Production line change propagation control method and system based on model predictive control

Abstract

A production line change propagation control method based on model predictive control is characterized by comprising the following steps: abstracting key parameters of production on a production line into a network based on a multilayer network theory, and constructing a multi-dimensional network model based on a relationship among the key parameters; based on the multi-dimensional network model, finding out changed parameter nodes and other key nodes influenced by the changed parameter nodes as first nodes, and constructing a prediction model by using the first nodes; setting a control period and a data threshold value; and recording a first moment of the control cycle, and obtaining prediction data of each moment in the remaining control cycle through the prediction model. The method aims to ensure that the implementation process of production line design change is controllable and efficient, chain reactions generated in the change process are greatly reduced, and unnecessary large-scale redesign is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of production regulation, and particularly to a production line change propagation control method and system based on model predictive control. Background Art

[0002] In modern manufacturing, design changes to production lines are made to improve production efficiency, reduce costs, adapt to changes in market demand or technological progress. As production line designs become increasingly complex, the number of parameters involved, such as equipment, processes, procedures, and materials, also increases. Whenever a design change occurs, the propagation and update of relevant parameters become particularly crucial. If the changed parameters cannot be transmitted to all links of the production line in a timely and accurate manner, or if a series of chain reactions are triggered during the propagation process, it may lead to the collapse of the design plan and even force a large-scale redesign of the entire production line, thereby greatly extending the design cycle, increasing design costs, and affecting production efficiency and product quality.

[0003] In particular, during the production line design process, parameter change propagation often triggers the "avalanche effect", that is, a small design change may trigger a chain of adjustments in multiple links and parameters, resulting in the invalidation of the system's design plan and even the need to re-evaluate the layout and process of the entire production line. Such situations not only increase design time and costs but also may lead to production interruptions or quality problems, seriously affecting the market competitiveness of enterprises.

[0004] Therefore, how to effectively control the parameter propagation during production line design changes, avoid avalanche-like chain propagation, and ensure the controllability, timeliness, and rationality of changes have become key technical problems that need to be solved urgently in modern manufacturing. Summary of the Invention

[0005] Aiming at the above defects, the purpose of the present invention is to propose a production line change propagation control method and system based on model predictive control to avoid the problem of avalanche-like chain propagation caused by changing a single parameter of the production line.

[0006] To achieve this purpose, the present invention adopts the following technical solutions: A production line change propagation control method based on model predictive control, including the following steps: Step S1: Based on the multi-layer network theory, abstract the key parameters in the production line into a network, and construct a multi-dimensional network model based on the relationships between the key parameters;

[0007] Based on the multi-dimensional network model, find the changed parameter node and other key nodes affected by it as the first nodes, and use the first nodes to construct a prediction model;

[0008] Step S2: Set the control period and data threshold;

[0009] Step S3: Record the first moment of the control cycle, and obtain the prediction data of each moment in the remaining control cycles through the prediction model, and determine whether all the prediction data are less than the data threshold. If all the prediction data are less than the data threshold, execute step S4;

[0010] If all the predicted data are smaller than the data threshold, the control variables in the prediction model are adjusted by the objective function to obtain an updated prediction model, and step S4 is executed;

[0011] Step S4: Delete the first moment, and update the next moment as the first moment, and re-execute step S3 until the first moment is the last moment in the control cycle, and output the control variables in the prediction model.

[0012] Preferably, the key parameters include the configuration, dynamic type, control type and optimization type of the production line, and each key parameter includes at least one parameter node;

[0013] The construction standard of the multi-dimensional network model in step S1 is as follows:

[0014] Each layer of the network is composed of parameter nodes of the same nature. The relationship between different parameter nodes of the production line is abstracted as edges. The parameter nodes and edges constitute a multi-dimensional network model of the production line.

[0015] Preferably, the prediction model is as follows:

[0016] L(t+1)=AL(t)+Bβ(t)L(t) α +Du(t);

[0017] Where A represents the inherent coupling relationship coefficient between parameter nodes, B represents the allocation strategy weight coefficient, D represents the path blocking influence coefficient, α is the capacity parameter, β(t) and u(t) are both control variables, where β(t) represents the load distribution coefficient at time t, and u(t) is the propagation path switch of the 0 / 1 vector;

[0018] L(t)=[L1(t),L1(t),...L N (t)] T ;

[0019] Among them L N (t) represents the load of the Nth parameter node at time t, where N is the total number of parameter nodes;

[0020] There is a nonlinear relationship between the load of the parameter node and the load of the parameter node at the next moment, wherein the nonlinear relationship is as follows:

[0021] L i (t+1)=ωL i (t)+Ση[Lj (t)] θ ,

[0022] ω, θ, and η are all capacity parameters.

[0023] Preferably, the objective function is specifically as follows:

[0024]

[0025] where

[0026]

[0027] where H p is the total length of the control period, Q i represents the load tracking weight of the i-th parameter node, L ref,i represents the tracking load balancing target of the i-th parameter node, R i represents the threshold adjustment weight of the i-th parameter node, ΔC i (t) is also a control variable, ΔC i (t) represents the capacity threshold of the i-th parameter node at the t-th moment, S i represents the path blocking weight of the i-th parameter node.

[0028] Preferably, the objective function also has the following constraint conditions:

[0029]

[0030] ∑ΔC i (t + k) ≤ ΔC max ;

[0031] where β max represents the maximum load distribution coefficient, C max represents the capacity threshold of the maximum parameter node.

[0032] A production line change propagation control system based on model predictive control, using the above production line change propagation control method based on model predictive control, includes: a construction module, a setting module, a monitoring module, and an output module;

[0033] The construction module is used to abstract the key parameters produced on the production line into a network based on the multi-layer network theory, and construct a multi-dimensional network model based on the relationships between the key parameters;

[0034] Based on the multi-dimensional network model, find the changed parameter nodes and other key nodes affected by them as the first nodes, and construct a prediction model using the first nodes;

[0035] The setting module is used to set the control period and the data threshold;

[0036] The monitoring module is used to record the first moment of the control period, obtain the predicted data at each moment within the remaining control period through the prediction model, and determine whether all the predicted data are less than the data threshold. If all the predicted data are less than the data threshold, the output module is called;

[0037] If not all the predicted data are less than the data threshold, the control variables in the prediction model are adjusted through the objective function to obtain an updated prediction model, and the output module is called;

[0038] The output module is used to delete the first moment, update the next moment as the first moment, and call the monitoring module again until the first moment is the last moment within the control period, and output the control variables in the prediction model.

[0039] One of the above technical solutions has the following advantages or beneficial effects: The production line change propagation control method of model predictive control combines real-time collaboration and feedback mechanisms. The present invention aims to ensure that the implementation process of production line design changes is controllable, efficient, and greatly reduces the chain reaction generated during the change process, avoiding unnecessary large-scale re-design. Description of the Drawings

[0040] Figure 1 is a flowchart of an embodiment of the present invention.

[0041] Figure 2 is a schematic structural diagram of an embodiment of the system of the present invention.

[0042] Figure 3 is a schematic diagram of a multi-dimensional network model of an embodiment of the present invention. Detailed Embodiments

[0043] The following details the embodiments of the present invention. The examples of the embodiments are shown in the drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary only for explaining the present invention and should not be construed as limiting the present invention.

[0044] In the description of the embodiments of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of the described features. In the description of the embodiments of the present invention, "a plurality" means two or more unless otherwise specifically defined.

[0045] In addition, the terms "first" and "second" are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "a plurality of" means two or more. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0046] As Figures 1 to 3 shown, a production line change propagation control method based on model predictive control includes the following steps: Step S1: Based on the multi-layer network theory, abstract the key parameters produced on the production line into a network, and construct a multi-dimensional network model based on the relationships between the key parameters;

[0047] Based on the multi-dimensional network model, find the changed parameter nodes and other key nodes affected by them as the first nodes, and use the first nodes to construct a prediction model;

[0048] Step S2: Set the control period and the data threshold;

[0049] Step S3: Record the first moment of the control period, and obtain the predicted data at each moment within the remaining control period through the prediction model. Judge whether all the predicted data are less than the data threshold. If all the predicted data are less than the data threshold, then execute Step S4;

[0050] If not all the predicted data are less than the data threshold, adjust the control variables in the prediction model through the objective function to obtain an updated prediction model, and execute Step S4;

[0051] Step S4: Delete the first moment, update the next moment as the first moment, and re-execute Step S3 until the first moment is the last moment within the control period, and output the control variables in the prediction model.

[0052] To solve the problem in the prior art that parameter change propagation causes avalanche-like propagation, resulting in the unavailability of the design solution or forcing large-scale changes to the production line. A production line change propagation control method based on model predictive control is proposed. Combining real-time collaboration and feedback mechanisms, the present invention aims to ensure that the implementation process of production line design changes is controllable, efficient, and significantly reduce the chain reaction generated during the change process, and avoid unnecessary large-scale re-design.

[0053] The following is illustrated by an example: In a parameter network of a certain automobile assembly line, the core welding parameter node (Node 1) has a sharp increase in load (control variable) due to design changes, and it is necessary to dynamically adjust the load distribution through MPC to avoid cascading failures;

[0054] At this time, it is necessary to abstract the key parameters in the automotive assembly line parameter network into a network, and based on the relationships between the key parameters, construct a multi-dimensional network model. Through the multi-dimensional network, it can be learned which other parameter nodes need to be changed due to the change of node 1. Then, collect the other parameter nodes that need to be changed due to the change of node 1. Use node 1 and other parameter nodes as the first node to construct a corresponding prediction model. Through the prediction model, the overall load situation of the production line at the next moment can be predicted. Determine whether it exceeds the load situation (data threshold) through the prediction model. If it exceeds the load situation (data threshold), it means that the modification of node 1 has an adverse impact on the production line at this time. Then, it is necessary to modify the corresponding control variables (such as load distribution coefficient, propagation path switch, or the load of node 1), etc. The control variables can be obtained through the corresponding set objective function. Continuously modify the control variables through the prediction model and the objective function within a control cycle, so that the production line meets the production requirements and no adverse effects occur.

[0055] It is worth mentioning that in the present invention, the cycle regulation is carried out through the "execution of the first-step control" mechanism.

[0056] The following is illustrated by an example. Assume that the control cycle is set to 5, the first moment t = 1, and the current prediction model is used to predict t = 1, 2, 3, 4, 5 respectively. If the load is exceeded, update the prediction model. At this time, the first moment is modified to t = 2, and the new prediction model is used to predict t = 2, 3, 4, 5 respectively, and repeat the judgment of whether the load exceeds the threshold. That is, only the optimal control action at the first moment (t) is implemented in each control cycle, rather than executing all control instructions within the entire control cycle. This mechanism realizes the adaptive adjustment of complex dynamic systems by dynamically updating the system state and re-optimizing.

[0057] Preferably, the key parameters include the configuration, motion type, control type, and optimization type of the production line, and each key parameter contains at least one parameter node;

[0058] The construction criteria of the multi-dimensional network model in step S1 are as follows:

[0059] Each layer of the network is composed of parameter nodes of the same nature. Abstract the relationships between different parameter nodes of the production line as edges. The parameter nodes and edges form the multi-dimensional network model of the production line.

[0060] The production line is divided into four dimensions: configuration, motion type, control type, and optimization type.

[0061] Configuration: It mainly focuses on the physical scheme planning and configuration of the production line. It involves determining the overall structure of the production line according to the production requirements of products, process requirements, and the layout of resources. The design of configuration not only includes the selection and arrangement of equipment, but also involves the connection methods between equipment, transmission paths, station configurations, etc. The parameters of configuration include: dimensional parameters, length, width, area, etc.;

[0062] Dynamic type: Focuses on the action coordination and planning of each working unit and equipment in the production line. Specifically, it refers to how to ensure the smoothness and efficiency of the production process by reasonably arranging various tasks (such as equipment startup, operation, transfer, shutdown, etc.) during the production process. In the design of behavior, factors such as the working sequence of each equipment, task scheduling, and action timing need to be considered to avoid unnecessary waiting and conflicts. The parameters of the dynamic type include: movement speed, acceleration, etc.

[0063] Control type: Aims to achieve information exchange and coordinated control between various equipment and stations in the production line. The core of control design lies in how to achieve precise interaction between the information system and physical equipment. It includes the construction of the control network, data collection and analysis, decision-making and instruction issuance, etc. Effective control design can achieve automatic monitoring, real-time scheduling, and feedback adjustment of the production line, thereby enhancing the flexibility and response ability of the production process. The seamless connection between the information and physical worlds makes the entire production process more intelligent and adaptive. The parameters of the control type include: control accuracy, response time, etc.

[0064] Optimal type: The core of the production line optimization process. It achieves the optimal operation of the production process through the optimization of the whole-line drive engine. The key point of execution lies in precisely controlling each link of the production line through multi-dimensional data analysis and model optimization to achieve the best balance of production efficiency, quality, and cost. It involves optimization in multiple aspects such as production scheduling, equipment maintenance, energy management, and quality control. The parameters of the optimal type include: production capacity, yield, etc. The multi-link network model formed is as Figure 3 shown.

[0065] Preferably, the prediction model is specifically as follows:

[0066] L(t + 1) = AL(t) + Bβ(t)L(t) α + Du(t);

[0067] where A represents the inherent coupling relationship coefficient between parameter nodes, B represents the distribution strategy weight coefficient, D represents the path blockage influence coefficient, α is the capacity parameter, and both β(t) and u(t) are control variables, where β(t) represents the load distribution coefficient at time t, and u(t) is the propagation path switch of the 0 / 1 vector;

[0068] L(t) = [L1(t), L1(t),...LN (t)] T ;

[0069] where L N (t) represents the load of the Nth parameter node at time t, and N is the total number of parameter nodes;

[0070] There is a non-linear relationship between the load of the parameter node and the load of the parameter node at the next moment. The specific non-linear relationship is as follows:

[0071] L i (t + 1) = ωL i (t) + ∑η[L j (t)] θ ,

[0072] ω, θ, and η are all capacity parameters.

[0073] During the control period, the prediction model is predicted, and the control period requires predicting all moments within the period. At this time, the data of the next moment can be obtained through the non-linear relationship, and the non-linear relationship can be repeatedly used to obtain the load data of the parameter node at any moment within the control period.

[0074] Preferably, the objective function is specifically as follows:

[0075]

[0076] where

[0077]

[0078] where H p is the total length of the control period, Q i represents the load tracking weight of the ith parameter node, L ref,i represents the tracking load balancing target of the ith parameter node, R i represents the threshold adjustment weight of the ith parameter node, ΔC i (t) is also a control variable, and ΔC i (t) represents the capacity threshold of the ith parameter node at time t, and S i represents the path blocking weight of the ith parameter node.

[0079] The first term tracks the load balancing target L ref ; the load tracking deviation cost, which penalizes the load of the parameter node deviating from the ideal reference value. The second term is for penalizing frequent threshold adjustments; it penalizes the adjustment amplitude of the capacity threshold ΔC i (t) of the parameter node. The third term is For frequent adjustment of the penalty threshold; the number of decisions or criticality of the penalty to block the propagation path.

[0080] It is worth mentioning that Q is the load tracking weight, and a larger Q is given to high-sensitivity nodes (prone to avalanche). i , and strictly control their load deviation. S is the path blocking weight, and for the core process path (such as welding → painting), a very high S is set. i , and blocking can be avoided as much as possible.

[0081] Preferably, the objective function also has the following constraint conditions:

[0082] For the node capacity limit;

[0083] For the constraint of the control variable range;

[0084] For the path blocking logic;

[0085] ∑ΔC i (t + k) ≤ ΔC max For the total resource constraint;

[0086] Where β max represents the maximum load distribution coefficient, and C max represents the capacity threshold of the maximum parameter node.

[0087] A production line change propagation control system based on model predictive control, using the above-mentioned production line change propagation control method based on model predictive control, includes: a construction module, a setting module, a monitoring module, and an output module;

[0088] The construction module is used to abstract the key parameters in the production process into a network based on the multi-layer network theory, and construct a multi-dimensional network model based on the relationship between the key parameters;

[0089] Based on the multi-dimensional network model, find the changed parameter nodes and other key nodes affected by them as the first nodes, and construct a prediction model using the first nodes;

[0090] The setting module is used to set the control period and the data threshold;

[0091] The monitoring module is used to record the first moment of the control period, and obtain the prediction data of each moment within the remaining control period through the prediction model, and judge whether all the prediction data is less than the data threshold. If all the prediction data is less than the data threshold, the output module is called;

[0092] If not all of the predicted data are less than the data threshold, the control variables in the prediction model are adjusted through the objective function to obtain an updated prediction model, and the output module is called;

[0093] The output module is used to delete the first moment, update the next moment as the first moment, and call the monitoring module again until the first moment is the last moment within the control period, and output the control variables in the prediction model.

[0094] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0095] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A production line change propagation control method based on model predictive control, characterized in that, It includes the following steps: Step S1: Based on the multi-layer network theory, abstract the key parameters produced on the production line into a network, and construct a multi-dimensional network model based on the relationships between the key parameters. Based on the multi-dimensional network model, find the changed parameter nodes and other key nodes affected by them as the first nodes, and use the first nodes to construct a prediction model; Step S2: Set the control period and the data threshold; Step S3: Record the first moment of the control period, and obtain the predicted data at each moment within the remaining control period through the prediction model. Judge whether all the predicted data are less than the data threshold. If all the predicted data are less than the data threshold, then execute Step S4; If not all the predicted data are less than the data threshold, adjust the control variables in the prediction model through the objective function to obtain an updated prediction model, and execute Step S4; Step S4: Delete the first moment, update the next moment as the first moment, and re-execute Step S3 until the first moment is the last moment within the control period, and output the control variables in the prediction model.

2. The method for controlling the propagation of production line changes based on model predictive control according to claim 1, characterized in that, The key parameters include the configuration, dynamic type, control type, and optimal type of the production line, and each key parameter contains at least one parameter node; The construction criteria for the multi-dimensional network model in Step S1 are as follows:

3. A method for controlling production line change propagation based on model predictive control according to claim 1, characterized in that Each layer of the network consists of parameter nodes of the same nature. Abstract the relationships between different parameter nodes of the production line as edges. The parameter nodes and edges form the multi-dimensional network model of the production line. L(t + 1)=AL(t)+Bβ(t)L(t) α +Du(t); The prediction model is specifically as follows: L(t) = [L1(t), L1(t), … L N (t)] T ; where L N (t) represents the load of the Nth parameter node at time t, and N is the total number of parameter nodes; Where A represents the inherent coupling relationship coefficient between parameter nodes, B represents the allocation strategy weight coefficient, D represents the path blockage influence coefficient, α is the capacity parameter, β(t) and u(t) are both control variables, where β(t) represents the load distribution coefficient at time t, and u(t) is the propagation path switch of the 0 / 1 vector; L i (t + 1) = ωL i (t) + ∑η[L j (t)] θ , Where there is a non-linear relationship between the load of the parameter node and the load of the parameter node at the next moment. The non-linear relationship is specifically as follows:

4. A method for controlling the propagation of production line changes based on model predictive control according to claim 1, characterized in that, ω, θ, and η are all capacity parameters. Among them Among them, H p is the total length of the control period, Q i represents the load tracking weight of the i-th parameter node, L ref,i represents the tracking load balancing target of the i-th parameter node, R i represents the threshold adjustment weight of the i-th parameter node, ΔC i (t) is also a control variable, ΔC i (t) represents the capacity threshold of the i-th parameter node at the t-th moment, S i represents the path blocking weight of the i-th parameter node.

5. A method for controlling production line change propagation based on model predictive control according to claim 4, characterized in that The objective function is specifically as follows: ∑ΔC i (t + k) ≤ ΔC max ; where β max represents the maximum load distribution coefficient, and C max represents the capacity threshold of the maximum parameter node.

6. A production line change propagation control system based on model predictive control, characterized in that, The following constraint conditions also exist for the objective function: Using the method for controlling the propagation of production line changes based on model predictive control according to any one of claims 1 to 5, includes: a construction module, a setting module, a monitoring module, and an output module; The construction module is used to abstract the key parameters produced on the production line into a network based on the multi-layer network theory, and construct a multi-dimensional network model based on the relationships between the key parameters; Based on the multi-dimensional network model, find the changed parameter nodes and other key nodes affected by them as the first nodes, and use the first nodes to construct a prediction model; The setting module is used to set the control period and the data threshold; The monitoring module is used to record the first moment of the control period, and obtain the predicted data at each moment within the remaining control period through the prediction model, and judge whether all the predicted data are less than the data threshold. If all the predicted data are less than the data threshold, then call the output module; If not all the predicted data are less than the data threshold, adjust the control variables in the prediction model through the objective function to obtain an updated prediction model, and call the output module; The output module is used to delete the first moment, update the next moment as the first moment, and recall the monitoring module until the first moment is the last moment within the control period, and output the control variable in the prediction model.

Images