During the process of composting and fermentation for agricultural wastes, the water content of materials has changed, which will lead to the change of the load of the compost turning operation. However, the existing operation of the compost turning machine was mainly by manual operation, which led to low operation efficiency and easy to cause failure in the process of machine operation. Aiming at the problem of inaccurate regulation during manual operation of the compost turning machine, the relationship model for the operation load of the compost turning machine between the moisture content of the composting material and the moving speed of the machine was established through the experiments. Combined with the variable frequency speed control model of the machine, the speed control system of the compost turning machine was optimized by using DDPG (Deep Deterministic Policy Gradient) improved PID algorithm. The Simulink simulation results showed that compared with the traditional PID algorithm, the DDPG improved PID algorithm reduced the overshoot by 6.7%, the adjustment time by 2.5 s, and the anti-disturbance and follow-up performance were better. The field test results of compost turning operation showed that the control mode of DDPG improved PID algorithm was 4% lower than that of traditional PID algorithm under overshoot, and the adjustment time was reduced by 2 s. Compared with manual control, the adjustment time was reduced by 6 s.

在农业废弃物堆肥发酵过程中物料的含水率会发生变化，导致翻堆作业负荷的变化。而现有的翻堆作业主要通过人工操作，导致机器作业过程中作业效率低，容易产生故障。针对人工操作翻堆机作业时出现调控不精准的问题，通过试验构建翻堆机作业负荷与翻堆物料含水率、翻堆机行走速度关系模型，并结合翻堆机变频调速控制模型，利用DDPG(Deep Deterministic Policy Gradient)改进PID算法对翻堆作业调速系统进行优化控制。经过Simulink仿真结果表明，DDPG改进PID算法相比传统PID算法在超调量上减少6.7%，调节时间减少2.5 s，并且抗扰动与跟随性能均更优。翻堆作业现场测试结果表明：DDPG改进PID算法的控制方式相比传统PID算法超调量要降低4%、调节时间减少2 s，相比人工控制其调节时间减少6 s。