In　order　to　solve　the　problem　of　difficulty　in　determining　the　number　of　modes　and　quadratic　penalty　factors　in　the　process　of　variational　mode　decomposition,　a　GWO-VMD　algorithm　was　proposed.　Based　on　the　Longlong　Tunnel　gas　explosion　method　construction,　an　integrated　acquisition　module　was　used　to　collect　vibration　signals　during　the　rock-breaking　process.　The　phase　space　reconstruction　recursive　graph　(RP)　similarity　model　was　used　to　accurately　distinguish　the　GWO-VMD　principal　components　of　the　signals.　The　real　signal　with　interference　removed　was　reconstructed,　revealing　the　distribution　characteristics　of　the　energy　of　the　gas　explosion　signal　in　the　time-frequency　domain,　and　quantifying　the　accuracy　error　of　the　digital　electronic　detonator.　The　results　show　that　compared　with　traditional　variational　mode　algorithms,　the　GWO-VMD　algorithm　has　significant　advantages　in　improving　the　signal-to-noise　ratio　and　adaptive　feature　extraction　of　gas　explosion　signals,　and　has　strong　time-varying　frequency　tracking　performance.　It　can　accurately　identify　the　detonation　accuracy　of　digital　detonators　and　effectively　identify　the　related　of　tunnel-blasting　detonator　disaster　sources.　©　2025　Editorial　Department　of　China　Safety　Science　Journal.　All　rights　reserved.