Presentation is loading. Please wait.

Presentation is loading. Please wait.

1Peyton Spencer, 2Yang Liu, 2Dr. Kai Sun

Similar presentations


Presentation on theme: "1Peyton Spencer, 2Yang Liu, 2Dr. Kai Sun"โ€” Presentation transcript:

1 A Differential Transformation Toolbox for Solving Power System Differential Equations
1Peyton Spencer, 2Yang Liu, 2Dr. Kai Sun 1Auburn University 2The University of Tennessee, Knoxville ABSTRACT By modeling the grid with differential equations, utilities can predict the outcome of a fault and determine if the system remains resilient. In modern times, the grid is pushed closer to its limits to save money on infrastructure Faster simulation tools are needed to allow utilities to respond to faults proactively The differential transform method is one possible solution, but it has never been automated CONTRIBUTIONS Designed a storage method for polynomial differential equations Developed automatic transform code that can approximate solutions Tested a single machine infinite bus system METHOD The tool relies on the Differential Transform (DT) which provides a set of rules to explicitly transform a function to its Taylor series coefficients. The program is comprised of three large operations: Read the input and convert equations to MTM cell array format, allowing simple indexing ๐‘‘ ๐‘ฅ1 ๐‘‘๐‘ก =๐ด+๐ต+๐ถ ๐‘‘ ๐‘ฅ2 ๐‘‘๐‘ก =๐ทโˆ’๐ธ ๐‘‘ ๐‘ฅ3 ๐‘‘๐‘ก =๐น+๐บโˆ’๐ป Monomial Term Matrix (MTM) ๐ด .โˆ— ๐‘ฅ1 .โˆ— ๐‘ฅ2 .โˆ— ๐‘ฅ3 .โˆ—โ€ฆ.โˆ— ๐‘ฅ๐‘š = ๐ด1 ๐‘ฅ11 ๐‘ฅ12 โ€ฆ ๐‘ฅ1๐‘š ๐ด2 ๐‘ฅ21 ๐‘ฅ22 โ€ฆ ๐‘ฅ2๐‘š โ‹ฎ โ‹ฎ โ‹ฎ โ‹ฑ โ‹ฎ ๐ด๐‘› ๐‘ฅ๐‘›1 ๐‘ฅ๐‘›2 โ€ฆ ๐‘ฅ๐‘›๐‘š MTM cell array Sign cell array [๐ด] [๐ต] [๐ถ] [๐ท] [๐ธ] [๐น] [๐บ] [๐ป] [1 1] [2] [1 2] Monomials Construct a Taylor series approximation using initial values and the DT Taylor Approximation ๐‘ฅ ๐‘› ๐‘ก = ๐‘˜=0 โˆž ๐‘‹ ๐‘› ๐‘˜ ๐‘ก ๐‘˜ X(N,K) Transformed Matrix ๐‘‹ 1 (0) ๐‘‹ โ€ฆ ๐‘‹ 1 (๐พ) ๐‘‹ 2 (0) ๐‘‹ โ€ฆ ๐‘‹ 2 (๐พ) โ‹ฎ โ‹ฎ โ‹ฑ โ‹ฎ ๐‘‹ ๐‘ (0) ๐‘‹ ๐‘ (1) โ€ฆ ๐‘‹ ๐‘ (๐พ) MTM cell array Sign cell array Initial Values Transform Functions Automatically repeat the second process across multiple time windows The X(N,K) matrix is used to build a Taylor expansion for each differentiable variable. These approximations are accurate within a small time window. At the end of this window, the Taylor output becomes the new initial values and a new X(N,K) matrix is calculated. The next Taylor expansion is constructed with the new coefficient matrix, and the Taylor output is approximated for a time window of equal length. This process is repeated until the end of the time range specified. RESULTS This SMIB equation was tested against MATLAB ode45 solver: ๐‘‘๐‘ฅ ๐‘‘๐‘ก =๐‘Š๐‘ โˆ—๐‘ฆ, ๐‘ฅ=๐‘Ÿ๐‘œ๐‘ก๐‘œ๐‘Ÿ ๐‘Ž๐‘›๐‘”๐‘™๐‘’, ๐‘ฆ=๐‘“๐‘Ÿ๐‘’๐‘ž๐‘ข๐‘’๐‘›๐‘๐‘ฆ ๐‘‘๐‘ฆ ๐‘‘๐‘ก = 1 2๐ป (๐‘ƒ๐‘š โˆ’๐‘ƒ๐‘š๐‘Ž๐‘ฅโˆ— sin ๐‘ฅ โˆ’๐‘„โˆ—๐‘ฆ) Rotor Angle and Frequency of SMIB equation sin ๐‘ฅ was converted to a 9th order Taylor polynomial The solvers graphed ๐‘ก= 0, 5 with 4000 equal time steps DTsolver took 1.32 s while ode45 took .081 s Transforming high powers took the longest Increasing number of time steps affects accuracy more than increasing K order Can increase new methodโ€™s speed by allowing a varying time step that is small at points of inflection and large when the slope is changing less Acknowledgements This work was supported primarily by the ERC Program of the National Science Foundation and DOE under NSF Award Number EEC and the CURENT Industry Partnership Program.


Download ppt "1Peyton Spencer, 2Yang Liu, 2Dr. Kai Sun"

Similar presentations


Ads by Google