Indoor non-orthogonal multiple access (NOMA) visible light communications (VLC) systems have the potential to enable high-rate multi-user communication. Nonetheless, the multipath effect may lead to a substantial reduction in communication reliability and user fairness. Therefore, a virtual time reversal mirror (VTRM) technology for NOMA-VLC channel equalization to eliminate the influence of multipath effect on communication performance was proposed. Initially, the indoor NOMA-VLC system model and the characteristics of communication optical links in multi-user scenarios were analyzed. To leverage the sparsity characteristics of optical links gains, the sparsity adaptive matching pursuit (SAMP) algorithm was adopted to estimate the channel impulse response (CIR) of NOMA-VLC systems. On this basis, the VTRM method was introduced to equalize the channel of NOMA-VLC, reducing the impact of channel fading through the spatio-temporal focusing characteristics of VTRM, and the received signal was reconstructed to suppress the multipath effect. The theoretical analysis and simulation results show that the NOMA-VLC system equalized by the proposed algorithm improves the performance of user 1 by 4.4 dB, user 2 by 5.7 dB, improves the average signal-to-noise ratio (SNR) of the two users by 5.05 dB, and reduces the performance difference between users from 1.6 dB to 0.5 dB, when the forward error correction (FEC) bit error rate (BER) threshold is met. Overall, this method provides an effective approach for NOMA-VLC channel equalization.