-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathplacement_worker.py
571 lines (455 loc) · 24.6 KB
/
placement_worker.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
# -*- coding: utf-8 -*-
import json
from nfp_utls import almost_equal, rotate_polygon, get_polygon_bounds, polygon_area, is_convex, D_function, edge_point_distance
from settings import BIN_HEIGHT, MAX_MOVE_X, MAX_MOVE_Y, SCALING_FACTOR
import copy
import pyclipper
import time
import cvxpy as cp
TOL = 0.00001 * SCALING_FACTOR
class PlacementWorker():
def __init__(self, bin_polygon, paths, ids, rotations, config, nfp_cache):
self.bin_polygon = bin_polygon
self.paths = copy.deepcopy(paths)
self.ids = ids # 图形原来的ID顺序
self.rotations = rotations
self.config = config
self.nfpCache = copy.deepcopy(nfp_cache) or {}
def find_constraints(self,A,B,i,j,x,y,placements):
# -*- coding: utf-8 -*-
"""
Created on Thu Sep 5 11:22:11 2019
@author: chend
"""
'''
param: A is a stationary polygon
B is the obitating polygon
i,j 当前寻找约束的两个polygon在当前排序中的位置，从0开始，-1表示bin
x,y LP的决策变量组，为points的第一个点的真实坐标，例如x[i]表示第i个块料的第一个点的横轴位置
output: [] constraints的列表
'''
constraints = []
key = {
'A': A['p_id'],
'B': B['p_id'],
'inside': False,
'A_rotation': A['rotation'],
'B_rotation': B['rotation']
} #nfp是按 B的第一个点绕出来的
json_key = json.dumps(key)
nfp = self.nfpCache[json_key][0]
nfp_current_location = [{'x':point['x']+placements[i]['x'] , 'y':point['y']+placements[i]['y']} for point in nfp]
ref_point_B = B['current_location'][0]
# nfp逆时针，当ref_point_B在nfp上edge的右边时，便是valid_edge,找出所有valid_edge
valid_edge = []
valid_distance = []
valid_point_index = []
for k in range(0,len(nfp_current_location)):
point_1 = nfp_current_location[k-1]
point_2 = nfp_current_location[k]
D_func_value = D_function(point_1,point_2,ref_point_B)
dis = edge_point_distance([point_1,point_2],ref_point_B)
#D_func为负数说明在右边,0表示在向量上;由于精度问题，这里距离足够小的话，我们也认为是在外面
if (D_func_value <= 0 + TOL): #or (dis <= 0 + TOL):
valid_edge.append([{'x':point_1['x'],'y':point_1['y']}, {'x':point_2['x'],'y':point_2['y']}])
valid_distance.append( edge_point_distance(valid_edge[-1],ref_point_B) )
valid_point_index.append([k-1,k])
#有一点点由于精度引起的重叠，加一个constraint要求距离不能继续变大
if len(valid_edge) == 0:
min_dis,point_index = min( (edge_point_distance([nfp_current_location[k-1],nfp_current_location[k]] , ref_point_B),k)
for k in range(len(nfp_current_location)) )
print(u'(%s,%s) ref_point在nfp内部了'%(A['p_id'],B['p_id']),
u'离最近边的距离为:', min_dis )
print()
constraints.append( D_function(nfp_current_location[point_index-1],nfp_current_location[point_index],
{'x':x[j],'y':y[j]}) <=
D_function(nfp_current_location[point_index-1],nfp_current_location[point_index],
ref_point_B))
return constraints
elif len(valid_edge) > 0:
if is_convex(nfp_current_location):
max_distance = max(valid_distance)
#目前不在nfp上
if max_distance > 0:
max_distance_index = valid_distance.index(max_distance)
point_1 = valid_edge[max_distance_index][0]
point_2 = valid_edge[max_distance_index][1]
constraints.append( D_function(point_1,point_2,{'x':x[j],'y':y[j]}) <= 0 )
#目前在 nfp上，在nfp上的话，只可能在vertex上，或在edge上
elif max_distance == 0:
#在vertex上
if len(valid_edge) == 1:
edge = valid_edge[0]
constraints.append( D_function(edge[0],edge[1],{'x':x[j],'y':y[j]}) <= 0 )
#在edge上
if len(valid_edge) == 2:
edge_1 = valid_edge[0]
if edge_1[0]['x'] == edge_1[1]['x']:
slope_1 = 1000000
else:
slope_1 = (edge_1[1]['y'] - edge_1[0]['y'])/(edge_1[1]['x']-edge_1[0]['x'])
#slope_2
edge_2 = valid_edge[1]
if edge_2[0]['x'] == edge_2[1]['x']:
slope_2 = 1000000
else:
slope_2 = (edge_2[1]['y'] - edge_2[0]['y'])/(edge_2[1]['x']-edge_2[0]['x'])
if abs(slope_1) <= abs(slope_2):
constraints.append( D_function(edge_1[0],edge_1[1],{'x':x[j],'y':y[j]}) <= 0 )
else:
constraints.append( D_function(edge_2[0],edge_2[1],{'x':x[j],'y':y[j]}) <= 0 )
if not is_convex(nfp_current_location):
max_distance = max(valid_distance)
edge = valid_edge[ valid_distance.index(max_distance) ]
point_1_index,point_2_index = valid_point_index[valid_distance.index(max_distance)][0],valid_point_index[valid_distance.index(max_distance)][1]
#加入edge构成的约束
constraints.append( D_function(edge[0],edge[1],{'x':x[j],'y':y[j]}) <= 0 )
if len(edge) is not 0:
ON_RIGHT = True
else:
ON_RIGHT = False
#逆时针寻找，下一条边的终点在edge右侧的，则加入下一条边构成的约束
while ON_RIGHT:
if point_2_index < len(nfp_current_location)-1:
point_3_index = point_2_index + 1
if point_2_index == len(nfp_current_location)-1:
point_3_index = 0
point_1 = nfp_current_location[point_1_index]
point_2 = nfp_current_location[point_2_index]
point_3 = nfp_current_location[point_3_index]
ON_RIGHT = D_function( point_1,point_2,point_3) <= 0
point_1_index = point_2_index
point_1 = nfp_current_location[point_1_index]
point_2_index = point_3_index
point_2 = nfp_current_location[point_2_index]
if point_3_index == len(nfp_current_location)-1:
point_3_index = 0
point_3 = nfp_current_location[point_3_index]
if ON_RIGHT:
constraints.append( D_function(point_1,point_2,{'x':x[j],'y':y[j]}) <= 0 )
#顺时针寻找，下一条边的终点在上一条右侧的，则加入这条边构成的约束
point_1_index = valid_point_index[valid_distance.index(max_distance)][0]
point_2_index = valid_point_index[valid_distance.index(max_distance)][1]
point_0_index = point_1_index - 1
point_0 = nfp_current_location[point_0_index]
point_1 = nfp_current_location[point_1_index]
point_2 = nfp_current_location[point_2_index]
if len(edge) is not 0:
ON_RIGHT = True
while ON_RIGHT:
ON_RIGHT = D_function( point_0,point_1,point_2) <= 0
point_0_index = point_0_index - 1
point_1_index = point_1_index - 1
point_2_index = point_2_index - 1
point_0 = nfp_current_location[point_0_index]
point_1 = nfp_current_location[point_1_index]
point_2 = nfp_current_location[point_2_index]
if ON_RIGHT:
constraints.append( D_function(point_1,point_2,{'x':x[j],'y':y[j]}) <= 0 )
return constraints
def compact_layout(self,current_layout):
# -*- coding: utf-8 -*-
"""
Created on Thu Sep 5 11:22:11 2019
@author: chend
"""
'''
Param: current_layout = {'placed':[], 'placements':[]} 其中placed是经旋转的块料
output: placement: [{'x': ,'y':},{},....]
'''
placed = copy.deepcopy(current_layout['placed'])
placements = copy.deepcopy(current_layout['placements'])
if len(placed)<2 and len(placements)<2:
return placements
for i in range(0,len(placed)):
placed[i]['current_location'] = [{'x':point['x']+placements[i]['x'],
'y':point['y']+placements[i]['y']} for point in placed[i]['points']]
# Linear Programming
#variable
x = cp.Variable(shape=len(placed))#integer=True
y = cp.Variable(shape=len(placed))
z = cp.Variable(shape=1)
#obj
obj = cp.Minimize(z)
#constraints
constraints = []
constraints += [x[i]>=0 for i in range(len(placed))]
constraints += [y[i]>=0 for i in range(len(placed))]
constraints += [z>=0]
# i是固定的块， j是待移动方块
for j in range(0,len(placed)):
B = placed[j]
# 每个块的移动不能超过一定距离
ref_point = B['points'][0]
distance_from_ref_point_to_part_right = B['x'] + B['width'] - ref_point['x']
distance_from_ref_point_to_part_left = ref_point['x'] - B['x']
distance_from_ref_point_to_part_top = B['y'] + B['width'] - ref_point['y']
distance_from_ref_point_to_part_bottom= ref_point['y'] - B['y']
# x-axis
current_x = ref_point['x'] + placements[j]['x']
constraints.append( x[j] - current_x <= MAX_MOVE_X )
constraints.append( x[j] - current_x >= -MAX_MOVE_X )
# y-axis
current_y = ref_point['y'] + placements[j]['y']
constraints.append( y[j] - current_y <= MAX_MOVE_Y )
constraints.append( y[j] - current_y >= -MAX_MOVE_Y )
# 目标函数 z >= x_i + W_i, W_i是当前ref点到该块料最右侧距离
constraints.append( z >= x[j] + distance_from_ref_point_to_part_right )
# 每个块料都需要在 bin_offset里面
bin_offset = get_polygon_bounds(self.bin_polygon['points_offset'])
constraints.append( x[j] - distance_from_ref_point_to_part_left >= bin_offset['x'] )
constraints.append( y[j] + distance_from_ref_point_to_part_top <= bin_offset['y'] + bin_offset['width'] )
constraints.append( y[j] - distance_from_ref_point_to_part_bottom >= bin_offset['y'] )
for i in range(0,j):
A = placed[i]
constraints += self.find_constraints(A,B,i,j,x,y,placements)
# compaction problem
prob = cp.Problem(obj,constraints)
prob.solve(solver = cp.CVXOPT)#,**{'cplex_filename':'a.lp'}
print('x:',x.value)
print('y:',y.value)
if prob.status == 'optimal':
best_location = [{'x':x_loc,'y':y_loc} for x_loc,y_loc in zip(x.value,y.value)]
shift_vector = [{'x': best_location[k]['x'] - placed[k]['current_location'][0]['x'],
'y': best_location[k]['y'] - placed[k]['current_location'][0]['y']}
for k in range(0,len(placed))]
placements = [{'x':placements[k]['x'] + shift_vector[k]['x'],
'y':placements[k]['y'] + shift_vector[k]['y'],
'p_id':placements[k]['p_id'],
'rotation':placements[k]['rotation']}
for k in range(0,len(placements))]
print(u'找到最优解，已压缩一次\n')
elif prob.status == 'infeasible':
print(u'压缩模型无解,无法压缩\n')
else:
print(u'压缩模型出错\n')
return placements
def place_paths(self):
start = time.clock()
# 排列图形
if self.bin_polygon is None:
return None
# rotate paths by given rotation
rotated = list()
for i in range(0, len(self.paths)):
r = copy.deepcopy(self.paths[i][1])
#旋转本体
rot_return = rotate_polygon(self.paths[i][1]['points'], self.paths[i][2])
r['points'] = rot_return['points']
r['x'] = rot_return['x']
r['y'] = rot_return['y']
r['width'] = rot_return['width']
r['height'] = rot_return['height']
#旋转 offset 和 convex_hull
r['points_offset'] = rotate_polygon(self.paths[i][1]['points_offset'], self.paths[i][2])['points']
r['convex_hull'] = rotate_polygon(self.paths[i][1]['convex_hull'], self.paths[i][2])['points']
#其他信息
r['rotation'] = self.paths[i][2]
r['source'] = self.paths[i][1]['p_id']
r['p_id'] = self.paths[i][0]
rotated.append(r)
paths = rotated
# 保存所有转移数据
all_placements = list()
orders = list()
rotations = list()
# 基因组的适应值
fitness = []
bin_area = abs(polygon_area(self.bin_polygon['points']))
min_width = None
while len(paths) > 0:
placed = list()
placements = list()
# add 1 for each new bin opened (lower fitness is better)
#fitness += 1
for i in range(0, len(paths)):
path = paths[i]
# 图形的坐标
key = json.dumps({
'A': '-1',
'B': path['p_id'],
'inside': True,
'A_rotation': 0,
'B_rotation': path['rotation']
})
binNfp = self.nfpCache.get(key)
if binNfp is None or len(binNfp) == 0:
continue
# part unplaceable, skip
error = False
# ensure all necessary NFPs exist
for p in placed:
key = json.dumps({
'A': p['p_id'],
'B': path['p_id'],
'inside': False,
'A_rotation': p['rotation'],
'B_rotation': path['rotation']
})
nfp = self.nfpCache.get(key)
if nfp is None:
error = True
break
# part unplaceable, skip
if error:
continue
#把第一个图形放到布料左下角
position = None
if len(placed) == 0:
for j in range(0, len(binNfp)):
for k in range(0, len(binNfp[j])):
if position is None or (binNfp[j][k]['x']-path['points'][0]['x'] < position['x']):
position = {
'x': binNfp[j][k]['x'] - path['points'][0]['x'],
'y': binNfp[j][k]['y'] - path['points'][0]['y'],
'p_id': path['p_id'],
'rotation': path['rotation']
}
placements.append(position)
placed.append(path)
continue
clipper_bin_nfp = list()
for j in range(0, len(binNfp)):
clipper_bin_nfp.append([[p['x'], p['y']] for p in binNfp[j]])
# 找出 待放置方块 和 已放置方块 的 combined_NFP
clipper = pyclipper.Pyclipper()
for j in range(0, len(placed)):
p = placed[j]
key = json.dumps({
'A': p['p_id'],
'B': path['p_id'],
'inside': False,
'A_rotation': p['rotation'],
'B_rotation': path['rotation']
})
nfp = self.nfpCache.get(key)
if nfp is None:
continue
for k in range(0, len(nfp)):
clone = [[np['x'] + placements[j]['x'], np['y'] + placements[j]['y']] for np in nfp[k]]
clone = pyclipper.CleanPolygon(clone)
if len(clone) > 2:
clipper.AddPath(clone, pyclipper.PT_SUBJECT, True)
combine_nfp = clipper.Execute(pyclipper.CT_UNION, pyclipper.PFT_NONZERO, pyclipper.PFT_NONZERO)
if len(combine_nfp) == 0:
continue
# 以 combined_NFP 为 夹子(clip) 夹掉 待放方块和画布的内部NFP
clipper = pyclipper.Pyclipper()
clipper.AddPaths(combine_nfp, pyclipper.PT_CLIP, True)
try:
clipper.AddPaths(clipper_bin_nfp, pyclipper.PT_SUBJECT, True)
except:
print(u'图形坐标出错', clipper_bin_nfp)
# choose placement that results in the smallest bounding box
finalNfp = clipper.Execute(pyclipper.CT_DIFFERENCE, pyclipper.PFT_NONZERO, pyclipper.PFT_NONZERO)
if len(finalNfp) == 0:
continue
#这里clean nfp后会有很多重叠的地方，不能clean
#finalNfp = pyclipper.CleanPolygons(finalNfp)
for j in range(len(finalNfp)-1, -1, -1):
if len(finalNfp[j]) < 3:
finalNfp.pop(j)
if len(finalNfp) == 0:
continue
finalNfp = [[{'x': p[0], 'y': p[1]}for p in polygon] for polygon in finalNfp]
min_width = None
min_area = None
min_x = None
#all_points 是已摆好的方块的转移后摆好的坐标点集合
all_points = list()
for m in range(0, len(placed)):
for p in placed[m]['points']:
all_points.append({
'x': p['x']+placements[m]['x'],
'y': p['y']+placements[m]['y']
})
# 生成nfp多边形
for nf in finalNfp:
if abs(polygon_area(nf)) < 2:
continue
for p_nf in nf:
all_points_deepcopy = copy.deepcopy(all_points)
# path 坐标
shift_vector = {
'x': p_nf['x'] - path['points'][0]['x'],
'y': p_nf['y'] - path['points'][0]['y'],
'p_id': path['p_id'],
'rotation': path['rotation'],
}
# 找新坐标后的最小矩形
for m in range(0, len(path['points'])):
all_points_deepcopy.append({
'x': path['points'][m]['x'] + shift_vector['x'],
'y': path['points'][m]['y'] + shift_vector['y']
})
all_points_deepcopy.append({'x':0,'y':0})
rect_bounds = get_polygon_bounds(all_points_deepcopy)
# weigh width more, to help compress in direction of gravity
# ??? width*2 不就是对width做更多的惩罚吗？这样图形会尽量往左下边放
area = rect_bounds['height'] + rect_bounds['width']
if (min_area is None or area < min_area or almost_equal(min_area, area)) and (
min_x is None or shift_vector['x'] <= min_x):
min_area = area
min_width = rect_bounds['width']
position = shift_vector
min_x = shift_vector['x']
if position:
placed.append(path)
placements.append(position)
for p in placed:
p_id = paths.index(p)
if p_id >= 0:
paths.pop(p_id)
if placements and len(placements) > 0:
all_placements.append(placements)
else:
# something went wrong
break
placed_area = 0
all_points = []
for k,path in enumerate(placed):
placed_area += abs(polygon_area(path['points']))
for point in path['points']:
all_points.append({
'x': point['x'] + placements[k]['x'],
'y': point['y'] + placements[k]['y']
})
all_points.append({'x':0,'y':0})
min_rect_area = get_polygon_bounds(all_points)['width']*BIN_HEIGHT
#保存第一次heuristic排列得到的结果
all_placements.append(placements)
orders.append([ self.paths[i][0] for i in range(len(self.paths))])
rotations.append([self.rotations ] )
fitness.append(placed_area/min_rect_area)
print('由Heuristic得到的最小包络矩阵利用率: %.4f'%fitness[0])
print('排列一次耗时:%.1f'%(time.clock()-start),'\n-----------\n' )
print('排列得到：',all_placements)
#压缩
IMPROVED = True
while IMPROVED:
start = time.clock()
placements = self.compact_layout({'placed':placed,'placements':placements})
#new_fitness
placed_area = 0
all_points = []
for k,path in enumerate(placed):
placed_area += abs(polygon_area(path['points']))
for point in path['points']:
all_points.append({
'x': point['x'] + placements[k]['x'],
'y': point['y'] + placements[k]['y']
})
all_points.append({'x':0,'y':0})
min_rect_area = get_polygon_bounds(all_points)['width']*BIN_HEIGHT
new_fitness = placed_area/min_rect_area
if new_fitness > fitness[-1]:
print(u'已压缩现有排列方式,新排列方式的利用率为: %.4f' %new_fitness)
print('压缩一次用时: %.2f' %(time.clock()-start),'\n-------------\n')
fitness.append(new_fitness)
all_placements.append(placements)
else:
IMPROVED = False
print('压缩完成，已无法继续压缩')
print('压缩总用时:%.2f' %(time.clock()-start))
print('全部placements:',all_placements)
return {'placements': all_placements, 'fitness': fitness, 'paths': paths, 'area': bin_area}