In Sect. “Method“, the construction process of the integrated digital twin system for the “shape-performance-control” of the roadheader is described. Additionally, the specific implementation details are introduced in this section.
Dataset construction
To construct the dataset for training the twin model of the roadheader, the mechanical performance of the machine, including the stress and deformation, is calculated through numerical simulation. The cutting resistance generated by the cutting head when cutting coal and rock is the main load on the machine. This cutting resistance can be measured using multiple sensors. The mechanical performance of the roadheader is represented by the following eight variables: the vibration acceleration of the cutting arm, displacement of the piston rod of the rotating hydraulic cylinder, displacement of the piston rod of the lifting hydraulic cylinder, pressure of the left rotating hydraulic cylinder, pressure of the right rotating hydraulic cylinder, pressure of the lifting hydraulic cylinder, speed of the cutting head and torque of the hydraulic motor. The LS-DYNA module of ANSYS is used to conduct parametric numerical simulation of the roadheader, which involved multiple steps. Here is a brief explanation of the important steps in the basic process of finite element analysis:
Model preprocessing
The three-dimensional model of the roadheader is imported into Spaceclaim. In this step, parts such as bolts and nuts that have little influence on the mechanical properties of the roadheader are deleted. Additionally, chamfering is removed, and the hydraulic cylinder, cutting arm, and other parts are merged separately to simplify the model.
Material setting
After the preprocessing, materials are assigned to each part of the model. The roadheader base is made of Q235 material, the rotating table is made of ZG35 material, the lifting and rotating hydraulic cylinder is filled with No. 46 anti-wear hydraulic oil, the cutting head is treated as a rigid body, and other parts are made of No. 45 steel. The parameters of each material are provided in Table 1.
Meshing
The mesh type of hard structural parts are set to Tetrahedrons, and the liquid mesh type is set to Automatic. The cutting head is considered rigid and does not undergo plastic deformation, so its mesh is roughly divided. This means that the mesh is not as refined as other components. The rotating and lifting hydraulic cylinder is an important power component, and its mesh should be encrypted, and the mesh size was set to 15 mm. The contact surface of hydraulic cylinder, cutting arm and rotating table are encrypted with grid. The mesh size was set to 10 mm. For other parts, the mesh size was set to 30 mm. The total number of elements in the system is reported to be 278,747.
Contact setting
The friction contact is set between the piston of the lifting and rotating hydraulic cylinder and the inner wall of the cylinder. The friction coefficient for this contract is 0.15, and the frictionless contact is set between other components.
Boundary condition setting
The boundary condition is applied at the corresponding part position, and the set value of the boundary condition is determined based on the measurement from the sensor during the actual cutting process. Specifically, fixed constraints are imposed on the surface of the roadheader base, torque and speed are applied at the cutting head, and remote displacement is applied to the front end face of the cutting arm to simulate the vibration during the cutting process. It worth nothing that the cutting arm primarily vibrates in the plane parallel to the coal wall, with minimao vibration in the direction perpendicular to the coal wall. The specific values of the boundary conditions are provided in Table 2.
The stress results of each node of the roadheader are integrated, resulting in the mechanical performance dataset of the roadheader, which is required for twin model training. The dataset construction process is shown in Fig. 3.
The correlation between the twin model and the variables and stresses in the dataset can be mathematically described by Eq. (1).
$$ \begin{gathered} F([x_{1} ,x_{2} ,…,x_{8} ]_{i} ) = [data_{1} ,…,data_{272847} ]_{i} , \\ i \in (0,900] \\ \end{gathered} $$
(1)
In Eq. (1), F represents the twin model of the relationship between the predictive variable and the node stress of the roadheader. The variable x comprises 8 factors that influence the mechanical performance of the roadheader. These factors include the vibration acceleration of the cutting arm, the displacement of the piston rod of the…
Read More: Research on the “shape-performance-control” integrated digital twin system for