Method, apparatus, medium and product for generating plant layout

Abstract

A method, a production unit, a medium and a product for generating a plant layout are provided. The method comprises the following steps: obtaining layout parameters, wherein the layout parameters comprise workshop parameters and transportation parameters; determining a first position of a first production unit based on the layout parameter and a first production unit parameter of the first production unit; determining a second position of a second production unit based on the layout parameter, the first position and a second production unit parameter of the second production unit; and generating a layout of the workshop at least based on the first position and the second position. By using the position information of the previous production unit and combining the workshop parameters and the transportation parameters, the optimized workshop layout can be quickly and automatically generated, so that the efficiency of the workshop layout can be improved, and the production scheduling is more reasonable.

Description

Method, device, medium and product for generating shop layout Technical Field

[0001] Embodiments of the present disclosure relate generally to the field of computers, and more particularly to methods, devices, media, and products for generating shop floor layouts. Background Art

[0002] The workshop is the basic unit of space for organizing production activities in a factory or enterprise. Multiple production units are arranged within the workshop, and these units work together to complete the factory's or enterprise's production, processing, and logistics plans. Consequently, the quality of the workshop's layout design directly impacts the factory's or enterprise's operational efficiency and the rationality of production scheduling. An optimized layout design can improve operational efficiency, streamline production scheduling, and increase logistics speed. Therefore, workshop layout design is a crucial component of factory planning.

[0003] Summary of the Invention

[0004] Although there are some methods for workshop layout design in the prior art, the inventors of this application have noticed that traditional workshop layout design solutions are highly dependent on the experience of designers, perform poorly in global optimization capabilities, and take too long to calculate.

[0005] To this end, embodiments of the present disclosure provide methods, devices, media, and products for generating a workshop layout. According to embodiments of the present disclosure, a trained layout generation model is used to generate a workshop layout, enabling rapid and automatic generation of an optimized workshop layout, thereby improving the efficiency of the workshop layout and making production scheduling more rational.

[0006] In a first aspect of the present disclosure, a method for generating a workshop layout is provided. The method comprises: obtaining layout parameters, the layout parameters including workshop parameters and transportation parameters; determining a first position of a first production unit based on the layout parameters and first production unit parameters of a first production unit; determining a second position of a second production unit based on the first position of the layout parameters and second production unit parameters of a second production unit; and generating a workshop layout based on at least the first position and the second position. By utilizing the location information of previous production units, combined with the workshop parameters and transportation parameters, an optimized workshop layout can be automatically generated, thereby improving the efficiency of the workshop layout and making production scheduling more reasonable.

[0007] In some embodiments, the method further includes: obtaining parameters of the production unit to be arranged; and determining a current position of the production unit to be arranged based on the layout parameters, the parameters of the production unit to be arranged, and the determined previous position of the previous production unit. Generating a workshop layout based on at least the first position and the second position includes: generating a workshop layout based on the determined previous position of the previous production unit and the current position of the production unit to be arranged, wherein the previous position includes the first position and the second position. Similar operations can be performed on all generated units to be arranged, thereby enabling rapid and automated workshop layout.

[0008] In some embodiments, the previous location and the current location meet constraints, and the constraints are associated with transportation parameters. By utilizing the transportation parameters to further constrain the positions of the production units to be arranged and the previous production units in the workshop layout, the logistics information and the layout information complement each other, thereby achieving an optimized workshop layout.

[0009] In some embodiments, where the transport parameters include the length, width, speed, number, and turning radius of the transport devices used for the layout, the constraints include: in the layout, the minimum distance between the production unit to be arranged and the previous production order must be no less than the width of the transport device and no less than the turning radius of the transport device. By setting the minimum distance between two production units in conjunction with the constraints, the transport operation of the transport device during the logistics process can be facilitated.

[0010] In some embodiments, the method is performed by a trained layout generation model, and the layout generation model is trained by: generating a sample layout based on sample parameters in a received sample parameter set; evaluating the sample layout using an evaluation model and obtaining sample evaluation parameters from the evaluation model; constructing a sample reward function based on the sample evaluation parameters; and adjusting the layout generation strategy adopted by the layout generation model based on the constructed sample reward function. During the training of the layout generation model, constructing a reward function based on the evaluation parameters of the sample layout by the evaluation model and adjusting the adopted layout generation strategy based on the constructed sample reward function can enable the layout generation model to learn a strategy corresponding to generating an optimized layout.

[0011] In some embodiments, the sample layout includes multiple sample production units, the evaluation model includes a parameterized simulation model, and evaluating the sample layout using the evaluation model and obtaining sample evaluation parameters from the evaluation model includes: determining multiple sample positions corresponding to the multiple sample production units based on the sample layout; inputting the multiple sample positions and sample parameters into the parameterized simulation model; and obtaining sample evaluation parameters output by the parameterized simulation model. By constructing a sample reward function using the sample evaluation parameters obtained by simulating the sample layout with the parameterized simulation model, the feedback required for training the layout generation model can be accurately obtained, thereby making policy adjustments more effective and accurate.

[0012] In some embodiments, the parameterized simulation model includes a parameter input layer and a simulation layer, the parameter input layer is used to receive multiple sample locations and sample parameters, wherein the sample parameters include workshop sample parameters, transport sample parameters, sample production unit parameters corresponding to multiple sample production units, task sample parameters, and scheduling sample parameters, and the simulation layer constructs a simulation layout based on the workshop sample parameters, multiple sample locations, and sample production unit parameters, and the simulation layer simulates on the constructed simulation layout based on the task sample parameters and the scheduling sample parameters and the constraints associated with the transport sample parameters, and generates sample evaluation parameters. By dividing the parameterized simulation model into the parameter input layer and the simulation layer, the logic of the parameterized simulation model can be improved, and during the simulation, based on the task parameters and the production parameters, the logistics information and production information of the layout application can be effectively taken into account, thereby providing a more comprehensive evaluation for the layout and helping to optimize the layout design.

[0013] In some embodiments, the sample layout includes multiple sample production units, and the evaluation model includes a graph neural network-based evaluation model. Evaluating the sample layout using the evaluation model and obtaining sample evaluation parameters from the evaluation model includes: determining a sample layout graph based on the sample layout; inputting the sample layout graph into the graph neural network-based evaluation model; and receiving the sample evaluation parameters output by the graph neural network-based evaluation model. By evaluating the sample layout using the graph neural network-based evaluation model, the evaluation speed can be greatly improved, thereby increasing the speed of execution of the method.

[0014] In some embodiments, determining a sample layout graph based on the sample layout includes: determining a first sample position of a first sample production unit among a plurality of sample production units in the sample layout; determining a second sample position of a second sample production unit among the plurality of sample production units in the sample layout; and constructing a sample layout graph based on the first sample position and first sample production unit parameters of the first sample production unit and the second sample position and second sample production unit parameters of the second sample production unit, wherein the sample layout graph includes a first node, a second node, and an edge connecting the first node and the second node, the first node and the second node corresponding to the first sample production unit and the second sample production unit, respectively, and wherein the edge corresponds to a process production flow between the first sample production unit and the second sample production unit, the process production flow being obtained based on the first sample production unit parameters and the second sample production unit parameters. By corresponding the nodes and edges in the layout graph to the production units and the corresponding process production flows, a rapid and comprehensive evaluation of the sample layout can be performed using an evaluation model based on a graph neural network.

[0015] In some embodiments, the graph neural network-based evaluation model is obtained through training, and sample data used to train the graph neural network-based evaluation model is obtained based on sample device parameters of a sample device in combination with constraints, and the loss used to train the graph neural network-based evaluation model is obtained by simulation based on the sample data using a parameterized simulation model. By adopting the above-mentioned method to obtain sample data and loss for training the graph neural network-based evaluation model, the graph neural network-based evaluation model can more accurately evaluate the generated layout.

[0016] In some embodiments, the method for generating a layout further includes: validating the layout using an evaluation model and obtaining evaluation parameters from the evaluation model; constructing a reward function for generating the layout based on the evaluation parameters; and validating whether the generated layout meets predetermined conditions based on the reward function. Further validation of the generated layout can determine whether the layout generation model is capable of stably generating an optimized layout.

[0017] In some embodiments, wherein the layout includes multiple production units, the evaluation model includes a parameterized simulation model, and wherein validating the generated layout using the evaluation model and obtaining evaluation parameters from the evaluation model includes: determining, based on the layout, multiple positions corresponding to the multiple production units; inputting production unit parameters of the multiple positions and the multiple production units, as well as layout parameters, into the parameterized simulation model; and obtaining evaluation parameters output by the parameterized simulation model. By validating the generated layout using the parameterized model, the evaluation parameters of the generated layout can be obtained more accurately.

[0018] In some embodiments, wherein the layout includes multiple production units, and the evaluation model includes an evaluation model based on a graph neural network, and wherein verifying the generated layout using the evaluation model and obtaining evaluation parameters from the evaluation model includes: determining a layout graph based on the layout; inputting the layout graph into the evaluation model based on the graph neural network; and receiving evaluation parameters output by the evaluation model based on the graph neural network based on the layout graph, wherein the layout graph includes a first node, a second node, and an edge connecting the first node and the second node, and wherein the first node and the second node correspond to a first production unit and a second production unit, respectively, the edge corresponds to a process production flow between the first production unit and the second production unit, and the process production flow is determined based on the first production unit parameters and the second production unit parameters. By adopting the evaluation model based on the graph neural network, the evaluation parameters of the layout can be quickly obtained.

[0019] In a second aspect of the present disclosure, a device for generating a workshop layout is provided, comprising: a layout parameter acquisition module configured to acquire layout parameters, the layout parameters including workshop parameters and transportation parameters; a first determination module configured to determine a first position of a first production unit based on the layout parameters and a first production unit parameter of the first production unit; a second determination module configured to determine a second position of a second production unit based on the layout parameters, the first position, and the second production unit parameter of the second production unit; and a layout generation module configured to generate a workshop layout based at least on the first position and the second position.

[0020] In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes: one or more processors; and a memory for storing computer-executable instructions, wherein the computer-executable instructions, when executed by the one or more processors, enable the electronic device to implement any of the methods described in the first aspect above.

[0021] In a fourth aspect of the present disclosure, a computer-readable medium is provided, on which computer-executable instructions are stored. When the computer-executable instructions are executed, at least one processor is caused to perform the method according to any one of the above first aspects.

[0022] In a fifth aspect of the present disclosure, a computer program product is provided, comprising computer-executable instructions, which, when executed, cause at least one processor to perform the method according to any one of the above first aspects.

[0023] This summary is provided to introduce a selection of concepts in a simplified form that are further described in the detailed description below. This summary is not intended to identify key features or essential features of the disclosure, nor is it intended to limit the scope of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above-mentioned characteristics, technical features, advantages and implementation methods of the present disclosure will be further described below in a clear and understandable manner through the description of preferred embodiments in conjunction with the accompanying drawings, wherein:

[0025] FIG1A shows a schematic diagram of an example workshop layout according to an embodiment of the present disclosure;

[0026] FIG1B illustrates a block diagram showing components for generating a shop floor layout according to an embodiment of the present disclosure;

[0027] FIG2 shows a schematic flow chart of generating a workshop layout according to an embodiment of the present disclosure;

[0028] FIG3 shows a schematic diagram of generating a workshop layout according to an embodiment of the present disclosure;

[0029] FIG4 shows a schematic flowchart of a method for training a layout generation model according to an embodiment of the present disclosure;

[0030] FIG5 shows a schematic block diagram of a parameterized simulation model according to an embodiment of the present disclosure;

[0031] FIG6 shows a schematic diagram of modeling a production unit according to an embodiment of the present disclosure;

[0032] FIG7 shows a schematic diagram of modeling a transportation device according to an embodiment of the present disclosure;

[0033] FIG8 shows a flowchart 800 of a simulation control strategy for a transportation device according to an embodiment of the present disclosure;

[0034] FIG9 shows evaluation parameters for evaluating a sample layout according to an embodiment of the present disclosure;

[0035] FIG10 shows a process of generating sample data for training an evaluation model based on a graph neural network according to an embodiment of the present disclosure; and

[0036] FIG. 11 shows a block diagram of an example device that may be used to implement embodiments of the present disclosure.

[0037] List of reference numerals:

[0038] 110: Workshop area

[0039] 121: Layout parameters

[0040] 126: Layout Generation Model

[0041] 123: Workshop Layout

[0042] 122: Parametric Simulation Model

[0043] 124: Evaluation Model Based on Graph Neural Network

[0044] 320-1: Production Unit 1

[0045] 320-2: Production Unit 2

[0046] 320-(N-1): Production unit (N-1)

[0047] 320-N: Production unit N

[0048] 510: parameter input layer

[0049] 520: Simulation layer

[0050] 511: Workshop sample parameters

[0051] 512: Sample production unit attribute parameters

[0052] 513: Task sample parameters and scheduling sample parameters

[0053] 514: Transport sample parameters

[0054] 530: Physical parameter model

[0055] 531: Workshop Structure

[0056] 533: Modeling of Transport Devices

[0057] 535: Sample production unit

[0058] 540: Data layer

[0059] 542: Mission Data

[0060] 543: Evaluation indicators

[0061] 544: Process production data

[0062] 545: Transport device data

[0063] 546: Scheduling data

[0064] 548: Control Methods

[0065] 610: Functional logic layer

[0066] 611: Processing Unit

[0067] 612: Entry buffer

[0068] 613: Export buffer

[0069] 614: Interaction location

[0070] 620: Model layer

[0071] 621: Modeling Production Units

[0072] 622: Attribute parameters of production unit

[0073] 623: Other attributes

[0074] 630: Sample production unit attribute parameters

[0075] 710: Transport Device Physical Layer

[0076] 712: Transport device pool

[0077] 713: Transport Device

[0078] 720: Transport sample parameters

[0079] 722: Attribute parameters

[0080] 724: Transporter Task List

[0081] 726: Transport device scheduling method

[0082] 920: Evaluation indicators

[0083] 910: Cost Dimension

[0084] 920: Structural Dimension

[0085] 930: Production Dimension

[0086] 940: Logistics Dimension

[0087] 1010: Sample production unit parameters

[0088] 1020: Sample parameters

[0089] 1030: Sample Graphics

[0090] 1101: CPU

[0091] 1102: ROM

[0092] 1103: RAM

[0093] 1104: Bus

[0094] 1105: I / O interface

[0095] 1106: Input unit

[0096] 1107: Output unit

[0097] 1108: Storage unit

[0098] 109: Communication unit DETAILED DESCRIPTION

[0099] The principles of the present disclosure will be described below with reference to several exemplary embodiments shown in the accompanying drawings. Although preferred embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that these embodiments are described only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way.

[0100] As used herein, the term "including" and its variations represent open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "based at least in part on". The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". Other explicit and implicit definitions may also be included below. Some example embodiments of the present disclosure are described below with reference to Figures 1-11.

[0101] The workshop is the basic unit of space for organizing production activities in a factory or enterprise. Multiple production units are arranged within the workshop, and these units work together to complete the factory's or enterprise's production, processing, and logistics plans. Consequently, the quality of the workshop's layout design directly impacts the factory's or enterprise's operational efficiency and the rationality of production scheduling. An optimized layout design can improve operational efficiency, streamline production scheduling, and increase logistics speed. Therefore, workshop layout design is a crucial component of factory planning.

[0102] Although there are some methods for workshop layout design in the prior art, traditional workshop layout design solutions are highly dependent on the experience of designers, perform poorly in global optimization capabilities, and take too long to calculate.

[0103] To this end, embodiments of the present disclosure provide methods, production units, media, and products for generating a workshop layout. According to embodiments of the present disclosure, a trained layout generation model is used to generate a workshop layout, enabling rapid and automatic generation of an optimized workshop layout, thereby improving workshop layout efficiency and making production scheduling more rational.

[0104] FIG1A illustrates a schematic diagram of an example workshop layout according to an embodiment of the present disclosure. This example workshop layout can be applied to workshop layouts. As shown in FIG1 , area 110 represents a workshop area and has architectural boundaries defined according to workshop parameters, such as a length of L and a width of W. The interior of workshop area 110 includes multiple production units (as represented by the rectangular parallelepiped and cube in FIG1 , where different rectangular parallelepipeds and cubes can represent different types of production units). In one embodiment, a production unit can include production equipment; in another embodiment, a production unit can include production equipment and its associated equipment. Production units can be defined based on the actual application of the workshop, and this disclosure is not limited in this regard. The position of each production unit in the layout can be determined according to the layout generation method according to an embodiment of the present disclosure, and a workshop layout can be generated based on the determined positions, as shown in FIG1 . This automatically generates an optimized workshop layout, thereby improving workshop layout efficiency and making production scheduling more efficient. It should be understood that the layout diagram shown in FIG1A is merely illustrative, and the layout diagram 100 in FIG1A can have various other forms depending on actual application scenarios.

[0105] Figure 1B shows a block diagram 100 of layout generation according to an embodiment of the present disclosure. The layout generation model 126 can receive layout parameters 121 and generate a layout 123 of the workshop based on the layout parameters 121. The layout generation model 126 can be obtained based on training. During the training process, each time the layout generation model 126 generates a sample layout (which can be called an "iteration"), the generated sample layout can be input into the parameterized simulation model 122, and the parameterized simulation model 122 evaluates the sample layout to obtain evaluation parameters. The layout generation model 126 constructs a reward function based on the evaluation parameters and adjusts the layout generation strategy adopted for generating the layout based on the reward function. In one embodiment, a model 124 based on a graph neural network can also be used to evaluate the sample layout to obtain evaluation parameters. The layout generation model constructs a reward function based on the evaluation parameters and adjusts the layout generation strategy based on the reward function. The model 124 based on a graph neural network can be trained by the parameterized simulation model 122. Furthermore, to verify the stability of layout generation model 126 , after layout generation model 126 is trained, during the application phase, the layout generated by layout generation model 126 can be input into parameterized model 122 for simulation verification, or input into graph neural network-based model 124 for evaluation verification. This process will be described in detail below with reference to the accompanying figures.

[0106] FIG2 shows a flowchart of a method 200 for generating a layout according to an embodiment of the present disclosure. The method 200 may be executed by a computing device. The computing device may include, but is not limited to, a personal computer, a server computer, a handheld or laptop device, a mobile device (such as a mobile phone, a personal digital assistant (PDA), a media player, etc.), a multi-processor system, a consumer electronic product, a wearable electronic device, a smart home device, a minicomputer, a mainframe computer, an edge computing device, a distributed computing environment including any of the above systems or devices, etc. Furthermore, the order of the steps in the method 200 is only illustrative, and the order of the steps in the method 200 may vary depending on the actual application.

[0107] As shown in FIG2 , at step 210 , the computing device obtains layout parameters, wherein the layout parameters include workshop parameters and transportation parameters. In one embodiment, the workshop parameters include parameters for constructing the infrastructure of the workshop, which may include, but are not limited to:

[0108] ---The size parameters of the workshop, for example, the length L and width W of the workshop, represented by [L s ,W s ];

[0109] ---Warehouse location parameter, represented by [x w ,y w ], where the warehouse can be adjacent to the workshop and used as the material taking port or material discharge port of the workshop. It can be assumed that the coordinates of the upper right corner of the workshop are (0,0). Accordingly, the x in the location parameter of the warehouse is w =0 or y w =0, that is, the warehouse can be placed on one side of the workshop area 110 in FIG1 (not shown) and used as a material taking port or a material discharge port of the workshop; and

[0110] ---The stop position parameters of the transport device in the workshop [x A ,y A ], wherein the transport device may include a device for performing logistics transportation in the workshop, including but not limited to an automatic guided vehicle (AGV), a transport robot and other devices capable of performing logistics transportation functions, and the docking point of the transport device may be located on one side of the workshop, for example, the docking point of the transport device may be placed on one side of the workshop area 110 in FIG1 (not shown), and accordingly, the x in the docking point position parameter A =0 or y A =0.

[0111] In one embodiment, the transport parameters characterize the parameters of the transport device used for logistics transport in the workshop, including but not limited to the length, width, turning radius, speed, and number of the transport device, which can be expressed as [l A ,wA ,TR A ,s A ,num A ].

[0112] At block 204, the computing device determines a first position of the first production unit based on the layout parameters and the first production unit parameters of the first production unit. The first production unit parameters of the first production unit include attribute parameters inherent to the first production unit, and may include the length and width of the production unit, the process name OP in the process production flow, and the process processing time ct of the production unit. In one embodiment, the first production unit parameters may be expressed as [l D,1 ,w D,1 ,OP D,1 ,ct D,1 The first position after determination includes the position of the first production unit in the layout, which can be expressed as [x D,1 ,y D,1 ]. In addition, the first direction d of the first production unit can be determined based on the layout parameters and the first production unit parameters of the first production unit. D,1 Correspondingly, the pose parameters of the first production unit in the layout can be expressed as [x D,1 ,y D,1 ,d D,1 ], where the pose may include a position parameter and an orientation parameter. According to an embodiment of the present disclosure, the orientation parameter may be selected from the following set [0, 90, 180, 270]. It is understood that, depending on the specific practical application, the orientation parameter in the present disclosure may also be selected from any other angle, and the present disclosure is not limited thereto.

[0113] In one embodiment, the production units are mapped as rectangles (as shown in layout 100 in Figure 1) with a length of l and a width of w, and different types of production units can be represented by rectangles of different lengths and widths, thereby enabling the layout generation method according to an embodiment of the present disclosure to be applicable to layout design in various scenarios, thereby improving the applicability of the present disclosure.

[0114] At block 206, the computing device determines a second position of the second production unit based on the layout parameters, the first position, and the second production unit parameters of the second production unit. The second production unit parameters of the second production unit include attribute parameters inherent to the second production unit, and may include the length and width of the production unit, the name of the process step in the process production flow, and the processing time for the production unit to process the process step. In one embodiment, the second production unit parameters can be expressed as [l D,2 ,w D,2 ,OP D,2 ,ct D,2The second position after determination includes the position of the second production unit in the layout, which can be expressed as [x D,2 ,y D,2 ]. In addition, the second position and the second direction of the second production unit can be determined based on the layout parameters, the first position, the first direction and the second production unit parameters of the second production unit. D,2 Correspondingly, the pose parameters of the second production unit in the workshop layout can be expressed as [x D,2 ,y D,2 ,d D,2 ]. At block 208 , the computing device generates a plant layout based on at least the first location and the second location.

[0115] The process of generating a workshop layout will be described in more detail below in conjunction with FIG3. As shown in FIG3, in the initial state, no production unit is placed in the workshop area 310. The computing device can obtain layout parameters for the layout, as described in conjunction with FIG2, and the layout parameters include workshop parameters and transportation parameters. The workshop parameters may include the size parameters of the workshop 310 [L s ,W s ]、Warehouse location parameters[x w ,y w ] and the stop location parameters of the transport device [x A ,y A ], and the transportation parameters include the length, width, turning radius, speed, and number of the transportation devices used in the workshop 310, which can be expressed as [l A ,w A ,TR A ,s A ,num A ].

[0116] The schematic workshop layout in FIG3 is a schematic diagram of a layout for a group of production units (production units 1 to N). In one embodiment, the group of production units can be pre-processed to determine the placement order of each production unit in the group. The placement order can be determined based on the actual application of the layout and is not limited in this disclosure.

[0117] The computing device can obtain the first production unit 320-1 and the production unit parameters of the first production unit according to the determined placement order of the production units. D,1 ,w D,1 ,OP D,1 ,ct D,1 The computing device combines the layout parameters and the first production unit parameters of the first production unit 320-1 to determine the first position [x D,1 ,y D,1], the first position represents the position of the first production unit in the workshop layout, as shown by the placement of the first production unit in 310-1 in FIG3. In addition, the first direction d of the first production unit can also be determined based on the layout parameters and the first production unit parameters of the first production unit. D,1 , the pose parameters of the first production unit in the layout can be expressed as [x D,1 ,y D,1 ,d D,1 ].

[0118] The computing device may obtain the second production unit 320-2 and calculate the first position [x D,1 ,y D,1 ] and the second production unit parameter [l D,2 ,w D,2 ,OP D,2 ,ct D,2 ], determine the second position [x D,2 ,y D,2 ], the second position represents the position of the second production unit in the workshop layout, as shown by the placement of the second production unit in 310-2 in FIG3. In addition, the second position and the second direction d of the second production unit can also be determined based on the layout parameters, the first position, the first direction, and the second production unit parameters of the second production unit. D,2 Correspondingly, the pose parameters of the second production unit in the layout can be expressed as [x D,2 ,y D,2 ,d D,2 ].

[0119] In one embodiment, the first position and the second position meet the constraints, and the constraints are related to the transport parameters, such as the length, width, turning radius, speed, and number of transport devices (expressed as [l A ,w A ,TR A ,s A ,num A Specifically, the constraint condition includes: the minimum distance between the first production unit and the second production unit is not less than the width w of the transport device A and not less than the turning radius TR of the transport device A .

[0120] The computing device can continue to determine the position of the next production unit in the same manner as above. In one embodiment, the computing device can obtain the third production unit and calculate the position of the next production unit based on the layout parameters, the first position [x D,1 ,y D,1 ], the second position [x D,2 ,yD,2 ] and the third production unit parameters of the third production unit [l D,3 ,w D,3 ,OP D,3 ,ct D,3 ], determine the third position of the third production unit [x D,3 ,y D,3 In one embodiment, the first position, the second position and the third position all meet the above constraints, that is, the minimum distance between the first production unit, the second production unit and the third production unit is not less than the width w of the transport device. A and not less than the turning radius TR of the transport device A In addition, a third position and a third orientation of the third production unit, i.e., a third pose parameter, may be determined based on the layout parameters, the first position, the first orientation, the second position, the second orientation, and the third production unit parameters of the third production unit.

[0121] Similarly, the calculation production unit can determine the current position of the current production unit based on the layout parameters, the previous position of the previous production unit determined, and the production unit parameters of the current production unit. For example, the production unit to be laid out is the Nth production unit, and the calculation device can obtain the Mth production unit (1≤M≤N, and M is a positive integer), and based on the layout parameters, the position of the previous (M-1) production units, and the Mth production unit parameters of the Mth production unit [l D,M ,w D,M ,OP D,M ,ct D,M ], determine the Mth position of the Mth production unit [x D,M ,y D,M In one embodiment, the computing device may determine the position and orientation of the current production unit based on the layout parameters, the previous position and orientation of the previous production unit, and the production unit parameters of the current production unit.

[0122] In one embodiment, the determined plurality of positions (e.g., the first position, the second position, ..., the Nth position) meet the constraints, and the constraints are consistent with the transport parameters, such as the length, width, turning radius, speed, and number of transport devices (expressed as [l A ,w A ,TR A ,s A ,num A Specifically, the constraint condition includes: the minimum distance between the above-mentioned multiple production units (for example, the first production unit, the second production unit, ..., the Nth production unit) is not less than the width w of the transport device A and not less than the turning radius TR of the transport device A. By utilizing the width and turning radius of the transport device to constrain the minimum distance between the above-mentioned multiple production units, the transport device can be moved, rotated, and other transport operations in the workshop, thereby providing layout support for the logistics stage and effectively optimizing the layout design. Accordingly, the computing device can determine the workshop layout based on the acquired current position and the determined previous position of the previous production equipment, and in one embodiment, the previous position can include a first position and a second position, that is, the computing device can generate a workshop layout 310-N based on the determined first position, second position, ..., Nth position.

[0123] In one embodiment, the layout parameters may also include scheduling parameters and task parameters associated with the product. In one embodiment, the scheduling parameters may include scheduling information during the production process, for example, the scheduling situation for each day of the week, represented as [d1, ..., d a ,……,d7], where d a Indicates the scheduling situation of a certain day of the week; in addition, the scheduling information can also include the start time of the production schedule h s and end time h e , for example, [h s,i h e,i ] represent the start time and end time of the i-th production schedule respectively. The start time and end time of multiple production schedules can be expressed as {[h s,1， h e,1 ],[h s,2， h e,2 ,],……,[h s,i h e,i In one embodiment, the task parameters may include the number of products in each task, the process steps in the process flow required to produce the product, and the processing time of the corresponding production unit in the corresponding process steps. The task parameters may be expressed as o={o1, o2, ..., o j}, where o k represents the kth task (1≤k≤j, and k is a positive integer), and for each element o in task o k， o k ={n k ,[(OP k,1 ,ct k,1 ),……,(OP k,m ,ct k,m )]}, where n k Indicates the number of products required in the kth task, (OP k,v ,ct k,v ) represents the process steps OP required in the production process of the corresponding task kAnd the corresponding production unit performs the process step OP k The time required k,v In one embodiment, the task may include order information, such as the quantity of products required in each order and the production process required to produce the product. In one embodiment, the production process may also be determined based on the task parameters, for example, for the task o k ={n k ,[(OP k,1 ,ct k,1 ),……,(OP k,m ,ct k,m )]}, the process production flow of the corresponding product can be determined from the above parameters as OP k,1 →OP k,2 →……→OP k,m , and combined with the production unit parameters, the production unit that executes each process step in the process production flow can be determined, so that the process production flow can be executed by the corresponding production unit to complete the product production in the task.

[0124] By including scheduling parameters and task parameters in the layout parameters, production scheduling information is considered in the layout generation strategy, so that a reasonable workshop layout can be generated more optimally and efficiently to meet the subsequent production process and logistics process.

[0125] The method 200 for generating a layout according to an embodiment of the present disclosure can be executed by a trained layout generation model in a computing device. In one embodiment, the layout generation model can use a decision-making method such as a Markov decision process to perform layout. Specifically, for each iteration (i.e., each time the layout is determined), the previous position of the previously placed production unit can be used as a state, and the possible placement of the current production unit to be placed in a certain position can be used as an action. The reward of each state except the last state (corresponding to the generated layout) in each iteration is determined to be 0, and the reward function for the adopted layout strategy is determined based on the evaluation of the generated layout. In one embodiment, the layout generation model determines and generates the layout based on the decision-making process. Therefore, by training the layout generation model, the layout generation strategy can be continuously optimized and adjusted, so that the trained layout generation model can adopt an optimized layout generation strategy to generate the layout.

[0126] FIG4 illustrates a training method 400 for training a layout generation model according to an embodiment of the present disclosure. Method 400 may be performed by a computing device. At block 401, the computing device generates a sample layout based on sample parameters in a received sample parameter set. In one embodiment, the layout generation model to be trained receives a sample parameter set S. In one embodiment, the sample parameter set S may include multiple sets of sample parameters, each set of sample parameters may include:

[0127] ---Workshop sample parameters {[L ss ,W ss ],[x ws ,y ws ][x As ,y As ]};

[0128] ---Transport sample parameters {[l As ,w As ,TR As ,s As ,num As ]};as well as

[0129] --- Sample production unit parameters corresponding to multiple sample production units (for example, P sample production units) {[l Ds,1s ,w Ds,1s ,OP Ds,1s ,ct Ds,1s ],……,[l Ds,ps ,w Ds,ps ,OP Ds,ps ,ct Ds,ps ]}.

[0130] In addition, in one embodiment, each set of sample parameters may further include task sample parameters {o 1s ,o 2s ,……,o js}, where o ks ={n ks ,[(OP ks,1s ,ct ks,1s ),……,(OP ks,ms ,ct ks,ms )]}(1≤k≤j). In addition, each set of sample parameters may also include scheduling sample parameters {[d 1s ,……,d as ,……,d 7s ];[h ss,1s， h es,1s ],[h ss,2s， h es,2s ],……,[h s,i h e,iIt is understandable that the meaning of each parameter in each set of sample parameters can be understood in conjunction with the content described above, and for the sake of brevity, it will not be repeated here.

[0131] The layout generation model to be trained generates corresponding sample layouts for each set of sample parameters. s For example, assuming that there are n groups of sample parameters in the sample parameter set S, correspondingly, the sample parameter set S = {s1, s2, ..., sn}, and each group of sample parameters si may include the sample parameters described above, for example, si includes the following parameters:

[0132] ---{[L ss ,W ss ],[x ws ,y ws ],[x As ,y As ]};

[0133] ---{[l As ,w As ,TR As ,s As ,num As ]};

[0134] ---{[l Ds,1s ,w Ds,1s ,OP Ds,1s ,ct Ds,1s ],……,[l Ds,ps ,w Ds,ps ,OP Ds,ps ,ct Ds,ps ]};

[0135] ---{[d 1s ,……,d as ,……,d 7s ];[h ss,1s, h es,1s ],[h ss,2s, h es,2s ,],……,[h s,i h e,i ]};

[0136] ---{o 1s ,o 2s ,……,o js}, where o ks ={n ks ,[(OP ks,1s ,ct ks,1s ),……,(OP ks,ms ,ct ks,ms )]}(1≤k≤j).

[0137] The layout generation model generates a sample layout Layout based on the sample parameters si si In the sample layout si In the example, the sample positions of multiple sample production units are determined, including the sample positions of t sample production units: {[x Ds,1s ,y Ds,1s ],[x Ds,2s ,y Ds,2s ],……,[x Ds,ts ,y Ds,ts ]}.

[0138] In block 404, the layout generation model can evaluate the sample layout using the evaluation model and obtain sample evaluation parameters from the evaluation model. As described above, the layout generation model sets the reward for each state except the last state (corresponding to the generated layout) in each iteration to 0, and determines the reward function of the adopted layout strategy based on the evaluation of the generated layout. Accordingly, during the training phase, the layout generation model can evaluate the sample layout using the evaluation model and obtain sample evaluation parameters from the evaluation model. In one embodiment, the evaluation parameters may include workshop site utilization SiteUtili(p,g), production unit efficiency EquipEffic(p,g), and transportation device efficiency AGVEffic(p,g). Where (p,g) represents the generated layout p for the task parameter g. It should be understood that the above evaluation parameters are merely illustrative, and the generated layout can be evaluated and other types of evaluation parameters can be obtained based on the key aspects of focus in actual applications. This disclosure does not limit the evaluation parameters. According to embodiments of the present disclosure, samples can be evaluated in a variety of ways, which will be described in detail below.

[0139] In block 406, a sample reward function is constructed based on the sample evaluation parameters. In one embodiment, the sample reward function R p,g It can be a linear combination of evaluation parameters, such as shown in Formula 1:

[0140] R p,g =-SiteUtili(p,g)-EquipEffic(p,g)-AGVEffic(p,g) (Formula 1)

[0141] It is understandable that other combinations of evaluation parameters may be used to calculate the reward function, and this disclosure does not limit this.

[0142] At block 408, based on the constructed sample reward function, the layout generation strategy π adopted by the layout generation model is adjusted. θ In one embodiment, the sample reward function R can be calculated based on the following formula (2):p,g To further construct the objective function J(θ, G), so as to adjust the strategy adopted in generating the layout model according to the objective function.

[0143]

[0144] in, According to the strategy π θ The expected reward function for the layout P generated for a given task g is obtained by adjusting the policy π θ The above objective function is made to meet predetermined conditions, thereby obtaining a trained layout generation model.

[0145] The following describes in detail the process of evaluating a sample layout and obtaining sample evaluation parameters from an evaluation model. In one embodiment, the evaluation model comprises a parameterized simulation model. The parameterized simulation model can receive a set of parameters, construct a simulation layout based on the parameters, and further simulate within the simulation layout based on the parameters, generating output simulation evaluation metrics. The following describes in detail the process and implementation of simulating a sample layout using the parameterized simulation model, with reference to Figures 5-9.

[0146] In the process of training the layout generation model, each time a sample layout is generated (which can be called an "iteration"), the generated sample layout can be evaluated to adjust the layout generation strategy. In the case of using a parameterized simulation model, the layout generation model can determine the multiple sample positions and sample directions corresponding to the multiple sample production units in the sample layout according to the sample layout. The layout generation model generates a sample layout for the sample parameters si. si For example, the layout generation model can determine the sample layout Layout si The sample positions of multiple sample production units in , including the sample positions of t sample production units {[x Ds,1s ,y Ds,1s ],[x Ds,2s ,y Ds,2s ],……,[x Ds,ts ,y Ds,ts ]} (in one embodiment, it may also include the sample direction). The layout generation model converts multiple sample positions {[x Ds,1s ,y Ds,1s ],[x Ds,2s ,y Ds,2s ],……,[x Ds,ts ,y Ds,ts ]} (in one embodiment, it may also include the sample direction) and the sample parameters si (refer to the parameter list listed above) are input into the parameterized simulation model, and the parameterized simulation model is further obtained for the sample layout Layout siSample evaluation parameters output after simulation include, but are not limited to, workshop site utilization SiteUtili, production unit efficiency EquipEffic, and transportation device efficiency AGVEffic.

[0147] FIG5 is a schematic block diagram of a parameterized simulation model 500 according to an embodiment of the present disclosure. In one embodiment, as shown in FIG5 , the parameterized simulation model 500 includes a parameter input layer 510 and a simulation layer 520. The parameter input layer 510 is used to receive a plurality of sample positions (in one embodiment, it may also include sample directions) and sample parameters, where the sample parameters may include the following:

[0148] ---Workshop sample parameters {[L ss ,W ss ],[x ws ,y ws ],[x As ,y As ]};

[0149] ---Transport sample parameters {[l As ,w As ,TR As ,s As ,num As ]};

[0150] ---Sample production unit parameters for multiple sample production units {[l Ds,1s ,w Ds,1s ,OP Ds,1s ,ct Ds,1s ],……,[l Ds,ps ,w Ds,ps ,OP Ds,ps ,ct Ds,ps ]};

[0151] ---Scheduling sample parameters {[d 1s ,……,d as ,……,d 7s ];[h ss,1s, h es,1s ],[h ss,2s, h es,2s ,],……,[h s,i h e,i ]};

[0152] ---Scheduling sample parameters {o 1s ,o 2s ,……,o js}, where o ks ={n ks ,[(OP ks,1s ,ct ks,1s),……,(OP ks,ms ,ct ks,ms )]}(1≤k≤j).

[0153] In the parameter input layer 510, various types of data can be logically classified according to categories. For example, in the parameter input layer, the received sample parameters and sample positions (which may also include sample directions) can be logically classified as follows:

[0154] ---Workshop sample parameter 511: {[L ss ,W ss ],[x ws ,y ws ][x As ,y As ]};

[0155] ---Sample production unit attribute parameters 512:

[0156] Including sample production unit parameters and sample position (can also include sample direction d Ds,ts ) parameter combination: {[l Ds,1s ,w Ds,1s ,x Ds,1s ,y Ds,1s ,d Ds,1s ,OP Ds,1s ,ct Ds,1s ],……,[l Ds,ts ,w Ds,ts ,x Ds,ts ,y Ds,ts ,d Ds,ts ,OP Ds,ts ,ct Ds,ts ]};

[0157] ---Task sample parameters and scheduling sample parameters 513: including task sample parameters and scheduling sample parameters, expressed as: {[d 1s ,……,d as ,……,d 7s ];[h ss,1s, h es,1s ],[h ss,2s, h es,2s ,],……,[h s,i h e,i ]};{o 1s ,o 2s ,……,o js}, where o ks ={n ks ,[(OP ks,1s ,ct ks,1s ),……,(OP ks,ms ,ct ks,ms)]}(1≤k≤j); and

[0158] ---Transportation sample parameters 514: Characterize the relevant parameters of the transportation device, which can be expressed as {[l As ,w As ,TR As ,s As , num As ]}.

[0159] In one embodiment, the task sample parameters and scheduling sample parameters 513 may also include various tasks determined according to the task information. k Corresponding process production process OP k,1 →OP k,2 →……→OP k,m .

[0160] The simulation layer 520 constructs a simulation layout based on the received workshop sample parameters, multiple sample locations (which may also include sample directions), and sample production unit parameters. Furthermore, the simulation layer 520 performs simulation on the constructed simulation layout based on the task sample parameters, scheduling sample parameters, and constraints associated with the transport sample parameters, and generates the sample evaluation parameters. Specifically, the simulation layer 520 includes a data layer 540 and a physical parameter model 530. The physical parameter model 530 constructs a workshop architecture (as shown in 531 in FIG. 5 ) based on the workshop sample parameters 511 and places multiple sample production units (as shown in 535 in FIG. 5 ) based on the sample production unit attribute parameters 512. Thus, the simulation layer 520 constructs a simulated layout of the sample production units. Furthermore, the data layer 540 constructs task data 542, process production data 544, scheduling data 546, transport device data 545, and a control method 548 based on the task sample parameters, scheduling sample parameters 513, transport sample parameters 514, and constraints associated with the transport sample parameters. The transport device data 545 is input to the physical parameter model 530. When the simulation is initialized, the parameterized simulation model models the num As A transport device (shown as 533 in FIG5 ) is constructed. Accordingly, based on the constructed task data 542 , process production data 544 , scheduling data 546 , transport device data 545 , and control method 548 , the parameterized simulation model runs a simulation in the constructed simulation layout in conjunction with the modeled transport device 533 , and outputs evaluation indicators (e.g., sample evaluation parameters) 543 .

[0161] The following describes a schematic diagram 600 of modeling a sample production unit according to an embodiment of the present disclosure in conjunction with FIG6 . As shown in FIG6 , the modeled sample production unit 600 includes a front-end model layer 620 and a back-end functional logic layer 610. Based on the received sample production unit attribute parameters 630, for example, the sample production unit attribute parameters {[l Ds,is ,w Ds,is ,x Ds,is, y Ds,is, d Ds,is ,OP Ds,is ,ct Ds,is ], the front-end model layer 620 can model 621 the sample production unit i, and the sample production unit i can have attribute parameters including length, width, position and direction, as shown in 622. In addition, the sample production unit can also have other attributes, such as name, etc., as shown in Figure 623. The functional logic layer 610 of the sample production unit simulates the functional components of the sample production unit in the workshop, for example, including an interaction position 614 for loading and unloading materials with a transportation device (e.g., AGV), an entry buffer 612, an exit buffer 613 and a processing unit 611. The transportation device unloads the upstream materials to be processed into the entry buffer 612 and queues them for processing. After the processing is completed, the processing unit 611 releases the materials to the exit buffer 613, waiting for the transportation device to load the materials to the interaction position.

[0162] It should be understood that while the above description of FIG. 6 is based on a sample production unit, it is understood that the production unit modeling in FIG. 6 can be applied to any production unit, including but not limited to the modeling of a workshop production unit when evaluating a layout design. Other modeling approaches may also be employed depending on the actual application of the production unit, and this disclosure is not intended to limit such approaches.

[0163] FIG7 shows a schematic diagram of modeling a transport device according to an embodiment of the present disclosure. In one embodiment, a transport device, such as an AGV, can be modeled based on transport sample parameters 720. For example, the transport sample parameters 720 may include attribute parameters 722, a transport device task list 724, and a transport device scheduling method 726. The transport device physical layer 710 can model transport devices 713-1, 713-2, ..., 713-num that meet the transport sample parameters 720 from the transport device pool 712 based on the transport sample parameters. A , where num A It should be understood that Figure 7 is only schematic, and although the above description is based on the transport sample parameters, it can be understood that the production unit modeling in Figure 7 can be for any transport parameters.

[0164] After the transport device is modeled, the parametric simulation device can perform simulated control of the transport device within the constructed simulation layout according to a predefined control strategy. Below, a flowchart 800 of a simulated control strategy for a transport device according to an embodiment of the present disclosure will be described in conjunction with FIG8 . It should be understood that the control flow in FIG8 is merely illustrative, and other control flows may be selected based on specific plant applications, production logistics, and other information, and this disclosure is not limited thereto.

[0165] As shown in FIG8 , control flow 800 begins at block 810 . At block 811 , the parameterized simulation model receives input sample data. At block 812 , the parameterized simulation model executes scheduling operations based on the scheduling results. At block 820 , the parameterized simulation model determines whether the current operation is the first operation of the job. If the determination is "yes," the model checks at block 821 whether the transporter will arrive at the loading / unloading station ahead of schedule. If the determination is "no," the model checks at block 822 whether the transporter will arrive at the machine and demand point ahead of schedule. At block 823 , the parameterized simulation model moves the transporter from its current location to the demand point of the next job. Next, at block 830 , the parameterized simulation model checks whether the job is ready. If so, the model moves the job to the machine to be scheduled for the next job at block 832 . If not, the model waits for the job to become ready at block 831 before proceeding to block 832 to move the job to the machine to be scheduled for the next job. Next, at block 840, the parametric simulation model checks whether the machine's buffer is full. If this check is "no," the transporter unloads the job into the buffer at block 841 and waits until the machine has a free job before loading it onto the machine. If this check is "yes," the transporter loads the job onto the machine at block 842. The parametric simulation model then determines at block 850 whether all operations have been completed. If so, the model outputs the evaluation parameters. Otherwise, the model returns to block 812 to continue scheduling operations.

[0166] After the above simulation, the evaluation parameters of the sample layout of the workshop can be determined. The layout of the workshop not only considers the occupied area, but also needs to consider the production efficiency and logistics operation efficiency. Accordingly, the evaluation parameters can be considered from multiple dimensions. Figure 9 shows the evaluation parameters for evaluating the sample layout according to an embodiment of the present disclosure. As shown in Figure 9, the evaluation index 920 may include, for example, four dimensions:

[0167] 910: Cost dimension, including, but not limited to, operating costs;

[0168] 920: Structural dimensions, including but not limited to site utilization SiteUtili;

[0169] 930: Production dimensions, including but not limited to, average work in process (WIP), average completion time, average due time, production rate, average utilization of logistics equipment; and

[0170] 940: Logistical dimensions, including but not limited to spatial scalability.

[0171] The above describes the process of evaluating sample layouts using a parameterized model. It is understandable that during the training process, the parameterized model can be used iteratively to evaluate sample layouts, thereby completing the training of the layout generation model.

[0172] In an alternative embodiment, the evaluation model may further include an evaluation model based on a graph neural network. A sample layout graph may be determined based on the sample layout, the sample layout graph may be input into the evaluation model based on the graph neural network to obtain sample evaluation parameters from the evaluation model, and the sample evaluation parameters output by the evaluation model based on the graph neural network may be received.

[0173] Specifically, when determining a sample layout graph based on a sample layout, a first sample position loc1 of a first sample production unit among a plurality of sample production units in the sample layout can be determined, and a second sample position loc2 of a second sample production unit among a plurality of sample production units in the sample layout can be determined. Based on the first sample position loc1 and the first sample production unit parameters of the first sample production unit and the second sample position loc2 and the second sample production unit parameters of the second sample production unit, a sample layout graph is constructed. In one embodiment, the sample layout graph includes a first node, a second node, and an edge connecting the first node and the second node, the first node and the second node corresponding to the first sample production unit and the second sample production unit, respectively, and the edge corresponding to the process production flow between the first sample production unit and the second sample production unit, and the process production flow is based on the first sample production unit parameters [l Ds,1s ,w Ds,1s ,OP Ds,1s ,ct Ds,1s ] and the second sample production unit parameters [l Ds,2s ,w Ds,2s ,OP Ds,2s ,ct Ds,2s The evaluation model based on graph neural networks can evaluate sample layout graphs, thereby generating and outputting evaluation parameters, such as workshop site utilization (SiteUtili), production unit efficiency (EquipEffic), and transportation device efficiency (AGVEffic).

[0174] In one embodiment, the evaluation model based on the graph neural network is obtained through training, and can be obtained through the parameterized simulation model described above. When training the evaluation model based on the graph neural network, the required sample data can be obtained based on the sample production unit parameters of the sample production unit and in combination with the constraints. As shown in Figure 10, a process 1000 for generating sample data for training a model based on a graph neural network according to an embodiment of the present disclosure is illustrated.

[0175] 1010 in Figure 10 is the sample production unit parameters of the sample production unit, for example, the length, width, process step name in the process flow, processing time, etc. of the sample production unit. Based on predetermined constraints, the position and direction of the sample production unit can be set. The agreed conditions may include: the production unit is within the boundaries defined by the size of the workshop; the production units cannot overlap with each other; and there is a certain distance between the production units, where the distance is related to the parameters of the transportation device. Through the above constraints, the corresponding position and direction (i.e., posture) can be set for each sample production unit, and the posture parameters and production unit parameters are further combined to obtain sample parameters 1020. Further, based on the generated sample parameters, i.e., the production unit parameters and posture parameters, a sample graphic 1030 can be obtained. As shown in Figure 10, the sample graph 1030 may include multiple nodes (as represented by circles in 1030) and edges connecting two nodes (as represented by line segments in 1030). Each node in the graph 1030 corresponds to each sample production unit in the sample production unit, and the edge connecting two sample production units corresponds to the process production flow between the corresponding sample production units. The process production flow between the corresponding sample production units corresponding to the edge connecting two nodes can be determined based on the production unit parameters such as the process step name and processing time in the process flow in the sample production unit parameters.

[0176] The parameterized model described above can be used to train the evaluation model based on the graph neural network. The evaluation model based on the graph neural network generates an evaluation parameter P1 based on the sample parameters, and the parameterized model generates an evaluation parameter P0 based on the sample parameters. The evaluation parameter P0 generated by the parameterized model can be regarded as the ground-truth. A loss function can be further constructed based on P1 and P0, thereby minimizing the loss function to adjust the parameters of the graph neural network model. During the parameter adjustment process, the following formulas (3) and (4) can be used:

[0177]

[0178]

[0179] Among them, vi and vj represent the feature vectors of the i-th node and the j-th node in the layout graph, w ij e is the weight between two nodes. By applying full connection and cascade to the two nodes, the edge e between the two nodes is realized. ji Each node updates its feature vector by passing the average of the feature vectors of its input edges and output edges (i.e., adjacent edges) to another fully connected network.

[0180] Because graph neural network-based evaluation models have a high processing speed, using graph neural network-based evaluation models during the training of the layout generation model can significantly reduce training time. The above describes the training process for the generation layout model. It is understood that other methods can be used to train the generation layout model, and this disclosure is not limited to this.

[0181] Returning to Figure 2 , after the workshop layout is generated in block 208 , to ensure the stability of the output of the generated layout model, the generated workshop layout can be further evaluated. This can be done using a process similar to that used to evaluate sample layouts during training. Evaluation parameters can be obtained by employing the aforementioned parameterized model or a graph neural network-based model. A reward function is then constructed based on the evaluation parameters to determine whether the generated workshop layout meets expectations.

[0182] In one embodiment, the evaluation model includes a parameterized simulation model, and wherein the evaluation model is used to verify the generated layout and obtain evaluation parameters from the evaluation model, including: the computing device can determine multiple positions (and may also include corresponding multiple directions) corresponding to multiple production units according to the layout, and input the multiple positions (correspondingly, may also include multiple directions), the production unit parameters of the multiple production units, and the layout parameters into the parameterized simulation model, so that the parameterized simulation model performs simulation layout construction and simulation according to the received data. The computing device can obtain the evaluation parameter P output by the parameterized simulation model. Based on the obtained evaluation parameter P, the computing device can construct a reward function and determine whether the output of the layout generation model meets expectations based on the constructed reward function. The structure and specific simulation process of the parameterized simulation model used to evaluate the layout are similar to the parameterized model described above in the training process. For the sake of simplicity, they will not be repeated here.

[0183] In addition, in one embodiment, the evaluation model may also include an evaluation model based on a graph neural network. When the generated layout is verified using the graph neural network-based evaluation model and evaluation parameters are obtained, the computing device may determine a layout graph (similar to the graph illustrated in 1030 in FIG. 10 ) based on the layout. The computing device may input the layout graph into the graph neural network-based evaluation model and receive evaluation parameters output by the graph neural network-based evaluation model based on the layout graph. The evaluation parameters may include workshop site utilization (SiteUtili), production unit efficiency (EquipEffic), and transportation equipment efficiency (AGVEffic). In one embodiment, the layout graph may include a first node, a second node, and an edge connecting the first and second nodes. The first and second nodes correspond to a first and second production unit, respectively, and the edge corresponds to a process flow between the first and second production units. The process flow may be determined based on the first and second production unit parameters. It is understood that the computing device may generate corresponding nodes and edges for each of the multiple production units in a similar manner. This section can be understood in conjunction with the description of the sample graph above and will not be further elaborated here for the sake of simplicity.

[0184] According to an embodiment of the present disclosure, there is also provided an apparatus for generating a workshop layout, comprising: a layout parameter acquisition module configured to acquire layout parameters, the layout parameters including workshop parameters and transportation parameters; a first determination module configured to determine a first position of a first production unit based on the layout parameters and a first production unit parameter of the first production unit; a second determination module configured to determine a second position of a second production unit based on the layout parameters, the first position, and a second production unit parameter of the second production unit; and a layout generation module configured to generate a workshop layout based on at least the first position and the second position. It will be understood that the apparatus for generating a workshop layout according to an embodiment of the present disclosure can be implemented in combination with the various method embodiments described above, and will not be further described herein.

[0185] FIG11 shows a schematic block diagram of an example device 1100 that can be used to implement an embodiment of the present disclosure. The device 1100 can be used to implement the method 200 described in FIG2 and / or the method 400 described in FIG4. For example, the local device and / or cloud server described above can be implemented by one or more components in the example device 1100. As shown in FIG11, the device 1100 includes a processor, such as a central processing unit (CPU) 1101, which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) 1102 or computer program instructions loaded from a storage unit 1108 into a random access memory (RAM) 1103. In RAM 1103, various programs and data required for the operation of the device 1100 can also be stored. CPU 1101, ROM 1102, and RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0186] Various components in device 1100 are connected to I / O interface 1105, including an input unit 1106, such as a keyboard and mouse; an output unit 1107, such as various types of displays and speakers; a storage unit 1108, such as a magnetic disk and optical disk; and a communication unit 1109, such as a network card, a modem, a wireless communication transceiver, etc. Communication unit 1109 allows device 1100 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0187] The processing unit 1101 performs the various methods and processes described above, such as method 200 and / or method 400 described in Figure 4. For example, in some embodiments, method 200 and / or method 400 can be implemented as a computer software program or computer program product, which is tangibly contained in a computer-readable medium, such as a non-transient computer-readable medium (e.g., storage unit 1108). In some embodiments, part or all of the computer program can be loaded and / or installed on the device 1100 via ROM 1102 and / or communication unit 1109. When the computer program is loaded into RAM 1103 and executed by CPU 1101, one or more steps in the method or process described above can be performed. Alternatively, in other embodiments, CPU 1101 can be configured to perform the method or process of an embodiment of the present disclosure in any other appropriate manner (e.g., by means of firmware).

[0188] It should be understood by those skilled in the art that the various steps of the method of the present disclosure described above can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Alternatively, they can be implemented using program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the present disclosure is not limited to any specific combination of hardware and software. For example, certain embodiments of the present disclosure also include various program modules and / or integrated circuit modules for executing one or more steps of method 200 and / or one or more other steps described in other embodiments of the present disclosure. These program modules can be included or embodied in a device, such as the device 800 in Figure 8.

[0189] According to an exemplary implementation of the present disclosure, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein the computer-executable instructions are executed by a processor to implement the method described above. According to an exemplary implementation of the present disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the method described above.

[0190] Various aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, apparatuses, devices, and computer program products implemented according to the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer-readable program instructions.

[0191] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable information presentation device, thereby producing a machine such that when these instructions are executed by the processing unit of the computer or other programmable information presentation device, a device is generated that implements the functions / actions specified in one or more blocks in the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium, where these instructions cause the computer, programmable information presentation device, and / or other device to operate in a specific manner, such that the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks in the flowchart and / or block diagram.

[0192] Computer-readable program instructions can be loaded onto a computer, other programmable information presentation apparatus, or other device, so that a series of operational steps are performed on the computer, other programmable information presentation apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable information presentation apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowcharts and / or block diagrams of the embodiments of the present disclosure.

[0193] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple implementations of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a part for a module, program segment or instruction, and a part for a module, program segment or instruction comprises one or more executable instructions for realizing the logical function of the specification. In some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be realized by a special hardware-based system that performs the function or action of the specification, or can be realized by a combination of special hardware and computer instructions.

[0194] While various implementations of the present disclosure have been described above, the foregoing description is intended to be illustrative, not exhaustive, and not limited to the disclosed implementations. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described implementations. The terminology used herein is selected to best explain the principles of the implementations, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the various implementations disclosed herein.

Claims

1. A method for generating a workshop layout, comprising: Acquiring (202) layout parameters, the layout parameters including workshop parameters and transportation parameters; determining ( 204 ) a first position of the first production unit based on the layout parameters and a first production unit parameter of the first production unit; determining ( 206 ) a second position of the second production unit based on the layout parameters, the first position, and second production unit parameters of the second production unit; as well as The plant layout is generated (208) based on at least the first position and the second position.

2. The method according to claim 1, further comprising: Obtaining parameters of the production unit to be arranged; Determine the current position of the production unit to be arranged based on the layout parameters, the parameters of the production unit to be arranged and the previous position of the previous production unit determined, Wherein, generating (208) the workshop layout based on at least the first position and the second position includes: generating (208) the shop floor layout based on the determined previous positions of the previous production units and the current positions of the production units to be laid out, The previous position includes the first position and the second position. 3 . The method of claim 2 , wherein the previous location and the current location comply with constraints, and wherein the constraints are associated with the transportation parameters.

4. The method according to claim 3, wherein the transport parameters include the length, width, speed, number and turning radius of the transport devices used for the layout, and the constraints include: In the layout, the minimum distance between the previous production unit and the production unit to be arranged is not less than the width of the transport device and not less than the turning radius of the transport device.

5. The method according to claim 1, wherein the method is performed by a trained layout generation model (126), and wherein the layout generation model is trained by: generating a sample layout based on the sample parameters in the received sample parameter set; Using an evaluation model (122; 124) to evaluate the sample layout, and obtaining sample evaluation parameters from the evaluation model (122; 124); Constructing a sample reward function based on the sample evaluation parameters; and The layout generation strategy adopted by the layout generation model (126) is adjusted based on the constructed sample reward function.

6. The method according to claim 5, wherein: The sample layout includes a plurality of sample production units, the evaluation model (122; 124) includes a parameterized simulation model (122), and the evaluation model (122; 124) is used to evaluate the sample layout and obtain the sample production units from the evaluation model (122; 124) sample evaluation parameters include: determining, according to the sample layout, a plurality of sample positions corresponding to the plurality of sample production units; inputting the plurality of sample positions and the sample parameters into the parameterized simulation model (122); and The sample evaluation parameters output by the parameterized simulation model (122) are obtained.

7. The method according to claim 6, wherein: The parameterized simulation model (122) includes a parameter input layer (510) and a simulation layer (520), wherein the parameter input layer (510) is used to receive the multiple sample locations and the sample parameters, wherein the sample parameters include workshop sample parameters, transportation sample parameters, sample production unit parameters corresponding to the multiple sample production units, task sample parameters and scheduling sample parameters, and the simulation layer (520) constructs a simulation layout based on the workshop sample parameters, the multiple sample locations and the sample production unit parameters, and the simulation layer (510) performs simulation on the constructed simulation layout based on the task sample parameters and the scheduling sample parameters and the constraints associated with the transportation sample parameters, and generates the sample evaluation parameters.

8. The method according to claim 5, wherein The sample layout includes a plurality of sample production units, and the evaluation model (122; 124) includes an evaluation model (124) based on a graph neural network, and wherein evaluating the sample layout using the evaluation model (122; 124) and obtaining sample evaluation parameters from the evaluation model (122; 124) includes: determining a sample layout graphic according to the sample layout; Inputting the sample layout graph into the graph neural network-based evaluation model (124); and Receive the sample evaluation parameters output by the graph neural network-based evaluation model (124).

9. The method according to claim 8, wherein Determining a sample layout graphic according to the sample layout includes: determining a first sample position of a first sample production unit of the plurality of sample production units in the sample layout; determining a second sample position of a second sample production unit of the plurality of sample production units in the sample layout; constructing the sample layout graph based on the first sample position and first sample production unit parameters of the first sample production unit and the second sample position and second sample production unit parameters of the second sample production unit, The sample layout graph includes a first node, a second node, and an edge connecting the first node and the second node, the first node and the second node correspond to the first sample production unit and the second sample production unit, respectively, and the edge corresponds to a process production flow between the first sample production unit and the second sample production unit, and the process production flow is obtained based on the first sample production unit parameters and the second sample production unit parameters.

10. The method according to claim 8, wherein The evaluation model (124) based on the graph neural network is obtained through training, and wherein, the sample data used to train the evaluation model (124) based on the graph neural network is obtained based on the sample production unit parameters of the sample production unit and in combination with the constraint conditions, and the loss used to train the evaluation model (124) based on the graph neural network is obtained by simulation based on the sample data according to the parameterized simulation model (122).

11. The method according to claim 1 , further comprising: The workshop layout is verified using an evaluation model (122; 124), and the evaluation model (122; 124) evaluation parameters; Constructing a reward function for generating the workshop layout based on the evaluation parameters; as well as Verify whether the generated workshop layout satisfies a predetermined condition based on the reward function.

12. The method according to claim 11, wherein The workshop layout includes a plurality of production units, the evaluation model (122; 124) includes a parameterized simulation model (122), and the evaluation model (122; 124) is used to verify the generated layout and obtain the evaluation results from the evaluation model (122; 124) include: Determining, according to the workshop layout, a plurality of locations corresponding to the plurality of production units; and inputting the plurality of locations, the production unit parameters of the plurality of production units, and the layout parameters into the parameterized simulation model (122; 124); The evaluation parameter output by the parameterized simulation model (122; 124) is obtained.

13. The method according to claim 11, wherein The layout includes a plurality of production units, and the evaluation model (122; 124) includes an evaluation model (124) based on a graph neural network, and wherein verifying the generated layout using the evaluation model (122; 124) and obtaining evaluation parameters from the evaluation model (122; 124) includes: determining a layout graphic according to the workshop layout; Inputting the layout graph into the graph neural network-based evaluation model (124); and receiving evaluation parameters outputted by the graph neural network-based evaluation model (124) according to the layout graph, The layout graph includes a first node, a second node, and an edge connecting the first node and the second node, and the first node and the second node correspond to the first production unit and the second production unit respectively, and the edge corresponds to the process production flow between the first production unit and the second production unit, and the process production flow is determined based on the first production unit parameters and the second production unit parameters.

14. A device for generating a workshop layout, comprising: A layout parameter acquisition module is configured to acquire layout parameters, wherein the layout parameters include workshop parameters and transportation parameters; a first determining module configured to determine a first position of the first production unit based on the layout parameter and a first production unit parameter of the first production unit; a second determining module configured to determine a second position of the second production unit based on the layout parameter, the first position, and a second production unit parameter of the second production unit; as well as The layout generation module is configured to generate the workshop layout based on at least the first position and the second position.

15. An electronic device (1100), comprising: one or more processors (1101); as well as A memory (1103) is configured to store computer-executable instructions, wherein the computer-executable instructions, when executed by the one or more processors (1101), enable the electronic device (1100) to implement the method according to any one of claims 1 to 13.

16. A computer-readable storage medium (1103) having computer-executable instructions stored thereon, wherein the computer-executable instructions, when executed by a processor (1101), implement the method according to any one of claims 1 to 13.

17. A computer program product comprising computer executable instructions which, when executed by a processor (1101), implement the method according to any one of claims 1 to 13.

Images