A New FPGA-Based Real-Time Digital Solver for Power System Simulation

Abstract

Considering the rational use of field programmable gate array (FPGA) resources, this paper proposes a new FPGA-based real-time digital solver (FRTDS) for power system simulation. Based on the relationship between the number of computing components, the operating frequency, and the pipeline length, the best selection principle is given. By analyzing the implementation method of the Multi-Port Read/Write Circuit, the computing formula of the Look-Up-Table (LUT) consumption was derived. Given the excessive use of LUTs in the original computing components, the computing components were assembled in a single typical arithmetic expression of the power system simulation program, as the basic computing formula was characterized by a subset of the typical computing formula and multiple uses of the same variable. Data communication between different computing components was realized by using Multi-Port Input Circuits that share some outputs of read controller, and Multi-Port Output Circuits, which share some outputs of computing cores. According to the test results of original FRTDS and new FRTDS, it was found that the solution proposed in this paper had a shorter ideal simulation time and a higher parallel computing capability, which was very suitable for real-time digital simulation of power systems.

An FRTDS Real-Time Simulation Optimized Task Scheduling Algorithm Based on Reinforcement Learning

“…As seen in Figure 1, multiple selected read-write control I/O ports are connected to all RAM in the RAM area, and the remaining I/O ports only communicate with the private RAM area of their own computing components. According to the design of FRTDS [21], internal RAM can store up to 1024 double floating-point numbers. Let FRTDS enable L+1 BRAMs labeled n 0 to n L , and let the system contains P+1 computing component modules labeled m 0 to m P .…”

“…Our laboratory presents an FRTDS system, and we decompose power systems into subsystems to solve them. This is an effective approach [15], [17]- [21]. FRTDS casts the simulation process into many steps.…”

“…Power system electromagnetic transient simulation calculation task scheduling software was developed to directly generate the binary stream required by FRTDS, eliminating concern about the coding of simulation programs [20]. The hardware platform we used to run examples was designed by [21], and it promises a reduction in LUT usage in FPGA. Figures 1 and 3 show the system structure of a computing component in FRTDS.…”

“…If the I/O required by the computing task does not occupy all of the needed I/O ports, then the data are transferred from other RAM storage areas through the F or G port to complete the calculation. According to the hardware design [21], the A and E ports are directly connected to all RAM memory blocks, so data in any RAM area can be obtained directly by reading the control layer, whether in the private RAM area or that of the computing component. The reading and writing relation between RAMs and computing components can be seen in Figure 2.…”

Assessment of Methods for the Real-Time Simulation of Electronic and Thermal Circuits

“…Moreover, strategies like index reduction [17] and model order reduction [18,19] have been proposed to streamline the solution of the system equations. In the latest decades, the need for RT performance has motivated the development of simulation tools and platforms able to meet this requirement [20][21][22]. The capability of the proposed solution approaches to deliver RT execution, however, needs further evaluation.…”