* Further changes to allow certain subsystems to be not varied, #33

pyDMPC/ControlFramework/algorithm/BExMoC.py

@@ -33,15 +33,19 @@ def CalcBCvalues(amount_vals_BCs, exp_BCs, center_vals_BCs, factors_BCs, amount_
  amount_lower_vals_BCs =[x/2 for x in amount_vals_BCs]
  amount_lower_vals_BCs = [math.floor(x) for x in amount_lower_vals_BCs]
  amount_upper_vals_BCs = [amount_vals_BCs[x]-amount_lower_vals_BCs[x]-1 for x in range(len(amount_vals_BCs))]
- values_BCs = []
- for i in range(len(amount_vals_BCs)):
- temp_list = []
- for j in range(amount_lower_vals_BCs[i]):
- temp_list.append(center_vals_BCs[i] - (amount_lower_vals_BCs[i]-j)**exp_BCs[i]*factors_BCs[i])
- for j in range(amount_upper_vals_BCs[i]+1):
- temp_list.append(center_vals_BCs[i] + j**exp_BCs[i]*factors_BCs[i])
- values_BCs.append(temp_list)
- values_BCs[0].insert(0,0)
+ if amount_lower_vals_BCs + amount_lower_vals_BCs != 0:
+ values_BCs = []
+ for i in range(len(amount_vals_BCs)):
+ temp_list = []
+ for j in range(amount_lower_vals_BCs[i]):
+ temp_list.append(center_vals_BCs[i] - (amount_lower_vals_BCs[i]-j)**exp_BCs[i]*factors_BCs[i])
+ for j in range(amount_upper_vals_BCs[i]+1):
+ temp_list.append(center_vals_BCs[i] + j**exp_BCs[i]*factors_BCs[i])
+ values_BCs.append(temp_list)
+ values_BCs[0].insert(0,0)
+ #print(values_BCs)
+ else:
+ values_BCs = [[25], [0.007]]
 
  return values_BCs
 
@@ -107,16 +111,16 @@ def CalcLookUpTables(s, time_storage, init_conds):
  boundaries = s._bounds_DVs #[low, high]
 
  if boundaries[0] != boundaries[1]:
- ranges = [slice(boundaries[0],boundaries[1]+5, 5)]
+ ranges = [slice(boundaries[0],boundaries[1]+5, 10)]
 
  """ First conduct a brute force search """
  obj_fun_val = brute(Objective_Function.Obj,ranges,args = (BC, s), disp=True, full_output=True, finish = None)
- init_conds = obj_fun_val[0]-5
+ init_conds = obj_fun_val[0]
  cons = ({'type':'ineq','fun': lambda x: x-boundaries[0]},
  {'type':'ineq','fun': lambda x: boundaries[1]-x})
 
  """ Perform an addtional optimization to refine the previous findings """
- obj_fun_val = minimize(Objective_Function.Obj,init_conds,args = (BC, s),method='COBYLA',constraints=cons, options={'maxiter':100, 'catol':0.0002, 'rhobeg':5})
+ obj_fun_val = minimize(Objective_Function.Obj,init_conds,args = (BC, s),method='SLSQP',constraints=cons, options={'maxiter':100, 'ftol':0.01})
  else:
  ranges = [slice(boundaries[0],boundaries[1]+1, 1)]
  obj_fun_val = brute(Objective_Function.Obj,ranges,args = (BC, s), disp=True, full_output=True, finish = None)
@@ -171,9 +175,8 @@ def CalcLookUpTables(s, time_storage, init_conds):
  storage_cost[counter,len(s.values_BCs)] = obj_fun_val.get('fun')
  exDestArr[index1[0].tolist(), index2[1].tolist()] = obj_fun_val.get('fun')
 
- # Get Output Variables
+ """ Get Output Variables """
  output_vals = Objective_Function.GetOutputVars()
-
  """ Fill look-up table Out """
  k = len(s.values_BCs)
  while k < (s.num_VarsOut+len(s.values_BCs)):
@@ -187,12 +190,13 @@ def CalcLookUpTables(s, time_storage, init_conds):
  else:
  storage_grid = np.append(storage_grid_alt2,res_grid,axis=1)
 
+
  storage_grid = res_grid
+ #print(exDestArr)
  return [storage_cost, storage_DV, storage_out, exDestArr, storage_grid]
 
 
-def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost, 
- storage_out, variation):
+def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost, storage_out, variation):
  """
  Interpolate the values of the decision variables, costs and outputs
 
@@ -212,7 +216,7 @@ def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost,
  Reformat the boundary conditions, decision variables, outputs and
  costs
  """
- if variation: 
+ if variation:
  cond_BC = [True if L<len(measurements_SubSys) else False for L in range(len(storage_DV[0]))]
  cond_DV = [False if L<len(measurements_SubSys) else True for L in range(len(storage_DV[0]))]
  cond_Out = [False if L<len(measurements_SubSys) else True for L in range(len(storage_out[0]))]
@@ -223,24 +227,21 @@ def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost,
  grid_measurements = measurements_SubSys[::-1]
  grid_point_values_costs = np.compress(cond_Costs,storage_cost, axis = 1)
  grid_point_values_out = np.compress(cond_Out,storage_out, axis = 1)
-
+  """
  print("Grid points:")
  print(grid_points)
  print("values:")
  print(grid_point_values)
  print("measurements:")
  print(grid_measurements)
+ """
 
  """ Interpolation of reformatted data """
  try:
  commands = interpolate.griddata(grid_points, grid_point_values,grid_measurements ,method='linear')
  costs = interpolate.griddata(grid_points, grid_point_values_costs,grid_measurements ,method='linear')
  out = interpolate.griddata(grid_points, grid_point_values_out, grid_measurements ,method='linear')
 
- print("commands: " + str(commands))
- print("costs: " + str(costs))
- print("outputs: " + str(out))
-
  # Check if commands are in range, else set to boundary values
  for i, val in enumerate(commands):
  if val < bounds_DVs[i]:
@@ -262,11 +263,15 @@ def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost,
  out = []
 
  for i in range(0,len(storage_DV)):
- commands.append(storage_DV[0,0])
- costs.append(0)
- out.append(0)
+ commands.append(storage_DV[0,2])
+ costs.append(storage_cost[0,2])
+ out.append(storage_out[0,2])
  print('interpolation failed!')
-
+ """
+ print("commands: " + str(storage_DV))
+ print("costs: " + str(storage_cost))
+ print("outputs: " + str(storage_out))
+ """
  else:
  commands = []
  costs = []
@@ -276,5 +281,5 @@ def Interpolation(measurements_SubSys, storage_DV, bounds_DVs, storage_cost,
  commands.append(storage_DV[0,2])
  costs.append(storage_cost[0,2])
  out.append(storage_out[0,2])
-  
- return [commands, costs, out]
+
+ return [commands, costs, out]