Real- Time Recognition Algorithm of Tractor Random Dynamic Load Characteristics for Control

The tractor load is a continuous and dynamic random load with time-varying amplitude and frequency, the time domain characteristics of which are represented by steady-state value, instantaneous value and dynamic load variation coefficient. The time domain characteristics of the random load reflect tractor working state, which are important parameters of tractor control systems for ploughing depth, selection of operating gears and timing of shifting, and load distribution of hybrid power. This kind of dynamic load signal is a colored random signal mixed in other random noises, and its time domain characteristics are difficult to obtain online, which becomes the bottleneck influencing the random load characteristics applied to the field of tractor real-time control. In this paper, based on the study of time domain and frequency domain characteristics of tractor random load signal, the triggered quadratic frequency conversion adaptive Kalman filter algorithm is used to realize the real-time extraction of random load characteristics. Then white noise is eliminated by Kalman filter to extract colored random load signals, and the load steady-state value is obtained after the colored noise is whitened by reducing sampling frequency of secondary filtering. In view of the unknown statistical characteristics of secondary filtering noise, a triggered fading adaptive filter combined with vehicle bus network information sharing technology is adopted to ensure the ability of filter to quickly track the load change. The simulation results show that the random load signal obtained by the algorithm is close to the reference value, and the load variation coefficient can reflect the characteristic changes of the random load, which lays a foundation for the dynamic control of tractor operation process.