3. Derivative gain
4. Second derivative gain
5. Integrator gain
6. velocity feedforward
7. Acceleration feedforward
8. Jerk feedforward
关于"AUTO TUNING"
我猜测：将VCMM方程式作为算法，植入控制器固件，通过检测和计算每个不同时刻的速度V，用v%作为不同增益，是这样实现“AUTO TUNINGgains”，我这样理解对吗？"
VCCM方程用于计算开环增益。 开环传递函数用于计算闭环传递函数。 我在这里展示
https://www.iyeya.cn/thread-66648-1-1.html
我还没做完。
我们需要讨论添加微分控制器增益。
我们需要讨论添加积分器控制器增益。
每次更改后，我们都可以将闭环极点移动到更好的位置，以获得更快的响应
The VCCM equation is used to calculate the open loop gain.The open loop transfer function is used to calcuate the closed loop transfer function.I show this here
https://www.iyeya.cn/thread-66648-1-1.html
I am not done yet.
We need to talk about adding the derivative controller gain.
We need to talk about adding the integrator controller gain.
After each change we can move the closed loop poles to a better position for a faster response.
本帖最后由 蜻蜓 于 2019-6-6 23:49 编辑
这个帖子没及时发出来，和后面帖子重复了。 Posting is not shown, re-sent, look at this.
My understanding of feedforward is that if we know the fixed size parameters, such as gravity, Coulomb friction, its magnitude and value, we can use feedforward compensation. But for variable load, the parameters are variable, and the feedforward value should also change in real time. How can the controller predict (know in advance) the feedforward compensation value that needs to be output?
By calculating the first derivative (velocity), the second derivative (acceleration), the second derivative gain (Jerk acceleration rate of change), multiplied by the reduced coefficient, the variable feed-forward value is obtained.
Or the feed-forward value can be obtained by multiplying the velocity and acceleration by a coefficient varying according to the curve.
If the feed-forward value can be curved and the feed-forward value can be read according to the curve, the controller will be very advanced.
Last year in this forum, last year, I knew nothing about servo control, so I often misunderstood it.发帖子不显示，重新发了一次。
我对前馈的理解，对已知固定大小参数，比如重力，库仑摩擦力，知道它的大小数值，可用前馈补偿，但对于变负载，参数是变化的，前馈值也应该实时变化的，控制器如何预知（提前知道）需要输出的前馈的补偿值？
是通过求取一阶导数（速度），二阶导数（加速度），二阶导数增益（“Jerk加速度的变化率”），乘以缩小的系数，得到的变化的前馈值？
或者用速度，加速度乘以一个按曲线变化的系数得到前馈值。
把前馈值能做出曲线来，按曲线读取前馈值，那控制器会很先进。
我去年来这个论坛，去年，我对伺服控制一无所知，所以经常会理解错。
本帖最后由 蜻蜓 于 2019-6-6 23:51 编辑
PEN 发表于 2018-7-20 12:00
这是一种简化。
很多时候都有妥协。
我们使用的算法允许使用小直径气缸，这是其他控制器所需要的。 较小 ...
We have a customer that can use 5 inch diameter cylinders with our 3rd generation controllers but must use 6 inch cylinders with our 2nd generation controllers.The decision is simple.Use our 3rd generation controller with advanced algorithms. The 2nd generation controller used a more traditional PID with velocity and acceleration feed forward. ，Penteacher：
小直径缸带来的问题1是输出力降低2是固有频率的降低，第三代控制器什么算法能改善这两点？能降低阀口压差，让更多的压力作用在油缸上？是挺举加大弹性模量？
蜻蜓 发表于 2019-6-5 19:00
Posting is not shown, re-sent, look at this.
My understanding of feedforward is that if we know the ...
很少需要使用曲线。
几个月前，蜻蜓询问负荷从1吨变为100吨。
可以使用负载引起的压力来计算质量。 控制器增益可以完美计算。
但是，大多数情况下负载不会发生太大变化。 前馈收益不一定是完美的。 如果前馈增益完美，那么就不需要闭环控制。 如果前馈增益在完美的5％以内，则会显着降低流动误差。 闭环控制只需要纠正5％的误差。 这比纠正100％误差的闭环控制要好得多。 我总是尝试用平均负载来优化控制器增益。
不要让完美成为善的敌人。
但接近完美总是好的。
Rarely does one need to use curves.
Months ago dragonfly asked about a load that changed from 1 ton to 100 tons.
The mass can be calculated using the load induced pressure. The controller gains can be calculated perfectly.
However, most of the time the loads do not change that much. The feedforward gains do not need to be perfect. If the feedforward gains were perfect then there would be no need for closed loop control. If the feedforward gain is within 5% of perfect it will reduce the flowing error significantly. The closed loop control will only need to correct for the 5% of error. This is much better than the closed loop control correcting for 100% of the error. I always try to optimize controller gains with an average load.
Do not let perfection be the enemy of the good.
But is is always good to be close to perfect.
页:
1
[2]