模型训练

train_corrector.py

@@ -14,7 +14,7 @@ feature_columns = [
     "T_air", "T_initial", "T_m", "a", "b", "c"
 ]
 
-# 定义定时任务的训练函数1
+# 定义定时任务的训练函数
 def train_and_save_model():
     print("🔄 定时任务开始：重新训练模型...")
 
@@ -26,14 +26,31 @@ def train_and_save_model():
         print(f"❌ 读取CSV失败: {e}")
         return
 
+    # 强制转换为 datetime 类型
+    df["t_real"] = pd.to_datetime(df.get("t_real"), errors="coerce")
+    df["t_formula"] = pd.to_datetime(df.get("t_formula"), errors="coerce")
+
+    # 删除无法转换时间的行
+    before_drop = len(df)
+    df = df.dropna(subset=["t_real", "t_formula"])
+    after_drop = len(df)
+    if before_drop != after_drop:
+        print(f"⚠️ 删除了 {before_drop - after_drop} 行非法时间数据")
+
+    # 计算真实解冻时长（小时）
+    df["t_real_hours"] = (df["t_real"] - df["t_formula"]).dt.total_seconds() / 3600
+
+    # 再次检查并去除无效目标值
+    if df["t_real_hours"].isna().any():
+        invalid_count = df["t_real_hours"].isna().sum()
+        df = df.dropna(subset=["t_real_hours"])
+        print(f"⚠️ 删除了 {invalid_count} 行无效的解冻时长")
+
     # 确保字段类型正确（如果这两列存在）
     for col in ["material_name", "manufactured_goods"]:
         if col in df.columns:
             df[col] = df[col].astype(str)
 
-    # 计算真实解冻时长（小时）
-    df["t_real_hours"] = (df["t_real"] - df["t_formula"]).dt.total_seconds() / 3600
-
     # 检查有没有缺失特征
     missing_features = [col for col in feature_columns if col not in df.columns]
     if missing_features:

参考.py

@@ -3,22 +3,26 @@ from sklearn.model_selection import train_test_split
 from sklearn.ensemble import RandomForestRegressor
 from sklearn.metrics import mean_squared_error
 from scipy.optimize import minimize
+from flask import Flask, request, jsonify
+
+app = Flask(__name__)
 
 
 # 解冻时间计算函数
-def calculate_defrost_time(w, T_initial):
+def calculate_defrost_time(w, data):
     # 物理参数1
-    rho_coal = 1400  # 煤的密度
-    rho_ice = 917  # 冰的密度
-    C_coal = 1000  # 煤的比热容
-    C_ice = 2100
-    L = 334000  # 物体长度（米）
-    k_coal = 0.3  # 煤的导热系数
-    k_ice = 2.2
-    h = 10  # 近似导热系数（W/m²·K）
-    T_air = 20  # 空气温度（摄氏度）
-    T_m = 0  # 熔点温度（摄氏度）
-    a, b, c = 13, 2.72, 1.6  # 物体的尺寸（长、宽、高）
+    rho_ice = data['rho_ice']  # 冰的密度
+    C_ice = data['C_ice']  # 冰的比热容
+    L = data['L']  # 冰的潜热
+    k_ice = data['k_coal']  # 冰的导热系数
+    T_m = data['T_m']  # 熔点温度（摄氏度）
+    rho_coal = data['rho_coal']  # 物料的密度
+    C_coal = data['C_coal']  # 物料的比热容
+    k_coal = data['k_coal']  # 物料的导热系数
+    h = data['h']  # 对流换热系数
+    T_air = data['T_air']  # 空气温度（摄氏度）
+    T_initial = data['T_initial']  # 物料初始温度
+    a, b, c = data['a'], data['b'], data['c']  # 车厢的尺寸（长、宽、高）
     # 计算体积和表面积
     V = a * b * c
     A = 2 * (a * b) + 2 * (b * c) + 2 * (a * c)
@@ -45,68 +49,43 @@ def calculate_defrost_time(w, T_initial):
     return t_total
 
 
-# 模拟生成数据
-np.random.seed(42)
-n_samples = 500
-# 含水量，范围 10% - 15%
-w = np.random.uniform(0.1, 0.15, n_samples)
-# 初始温度，范围 -10 到 20 摄氏度
-T_initial = np.random.uniform(-10, 20, n_samples)
-# 计算解冻时间
-original_times = []
-for i in range(n_samples):
-    original_times.append(calculate_defrost_time(w[i], T_initial[i]))
-
-# 模拟实际作业时长（实际时长和计算时长有一定随机误差）
-y = np.array(original_times) + np.random.normal(0, 0.2, n_samples)
-
-# 准备特征
-X = np.column_stack((w, T_initial, original_times))
-
-# 划分训练集和测试集
-X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
-
-# 选择模型并训练
-model = RandomForestRegressor(n_estimators=100, random_state=42)
-model.fit(X_train, y_train)
-
-# 模型评估
-y_pred = model.predict(X_test)
-mse = mean_squared_error(y_test, y_pred)
-print(f"Mean Squared Error: {mse}")
-
-
 # 反推函数
-def reverse_calculate(time):
+def reverse_calculate(time, data):
     def objective_function(vars):
-        w, T_initial = vars
-        calculated_time = calculate_defrost_time(w, T_initial)
+        w = vars
+        calculated_time = calculate_defrost_time(w, data)
         return abs(calculated_time - time)
 
-    bounds = [(0.1, 0.15), (-10, 20)]  # 含水量和初始温度的范围
-    num_trials = 5  # 尝试的次数
+    bounds = [(0.1, 0.5)]  # 含水量和初始温度的范围
+    num_trials = 10  # 尝试的次数
     best_result = None
     best_error = np.inf
     for _ in range(num_trials):
         # 随机生成初始猜测值
-        initial_guess = [np.random.uniform(0.1, 0.15), np.random.uniform(-10, 20)]
+        initial_guess = [np.random.uniform(0.1, 0.15)]
         result = minimize(objective_function, initial_guess, bounds=bounds)
         error = result.fun
         if error < best_error:
             best_error = error
             best_result = result
-    return best_result.x[0], best_result.x[1]
+    return best_result.x[0]
+
 
+# API 入口
+@app.route('/reverse_calculate', methods=['POST'])
+def handle_request():
+    try:
+        data = request.get_json()
+        time = float(data['realThawTime'])
+        w_result = reverse_calculate(time, data)
+        print(f"实际用时{time}")
+        print(f"含水量{round(w_result, 4) * 100}")
+        return jsonify({
+            "water_content": round(w_result, 4) * 100
+        })
+    except Exception as e:
+        return jsonify({"error": str(e)}), 400
 
-# 输入实际解冻时间
-try:
-    test_time = float(input("请输入实际解冻时间（小时）: "))
-    w_result, T_result = reverse_calculate(test_time)
-    print(f"反推得到的含水量: {w_result}")
-    print(f"反推得到的初始温度: {T_result}")
-    recalculated_time = calculate_defrost_time(w_result, T_result)
-    print(f"重新计算的解冻时间: {recalculated_time}")
-except ValueError:
-    print("输入无效，请输入有效的数字。")
 
-    
+if __name__ == '__main__':
+    app.run(host='127.0.0.1', port=9993, debug=True)