With the rapid growth of internet traffic, the demand for high-bandwidth, low-latency, and reliable network services is becoming increasingly critical. Multipath transmission technologies, which leverage heterogeneous networks such as 4G, 5G, and WiFi, have emerged as key solutions for improving transport efficiency. Multipath QUIC (MPQUIC), built on the UDP protocol, stands out for its fast connection establishment, agile path switching, and non-blocking stream multiplexing. However, MPQUIC still faces several challenges, including packet reordering, path scheduling strategies that struggle to adapt to dynamic network conditions, and delayed network state awareness in schedulers. To address these issues, this paper proposes an adaptive multi-weight stream scheduling algorithm. It prioritizes time-sensitive data through urgency-based stream scheduling, dynamically adjusts path weights according to real-time network metrics, and incorporates differentiated retransmission strategies for various stream types. Simulation experiments across multiple transmission scenarios demonstrate that the proposed algorithm significantly reduces file transfer time and packet delay compared to existing MPQUIC schedulers, thereby improving overall Quality of Service (QoS).