The residual hardware impairments（HWIs） caused by the non-ideal characteristics of the transceiver hardware is unavoidable in the intelligent reflecting surface （IRS） assisted massive multiple-input multiple-output （MIMO） system， which seriously affects the uplink achievable rate. To solve this problem， a passive beamforming scheme based on the adaptive quantum bacterial foraging optimization （AQBFO） algorithm is proposed to suppress the negative impact of HWIs on the system performance. Firstly， an approximate analytical expression of the uplink achievable rate is derived based on statistical channel state information （CSI）. Then，the passive beamforming optimization scheme based on AQBFO algorithm is carried out to maximize the sum rate. Simulation results show that in IRS-assisted massive MIMO system， the passive beamforming scheme based on AQBFO algorithm can effectively suppress the influence of residual HWIS and significantly improve the uplink ergodic sum rate.