Good　clustering　algorithm　can　effectively　reduce　network　energy　consumption　and　improve　the　reliability　of　network.　However,　load　unbalance　and　communication　unreliability　between　the　clusters　have　significant　impacts　on　the　performance　of　the　clustering　algorithm.　In　this　paper,　a　fault-tolerance　clustering　algorithm　with　load-balance　aware　was　proposed　to　solve　these　problems.　Base　on　the　quality　of　the　particles　in　the　population,　an　adaptive　discrete　particle　swarm　optimization　(ADPSO)　with　an　adaptive　adjustment　strategy　for　inertia　weight　was　designed,　using　the　randomly　two-point　crossover　operator　and　random　one-point　mutation　operator　of　the　genetic　algorithm.　We　introduced　a　cluster　head　selection　mechanism　based　on　ADPSO　for　optimizing　both　of　the　two　objects,　load　balancing　and　energy　consumption　in　the　algorithm.　Moreover,　to　guarantee　the　reliability　of　data　transmission,　an　inter-cluster　connectivity　algorithm　based　on　local　minimum　spanning　tree　was　constructed,　which　ensured　the　two-connectivity　by　eliminating　cut　point　in　the　network.　The　experimental　results　demonstrate　that　the　proposed　algorithm　can　achieve　better　performance　on　load　balancing　and　two-connectivity,　effectively　reducing　the　energy　consumption,　prolonging　the　lifetime　of　the　network　and　improving　the　network　reliability.