In　order　to　get　actual　moving　load　on　bridges,　an　effective　method　for　moving　load　identification　applied　in　long　span　continuous　rigid　frame　bridges　was　presented　with　the　linear　regression　equation　between　moving　loads　and　strain　signals　acquired　from　health　monitoring　system.　Firstly,　the　strain　signal　was　de-noised　with　wavelet　transform.　And　then,　the　speed　and　time　of　the　vehicles　passing　the　bridge　were　estimated　by　time　history　of　strain　response.　Finally,　the　binary　linear　regression　equation　between　moving　loads　and　strain　signals　was　established.　The　results　show　that　the　error　rate　between　calculated　and　measured　vehicle　load　is　almost　below　17%,　and　the　axle-load　distribution　can　be　obtained　with　the　standard　vehicle　model.