Model description:
Attached image shows the principal structure of the three-tank system. The plant consists of three cylinders $T_1$, $T_2$, and $T_3$ with the cross section $S_A$. These tanks are connected serially with each other by pipes with the cross section $S_n$. A single outflow valve with the cross section $S_n$ is located at tank 2. The outflowing liquid (usually distilled water) is collected in a reservoir, which supplies the pumps 1 and 2. $H_{max}$ denotes the highest possible liquid level. The control input signals are the pump liquid flow rates $Q_1$ and $Q_2$ , the output signals are the liquid levels $h_1$ and $h_2$ .
Define the following variables and the parameters: $az_i$ : outflow coefficients of tank $i$ ; $h_1$ , $h_2$ , $h_3$ : liquid levels (m); $Q_{13}$ : flow rate from tank 1 to tank 3 $(m^3/sec)$ ; $Q_{32}$ : flow rate from tank 3 to tank 2 $(m^3/sec)$ ; $Q_{20}$ : flow rate from tank 2 to reservoir $(m^3/sec)$ ; $Q_1$ , $Q_2$ : supplying flow rates $(m^3/sec)$ ; $S_A$ : section of cylinder $(m^2)$ ; $S_1$ : section of leak opening $(m^2)$ ; $S_n$ : section of connection pipe $(m^2)$. Then, the dynamics of the three-tank system is expressed by a set of differential equations
$$\eqalignno{S_{A}\displaystyle{{dh_{1}}\over{dt}}=&\,Q_{1}(t)-Q_{13}(t)\cr S_{A}\displaystyle{{dh_{3}}\over{dt}}=&\, Q_{13}(t)-Q_{32}(t)\cr S_{A}\displaystyle{{dh_{2}}\over{dt}}=&\, Q_{2}(t)+Q_{32}(t)-Q_{20}(t)\cr Q_{13}(t)=&\,\alpha z_{1}S_{n}sgn(h_{1}(t)-h_{3}(t))\sqrt{(2g\left\vert h_{1}(t)-h_{3}(t)\right\vert}\cr Q_{32}(t)=&\,\alpha z_{3}S_{n}sgn(h_{3}(t)-h_{2}(t))\sqrt{2g\left\vert h_{3}(t)-h_{2}(t)\right\vert}\cr Q_{20}(t)=&\,\alpha z_{2}S_{n}sgn(h_{2}(t))\sqrt{2g\left\vert h_{2}(t)\right\vert},}$$
where $\alpha_1$, $\alpha_2$, $\alpha_3$: outflow coefficients (dimensionless, real values ranging from 0 to 1), $g$: earth acceleration $(m/s^2)$, $sgn(z)$: sign of the argument $z$.
In the simulation, the discretized model of three-tank system is obtained by first-order Euler's method, the sampling period is 1 s and the simulation time is 1500 s. The parameters of three-tank system are given in the table below. The initial conditions are $h_1(1)=0$, $h_2(1)=0$, $h_3(1)=0$, $Q_1(1)=0$, and $Q_2(1)=0$.
| $S_A$ | $0.0154m^2$ |
| $S_n$ | $5 \times 10^{-5}m^2$ |
| $H_{max}$ | $0.6m$ |
| $Q_{max}$ | $0.0001m^3/s$ |
| $\alpha_1$ | $.22$ |
| $\alpha_2$ | $.28$ |
| $\alpha_3$ | $.27$ |
Type:
Form:
Time domain:
Linearity:
Publication details:
| Title | Data-Driven Model-Free Adaptive Control for a Class of MIMO Nonlinear Discrete-Time Systems |
| Publication Type | Journal Article |
| Year of Publication | 2011 |
| Authors | Hou, Zhongsheng, and Jin ShangTai |
| Journal | IEEE Transactions on Neural Networks |
| Volume | 22 |
| Start Page | 2173 |
| Issue | 12 |
| Pagination | 2173-2188 |
| Date Published | 11/2011 |
| ISSN | 1045-9227 |
| ISBN Number | 12409274 |
| Keywords | adaptive control, control system synthesis, convergence, discrete time systems, linearisation techniques, MIMO systems, nonlinear control systems, stability, tracking |
| Abstract | In this paper, a data-driven model-free adaptive control (MFAC) approach is proposed based on a new dynamic linearization technique (DLT) with a novel concept called pseudo-partial derivative for a class of general multiple-input and multiple-output nonlinear discrete-time systems. The DLT includes compact form dynamic linearization, partial form dynamic linearization, and full form dynamic linearization. The main feature of the approach is that the controller design depends only on the measured input/output data of the controlled plant. Analysis and extensive simulations have shown that MFAC guarantees the bounded-input bounded-output stability and the tracking error convergence. |
| DOI | 10.1109/TNN.2011.2176141 |
