Data-driven solver for truss structures

class ddtruss.Truss(points, lines)

Simple static equilibrium solver for truss structures

Parameters:
  • points (ndarray, shape (n_points, dim)) – Point coordinates with spatial dimension dim
  • lines (ndarray, shape (n_lines, 2)) – Connectivity of bars
integrate(fun)

Integrate a function defined on bars

Parameters:fun (ndarray, shape (n_lines, ..)) – Function to be integrated
solve(A=1, E=1, U_dict={}, F_dict={}, sig0=None, construct_K=False)

Solve the static equilibrium problem for the truss structure

Parameters:
  • A (float or ndarray, shape (n_lines, )) – Cross section area
  • E (float) – Young’s modulus
  • U_dict (dict) – Prescribed displacement {point_id: (Ux, Uy), ...}, use None for Ux or Uy when this component is not concerned
  • F_dict (dict) – Prescribed nodal force {point_id: (Fx, Fy), ...}
  • sig0 (ndarray, shape (n_lines, )) – Initial stress
  • construct_K (bool) – Whether force reconstructing and performing LU factorization of the stiffness matrix
Returns:

  • u (ndarray, shape (n_ddl, )) – Displacement solution
  • eps (ndarray, shape (n_lines, )) – Strain
  • sig (ndarray, shape (n_lines, )) – Stress
class ddtruss.DataDrivenSolver(truss)

Data-driven solver for truss structures

Parameters:truss – Object defining the truss structure
load_material_data(material_data)

Load one-dimensional material data

Parameters:material_data (ndarray, shape (n_data, 2)) – Experimentally measured (strain, stress) pairs
solve(A=1, U_dict={}, F_dict={}, n_iterations=100, E_num=None, n_neighbors=1, idx=None, save_history=False)

Solve the static equilibrium problem for the truss structure using a data-driven approach

Parameters:
  • A (float or ndarray, shape (n_lines, )) – Cross section area
  • U_dict (dict) – Prescribed displacement {point_id: (Ux, Uy), ...}
  • F_dict (dict) – Prescribed nodal force {point_id: (Fx, Fy), ...}
  • n_iterations (int) – Maximimum iteration for the data-driven solver
  • E_num (float) – Numerical value
  • n_neighbors (int) – Number of nearest local states to look for in the material data
  • idx (ndarray, shape (n_lines, )) – Initial local states
  • save_history (bool) – Whether also save more iteration history
Returns:

  • u (ndarray, shape (n_ddl, )) – Displacement solution
  • eps (ndarray, shape (n_lines, )) – Strain
  • sig (ndarray, shape (n_lines, )) – Stress
  • iter_history (ndarray) – Iteration history for objective function and if asked strain/stress

Indices and tables