ADRC learning,

ADRC learning,

Debugging four-wheeled smart vehicles, cricket control systems, two-wheeled upright vehicles, steering gear control, these control systems use PID control, although I already have many ways to improve, however, it is still difficult to break through the limitations of the traditional PID, and the adjustment speed and superadjustment must exist at the same time. To achieve better control effects, we need to know the precise system model to solve the problem with modern control theory. I can see from the Internet that ADRC is a control algorithm that combines the advantages of PID and modern models. I want to learn about it.
I,Let's take a look at the traditional PID control technology.
PID control technology is relatively simple and easy to understand. If there is an error, it will be adjusted to infinitely close to zero.

The advantage of PID is obvious, simple, simple, and can be controlled by such a formula.
However, there are still some tips during use!
1. Although I can improve the control accuracy, its existence may easily lead to over-harmonic lag. For systems with high response speed (close to the system's Open-Loop Response frequency, basically, you need to use PD control. If you want to improve the response and ignore the excessive adjustment, you can apply the I limitation, incomplete points, and other methods to the control, it can obviously improve the system response, but the overhead will increase sharply.
2. The control of P is the embodiment of comprehensive system capabilities. When P is small, the system is relatively stable, but the tracking capability is poor. When P is large, the system is always fluctuating. If you debug a system, it will often drive you crazy, neither big nor small. At this time, we can use some methods to change P to increase the order of P and increase the system response. At the same time, we can take into account the stability P = a * Error + B when there is a small Error. Using a standard second-order function can significantly improve the quality of the control system.
3. The amount of D does play a significant role in the pre-judgment of control, especially for systems that are lagging behind. After D is added, the response can be significantly improved and the sequelae caused by point control can be restrained. However, after many debugging, I found that the upper limit of the improvement of the differential variable to the system was reached quickly. Basically, after PI adjustment was completed, D could increase the system quality by 10%-20%, then it will have no effect or cause a shock.

Therefore, I used to try my best to change P to I to D, but it is always difficult to take into account the running of the system in different States. In particular, a lot of debugging parameters have been added, and they are all messy at last, so we cannot say why. After reading this article, I suddenly realized that I was hitting the bottom of my heart.
Disadvantages of PID:
① Method for obtaining errors;
(2) de/dt extraction by error e;
③ The weighted sum strategy is not necessarily the best;
④ Credit feedback has many side effects
Many of the problems I mentioned above are related to the four points in the Summary. Let's take a look at the solution.


In my understanding, the 'Tao' here (the character will not be displayed) is the sampling interval. Sometimes it is true that the shorter the sampling period, the greater the noise, system Shock. For example, when the speed of an upright vehicle is adjusted, because the upright, speed, and steering are all about adjusting the rotation speed of the wheel, the measurement speed and noise are extremely high. Only after the sampling period is improved can the noise be significantly reduced.