To address the collaborative optimization challenges of microgrids containing distributed photovoltaics（PV）in cold chain logistics parks, where the refrigeration systems require 24-hour continuous operation to maintain strict temperature control（-18°C±2°C）while PV output concentrates during 10:00—15:00 and logistics operations mainly occur in early morning and evening, this temporal mismatch between power demand and PV generation leads to severe PV curtailment. Proposing a scheduling strategy that balances PV consumption rate and operational economy. This strategy tackles the issues of high PV curtailment caused by strong PV output randomness and the complexity of multi-objective cooperative optimization. A tripartite collaborative architecture of“source-load-storage”is constructed. Typical PV output scenarios and their probability distributions are generated using the k-means clustering algorithm. Downside risk constraints are employed to quantify the cost fluctuation risk under these scenario probabilities, providing a continuous regulation space spanning from risk aversion to risk neutrality. A multi-objective optimization model is established to maximize the PV consumption rate and minimize the comprehensive operational cost. A dynamic-weight ideal point method based on the rate of change of the objective functions is proposed.An improved escape algorithm is designed to solve this model. Case studies demonstrate that under a defined risk regulation level, the park’s PV consumption rate reaches 95.13% , and the daily operating cost is reduced to RMB 7 356, representing a cost reduction of 21.60% compared to pre- optimization levels. The results verify that the proposed collaborative optimization method can effectively enhance distributed PV consumption capability and operational economy. It provides a viable solution and theoretical foundation for scheduling complex microgrids containing high-penetration renewable energy and thermostatically controlled loads.