Method and device for providing samples for machine learning and storage medium
By establishing a production organization simulation model to generate a high-quality machine learning sample set, the problems of insufficient and poor sample quality in production organization decision-making by machine learning models are solved, thereby improving the model's decision-making ability and adaptability to environmental changes.
Patent Information
- Application Number
- CN202410629009.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-21
- Publication Date
- 2025-11-21
AI Technical Summary
Existing machine learning models face challenges in production organization decision-making, such as limited sample size, poor data quality, and poor ability to adapt to environmental changes, resulting in poor performance in practical applications.
By establishing a production organization simulation model, the production process is simulated, a large number of machine learning samples are generated, and the best data is selected by evaluating the model to generate a high-quality machine learning sample set.
This solves the problems of insufficient sample quantity and poor sample quality, laying the foundation for the practical application of machine learning in the field of production decision-making, and improving the model's decision-making ability and adaptability to environmental changes.
Smart Images
Figure CN120996591A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of digitization, and in particular to a method for providing samples for machine learning, a device and a storage medium. BACKGROUND
[0002] At present, artificial intelligence technology related to machine learning is developing rapidly, and there are a large number of successful application cases in the fields of image recognition, semantic analysis, etc. Some scholars now propose to apply machine learning to production decision-making fields, for example, in the job shop scheduling problem, by importing environmental variables into a machine learning model to obtain a decision output, thereby solving production organization decision-making problems such as workpiece selection of equipment and equipment selection of workpieces.
[0003] Among them, the machine learning model is mainly trained through historical data or online learning mode, which is feasible in theoretical research level, but faces great resistance in landing application stage. On the one hand, the data problem in the process of industrial production organization decision-making, which is characterized by small quantity, missing field, data distortion, etc., leads to poor decision-making ability of the model established by machine learning, and the effect is very poor, on the other hand, the ability to cope with environmental changes is poor, such as temporary maintenance of a device, urgent contract insertion, etc. The machine learning model trained for the conventional production environment cannot support the corresponding production organization decision-making.
[0004] Therefore, it is hoped that a high-precision production organization simulation model can be established to generate machine learning samples, which can better train machine learning models in various scenarios, thereby solving the above two problems and laying a foundation for the landing application of machine learning in the field of production decision-making. SUMMARY
[0005] In order to solve the above defects, the present application provides a method for providing samples for machine learning, the purpose of which is to establish a machine learning model for solving production organization decision-making problems, comprising the following steps:
[0006] Establishing a simulation model according to the characteristics of the production organization decision-making problem, which includes a simulation module of the equipment or rules required to solve the production organization decision-making problem;
[0007] Setting simulation parameters for the simulation model;
[0008] Repeating the simulation model to generate an initial data set of specified data required to solve the production organization decision-making problem;
[0009] Evaluating the data in the initial data set, and filtering out preferred data based on the set evaluation conditions to generate a preferred data set;
[0010] Numerical transformation is performed on the preferred data set to transform it into feature data and label data suitable for machine learning, thereby generating a machine learning sample set.
[0011] In the method described above, the evaluation of the data in the initial data set includes establishing a benefit evaluation model and a screening criterion, the benefit evaluation model evaluates the data in the initial data set, and the preferred data is selected based on the screening criterion.
[0012] In the method described above, the production organization decision problem is a production organization decision, and the simulation model includes:
[0013] A product process flow simulation module is used to simulate the processing path of products in the production line and carry the information flow of products among multiple simulation modules.
[0014] A device workflow simulation module is used to simulate the operation process of the device.
[0015] A device space layout simulation module is used to simulate the spatial layout of the device.
[0016] A scheduling rule simulation module is used to simulate the production organization process, organically combine products, devices, etc. to achieve the final production purpose, and must include rules for production organization decisions.
[0017] In the method described above, the scheduling rules include device calling rules and product processing priorities.
[0018] In the method described above, a plurality of modes are combined to set the scheduling rules of the production organization decision.
[0019] In the method described above, based on the four dimensions of processing device operation information, transfer device operation information, product information, and team information, simulation models under different production scenarios are established.
[0020] In the method described above, the specified data required to solve the production organization decision problem includes environment data, decision data, and index information related to the production organization decision task, the environment data refers to instantaneous information that affects production organization decisions, which is composed of device information and product information, the decision data corresponds to the production organization decision problem, and the index information refers to statistical information used to evaluate the pros and cons of decisions.
[0021] In the method described above, when the simulation model is triggered, the following process is performed:
[0022] The current instantaneous environment data is obtained and output.
[0023] The current decision data is output according to the scheduling rules.
[0024] Output statistically significant index information in a given time range.
[0025] Accordingly, the present application also proposes a computer readable medium, which stores instructions, and the instructions make the computer execute the method when executed on the computer.
[0026] Accordingly, the present application also proposes an electronic device, which comprises a memory for storing instructions executed by one or more processors of the electronic device, and a processor, which is one of the processors of the electronic device, for executing the method.
[0027] Compared with the prior art, the present application simulates the production process by establishing a simulation model of the production organization, thereby generating a large number of machine learning samples, and then scoring the large number of machine learning samples by evaluating the model, and taking the samples with high scores or filtering out the data with low scores, so that the method can generate a large number of high-quality learning samples for any production scenario, to a certain extent, solve the problems of small sample quantity and poor sample data quality, and lay a foundation for the application of machine learning in the field of production decision-making. BRIEF DESCRIPTION OF DRAWINGS
[0028] Figure 1 The application scenario diagram of the embodiments of the present application is shown in the figure;
[0029] Figure 2 The flowchart of the method for providing samples for machine learning according to some embodiments of the present application is shown in the figure;
[0030] Figure 3 The block diagram of the electronic device according to some embodiments of the present application is shown in the figure. DETAILED DESCRIPTION
[0031] The embodiments of the present application will be described by specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the description. Although the description of the present application will be introduced in combination with the preferred embodiments, it does not mean that the features of the present application are limited to the embodiments. On the contrary, the purpose of introducing the present application in combination with the embodiments is to cover other options or modifications that can be extended based on the claims of the present application. In order to provide a deep understanding of the present application, many specific details will be included in the following description. The present application can also be implemented without using these details. In addition, in order to avoid confusion or obscure the focus of the present application, some specific details will be omitted in the description. It should be noted that the embodiments and features in the embodiments in the present application can be combined with each other without conflict.
[0032] It should be noted that in the present specification, similar reference numbers and letters represent similar items in the following drawings, and therefore, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings, and the same definition applies.
[0033] It should also be declared that the numbering of the methods and processes in the present application is for ease of reference, and not to limit the sequence, and if there is a sequence between the steps, the written description shall prevail. Figure 2 Figure 2 The application scenario of the embodiments in the present application is shown in the figure. The steel plant processes the original steel into a certain shape and size of casting blank, which needs to go through three processes of converter processing 10, refining processing 11 and continuous casting processing 12. The converter processing 10 mixes the molten iron and scrap steel mixture for oxygen blowing smelting, and produces molten steel that meets the requirements of the next process by adding certain auxiliary materials and alloys. The refining processing 11 removes impurities from the molten steel. Specifically, the RH processing 111 removes gas, decarburizes, blows oxygen, heats up, controls composition, removes sulfur, phosphorus and inclusions, etc. The main function of the LF processing 112 is to remove inclusions, desulfurize, homogenize the composition of the molten steel and adjust the temperature of the molten steel, etc. Moreover, the refining processing can be performed more than once. When the system determines that the molten steel after the refining processing 11 is completed cannot meet the use requirements of the next process, the refining processing can be repeated several times. The continuous casting processing 12 continuously casts the high-temperature molten steel into a casting blank with a certain cross-sectional shape and size.
[0035] It can be seen that the molten steel needs to be transferred between the converter processing 10 and the refining processing 11, and between the refining processing 11 and the continuous casting processing 12, that is, the transfer equipment needs to be arranged to transport the molten steel. In actual production, there are usually multiple converter devices, multiple refining devices and multiple continuous casting devices. Moreover, the number of various devices is not one-to-one, and based on the time required for each device to complete processing, the transfer equipment needs to be reasonably arranged to maximize efficiency.
[0036] Based on the rapid development of machine learning technology, it is hoped that the work of reasonably arranging the transfer equipment can be solved by machine learning. However, the basis of machine learning is a sample set, and if there are not a large number of diversified samples, the decision model established by machine learning is defective and cannot be truly used in actual production.
[0037] The application idea of the present application is to establish simulation models of converter equipment, refining equipment and continuous casting equipment, combine the position maps of multiple converter equipment, refining equipment and continuous casting equipment in the factory and the layout map of transfer equipment on the track, simulate the running path of the transfer equipment when molten steel is transferred from the converter equipment to the refining equipment and the refining equipment to the continuous casting equipment, and then according to the coordination of the converter equipment, the refining equipment and the continuous casting equipment with the transfer equipment in a period of time, filter out the feasible or efficient running path that meets the demand from all simulated running paths, save it as a positive sample for subsequent machine learning. Through the simulation running of the converter equipment, the refining equipment and the continuous casting equipment, a large amount of data can be obtained in a short time, and these data are selectively optimized for a purpose, thereby effectively solving the problem of lack of samples in the prior art.
[0038] Specifically, as shown in Figure 2 , the figure is a flowchart of a method for providing samples for machine learning according to some embodiments of the present application. The purpose of the machine learning is to establish a machine learning model to solve the production organization decision problem. The following will be described in combination with the optimization calling of the transfer equipment in the steelmaking plant between the converter, the refining and the continuous casting process to explain the flowchart. Figure 2
[0039] S1, a simulation model is established according to the characteristics of the production organization decision problem, which includes simulation modules of equipment or rules required to solve the production organization decision problem.
[0040] The simulation model of the production organization in steelmaking includes four simulation modules of product process flow, equipment workflow, equipment space layout and scheduling rules.
[0041] The product process flow simulation module is used to simulate the processing path of the product in the production line and carry the information flow of the product among multiple simulation modules. For each product, the process flow is determined based on the product parameters. The product parameters include tapping mark, weight, process path and processing time of each process. The process path represents the process / equipment type that the product needs to go through. Assuming that BOF represents converter steelmaking, RH represents RH refining, LF represents LF refining and CCM represents continuous casting machine processing. The process path of a certain steelmaking is represented as processList <process>P1 = {BOF, RH, CCM}, meaning the product needs to be refined in a converter, then in the RH refining process, and finally processed in a continuous casting machine.
[0042] The product in the steelmaking area is molten steel. The flow of molten steel between various pieces of equipment uses ladles as the carrier; therefore, the product process flow here refers to the ladle process flow. Each ladle corresponds to molten steel with the same tapping mark. The ladle process flow in the steelmaking area can be found by referring to [reference needed]. Figure 1 and to Figure 1 Explanation.
[0043] The equipment workflow simulation module is used to simulate the operation of equipment. For each piece of equipment, a workflow simulation model is established based on the equipment parameters. In this embodiment, the processing equipment includes:
[0044] There are three converters, namely 1#BOF, 2#BOF and 3#BOF. After molten iron and scrap steel enter the converters, they are smelted by oxygen blowing and certain auxiliary materials and alloys are added to produce molten steel that meets the requirements of the next process.
[0045] Five refining units are designated as 1#RH, 2#RH, 3#RH, 1#LF, and 2#LF. The RH refining unit can perform degassing, decarburization, oxygen blowing, heating, composition control, desulfurization, phosphorus removal, and inclusion removal on molten steel. The main function of the LF refining unit is to remove inclusions from molten steel, desulfurize, homogenize the composition of molten steel, and adjust the temperature of molten steel.
[0046] There are three continuous casting machines, namely No. 1 CCM, No. 2 BCCM and No. 3 CCM. The continuous casting machines can continuously cast high-temperature molten steel into billets with certain cross-sectional shapes and dimensions.
[0047] The processing equipment parameters include: ① Status parameters, including Idle, Prepare, Work, Breakdown, and Repair; ② Time parameters, including Prepare time, Work time, and Repair time. The Prepare time and Work time are related to the current product's steel tapping mark and weight; ③ Other parameters: Failure rate and scheduled inspection frequency.
[0048] The transfer equipment consists of three overhead cranes, namely CRANE 1, CRANE 2 and CRANE 3, which are responsible for the transfer of steel ladles between the various pieces of equipment.
[0049] Transportation equipment parameters include: ① state parameters, including IDLE, MOVE, LOAD, UNLOAD, BREAKDOWN, REPAIR; ② time parameters: MOVE time, LOAD time, UNLOAD time, REPAIR time; ③ other parameters: failure rate, inspection frequency.
[0050] The equipment space layout simulation module is used to simulate the spatial layout of the equipment, including simulating the physical size of the equipment, the spatial distance between the equipment and the equipment, the crane travel area, and the crane travel path.
[0051] The scheduling rule simulation module is used to simulate the production organization process, organically combines products, equipment, etc. to achieve the final production purpose, which contains scheduling rules for solving the production organization decision problem, and the scheduling rules can include equipment calling rules, product processing priority, etc., such as refining equipment selection rules, continuous casting machine selection rules, crane transportation task priority rules. In this embodiment, the initial refining equipment selection rule adopts the shortest queue first principle, the continuous casting machine selection rule adopts the principle of selecting the equipment that finishes casting first, and the crane transportation task adopts the FIFO principle. It should be noted that here it is not limited to simple priority rules, but also mathematical optimization models, machine learning models, etc.
[0052] S2, set simulation parameters for the simulation model.
[0053] Here, two specific production scenarios are given as examples, and corresponding simulation parameter setting examples are given.
[0054] First, a crane is in a long-term maintenance state, and the parameter settings are as follows:
[0055] Product parameter setting: set according to historical production actual data;
[0056] Equipment parameter setting: set the maintenance time of the crane to infinity, i.e. REPAIRtime=infinity;
[0057] Other equipment parameters are set according to historical data; note that the actual crane maintenance time may be only 3 days, but in order to obtain sufficient sample size under this production scenario, the maintenance time can be set to infinity;
[0058] Dispatching rule setting: Since the learning sample of ladle selection of refining equipment needs to be generated in this embodiment, the refining equipment selection rule adopts a combined mode of "the shortest task queue first + the first idle equipment first + random selection", and the weighting coefficients of the three modes are set to 0.4, 0.4 and 0.2;
[0059] Simulation duration: 30 days;
[0060] Simulation running times: 1000 times;
[0061] Random seed generation mode: random.
[0062] Secondly, if a certain type of special product needs to be urgently produced recently, the parameter settings are as follows:
[0063] Product parameter setting: The tapping mark, weight, process path and processing time of each process of the product are all set to the type of special product;
[0064] Equipment parameter setting: Set according to historical data;
[0065] Dispatching rule setting: Since the learning sample of ladle selection of refining equipment needs to be generated in this embodiment, the refining equipment selection rule adopts a combined mode of "the shortest task queue first + random selection", and the proportion of ladles selected randomly is set to 0.3. The continuous casting machine selection rule and the crane transportation task priority rule are set according to the actual situation;
[0066] Simulation duration: 30 days;
[0067] Simulation running times: 1000 times;
[0068] Random seed generation mode: random.
[0069] S3, repeatedly running the simulation model to generate an initial data set of specified data required to solve the production organization decision-making problem.
[0070] The output specified data includes environmental data, decision-making data and index data. Whenever a product ends BOF steelmaking and raises the demand for "refining equipment selection", the specific process of outputting the specified data is as follows:
[0071] (1) Obtain the current instantaneous environmental data and output.
[0072] The environmental data refers to the instantaneous information affecting the refining equipment selection decision, which is composed of equipment information and product information, and the expression is Env={equipEnv, productEnv};
[0073] The device information includes device type, device number, current status, current status remaining time, processing task list, and is expressed as equipEnv = { equipType, equipNo, status, time, taskList <product>} If the current state is IDLE or BREAKDOWN, the time value is null.
[0074] The product information includes the current process, the current equipment number, the time left for the current process, the next process, and the subsequent process flow. The expression is productEnv = {currentProcess, currentEquipNo, time, nxtProcess, nxtProcessList <process>} If time is 0, it means that the product has finished the current process. Since in this embodiment, the information is only outputted when the refining equipment is selected, the ngtProcess in the product information only contains the RH and LF refining equipment.
[0075] (2) According to the decision returned by the scheduling rule, output the decision data of "refining equipment selection".
[0076] The scheduling rule selects a specific equipment number as the decision of equipment selection, and the decision data expression is Action = {equipType, equipNo}, which contains the information of equipment type and equipment number.
[0077] (3) Output the index data within a given time range.
[0078] In this embodiment, the average waiting time (waitTime) of the ladle in the next 30 minutes and the equipment utilization (equipUtil) are used as evaluation indexes to evaluate the decision benefit, and the expression is Eval = {waitTime, equipUtil | 30min}.
[0079] For each "refining equipment selection" requirement of the product, the above three steps are repeated, and the corresponding output environment data, decision data and index data form an initial data, and multiple initial data are collected into an initial data set.
[0080] S4, evaluate the data in the initial data set, and select the preferred data based on the set evaluation condition to generate a preferred data set. In one embodiment, a decision benefit evaluation model can be established to evaluate the data, and each data in the initial data set is quantitatively scored to select the data meeting the given requirements. Specifically, it can include:
[0081] (1) Establish a decision benefit evaluation model so that the final score falls between [0, 1]. In this embodiment, the evaluation model is represented by the following function:
[0082] Reward = f(Eval)
[0083] Reward is a floating point number related to index data, and its value falls between [0, 1]. The specific form of f(Eval) is:
[0084]
[0085] wherein, waitTime represents the waiting time, equipUtil represents the equipment utilization rate, minWaitTime and maxWaitTime correspond to the minimum and maximum waitTime in the initial data set, and minEquipUtil and maxEquipUtil correspond to the minimum and maximum equipUtil in the initial data set.
[0086] According to the business needs, the evaluation model can also consider the environmental data and the decision data, that is, in the following form: Reward1=f(Eval|Env), Reward2=f(Eval|Action), Reward3=f(Eval|Env,Action), and the specific form of the function is more complex, which is not described here.
[0087] (2) Determine the screening criteria to retain the data that meets the requirements to form the initial learning sample. In this embodiment, the screening criteria is: if Reward >= 0.7, the data is retained.
[0088] For the evaluation model considering the environmental data and the decision data, that is, the model outputs Reward1, Reward2, and Reward3, which are three-dimensional scores, the three can be averaged or weighted averaged to obtain a comprehensive score, or samples that meet the three-dimensional score can be selected.
[0089] S5, numerical conversion is performed on the preferred data set to convert it into feature data and label data suitable for machine learning, thereby generating a machine learning sample set.
[0090] (1) Numerical conversion of environmental data
[0091] The environmental data refers to the instantaneous information that affects the selection decision of the refining equipment, which is composed of equipment information and product information.
[0092]
[0093] (2) Numerical conversion of decision data;
[0094] In this embodiment, the decision is to select a refining equipment, and the refining equipment category only has RH and LF two categories, therefore, the conversion table of the decision data table Action={equipType,equipNo} is as follows:
[0095]
[0096] (3) The environmental data is taken as the feature data, and the decision data is taken as the label data to obtain the final machine learning sample.
[0097] The above embodiments obtain sufficient initial data samples by simulating the production process, and then screen them according to predetermined rules to obtain preferred data samples, which can enable the decision model established by machine learning to make more accurate or more appropriate decisions. The data samples obtained through simulation are huge, and the data retained after screening is the data with more optimal equipment utilization or waiting time, or other optimal characteristics, depending on the screening criteria in S4. Under the training of the optimal data samples, the decision model established by machine learning can also accurately select the optimal scheme / option. That is, the established decision model can more efficiently guide production activities, for example, which refining equipment to select.
[0098] Reference is now made to Figure 3 , which shows a block diagram of an electronic device 400 according to an embodiment of the present application. The electronic device 400 is used to implement the method of the present application for providing samples for machine learning, and the implementation entity can be a desktop computer device, a notebook computer device, a tablet computer device, a mobile terminal, etc., or a special control device on a large production line.
[0099] The electronic device 400 can include one or more processors 401 coupled to a controller hub 403. For at least one embodiment, the controller hub 403 communicates with the processor(s) 401 via a bus, such as a Front Side Bus (FSB), a Point-to-Point (PtP) interface such as QuickPath Interconnect (QPI), or similar connection 406. The processor(s) 401 execute instructions to perform a general type of processing operation. In one embodiment, the controller hub 403 includes, but is not limited to, a Graphics & Memory Controller Hub (GMCH) (not shown), which includes a memory controller and a graphics controller, and an Input / Output Hub (IOH) (which can be on a separate chip) (not shown). The GMCH is coupled to the IOH via an interconnect 406.
[0100] The electronic device 400 can also include a coprocessor 402 and a memory 404 coupled to the controller hub 403. Alternatively, one or both of the memory and the GMCH can be integrated within the processor (as described herein), and the memory 404 and coprocessor 402 are directly coupled to the processor 401 and the controller hub 403, with the controller hub 403 being in a single chip with the IOH.
[0101] The memory 404 can be, for example, a Dynamic Random Access Memory (DRAM), a Phase Change Memory (PCM), or a combination of both. One or more tangible, non-transitory computer-readable media for storing data and / or instructions can be included in the memory 404.
[0102] The computer-readable storage medium stores instructions therein, in particular, a transient and a permanent copy of the instructions. The instructions can include instructions that, when executed by at least one of the processors, cause the electronic device 400 to carry out the methods according to the present application. When the instructions are run on the computer, the computer is caused to perform the methods according to the present application as described above.
[0103] In one embodiment, the coprocessor 402 is a special-purpose processor, such as, for example, a high-throughput MIC (Many Integrated Core) processor, a network or communication processor, compression engine, graphics processor, a GPGPU (General- purpose graphics processing unit), embedded processor, or the like. The coprocessor 402 can be a Figure 3 special-purpose processor in one embodiment.
[0104] In one embodiment, the electronic device 400 can further include a network interface controller (NIC) 406. The network interface 406 can include a transceiver to provide a radio interface for the electronic device 400 to communicate to any other suitable device (e.g., a front end module, an antenna, etc.). In various embodiments, the network interface 406 can be integrated with other components of the electronic device 400. The network interface 406 can implement the functionality of the communication unit in the above-described embodiments.
[0105] The electronic device 400 can further include an input / output (I / O) device 405. The I / O 405 can include a user interface designed to enable a user to interact with the electronic device 400, a peripheral component interface designed to enable peripheral components to also interact with the electronic device 400, and / or a sensor designed to determine environmental conditions and / or location information related to the electronic device 400.
[0106] Notably, Figure 3 are merely exemplary. That is, although Figure 3 The electronic device 400 includes a processor 401, a controller hub 403, a memory 404, and so on, as shown in the figure, but in actual applications, the device using the methods of the present application can include only a part of the components of the electronic device 400, for example, can include only the processor 401 and the network interface 406. Figure 3 Optional nature of the components in the figure is shown by dashed lines.
[0107] The embodiments of the methods of the present application can be implemented in software, firmware, hardware, or any combination thereof.
[0108] The program code can be applied to input instructions to perform the functions described herein and generate output information. The output information can be applied to one or more output devices in a known manner. For the purposes of this application, a processing system includes any system that has a processor, such as for example a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC), or a microprocessor.
[0109] The program code can be implemented in a high level procedural or object oriented programming language to communicate with a processing system. The program code can also be implemented in assembly or machine language, if desired. In fact, the mechanisms described herein are not limited in scope to any particular programming language. In any case, the language can be a compiled or interpreted language.
[0110] One or more aspects of at least one embodiment can be implemented by representative instructions stored on a machine-readable storage medium which represents various logic within the processor, which when read by a machine causes the machine to fabricate logic to perform the techniques described herein. Such representations, known as "IP cores" can be stored on a tangible, machine-readable medium and supplied to various customers or manufacturing facilities to load into the fabrication machines that make the logic or processor.
[0111] In some cases, an instruction translator can be used to translate instructions from a source instruction set to a target instruction set. For example, an instruction translator can transform (e.g., using static binary translation, dynamic binary translation including dynamic compilation), morph, emulate, or otherwise translate instructions to one or more other instructions to be processed by a core. An instruction translator can be implemented in software, hardware, firmware, or combinations thereof. An instruction translator can be on a processor, off a processor, or partially on and partially off a processor.
[0112] In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the application can be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been described in detail in order not to obscure the understanding of this description.
[0113] Similarly, it is to be understood that the embodiments of the present application can be used in the exact form disclosed herein, carried in parts, used, or carried out but not used, in variations of one or more embodiments suggested herein, and / or in variations of one or more embodiments of the prior art disclosed in this description. It is also to be understood that such embodiments of the present application as can have anywhere been discussed are presented for purposes of example and that the present application is limited to but is not limited to these, but is defined by the claims.
[0114] Those skilled in the art will appreciate that the modules in the apparatuses in the embodiments can be adapted and placed in one or more apparatuses other than the embodiments. The modules or units or components in the embodiments can be combined into one module or unit or component, and further can be divided into multiple sub-modules or sub-units or sub-components. Except that at least some of such features and / or processes or units are mutually exclusive, all combinations of all features disclosed in this specification (including accompanying claims, abstract and drawings) and all processes or units of any methods or apparatuses disclosed in this specification are enabled. Unless explicitly stated, each feature disclosed in this specification (including accompanying claims, abstract and drawings) can be replaced by alternative features that serve the same, equivalent or similar purpose.
[0115] Further, those skilled in the art will appreciate that a combination of features of different embodiments can mean within the scope of the present application and form a different embodiment. For example, in the claims, any one of the claimed embodiments can be used in any combination.< / process> < / product> < / process>
Claims
1. A method of providing samples for machine learning, characterized by, The purpose of the machine learning is to establish a machine learning model for solving production organization decision-making problems, including the following steps: establishing a simulation model according to the characteristics of the production organization decision-making problem, which includes a simulation module of equipment or rules required to solve the production organization decision-making problem; setting simulation parameters for the simulation model; repeatedly running the simulation model to generate an initial data set of specified data required to solve the production organization decision-making problem; evaluating the data in the initial data set, selecting preferred data based on the set evaluation conditions to generate a preferred data set; numerical conversion of the preferred data set to feature data and label data suitable for machine learning, thereby generating a machine learning sample set.
2. The method of claim 1, wherein, The evaluation of the data in the initial data set includes establishing a benefit evaluation model and a screening criterion, and the benefit evaluation model evaluates the data in the initial data set, and the preferred data is selected based on the screening criterion.
3. The method of claim 1, wherein, The simulation model includes: a product process flow simulation module for simulating the processing path of products in the production line and carrying the information flow of products among multiple simulation modules; a device workflow simulation module for simulating the operation process of the device; a device space layout simulation module for simulating the spatial layout of the device; a scheduling rule simulation module for simulating the production organization process, organically combining products, devices, etc. to achieve the final production purpose, which contains rules for solving the production organization decision-making problem.
4. The method of claim 3, wherein, The scheduling rules include device calling rules and product processing priorities.
5. The method of claim 4, wherein, The scheduling rules of the production organization decision-making are set in a combination of multiple modes.
6. The method of claim 3, wherein, Based on the operation and maintenance information of the processing equipment, the operation and maintenance information of the transfer equipment, product information, and team information, simulation models under multiple production scenarios are established.
7. The method of claim 1, wherein, The specified data required to solve the production organization decision-making problem includes environment data, decision data, and index information related to the production organization decision-making task, the environment data is the instantaneous information that affects the production organization decision-making, which is composed of device information and product information, the decision data corresponds to the production organization decision-making problem, and the index information is statistical information for evaluating the pros and cons of the decision.
8. The method of claim 7, wherein, When the production organization decision-making task in the simulation model is triggered, the following processes are performed: obtain the current instantaneous environment data and output; output the current decision data according to the scheduling rules; output the index information with statistical significance within a given time range.
9. A computer readable medium characterized by The computer readable medium stores instructions that, when executed on a computer, cause the computer to perform the method of any one of claims 1-8.
10. An electronic device, comprising: including: a memory for storing instructions executed by one or more processors of the electronic device, and a processor, which is one of the processors of the electronic device, for executing the method of any one of claims 1-8.