Plant intelligent factory multi-dimensional data monitoring and collaborative management method and system

Abstract

The application discloses a plant intelligent factory multidimensional data monitoring and collaborative control method and system, and belongs to the technical field of collaborative control. The method specifically comprises the following steps: collecting multidimensional data of a plant intelligent factory in real time, wherein the multidimensional data comprises plant physiological parameters, environmental factor parameters, nutrient solution parameters, and device and operation state parameters; based on the multidimensional data, a dynamic multi-level causal correlation model between devices, plant physiology, environment and operation is constructed; potential collaborative influence paths in an abnormal state are identified; when an abnormality is detected or a growth process deviates from a trend is predicted, cross-level control is automatically performed; cross-level control instructions are distributed to the device level, the cultivation unit level and the workshop level according to the levels; and the plant intelligent factory is subjected to multi-link joint regulation and control. Through the establishment of the collaborative mechanism of the causal correlation and the path, the regulation and control are more accurate, flexible and adaptive, and the stability, resource utilization rate and intelligent level of plant production are effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of collaborative control technology, specifically a method and system for multi-dimensional data monitoring and collaborative management of intelligent plant factories. Background Technology

[0002] With the deep integration of modern agriculture and intelligent manufacturing technologies, plant factories are gradually becoming an important model for achieving efficient agricultural production. Plant factories achieve controllability and standardization of the entire plant growth process through artificial light sources, climate control, nutrient supply, and intelligent monitoring. However, most existing plant factories rely on single-dimensional monitoring methods, such as monitoring environmental temperature and humidity or nutrient solution concentration, which cannot comprehensively integrate multi-dimensional information such as plant physiological state, environmental factors, equipment operation, and management personnel operation. Therefore, it is difficult to achieve stable and efficient production control under complex dynamic conditions.

[0003] Currently, some plant factories have introduced IoT, big data, and automated control systems, achieving data collection and centralized management to a certain extent. However, most of these systems rely on static thresholds and empirical rules for judgment, lacking cross-level data fusion and causal relationship modeling. For example, when plants exhibit abnormal growth, it is difficult to simultaneously analyze the linkage between lighting, carbon dioxide concentration, nutrient solution supply, and equipment operating status, leading to inaccurate anomaly localization, delayed control strategies, and potentially even resource waste and production risks.

[0004] Furthermore, most existing control methods focus on optimizing single links and lack multi-level synergistic mechanisms. In the complex plant factory production chain, plant physiological responses, environmental disturbances, equipment operation and maintenance, and human operation are coupled together, and adjustments to a single link are insufficient to cope with production deviations under the combined effects of multiple factors. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a multi-dimensional data monitoring and collaborative management method and system for intelligent plant factories. This method is capable of dynamic causal modeling based on multi-dimensional data, combined with cross-level collaborative management and control, and possesses continuous optimization and evolution capabilities, thereby comprehensively improving the stability, flexibility, and intelligence level of plant factories.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A multi-dimensional data monitoring and collaborative management method for plant-based smart factories, including:

[0008] Real-time collection of multi-dimensional data from the intelligent plant factory, including plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operating status parameters;

[0009] Based on the multi-dimensional data, a dynamic multi-level causal relationship model is constructed among equipment, plant physiology, environment and operation to identify potential synergistic impact paths under abnormal conditions.

[0010] When an anomaly is detected or a deviation from the growth trend is predicted, cross-level control is automatically implemented, distributing cross-level control instructions to the equipment level, cultivation unit level, and workshop level, enabling multi-stage linkage control of the plant intelligent factory.

[0011] Specifically, based on the aforementioned multi-dimensional data, a dynamic, multi-level causal relationship model is constructed among equipment, plant physiology, environment, and operation to identify potential synergistic impact paths under abnormal conditions, including:

[0012] The multi-dimensional data is preprocessed in layers according to equipment parameter layer, plant physiological layer, environmental factor layer and operation layer, and the data in the same layer is normalized and time-synchronized.

[0013] By establishing a multi-level feature index, key factors with potential coupling relationships between different levels are identified, and a cross-level correlation factor set is generated.

[0014] Based on the cross-layer associated factor set, a temporal interaction sequence between factors at different levels is constructed, wherein the temporal interaction sequence represents the sequential relationship and transmission path of factors in the dynamic process.

[0015] Based on the time-series interaction sequence, key factors of the device parameter layer, plant physiology layer, environmental factor layer and operation layer are mapped to a multi-level causal graph, and cross-layer interaction links are marked in the graph.

[0016] In the multi-level causal graph, potential collaborative influence paths corresponding to abnormal states are extracted according to a preset path traversal logic.

[0017] Specifically, based on the aforementioned temporal interaction sequence, key factors at the device parameter layer, plant physiological layer, environmental factor layer, and operational layer are mapped to a multi-level causal graph, and cross-layer interaction links are marked in the graph, including:

[0018] The device factors, plant physiological factors, environmental factors, and operational factors in the time-series interaction sequence are abstracted into graph nodes, and each node is assigned a unique identifier.

[0019] Based on the temporal dependencies between factors at the same level, intra-layer causal links are generated in the causal graph, and the intra-layer causal links reflect the dynamic interaction logic between key factors at the same level.

[0020] The factor pairs with cross-layer temporal dependencies are traversed, and cross-layer action links are established in the causal graph to mark the causal connections between different layers.

[0021] The intra-layer causal links and cross-layer interaction links are structurally integrated to form a multi-layer causal graph with hierarchical topological features.

[0022] Specifically, in the multi-level causal graph, potential collaborative influence paths corresponding to abnormal states are extracted according to a preset path traversal logic, including:

[0023] In the multi-level causal graph, based on the triggering information corresponding to the abnormal state, the node related to the abnormal state is determined as the starting anchor point for traversal.

[0024] Starting from the traversal starting anchor point, the path traversal logic is recursively applied layer by layer, and successive nodes that have a causal relationship with the traversal starting anchor point are visited in turn to form a candidate path set.

[0025] The causal consistency test is performed on each candidate path set to exclude paths that are inconsistent with the graph logic or unrelated to the abnormal state, and to generate valid paths that meet the constraints.

[0026] The effective paths are structured and aggregated to generate potential collaborative impact paths corresponding to abnormal states.

[0027] Specifically, when an anomaly is detected or a deviation from the growth trend is predicted, automatic cross-level control is implemented, distributing cross-level control instructions to the equipment level, cultivation unit level, and workshop level, enabling multi-stage coordinated control of the plant intelligent factory, including:

[0028] The multi-source triggering information generated by anomaly detection or growth process prediction is merged to generate cross-level control events, and the scope, time limit and priority labels are marked for the cross-level control events.

[0029] Based on potential synergistic impact paths, the set of controlled objects and prohibited boundaries are determined, forming a control intention that includes target state constraints, resource constraints, and temporal constraints;

[0030] Based on the stated control intent, the control actions are decomposed into atomic instruction sequences at the equipment level, cultivation unit level, and workshop level, establishing upstream and downstream dependencies and execution windows, and generating complete instructions for cross-layer scheduling and invocation;

[0031] The atomic instruction sequence is checked for resource usage, timing, and mutual exclusion. When mutual exclusion is detected, priority arbitration and alternative sequence reconstruction are performed to generate an executable list that meets the constraints.

[0032] The executable list is distributed to the corresponding execution terminals according to the hierarchy, enabling multi-stage linkage control of the plant intelligent factory.

[0033] Specifically, the process of determining the set of controlled objects and prohibited boundaries based on potential collaborative influence paths, and forming a control intention that includes target state constraints, resource constraints, and temporal constraints, includes:

[0034] In the potential synergistic impact path, the set of controlled objects that need to be controlled is determined based on the causal relationship of the nodes, and an independent index is established for each controlled object;

[0035] For the set of controlled objects, based on the conflict conditions existing in the cross-layer action link, a prohibition boundary is set to limit the scope of operations that cannot be triggered;

[0036] By combining the set of controlled objects and the prohibited boundaries, a set of control constraints is generated, which includes target state constraints, resource constraints, and time sequence constraints.

[0037] The aforementioned set of control constraints is structured and integrated to form a complete control intent.

[0038] Specifically, based on the stated control intent, the control actions are decomposed into atomic instruction sequences at the equipment, cultivation unit, and workshop levels. Upstream and downstream dependencies and execution windows are established, and complete instructions for cross-level scheduling and invocation are generated, including:

[0039] The control intent is parsed into atomic instruction sequences at the equipment level, cultivation unit level, and workshop level, and each instruction is assigned a unique identifier;

[0040] Based on the triggering order and resource usage conditions in the atomic instruction sequence, an upstream and downstream dependency chain is established;

[0041] Based on the aforementioned dependency chain, and according to time intervals and parallelism constraints, corresponding execution windows are divided, and each instruction is mapped to the corresponding execution window;

[0042] The mapped atomic instructions, their dependencies, and execution windows are serialized and aggregated to generate complete instructions for cross-layer scheduling and invocation.

[0043] Specifically, the atomic instruction sequence is checked for resource usage, timing, and mutual exclusion. Upon detection of mutual exclusion, priority arbitration and alternative sequence reconstruction are performed to generate an executable list that meets the constraints, including:

[0044] The atomic instruction sequence is checked for resource usage, timing, and security mutual exclusion to generate a set of verification results containing conflict markers and triggering conditions.

[0045] The test result set is mapped to the instruction dependency chain to locate the conflicting instruction node and its upstream and downstream related nodes, and the non-parallelizable domain and reorderable domain are divided accordingly.

[0046] Within the non-parallel domain, relevant instructions are prioritized and arbitrated based on control intent, prohibition boundaries, and execution windows to determine the set of instructions to be retained, postponed, and replaced.

[0047] Under the constraints of the reorderable domain and the set of replacement instructions, the instruction sequence that satisfies the dependency relationship and execution window is reconstructed, and an executable list containing the instruction order, execution window and mutual exclusion masking rules is output.

[0048] A multi-dimensional data monitoring and collaborative management system for intelligent plant factories is used to implement the aforementioned multi-dimensional data monitoring and collaborative management method for intelligent plant factories, including: a data acquisition module, an impact path identification module, and a linkage control module;

[0049] The data acquisition module is used to collect multi-dimensional data of the plant intelligent factory in real time. The multi-dimensional data includes plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operation status parameters.

[0050] The influence path identification module, based on the multi-dimensional data, constructs a dynamic multi-level causal relationship model between equipment, plant physiology, environment and operation, and identifies potential collaborative influence paths under abnormal conditions.

[0051] The linkage control module is used to automatically perform cross-level control when an anomaly is detected or a deviation from the growth process is predicted. It distributes cross-level control instructions to the equipment level, cultivation unit level and workshop level according to the level, so as to carry out multi-link linkage control of the plant intelligent factory.

[0052] Specifically, the linkage control module includes: an intent recognition unit, an instruction generation unit, and a linkage control unit;

[0053] The intent recognition unit determines the set of controlled objects and prohibited boundaries based on potential collaborative influence paths, forming a control intent that includes target state constraints, resource constraints, and temporal constraints.

[0054] The instruction generation unit is used to decompose the control action into atomic instruction sequences at the equipment level, cultivation unit level and workshop level according to the control intention, and generate complete instructions for cross-level scheduling and invocation.

[0055] The linkage control unit is used to distribute the executable list to the corresponding execution terminal according to the hierarchy, so as to carry out multi-stage linkage control of the plant intelligent factory.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] This invention proposes a multi-dimensional data monitoring and collaborative management method and system for intelligent plant factories. It integrates plant physiological parameters, environmental factors, nutrient solution parameters, and equipment and operational status into a unified data framework. Based on causal modeling, it constructs cross-level dynamic correlations. When abnormal states or deviations from growth trends occur, it can quickly identify potential collaborative influence paths and generate cross-level control intentions and actionable lists, achieving multi-stage linkage control from the equipment level, cultivation unit level, to the workshop level. The overall method not only achieves global perception and dynamic control of complex elements in the plant factory but also, by establishing a collaborative mechanism of causal relationships and paths, makes control more precise, flexible, and adaptive, thereby effectively improving the stability, resource utilization, and intelligence level of plant production. Attached Figure Description

[0058] Figure 1 Flowchart of the multi-dimensional data monitoring and collaborative management method for intelligent plant factories provided by this invention;

[0059] Figure 2 A schematic diagram illustrating the principle structure of the multi-level causal relationship model provided by this invention;

[0060] Figure 3 The linkage control flowchart provided by this invention;

[0061] Figure 4 This is an architecture diagram of the multi-dimensional data monitoring and collaborative management system for intelligent plant factories provided by the present invention. Detailed Implementation

[0062] 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.

[0063] 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.

[0064] It should be noted that, unless there is a 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 executed in a different order than the module division in the device or the order in the flowchart. In addition, the terms "first," "second," and "third" used in this application do not limit the data or execution order, but only distinguish identical or similar items with essentially the same function and effect.

[0065] 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.

[0066] Example 1

[0067] Please see Figures 1-3 The present invention provides an embodiment of a multi-dimensional data monitoring and collaborative management method for intelligent plant factories, comprising the following specific steps:

[0068] Step S1: Collect multi-dimensional data of the plant intelligent factory in real time. The multi-dimensional data includes plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operation status parameters.

[0069] In this embodiment, multi-source sensors and monitoring nodes need to be deployed within the intelligent plant factory to acquire plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operational status parameters. Plant physiological parameters reflect plant growth status through indicators such as chlorophyll fluorescence, leaf surface temperature, or transpiration intensity. Environmental factor parameters describe the effects of the external environment on plants through data such as light intensity, air temperature and humidity, and carbon dioxide concentration. Nutrient solution parameters characterize the root absorption environment through information such as pH, conductivity, and nutrient concentration. Equipment and operational status parameters consist of lighting parameters, irrigation parameters, climate control equipment parameters, and manual operation record parameters.

[0070] Step S2: Based on the multi-dimensional data, construct a dynamic multi-level causal relationship model between equipment, plant physiology, environment and operation, and identify potential synergistic influence paths under abnormal conditions.

[0071] The specific steps of step S2 are as follows:

[0072] Step S201: Perform layered preprocessing on the multi-dimensional data according to the equipment parameter layer, plant physiology layer, environmental factor layer and operation layer, and perform normalization and time synchronization processing on the data in the same layer respectively.

[0073] In this embodiment, based on the data source and physical attributes, the collected multi-dimensional data is divided into an equipment parameter layer, a plant physiology layer, an environmental factor layer, and an operational layer to ensure clear hierarchical boundaries for different categories of data during modeling. Subsequently, within each layer, to address the issue of inconsistent dimensions and sampling frequencies, numerical data needs to be normalized to ensure comparability within the same numerical range and avoid weight shifts caused by scale differences. Simultaneously, considering the possibility of timestamp misalignment at different acquisition nodes, time synchronization is achieved through interpolation, resampling, or time window alignment, enabling data points within the same layer to correspond to a unified time series benchmark. Through this hierarchical preprocessing principle, a normalized dataset with four levels is finally obtained.

[0074] Step S202: By establishing a multi-level feature index, identify key factors with potential coupling relationships between different levels, and generate a cross-level correlation factor set.

[0075] In this embodiment, after completing the hierarchical preprocessing, a multi-level feature index is established for the data of the equipment parameter layer, plant physiology layer, environmental factor layer, and operation layer. This index, by defining the semantic labels, temporal identifiers, and functional attributes of each type of factor, enables the location and correlation of cross-layer data. Specifically, by comparing the response patterns and changing trends of data from different layers over time, key factors with mutually coupled characteristics are identified, such as the linkage between plant leaf photosynthetic rate and light intensity, and the synchronous changes between nutrient solution conductivity and root transpiration rate. It should be noted that the identification process not only considers the direct correlation between factors but also needs to combine upstream and downstream logical relationships to extract factor combinations with indirect interaction links. Through this process, a set of cross-layer correlated factors is finally formed, which can characterize the potential coupling relationships between multi-layer data.

[0076] Step S203: Based on the cross-level associated factor set, construct a temporal interaction sequence between factors at different levels. The temporal interaction sequence represents the sequential relationship and transmission path of factors in the dynamic change process.

[0077] In this embodiment, after obtaining the cross-level correlation factor set, these factors need to be logically arranged in chronological order to generate a temporal interaction sequence that reflects the causal progression. Specifically, by aligning and comparing the changing trends of factors at different levels on the time axis, the triggering relationship between the changes in the factors that occurred first and the subsequent response factors is identified, and this sequential connection is recorded in a serialized manner. It should be noted that the construction of temporal interactions is not limited to a single direct cause and effect, but should also cover multiple transmission paths across levels. For example, changes in light intensity in the environmental factor layer first cause fluctuations in the transpiration rate of the plant physiological layer, and then further affect the circulation flow rate of the nutrient solution layer. Through this layer-by-layer progressive approach, a set of dynamic interaction sequences is finally obtained.

[0078] Step S204: Based on the time-series interaction sequence, map the key factors of the device parameter layer, plant physiology layer, environmental factor layer and operation layer to a multi-level causal map, and mark the cross-layer interaction links in the map.

[0079] The specific steps of step S204 are as follows:

[0080] Step S2041: Abstract the device factors, plant physiological factors, environmental factors and operational factors in the time-series interaction sequence into graph nodes, and assign a unique identifier to each node.

[0081] In this embodiment, to achieve a unified representation across levels in the causal graph, it is necessary to first abstract the various factors in the temporal interaction sequence. Specifically, equipment factors such as lighting power and irrigation pump rate, plant physiological factors such as photosynthetic rate and leaf temperature, environmental factors such as air humidity and carbon dioxide concentration, and operational factors such as artificial fertilization and maintenance operations are all abstracted into independent nodes in the graph. To ensure the uniqueness of subsequent retrieval and retrieval, each node needs to be assigned a unique identifier, which includes information such as factor category, timestamp, and attribute label, ensuring that the same type of factor can be accurately distinguished in different times or different operational scenarios. Through this abstraction and identification operation, the temporal interaction sequence is transformed from the original multidimensional data stream into a structured set of nodes.

[0082] Step S2042: Based on the temporal dependencies between factors at the same level, generate intra-layer causal links in the causal graph. The intra-layer causal links reflect the dynamic interaction logic between key factors at the same level.

[0083] In this embodiment, during the construction of the causal graph, it is necessary to establish intra-layer causal links for factors within the same level based on their time-series dependencies. Specifically, by analyzing the change patterns of factors within the same level on the time axis, the logical relationship between pre-trigger and subsequent response is identified. For example, in the plant physiological layer, the increase in photosynthetic rate usually precedes the change in transpiration intensity; in the equipment layer, the start-up of irrigation pumps often precedes the increase in nutrient solution flow rate. It should be noted that this dependency relationship is not a simple numerical correlation, but a dynamic interaction structure extracted based on the functional logic and temporal patterns between factors. By mapping these dependencies into directional links in the graph, a network representing the dynamic logic of factors can be formed within the layer.

[0084] Step S2043: Traverse the factor pairs that have cross-layer temporal dependencies, establish cross-layer action links in the causal graph, and mark the causal connections between different layers.

[0085] In this embodiment, after generating intra-layer causal links, it is necessary to further identify the temporal dependencies between cross-level factors and establish cross-layer interaction links in the causal graph. Specifically, by iterating through the factor pairs of the device layer, plant physiology layer, environment layer, and operation layer one by one, and comparing their sequential changes over time, if it is found that the fluctuation of a factor in one layer consistently leads the response of a factor in another layer, it is determined that there is a cross-layer dependency between the two. For example, changes in light intensity in the environment layer may precede the chlorophyll fluorescence response in the plant physiology layer, and fertilization actions in the operation layer may precede changes in conductivity in the nutrient solution layer. It should be noted that when establishing cross-layer interaction links, inter-layer logical constraints should be combined to ensure that the links not only reflect temporal relationships but also embody the rationality of causal driving. Finally, the identified factor pairs are mapped to directional cross-layer connections in the graph, thereby forming a causal topological framework covering different levels.

[0086] Step S2044: The intra-layer causal links and cross-layer interaction links are structurally integrated to form a multi-level causal graph with hierarchical topological features.

[0087] In this embodiment, after constructing the intra-layer causal links and cross-layer interaction links, they need to be structurally integrated to form a multi-level causal graph with hierarchical topological characteristics. Specifically, firstly, the causal links within the same layer are treated as local subgraphs, maintaining the internal logical relationships of factors at each layer. Then, cross-layer interaction links are introduced and matched with the nodes of the corresponding subgraphs, so that cross-layer dependencies can act as bridges to connect different subgraphs into a whole network. It should be noted that during the integration process, node identifiers, time order, and logical constraints should be unified to avoid duplicate mappings or logical conflicts. Through this structural integration operation, a causal graph composed of multi-level subgraphs and cross-layer connections is finally obtained.

[0088] Step S205: In the multi-level causal graph, extract the potential collaborative influence paths corresponding to the abnormal state according to the preset path traversal logic.

[0089] The specific steps of step S205 are as follows:

[0090] Step S2051: In the multi-level causal graph, based on the trigger information corresponding to the abnormal state, determine the node related to the abnormality as the starting anchor point for traversal.

[0091] In this embodiment, given that a multi-level causal graph has been established, it is necessary to locate the starting anchor point for traversal based on the triggering information corresponding to the abnormal state. Specifically, firstly, the abnormal state is analyzed to extract the factor categories, occurrence time, and associated hierarchical information involved. Then, this information is matched with the node identifiers in the causal graph to filter out target nodes directly related to the abnormality. For example, when an abnormal decrease in photosynthetic rate is detected in the plant physiological layer, the node is marked as an abnormal anchor point. If a sudden change in light intensity in the environmental layer is detected at the same time, the node is used as a candidate anchor point. It should be noted that when determining the anchor point, the triggering position of the factor in the causal graph and its possible downstream influence range are given priority to ensure that the subsequent path traversal can unfold along the causal chain related to the abnormality. Through this process, the traversal starting anchor point corresponding to the abnormal state is finally obtained.

[0092] Step S2052: Starting from the traversal starting anchor point, proceed layer by layer according to the preset path traversal logic, and visit the successor nodes that have a causal relationship with the traversal starting anchor point in turn to form a candidate path set.

[0093] In this embodiment, after determining the starting anchor point for traversal, it is necessary to recursively traverse from the anchor point layer by layer according to the preset path traversal logic to identify successor nodes with causal relationships. Specifically, firstly, based on the directionality of the links in the causal graph, the nodes are expanded from the anchor point to downstream nodes according to the chronological order or logical triggering order. At each step of the recursion, it is necessary to determine whether the node satisfies the cross-layer or intra-layer dependency constraints. If it does, it is included in the traversal path and the expansion continues. Through continuous iterative expansion, candidate path branches consisting of multiple nodes are gradually formed. It should be noted that the formation of candidate paths not only considers directly connected successor nodes, but also covers paths that may have multiple indirect connections, thereby ensuring that the causal propagation chain related to abnormal states is completely captured. Finally, a set of candidate paths is generated by the traversal logic.

[0094] Step S2053: Perform causal consistency checks on each candidate path in the set, exclude paths that are inconsistent with the graph logic or unrelated to the abnormal state, and generate valid paths that meet the constraints.

[0095] In this embodiment, after obtaining the candidate path set, each path needs to undergo a causal consistency check to ensure its logical rationality and relevance to the abnormal state. Specifically, firstly, the causal links between adjacent nodes in the path are compared one by one to check whether their directionality is consistent with the existing topological rules in the causal graph. If there are reverse or broken situations, they are excluded. Secondly, the node states in the path are matched with the abnormal triggering information to determine whether the path truly reflects the causal chain of abnormal propagation. If it is found that the node changes are not directly or indirectly related to the abnormal state, the path is deemed invalid. Finally, according to preset constraints, such as time window consistency and cross-layer logic constraints, the entire path is comprehensively evaluated, and paths that meet the conditions are selected. Through this process, a set of effective paths is finally obtained, which can accurately represent the transmission logic of the abnormal state among multiple factors.

[0096] Step S2054: The effective paths are structurally aggregated to generate potential collaborative impact paths corresponding to the abnormal state.

[0097] In this embodiment, after obtaining valid paths that meet the causal consistency test, these paths are structured and aggregated to form potential synergistic influence paths that can directly correspond to abnormal states. Specifically, firstly, the valid paths are reordered and grouped according to the hierarchical attributes and time order of the nodes in the paths to ensure that the causal relationships of factors at different levels are clearly presented. Then, through a unified path coding method, the nodes, links, and constraints involved are encapsulated into standardized path units, thereby achieving the indexability and reusability of the paths. In this process, the overlapping parts between different valid paths also need to be merged to avoid duplicate expressions or logical conflicts. It should be noted that this aggregation operation not only preserves the independence of each path, but also integrates them to form a synergistic influence set with a wider coverage, ultimately generating potential synergistic influence paths.

[0098] Figure 2 This diagram illustrates the principle structure of a multi-level causal relationship model. The model uses equipment parameter layer, plant physiology layer, environmental factor layer, and operational layer as basic units, representing the interactions between factors through nodes and links. Each layer contains several factors within the same layer, and the dynamic logical connections between them are described using intra-layer causal links. Different layers are connected through cross-layer interaction links. Dashed lines between factors within the same layer indicate temporal dependencies or logical correlations between key factors within the same layer. For example, within the equipment parameter layer, there is a sequential triggering dependency between irrigation pump start / stop signals and nutrient solution flow data; within the plant physiology layer, there is also intra-layer dynamic coupling between changes in photosynthetic rate and changes in transpiration intensity. Dashed lines with arrows represent causal interaction paths across layers, with the arrow direction reflecting the triggering order or causal transmission direction between factors. For example, changes in light intensity in the environmental factor layer may precede changes in leaf photosynthetic rate in the plant physiology layer, thus being marked with a dashed line with an arrow in the causal diagram, indicating the action from the environmental factor layer to the plant physiology layer.

[0099] Specifically, this application constructs a causal network with hierarchical topological features. When an abnormal state occurs, it can traverse along the causal paths marked in the graph and quickly identify potential collaborative influence paths related to the abnormality.

[0100] Step S3: When an anomaly is detected or a deviation from the growth trend is predicted, cross-level control is automatically implemented, and cross-level control instructions are distributed to the equipment level, cultivation unit level and workshop level according to the level, so as to carry out multi-link linkage control of the plant intelligent factory.

[0101] like Figure 3 As shown, the specific steps of step S3 are as follows:

[0102] Step S301: Merge the multi-source triggering information generated by anomaly detection or growth process prediction to generate cross-level control events, and assign scope, time limit and priority labels to the cross-level control events.

[0103] In this embodiment, during the operation of the plant smart factory, the anomaly detection module and the growth process deviation prediction module continuously output trigger information from different sources. This information may come from plant physiological monitoring, environmental factor fluctuations, nutrient solution supply anomalies, and equipment operation deviations. To avoid redundancy or conflicts caused by information fragmentation, these multi-source trigger information needs to be merged. Specifically, firstly, trigger events within the same time window are aggregated to identify whether they point to the same anomaly category or have overlapping causal relationships. Then, the aggregated events are abstracted into unified cross-level control events, and key metadata is added to their structure, including the event scope, such as the level and object set of the impact, the time limit, such as constraints that must be handled within a specific period, and priority labels, such as sorting according to the severity of the anomaly or the risk to yield and quality. Through this merging and calibration process, cross-level control events with unified semantics and operational attributes are formed.

[0104] Step S302: Based on the potential collaborative impact path, determine the set of controlled objects and the prohibited boundary, and form a control intention that includes target state constraints, resource constraints and time sequence constraints.

[0105] The specific steps of step S302 are as follows:

[0106] Step S3021: In the potential synergistic impact path, determine the set of controlled objects that need to be controlled based on the causal relationship of the nodes, and establish an independent index for each controlled object.

[0107] In this embodiment, after the potential synergistic impact path has been extracted, it is necessary to further identify the set of objects that actually need to be controlled in the path. Specifically, firstly, the nodes in the path are traversed, and their position and strength of influence in the abnormal propagation are analyzed in combination with the directionality of the causal link. Nodes that are in the key driving link or have a decisive role in the spread of abnormality are screened out, such as the leaf photosynthetic efficiency node in the plant physiological layer, the carbon dioxide concentration node in the environmental layer, the conductivity node in the nutrient solution layer, or the irrigation pump start / stop node in the equipment layer. Subsequently, an independent index is established for each controlled object. This index consists of information such as object category, level, timestamp, and location identifier to ensure that different objects can be accurately located and scheduled in the same scenario. Through this identification and indexing process, a set of controlled objects with clear boundaries and operable attributes is finally obtained.

[0108] Step S3022: For the set of controlled objects, based on the conflict conditions existing in the cross-layer action link, set a prohibition boundary to limit the scope of operations that cannot be triggered.

[0109] In this embodiment, after determining the set of controlled objects, the potential conflict conditions in the cross-layer action chain are further considered to set prohibition boundaries for these objects, limiting their operational scope. Specifically, the cross-layer chain is first traversed to analyze whether there are resource occupation conflicts, temporal dependency contradictions, or logical mutual exclusions during the execution of factors at different levels. For example, the high-frequency start-stop of irrigation pumps may conflict with the stability of nutrient solution circulation, and the operation of increasing light intensity may be mutually constrained with the start of cooling fans at specific times. Subsequently, these conflict conditions are mapped to the set of controlled objects, marking the non-triggerable operation boundaries for the relevant objects, and solidifying them in the control framework in the form of restriction rules. It should be noted that the prohibition boundaries not only cover hard prohibition conditions, but also include restrictions that take effect dynamically in specific environmental contexts, ensuring that no new contradictions or risks are introduced when executing control actions. In this way, a set of prohibition boundaries corresponding to the set of controlled objects is finally formed.

[0110] Step S3023: Combine the controlled object set and the prohibited boundary to generate a control constraint set that includes target state constraints, resource constraints and timing constraints.

[0111] In this embodiment, after obtaining the set of controlled objects and their corresponding prohibited boundaries, a unified set of control constraints is further constructed to ensure that the generation of subsequent instructions conforms to the restrictions of multi-dimensional conditions. Specifically, firstly, for each controlled object, a clear target state constraint is set based on the expected value of the target state and the abnormal correction requirements to limit the final control direction of the object. Secondly, based on the resource consumption relationship of each controlled object in the cross-layer causal chain, resource constraints are defined, such as the nutrient solution supply must not exceed the maximum flow of the circulation system, and the lighting power must not exceed the upper limit of power allocation. Thirdly, based on the logical dependencies and temporal order between objects, temporal constraints are generated to specify the sequential connection and parallelism of instruction execution. It should be noted that this set of control constraints is organized in a structured manner and can simultaneously cover the requirements of target orientation, resource allocation, and time sequence.

[0112] Step S3024: The control constraint set is structurally integrated to form a complete control intent.

[0113] In this embodiment, after the target state constraints, resource constraints, and timing constraints are generated separately, they need to be structurally integrated to form a complete control intent that can directly drive the generation of subsequent instructions. Specifically, firstly, the three types of constraints are mapped according to the index of the controlled object, so that each object corresponds to a complete set of constraint rules. Then, the constraints between different objects are merged through the logic rule engine, and priority overlap or condition conflict is handled to ensure that the overall constraint system remains consistent globally. Next, the integrated result is structurally encapsulated to form an intent unit containing constraint hierarchy, scope of action, and triggering conditions. Finally, the intent unit is associated with cross-level control events to realize a traceable logical chain from abnormal state to control action.

[0114] Step S303: Based on the control intent, decompose the control action into atomic instruction sequences at the equipment level, cultivation unit level, and workshop level, establish upstream and downstream dependencies and execution windows, and generate complete instructions for cross-layer scheduling and invocation.

[0115] The specific steps of step S303 are as follows:

[0116] Step S3031: Parse the control intent into atomic instruction sequences at the device level, cultivation unit level, and workshop level, and assign a unique identifier to each instruction.

[0117] In this embodiment, after a complete control intention is formed, it is translated into an executable sequence of atomic instructions and then decomposed hierarchically. Specifically, based on the target state, resource allocation, and timing constraints defined in the control intention, the overall control requirements are first decomposed into control actions of the smallest granularity. For example, equipment-level instructions correspond to the start / stop of irrigation pumps or lighting adjustment, cultivation unit-level instructions correspond to the nutrient solution supply or air circulation control of a specific crop area, and workshop-level instructions correspond to overall environmental regulation or energy consumption allocation. Subsequently, a unique identifier is assigned to each atomic instruction, which consists of elements such as hierarchical attributes, object index, action type, and timestamp. It should be noted that the instruction parsing process not only ensures the logical integrity of the control intention but also converts the complex multi-level control requirements into schedulable and traceable instruction units.

[0118] Step S3032: Establish an upstream and downstream dependency chain based on the triggering order and resource occupancy conditions in the atomic instruction sequence.

[0119] In this embodiment, after the atomic instruction sequence is generated and assigned a unique identifier, it is necessary to further establish an upstream and downstream dependency chain based on its execution logic. Specifically, firstly, the triggering order of each instruction is analyzed. If the execution result of a certain instruction constitutes a prerequisite for a subsequent instruction, a sequential dependency relationship is established between the two. For example, the irrigation valve opening instruction must precede the nutrient solution flow adjustment instruction. Secondly, conflicts in resource usage of the instructions are checked. When multiple instructions involve the same resource unit and cannot be executed in parallel, their execution order needs to be constrained through the dependency chain. For example, lighting power allocation and ventilation power adjustment should avoid simultaneous overload requests within the same time period. It should be noted that this dependency chain not only includes direct sequential connections but also covers logical interlocks between cross-layer instructions, thereby organizing the scattered atomic instruction sequence into an ordered chain structure. Through this process, an upstream and downstream dependency chain that reflects the execution logic and resource constraints is finally obtained.

[0120] Step S3033: Based on the dependency chain, divide the corresponding execution windows according to the time interval and parallelism constraints, and map each instruction to the corresponding execution window.

[0121] In this embodiment, after establishing upstream and downstream dependency chains, the execution process of atomic instructions is mapped to specific time frames to ensure that control actions are implemented in an orderly manner under the conditions of logical correctness and resource availability. Specifically, firstly, based on the constraints in the dependency chain, the earliest executable time and the latest allowed completion time of each instruction are determined, and an initial time interval is defined accordingly. Subsequently, based on parallelism constraints, instructions that can be executed simultaneously and do not conflict are grouped and mapped to the same execution window, while instructions with conflicts or dependencies are assigned to adjacent or sequential execution windows. During this process, the scheduling of cross-layer instructions is corrected to ensure that actions at different levels remain coordinated and consistent in the time dimension. It should be noted that the final generated execution window not only clarifies the start and end intervals of instruction execution but also provides operable boundary conditions for the parallel and serial relationships between instructions, forming a structured time scheduling framework.

[0122] Step S3034: Serialize and aggregate the mapped atomic instructions, their dependencies, and execution windows to generate complete instructions for cross-layer scheduling calls.

[0123] In this embodiment, after atomic instructions are mapped to execution windows, these instructions, their corresponding dependencies, and execution window information are uniformly collected and serialized to form a complete instruction that can be scheduled and invoked across layers. Specifically, firstly, the unique identifier of each instruction is bound to its respective execution window and upstream and downstream dependent nodes to ensure that the instruction can be correctly parsed when invoked. Then, all the bound instructions are concatenated according to chronological and logical order to form a structured serialized set. During this process, concurrency conflicts and timing contradictions between cross-layer instructions also need to be adjusted so that the final sequence conforms to both the constraints of the dependency chain and the scheduling requirements of the execution window. It should be noted that this serialization and collection operation not only transforms scattered atomic instructions into a complete instruction set.

[0124] Step S304: Perform resource usage, timing, and security mutual exclusion checks on the atomic instruction sequence. When mutual exclusion is detected, perform priority arbitration and alternative sequence reconstruction to generate an executable list that meets the constraints.

[0125] The specific steps of step S304 are as follows:

[0126] Step S3041: Perform resource usage verification, timing verification, and security mutual exclusion verification on the atomic instruction sequence to generate a set of verification results containing conflict markers and triggering conditions.

[0127] In this embodiment, after generating a set of serialized atomic instructions, it is checked to ensure that execution abnormalities will not occur due to resource conflicts or logical contradictions during subsequent scheduling. Specifically, firstly, in the resource occupancy check, instructions involving the same resource unit are compared to determine whether there is resource overlap or exceeding the threshold, such as the same irrigation pipeline being occupied by different instructions at the same time. Secondly, in the timing check, the execution order of instructions is checked according to the dependency chain. If reverse calls or executions that do not meet the preconditions are found, they are marked as timing conflicts. Thirdly, in the safety mutual exclusion check, instruction combinations that may cause danger or violate operational constraints are detected, such as the situation where the lighting power increases and the cooling fan is turned off at the same time. Through the above multi-dimensional checks, a set of verification results is finally generated, which includes the markings and triggering conditions of various conflicts.

[0128] Step S3042: Map the test result set to the instruction dependency chain, locate the conflicting instruction node and its upstream and downstream related nodes, and divide the non-parallel domain and reorderable domain accordingly.

[0129] In this embodiment, after obtaining the verification result set containing conflict markers and triggering conditions, it is mapped to the existing instruction dependency chain to accurately locate the problem area in the global scheduling framework. Specifically, the conflict information identified in the verification result set is first matched with the unique identifier of the instruction to determine the specific instruction node where the conflict occurred. Then, the upstream and downstream associations of the node in the dependency chain are traced to identify the relevant instruction set affected by it. Based on this, the node containing the conflict and its dependency chain segment is defined as a non-parallel domain, indicating that the instructions in this region must be executed in strict order. Nodes and chain segments that do not involve conflicts are defined as reorderable domains, allowing flexible adjustment of the execution order under the premise of satisfying dependency constraints.

[0130] Step S3043: Within the non-parallel domain, priority arbitration is performed on relevant instructions based on the control intent, prohibition boundary, and execution window to determine the instruction set of retained instructions, delayed instructions, and replacement instructions.

[0131] In this embodiment, given the clear existence of instruction conflicts in the non-parallel domain, priority arbitration is needed to determine the execution priority of different instructions. Specifically, firstly, conflicting instructions are sorted according to the target state constraints defined in the control intent, prioritizing the retention of instructions directly corresponding to the global control target. Secondly, based on prohibited boundary conditions, instructions that violate security mutual exclusion rules or are triggered out of bounds are eliminated and marked as non-executable. Subsequently, according to the timing arrangement of the execution window, the remaining instructions are scheduled in the time dimension. If resource conflicts cannot be resolved within the same window, some instructions are postponed to subsequent windows for execution. For instructions with functional substitution relationships, logically equivalent but more resource-efficient alternative instructions are selected to alleviate conflicts. It should be noted that the final result of this arbitration process is a set of instructions containing retained instructions, postponed instructions, and replacement instructions.

[0132] Step S3044: Under the constraints of the reorderable domain and the set of replacement instructions, reconstruct the instruction sequence that satisfies the dependency relationship and execution window, and output an executable list containing the instruction order, execution window and mutual exclusion masking rules.

[0133] In this embodiment, after priority arbitration, the instructions in the reorderable domain and the set of replacement instructions are uniformly scheduled to generate a final list of executable instructions. Specifically, firstly, the instructions in the reorderable domain are reordered according to the dependency chain to ensure that all preconditions are met before the execution of subsequent instructions. Secondly, the instructions in the set of replacement instructions are merged with the original sequence to replace the removed or delayed instruction nodes, thereby maintaining the integrity of the overall logic. Subsequently, under the constraints of the execution window, the instructions that can be executed in parallel are grouped, and mutual exclusion rules are introduced for instructions with potential conflicts to avoid resource contention or logical contradictions. Finally, the processed instructions are output in a structured form to form a list containing the execution order, corresponding window, and shielding constraints.

[0134] It should be noted that this list not only integrates the execution priorities and replacement strategies after arbitration, but also realizes an operational expression of complex regulatory tasks through timing and mutual exclusion rules.

[0135] Step S305: Distribute the executable list to the corresponding execution terminals according to the hierarchy to carry out multi-stage linkage control of the plant intelligent factory.

[0136] In this embodiment, after generating an executable list containing instruction sequence, execution window, and mutual exclusion rules, it needs to be distributed to the corresponding execution terminals according to their levels to achieve cross-level linkage control. Specifically, firstly, based on the preset level labels in the instruction list, device-level instructions are allocated to single-machine control modules, cultivation unit-level instructions are allocated to regional control nodes, and workshop-level instructions are allocated to the global scheduling center. Subsequently, the instructions are synchronously corrected during the distribution process to ensure that execution terminals at different levels receive and execute operations under the same time reference. Next, the execution interface is called within each execution terminal to convert the received atomic instructions into actual operation actions. For example, the device-level execution terminal adjusts the light power or water pump flow, the cultivation unit-level execution terminal adjusts the local nutrient solution concentration, and the workshop-level execution terminal uniformly controls the environmental temperature and humidity. Finally, the feedback information generated during the execution process is collected and recorded to form the input basis for closed-loop control.

[0137] It should be noted that this hierarchical distribution not only enables parallel linkage of multiple links, but also ensures consistency and traceability of control actions across the entire plant smart factory through upstream and downstream coordination.

[0138] Example 2

[0139] Please see Figure 4 Another embodiment of the present invention provides a multi-dimensional data monitoring and collaborative management system for intelligent plant factories, comprising: a data acquisition module, an impact path identification module, and a linkage control module;

[0140] The data acquisition module is used to collect multi-dimensional data of the plant intelligent factory in real time. The multi-dimensional data includes plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operation status parameters.

[0141] The influence path identification module, based on the multi-dimensional data, constructs a dynamic multi-level causal relationship model between equipment, plant physiology, environment and operation, and identifies potential collaborative influence paths under abnormal conditions.

[0142] The linkage control module is used to automatically perform cross-level control when an anomaly is detected or a deviation from the growth process is predicted. It distributes cross-level control instructions to the equipment level, cultivation unit level and workshop level according to the level, so as to carry out multi-link linkage control of the plant intelligent factory.

[0143] The linkage control module includes: an intent recognition unit, an instruction generation unit, and a linkage control unit;

[0144] The intent recognition unit determines the set of controlled objects and prohibited boundaries based on potential collaborative influence paths, forming a control intent that includes target state constraints, resource constraints, and temporal constraints.

[0145] The instruction generation unit is used to decompose the control action into atomic instruction sequences at the equipment level, cultivation unit level and workshop level according to the control intention, and generate complete instructions for cross-level scheduling and invocation.

[0146] The linkage control unit is used to distribute the executable list to the corresponding execution terminal according to the hierarchy, so as to carry out multi-stage linkage control of the plant intelligent factory.

[0147] 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.

[0148] 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 method for multi-dimensional data monitoring and collaborative management of a plant smart factory, characterized in that, The method comprises the following steps: Real-time acquisition of multi-dimensional data of a plant intelligent factory, including plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operation state parameters; Based on the multi-dimensional data, a dynamic multi-level causal correlation model between equipment, plant physiology, environment, and operation is constructed, and potential synergistic influence paths in abnormal states are identified; When an anomaly is detected or a growth process is predicted to deviate from a trend, automatic cross-level management and control is performed, cross-level management and control instructions are distributed to the equipment level, the cultivation unit level, and the workshop level according to the level, and multi-link joint regulation and control of the plant intelligent factory is performed; Based on the multi-dimensional data, a dynamic multi-level causal correlation model between equipment, plant physiology, environment, and operation is constructed, and potential synergistic influence paths in abnormal states are identified, comprising: The multi-dimensional data is hierarchically preprocessed according to the equipment parameter layer, the plant physiology layer, the environmental factor layer, and the operation layer, and the data in the same layer is normalized and time-synchronized; By establishing a multi-level feature index, key factors with potential coupling relationships between layers are identified, and a cross-layer correlation factor set is generated; According to the cross-layer correlation factor set, a time sequence interaction sequence between different level factors is constructed, which represents the factor sequence and transmission path in the dynamic change process; Based on the time sequence interaction sequence, the key factors of the equipment parameter layer, the plant physiology layer, the environmental factor layer, and the operation layer are mapped to a multi-level causal graph, and cross-layer action links are marked in the graph; In the multi-level causal graph, according to a preset path traversal logic, potential synergistic influence paths corresponding to abnormal states are extracted.

2. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 1, wherein, Based on the time sequence interaction sequence, the key factors of the equipment parameter layer, the plant physiology layer, the environmental factor layer, and the operation layer are mapped to a multi-level causal graph, and cross-layer action links are marked in the graph, comprising: The equipment factors, plant physiology factors, environmental factors, and operation factors in the time sequence interaction sequence are abstracted into graph nodes respectively, and each node is assigned a unique identifier; According to the time sequence dependency relationship between factors in the same level, intra-layer causal links are generated in the causal graph, which reflect the dynamic interaction logic between key factors in the same layer; Iterate the factor pairs with cross-layer time sequence dependency relationship, and establish cross-layer action links in the causal graph to mark the causal connection between different levels; The intra-layer causal links and cross-layer action links are structurally integrated to form a multi-level causal graph with hierarchical topological characteristics.

3. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 2, wherein, In the multi-level causal graph, according to a preset path traversal logic, potential synergistic influence paths corresponding to abnormal states are extracted, comprising: In the multi-level causal graph, according to the trigger information corresponding to the abnormal state, determine the nodes related to the abnormality as the traversal starting anchor point; Starting from the traversal starting anchor point, recursively according to the preset path traversal logic, sequentially access the successor nodes that have causal connection with the traversal starting anchor point to form a candidate path set; The causal consistency of the candidate path set is checked one by one, and paths inconsistent with the atlas logic or irrelevant to the abnormal state are excluded to generate effective paths meeting the constraint conditions; The effective paths are structured and collected to generate potential synergistic influence paths corresponding to the abnormal state.

4. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 1, wherein, When the abnormality is detected or the growth process is predicted to deviate from the trend, automatic cross-level management is performed, and cross-level management instructions are distributed to the device level, the cultivation unit level, and the workshop level according to the level, and multi-link joint regulation and control of the plant intelligent factory is performed, including: The multi-source trigger information generated by abnormality detection or growth process prediction is merged and processed to generate a cross-level management event, and the scope, time limit, and priority label of the cross-level management event are labeled; Based on the potential synergistic influence path, a set of controlled objects and a forbidden boundary are determined to form a management intent containing target state constraints, resource constraints, and timing constraints; According to the management intent, the management action is decomposed into atomic instruction sequences at the device level, the cultivation unit level, and the workshop level, the upstream and downstream dependency relationships and the execution window are established, and complete instructions for cross-layer scheduling are generated; The resource occupation, timing sequence, and safety mutual exclusion of the atomic instruction sequence are checked, and when mutual exclusion is detected, priority arbitration and alternative sequence reconstruction are performed to generate an executable list that meets the constraint conditions; The executable list is distributed to the corresponding execution end according to the level, and multi-link joint regulation and control of the plant intelligent factory is performed.

5. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 4, wherein, Based on the potential synergistic influence path, a set of controlled objects and a forbidden boundary are determined to form a management intent containing target state constraints, resource constraints, and timing constraints, including: In the potential synergistic influence path, the set of controlled objects that need to be implemented is determined according to the node causal relationship, and each controlled object is independently indexed; For the set of controlled objects, the forbidden boundary is set to limit the range of operations that cannot be triggered according to the conflict conditions existing in the cross-level action link; Combined with the set of controlled objects and the forbidden boundary, a control constraint set containing target state constraints, resource constraints, and timing constraints is generated; The control constraint set is structured and integrated to form a complete control intent.

6. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 5, wherein, According to the management intent, the management action is decomposed into atomic instruction sequences at the device level, the cultivation unit level, and the workshop level, the upstream and downstream dependency relationships and the execution window are established, and complete instructions for cross-layer scheduling are generated, including: The management intent is parsed into atomic instruction sequences at the device level, the cultivation unit level, and the workshop level, and each instruction is assigned a unique identifier; According to the trigger sequence and resource occupation conditions in the atomic instruction sequence, an upstream and downstream dependency relationship chain is established; Combined with the dependency relationship chain, the corresponding execution window is divided according to the time interval and parallelism constraint, and each instruction is mapped to the corresponding execution window; The mapped atomic instructions, their dependency relationships, and execution windows are serialized and collected to generate complete instructions for cross-layer scheduling.

7. The plant smart factory multi-dimensional data monitoring and collaborative management method of claim 6, wherein, The resource occupation, timing sequence, and safety mutual exclusion of the atomic instruction sequence are checked, and when mutual exclusion is detected, priority arbitration and alternative sequence reconstruction are performed to generate an executable list that meets the constraint conditions, including: The atomic instruction sequence is checked for resource occupation, timing sequence and safety mutual exclusion, and a verification result set containing conflict markers and trigger conditions is generated; The verification result set is mapped to an instruction dependency chain, and the conflicting instruction nodes and their upstream and downstream associated nodes are located, and the non-parallel domain and the rearrangeable domain are divided accordingly; In the non-parallel domain, priority arbitration is performed on related instructions according to control intention, prohibited boundary and execution window, and the instruction set of reserved instructions, delayed instructions and replacement instructions is determined; Under the constraints of the rearrangeable domain and the replacement instruction set, the instruction sequence that meets the dependency relationship and the execution window is reconstructed, and the executable manifest containing the instruction sequence, the execution window and the mutual exclusion shielding rule is output.

8. The plant intelligent factory multi-dimensional data monitoring and collaborative management system, used for realizing the plant intelligent factory multi-dimensional data monitoring and collaborative management method in any one of claims 1-7, characterized in that, It comprises: a data acquisition module, an influence path identification module, and a linkage control module; The data acquisition module is configured to acquire multi-dimensional data of the plant intelligent factory in real time, wherein the multi-dimensional data comprises plant physiological parameters, environmental factor parameters, nutrient solution parameters, and equipment and operation state parameters; The influence path identification module is configured to construct a dynamic multi-level causal association model among equipment, plant physiology, environment, and operation based on the multi-dimensional data, and identify potential synergistic influence paths in abnormal states; The linkage control module is configured to automatically perform cross-level control when an abnormality is detected or a growth process deviating trend is predicted, and distribute cross-level control instructions to the equipment level, the cultivation unit level, and the workshop level according to the level, and perform multi-linkage control of the plant intelligent factory. 9.The plant smart factory multi-dimensional data monitoring and collaborative management system of claim 8, wherein, The linkage control module comprises an intention identification unit, an instruction generation unit, and a linkage control unit; The intention identification unit is configured to determine a controlled object set and a prohibited boundary based on the potential synergistic influence path, and form a control intention containing target state constraints, resource constraints, and timing constraints; The instruction generation unit is configured to decompose control actions into atomic instruction sequences of the equipment level, the cultivation unit level, and the workshop level according to the control intention, and generate complete instructions for cross-level scheduling calls; The linkage control unit is configured to distribute executable manifests to corresponding execution ends according to the level, and perform multi-linkage control of the plant intelligent factory.

Images