这里以#号拆分多行：
SELECT cy.zjhm, REGEXP_SUBSTR(cy.jtcy, '[^#]+', 1, LEVEL) NAME FROM t_xsgl_xsgl_xsxx cy CONNECT BY LEVEL <= REGEXP_COUNT(cy.jtcy, '[^#]+') AND ROWID = PRIOR ROWID AND PRIOR DBMS_RANDOM.VALUE IS NOT NULL ORDER BY cy.zjhm;