Process card optimization method and device for intelligent cloud simulation, equipment and storage medium

Abstract

The invention relates to the field of data processing, and discloses a process card optimization method and device for intelligent cloud simulation, equipment and a storage medium. The method comprises the steps of obtaining equipment operation data in a production workshop; performing data cleaning on the production workshop data in the cloud platform by using a big data processing technology, and removing abnormal data in the production workshop data; establishing a process card multi-level simulation model based on an FEA finite element analysis model, setting a production scene and process parameters according to historical data of a production workshop, and performing simulation calculation on the production scene and the process parameters by using the process card multi-level simulation model; learning the initial production workshop data and the initial process card production scheme based on an improved GAN generative adversarial network; learning and defining states, actions and rewards in the learning process of the GAN generative adversarial network by utilizing an enhancement algorithm Q; and the process card is adaptively adjusted according to the production data, so that the problems of insufficient real-time performance, flexibility and intelligence in the traditional method are solved.

Description

Technical Field

[0001] The present application relates to the field of intelligent manufacturing technology, and in particular to a process card optimization method, device, equipment and storage medium for intelligent cloud simulation. Background Art

[0002] As the manufacturing industry transforms towards intelligence and digitalization, the importance of process cards in the production process continues to increase. Process cards are detailed descriptions of production processes that guide production operations, quality control, and resource allocation. However, traditional process card optimization methods often rely on static data analysis and manual adjustments, and cannot respond to changes in the production site in real time. In addition, process card optimization is mostly based on preset rules and historical data, lacks self-learning and adaptive capabilities, and cannot automatically optimize the process according to real-time data in production. In this complex and changing production demand, traditional process card optimization methods often make it difficult to meet flexibility requirements, resulting in reduced production efficiency and resource utilization. Therefore, how to improve process cards to cope with rapidly changing production needs and complex process environments is a technical problem that needs to be solved at this stage. Summary of the invention

[0003] The present application provides a process card optimization method for intelligent cloud simulation, which can store asset registration in a database and audit asset change request information through a multiple audit mechanism to ensure the security of assets.

[0004] In a first aspect, an embodiment of the present application provides a process card optimization method for intelligent cloud simulation, and the process card optimization method comprises the following steps:

[0005] Acquire equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0006] Using big data processing technology in the cloud platform to clean the production workshop data, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0007] Establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain an initial process card production plan;

[0008] Learning the initial production workshop data and the initial process card production plan based on an improved GAN generative adversarial network;

[0009] The reinforcement algorithm Q learning is used to define the state, action and reward in the learning process of the GAN generative adversarial network to obtain the target process card production plan;

[0010] The target process card production plan is multi-threaded managed and data shared through the Hadoop parallel computing framework, and the target process card production plan is pushed to production equipment and operators for execution through the production control system.

[0011] In a possible implementation, the acquiring of equipment operation data, environmental data, and production progress data in a production workshop to obtain production workshop data, and transmitting the production workshop data to a cloud platform via a wireless network includes:

[0012] Temperature sensors, pressure sensors, and vibration sensors are deployed in the production workshop. The motor speed and temperature change parameters of equipment components are obtained through the sensors every 5 seconds. The collected data are sorted according to the equipment number and collection time information to obtain the equipment operation data;

[0013] The environment data at different locations in the workshop are collected through temperature and humidity sensors, light sensors, and air quality sensors. The collected data are marked with spatial locations to obtain environmental data.

[0014] The production progress data is obtained by reading the task completion mark of each production process node, the product flow counting device, and the time node information of the material preparation and product quality inspection links entered manually;

[0015] Aligning and merging the equipment operation data, environmental data, and production progress data according to a unified timestamp to obtain production workshop data;

[0016] The production workshop data is encrypted using the AES encryption algorithm, and after encryption, the production workshop data is transmitted to the cloud platform via a wireless network.

[0017] In a possible implementation, the data cleaning of the production workshop data is performed using big data processing technology in the cloud platform to remove abnormal data in the production workshop data to obtain initial production workshop data, including:

[0018] Acquire production workshop data, and use linear interpolation to supplement missing values ​​of the production workshop data to obtain first production workshop data;

[0019] The K-means clustering algorithm is used to cluster the first production workshop data.

[0020]

[0021] Where E is the total clustering error sum of squares, k is the number of clusters, and Ci represents the i-th cluster, x is cluster C i The data points in i The centroid of the cluster;

[0022] The data points with similar characteristics in the production workshop data are divided into the same category, and the centroid positions are continuously updated iteratively, and the data points are reallocated to the cluster where the nearest centroid is located until convergence to the minimum value, so as to obtain normal data clustering, and the data points far away from the centroids of each cluster are removed to obtain the second production workshop data;

[0023] The equipment operation data, environmental data and production progress data in the second production workshop data are stored in separate tables to obtain initial production workshop data.

[0024] In a possible implementation, the multi-level simulation model of the process card is established based on the FEA finite element analysis model, the production scene and process parameters are set according to the historical data of the production workshop, and the production scene and process parameters are simulated and calculated using the multi-level simulation model of the process card to obtain an initial process card production plan, including:

[0025] According to the product characteristics and process flow of the workshop production, a multi-level simulation model of the process card is established based on the FEA finite element analysis principle in the ANSYS simulation software;

[0026] Obtain historical production data of the production workshop, analyze the production scenario details and process parameters of different order products in the historical production data, and extract the production scenario mode;

[0027] According to the production scenario mode, the corresponding initial values ​​of the process parameters are set, and according to the initial values ​​of the process parameters, the multi-level simulation model of the process card is started to perform simulation calculations;

[0028] Numerical calculations are performed on the material deformation index, processing energy consumption index and production cycle index in the initial values ​​of the process parameters according to the mechanical equations and thermodynamic equations in the multi-level simulation model of the process card;

[0029] Combining production efficiency, product quality and cost control, the process parameter combination and production process arrangement are screened out to obtain the initial process card production plan.

[0030] In a possible implementation, the learning of the initial production workshop data and the initial process card production plan based on the improved GAN generative adversarial network includes:

[0031] A fully connected neural network structure is used to establish a generator of a GAN generative adversarial network, which includes at least an input layer, an intermediate layer and an output layer; multiple hidden layers are set in the intermediate layer, the Re lu function is used as the activation function, a nonlinear transformation is introduced, and the number of neurons in the output layer is matched with the data dimension of the process card production plan;

[0032] The discriminator of the GAN generative adversarial network is established using a fully connected neural network structure. The input layer is used to receive data with the same dimension as the generator output. The middle layer is set as a hidden layer, the LeakyRe lu function is set as the activation function, and the output layer uses the Sigmoid function to compress the output to between (0,1);

[0033] The gradient penalty is introduced to improve the established GAN generative adversarial network, and the gradient penalty term is added to the discriminator loss function to obtain the WGAN-GP generative adversarial network;

[0034] A batch of real workshop data and process card production plans are randomly selected from the initial workshop data and the initial process card production plan, and the output and loss of the discriminator for the real data are calculated. A batch of noise vectors are collected from the prior noise distribution, and the generated data is obtained through the generator, and the output and loss of the discriminator for the generated data are calculated.

[0035] The total loss of the discriminator is calculated based on the gradient penalty, and the discriminator weights are updated based on the gradient back propagation, keeping the discriminator weights unchanged, and the generator loss is calculated again by sampling from the noise distribution, and the generator weights are updated by back propagation;

[0036] Constantly adjust the learning rate, the ratio of discriminator and generator training times, and the noise distribution parameters, and optimize the hyperparameter settings by monitoring the loss curve during training and the quality of generated samples.

[0037] In a possible implementation, the reinforcement algorithm Q learning is used to define the state, action and reward in the GAN generation adversarial network learning process to obtain the target process card production plan, including:

[0038] Create a Q table, where rows correspond to different states in the state space S, and columns correspond to various actions in the action space A. The initial value of the Q table is randomly set in the range of (-0.1-0.1);

[0039] In each round of learning iteration, the agent is in the current state s t ; Select the greedy strategy according to the Q table and randomly select an action a with probability э t Explore and select the work with the largest Q value in the current state with probability 1-э, that is:

[0040]

[0041] Among them, the initial value of э is 0.5;

[0042] Execute action a t , if the process parameters are adjusted in the GAN generative adversarial network, the parameters of the generator generating the process card are modified according to the set step size, and if the generator structure adjustment is involved, the generator structure is changed;

[0043] Get new status by simulating production or applying the newly generated process card in a small range in the actual production environment t+1 and instant rewards t+1 ;

[0044] Update the Q table according to the Q learning update formula,

[0045]

[0046] Among them, α is the learning rate, which determines the speed at which new information covers the old Q value, and its value range is between (0.01-0.1); γ is the discount factor, which is used to measure the importance of future rewards, and its value range is between (0-1);

[0047] Through continuous cycles, the Q table gradually converges, and the optimal actions are taken for different production states, guiding the GAN to generate adversarial networks to generate results close to the target process card production plan, and obtain the target process card production plan.

[0048] In a possible implementation, the target process card production plan is managed and data shared by multiple threads through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system, including:

[0049] Determine the number and specifications of computing nodes and storage nodes of the Hadoop cluster based on production scale and data processing volume;

[0050] Organize various data in the target process card production plan, the target process card production plan at least including the numerical sequence of cutting speed and feed rate of mechanical processing, the time stamp record of temperature and pressure setting value of chemical process, product design specification document, and equipment adaptation information, into structured data;

[0051] Analyze the data processing logic of the target process card production plan, determine the MapReduce task structure, and the YARN resource manager allocates the Map and Reduce subtasks of the MapReduce task to multiple threads of different computing nodes for execution according to the idle status of cluster node resources and task priority;

[0052] Through the built-in communication module of the production control system, the process card plan data is pushed to the production equipment controller and operator terminal using the Kafka message queue mechanism.

[0053] In a second aspect, the present application provides a process card optimization device for intelligent cloud simulation, wherein the process card optimization device comprises the following modules:

[0054] A data acquisition module is used to acquire equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0055] A data processing module, used to clean the production workshop data by using big data processing technology in the cloud platform, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0056] A process card acquisition module is used to establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain an initial process card production plan;

[0057] A process card optimization module, used for learning the initial production workshop data and the initial process card production plan based on an improved GAN generative adversarial network;

[0058] A process card obtaining module is used to define the state, action and reward in the learning process of the GAN generation adversarial network by using the reinforcement algorithm Q learning to obtain the target process card production plan;

[0059] The process card execution module is used to perform multi-thread management and data sharing on the target process card production plan through the Hadoop parallel computing framework, and push the target process card production plan to the production equipment and operators for execution through the production control system.

[0060] In the third aspect, the present application provides a process card optimization device for intelligent cloud simulation, comprising: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory so that the process card optimization device for intelligent cloud simulation executes each step of the above-mentioned process card optimization method for intelligent cloud simulation.

[0061] In a fourth aspect, the present application provides a computer-readable storage medium, which stores instructions, and when the computer-readable storage medium is run on a computer, it enables the computer to execute the various steps of the above-mentioned process card optimization method of intelligent cloud simulation.

[0062] Based on the method provided in the present application, the production workshop data is obtained by acquiring the equipment operation data, environmental data and production progress data in the production workshop, and the production workshop data is transmitted to the cloud platform through a wireless network; the production workshop data is cleaned by using big data processing technology in the cloud platform to remove abnormal data in the production workshop data to obtain the initial production workshop data; a multi-level simulation model of the process card is established based on the FEA finite element analysis model, the production scene and process parameters are set according to the historical data of the production workshop, and the production scene and process parameters are simulated and calculated by using the multi-level simulation model of the process card to obtain the initial process card production plan; the initial production workshop data and the initial process card production plan are learned based on the improved GAN generative adversarial network; the reinforcement algorithm Q learning is used to define the state, action and reward in the learning process of the GAN generative adversarial network to obtain the target process card production plan; the target process card production plan is multi-threaded managed and data shared through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system. It can not only optimize the production process in real time, but also adaptively adjust the process card according to the production data, solving the problems of insufficient real-time, flexibility and intelligence in traditional methods, improving production efficiency, reducing costs and ensuring product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of an embodiment of the first intelligent cloud simulation process card optimization method provided in the embodiments of the present application;

[0064] Figure 2 This is a flow chart of an embodiment of a second intelligent cloud simulation process card optimization method provided in an embodiment of the present application;

[0065] Figure 3 This is a flow chart of an embodiment of a third intelligent cloud simulation process card optimization method provided in an embodiment of the present application;

[0066] Figure 4 A schematic diagram of the structure of a first intelligent cloud simulation process card optimization device provided in an embodiment of the present application;

[0067] Figure 5 A structural schematic diagram of a process card optimization device for intelligent cloud simulation provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] The embodiment of the present application provides a process card optimization method for intelligent cloud simulation, which can store asset registration in a database and audit asset change request information through a multiple audit mechanism, thereby solving the problem of poor security of assets in daily management.

[0069] The technical scheme in the embodiment of the present application will be described below in conjunction with the accompanying drawings in the embodiment of the present application. Wherein, the terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable in appropriate circumstances, so that the embodiments described here can be implemented in a sequence other than the content illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0070] It is understandable that any part of the data acquisition or collection involved in this application has been authorized by the user.

[0071] It is understandable that the executor of the present application may be a process card optimization device for intelligent cloud simulation, or a mobile terminal or a server, which is not specifically limited here.

[0072] The specific process of the embodiment of the present application is described below. Figure 1 , Figure 1 The flowchart of an embodiment of a process card optimization method for intelligent cloud simulation provided in an embodiment of the present application includes:

[0073] 101. Obtain equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0074] 102. Use big data processing technology in the cloud platform to clean the production workshop data, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0075] It is understandable that temperature sensors, pressure sensors and vibration sensors are deployed in the production workshop, and the speed of the motor and the temperature change parameters of the equipment components are obtained through the sensors every 5 seconds. The collected data are sorted according to the equipment number and collection time information to obtain the equipment operation data;

[0076] The environment data at different locations in the workshop are collected through temperature and humidity sensors, light sensors, and air quality sensors. The collected data are marked with spatial locations to obtain environmental data.

[0077] The production progress data is obtained by reading the task completion mark of each production process node, the product flow counting device, and the time node information of the material preparation and product quality inspection links entered manually;

[0078] Align and merge equipment operation data, environmental data, and production progress data based on a unified timestamp to obtain production workshop data;

[0079] The AES encryption algorithm is used to encrypt the production workshop data, and after encryption, the production workshop data is transmitted to the cloud platform via the wireless network.

[0080] 103. Establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain the initial process card production plan;

[0081] It can be understood that the production workshop data is obtained, and the missing values ​​of the production workshop data are supplemented by linear interpolation to obtain the first production workshop data;

[0082] The K-means clustering algorithm is used to cluster the data of the first production workshop.

[0083]

[0084] Where E is the total clustering error sum of squares, k is the number of clusters, and C i represents the i-th cluster, x is cluster C i The data points in i The centroid of the cluster;

[0085] The data points with similar characteristics in the production workshop data are divided into the same category. The centroid position is updated through continuous iteration, and the data points are reallocated to the cluster with the nearest centroid until it converges to the minimum value to obtain normal data clustering. The data points far away from the centroid of each cluster are removed to obtain the second production workshop data.

[0086] The equipment operation data, environmental data and production progress data in the second production workshop data are stored in separate tables to obtain the initial production workshop data.

[0087] 104. Based on the improved GAN generative adversarial network, the initial production workshop data and the initial process card production plan are learned;

[0088] It can be understood that according to the characteristics of the products and process flow produced in the workshop, a multi-level simulation model of the process card is established based on the FEA finite element analysis principle in the ANSYS simulation software;

[0089] Obtain historical production data from the production workshop, analyze the production scenario details and process parameters of different order products in the historical production data, and extract production scenario patterns;

[0090] Set the corresponding initial values ​​of process parameters according to the production scenario mode, and start the multi-level simulation model of the process card to perform simulation calculations according to the initial values ​​of the process parameters;

[0091] According to the mechanical equations and thermodynamic equations in the multi-level simulation model of the process card, the material deformation index, processing energy consumption index and production cycle index in the initial value of the process parameters are numerically calculated;

[0092] Combining production efficiency, product quality and cost control, the process parameter combination and production process arrangement are screened out to obtain the initial process card production plan.

[0093] 105. Use the reinforcement algorithm Q learning to define the state, action and reward in the GAN generative adversarial network learning process to obtain the target process card production plan;

[0094] It can be understood that the generator of the GAN generative adversarial network is established using a fully connected neural network structure, which includes at least an input layer, an intermediate layer, and an output layer; multiple hidden layers are set in the intermediate layer, the Re lu function is used as the activation function, a nonlinear transformation is introduced, and the number of neurons in the output layer is matched with the data dimension of the process card production plan;

[0095] The discriminator of the GAN generative adversarial network is established using a fully connected neural network structure. The input layer is used to receive data with the same dimension as the generator output. The middle layer is set as a hidden layer, the LeakyRe lu function is set as the activation function, and the output layer uses the Sigmoid function to compress the output to between (0,1);

[0096] The gradient penalty is introduced to improve the established GAN generative adversarial network, and the gradient penalty term is added to the discriminator loss function to obtain the WGAN-GP generative adversarial network;

[0097] A batch of real workshop data and process card production plans are randomly selected from the initial workshop data and the initial process card production plan, and the output and loss of the discriminator for the real data are calculated. A batch of noise vectors are collected from the prior noise distribution, and the generated data is obtained through the generator, and the output and loss of the discriminator for the generated data are calculated.

[0098] The total loss of the discriminator is calculated based on the gradient penalty, and the discriminator weights are updated based on the gradient back propagation, keeping the discriminator weights unchanged, and the generator loss is calculated again by sampling from the noise distribution, and the generator weights are updated by back propagation;

[0099] Constantly adjust the learning rate, the ratio of discriminator and generator training times, and the noise distribution parameters, and optimize the hyperparameter settings by monitoring the loss curve during training and the quality of generated samples.

[0100] 106. The target process card production plan is managed and data shared through multi-threading through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system.

[0101] It can be understood that a Q-table is established, with rows corresponding to different states in the state space S and columns corresponding to various actions in the action space A. The initial value of the Q-table is randomly set in the range of (-0.1-0.1);

[0102] In each round of learning iteration, the agent is in the current state s t ; Select the greedy strategy according to the Q table and randomly select an action a with probability э t Explore and select the work with the largest Q value in the current state with probability 1-э, that is:

[0103]

[0104] Among them, the initial value of э is 0.5;

[0105] Execute action a t , if the process parameters are adjusted in the GAN generative adversarial network, the parameters of the generator generating the process card are modified according to the set step size. If the generator structure adjustment is involved, the generator structure is changed;

[0106] Get new status by simulating production or applying the newly generated process card in a small range in the actual production environment t+1 and instant rewards t+1 ;

[0107] Update the Q table according to the Q learning update formula,

[0108]

[0109] Among them, α is the learning rate, which determines the speed at which new information covers the old Q value, and its value range is between (0.01-0.1); γ is the discount factor, which is used to measure the importance of future rewards, and its value range is between (0-1);

[0110] Through continuous cycles, the Q table gradually converges, and the optimal actions are taken for different production states, guiding the GAN to generate adversarial networks to generate results close to the target process card production plan, and obtain the target process card production plan.

[0111] Based on the method provided by this application, the production workshop data is obtained by obtaining the equipment operation data, environmental data and production progress data in the production workshop, and the production workshop data is transmitted to the cloud platform through the wireless network; the data of the production workshop is cleaned by using big data processing technology in the cloud platform, and the abnormal data in the production workshop data is removed to obtain the initial production workshop data; a multi-level simulation model of the process card is established based on the FEA finite element analysis model, and the production scene and process parameters are set according to the historical data of the production workshop, and the production scene and process parameters are simulated and calculated using the multi-level simulation model of the process card to obtain the initial process card production plan; the initial production workshop data and the initial process card production plan are learned based on the improved GAN generation adversarial network; the state, action and reward in the learning process of the GAN generation adversarial network are defined by the reinforcement algorithm Q learning, and the target process card production plan is obtained; the target process card production plan is multi-threaded managed and data shared through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system. It can not only optimize the production process in real time, but also adaptively adjust the process card according to the production data, solve the problems of real-time, flexibility and intelligence in the traditional method, improve production efficiency, reduce costs and ensure product quality.

[0112] See also Figure 2 , Figure 2 The flowchart of the embodiment of the third intelligent cloud simulation process card optimization method provided in the embodiment of the present application includes:

[0113] 201. Obtain equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0114] 202. Use big data processing technology in the cloud platform to clean the production workshop data, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0115] Among them, steps 201-202 are similar to the execution steps of the above steps 101-102, and the details are not repeated here.

[0116] 203. According to the product characteristics and process flow of the workshop, a multi-level simulation model of the process card is established based on the FEA finite element analysis principle in the ANSYS simulation software;

[0117] It is understandable that the basic construction of FEA finite element analysis model: according to the characteristics of the products produced in the workshop and the process flow, a multi-level simulation model of the process card is constructed based on the FEA finite element analysis principle in professional simulation software (such as ANSYS). For example, for a machining workshop, the workpiece material properties (elastic modulus, Poisson's ratio, etc.), tool cutting parameters (cutting speed, feed rate, cutting depth) and mechanical properties of the processing equipment are included in the model, and key parts such as the contact area between the workpiece and the tool are gridded to improve the simulation accuracy.

[0118] 204. Obtain historical production data of the production workshop, analyze the production scene details and process parameters of different order products in the historical production data, and extract the production scene mode;

[0119] It is understandable that when running the simulation model, various possible production scenarios and process parameter combinations are fully considered. For example, a variety of simulation scenarios are set for different product models, raw material batches, production environment conditions and other factors. In each scenario, the process parameters such as processing speed, cutting depth, heating temperature, etc. are further adjusted to perform large-scale multiple simulation calculations. Through a comprehensive evaluation of the simulation results, including comprehensive consideration of multiple indicators such as production efficiency, product quality, equipment loss, energy consumption, etc., the optimal initial process card solution is screened out.

[0120] 205. Set the corresponding initial values ​​of the process parameters according to the production scenario mode, and start the multi-level simulation model of the process card to perform simulation calculation according to the initial values ​​of the process parameters;

[0121] It is understandable that historical data drives the scenario and parameter settings: deeply explore the historical data accumulated in the production workshop, analyze the production scenario details of different order products in the past, such as the impact of batch size and raw material batch differences on production, and the process parameter combination for stable operation of each process. Extract typical production scenario patterns from them, and set the production scenario of this simulation in combination with the current order demand, such as small batch trial production scenarios of new products, large-scale production scenarios in peak seasons, etc.; at the same time, accurately set the corresponding initial values ​​of process parameters to provide basic conditions for simulation calculations.

[0122] 206. According to the mechanical equations and thermodynamic equations in the multi-level simulation model of the process card, numerical calculations are performed on the material deformation index, processing energy consumption index and production cycle index in the initial values ​​of the process parameters;

[0123] It is understandable that simulation calculation and initial plan generation: start the multi-level simulation model of the process card for simulation calculation. The model performs numerical calculations on key indicators such as material deformation, processing energy consumption, and production rhythm based on built-in physical equations such as mechanics and thermodynamics and production logic rules. After multiple iterative simulations (such as more than 100 times), comprehensive consideration of multi-dimensional goals such as production efficiency, product quality, and cost control, select the optimal process parameter combination and production process arrangement, and form an initial process card production plan, which covers detailed operation steps of each process, equipment scheduling plan, quality control nodes, and other contents.

[0124] 207. Combined with production efficiency, product quality and cost control, the process parameter combination and production process arrangement are screened out to obtain the initial process card production plan.

[0125] 208. Based on the improved GAN generative adversarial network, the initial production workshop data and the initial process card production plan are learned;

[0126] 209. Use the reinforcement algorithm Q learning to define the state, action and reward in the GAN generation adversarial network learning process to obtain the target process card production plan;

[0127] 210. The target process card production plan is managed and data shared through multi-threading through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system.

[0128] Among them, steps 207-210 are similar to the execution steps of the above steps 104-106, and will not be repeated here.

[0129] Based on the method provided in the embodiment of the present application, a multi-level simulation model of the process card is established based on the FEA finite element analysis model, the production scenarios and process parameters are set according to the historical data of the production workshop, and the production scenarios and process parameters are simulated and calculated using the multi-level simulation model of the process card. The FEA model can accurately predict the deformation of the equipment under different cutting forces, thereby ensuring that the equipment always maintains a stable operating state in the actual production process, avoiding processing errors or equipment failures caused by structural deformation, and providing a scientific basis for the optimization of equipment parameters in the process card.

[0130] See also Figure 3 , Figure 3 The flowchart of the embodiment of the third intelligent cloud simulation process card optimization method provided in the embodiment of the present application includes:

[0131] 301. Acquire equipment operation data, environmental data, and production progress data in the production workshop to obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0132] 302. Use big data processing technology in the cloud platform to clean the production workshop data, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0133] 303. Establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain the initial process card production plan;

[0134] Among them, steps 301-303 are similar to the execution steps of the above steps 101-103, and the details are not repeated here.

[0135] 304. A generator of a GAN generative adversarial network is established using a fully connected neural network structure, which includes at least an input layer, an intermediate layer, and an output layer; multiple hidden layers are set in the intermediate layer, the Re lu function is used as the activation function, a nonlinear transformation is introduced, and the number of neurons in the output layer is matched with the data dimension of the process card production plan;

[0136] It is understandable that the improved GAN generative adversarial network architecture is constructed: Wasserstein GAN (WGAN) is used as the basic framework, and the gradient penalty term is introduced for improvement to enhance the stability and convergence of network training. The network consists of a generator and a discriminator. The generator is responsible for generating new process card scheme variants based on the input initial production workshop data, and the discriminator determines whether the input process card scheme comes from the real initial scheme or the fake scheme generated by the generator.

[0137] 305. Use a fully connected neural network structure to establish a discriminator of the GAN generative adversarial network. The input layer is used to receive data with the same dimension as the generator output. The middle layer is set as a hidden layer. The LeakyRe lu function is set as the activation function. The output layer uses the Sigmoid function to compress the output to between (0,1).

[0138] 306. The gradient penalty is introduced to improve the established GAN generative adversarial network, and the gradient penalty term is added to the discriminator loss function to obtain the WGAN-GP generative adversarial network;

[0139] It is understandable that data input and learning training: the initial production workshop data and the initial process card production plan are input into the improved GAN network in pairs. The generator adjusts the process parameters based on the real-time status of the workshop equipment, dynamic changes in the environment and other information, and the discriminator continuously improves the ability to distinguish true and false plans. The two are gradually optimized in the adversarial game. The training process uses the Adam optimization algorithm, setting the learning rate to 0.0002, β1=0.5, β2=0.999, and after thousands of rounds of iterative training (such as 5000 rounds), the generator can accurately generate a process card plan that fits actual production and is better.

[0140] 307. Randomly extract a batch of real production workshop data and process card production plans from the initial production workshop data and the initial process card production plan, calculate the output and loss of the discriminator for the real data, collect a batch of noise vectors from the prior noise distribution, obtain generated data through the generator, and calculate the output and loss of the discriminator for the generated data;

[0141] 308. The total loss of the discriminator is calculated based on the gradient penalty, and the discriminator weight is updated based on the gradient back propagation, keeping the discriminator weight unchanged, and the generator loss is calculated again by sampling from the noise distribution, and the generator weight is updated by back propagation;

[0142] 309. Continuously adjust the learning rate, the ratio of discriminator and generator training times, and the noise distribution parameters, and optimize the hyperparameter settings by monitoring the loss curve during training and the quality of generated samples;

[0143] 310. Use the reinforcement algorithm Q learning to define the state, action and reward in the GAN generation adversarial network learning process to obtain the target process card production plan;

[0144] 311. The target process card production plan is managed and data shared through multi-threading through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system.

[0145] Among them, steps 310-311 are similar to the execution steps of the above steps 105-106, and will not be repeated here.

[0146] Based on the method provided in the embodiment of the present application, the initial production workshop data and the initial process card production plan are learned based on the improved GAN generative adversarial network. The generative artificial intelligence module can immediately automatically adjust the equipment parameter settings in the process card based on the learned knowledge and real-time data, such as reducing the equipment operating speed or adjusting the cooling system parameters, to ensure that the equipment operates in a safe and stable state while ensuring that product quality is not affected.

[0147] The above describes the process card optimization method of intelligent cloud simulation in the embodiment of the present application. The following describes the process card optimization device of intelligent cloud simulation in the embodiment of the present application. Figure 4 , Figure 4 A schematic diagram of the structure of a process card optimization device for intelligent cloud simulation provided in an embodiment of the present application includes:

[0148] The data acquisition module 401 is used to acquire the equipment operation data, environmental data and production progress data in the production workshop, obtain the production workshop data, and transmit the production workshop data to the cloud platform via a wireless network;

[0149] The data processing module 402 is used to clean the production workshop data using big data processing technology in the cloud platform, remove abnormal data in the production workshop data, and obtain initial production workshop data;

[0150] The process card acquisition module 403 is used to establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain an initial process card production plan;

[0151] A process card optimization module 404 is used to learn the initial production workshop data and the initial process card production plan based on the improved GAN generative adversarial network;

[0152] The process card obtaining module 405 is used to define the state, action and reward in the GAN generation adversarial network learning process by using the reinforcement algorithm Q learning to obtain the target process card production plan;

[0153] The process card execution module 406 is used to perform multi-thread management and data sharing on the target process card production plan through the Hadoop parallel computing framework, and push the target process card production plan to the production equipment and operators for execution through the production control system.

[0154] Based on the device provided by the embodiment of the present application, the equipment operation data, environmental data and production progress data in the production workshop are received, and the cloud platform receives the production workshop data through a wireless network; the big data processing technology is used in the cloud platform to clean the production workshop data, remove the abnormal data in the production workshop data, establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters, and learn the initial production workshop data and the initial process card production plan based on the improved GAN generation adversarial network; use the reinforcement algorithm Q learning to define the state, action and reward in the learning process of the GAN generation adversarial network to obtain the target process card production plan; the target process card production plan is multi-threaded managed and data shared through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system. Not only can the production process be optimized in real time, but also the process card can be adaptively adjusted according to the production data, solving the problems of insufficient real-time, flexibility and intelligence in traditional methods, improving production efficiency, reducing costs and ensuring product quality.

[0155] See also Figure 5 , Figure 5 It is a structural schematic diagram of a process card optimization device for intelligent cloud simulation provided by an embodiment of the present application. The process card optimization device 500 for intelligent cloud simulation may have relatively large differences due to different configurations or performances, and may include one or more processors 510 (for example, one or more processors) and a memory 520, and one or more storage media 530 storing application programs 533 or data 532. Among them, the memory 520 and the storage medium 530 may be short-term storage or persistent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the process card optimization device 500 for intelligent cloud simulation. Furthermore, the processor 510 may be configured to communicate with the storage medium 5630 to execute a series of instruction operations in the storage medium 530 on the process card optimization device 500 for intelligent cloud simulation.

[0156] The process card optimization device 500 of intelligent cloud simulation may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input and output interfaces 560, and / or one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It can be understood by those skilled in the art that Figure 5The structure of the process card optimization equipment for intelligent cloud simulation shown does not constitute a limitation of the process card optimization equipment for intelligent cloud simulation, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0157] The present application also provides a process card optimization device for intelligent cloud simulation, wherein the computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes each step of the process card optimization method for intelligent cloud simulation in the above-mentioned embodiments.

[0158] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes each step of the process card optimization method of intelligent cloud simulation.

[0159] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0160] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0161] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0162] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A process card optimization method for intelligent cloud simulation, characterized in that: The process card optimization method comprises the following steps: Acquire equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network; Using big data processing technology in the cloud platform to clean the production workshop data, remove abnormal data in the production workshop data, and obtain initial production workshop data; Establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain an initial process card production plan; Learning the initial production workshop data and the initial process card production plan based on an improved GAN generative adversarial network; The reinforcement algorithm Q learning is used to define the state, action and reward in the learning process of the GAN generative adversarial network to obtain the target process card production plan; The target process card production plan is multi-threaded managed and data shared through the Hadoop parallel computing framework, and the target process card production plan is pushed to production equipment and operators for execution through the production control system.

2. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The obtaining of equipment operation data, environmental data and production progress data in the production workshop to obtain production workshop data, and transmitting the production workshop data to the cloud platform via a wireless network includes: Temperature sensors, pressure sensors, and vibration sensors are deployed in the production workshop. The motor speed and temperature change parameters of equipment components are obtained through the sensors every 5 seconds. The collected data are sorted according to the equipment number and collection time information to obtain the equipment operation data; The environment data at different locations in the workshop are collected through temperature and humidity sensors, light sensors, and air quality sensors. The collected data are marked with spatial locations to obtain environmental data. The production progress data is obtained by reading the task completion mark of each production process node, the product flow counting device, and the time node information of the material preparation and product quality inspection links entered manually; Aligning and merging the equipment operation data, environmental data, and production progress data according to a unified timestamp to obtain production workshop data; The production workshop data is encrypted using the AES encryption algorithm, and after encryption, the production workshop data is transmitted to the cloud platform via a wireless network.

3. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The data cleaning of the production workshop data by using big data processing technology in the cloud platform to remove abnormal data in the production workshop data to obtain initial production workshop data includes: Acquire production workshop data, and use linear interpolation to supplement missing values ​​of the production workshop data to obtain first production workshop data; The K-means clustering algorithm is used to cluster the first production workshop data. Where E is the total clustering error sum of squares, k is the number of clusters, and C i represents the i-th cluster, x is cluster C i The data points in i The centroid of the cluster; The data points with similar characteristics in the production workshop data are divided into the same category, and the centroid positions are continuously updated iteratively, and the data points are reallocated to the cluster where the nearest centroid is located until convergence to the minimum value, so as to obtain normal data clustering, and the data points far away from the centroids of each cluster are removed to obtain the second production workshop data; The equipment operation data, environmental data and production progress data in the second production workshop data are stored in separate tables to obtain initial production workshop data.

4. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The multi-level simulation model of the process card is established based on the FEA finite element analysis model, the production scene and process parameters are set according to the historical data of the production workshop, and the production scene and process parameters are simulated and calculated using the multi-level simulation model of the process card to obtain the initial process card production plan, including: According to the product characteristics and process flow of the workshop production, a multi-level simulation model of the process card is established based on the FEA finite element analysis principle in the ANSYS simulation software; Obtain historical production data of the production workshop, analyze the production scenario details and process parameters of different order products in the historical production data, and extract the production scenario mode; According to the production scenario mode, the corresponding initial values ​​of the process parameters are set, and according to the initial values ​​of the process parameters, the multi-level simulation model of the process card is started to perform simulation calculations; Numerical calculations are performed on the material deformation index, processing energy consumption index and production cycle index in the initial values ​​of the process parameters according to the mechanical equations and thermodynamic equations in the multi-level simulation model of the process card; Combining production efficiency, product quality and cost control, the process parameter combination and production process arrangement are screened out to obtain the initial process card production plan.

5. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The improved GAN generative adversarial network is used to learn the initial production workshop data and the initial process card production plan, including: A fully connected neural network structure is used to establish a generator of a GAN generative adversarial network, which includes at least an input layer, an intermediate layer and an output layer; multiple hidden layers are set in the intermediate layer, the Relu function is used as the activation function, a nonlinear transformation is introduced, and the number of neurons in the output layer is matched with the dimension of the process card production plan data; The discriminator of the GAN generative adversarial network is established using a fully connected neural network structure. The input layer is used to receive data with the same dimension as the generator output. The middle layer is set as a hidden layer, the LeakyRelu function is set as the activation function, and the output layer uses the Sigmoid function to compress the output to between (0,1); The gradient penalty is introduced to improve the established GAN generative adversarial network, and the gradient penalty term is added to the discriminator loss function to obtain the WGAN-GP generative adversarial network; A batch of real workshop data and process card production plans are randomly selected from the initial workshop data and the initial process card production plan, and the output and loss of the discriminator for the real data are calculated. A batch of noise vectors are collected from the prior noise distribution, and the generated data is obtained through the generator, and the output and loss of the discriminator for the generated data are calculated. The total loss of the discriminator is calculated based on the gradient penalty, and the discriminator weights are updated based on the gradient back propagation, keeping the discriminator weights unchanged, and the generator loss is calculated again by sampling from the noise distribution, and the generator weights are updated by back propagation; Constantly adjust the learning rate, the ratio of discriminator and generator training times, and the noise distribution parameters, and optimize the hyperparameter settings by monitoring the loss curve during training and the quality of generated samples.

6. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The reinforcement algorithm Q learning is used to define the state, action and reward of the GAN generative adversarial network learning process, and obtain the target process card production plan, including: Create a Q table, where rows correspond to different states in the state space S, and columns correspond to various actions in the action space A. The initial value of the Q table is randomly set in the range of (-0.1-0.1); In each round of learning iteration, the agent is in the current state s t ; Select the greedy strategy according to the Q table and randomly select an action a with probability э t Explore and select the work with the largest Q value in the current state with probability 1-э, that is: Among them, the initial value of э is 0.5; Execute action a t , if the process parameters are adjusted in the GAN generative adversarial network, the parameters of the generator generating the process card are modified according to the set step size, and if the generator structure adjustment is involved, the generator structure is changed; Get new status by simulating production or applying the newly generated process card in a small range in the actual production environment t+1 and instant rewards t+1 ; Update the Q table according to the Q learning update formula, Among them, α is the learning rate, which determines the speed at which new information covers the old Q value, and its value range is between (0.01-0.1); γ is the discount factor, which is used to measure the importance of future rewards, and its value range is between (0-1); Through continuous cycles, the Q table gradually converges, and the optimal actions are taken for different production states, guiding the GAN to generate adversarial networks to generate results close to the target process card production plan, and obtain the target process card production plan.

7. The process card optimization method of intelligent cloud simulation as claimed in claim 1, characterized in that: The target process card production plan is managed and data shared by multiple threads through the Hadoop parallel computing framework, and the target process card production plan is pushed to the production equipment and operators for execution through the production control system, including: Determine the number and specifications of computing nodes and storage nodes of the Hadoop cluster based on production scale and data processing volume; Organize various data in the target process card production plan, the target process card production plan at least including the numerical sequence of cutting speed and feed rate of mechanical processing, the time stamp record of temperature and pressure setting value of chemical process, product design specification document, and equipment adaptation information, into structured data; Analyze the data processing logic of the target process card production plan, determine the MapReduce task structure, and the YARN resource manager allocates the Map and Reduce subtasks of the MapReduce task to multiple threads of different computing nodes for execution according to the idle status of cluster node resources and task priority; Through the built-in communication module of the production control system, the process card plan data is pushed to the production equipment controller and operator terminal using the Kafka message queue mechanism.

8. A process card optimization device for intelligent cloud simulation, characterized in that: The process card optimization device includes the following modules: A data acquisition module is used to acquire equipment operation data, environmental data and production progress data in the production workshop, obtain production workshop data, and transmit the production workshop data to the cloud platform via a wireless network; A data processing module, used to clean the production workshop data by using big data processing technology in the cloud platform, remove abnormal data in the production workshop data, and obtain initial production workshop data; A process card acquisition module is used to establish a multi-level simulation model of the process card based on the FEA finite element analysis model, set the production scene and process parameters according to the historical data of the production workshop, and use the multi-level simulation model of the process card to simulate and calculate the production scene and process parameters to obtain an initial process card production plan; A process card optimization module, used for learning the initial production workshop data and the initial process card production plan based on an improved GAN generative adversarial network; A process card obtaining module is used to define the state, action and reward in the learning process of the GAN generation adversarial network by using the reinforcement algorithm Q learning to obtain the target process card production plan; The process card execution module is used to perform multi-thread management and data sharing on the target process card production plan through the Hadoop parallel computing framework, and push the target process card production plan to the production equipment and operators for execution through the production control system.

9. An intelligent cloud simulation process card optimization device, characterized in that: The process card optimization device of intelligent cloud simulation includes: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the process card optimization device of the intelligent cloud simulation performs each step of the process card optimization method of the intelligent cloud simulation as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the various steps of the process card optimization method for intelligent cloud simulation as described in any one of claims 1-7 are implemented.

Images