试试这个：

import networkx as nx

L = df.groupby('cluster')['cluster2'].agg(list)
G=nx.Graph()
for l in L:
    nx.add_path(G, l)
df = (df.assign(cluster_new = df.groupby(df['cluster2'].map(
    {v2:k for k,v in enumerate(list(nx.connected_components(G))) for v2 in v}
    ))['cluster'].transform('first')))

输出量：

zip zip_splitted cluster cluster2 cluster_new
0    A          A_1     111      991         111
1    A          A_1     111      991         111
2    A          A_1     111      994         111
3    A          A_1     112      991         111
4    A          A_1     112      882         111
5    A          A_1     112      991         111
6    A          A_2     113      993         111
7    A          A_2     113      991         111
8    A          A_2     113      994         111
9    A          A_2     114      992         111
10   A          A_2     114      991         111
11   A          A_2     114      991         111
12   C          C_1     115      889         111
13   C          C_1     115      889         111
14   C          C_1     115      992         111
15   C          C_1     116      998         116
16   C          C_1     116      997         116
17   C          C_1     116      999         116

用pandas iterrows（）创建的解决方案也在工作，但图形方法看起来更好，可能更快。

new_cluster_ids = {}

# Iterate over each row in the dataframe
for index, row in df.iterrows():
    cluster = row['cluster']
    cluster2 = row['cluster2']

    # Check if the cluster already has a new ID assigned
    if cluster in new_cluster_ids:
        new_cluster_id = new_cluster_ids[cluster]
    else:
        new_cluster_id = cluster
    
    # Find other rows with the same cluster2 value
    same_cluster2_rows = df[df['cluster2'] == cluster2]
    
    # Update the new cluster ID for all rows with the same cluster2 value
    for _, same_cluster2_row in same_cluster2_rows.iterrows():
        new_cluster_ids[same_cluster2_row['cluster']] = new_cluster_id
    
df['cluster_new'] = df['cluster'].map(new_cluster_ids)

print(df)