Flow control method and device for intelligent manufacturing

Abstract

The invention discloses an intelligent manufacturing process control device and method. The device comprises a production end and a cloud end, the production end is used for controlling production resources according to the production flow information; wirelessly communicating with a production resource; acquiring production situation information; the cloud is used for acquiring production demand information of a user and sample data associated with the production demand information; and processing the production demand information and the production situation information by using a production distribution model, generating production flow information meeting the production demand information, and sending the production flow information to a production end. According to the method, an enterprise can be helped to operate in a data-driven decision mode, a data chain with data flowing automatically can be formed, efficient collaboration of all links of production and manufacturing is promoted, and the production efficiency can be generally improved; the production resources and the production management control system are decoupled, the production resources are deployed locally, and the production management control system is deployed at the cloud, so that flexible deployment and management of the production resources can be realized.

Description

Technical Field

[0001] The present invention relates to the fields of industrial artificial intelligence and industrial Internet, and in particular to a process control device and method for intelligent manufacturing. Background Art

[0002] At present, in the field of industrial manufacturing, its manufacturing process still mainly follows the traditional order-distribution-production model, which is difficult to dynamically upgrade and adjust according to market demand or technology upgrades. Due to the high efficiency of assembly line production in meeting large-scale and batch production, it is still widely used in the manufacturing industry. As the practice of artificial intelligence in industrial production in recent years has proved, regardless of whether the trend of increasing personalized variables in industrial production will continue, or whether companies will re-emphasize the return to hardware standardization due to continued business considerations and growing pressure costs, flexible production and processing methods will likely show greater advantages.

[0003] With the rapid development of artificial intelligence, many artificial intelligence systems are mainly used in links such as commodity marketing or market forecasting. Using artificial intelligence systems to analyze the behavior of buyers or search engine users and obtain the highest possible advertising revenue from them is very different from quickly obtaining correct fault diagnosis results based on abnormal temperature data obtained from machines. The former case is to find patterns in a large amount of data, while in the latter case, when training artificial intelligence, it is necessary to know the connection between the temperature data of a specific machine produced by the manufacturer and the process of using the machine by the relevant enterprise. Unlike the former, which identifies patterns in the data, the latter must be able to combine data with industrial processes. Therefore, in different production scenarios, it is necessary to find matching process control methods.

[0004] Since manufacturing systems are generally complex systems, it is often difficult to describe them using any single modeling method. For example, how to complement and coordinate multiple types of models to achieve accurate description and control of the production process is a problem that needs to be solved in the current field of intelligent manufacturing process control. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a process control device and method for intelligent manufacturing, which is conducive to improving the adaptability of industrial production processes, thereby improving the intelligence and universality of industrial manufacturing systems. In order to solve the above technical problems, the embodiment of the present invention discloses a process control device and method for intelligent manufacturing.

[0006] In a first aspect of an embodiment of the present invention, a process control device for intelligent manufacturing is disclosed, comprising: a production end and a cloud; the production end is used to control production resources according to production process information; wirelessly communicate with production resources; and collect production situation information; the production situation information includes raw material consumption information, production resource location information, production resource utilization information, completed product information, and to-be-finished product information;

[0007] The cloud end is used to obtain the production demand information of the user; use the production status information to identify and process the pre-stored production resource allocation model to obtain a production allocation model; use the production allocation model to process the production demand information and production status information to generate production process information that meets the production demand information; and send the production process information to the production end.

[0008] The production end includes a radio frequency unit and a signal processing unit; the signal processing unit is connected to the radio frequency unit;

[0009] The signal processing unit includes a digital-to-object mapping module and a state monitoring control module;

[0010] The digital-to-object mapping module is used to encode and model the production resources to obtain a digital model of the production resources; the digital model includes function information, attribute information and status information;

[0011] The state monitoring control module is used to collect the state of production resources to obtain production situation information; according to the production process information, the state of production resources is controlled to change so that the production resources complete the production instructions in the production process information.

[0012] The radio frequency unit includes a positioning module and a communication module; the communication module is used to establish a communication link with the signal processing unit and the cloud, send the production status information collected by the status monitoring and control module to the cloud, and send the production process information issued by the cloud to the status monitoring and control module; the positioning module is used to sense the location information of the production resources.

[0013] The state monitoring control module of the signal processing unit includes an edge computing server, which has image recognition function, real-time encoding and decoding function, network platform function and platform management function.

[0014] The cloud comprises: a process generation unit, a process update unit and a demand acquisition unit; the process generation unit is connected to the demand acquisition unit and the process update unit; the demand acquisition unit is connected to the process update unit;

[0015] The demand acquisition unit is used to acquire the production demand information of the user and / or the sample data associated with the production demand information;

[0016] The process generation unit is used to use the production situation information and the production process information to identify and process the pre-stored production resource allocation model to obtain a production allocation model, use the production allocation model to process the production demand information and the production situation information, generate production process information that meets the production demand information, and send the production process information to the production end;

[0017] The process updating unit is used to store and update the production resource allocation model.

[0018] The method of using the production situation information and the production process information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes:

[0019] Obtain the production status information and corresponding production process information of the completed production process;

[0020] Using each pre-stored production resource allocation model, the production situation information of the completed production process is processed to obtain corresponding predicted production process information;

[0021] Calculate the difference between the predicted production process information obtained by each production resource allocation model and the production process information corresponding to the production situation information of the completed production process to obtain the corresponding difference value;

[0022] The production resource allocation model with the smallest difference value is determined as the production allocation model.

[0023] The pre-stored production resource allocation model includes a linear model and a neural network model;

[0024] The linear model is expressed as:

[0025] x(t+1)=Ax(t)+Bu(t)

[0026] y(t)=Cx(t),

[0027] Among them, x(t) is the production situation information obtained at time t, x(t+1) is the production situation information obtained at time t+1, u(t) is the production demand information obtained at time t, y(t) is the production process information predicted at time t, A, B, and C are the known system matrix, control matrix, and prediction matrix respectively;

[0028] The neural network model adopts one of the back propagation neural network model, convolutional neural network, recurrent neural network, long short-term memory neural network, self-attention neural network or process control neural network.

[0029] The process generation unit and the process update unit are implemented using an industrial cloud platform; the process generation unit includes a cloud server; the cloud server of the process generation unit and the edge computing server of the signal processing unit together constitute a cloud-edge collaborative computing network.

[0030] A second aspect of an embodiment of the present invention discloses a process control method for intelligent manufacturing, comprising:

[0031] S1, collect production situation information, production demand information and historical production process information;

[0032] S2, using the production situation information and historical production process information, identifying and processing the pre-stored production resource allocation model to obtain a production allocation model;

[0033] S3, using the production allocation model to process the production situation information and production demand information to generate production process information that meets the production demand information.

[0034] The method of using the production situation information and the historical production process information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes:

[0035] S21, obtaining production status information and corresponding production process information of a completed production process; the corresponding production process information is historical production process information;

[0036] S22, using each pre-stored production resource allocation model, processing the production situation information of the completed production process to obtain corresponding predicted production process information;

[0037] S23, performing difference calculation between the predicted production process information obtained by each production resource allocation model and the production process information corresponding to the production situation information of the completed production process to obtain a corresponding difference value;

[0038] S24, determining the production resource allocation model with the smallest difference value as the production allocation model.

[0039] The beneficial effects of the present invention are:

[0040] In terms of production operations, the method of the present invention can help enterprises operate in a "data-driven decision-making" mode, form a data chain with automatic data flow, promote efficient collaboration in all aspects of production and manufacturing, greatly reduce the complexity and uncertainty of the manufacturing system, and generally improve production efficiency.

[0041] The method of the present invention decouples production resources and production management and control systems, deploys production resources locally, and deploys production management and control systems in the cloud, which can realize flexible deployment and management of production resources, and uses the cloud to build a production demand matching and distribution platform similar to "Didi Taxi", so as to achieve accurate matching of various production demands and production resources, and truly realize flexible production. Taking clothing companies as an example, China currently has more than 400,000 clothing companies of all sizes, of which 90% are small and micro enterprises. These companies have weak sales capabilities and survive by OEM. They have almost no digital capabilities and use Excel tables to manage production, which is almost in chaos to respond to the production needs of small orders, urgent orders, and fast orders, with out-of-control costs and meager profits. The present invention can realize technical transformation of flexible production for these companies at a lower cost, so that their production equipment can be put on the cloud without the need to purchase IT equipment, which greatly saves production costs.

[0042] The present invention adopts satellite communication technology at the production end, so that all production equipment can achieve "one-click access to the cloud" through satellite Internet, solving the monitoring and control problems of remote equipment; at the same time, the sensing function is integrated at the radio frequency end to realize real-time monitoring of the location of production resources.

[0043] In order to achieve the integration of traditional models and neural network models, when controlling production resources according to production needs and equipment status, the present invention uses historical successful production data to identify various models, finds the model that is most suitable for the current scenario, and uses the model to achieve the allocation of production resources. For scenarios suitable for traditional state transition models, the present application can use state transition models to achieve the allocation of production resources. For complex systems with multiple production factors, the present application can use neural network models to allocate production resources.

[0044] The process control neural network established by the present invention processes the input production situation information in parallel through eight feature processing modules, quickly realizes the accurate extraction of production variables of multiple types and multiple industry modes, suppresses the interference of incoherent quantities, effectively extracts feature variables, and finally uses linear modules and activation modules to fuse the eight types of feature variables to obtain the final production process information; through the above operations, the accuracy and real-time performance of the production process information are ensured. In the process of training the process control neural network, the present invention establishes a training update model: The introduction of multiple preset convergence variables enables rapid convergence of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is the composition principle diagram of the device of the present invention;

[0046] Figure 2 It is a flow chart for implementing the method of the present invention. DETAILED DESCRIPTION

[0047] In order to better understand the content of the present invention, an embodiment is given here.

[0048] Figure 1 It is the composition principle diagram of the device of the present invention, Figure 2 It is a flow chart for implementing the method of the present invention.

[0049] In order to organically combine the traditional manufacturing control process and artificial intelligence methods to quickly adapt to the production process, realize operation in a "data-driven decision-making" mode, and improve production efficiency, this embodiment discloses a process control device and method for intelligent manufacturing.

[0050] The embodiment of the present invention discloses an intelligent manufacturing process control device, including: a production end and a cloud; the production end is used to control production resources according to production process information; wirelessly communicate with production resources; collect production situation information; the production situation information includes raw material consumption information, production resource location information, production resource utilization information, completed product information, to-be-finished product information, etc.;

[0051] The cloud end is used to obtain the production demand information of the user and the sample data associated with the production demand information; use the production situation information to identify and process the pre-stored production resource allocation model to obtain the production allocation model; use the production allocation model to process the production demand information and the production situation information to generate production process information that meets the production demand information, and send the production process information to the production end; the production process information includes production instructions for each production resource;

[0052] The production end includes a radio frequency unit and a signal processing unit; the signal processing unit is connected to the radio frequency unit;

[0053] The signal processing unit includes a digital-to-object mapping module and a state monitoring control module;

[0054] The digital-to-object mapping module is used to encode and model the production resources to obtain a digital model of the production resources; the digital model includes function information, attribute information and status information;

[0055] The state monitoring control module is used to collect the state of production resources to obtain production situation information; according to the production process information, the state of production resources is controlled to change so that the production resources complete the production instructions in the production process information.

[0056] The modeling process may be directed to the production process, using a Markov model to model the production process to obtain a digital model of production resources, or using 3D-MAX simulation software to establish a digital model of production resources;

[0057] The digital-physical mapping module includes a processor and manufacturing software running on the processor; the manufacturing software includes CAD or CAE or CAM software matching a specific industrial field of production. For example, for the wind power field, GHT-MON or DIgSILENT or WTD can be used.

[0058] The radio frequency unit includes a positioning module and a communication module; the communication module is used to establish a communication link with the signal processing unit and the cloud, send the production status information collected by the status monitoring and control module to the cloud, and send the production process information sent by the cloud to the status monitoring and control module; the positioning module is used to sense the location information of the production resources and send the location information of the production resources to the status monitoring and control module.

[0059] The communication module is a low-latency communication module and / or a satellite communication module;

[0060] The state monitoring and control module of the signal processing unit includes an edge computing server, which has image recognition function, real-time encoding and decoding function, network platform function and platform management function; each achievable function of the signal processing unit corresponds to one of its submodules; the edge computing server can be implemented by a micro server that supports a deep learning reasoning framework; the deep learning reasoning framework includes TensorFlow, PyTorch, HiAI, MNN, etc.

[0061] The network platform functions include location service function, business offloading capability, wireless network information service capability, access user identity recognition service capability, bandwidth service management capability, QoS service capability, and IoT service capability;

[0062] The IoT service capabilities include supporting various types of production resource terminals to access through various protocols, device management, data analysis, ToD services, rule management, and collaborative management; the protocols include TCP / UDP protocol, MQTT protocol, CoAP protocol, etc.; the communication protocol types of production resource terminals include Lora / ZigBee / NB-IoT, etc. The ToD service includes receiving ToD information of production resource terminals; the ToD information includes time notification, timestamp, etc.;

[0063] The signal processing unit may perform TDOA positioning on the production resource terminal by utilizing ToD information of the same production resource terminal received by multiple processing terminals included therein.

[0064] The cloud comprises: a process generation unit, a process update unit and a demand acquisition unit; the process generation unit is connected to the demand acquisition unit and the process update unit; the demand acquisition unit is connected to the process update unit;

[0065] The demand acquisition unit is used to acquire the production demand information of the user and the sample data associated with the production demand information;

[0066] The process generation unit is used to use the production situation information to identify and process the pre-stored production resource allocation model to obtain a production allocation model, use the production allocation model to process the production demand information and the production situation information, generate production process information that meets the production demand information, and send the production process information to the production end;

[0067] The step of processing the production demand information and the production situation information using the production allocation model to generate production process information that meets the production demand information includes:

[0068] Using the production demand information to update the information of the unfinished products in the production situation information, obtaining updated production situation information, and inputting the updated production situation information into the production allocation model;

[0069] The updating of the unfinished product information in the production situation information by using the production demand information is to add the production demand information to the unfinished product information.

[0070] The process updating unit is used to store and update the production resource allocation model;

[0071] The process of updating the production resource allocation model of the process update unit includes: when the production resource allocation model is a neural network model, using the production situation information and production process information collected recently, adopting an update algorithm to update the neural network model; the update algorithm is an update algorithm in the neural network, including BP algorithm, gradient descent method, momentum method, adaptive learning rate algorithm, etc.

[0072] When the production resource allocation model is not a neural network model, the production resource allocation model is remodeled at regular intervals using the most recently collected production situation information and production process information.

[0073] The method of using the production situation information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes:

[0074] Obtain the production status information and corresponding production process information of the completed production process;

[0075] Using each pre-stored production resource allocation model, the production situation information of the completed production process is processed to obtain corresponding predicted production process information;

[0076] Calculate the difference between the predicted production process information obtained by each production resource allocation model and the production process information corresponding to the production situation information of the completed production process to obtain the corresponding difference value;

[0077] Determine the production resource allocation model with the smallest difference value as the production allocation model;

[0078] The method of using the production situation information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes:

[0079] The pre-stored production resource allocation model is a TS fuzzy model;

[0080] Obtain the production status information and corresponding production process information of the completed production process;

[0081] According to the production process, fuzzy rules are constructed;

[0082] Using the production situation information and the corresponding production process information, the fuzzy rules are structurally identified to obtain the structural information of the production resource allocation model; using the production situation information and the corresponding production process information, the fuzzy rules are parameter identified to obtain the parameter information of the production resource allocation model;

[0083] Substituting parameter information and structural information into the TS fuzzy model, the production allocation model is obtained.

[0084] The pre-stored production resource allocation model may also be an Agent intelligent entity model, which is constructed using a MAS modeling method;

[0085] The pre-stored production resource allocation model includes a linear model and a neural network model;

[0086] The linear model is expressed as:

[0087] x(t+1)=Ax(t)+Bu(t),

[0088] y(t)=Cx(t)

[0089] Among them, x(t) is the production situation information obtained at time t, x(t+1) is the production situation information obtained at time t+1, u(t) is the production demand information obtained at time t, y(t) is the production process information predicted at time t, A, B, and C are the known system matrix, control matrix, and prediction matrix respectively;

[0090] The neural network model adopts one of the back propagation neural network model, convolutional neural network, recurrent neural network, long short-term memory neural network, self-attention neural network and process control neural network.

[0091] The expression for the difference calculation is:

[0092]

[0093] Among them, pb is the calculated difference value, N is the number of control variables contained in the production process information, and p i is the i-th control variable in the production process information corresponding to the production status information of the completed production process, is the i-th control variable of the predicted production process information;

[0094] The process control neural network includes a first feature processing module, a second feature processing module, a third feature processing module, a fourth feature processing module, a fifth feature processing module, a sixth feature processing module, a seventh feature processing module, an eighth feature processing module, a linear module and an activation module;

[0095] The input end of the first feature processing module is configured as the model input of the process control neural network, the output end of the first feature processing module is connected to the input end of the second feature processing module; the output end of the second feature processing module is connected to the input end of the third feature processing module; the output end of the third feature processing module is connected to the input end of the fourth feature processing module; the output end of the fourth feature processing module is connected to the input end of the fifth feature processing module; the output end of the fifth feature processing module is connected to the input end of the sixth feature processing module; the output end of the sixth feature processing module is connected to the input end of the seventh feature processing module; the input end of the seventh feature processing module is connected to the input end of the eighth feature processing module; the output end of the eighth feature processing module is connected to the input end of the linear module; the output end of the linear module is connected to the input end of the activation module; the output end of the activation module is configured as the model output of the process control neural network.

[0096] The first feature processing module includes a first multi-head attention unit, a second multi-head attention unit, a third multi-head attention unit, a first fusion unit, a second fusion unit, a third fusion unit, a fourth fusion unit, a fifth fusion unit, a first normalization unit, a second normalization unit, a third normalization unit, a fourth normalization unit, a fifth normalization unit, a first network unit, and a second network unit; wherein,

[0097] The input end of the first multi-head attention unit and the input end of the first fusion unit are both configured as the input end of the first feature processing module, the output end of the first multi-head attention unit is connected to the input end of the first fusion unit; the output end of the first fusion unit is connected to the input end of the first normalization unit; the output end of the first normalization unit is respectively connected to the input end of the first network unit and the input end of the second fusion unit; the output end of the first network unit is connected to the input end of the second fusion unit; the output end of the second fusion unit is connected to the input end of the second normalization unit; the output end of the second normalization unit is respectively connected to the input end of the second multi-head attention unit, the input end of the third multi-head attention unit and the input end of the third fusion unit; The output end is connected to the input end of the third fusion unit; the output end of the third fusion unit is connected to the input end of the third normalization unit; the output end of the third normalization unit is respectively connected to the input end of the third multi-head attention unit and the input end of the fourth fusion unit; the output end of the third multi-head attention unit is connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is connected to the input end of the fourth normalization unit; the output end of the fourth normalization unit is respectively connected to the input end of the second network unit and the input end of the fifth fusion unit; the output end of the second network unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the fifth normalization unit; the output end of the fifth normalization unit is configured with the output end of the first feature processing module.

[0098] It should be noted that the model structures of the above-mentioned first feature processing module, second feature processing module, third feature processing module, fourth feature processing module, fifth feature processing module, sixth feature processing module, seventh feature processing module, and eighth feature processing module are consistent, and the embodiments of the present invention do not limit this. Furthermore, the production situation information can be deeply extracted by processing the above-mentioned first feature processing module, second feature processing module, third feature processing module, fourth feature processing module, fifth feature processing module, sixth feature processing module, seventh feature processing module, and eighth feature processing module in sequence, thereby realizing the deep capture of the logical relationship of the production process and identifying the process variables that play a major role therein, and the embodiments of the present invention do not limit this.

[0099] It should be noted that the above-mentioned linear module is constructed based on the fully connected layer, which is composed of a series of weight matrices and bias vectors, and is used to perform linear transformation on the input features, which is not limited in the embodiment of the present invention. Furthermore, the linear module provides powerful feature processing for the process control neural network through simple mathematical operations to cooperate with subsequent activation operations and increase the nonlinear expression ability of the process control neural network, which is not limited in the embodiment of the present invention.

[0100] It should be noted that the above activation module is constructed based on the softmax activation function, which is not limited in the embodiment of the present invention.

[0101] It should be noted that the above-mentioned first fusion unit, second fusion unit, third fusion unit, fourth fusion unit, and fifth fusion unit are constructed based on the add operation, which is not limited in the embodiment of the present invention.

[0102] It should be noted that the above-mentioned first normalization unit, second normalization unit, third normalization unit, fourth normalization unit, and fifth normalization unit are constructed based on the Layer Normalization layer, and the embodiment of the present invention does not limit this.

[0103] It should be noted that the first network unit and the second network unit are constructed based on a feedforward neural network, which is not limited in the embodiment of the present invention. Further, the size of the first network unit and the second network unit is 4×768, which is not limited in the embodiment of the present invention.

[0104] It should be noted that the above-mentioned first multi-head attention unit, second multi-head attention unit, and third multi-head attention unit are constructed based on a multi-head attention mechanism (Multi-Head Attention), which is not limited in the embodiment of the present invention.

[0105] It should be noted that the output of the second normalization unit mentioned above is input to the third multi-head attention unit respectively as query (Query, Q) (i.e., query vector) and key (Key, K) (i.e., key vector). Furthermore, the query vector represents the currently processed input element, which is used to compare with the key vector to determine the attention score. In multi-head attention, the query is copied to each head, but the query vector of each head may be transformed through different linear layers to capture different information. Furthermore, key vectors are similar to query vectors, but they represent other elements in the input sequence. The dot product of the query vector and the key vector is used to calculate the attention scores, which determine the weight of the value vector in the final output.

[0106] It should be noted that the output of the third normalization unit is input to the third multi-head attention unit as value (Value, V) (i.e., value vector). Furthermore, the value vector contains information about the input elements, which will be weighted according to the attention scores calculated by the query and the key. In multi-head attention, the value vector is also transformed through different linear layers to capture information in different representation subspaces.

[0107] It should be noted that the output data from the second normalization unit must first undergo position encoding processing before being input into the second multi-head attention unit, which is not limited in this embodiment of the present invention.

[0108] For the neural network model in the pre-stored production resource allocation model, before use, it is necessary to use the production situation information and production process information of the completed production process to train the neural network model, specifically including:

[0109] Using the production situation information and production process information of the completed production process, a training data set is constructed; the training data in the training data set is the production situation information, and the label information in the training data set is the production process information corresponding to the production situation information;

[0110] Initialize the number of training iterations;

[0111] Input the training data in the training data set as input data into the neural network model;

[0112] Using the neural network model, the input data is processed to obtain a predicted value;

[0113] Performing difference calculation processing on the obtained predicted value and the label information corresponding to the input data to obtain a difference value;

[0114] Determine whether the difference value satisfies a convergence condition, and obtain a first determination result;

[0115] When the first judgment result is no, judging whether the training iteration number value is equal to the training number threshold, and obtaining a second judgment result;

[0116] When the second judgment result is no, determining that the model training state does not meet the training termination condition;

[0117] When the second judgment result is yes, determining that the model training state satisfies the training termination condition;

[0118] When the first judgment result is yes, determining that the model training state satisfies a training termination condition;

[0119] When the model training state does not meet the training termination condition, the neural network model is updated with the parameter update model, the training iteration number value is increased by 1, and the execution is triggered to input the training data in the training data set as input data into the neural network model;

[0120] When the model training state satisfies the training termination condition, the training process of the neural network model is completed to obtain a trained neural network model.

[0121] The difference value satisfies the convergence condition, which means that the difference value is less than a preset convergence threshold; the difference value does not satisfy the convergence condition, which means that the difference value is not less than the preset convergence threshold.

[0122] The difference calculation process may be implemented using a loss function.

[0123] The loss function may adopt a cross entropy loss function.

[0124] The parameter updating model is:

[0125]

[0126] θ←θ+v;

[0127] In the formula, is the difference value calculated for the i-th training data in the training data set, v is the parameter update value, θ is the parameter of the module to be updated, η is the initial parameter learning rate, α is the momentum angle parameter, 0≤α≤π / 4, ▽ θ It means to find partial derivatives with respect to variable θ;

[0128] The process generation unit and the process update unit are implemented using an industrial cloud platform;

[0129] The process generation unit includes a cloud server; the cloud server of the process generation unit and the edge computing server of the signal processing unit together constitute a cloud-edge collaborative computing network,

[0130] The signal processing unit, the process generation unit, the demand acquisition unit and the process update unit together constitute a data center network. The data network adopts congestion control mechanism and / or link flow control mechanism and / or load balancing technology to perform network flow control to ensure low latency, high bandwidth and losslessness of data transmission in the data network.

[0131] The industrial cloud platform may be a MindSphere platform.

[0132] The communication network between the cloud end and the production end can be realized through a time-clear network (TAN).

[0133] The cloud also includes a data aggregation module; the data aggregation module is used to collect data generated during the production process, classify and extract the data, and provide a data set for the model training process of the process generation unit and the process update unit;

[0134] The digital model of production resources in the digital-physical mapping module can be implemented using industrial APPs; industrial APPs encapsulate industrial technologies such as processes, logic, data and data flows, experience, algorithms, knowledge, etc. for solving specific problems. Each industrial APP is a collection and carrier of some specific industrial technologies / knowledge, which is applied in a given scenario. For example, if bolts / nuts are required to be suitable for corrosion-resistant and high-temperature-resistant scenarios, a special design APP for the bolt / nut design is required.

[0135] Industrial APP determines the various functions of industrial software required in the process, tooling, manufacturing and trial stages based on the decoupled levels and specific physical product units of complex products, and decomposes and develops them, breaking down the original large-scale industrial software into software with a unified interface, complete functions and strong expansibility.

[0136] The cloud server of the process generation unit and the edge computing server of the signal processing unit together constitute a cloud-edge collaborative computing network. The cloud-edge collaborative computing network is constructed using the KubeEdge architecture. The edge computing server of the signal processing unit is the edge side of the cloud-edge collaborative computing network, and the cloud server of the process generation unit is the cloud side of the cloud-edge collaborative computing network. Based on the KubeEdge architecture, the edge side includes the Edged module, MetaManager module, DeviceTwin module, EdgeHub module, ServiceBus module and EventBus module. The cloud side includes the CloudHub module and the Controller module.

[0137] The demand acquisition unit includes a transaction platform module, which uses blockchain technology to build a transaction platform between computing resource demanders and computing resource providers.

[0138] The signal processing unit deploys deep learning inference frameworks, including MACE, HiAI, NCNN, MNN, TensorFlow Lite, Paddle Lite, etc.

[0139] The reasoning implementation of the signal processing unit includes online reasoning and offline reasoning. The online reasoning means that the signal processing unit sends the collected data to the process generation unit and the process update unit for reasoning to obtain model update data, and the signal processing unit uses the model update data to update its deep learning reasoning framework; the offline reasoning means that the signal processing unit uses the collected data to update its deep learning reasoning framework.

[0140] The signal processing unit is used to complete edge reasoning;

[0141] The process generation unit is used to complete cloud-based reasoning;

[0142] The process generation unit includes a digital twin module;

[0143] The process update unit is used to complete cloud-based training and cloud-based reasoning;

[0144] The process update unit includes a digital twin model of production resources and generation environment, and a large model;

[0145] The pre-stored production resource allocation model is a self-built processing model, and the parameter and structure adjustment of the model are achieved through an automated machine learning platform (AutoML platform). The AutoML platform includes CloudAutoML, Baidu EasyDL, Alibaba Cloud PAI, etc.

[0146] The software systems of the process generation unit, the demand acquisition unit and the process update unit are designed using an evolutionary architecture;

[0147] The production resources include production robots, automated guided vehicles (AGVs), collaborative robots (Cobots) and drones.

[0148] The production end further includes a storage unit; the storage unit is connected to the process generation unit and the signal processing unit;

[0149] It should be noted that since the method of the embodiment of the present application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time is actually time information. It can be understood that if size, quantity, position, etc. are mentioned in subsequent embodiments, they are all corresponding data for processing by the computer device. The details will not be repeated here.

[0150] It should be noted that the artificial intelligence related technologies that may be involved in this application are briefly described. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0151] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0152] The image processing processes involved in the embodiments of the present invention can all be implemented using computer vision technology. Computer vision technology (CV) is a technology that studies how to make machines "see". To put it more concretely, it refers to machine vision such as using cameras and computers to replace human eyes to identify and measure targets, and further perform graphic processing to make computer processing into images that are more suitable for human eye observation or transmission to instrument detection. As a scientific discipline, computer vision studies related theories and technologies, and attempts to establish an artificial intelligence system that can obtain information from images or multidimensional data. Computer vision technology generally includes image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, three-dimensional object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous positioning and map construction, and other technologies, as well as common biometric recognition technologies such as face recognition and fingerprint recognition.

[0153] Unimodal information is data of only one type, such as text, image, audio, video, electromagnetic signal, etc. Multimodal information is data that includes at least two types of unimodal information. Furthermore, multimodal information is suitable for complex tasks that require the integration of multiple information sources, such as sentiment analysis, robot interaction, autonomous driving, etc. By integrating information from multiple modalities, higher performance and accuracy can usually be achieved on the task.

[0154] A large model refers to an artificial neural network model with a very large number of parameters. In the field of artificial intelligence, a large model generally refers to a model with hundreds of millions to trillions of parameters. Models usually need to be trained on large-scale data sets and require a large amount of computing resources to be optimized and adjusted. Large models are often used to solve complex tasks such as natural language processing, computer vision, and speech recognition. Generative AI is an AI that can create new content and ideas, including conversations, stories, images, videos, and music. In an embodiment of the present application, the large model may be ChatGPT, BERT, XLNet, Zhipu model, Claude, Moonshot AI model, ChatGLM model, Qianyi Tongwen model, MiniMax model, Spark model, Llama model, 360GPT model, Qwen model, Baichuan model, Skylark model, vivoLM model, and Wenxin Yiyan scale language models, which are not limited in the embodiments of the present application.

[0155] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A process control device for intelligent manufacturing, characterized in that: include: Production and cloud; The production end is used to control production resources according to production process information; Wireless communication with production resources; Collect production situation information; The production situation information includes raw material consumption information, production resource location information, production resource utilization information, completed product information, and pending product information; The cloud end is used to obtain the production demand information of the user; use the production status information to identify and process the pre-stored production resource allocation model to obtain a production allocation model; use the production allocation model to process the production demand information and production status information to generate production process information that meets the production demand information; and send the production process information to the production end.

2. The process control device for intelligent manufacturing according to claim 1, characterized in that: The production end includes a radio frequency unit and a signal processing unit; the signal processing unit is connected to the radio frequency unit; The signal processing unit includes a digital-to-object mapping module and a state monitoring control module; The digital-to-object mapping module is used to encode and model the production resources to obtain a digital model of the production resources; the digital model includes function information, attribute information and status information; The state monitoring control module is used to collect the state of production resources to obtain production situation information; according to the production process information, the state of production resources is controlled to change so that the production resources complete the production instructions in the production process information.

3. The process control device for intelligent manufacturing according to claim 2, characterized in that: The radio frequency unit includes a positioning module and a communication module; the communication module is used to establish a communication link with the signal processing unit and the cloud, send the production status information collected by the status monitoring and control module to the cloud, and send the production process information issued by the cloud to the status monitoring and control module; the positioning module is used to sense the location information of the production resources.

4. The process control device for intelligent manufacturing according to claim 2, characterized in that: The state monitoring control module of the signal processing unit includes an edge computing server, and the edge computing server has image recognition function, real-time encoding and decoding function, network platform function and platform management function.

5. The process control device for intelligent manufacturing according to claim 1, characterized in that: The cloud comprises: a process generation unit, a process update unit and a demand acquisition unit; the process generation unit is connected to the demand acquisition unit and the process update unit; the demand acquisition unit is connected to the process update unit; The demand acquisition unit is used to acquire the production demand information of the user and / or the sample data associated with the production demand information; The process generation unit is used to use the production situation information and the production process information to identify and process the pre-stored production resource allocation model to obtain a production allocation model, use the production allocation model to process the production demand information and the production situation information, generate production process information that meets the production demand information, and send the production process information to the production end; The process updating unit is used to store and update the production resource allocation model.

6. The process control device for intelligent manufacturing according to claim 5, characterized in that: The method of using the production situation information and the production process information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes: Obtain the production status information and corresponding production process information of the completed production process; Using each pre-stored production resource allocation model, the production situation information of the completed production process is processed to obtain corresponding predicted production process information; Calculate the difference between the predicted production process information obtained by each production resource allocation model and the production process information corresponding to the production situation information of the completed production process to obtain the corresponding difference value; The production resource allocation model with the smallest difference value is determined as the production allocation model.

7. The process control device for intelligent manufacturing according to claim 6, characterized in that: The pre-stored production resource allocation model includes a linear model and a neural network model; The linear model is expressed as: x(t+1)=Ax(t)+Bu(t) y(t)=Cx(t), Among them, x(t) is the production situation information obtained at time t, x(t+1) is the production situation information obtained at time t+1, u(t) is the production demand information obtained at time t, y(t) is the production process information predicted at time t, A, B, and C are the known system matrix, control matrix, and prediction matrix respectively; The neural network model adopts one of the back propagation neural network model, convolutional neural network, recurrent neural network, long short-term memory neural network, self-attention neural network or process control neural network.

8. The process control device for intelligent manufacturing according to claim 4, characterized in that: The process generation unit and the process update unit are implemented using an industrial cloud platform; the process generation unit includes a cloud server; the cloud server of the process generation unit and the edge computing server of the signal processing unit together constitute a cloud-edge collaborative computing network.

9. A process control method for intelligent manufacturing, characterized in that: include: S1, collect production situation information, production demand information and historical production process information; S2, using the production situation information and historical production process information, identifying and processing the pre-stored production resource allocation model to obtain a production allocation model; S3, using the production allocation model to process the production situation information and production demand information to generate production process information that meets the production demand information.

10. The intelligent manufacturing process control method according to claim 9, characterized in that: The method of using the production situation information and the historical production process information to identify and process the pre-stored production resource allocation model to obtain the production allocation model includes: S21, obtaining the production status information and corresponding production process information of the completed production process; S22, using each pre-stored production resource allocation model, processing the production situation information of the completed production process to obtain corresponding predicted production process information; S23, performing difference calculation between the predicted production process information obtained by each production resource allocation model and the production process information corresponding to the production situation information of the completed production process to obtain a corresponding difference value; S24, determining the production resource allocation model with the smallest difference value as the production allocation model.

Images