目录
做工厂上位机开发这些年，有个场景我见过太多次：车间主任每周一早上，对着一堆设备日志，手工往Excel里敲数字，统计上周的产量、良品率、各班次的完成情况。整个过程少则半小时，多则两个钟头。数据还不一定准——因为有些设备的日志格式不统一，复制粘贴出错是常事。
这件事，本来可以用一个界面解决。
周期统计界面，说白了就是：按时间维度（班次/日/周/月）聚合生产数据，用图表和表格同时呈现，让管理人员一眼看清楚产线状态。需求不复杂，但实现起来有几个绕不开的坑——数据聚合逻辑怎么写才不乱、Tkinter的Canvas图表性能够不够用、多周期切换时界面怎么刷新不闪烁。这篇文章把这些问题一次说清楚。
🧩 先把需求拆开看
周期统计界面，功能上大概分三块：
- 周期选择器：支持按班次、日、周、月切换，切换后数据自动刷新
- 汇总卡片区：显示总产量、良品率、设备稼动率等核心KPI
- 趋势图 + 明细表：折线图展示趋势，下方表格展示每个周期的明细数据
这三块的刷新逻辑是联动的——切换周期时，三块同时更新。如果设计不好，每次切换都重建所有控件，界面会明显闪烁，用户体验很差。
正确的做法是：控件只建一次，切换周期时只更新数据和重绘图表。这个原则贯穿整个设计。
🏗️ 数据层：周期聚合逻辑
先把数据层写清楚，UI层才好对接。生产数据通常存在SQLite或者CSV里，咱们用一个DataAggregator类封装所有聚合操作。
pythonimport sqlite3
import pandas as pd
from datetime import datetime, timedelta
from typing import Literal
PeriodType = Literal["shift", "day", "week", "month"]
class DataAggregator:
"""
生产数据聚合器
负责按不同时间粒度汇总产量、良品数、设备运行时长
"""
def __init__(self, db_path: str):
self.db_path = db_path
def _fetch_raw(self, start: datetime, end: datetime) -> pd.DataFrame:
conn = sqlite3.connect(self.db_path)
sql = """
SELECT record_time, product_count, good_count,
machine_id, run_minutes
FROM production_log
WHERE record_time BETWEEN ? AND ?
ORDER BY record_time
"""
df = pd.read_sql_query(
sql, conn,
params=(start.strftime("%Y-%m-%d %H:%M:%S"),
end.strftime("%Y-%m-%d %H:%M:%S")),
parse_dates=["record_time"]
)
conn.close()
return df
def get_period_data(self, period: PeriodType,
anchor: datetime) -> pd.DataFrame:
"""
根据周期类型和锚点时间，返回聚合后的DataFrame
列：period_label, total_count, good_count, yield_rate, run_hours
"""
start, end, freq = self._calc_range(period, anchor)
raw = self._fetch_raw(start, end)
if raw.empty:
return pd.DataFrame(columns=[
"period_label", "total_count",
"good_count", "yield_rate", "run_hours"
])
raw.set_index("record_time", inplace=True)
grouped = raw.resample(freq).agg(
total_count=("product_count", "sum"),
good_count=("good_count", "sum"),
run_minutes=("run_minutes", "sum")
).reset_index()
grouped["yield_rate"] = (
grouped["good_count"] / grouped["total_count"]
.replace(0, float("nan"))
).round(4)
grouped["run_hours"] = (grouped["run_minutes"] / 60).round(2)
grouped["period_label"] = grouped["record_time"].apply(
lambda t: self._format_label(t, period)
)
return grouped[[
"period_label", "total_count",
"good_count", "yield_rate", "run_hours"
]]
def _calc_range(self, period: PeriodType,
anchor: datetime) -> tuple:
if period == "shift":
# 按8小时班次，取当天三个班
day_start = anchor.replace(hour=0, minute=0, second=0)
return day_start, day_start + timedelta(hours=24), "8h"
elif period == "day":
week_start = anchor - timedelta(days=anchor.weekday())
week_start = week_start.replace(hour=0, minute=0, second=0)
return week_start, week_start + timedelta(days=7), "D"
elif period == "week":
month_start = anchor.replace(day=1, hour=0, minute=0, second=0)
if month_start.month == 12:
month_end = month_start.replace(year=month_start.year + 1, month=1)
else:
month_end = month_start.replace(month=month_start.month + 1)
return month_start, month_end, "W"
else: # month
year_start = anchor.replace(month=1, day=1, hour=0, minute=0, second=0)
year_end = year_start.replace(year=year_start.year + 1)
return year_start, year_end, "ME"
def _format_label(self, t: datetime, period: PeriodType) -> str:
if period == "shift":
hour = t.hour
if hour < 8: return "夜班 (00-08)"
elif hour < 16: return "早班 (08-16)"
else: return "中班 (16-24)"
elif period == "day": return t.strftime("%m/%d")
elif period == "week": return f"第{t.isocalendar()[1]}周"
else: return t.strftime("%Y-%m")
def get_summary(self, period: PeriodType,
anchor: datetime) -> dict:
"""返回当前周期的汇总KPI"""
df = self.get_period_data(period, anchor)
if df.empty:
return {"total": 0, "good": 0,
"yield_rate": 0.0, "run_hours": 0.0}
return {
"total": int(df["total_count"].sum()),
"good": int(df["good_count"].sum()),
"yield_rate": round(df["good_count"].sum() /
max(df["total_count"].sum(), 1) * 100, 2),
"run_hours": round(df["run_hours"].sum(), 1)
}
get_period_data和get_summary是UI层唯一需要调用的两个接口。数据怎么聚合、时间范围怎么算——这些细节全封在里面，UI层不用操心。
🎨 界面层：周期选择器 + KPI卡片
界面整体用ttk.Notebook分Tab，或者直接用Frame堆叠都行。这里选更灵活的Frame方案，方便后续嵌入到更大的监控主界面里。
pythonimport tkinter as tk
from tkinter import ttk
from datetime import datetime
class PeriodSelector(tk.Frame):
"""
周期切换器：班次 / 日 / 周 / 月
切换时触发回调，通知父界面刷新数据
"""
PERIODS = [
("班次", "shift"),
("按日", "day"),
("按周", "week"),
("按月", "month"),
]
def __init__(self, parent, on_change_callback, **kwargs):
super().__init__(parent, bg="#ECEFF1", **kwargs)
self._callback = on_change_callback
self._var = tk.StringVar(value="day")
self._build()
def _build(self):
tk.Label(self, text="统计周期：",
font=("微软雅黑", 10),
bg="#ECEFF1", fg="#546E7A").pack(side="left", padx=(10, 4))
for label, value in self.PERIODS:
rb = tk.Radiobutton(
self, text=label, value=value,
variable=self._var,
font=("微软雅黑", 10),
bg="#ECEFF1", fg="#37474F",
activebackground="#CFD8DC",
selectcolor="#B0BEC5",
command=self._on_change
)
rb.pack(side="left", padx=6, pady=6)
def _on_change(self):
self._callback(self._var.get())
@property
def current(self) -> str:
return self._var.get()
KPI卡片区，每张卡片显示一个指标。颜色用参数传入，方便根据数值动态变色（比如良品率低于95%时变红）。
pythonclass KpiCard(tk.Frame):
"""单个KPI卡片"""
def __init__(self, parent, title: str,
accent_color: str = "#1565C0", **kwargs):
super().__init__(parent, bg="white",
relief="flat", bd=1, **kwargs)
self._accent = accent_color
self._title = title
# 顶部色条
bar = tk.Frame(self, bg=accent_color, height=4)
bar.pack(fill="x")
tk.Label(self, text=title,
font=("微软雅黑", 9), fg="#78909C",
bg="white").pack(pady=(10, 2))
self._value_label = tk.Label(
self, text="--",
font=("微软雅黑", 22, "bold"),
fg=accent_color, bg="white"
)
self._value_label.pack()
self._unit_label = tk.Label(
self, text="",
font=("微软雅黑", 9), fg="#90A4AE",
bg="white"
)
self._unit_label.pack(pady=(0, 10))
def update(self, value: str, unit: str = "",
alert: bool = False):
color = "#F44336" if alert else self._accent
self._value_label.config(text=value, fg=color)
self._unit_label.config(text=unit)
📈 图表绘制：用Canvas画折线图
Tkinter没有内置图表控件。方案有两个：用matplotlib嵌入，或者自己用Canvas画。我更倾向于后者——matplotlib嵌入后启动慢、内存占用高，在工控现场那种配置不高的工控机上，明显感觉卡。自己用Canvas画，轻量，刷新也快。
pythonclass TrendChart(tk.Canvas):
"""
轻量折线图，基于Canvas绘制
支持多系列、坐标轴、网格线、数据点悬停提示
"""
PADDING = {"left": 55, "right": 20, "top": 20, "bottom": 40}
def __init__(self, parent, **kwargs):
kwargs.setdefault("bg", "#FAFAFA")
kwargs.setdefault("highlightthickness", 0)
super().__init__(parent, **kwargs)
self._data: list[dict] = [] # [{"label":..., "values":..., "color":...}]
self._labels: list[str] = []
self._tooltip = None
self.bind("<Configure>", lambda e: self._redraw())
self.bind("<Motion>", self._on_mouse_move)
self.bind("<Leave>", self._hide_tooltip)
def set_data(self, labels: list[str], series: list[dict]):
"""
series格式：[{"name": "产量", "values": [100, 120, ...], "color": "#1565C0"}]
"""
self._labels = labels
self._data = series
self._redraw()
def _redraw(self):
self.delete("all")
if not self._data or not self._labels:
self._draw_empty()
return
w = self.winfo_width()
h = self.winfo_height()
if w < 10 or h < 10:
return
p = self.PADDING
chart_w = w - p["left"] - p["right"]
chart_h = h - p["top"] - p["bottom"]
# 计算Y轴范围
all_vals = [v for s in self._data for v in s["values"] if v is not None]
if not all_vals:
self._draw_empty()
return
y_min, y_max = min(all_vals), max(all_vals)
y_range = max(y_max - y_min, 1)
y_min -= y_range * 0.1
y_max += y_range * 0.1
n = len(self._labels)
x_step = chart_w / max(n - 1, 1)
def to_canvas(i, val):
cx = p["left"] + i * x_step
cy = p["top"] + chart_h * (1 - (val - y_min) / (y_max - y_min))
return cx, cy
# 网格线
for k in range(5):
y_val = y_min + (y_max - y_min) * k / 4
_, cy = to_canvas(0, y_val)
self.create_line(p["left"], cy, w - p["right"], cy,
fill="#E0E0E0", dash=(4, 4))
self.create_text(p["left"] - 6, cy,
text=f"{y_val:.0f}",
anchor="e", font=("微软雅黑", 8),
fill="#90A4AE")
# X轴标签
for i, label in enumerate(self._labels):
cx, _ = to_canvas(i, y_min)
self.create_text(cx, h - p["bottom"] + 12,
text=label, anchor="n",
font=("微软雅黑", 8), fill="#78909C")
# 坐标轴
self.create_line(p["left"], p["top"],
p["left"], h - p["bottom"],
fill="#B0BEC5", width=1)
self.create_line(p["left"], h - p["bottom"],
w - p["right"], h - p["bottom"],
fill="#B0BEC5", width=1)
# 折线 + 数据点
self._points_map = [] # 用于悬停检测
for series in self._data:
color = series.get("color", "#1565C0")
values = series["values"]
coords = []
for i, val in enumerate(values):
if val is None:
continue
cx, cy = to_canvas(i, val)
coords.extend([cx, cy])
self._points_map.append((cx, cy, val, series["name"]))
if len(coords) >= 4:
self.create_line(*coords, fill=color,
width=2, smooth=True)
for cx, cy, val, _ in self._points_map:
self.create_oval(cx-4, cy-4, cx+4, cy+4,
fill=color, outline="white", width=1.5)
def _draw_empty(self):
w, h = self.winfo_width(), self.winfo_height()
self.create_text(w // 2, h // 2, text="暂无数据",
font=("微软雅黑", 11), fill="#BDBDBD")
def _on_mouse_move(self, event):
if not hasattr(self, "_points_map"):
return
for cx, cy, val, name in self._points_map:
if abs(event.x - cx) < 10 and abs(event.y - cy) < 10:
self._show_tooltip(event.x, event.y,
f"{name}: {val:.0f}")
return
self._hide_tooltip(None)
def _show_tooltip(self, x, y, text):
self._hide_tooltip(None)
self._tooltip = self.create_rectangle(
x + 8, y - 18, x + 8 + len(text) * 7 + 10, y + 2,
fill="#37474F", outline=""
)
self._tooltip_text = self.create_text(
x + 13, y - 8, text=text,
anchor="w", font=("微软雅黑", 8), fill="white"
)
def _hide_tooltip(self, _):
if self._tooltip:
self.delete(self._tooltip)
self.delete(self._tooltip_text)
self._tooltip = None
鼠标悬停到数据点上会出现小提示框，这个细节做出来之后，用户反馈体验好了不少——毕竟看图的人不只是开发者，车间主任和品质工程师也要用，数值能直接看到比什么都强。
🔗 整合：主界面组装
把上面的模块拼起来，加上明细数据表格：
pythonclass ProductionStatsPanel(tk.Frame):
"""
生产数据周期统计主面板
包含：周期选择器 / KPI卡片 / 趋势图 / 明细表格
"""
def __init__(self, parent, db_path: str, **kwargs):
super().__init__(parent, bg="#F5F5F5", **kwargs)
self.aggregator = DataAggregator(db_path)
self.anchor = datetime.now()
self._build_ui()
self._refresh("day") # 默认按日视图
def _build_ui(self):
# ── 顶部：周期选择器 ────────────────────
self.selector = PeriodSelector(self, self._refresh)
self.selector.pack(fill="x", padx=10, pady=(10, 0))
# ── KPI卡片区 ────────────────────────────
card_frame = tk.Frame(self, bg="#F5F5F5")
card_frame.pack(fill="x", padx=10, pady=8)
kpi_defs = [
("总产量", "#1565C0"),
("良品数", "#2E7D32"),
("综合良率", "#6A1B9A"),
("运行时长", "#E65100"),
]
self.kpi_cards = {}
for title, color in kpi_defs:
card = KpiCard(card_frame, title, accent_color=color)
card.pack(side="left", fill="both",
expand=True, padx=5)
self.kpi_cards[title] = card
# ── 图表 ─────────────────────────────────
chart_frame = tk.LabelFrame(
self, text=" 产量趋势 ",
font=("微软雅黑", 10),
bg="#F5F5F5", fg="#546E7A"
)
chart_frame.pack(fill="both", expand=True,
padx=10, pady=(0, 6))
self.chart = TrendChart(chart_frame, height=200)
self.chart.pack(fill="both", expand=True, padx=8, pady=8)
# ── 明细表格 ─────────────────────────────
table_frame = tk.LabelFrame(
self, text=" 分周期明细 ",
font=("微软雅黑", 10),
bg="#F5F5F5", fg="#546E7A"
)
table_frame.pack(fill="x", padx=10, pady=(0, 10))
cols = ("周期", "总产量", "良品数", "良率(%)", "运行(h)")
self.tree = ttk.Treeview(
table_frame, columns=cols,
show="headings", height=6
)
for col in cols:
self.tree.heading(col, text=col)
self.tree.column(col, width=110, anchor="center")
scrollbar = ttk.Scrollbar(table_frame, orient="vertical",
command=self.tree.yview)
self.tree.configure(yscrollcommand=scrollbar.set)
self.tree.pack(side="left", fill="x",
expand=True, padx=8, pady=8)
scrollbar.pack(side="right", fill="y", pady=8)
# 隔行变色
self.tree.tag_configure("odd", background="#FAFAFA")
self.tree.tag_configure("even", background="#E8F5E9")
def _refresh(self, period: str):
summary = self.aggregator.get_summary(period, self.anchor)
df = self.aggregator.get_period_data(period, self.anchor)
# 更新KPI卡片
self.kpi_cards["总产量"].update(
f"{summary['total']:,}", "件")
self.kpi_cards["良品数"].update(
f"{summary['good']:,}", "件")
yr = summary["yield_rate"]
self.kpi_cards["综合良率"].update(
f"{yr}", "%", alert=(yr < 95.0))
self.kpi_cards["运行时长"].update(
f"{summary['run_hours']}", "h")
# 更新图表
if not df.empty:
self.chart.set_data(
labels=df["period_label"].tolist(),
series=[
{"name": "总产量", "color": "#1565C0",
"values": df["total_count"].tolist()},
{"name": "良品数", "color": "#2E7D32",
"values": df["good_count"].tolist()},
]
)
# 更新表格（清空后重填，不重建控件）
for row in self.tree.get_children():
self.tree.delete(row)
for i, row in df.iterrows():
yr_val = f"{row['yield_rate'] * 100:.2f}" \
if pd.notna(row["yield_rate"]) else "--"
tag = "even" if i % 2 == 0 else "odd"
self.tree.insert("", "end", values=(
row["period_label"],
f"{int(row['total_count']):,}",
f"{int(row['good_count']):,}",
yr_val,
f"{row['run_hours']:.1f}"
), tags=(tag,))
注意_refresh里的表格更新——用delete清空再重填，而不是销毁Treeview重建。这个区别在数据量大的时候非常明显，重建控件会有明显的闪烁感，清空重填则几乎感觉不到刷新。
需要初使化db
pythonimport sqlite3
from datetime import datetime, timedelta
db_path = "production.db"
conn = sqlite3.connect(db_path)
cursor = conn.cursor()
# 创建表
cursor.execute("""
CREATE TABLE IF NOT EXISTS production_log (
record_time DATETIME NOT NULL, product_count INTEGER NOT NULL, good_count INTEGER NOT NULL, machine_id TEXT NOT NULL, run_minutes INTEGER NOT NULL)
""")
# 清空旧数据
cursor.execute("DELETE FROM production_log")
# 使用当前日期及前7天的测试数据
base_date = datetime.now().replace(hour=0, minute=0, second=0)
test_data = []
for day_offset in range(7):
date = base_date - timedelta(days=day_offset)
for shift_hour in [0, 8, 16]:
test_data.append((
date.replace(hour=shift_hour).strftime("%Y-%m-%d %H:%M:%S"),
100 + day_offset * 10,
90 + day_offset * 8,
f"M{shift_hour//8 + 1}",
480
))
cursor.executemany("""
INSERT INTO production_log (record_time, product_count, good_count, machine_id, run_minutes)
VALUES (?, ?, ?, ?, ?)
""", test_data)
conn.commit()
conn.close()
print("✓ 测试数据已插入（最近7天）")
⚠️ 三个容易踩的坑
坑一：Canvas尺寸在pack之后才确定。 刚pack完就调winfo_width()，拿到的是1（或者0）。正确做法是绑定<Configure>事件，在事件触发后再绘图——上面的TrendChart已经这样处理了。
坑二：pandas的resample对时区敏感。 如果数据库里存的时间戳带时区信息，resample可能报错或者结果不对。统一在读取时用pd.to_datetime(..., utc=False)，或者在存入时就去掉时区信息，保持一致。
坑三：ttk.Treeview的隔行变色在Windows上默认主题下不生效。 需要先用ttk.Style().theme_use("clam")切换主题，tag_configure才能正常渲染背景色。加一行ttk.Style().theme_use("clam")在程序初始化时调用即可。
📌 小结
这套周期统计界面的核心设计思路，可以用三句话概括：
数据层和UI层彻底分离，DataAggregator只管聚合逻辑，界面控件只管渲染——这让后续替换数据源（从SQLite换成MySQL，或者接实时OPC-UA数据）变得非常简单。Canvas自绘图表在工控场景下比嵌入matplotlib更实用，轻量、启动快、刷新流畅。控件复用而非重建，是解决界面闪烁问题最直接的手段。
完整源码已开源，供学习参考。如果你在实际项目中遇到过多周期切换的性能问题，或者有更好的Canvas图表绘制思路，欢迎在评论区交流。
#Python #Tkinter #工控开发 #生产数据统计 #上位机开发
本文作者:技术老小子
本文链接:
版权声明:本博客所有文章除特别声明外，均采用 BY-NC-SA 许可协议。转载请注明出处！
目录