diff options
| -rw-r--r-- | README.rst | 6 | ||||
| -rw-r--r-- | src_py/opcut/csp.py | 67 |
2 files changed, 70 insertions, 3 deletions
@@ -1,9 +1,9 @@ -optcut -====== +opcut +===== `https://opcut.herokuapp.com/` -`optcut` is cutting stock problem optimizer +`opcut` is cutting stock problem optimizer (`https://en.wikipedia.org/wiki/Cutting_stock_problem`) utilizing multiple panels and guillotine cuts (end-to-end cuts). This project includes multiple back-end optimizer implementations and single-page web application front-end. diff --git a/src_py/opcut/csp.py b/src_py/opcut/csp.py new file mode 100644 index 0000000..da8906a --- /dev/null +++ b/src_py/opcut/csp.py @@ -0,0 +1,67 @@ +import collections +import enum + + +InputPanel = collections.namedtuple('InputPanel', ['width', 'height', 'label']) +InputItem = collections.namedtuple('InputItem', ['width', 'height', 'label', + 'rotate']) +OutputPanel = collections.namedtuple('OutputPanel', ['width', 'height', + 'label', 'items']) +OutputItem = collections.namedtuple('OutputItem', ['width', 'height', 'label', + 'x', 'y']) + +Method = enum.Enum('Method', ['FORWARD_GREEDY']) + + +def calculate(panels, items, cut_width, method): + """Calculate cutting stock problem + + Args: + panels (List[InputPanel]): input panels + items (List[InputItem]): input items + cut_width (float): cut width + method (Method): calculation method + + Returns: + List[OutputPanel] + + """ + return { + Method.FORWARD_GREEDY: _calculate_forward_greedy + }[method](panels, items) + + +def _calculate_forward_greedy(panels, items): + pass + + +_Part = collections.namedtuple('_Part', ['panel', 'width', 'height', 'x', 'y']) +_State = collections.namedtuple('_State', ['inputs', 'outputs']) +_fitness_K = 0.03 + + +def _fitness(state): + panel_states = {} + for i in state.inputs: + if i.panel not in panel_states: + panel_states[i.panel] = _State([i], []) + else: + panel_states[i.panel].inputs.append(i) + for i in state.outputs: + if i.panel not in panel_states: + panel_states[i.panel] = _State([], [i]) + else: + panel_states[i.panel].outputs.append(i) + + total_area = sum(i.width * i.height for i in panel_states.keys()) + result = 0 + for panel, panel_state in panel_states.items(): + result += (panel.width * panel.height - + sum(i.width * i.height + for i in panel_state.outputs)) / total_area + result -= (_fitness_K * + min(i.width * i.height for i in panel_state.outputs, + default=0) * + max(i.width * i.height for i in panel_state.inputs, + default=0)) / (total_area * total_area) + return result |