Research on Model-free Control of Switching Power Supply

1. Introduction

With the rapid development of power electronics technology, power electronic equipment is increasingly closely related to people's work and life, and electronic equipment cannot do without reliable power supply . Switching power supply is a power supply that uses modern power electronics technology to control the time ratio of switching transistor on and off to maintain a stable output voltage . The switching power supply is generally composed of pulse width modulation (PWM) control IC and MOSFET . Most of the switching power supply control parts are designed and operated according to analog signals, and the disadvantage is that the anti-interference ability is very poor. Due to the rapid development of computer control technology, the processing and control of digital signals have shown obvious advantages: it is easy to process and control by computers, the flexibility of design is greatly improved, and the software debugging is convenient, etc., and PID control has emerged.

It makes the switching power supply develop in the direction of digitalization, intelligence and multi-function. This undoubtedly improves the performance and reliability of the switching power supply. However, since the switching power supply itself is a nonlinear object, it is quite difficult to establish its accurate model, and approximate processing is often used. In addition, its power supply system and load changes are uncertain, so it is often difficult to make the parameters of the PID regulator change accordingly using the above-mentioned analog or digital PID control method. The control effect is not ideal. The recently developed model-free control [1> is a promising control method. It does not rely on the mathematical model of the controlled object, and integrates modeling and control. This is very suitable for some complex and variable systems or systems with uncertain structures that are difficult to describe with accurate mathematical models. It improves the control system of the switching power supply and not only meets the requirements of high performance and high reliability of the switching power supply.

2. Working principle of switching power supply

The principle block diagram of the switching power supply is shown in Figure 1. The grid voltage is converted into a DC voltage through the rectifier and filter in the input circuit and input into the high-frequency converter. The high-frequency converter converts the input DC voltage into a high-frequency pulse square wave voltage, which is converted into a DC voltage through the high-frequency rectifier and filter in the output circuit and supplied to the load.

Figure 1 Working principle of switching power supply

The control loop with microcomputer as the core samples the output voltage and current of the switching power supply with the support of control software , and compares it with the given data, then adjusts and controls the inverter, changes the conduction frequency or conduction/cutoff time of the MOSFET tube to stabilize the output, and monitors the working status of the switching power supply.

3. Switching power supply hardware system composition

The control system of the switching power supply can select different microprocessors according to the actual situation of the project. The block diagram of its composition principle is shown in Figure 2. The power-on/reset circuit provides a stable power supply and reset function to the microprocessor. The output voltage feedback is used to adjust the output voltage value and maintain the stability of the output voltage. The current feedback circuit is similar to the voltage feedback function. The digital tube display circuit and the keyboard input circuit realize the function of human-computer interaction. The PWM output drive circuit outputs pulses to control the opening and closing of the switch tube . When the output voltage is higher than the required voltage, the width of the output pulse is reduced, thereby reducing the output voltage; when the output voltage is lower than the required voltage, the width of the output pulse is increased, thereby increasing the output voltage.

Figure 2

4. Model-free control principle

4.1 Overview of Model-free Control

In the design of control laws, it is generally necessary to establish a mathematical model of the dynamic system. The classical method requires that this mathematical model must be established in advance or at least its structure must be determined in advance. And the more accurate the model is, the better. In the design of model-free control laws, the limitation of the control law on the establishment of the mathematical model as accurately as possible in advance is broken through.

Our modeling procedure is carried out with feedback control. The initial mathematical model can be inaccurate, but the designed control law must have a certain convergence. The model-free control law we designed is to control while modeling. After obtaining new observation data, we will re-model and control. This process will continue, making the mathematical model obtained each time gradually more accurate, and thus the performance of the control law will also be improved. We call this procedure the integrated procedure of real-time modeling and feedback control.

4.2 Integrated approach to modeling and adaptive control

In the reference, the following pan-model is proposed:

y(k)-y(k-1)=φ(k-1)[u(k-1)-u(k-2)> (4-1)

Without loss of generality, it is assumed here that the time delay of the controlled dynamic system S is 1, y(k) is the one-dimensional output of the system S, and u(k-1) is the P-dimensional input. φ(k) is the characteristic parameter, which is estimated online using a certain identification algorithm, and k is the discrete time. We will see that in the identification and control integration procedure of real-time identification-real-time feedback correction, φ(k) has obvious mathematical and engineering significance.

4.3 Integration of real-time modeling and feedback control

Specifically, our framework for integrating modeling and feedback control is as follows:

(1) Based on observational data and universal models

y(k)-y(k-1)=φ(k-1)[u(k-1)-u(k-2)>

Using appropriate estimation methods, we obtain the estimation φ(k-1) of φ(k-1).

(2) To find the one-step-ahead forecast value φ*(k) of φ(k-1), a simple method is to take

φ*(k) = φ*(k-1)

When seeking the control law, we still record φ*(k) as φ(k).

(3) Control law

Acting on system S, we get a new output y(k+1). Thus, we get a new set of data {y(k+1),u(k)}.
Repeating (1), (2) and (3) on the basis of this new set of data, we can get new data {y(k+2),u(k+1)} and so on. As long as system S meets certain conditions, under the action of this procedure, the output y(k) of system S will gradually approach y0.

4.4 Design of controller program

At present, most of the controllers used in industrial production process control are classic PID regulators and their variants. For systems with less severe coupling, the control effect of the PID regulator is satisfactory, but for systems with severe coupling, the PID regulator is powerless. The following uses the PID regulator as a benchmark to compare the model-free controller with the PID regulator to illustrate that the model-free controller has better decoupling and anti-interference capabilities.

The model-free control flow chart is as follows:

Figure 3 Model-free control flow chart

5. Test results

Here, the decoupling capabilities of the model-free controller and the PID regulator are simulated and compared. In order to ensure fairness in the comparison, the parameters of the model-free controller and the PID regulator are adjusted to the optimal state, and the following system [1> (4-5) is controlled:
The control results are shown in Figures 4 and 5

u（t） y（t）
Fig.4 Simulation results of PID control situation

u（t） y（t）
Fig.5 Simulation diagram of model-free control

It can be clearly seen from the simulation results that both the model-free controller and the PID regulator have achieved good control effects on the nonlinear system, but the control ability of the model-free control method for nonlinear coupling situations is much stronger than that of the PID regulator.

6. Conclusion

Model-free control is suitable for nonlinear control. Its control rules do not need to determine the model of the specific object. It has good stability and anti-interference ability for the control of nonlinear objects such as switching power supplies . The introduction of model-free control strategy has opened up a broad space for the development of switching power supplies.