feat: 新增 Physical_Only 预处理分支 — 纯物理特征+标准化，与 MNF A/B 对比

spectral_Preprocessing.py:
- 新增 Physical_Only 分支: Pipeline[PhysicalFeatureExtractor + StandardScaler]
- 纯物理特征流: 20个指数 → StandardScaler → SVR
- DualStream_MNF 逻辑完全不变

modeling_batch.py:
- preprocessing_methods 新增 Physical_Only
- 训练时 Physical_Only 也只取纯光谱50列
- 波长列表传递覆盖 Physical_Only

GUI:
- 两个面板同步新增「纯物理特征 (指数+标准化)」选项

src/core/modeling/modeling_batch.py

@@ -241,7 +241,7 @@ class WaterQualityModelingBatch:
         # 预处理方法列表
         self.preprocessing_methods = [
             "None", "MMS", "SS", "CT", "SNV", "MA", "SG", "MSC", "D1", "D2", "DT", "WVAE",
-            "DualStream_MNF"
+            "DualStream_MNF", "Physical_Only"
         ]
 
         # 样本划分方法列表
@@ -609,10 +609,10 @@ class WaterQualityModelingBatch:
         # ═══════════════════════════════════════════════════════════
         # ★ DualStream_MNF：只取纯光谱列，WQI 由 Pipeline 内 PhysicalExtractor 动态计算
         # ═══════════════════════════════════════════════════════════
-        if preprocess_method == "DualStream_MNF":
+        if preprocess_method in ("DualStream_MNF", "Physical_Only"):
             _spec_cols = [c for c in X_raw.columns if self._is_wavelength_column(c)]
             X_raw = X_raw[_spec_cols]
-            print(f"[DualStream_MNF] 精简为纯光谱: {X_raw.shape[1]} 列 "
+            print(f"[{preprocess_method}] 精简为纯光谱: {X_raw.shape[1]} 列 "
                   f"({X_raw.columns[0]} ~ {X_raw.columns[-1]} nm)")
 
         # 使用指定方法分割训练集和测试集
@@ -637,9 +637,9 @@ class WaterQualityModelingBatch:
             base_model.set_params(verbose=-1)
 
         # ============ 关键：把预处理器塞进 Pipeline ============
-        # DualStream_MNF 需要传入波长列表，供 PhysicalFeatureExtractor 定位波段
+        # DualStream_MNF / Physical_Only 需要波长列表，供 PhysicalFeatureExtractor 定位波段
         _wl_list = None
-        if preprocess_method == "DualStream_MNF":
+        if preprocess_method in ("DualStream_MNF", "Physical_Only"):
             _wl_list = self._extract_train_wavelengths(X_raw.columns)
         preproc = get_preprocessing_transformer(preprocess_method, wavelengths=_wl_list)
         pipeline = Pipeline([

src/gui/panels/step8_ml_train_panel.py

@@ -41,6 +41,7 @@ PREPROC_CHINESE = {
     'CT': '中心化 (CT)',
     'WVAE': '小波去噪 (WVAE)',
     'DualStream_MNF': '双流MNF (物理指数+降维)',
+    'Physical_Only': '纯物理特征 (指数+标准化)',
 }
 
 # 模型类型：内部键 -> 显示文本
@@ -177,7 +178,7 @@ class Step8MlTrainPanel(QWidget):
 
         preproc_grid = QGridLayout()
         self.preproc_checkboxes = {}
-        preproc_methods = ['None', 'MMS', 'SS', 'SNV', 'MA', 'SG', 'MSC', 'D1', 'D2', 'DT', 'CT', 'WVAE', 'DualStream_MNF']
+        preproc_methods = ['None', 'MMS', 'SS', 'SNV', 'MA', 'SG', 'MSC', 'D1', 'D2', 'DT', 'CT', 'WVAE', 'DualStream_MNF', 'Physical_Only']
 
         for i, method in enumerate(preproc_methods):
             checkbox = QCheckBox(PREPROC_CHINESE.get(method, method))

src/new/views/step8_view.py

@@ -43,6 +43,7 @@ PREPROC_CHINESE = {
     "CT": "中心化 (CT)",
     "WVAE": "小波去噪 (WVAE)",
     "DualStream_MNF": "双流MNF (物理指数+降维)",
+    "Physical_Only": "纯物理特征 (指数+标准化)",
 }
 
 MODEL_CHINESE = {
@@ -70,7 +71,7 @@ SPLIT_CHINESE = {
     "random": "随机划分 (Random)",
 }
 
-PREPROC_METHODS = ["None", "MMS", "SS", "SNV", "MA", "SG", "MSC", "D1", "D2", "DT", "CT", "WVAE", "DualStream_MNF"]
+PREPROC_METHODS = ["None", "MMS", "SS", "SNV", "MA", "SG", "MSC", "D1", "D2", "DT", "CT", "WVAE", "DualStream_MNF", "Physical_Only"]
 MODEL_GROUPS = [
     ("【线性模型】", ["LinearRegression", "Ridge", "Lasso", "ElasticNet", "PLS"]),
     ("【树模型】", ["DecisionTree", "RF", "ExtraTrees", "XGBoost", "LightGBM", "CatBoost"]),

src/preprocessing/spectral_Preprocessing.py

@@ -553,7 +553,13 @@ def get_preprocessing_transformer(method: str, wavelengths=None):
         from sklearn.pipeline import FeatureUnion
         return FeatureUnion([
             ('physical', PhysicalFeatureExtractor(wavelengths=wavelengths)),
-            ('mnf',      MNFTransformer()),   # n_components=0.95 自动确定
+            ('mnf',      MNFTransformer()),
+        ])
+    if method == "Physical_Only":
+        from sklearn.pipeline import Pipeline as _Pipeline
+        return _Pipeline([
+            ("physical", PhysicalFeatureExtractor(wavelengths=wavelengths)),
+            ("scaler",   StandardScaler()),
         ])
     if method not in _PREPROCESSING_TRANSFORMERS:
         print(f"未知预处理方法 '{method}'，回退为 IdentityTransformer")