The flexible process planning problem based on machining time and energy consumption prediction model studied in this paper is shown in Fig. 1. While considering the process planning flexibility problem, machining time and energy consumption prediction are combined with process planning, in which the flexibility problem will make the energy consumption and machining time for the same feature machining completion different, and also affect the total energy consumption and time for machining resource conversion. Therefore, in view of the complex process content in process planning, the process is decomposed into independent operations. In this paper, the operation is defined as one machining process for different accuracy levels of a feature, for example rough milling plane F1 using machine M1 and tool T1 is defined as an operation. And the machining time and energy consumption calculation is focused on the operation, and using the historical machining data, the machining time and energy consumption prediction of the operation is made, and the prediction results are applied to the process planning.
2.1 Process time and energy consumption forecasting for process planning
Due to flexibility issues and the influence of the manufacturing environment on machining, traditional machining time and energy estimation methods are difficult to adapt to these complex situations. This leads to the process routes generated based on the traditional machining time and energy consumption estimation methods are often not executed smoothly or executed poorly, which eventually affects the performance of the whole manufacturing system. Therefore, it is of great significance to study the processing time and energy consumption prediction problems oriented to process planning.
Machining time and energy consumption prediction is a manufacturing consumption control tool commonly used in manufacturing industries. Previous research methods lack the consideration of the impact of part geometry information on machining time and energy consumption, lack the refinement and utilization of historical data, and do not apply machining time and energy consumption prediction models in process planning. In order to solve these problems, this paper proposes a machining time and energy consumption prediction method based on historical data. This method decomposes processes into independent operations and uses them as carriers of machining data, effectively connecting part geometry, process information, NC programs and machining resources, as shown in Fig. 2.
Firstly, we decompose the historical data and extract the key information for the problem that the process data and machining data in the manufacturing plant are cluttered and there is no connection between them. Relevant process information is extracted from the process protocol card and decomposed to obtain the operation and its process information, such as the machine tool, the tool and the geometric parameters of the corresponding features. The operation is used to extract parameters such as depth of cut from the NC code, and the time and energy consumption data required to complete the NC code are extracted from the machine and the smart meter. Second, the machining process of CNC machine tools is very complex, and there are many factors that affect the energy consumption of workpiece machining. Among the factors affecting the machining time and energy consumption of the workpiece, despite the differences in geometry, quality, accuracy and other information of the workpiece, but the workpiece of the same shape features of the processing scheme has similarity, and different types of machining (operation for a class of features of the processing method, excluding equipment and other process information, such as rough milling plane) to complete the processing of the machining time and energy consumption indicators exist in a large difference. And since the machining process of a workpiece geometry usually consists of many machining steps, data on the machining time and energy consumption of the operation are important for the process planning phase. And because very few detailed parameters are available at the process planning stage, it is difficult to predict machining times and energy consumption from some specific cutting parameters, machine and tool parameters. Therefore, in this paper, a processing time and energy consumption prediction model was constructed from a macro perspective, and five decision variables were identified as. Machine (\({M_i}\)), Tool (\({T_i}\)), Machining type(\({P_i}\)), depth of cut (\({D_i}\)), cutting area (\({A_i}\)). The decision variables are denoted by X, as shown in Eq. (1), and two prediction objectives are identified: machining time S, and energy consumption E, as shown in Eq. (2).
$$X=[{M_i},{T_i},{P_i},{D_i},{A_i}]$$ 1
2.2 Flexible process planning problem description and modeling
To clearly describe the flexible process planning problem addressed in this paper, an example is given in Table 1, where the part \({P_1}\) contains 7 features and is machined on 5 machines with 7 different tools, for a total of 12 optional operations. In the example, due to constraints, feature F2 must be machined before F4 and feature F4 must be machined before F7. Both features F2 and F4 can be machined by multiple different processes to the desired effect, and each individual operation includes multiple selectable machines and multiple selectable tools. Moreover, the multi-flexibility problem leads to a large number of feasible processes for the part and multiple results in terms of machining energy and machining time for the whole machining process. Therefore, in this paper, when solving the flexible process planning problem, the energy consumption and machining time of the machining process are obtained from the prediction model, so as to obtain the process route that meets the actual machining environment.
Table 1
Machining information for processing \({P_1}\) on 5 machines
Feature
|
Optional Processes
|
Machine
|
Tool
|
Process constraints
|
F1
|
O1
|
M1, M2
|
T2, T3
|
|
F2
|
O2-O3
|
M2, M4/ M1, M2
|
T1, T2/T2, T3
|
Before F4
|
O4
|
M3, M4
|
T2, T3
|
F3
|
O5
|
M4, M5
|
T3, T5
|
|
F4
|
O6
|
M2, M3
|
T2, T3
|
Before F7
|
O7-O8
|
M1, M3/M3, M4
|
T1, T3/T2, T3
|
F5
|
O9
|
M2, M5
|
T2, T7
|
|
F6
|
O10
|
M2, M5
|
T4, T6
|
|
F7
|
O11-O12
|
M1, M3/M1, M2
|
T5, T6/T6, T7
|
|
Based on the above description, the mathematical model of flexible process planning proposed in this paper is shown below, and the minimization of completion time and minimization of completion energy consumption are the two objectives of the optimization problem. Completion time and completion energy consumption are optimization objectives often considered in process planning problems. Unlike previous studies, in this paper, not only the multiple components of completion time and completion energy consumption are considered, but also the machining time and machining energy consumption are affected by many factors, and a prediction model is used for accurate prediction of machining time and energy consumption.
$${E_{process}}{\text{=}}\sum\limits_{{i=1}}^{n} {{Q_i}}$$
3
$${E_{chgM}}{\text{=}}\sum\limits_{{i=1}}^{{n - 1}} {E{M_{({M_i},{M_{i+1}})}} \times \Omega ({M_i}} ,{M_{i+1}})$$
4
$$\Omega (x,y)=\left\{ \begin{gathered} 0,x \ne y \hfill \\ 1,x=y \hfill \\ \end{gathered} \right.$$
5
$${E_{chgT}}{\text{=}}\left\{ \begin{gathered} \sum\limits_{{i=1}}^{{n - 1}} {E{T_{{M_i}}} \times (1 - \Omega ({T_i},{T_{i+1}})} ) {M_i}={M_{i+1}} \hfill \\ \sum\limits_{{i=1}}^{{n - 1}} {E{T_{{M_{i+1}}}} \times (1 - \Omega (T({{M^{\prime}}_{i+1}}),{T_{i+1}})} ) {M_i} \ne {M_{i+1}} \hfill \\ \end{gathered} \right.$$
6
Equation (3) represents the energy consumption generated by the machine tool to machine the part, n represents the total number of operations required to machine the part, and \({Q_i}\) represents the energy consumption required to complete the i-th operation which obtained from the prediction model; Eq. (4) represents the energy consumption resulting from the transfer of the part between machines throughout the machining process, where \({M_i}\) denotes the machine selected for the i-th operation and \(E{M_{({M_i},{M_{i+1}})}}\) denotes the energy required to transport the part from \({M_i}\) to \({M_{i{\text{+}}1}}\); Eq. (5) represents whether the equipment selected for two adjacent operations is the same; Eq. (6) represents the energy consumption required for tool change operations, divided into two cases: same machine and different machines, where \(E{T_{{M_i}}}\) represents the energy consumption required for one tool change on \({M_i}\), and \(T({M^{\prime}_{i+1}})\) represents the tool used by \({M_{i+1}}\) during the last operation.
$${T_{process}}{\text{=}}\sum\limits_{{i=1}}^{n} {{H_i}}$$
7
\({T_{chgM}}{\text{=}}\sum\limits_{{i=1}}^{{n - 1}} {T{M_{({M_i},{M_{i+1}})}} \times \Omega ({M_i}} ,{M_{i+1}})\) (8)
\({T_{asm}}{\text{=}}\sum\limits_{{i=1}}^{{n - 1}} {S(i} ,i+1)\) (9)
$${T_{chgT}}{\text{=}}\left\{ \begin{gathered} \sum\limits_{{i=1}}^{{n - 1}} {T{T_{{M_i}}} \times (1 - \Omega ({T_i},{T_{i+1}})} ) {M_i}={M_{i+1}} \hfill \\ \sum\limits_{{i=1}}^{{n - 1}} {T{T_{{M_{i+1}}}} \times (1 - \Omega (T({{M^{\prime}}_{i+1}}),{T_{i+1}})} ) {M_i} \ne {M_{i+1}} \hfill \\ \end{gathered} \right.$$
10
Equation (7) represents the time required by the machine to machine the part, where n represents the total number of operations required to machine the part, and \({H_i}\) denotes the time required to complete the i-th operation, obtained from the prediction model. Eq. (8) represents the time required to transfer a part between machines, and \(T{M_{({M_i},{M_{i+1}})}}\) represents the time required to transport two adjacent operations from \({M_i}\) to \({M_{i+1}}\). Eq. (9) represents the time required for the adjustment and installation of the part, and t \(S(i,i+1)\) indicates the adjustment and installation time between two adjacent processes. Eq. (10) represents the time required for tool change, divided into two cases, same machine and different machine, where \(T{T_{{M_i}}}\) indicates the time required for tool change for machine \({M_i}\).
$$Min{\text{ }}{f_1}=Min({E_{process}}+{E_{chgM}}+{E_{chgT}})$$
11
$$Min{\text{ }}{f_2}=Min({T_{process}}+{T_{chgM}}+{T_{asm}}+{T_{chgT}})$$
12
The energy consumption and processing time described above are considered to be optimized simultaneously, where the energy consumption is affected by time but there are many other influencing factors. The two optimization objectives of this paper are shown in equations (11) and (12). And the mathematical model in this paper needs to satisfy the following assumptions:
-
The workpiece is transported between machines using electric equipment with constant power and energy consumption that is only time dependent;
-
Workpiece adjustment and installation time is defined as a constant, and energy consumption is linearly related to time;
-
Only one process can be selected for each feature to be machined;
-
Only one machine and tool can be selected for each operation;
-
The machine is always started, and the standby energy consumption is not considered;
-
The first machining of a machine does not require a tool change by default.