Brief introduction to two configurations of large-capacity UPS power supply system

With the development of social economy and the continuous increase of electrical equipment, various industries have greater and greater requirements for the capacity of switching power supply module UPS. There are two ways to construct a large-capacity UPS power supply system: one is to use a single large-capacity UPS, and the other is to use an "N+m" redundant parallel structure for the UPS inverter. The disadvantages of the former are high cost, large volume and weight, difficult transportation and installation, poor reliability, and once a fault occurs, it will cause power supply paralysis. The advantage of the latter is that it improves the flexibility of power supply, and can increase the switching frequency of the small-power UPS inverter module to the MHz level, thereby increasing the power density of a single machine (or inverter module), reducing the volume and weight of the inverter module of the switching power supply module UPS, and reducing the current stress of the power switch devices of each UPS inverter module, improving the reliability of the UPS, and at the same time fast dynamic response, easy maintenance, etc.

"N+m" redundant parallel technology is a new technology specially used to improve the reliability and hot maintenance (also known as hot plug and hot replacement) of UPS. In normal operation, the switching power supply module UPS is powered by "N+m" inverter modules in parallel. Each inverter module bears an average load current of 1/(N+m). When one or k (k≤m) of the inverter modules fail, they will automatically quit power supply, and the remaining N+(mk) inverter modules will continue to provide 100% current to the load, thus ensuring uninterrupted power supply of the UPS system.

Common UPS redundancy adopts the "N+1" (m=1) parallel connection mode, or the master-slave power supply system in which the UPS inverter modules are connected in parallel through the system control cabinet and then supply power to the outside, as well as the decentralized logic power supply scheme in which the parallel function is directly designed into the inverter module unit of each UPS. Regardless of which method is adopted, each UPS inverter module must evenly distribute the load current during normal operation. During operation, if one of the UPS inverter modules fails, the parallel system automatically takes the failed inverter module offline. At this time, the entire load is evenly shared by the remaining inverter modules in proportion. Obviously, the use of such a power supply system greatly enhances the reliability of the UPS power supply system.

1. Conditions for UPS to achieve "N+1" redundant parallel operation

The "N+1" redundant parallel operation technology of UPS is the key technology to improve the reliability and availability of UPS. The parallel connection of each UPS module must meet the following three conditions:

(1) The frequency, phase, phase sequence, voltage amplitude and waveform of each UPS inverter module must be the same;

(2) Each UPS inverter module must be able to achieve uniform distribution of active and reactive currents within the range of input voltage and load. To this end, the current sharing circuit must have good dynamic response characteristics and high stability.

(3) When an abnormality occurs in current sharing or synchronization or a UPS inverter module fails, the faulty module should be automatically detected and quickly removed without affecting the normal operation of other inverter modules.

There are two key technologies: one is synchronization technology, and the other is current sharing technology. The former is mainly to solve the problem of frequency, phase, waveform and phase sequence consistency of each module, and the latter is mainly to solve the problem of uniform load power of each inverter module. Since the inverter modules of each UPS are synchronously connected in parallel with the mains power grid, there is a same corresponding circuit in the middle of each UPS or the inverter modules of each switching power module UPS have a common corresponding circuit to achieve synchronization with the mains. After synchronization, the frequency, phase, waveform and phase sequence of the inverter modules of each UPS are the same as the mains power grid, and four of the five parameters in the condition are met. There may be some differences in the output voltage between each inverter module. This difference is mainly caused by different DC voltages or different internal resistance voltage drops of the inverter modules of a single UPS. Therefore, current sharing has become the main problem of parallel operation of each inverter module, and a current sharing method must be used to make the output voltage of each inverter module consistent. Since the output of each inverter module is added to the load through a common bus, it is equivalent to each inverter module sharing the same load. Therefore, the output load power factor of each inverter module only depends on the power factor of the total load on the bus. Therefore, the output power factor of each inverter module is the same. When balancing the current, there is no need to distinguish between active and reactive components. It is sufficient to balance the total output current of the module.

2. UPS parallel connection control method

The parallel connection of UPS is generally divided into centralized control, master-slave control, distributed logic control, 3C connection control and non-interconnection line control according to its connection method.

(1) Centralized control

Centralized control can be divided into direct centralized control and indirect centralized control. In the direct centralized control mode, the parallel unit detects the frequency and phase of the mains power and sends a synchronization pulse to each UPS inverter. When there is no mains power, the crystal oscillator can generate a synchronization pulse through the phase-locked loop control of each inverter unit to ensure the synchronization of the output voltage of each unit. The parallel unit also needs to detect the total current of the load, and then divide it by the number of parallel units as the current reference of each unit, and compare it with the current of the unit to find the deviation and control it to minimize it. However, due to the detection error, there may still be errors in the actual output voltage phase. In order to eliminate this defect, an indirect centralized control method can be used. This method uses the current error △I and the output voltage u to calculate △P and △Q, where △P is used as the phase compensation amount and △Q is used as the voltage amplitude compensation amount, which can further improve the accuracy of current balancing during parallel operation.

However, since the system still uses a centralized control unit, if the control unit fails, the entire switching power module UPS parallel system will be paralyzed, there is a single point of failure, and it cannot truly achieve the purpose of high reliability and true redundancy. Therefore, the current parallel system rarely uses this method.

(2) Master-slave control

The master-slave control method is to make the parallel control unit on each module, and select a UPS module as the master through the working mode selection switch, and the other units as slaves. Each UPS module unit detects the network status signal line, and its internal master-slave flag controls whether the switch is closed or not. When one unit in the system fails, the remaining units can still work. When the host has a problem, it can be switched so that another UPS module can continue to operate normally as the host system. Usually, the UPS module that serves as the host is in voltage control mode, while the other UPS are in current control mode.

Although this method has increased reliability, its synchronization signal is still a public centralized synchronization signal. Losing the synchronization signal during the switching process may cause the module to fail, and the complexity of the switching control circuit may also affect the normal operation of the system, thereby affecting the performance indicators of the entire system. Therefore, the master-slave parallel control system is not an ideal parallel redundant system.

3) Distributed logic control

Decentralized logic control is to disperse the control rights. When the inverter power supply is running in parallel, each power module detects its own active and reactive power, and transmits it to other parallel modules through the current sharing bus. At the same time, the power module itself also receives active and reactive signals from other modules for comprehensive judgment, determines the active and reactive benchmark of this module, and thus determines the reference value of the voltage and synchronization signal (frequency and phase) of each module.

Distributed logic control technology is an independent parallel control method. It uses the current and frequency signals of each power module in each inverter power supply to synthesize and derive the compensation signal control strategy of each frequency and voltage. This method can realize true redundant parallel connection. When one module fails and exits, it does not affect the parallel operation of other modules. It has been widely used in many fields with its good characteristics such as high reliability, decentralized hazards, and easy function expansion. It has become one of the main directions of the development of computer control systems and is a relatively complete distributed intelligent control technology. However, when multiple modules are connected in parallel, the number of interconnected lines is large, the amount of information is large, and the implementation is more complicated.

(4) 3C (Circular Chain Control) connection

The idea of ​​3C parallel connection is to reduce the number of interconnection lines and signal transmission to reduce the degree of dependence on other modules. It adds the output current feedback signal of the first switching power module inverter to the control loop of the second inverter, and the output current feedback signal of the second inverter to the third inverter, and then connects them in sequence. The output current feedback signal of the last inverter is returned to the control loop of the first inverter, so that the parallel system forms a ring structure in terms of signal and forms a parallel relationship in terms of power output.

The 3C scheme introduces signals from other modules in the control loop, which strengthens the influence between the switching power modules and makes the conventional scheme difficult to control. Therefore, the H∞ theory is generally used to design the controller to solve the stability problem. Each inverter unit uses a PI controller to obtain a fast dynamic response, and uses robust control to obtain the robustness of multiple module inverters to reduce the mutual influence between inverters. Compared with the previous scheme, the 3C parallel scheme of the switching power module only introduces the current signal of one module, and there is no need to simulate the signal averaging circuit, nor to know the number of parallel modules. However, the controller is complex and is mostly implemented by a digital control system, which is costly. In addition, the controller is designed using the H∞ method, the controller order is too high, and the technical difficulty is high.