Detailed Explanation of Oracle Multi table Cascade Update

With cursor implementation, I think this method is absolutely safer.
-First set a cursor to store the data needed in the cursor:

declare 
  CURSOR D_CURSOR_CUS_INFO IS
    select t3.id_           as id_,
           t3.owe_money_    as owe_money_,
           a.heatingArea    as heating_area_
    from T_CUS_OWE_MONEY_2 t2
    left join T_CUS_OWE_MONEY_3 t3 on t2.id_= t3.id_
    left join (select s.bh,  sum(
           case 
                 when s.stkbz='0' then nvl(s.mj,0)
                 when s.stkbz='1' then 0-nvl(s.mj,0)
           end 
      ) as heatingArea from  sk s  where s.nd = '2008-2009' group by s.bh) a on t2.bh_=a.bh
  where  t3.owe_money_- t2.owe_money_  = a.heatingArea*5 and t3.OWE_MONEY_ > 0;

--Then the loop cursor updates the data:

begin
        FOR everyRow IN D_CURSOR_CUS_INFO
        loop 
         update T_CUS_YEAR_STATUS t
             set t.HEATING_AREA_ = everyRow.HEATING_AREA_,
                 t.OWE_MONEY_    = everyRow.owe_money_
             where t.YEAR_ = '2008-2009'
                   and t.id_ = everyRow.id_;
       end loop;
      commit;
end;