diff --git a/model_pro/LSTM_model.py b/model_pro/LSTM_model.py
index bcd2fb9..7ec84d7 100644
--- a/model_pro/LSTM_model.py
+++ b/model_pro/LSTM_model.py
@@ -219,7 +219,15 @@ class LSTMModelManager:
     
     def __init__(self, bert_model_path, model_save_path=None, vocab_size=30522, 
                  embedding_dim=100, hidden_dim=64, output_dim=2, n_layers=1, 
-                 bidirectional=True, dropout=0.3, word2vec_path=None):
+                 bidirectional=True, dropout=0.3, word2vec_path=None, random_seed=42):
+        # 设置随机种子以确保可重现性
+        self.random_seed = random_seed
+        random.seed(random_seed)
+        np.random.seed(random_seed)
+        torch.manual_seed(random_seed)
+        if torch.cuda.is_available():
+            torch.cuda.manual_seed_all(random_seed)
+        
         self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
         self.tokenizer = BertTokenizer.from_pretrained(bert_model_path)
         self.vocab_size = vocab_size
@@ -305,13 +313,18 @@ class LSTMModelManager:
         
         if val_texts is None:
             X_train, X_val, y_train, y_val = train_test_split(
-                X_train, train_labels, test_size=0.2, stratify=train_labels
+                X_train, train_labels, test_size=0.2, 
+                stratify=train_labels, 
+                random_state=self.random_seed  # 添加随机种子
             )
         else:
             X_val = vectorizer.transform(val_texts)
             y_train, y_val = train_labels, val_labels
         
-        lr_model = LogisticRegression(class_weight='balanced')
+        lr_model = LogisticRegression(
+            class_weight='balanced',
+            random_state=self.random_seed  # 添加随机种子
+        )
         lr_model.fit(X_train, y_train)
         
         val_pred = lr_model.predict(X_val)