Factory production process digital management method and system based on Internet of Things

Abstract

The invention discloses a factory production process digital management method and system based on the Internet of Things, and belongs to the technical field of factory production management, and the method specifically comprises the steps: collecting related data of production equipment in a production line, carrying out the semantic fusion processing of the related data, generating a data stream with an incidence relation, and storing the data stream in a database; establishing a multi-relational graph structure based on the data flow, constructing a multi-dimensional digital model of the production process, receiving production intention information input by a manager, generating a production scheduling scheme according to the multi-dimensional digital model, selecting a scheduling scheme which is most matched with the production intention information, and sending the scheduling scheme to the manager; monitoring production data in the process of executing the scheduling scheme, determining influence factors according to a causal relationship when abnormality or deviation is detected, and updating the multi-dimensional digital model; according to the method, the production scheme can be dynamically adjusted under complex constraint conditions, so that a factory has self-adaptive management capability, and the intelligent level of production and the resource utilization efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of factory production management, and in particular to a factory production process digital management method and system based on the Internet of Things. BACKGROUND

[0002] In recent years, the development of the Internet of Things technology has provided new possibilities for the production process management of factories. By deploying sensors in production equipment, process links, logistics nodes and environmental control units, factories can realize real-time collection of running status, process parameters and environmental variables. However, existing Internet of Things applications mainly stay at the level of single-point monitoring or local optimization, such as device state detection, energy consumption statistics, or process automation scheduling, lacking unified modeling and collaborative management of cross-process, cross-equipment and cross-level data.

[0003] Existing production management systems often use fixed rules or linear optimization methods to schedule production, lacking consideration of complex objectives and adaptability to dynamic changes. In the face of multi-objective constraints (such as delivery time, energy consumption limit, device utilization balance) and uncertain factors (such as device failure, environmental fluctuations, raw material supply delay), traditional methods often can only intervene through artificial experience, resulting in insufficient flexibility and robustness of the system.

[0004] On the other hand, although the concepts of digital twinning and intelligent manufacturing are gradually applied in the industrial field, most current methods only stay at the level of virtual modeling and state mapping, failing to form a data-driven closed-loop evolution mechanism, i.e. how to use real-time data to continuously correct the model, and then optimize the scheduling and management strategy in reverse. However, the existing technology still has the following problems: lack of multi-objective scheduling generation mechanism for management personnel's intention, resulting in oversimplified digital model, making it difficult to make flexible decisions in a dynamic environment. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a factory production process digital management method and system based on the Internet of Things, which realizes a complete link from multi-dimensional data collection, semantic fusion, dynamic modeling, intention-driven scheduling to closed-loop self-evolution optimization.

[0006] To achieve the above purpose, the present application provides the following technical solutions: The factory production process digital management method based on the Internet of Things comprises: Collecting relevant data of production equipment in the production line, and performing semantic fusion processing on the relevant data to generate data streams with associated relationships, the relevant data including running data, energy consumption data, material flow data and environmental data; Establishing a multi-relationship graph structure based on the data streams to construct a multi-dimensional digital model of the production process; Receive production intention information input by management personnel, generate a production scheduling plan based on the multi-dimensional digital model, and select the scheduling plan that best matches the production intention information; During the execution of the scheduling scheme, production data is monitored. When an anomaly or deviation is detected, the influencing factors are determined based on the causal relationship, and the multidimensional digital model is updated.

[0007] Specifically, semantic fusion processing is performed on the relevant data to generate a data stream with related relationships, including: Timestamp alignment is performed on the collected production equipment-related data; Feature extraction is performed on the aligned production equipment-related data, and the features are classified according to the process stage based on the preset process stage labels; Establish semantic relationships across data types among the categorized data, and construct a data lookup table that reflects the logical dependencies between processes; The various types of data in the data lookup table are serialized and combined according to the order of the processes to generate a data stream with related relationships.

[0008] Specifically, based on the data flow, a multi-relationship graph structure is established to construct a multi-dimensional digital model of the production process, including: The data stream is divided into nodes according to equipment units, process links, energy consumption factors and environmental factors to obtain a basic node set; In the basic node set, different types of edges are established according to the time sequence, material transfer relationship, energy consumption dependence relationship and environmental coupling relationship, forming a graph structure with multiple semantics; The nodes and edges in the graph structure are iteratively combined to generate a multi-layered relationship network covering the equipment layer, process layer, and workshop layer. In the multi-layered relationship network, nodes are arranged in an orderly manner according to the logical sequence of the process stages to obtain a multi-dimensional digital model of the production process.

[0009] Specifically, in the multi-layered relationship network, nodes are arranged in an orderly manner according to the logical sequence of process stages to obtain a multi-dimensional digital model of the production process, including: The nodes in the multi-layered relationship network are grouped according to the stage of the process to form a stage set corresponding to the production process; Within each stage set, nodes are sorted according to material delivery sequence and energy consumption constraints to determine the arrangement order within the stage; Between adjacent stage sets, a sequential link is established according to the process connection relationship, connecting the output node of the previous stage to the input node of the next stage in sequence; The node sequence, after being sorted within stages and connected between stages, is combined into a whole structure to obtain a multidimensional digital model for representing the production process.

[0010] Specifically, the process of receiving production intention information input by management personnel, generating a production scheduling plan based on the multi-dimensional digital model, and selecting the scheduling plan that best matches the production intention information includes: Receive production intention information input by management personnel, and parse the production intention information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate; Based on the multidimensional digital model, several candidate scheduling schemes are generated according to the multidimensional target set, and each candidate scheduling scheme corresponds to a different process ordering and resource allocation method. The candidate scheduling schemes are compared, and the matching degree between each scheme and the multidimensional target set is calculated; A priority sequence is established between the candidate scheduling scheme with the highest matching degree and the remaining schemes, and the first scheme in the priority sequence is used as the execution scheduling scheme.

[0011] Specifically, based on the multidimensional digital model, several candidate scheduling schemes are generated according to the multidimensional target set, including: Map the multidimensional target set to the corresponding nodes and relationships in the multidimensional digital model; Based on the mapping results, the execution order of different processes is arranged and combined to form multiple process sorting sequences; For each process sequence, different equipment resources and energy consumption ratios are allocated to obtain the corresponding resource allocation scheme. The process sequence is combined with the corresponding resource allocation scheme to generate several candidate scheduling schemes.

[0012] Specifically, the candidate scheduling schemes are compared, and the matching degree between each scheme and the multidimensional target set is calculated, including: The multidimensional target set is divided into hard targets and flexible targets, and priority and allowable range are set for each target; Based on the multidimensional digital model, a time-series simulation is performed on each candidate scheduling scheme to extract the evaluation dataset corresponding to the target. The evaluation dataset is compared item by item with the allowable range of the corresponding target to generate a set of deviation information for each candidate scheduling scheme; The deviation information set is sorted and merged according to the priority to obtain the matching degree identifier of each candidate scheduling scheme, and recorded as a scheme matching degree lookup table.

[0013] Specifically, establishing a priority sequence between the candidate scheduling scheme with the highest matching degree and the remaining schemes, and using the first scheme in the priority sequence as the execution scheduling scheme, includes: The candidate scheduling schemes are sorted in descending order according to their matching degree to generate an initial sorted list; For candidate scheduling schemes with the same matching degree, the parallel relationship is resolved in turn based on the satisfaction of hard objectives, the number of constraint violations, and the continuity of key processes to obtain a priority draft; Based on the priority draft, each candidate scheduling scheme is adjusted within the boundary around each objective of the multidimensional objective set, the order changes caused by the adjustment are recorded, and a check record is generated. Based on the inspection records, the priority draft is corrected. When the first candidate scheduling scheme has a critical resource unavailability during the planned time period, its priority is downgraded, and the priority of the other candidate scheduling schemes is adjusted relative to the resource substitutability to obtain the priority sequence. The first candidate scheduling scheme in the priority sequence is used as the execution scheduling scheme, and the second candidate scheduling scheme and its triggering conditions are registered as backup schemes.

[0014] The Internet of Things (IoT)-based digital management system for factory production processes is used to implement the IoT-based digital management method for factory production processes, and includes: a data acquisition module, a model building module, a scheduling scheme matching module, and an update and optimization module. The data acquisition module is used to collect relevant data from production equipment in the production line, and to perform semantic fusion processing on the relevant data to generate a data stream with correlation. The relevant data includes operating data, energy consumption data, material flow data and environmental data. The model building module is used to establish a multi-relationship graph structure based on the data flow and construct a multi-dimensional digital model of the production process. The scheduling scheme matching module is used to receive production intention information input by the management personnel, generate a production scheduling scheme according to the multi-dimensional digital model, and select the scheduling scheme that best matches the production intention information. The update and optimization module is used to monitor production data during the execution of the scheduling scheme. When an anomaly or deviation is detected, it determines the influencing factors based on causal relationships and updates the multidimensional digital model.

[0015] Specifically, the scheduling scheme matching module includes: an intent analysis unit, a scheme generation unit, and a matching unit; The intent analysis unit is used to receive production intent information input by the manager and parse the production intent information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate. The scheme generation unit is used to generate several candidate scheduling schemes based on the multidimensional digital model and according to the multidimensional target set. The matching unit is used to compare the candidate scheduling schemes, calculate the matching degree between each scheme and the multidimensional target set, and select the first and best scheduling scheme.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention proposes a digital management method and system for factory production processes based on the Internet of Things (IoT). It collects multi-dimensional data from the operation of IoT sensors throughout the entire production process, constructs a multi-dimensional digital model, and generates candidate scheduling schemes based on the production intentions input by management personnel. The execution scheme is determined through comparison and priority ranking. During operation, anomaly diagnosis and model updates are achieved through causal reasoning, ultimately realizing a complete link from data collection, modeling, scheduling to closed-loop optimization. This method not only breaks through the limitations of data isolation, rigid scheduling, and lagging anomaly handling in traditional factories, achieving unified management across levels and elements, but also dynamically adjusts production schemes under complex constraints, enabling factories to have adaptive management capabilities, thereby significantly improving the level of intelligent production and resource utilization efficiency. Attached Figure Description

[0017] Figure 1 A flowchart of the Internet of Things-based digital management method for factory production processes provided by this invention; Figure 2 This invention provides an architecture diagram for a digital management system for factory production processes based on the Internet of Things. Detailed Implementation

[0018] The present application will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0020] It should be noted that, unless there is conflict, the various features in the embodiments of this application can be combined with each other, all of which are within the protection scope of this application. Furthermore, although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described can be performed in a different order than the module division in the device or the order in the flowchart. In addition, the " The terms "first," "second," and "third" do not limit the data or execution order; they are merely used to distinguish identical or similar items with essentially the same function and purpose.

[0021] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0022] Example 1 Please see Figure 1 The present invention provides an embodiment of a digital management method for factory production processes based on the Internet of Things, comprising the following specific steps: Step S1: Collect relevant data from production equipment in the production line and perform semantic fusion processing on the relevant data to generate a data stream with correlation. The relevant data includes operation data, energy consumption data, material flow data and environmental data.

[0023] The specific steps of step S1 are as follows: Step S101: Timestamp alignment of the collected production equipment-related data.

[0024] In this embodiment, the raw time information acquired by various sensors during data collection is uniformly converted into a standardized time reference, such as using an absolute time marker in the same time zone. Then, the sampling intervals of different data streams are compared, and missing or redundant time points are filled in and merged through interpolation or segmentation. Furthermore, with the actual cycle time of the production process as a reference, the processed multi-source data is reordered and aggregated according to the cycle time window of the process, thereby ensuring that data from different sensors can maintain consistency in the time dimension at the same time and in the same process stage.

[0025] It should be noted that production equipment related data includes: operating status data, such as real-time operating parameters like equipment start / stop status, speed, vibration amplitude, pressure, temperature, current, and voltage; process data, such as process parameters directly reflecting process execution, like machining feed rate, cutting depth, welding current, spraying flow rate, and assembly torque; energy consumption data, such as energy consumption, gas consumption, coolant flow rate, and steam pressure and flow rate; fault and maintenance data, such as equipment alarm records, fault codes, downtime, component life prediction values, and maintenance cycle records; and production cycle data, such as single-piece process completion time, production cycle deviation, output count statistics, and work-in-process dwell time, which are related to production efficiency.

[0026] Step S102: Extract features from the aligned production equipment-related data, and classify the features according to the process stage based on the preset process stage labels.

[0027] In this embodiment, different types of data streams are divided into sliding windows, and statistical characteristics such as mean, variance, peak value, and frequency distribution are calculated within each window to reflect the stability and volatility of equipment operation. Secondly, process features strongly correlated with the execution of processes are extracted based on equipment type and process characteristics, such as the rate of temperature rise and fall, the dominant frequency of vibration signals, and the slope of energy consumption curves, to characterize typical patterns of process stages. Subsequently, the extracted features are compared with pre-defined process link labels, and features belonging to the same process stage are merged into the corresponding classification set. Finally, the classified feature sets are arranged in an orderly manner according to the order of process stages to obtain classification results that accurately reflect the staged state of the production process.

[0028] Step S103: Establish semantic relationships across data types among the categorized data and construct a data lookup table that reflects the logical dependencies between processes.

[0029] In this embodiment, the classified data is attribute-labeled and categorized into process input, process, and process output types to clarify the data's position in the production chain. Subsequently, between adjacent processes, cross-data type correspondences are established based on material transfer relationships, energy consumption patterns, and equipment operation sequences. For example, the output temperature of the previous process is mapped to the input temperature of the next process, and the energy consumption peak of the previous process is linked to the cycle time of the next process. Then, within the same process, data pairings representing causal relationships are established by matching process features with result features. Finally, the above cross-process and cross-type relationships are registered in a structured form to form a queryable data lookup table.

[0030] Step S104: Serialize and combine the multiple types of data in the data lookup table according to the order of the processes to generate a data stream with related relationships.

[0031] In this embodiment, the process nodes in the data lookup table are sorted according to the flow sequence of the production process to ensure that the input, process, and output features are arranged in the order of the processes. Then, within each process, different types of feature data are integrated according to the collection time and semantic tags to form ordered segments in the sequence. Next, connection symbols are introduced between processes to sequentially concatenate the output data of the previous process with the input data of the next process, thereby establishing a continuous chain across processes. Finally, the entire sequence is uniformly encoded so that the multiple types of data are logically represented as a continuous data stream.

[0032] Step S2: Based on the data flow, establish a multi-relationship graph structure to construct a multi-dimensional digital model of the production process.

[0033] The specific steps of step S2 are as follows: Step S201: Divide the data stream into nodes according to equipment units, process links, energy consumption factors and environmental factors to obtain a basic node set.

[0034] In this embodiment, semantic tagging is performed on the fields in the data stream. Features directly related to the operation of production equipment are tagged as equipment unit nodes, data describing the execution sequence and stage status of processes are tagged as process link nodes, data involving the consumption of electricity, gas, steam, or cooling media are tagged as energy consumption element nodes, and data reflecting temperature, humidity, dust concentration, or noise levels are tagged as environmental element nodes. Then, the data within each type of node is uniquely divided to ensure that different nodes can clearly refer to specific objects or features. Next, distinguishing boundaries are established between different categories to avoid confusion of data attributes in subsequent modeling. Finally, the nodes of each category are gathered into a set to obtain a basic node set containing equipment units, process links, energy consumption elements, and environmental elements.

[0035] Step S202: In the set of basic nodes, establish different types of edges according to the time sequence, material transfer relationship, energy consumption dependence relationship and environmental coupling relationship to form a graph structure with multiple semantics.

[0036] In this embodiment, based on the time sequence of process execution, temporal edges are established between preceding and following nodes to represent the continuity of the process flow. Secondly, based on the material transfer path between different processes, material transfer edges are established between nodes, enabling the input and output links to form a traceable link. Furthermore, considering the dependency between energy consumption data and equipment operation, energy consumption dependency edges are established between energy consumption element nodes and corresponding equipment nodes to reveal the coupling relationship between energy consumption and process execution. Finally, considering the impact of environmental conditions on equipment operation and process quality, environmental coupling edges are established between environmental element nodes and related process nodes to reflect the intervention of external factors in the production process. Through the establishment of the above different types of edges, the originally isolated nodes are organized into a multi-layered semantic relationship network, forming a graph structure that can simultaneously reflect temporal logic, material flow, energy consumption, and environmental effects.

[0037] Step S203: Iteratively combine the nodes and edges in the graph structure to generate a multi-layered relationship network covering the equipment layer, process layer, and workshop layer.

[0038] In this embodiment, based on equipment unit nodes, the time sequence edges, energy consumption dependency edges, and environmental coupling edges directly connected to them are aggregated to form an equipment layer sub-network reflecting the operating characteristics of a single piece of equipment. Subsequently, within the scope of the process links, the equipment layer sub-networks belonging to the same process stage are combined, and material transfer edges are introduced to connect the input and output nodes of different processes, thereby forming a process layer network covering the local production chain. Next, multiple process layer networks are iteratively spliced ​​according to the production process sequence, and cross-process energy consumption and environmental correlations are retained to enable the network to exhibit multi-process collaborative characteristics. Finally, all process layer networks are merged into the workshop scope to construct a workshop layer network that can reflect the overall resource allocation, energy consumption balance, and environmental coupling relationship. Through this layer-by-layer combination process, the original graph structure is expanded into a multi-layer relationship network covering the equipment layer, process layer, and workshop layer.

[0039] Step S204: In the multi-layered relationship network, the nodes are arranged in an orderly manner according to the logical sequence of the process stages to obtain a multi-dimensional digital model of the production process.

[0040] The specific steps of step S204 are as follows: Step S2041: Group the nodes in the multi-layer relationship network according to the stage to which the process belongs, so as to form a stage set corresponding to the production process.

[0041] In this embodiment, the semantic tags carried by each node in the multi-layer relationship network are identified, such as equipment unit, process link, energy consumption element and environmental element, and the identification information corresponding to the process stage is extracted; then, according to the process flow document or the predefined process sequence, the nodes are merged according to their respective stages, so that equipment nodes, energy consumption nodes and environmental nodes belonging to the same stage are divided into the same set; furthermore, the original connection relationship between nodes is preserved within the set to ensure that the internal structure of the network is not destroyed after grouping; finally, a stage set is formed by process stage, and each set can be mapped one-to-one to the stage division of the production process.

[0042] Step S2042: Within each stage set, sort the nodes according to the material transfer sequence and energy consumption constraints to determine the arrangement order within the stage.

[0043] In this embodiment, nodes involving material input and output in the identification stage set are identified. A sequential link is established based on the directionality of material flow, placing material input nodes at the starting point and material output nodes at the ending point. Then, energy consumption element nodes in the set are analyzed. If a node is dependent on energy consumption peaks, its corresponding equipment node is pre-sorted to ensure that energy supply can cover the equipment's operational needs. Next, the remaining equipment nodes are arranged sequentially according to their functional roles in the process, forming a sequence that conforms to the process logic. Finally, the original edge relationships between the sorted nodes are preserved, resulting in a stage node sequence with a clear arrangement order.

[0044] Step S2043: Establish sequential links between adjacent stage sets according to the process connection relationship, and connect the output node of the previous stage to the input node of the next stage in sequence.

[0045] In this embodiment, output nodes that serve as process results in the previous stage set are identified, and their corresponding material, energy consumption, or status attributes are extracted. Then, input nodes matching these attributes are located in the next stage set, confirming their logical dependency in production. Next, sequential links are established between output and input nodes according to the process execution order, ensuring that the process results of the previous stage can be smoothly transmitted to the starting point of the next stage. Finally, all sequential links between adjacent stages are added to the overall network structure to ensure logical continuity in material flow, energy supply, and environmental conditions between processes. Through this process, the originally dispersed stage sets are linked into a complete process sequence, generating an ordered network that reflects the entire chain of dependencies in the production process.

[0046] Step S2044: Combine the node sequence after intra-stage sorting and inter-stage connection into an overall structure to obtain a multi-dimensional digital model for representing the production process.

[0047] In this embodiment, the ordered node sequences of each stage set are sequentially spliced ​​together, maintaining the integrity of the material flow link, energy supply link, and environmental constraint link during the splicing process. Then, the node connection relationships across stages are integrated, and the sequential links between preceding and subsequent stages are unified into the overall sequence structure. Next, duplicate nodes or redundant relationships in the global network are deduplicated and merged to ensure the simplicity and consistency of the model. Finally, the integrated node sequence is transformed into a multidimensional overall representation, so that the process logic, resource dependence, and environmental coupling of different dimensions can be uniformly expressed under the same framework. Through this process, a multidimensional digital model covering the equipment layer, process layer, and workshop layer is finally formed.

[0048] Step S3: Receive production intention information input by the management personnel, generate a production scheduling plan based on the multi-dimensional digital model, and select the scheduling plan that best matches the production intention information.

[0049] The specific steps of step S3 are as follows: Step S301: Receive production intention information input by the management personnel, and parse the production intention information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate.

[0050] In this embodiment, the input production intention information is parsed to identify semantic elements related to output, delivery time, energy consumption, and equipment utilization, transforming qualitative descriptions into quantitative indicators. Then, a unified measurement standard is established for the parsed indicators. For example, the output target is refined into output per unit time, the delivery time target into the upper limit of the time for process completion, the energy consumption limit into the allowable energy consumption range for each stage, and the equipment utilization rate into utilization rate or utilization duration indicators. Next, these indicators are normalized to form a comparable target vector in a multi-dimensional space. Finally, the normalized indicators are aggregated according to category to form a multi-dimensional target set including output target, delivery time, energy consumption limit, and equipment utilization rate.

[0051] Step S302: Based on the multidimensional digital model, generate several candidate scheduling schemes according to the multidimensional target set. Each candidate scheduling scheme corresponds to a different process ordering and resource allocation method.

[0052] The specific steps of step S302 are as follows: Step S3021: Map the multidimensional target set to the corresponding nodes and relationships in the multidimensional digital model.

[0053] In this embodiment, various target dimensions in the multidimensional target set are identified, and their numerical ranges or critical conditions are extracted. Then, the nodes and relationships corresponding to the target dimensions are located in the multidimensional digital model. For example, the output target is mapped to the process link node representing the output quantity of the process, the delivery time target is mapped to the edge describing the time sequence, the energy consumption limit is mapped to the energy consumption element node and its connection relationship with the equipment node, and the equipment utilization rate target is mapped to the equipment unit node representing the equipment operating status. Next, corresponding constraint labels or weights are added to the above-mentioned corresponding nodes and relationships so that they can reflect the constraints of the target set. Finally, a set of mapping structures with target constraints is generated inside the model.

[0054] Step S3022: Based on the mapping results, the execution order of different processes is arranged and combined to form multiple process sorting sequences.

[0055] In this embodiment, key process nodes affected by target constraints in the model are identified, and their dependencies with preceding and following nodes are extracted. Then, under the condition that the mandatory dependencies are not broken, process nodes with flexible sorting space are arranged and combined. For example, parallel executable processes are swapped in the same stage, or some processes are advanced or delayed without violating the material flow logic. Next, a process sorting sequence is generated for each arrangement, and constraint labels related to energy consumption, equipment utilization rate, and delivery time are retained in the sequence. Finally, all possible process sorting sequences are gathered into a candidate set.

[0056] Step S3023: For each process sequence, allocate different equipment resources and energy consumption ratios to obtain the corresponding resource allocation scheme.

[0057] In this embodiment, the available equipment unit set for each process in the process sequencing sequence is retrieved one by one, and candidate equipment is selected based on the rated capacity, operating status and maintenance cycle of the equipment. Then, according to the energy consumption limit target, the energy demand under different equipment combinations is determined, and the energy, gas, steam or cooling medium is allocated in proportion to meet the minimum consumption requirements for process execution and not exceed the global energy consumption constraint. Next, the selected equipment is bound with the corresponding energy consumption ratio to form a resource allocation sub-scheme, and multiple differentiated resource allocation methods are generated under the same process sequencing sequence. Finally, the resource allocation sub-schemes of all processes are combined to obtain a set of resource allocation schemes that correspond one-to-one with the process sequencing sequence.

[0058] Step S3024: Combine the process sorting sequence with the corresponding resource allocation scheme to generate several candidate scheduling schemes.

[0059] In this embodiment, for each process sequence, a set of resource allocation schemes associated with it is retrieved, and process nodes, equipment units, and energy consumption ratios are matched item by item. Then, process logic constraints are introduced during the matching process to ensure that equipment resources and energy supply can cover the execution window of the corresponding process on the time axis. Next, the verified process sequence is bound with resource allocation to form an independent candidate scheduling scheme unit, and its attribute indicators related to the target set are recorded during the generation process. Finally, all the candidate scheduling schemes obtained by combination are gathered into a scheme library for selection in the subsequent comparison and priority ranking stages. Through this process, the originally separate process execution logic and resource allocation method are integrated into a candidate scheduling scheme that can be directly implemented, realizing the transition from target mapping to executable plan.

[0060] Step S303: Compare the candidate scheduling schemes and calculate the matching degree between each scheme and the multidimensional target set.

[0061] The specific steps of step S303 are as follows: Step S3031: Divide the multidimensional target set into hard targets and flexible targets, and set priority and allowable range for each target.

[0062] In this embodiment, semantic parsing is performed on each indicator in the target set. Indicators that directly determine whether production can be completed, such as the upper limit of delivery time, minimum output requirements, and availability of key equipment, are classified as hard targets. Indicators that can fluctuate within a certain range without affecting the overall task completion, such as energy consumption balance, resource utilization rate, and environmental load, are classified as flexible targets. Then, strict threshold boundaries are set for hard targets to prevent scheduling schemes from exceeding the limits. Furthermore, allowable ranges are defined for flexible targets to clarify their upper and lower fluctuation ranges so as to reconcile different schemes. Finally, priorities are assigned to hard targets and flexible targets respectively to ensure that hard targets are met first during the comparison process, while flexible targets are comprehensively considered according to weights within the allowable range.

[0063] Step S3032: Based on the multidimensional digital model, perform time-series simulation for each candidate scheduling scheme and extract the evaluation dataset corresponding to the target.

[0064] In this embodiment, the process sequence, resource allocation, and energy consumption ratio in the candidate scheduling scheme are embedded one by one into the corresponding nodes and relationships of the multi-dimensional digital model. Then, the model is driven by the time axis to run, and the execution process of each process is deduced in sequence. During the deduction process, indicators such as output, energy consumption, equipment load rate, and process completion time are calculated in real time. Then, the deduction results are extracted according to the target classification to form data subsets corresponding to output targets, delivery time, energy consumption limits, and equipment utilization. Finally, each subset is merged into a complete evaluation dataset to reflect the performance of the candidate scheduling scheme in different target dimensions.

[0065] Step S3033: Compare the evaluation dataset with the allowable interval of the corresponding target item by item to generate a set of deviation information for each candidate scheduling scheme.

[0066] In this embodiment, the indicators in the evaluation dataset are grouped according to category, and a mapping relationship is established with the corresponding hard and flexible targets in the target set. Then, an interval test is performed on each mapping pair. If the hard target indicator exceeds the set boundary, it is recorded as an absolute deviation. If the flexible target indicator falls outside the allowable range, it is recorded as a relative deviation, and the direction and magnitude of the deviation are marked in the record. Next, the deviation results of all target dimensions are summarized into a deviation information set, and associated according to the process link or resource allocation element to ensure that the deviation can be traced back to the specific execution unit. Finally, the output deviation information set not only retains the gap between each solution and the target.

[0067] Step S3034: Sort and merge the deviation information set according to the priority to obtain the matching degree identifier of each candidate scheduling scheme, and record it as a scheme matching degree comparison table.

[0068] In this embodiment, the deviation information of each scheme is split according to the target category, and a sorting rule is set according to the priority order of hard targets and flexible targets. The deviation of hard targets is examined first, and if there is an over-limit situation, the overall matching degree is directly reduced. Then, the deviation information of flexible targets is merged according to the set weight factor, and the magnitude and direction of exceeding the allowable range are quantified into scores, and weighted and superimposed between different target dimensions. Next, the comprehensive score obtained by each candidate scheme is converted into a matching degree identifier, so that each scheme has a unique matching degree representation. Finally, a correspondence is established between the matching degree identifiers of all candidate schemes and the corresponding scheme numbers to generate a scheme matching degree lookup table.

[0069] Step S304: Establish a priority sequence between the candidate scheduling scheme with the highest matching degree and the other schemes, and take the first scheme in the priority sequence as the execution scheduling scheme.

[0070] The specific steps of step S304 are as follows: Step S3041: Sort each candidate scheduling scheme in descending order according to the matching degree to generate an initial ranking list.

[0071] In this embodiment, all matching degree identifiers in the scheme matching degree comparison table are extracted and a one-to-one correspondence with the corresponding scheme is maintained. Then, the matching degree value is used as the main sorting criterion, and the schemes are compared in descending order. Schemes with higher matching degrees are placed at the front, and schemes with lower matching degrees are moved to the back. When the matching degrees are the same, the schemes are kept in parallel and no further detailed judgment is performed to ensure that the initial ranking list only reflects the results of the matching degree dimension. Finally, the sorted scheme numbers and their corresponding matching degrees are reorganized to generate the initial ranking list.

[0072] Step S3042: For candidate scheduling schemes with the same matching degree, the parallel relationship is resolved in turn based on the satisfaction of hard objectives, the number of constraint violations, and the continuity of key processes to obtain a priority draft.

[0073] In this embodiment, a set of solutions with the same matching degree is retrieved, and the fulfillment of their hard objectives is checked one by one. If a solution does not fully meet its hard objectives, its priority is directly lowered. Then, the number of constraint violations for the remaining solutions is counted, including energy consumption exceeding limits, resource conflicts, or time delays, and solutions with fewer violations are placed at the top. Next, for solutions that remain tied in the first two steps, the continuity of their key processes is analyzed. For example, it is checked whether there is unnecessary waiting time or resource switching between adjacent processes, and solutions with higher continuity and smoother transitions are promoted to the top. Finally, after all factors are determined, the ranking results are re-integrated to generate a priority draft after refining the solutions with the same matching degree.

[0074] Step S3043: Based on the priority draft, apply boundary adjustments to each candidate scheduling scheme around each objective of the multidimensional objective set, record the order changes caused by the adjustments, and generate a check record.

[0075] In this embodiment, each objective in the multidimensional objective set is selected, and slight adjustments are made to each objective within its permissible range, such as shortening the delivery time tolerance, lowering the energy consumption limit, or increasing the equipment utilization threshold. After each objective adjustment, the satisfaction of the candidate scheduling schemes is re-examined. If a scheme that was originally in the lead violates the adjustment, its ranking is downgraded. Then, the ranking changes of all schemes under each adjustment condition are recorded, and the change information is summarized into a check record, including the direction, magnitude, and triggering objective of the ranking change. Finally, the check record is attached to the priority draft to generate the check record.

[0076] Step S3044: Based on the check record, the priority draft is corrected. When the first candidate scheduling scheme has a critical resource unavailability during the planned time period, its priority is downgraded, and the priority of the other candidate scheduling schemes is adjusted relative to the resource substitutability to obtain the priority sequence.

[0077] In this embodiment, the resource availability of the first candidate scheduling scheme in the priority draft is verified. The key equipment, energy supply, and environmental conditions on which it depends during the planned time period are retrieved. If any unavailability or supply exceeding the available range is found, the scheme is determined to be unenforceable under the current conditions. Subsequently, the scheme is downgraded and a subsequent scheme is selected from the priority draft to fill the gap. Then, the order of the remaining candidate scheduling schemes is adjusted according to the substitutability of resources. For example, when the key equipment can be replaced by backup equipment with similar functions, only minor adjustments are made. When there are no alternative resources, the order of the relevant schemes is significantly downgraded. Finally, after all schemes are adjusted, a new ranking result is generated and recorded as a priority sequence.

[0078] Step S3045: The first candidate scheduling scheme in the priority sequence is used as the execution scheduling scheme, and the second candidate scheduling scheme and its triggering conditions are registered as backup schemes.

[0079] In this embodiment, the top-ranked candidate scheduling scheme is extracted from the priority sequence and determined as the execution scheduling scheme. This scheme will be directly invoked as the baseline plan during actual production. Subsequently, the second-ranked candidate scheduling schemes are screened to identify their advantageous dimensions in satisfying the target set. Triggering conditions are set in combination with resource availability and process connection conditions. For example, the top-ranked scheme is triggered when critical equipment fails, energy supply exceeds limits, or uncontrollable delays occur in delivery time. Then, the second-ranked schemes are bound to the corresponding triggering conditions and registered as backup schemes to ensure that they can be quickly switched without affecting the overall progress when production conditions change abruptly. Finally, the execution scheduling scheme and the backup schemes are archived together to form a two-layer scheduling system that combines execution and emergency response. Through this process, the scheduling decision not only obtains a clear execution scheme but also establishes a dynamic switching mechanism, providing institutional guarantees for the stable operation of the production process.

[0080] Step S4: Monitor production data during the execution of the scheduling scheme. When an anomaly or deviation is detected, determine the influencing factors based on the causal relationship and update the multidimensional digital model.

[0081] In this embodiment, during the execution of the scheduling plan, key parameters, including process completion time, output indicators, energy consumption curves, and equipment load status, are monitored in real time and compared with the preset benchmark values ​​in the plan. When significant deviations or abnormal fluctuations are detected in the data, causal relationship analysis is initiated to trace the deviations step by step to identify the process links, equipment units, and external environmental factors involved, determining whether they are caused by equipment performance degradation, resource allocation imbalance, or external environmental disturbances. Subsequently, the identified influencing factors are mapped back to the corresponding nodes and relationships in the multidimensional digital model, and the relevant parameter values ​​or connection weights are updated so that the model can reflect the latest production status. Finally, after the update is completed, the adjusted model is used as a new decision-making basis to provide input for subsequent scheduling optimization and anomaly prevention. Through this process, the digital model maintains dynamic consistency with actual production, ensuring that management strategies can be promptly corrected as environmental and resource conditions change.

[0082] Example 2 Please see Figure 2 Another embodiment of the present invention provides: a digital management system for factory production processes based on the Internet of Things, comprising: a data acquisition module, a model building module, a scheduling scheme matching module, and an update and optimization module; The data acquisition module is used to collect relevant data from production equipment in the production line, and to perform semantic fusion processing on the relevant data to generate a data stream with correlation. The relevant data includes operating data, energy consumption data, material flow data and environmental data. The model building module is used to establish a multi-relationship graph structure based on the data flow and construct a multi-dimensional digital model of the production process. The scheduling scheme matching module is used to receive production intention information input by the management personnel, generate a production scheduling scheme according to the multi-dimensional digital model, and select the scheduling scheme that best matches the production intention information. The update and optimization module is used to monitor production data during the execution of the scheduling scheme. When an anomaly or deviation is detected, it determines the influencing factors based on causal relationships and updates the multidimensional digital model.

[0083] The scheduling scheme matching module includes: an intent analysis unit, a scheme generation unit, and a matching unit; The intent analysis unit is used to receive production intent information input by the manager and parse the production intent information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate. The scheme generation unit is used to generate several candidate scheduling schemes based on the multidimensional digital model and according to the multidimensional target set. The matching unit is used to compare the candidate scheduling schemes, calculate the matching degree between each scheme and the multidimensional target set, and select the first and best scheduling scheme.

[0084] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0085] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A digital management method for factory production processes based on the Internet of Things, characterized in that: include: Collect relevant data from production equipment in the production line, and perform semantic fusion processing on the relevant data to generate a data stream with correlation. The relevant data includes operation data, energy consumption data, material flow data, and environmental data. Based on the data flow, a multi-relationship graph structure is established to construct a multi-dimensional digital model of the production process; Receive production intention information input by management personnel, generate a production scheduling plan based on the multi-dimensional digital model, and select the scheduling plan that best matches the production intention information; During the execution of the scheduling scheme, production data is monitored. When an anomaly or deviation is detected, the influencing factors are determined based on the causal relationship, and the multidimensional digital model is updated.

2. The IoT-based digital management method for factory production processes as described in claim 1, characterized in that, The relevant data is subjected to semantic fusion processing to generate a data stream with related relationships, including: Timestamp alignment is performed on the collected production equipment-related data; Feature extraction is performed on the aligned production equipment-related data, and the features are classified according to the process stage based on the preset process stage labels; Establish semantic relationships across data types among the categorized data, and construct a data lookup table that reflects the logical dependencies between processes; The various types of data in the data lookup table are serialized and combined according to the order of the processes to generate a data stream with related relationships.

3. The IoT-based digital management method for factory production processes as described in claim 2, characterized in that, Based on the data flow, a multi-relationship graph structure is established to construct a multi-dimensional digital model of the production process, including: The data stream is divided into nodes according to equipment units, process links, energy consumption factors and environmental factors to obtain a basic node set; In the basic node set, different types of edges are established according to the time sequence, material transfer relationship, energy consumption dependence relationship and environmental coupling relationship, forming a graph structure with multiple semantics; The nodes and edges in the graph structure are iteratively combined to generate a multi-layered relationship network covering the equipment layer, process layer, and workshop layer. In the multi-layered relationship network, nodes are arranged in an orderly manner according to the logical sequence of the process stages to obtain a multi-dimensional digital model of the production process.

4. The IoT-based digital management method for factory production processes as described in claim 3, characterized in that, In the aforementioned multi-layered relationship network, nodes are arranged in an orderly manner according to the logical sequence of process stages to obtain a multi-dimensional digital model of the production process, including: The nodes in the multi-layered relationship network are grouped according to the stage of the process to form a stage set corresponding to the production process; Within each stage set, nodes are sorted according to material delivery sequence and energy consumption constraints to determine the arrangement order within the stage; Between adjacent stage sets, a sequential link is established according to the process connection relationship, connecting the output node of the previous stage to the input node of the next stage in sequence; The node sequence, after being sorted within stages and connected between stages, is combined into a whole structure to obtain a multidimensional digital model for representing the production process.

5. The IoT-based digital management method for factory production processes as described in claim 4, characterized in that, The process of receiving production intention information input by management personnel, generating a production scheduling plan based on the multi-dimensional digital model, and selecting the scheduling plan that best matches the production intention information includes: Receive production intention information input by management personnel, and parse the production intention information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate; Based on the multidimensional digital model, several candidate scheduling schemes are generated according to the multidimensional target set, and each candidate scheduling scheme corresponds to a different process ordering and resource allocation method. The candidate scheduling schemes are compared, and the matching degree between each scheme and the multidimensional target set is calculated; A priority sequence is established between the candidate scheduling scheme with the highest matching degree and the remaining schemes, and the first scheme in the priority sequence is used as the execution scheduling scheme.

6. The IoT-based digital management method for factory production processes as described in claim 5, characterized in that, Based on the aforementioned multidimensional digital model, several candidate scheduling schemes are generated according to the aforementioned multidimensional target set, including: Map the multidimensional target set to the corresponding nodes and relationships in the multidimensional digital model; Based on the mapping results, the execution order of different processes is arranged and combined to form multiple process sorting sequences; For each process sequence, different equipment resources and energy consumption ratios are allocated to obtain the corresponding resource allocation scheme. The process sequence is combined with the corresponding resource allocation scheme to generate several candidate scheduling schemes.

7. The IoT-based digital management method for factory production processes as described in claim 5, characterized in that, The candidate scheduling schemes are compared, and the matching degree between each scheme and the multidimensional target set is calculated, including: The multidimensional target set is divided into hard targets and flexible targets, and priority and allowable range are set for each target; Based on the multidimensional digital model, a time-series simulation is performed on each candidate scheduling scheme to extract the evaluation dataset corresponding to the target. The evaluation dataset is compared item by item with the allowable range of the corresponding target to generate a set of deviation information for each candidate scheduling scheme; The deviation information set is sorted and merged according to the priority to obtain the matching degree identifier of each candidate scheduling scheme, and recorded as a scheme matching degree lookup table.

8. The IoT-based digital management method for factory production processes as described in claim 5, characterized in that, The step of establishing a priority sequence between the candidate scheduling scheme with the highest matching degree and the remaining schemes, and using the first scheme in the priority sequence as the execution scheduling scheme, includes: The candidate scheduling schemes are sorted in descending order according to their matching degree to generate an initial sorted list; For candidate scheduling schemes with the same matching degree, the parallel relationship is resolved in turn based on the satisfaction of hard objectives, the number of constraint violations, and the continuity of key processes to obtain a priority draft; Based on the priority draft, each candidate scheduling scheme is adjusted within the boundary around each objective of the multidimensional objective set, the order changes caused by the adjustment are recorded, and a check record is generated. Based on the inspection records, the priority draft is corrected. When the first candidate scheduling scheme has a critical resource unavailability during the planned time period, its priority is downgraded, and the priority of the other candidate scheduling schemes is adjusted relative to the resource substitutability to obtain the priority sequence. The first candidate scheduling scheme in the priority sequence is used as the execution scheduling scheme, and the second candidate scheduling scheme and its triggering conditions are registered as backup schemes.

9. A factory production process digital management system based on the Internet of Things (IoT), used to implement the factory production process digital management method based on the IoT as described in any one of claims 1-8, characterized in that, include: Data acquisition module, model building module, scheduling scheme matching module, and update and optimization module; The data acquisition module is used to collect relevant data from production equipment in the production line, and to perform semantic fusion processing on the relevant data to generate a data stream with correlation. The relevant data includes operating data, energy consumption data, material flow data and environmental data. The model building module is used to establish a multi-relationship graph structure based on the data flow and construct a multi-dimensional digital model of the production process. The scheduling scheme matching module is used to receive production intention information input by the management personnel, generate a production scheduling scheme according to the multi-dimensional digital model, and select the scheduling scheme that best matches the production intention information. The update and optimization module is used to monitor production data during the execution of the scheduling scheme. When an anomaly or deviation is detected, the influencing factors are determined based on the causal relationship, and the multidimensional digital model is updated.

10. The IoT-based digital management system for factory production processes as described in claim 9, characterized in that, The scheduling scheme matching module includes: an intent analysis unit, a scheme generation unit, and a matching unit; The intent analysis unit is used to receive production intent information input by the manager and parse the production intent information into a multi-dimensional target set including output target, delivery time, energy consumption limit and equipment utilization rate. The scheme generation unit is used to generate several candidate scheduling schemes based on the multidimensional digital model and according to the multidimensional target set. The matching unit is used to compare the candidate scheduling schemes, calculate the matching degree between each scheme and the multidimensional target set, and select the first and best scheduling scheme.

Images