To address the issue of low comformity between traditional steady-state test conditions of hydraulic torque converters and real vehicle operating conditions, which fail to reflect the actual state of the vehicle, a cyclic operating condition for high-power tracked off-road vehicles with hydraulic torque converters is proposed. Based on real vehicle operating data from a certain type of tracked vehicle, the characteristics of torque converter in real vehicle operation are statistically analyzed. Twelve statistical feature parameters and ten ratio feature parameters are selected. The dimensionality reduction of the operating data is carried out by using principal component analysis. Clustering analysis on data segments is completed by the K-means clustering algorithm and unsupervised learning. The dynamic programming methods are used to integrate blocking conditions and typical operating condition fragments are obtained for cyclic operating condition reconstruction. The segments are smoothed by using Hanning windows. The cyclic operating conditions based on real vehicle data that align with the operational characteristics of the hydraulic torque converters during vehicle operation are constructed. The opimization is performed by simulated annealing and multi-objective particle swarm algorithm to build a cyclic condition based on the real vehicle data to fit the real vehicle operation characteristics of the torque converter, taking the average error of the main eigenvalues of the cyclic condition and the overall data, the sum of the speed difference and the slope difference at the connection as the targets. The research shows that the final condition consists of impeller speed-time and blocking signal-time. The average relative error of the main feature parameters is 2.92%, which is consistent with the performance of the real vehicle data. This research provides a new approach to design reliabile test conditions for hydraulic torque converters.