Distributed control system and method for AI intelligent control cabinet

Abstract

The invention belongs to the technical field of distributed control, and discloses an AI intelligent control cabinet distributed control system and method. Comprising the steps of obtaining running state data and control task data; constructing a task priority vector of each intelligent control cabinet in each control period; generating a global task conflict graph based on the task priority vectors of all the intelligent control cabinets in the same control period; obtaining a task conflict probability and a task execution sequence of each intelligent control cabinet in the same control period according to the global task conflict graph; based on the task conflict probability and the task execution sequence, a control task execution strategy of each intelligent control cabinet is adjusted, and an optimized control task sequence of each intelligent control cabinet is obtained; and according to the optimization control task sequence, distributed cooperative operation of each intelligent control cabinet in the next control period is controlled, and the overall operation efficiency and the conflict occurrence rate of the distributed system can be effectively balanced.

Description

Technical Field

[0001] The present invention relates to the field of distributed control technology, and more specifically, to an AI smart control cabinet distributed control system and method. Background Art

[0002] As core equipment in power systems, intelligent control cabinets undertake key functions such as distribution automation, load control, and protection coordination. Their technological evolution is closely tied to the modernization of power grids. In the current smart grid construction, with the continuous advancement of ultra-high voltage transmission, large-scale grid interconnection, and intelligent distribution network development, traditional single-function distribution control cabinets have evolved into intelligent power control devices that integrate multiple functions, including remote measurement and control, relay protection, and fault diagnosis. In modern power grid architectures, control cabinets have gradually shifted from standalone operation to distributed, coordinated systems, forming a hierarchical distributed control system covering the entire power generation, transmission, transformation, distribution, and consumption process. However, with the widespread adoption of new technologies such as large-scale grid integration of new energy sources, load-side response, microgrid clusters, and flexible direct current transmission, traditional distributed control cabinet systems face severe challenges such as insufficient real-time performance, lack of coordination, and low intelligence.

[0003] Specifically, in large-scale regional power distribution networks, each intelligent control panel generally operates in a "single-point response" mode, lacking a coordinated linkage mechanism. This prevents flexible resource allocation during peak demand periods or sudden load changes, leading to local overloads and even cascading failures. In modern power grids, the intermittent and fluctuating nature of renewable energy sources like wind and photovoltaics makes traditional control systems inadequate. These controls are unable to accurately predict and smooth power fluctuations, resulting in reduced power quality and increased risks to system stability. Power dispatch centers lack a unified coordination mechanism for control panels distributed across distribution stations. When a grid disturbance or fault occurs, the protection actions of each control panel lack coordination, easily leading to protection mismatches and widespread faults. The fixed-priority response mechanism used by traditional power control systems is unable to cope with increasingly complex grid operating scenarios and is slow to adapt to new service requirements such as demand-side response and distributed generation scheduling. Furthermore, power equipment status monitoring and fault warnings rely on independent judgment by each control panel, lacking data sharing and joint analysis capabilities. This makes it difficult to detect equipment hazards in a timely manner, inefficient maintenance, and increases the risk of power outages and economic losses. In smart microgrids and industrial park power supply systems, power scheduling conflicts between multiple control cabinets often occur. However, existing technologies lack intelligent conflict resolution solutions, which often lead to uneven power distribution, energy waste, or fluctuations in power supply quality.

[0004] In view of this, the present invention proposes an AI intelligent control cabinet distributed control system and method to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned objectives, the present invention provides the following technical solutions: a distributed control method of an AI intelligent control cabinet, comprising:

[0006] Obtaining operating status data and control task data of each smart control cabinet in each control cycle; constructing a task priority vector of each smart control cabinet in each control cycle based on the operating status data and the control task data;

[0007] Generate a global task conflict graph based on the task priority vectors of all smart control cabinets in the same control cycle;

[0008] According to the global task conflict graph, the task conflict probability and task execution order of each smart control cabinet in the same control cycle are obtained;

[0009] Based on the task conflict probability and the task execution order, adjusting the control task execution strategy of each smart control cabinet to obtain an optimized control task sequence for each smart control cabinet;

[0010] According to the optimized control task sequence, each smart control cabinet is controlled to operate in a distributed collaborative manner in the next control cycle.

[0011] Preferably, said constructing of the task priority vector of each smart control cabinet in each control cycle includes: extracting the equipment load rate, operation stability index and fault risk value from said operation status data;

[0012] Extracting the task urgency, task complexity, and task dependency from the control task data; constructing a task feature matrix for each smart control cabinet in each control cycle based on the equipment load rate, the operation stability index, the fault risk value, the task urgency, the task complexity, and the task dependency;

[0013] The method for obtaining the operation stability index includes: obtaining a time series of the operation status data of each smart control cabinet in the current control cycle; performing Fourier transform on the time series to extract the frequency domain characteristics of the time series; calculating the fluctuation amplitude and periodic change rate of the time series based on the frequency domain characteristics; and taking the weighted sum of the fluctuation amplitude and the periodic change rate as the operation stability index.

[0014] A principal component analysis is performed on the task feature matrix to obtain the task priority vector of each smart control cabinet in each control cycle.

[0015] Preferably, generating a global task conflict graph includes:

[0016] Calculate the Euclidean distance between the task priority vectors of any two smart control cabinets according to the task priority vectors of all smart control cabinets in the same control cycle as the task competition degree of the corresponding two smart control cabinets;

[0017] Constructing a task conflict directed graph among all smart control cabinets according to the task competitiveness and the task dependency in the control task data;

[0018] The task conflict directed graph is topologically sorted to obtain the task conflict path of each smart control cabinet in the task conflict directed graph; the task conflict path is weightedly fused with the task competitiveness to generate a global task conflict graph.

[0019] Preferably, obtaining the task conflict probability and task execution order of each smart control cabinet in the same control cycle includes:

[0020] According to the global task conflict graph, extract the number of conflicting nodes with other smart control cabinets in the task conflict path of each smart control cabinet;

[0021] Calculate the task conflict probability of each smart control cabinet in the same control cycle according to the number of conflicting nodes and the task competition degree;

[0022] According to the task conflict path in the global task conflict graph, the task execution order of each smart control cabinet in the same control cycle is determined.

[0023] Preferably, determining the task execution order of each smart control cabinet within the same control cycle according to the task conflict path in the global task conflict graph includes:

[0024] Performing a depth-first traversal on each task conflict path in the global task conflict graph to generate a directed acyclic graph of task dependency relationships between the smart control cabinets;

[0025] Based on the task dependency directed acyclic graph, calculating the task criticality index of each smart control cabinet, wherein the task criticality index includes in-degree value, out-degree value and betweenness centrality;

[0026] Constructing a task execution priority matrix according to the task criticality indicator and the task priority vector;

[0027] applying a critical path algorithm to the task execution priority matrix to identify a task execution critical chain;

[0028] Allocate a task execution time window for each smart control cabinet according to the task execution critical chain and the time constraints in the control task data of each smart control cabinet;

[0029] Based on the task execution time window and the task conflict probability, an adaptive time slicing strategy is adopted to determine the final task execution order of each smart control cabinet within the same control cycle.

[0030] Preferably, the adjusting the control task execution strategy of each smart control cabinet to obtain the optimized control task sequence of each smart control cabinet includes:

[0031] Determine the task adjustment weight of each smart control cabinet within the same control cycle based on the task conflict probability; rearrange the control task data of each smart control cabinet based on the task execution order and the task adjustment weight to obtain a candidate control task sequence for each smart control cabinet; simulate the execution of the candidate control task sequence to obtain the overall system operation efficiency and conflict occurrence rate from the simulation execution results;

[0032] Based on the overall operating efficiency of the system and the conflict occurrence rate, the candidate control task sequence is optimized to obtain an optimized control task sequence for each smart control cabinet.

[0033] Preferably, the re-arranging of the control task data of each smart control cabinet includes:

[0034] Constructing a task description tuple for the control task data of each smart control cabinet, wherein the task description tuple includes a task identifier, an original priority, an execution time, and a resource requirement;

[0035] Applying the task adjustment weight to the original priority in the task description tuple to calculate a weighted priority value;

[0036] Constructing a sequence constraint matrix according to the task execution order to characterize the front-to-back execution relationship between tasks; constructing a priority weighted directed acyclic graph based on the weighted priority value and the sequence constraint matrix;

[0037] Applying a topological sorting algorithm on the priority-weighted directed acyclic graph to generate an initial sorting scheme; performing resource conflict detection on the initial sorting scheme to identify task pairs that compete for resources in time-overlapping intervals;

[0038] For the identified task pairs, conflict resolution is performed according to the weights of the respective tasks, and the relative execution positions are adjusted to obtain a sorting scheme after conflict resolution; the sorting scheme after conflict resolution is mapped to the time axis to generate a candidate control task sequence;

[0039] The candidate control task sequence is optimized for compactness to reduce idle waiting time between tasks while keeping all sequence constraints unchanged.

[0040] Preferably, the method for calculating the task competitiveness includes:

[0041] Obtaining the task priority vectors of any two smart control cabinets in the same control cycle; calculating the Euclidean distance between the task priority vectors as the initial competition degree;

[0042] According to the task dependency in the control task data, determine whether there is a direct dependency between the two smart control cabinets; if there is a direct dependency, perform a positive weighted adjustment on the initial competitiveness; if there is no direct dependency, perform a negative weighted adjustment on the initial competitiveness to obtain the task competitiveness.

[0043] Preferably, the optimizing the candidate control task sequence to obtain the optimized control task sequence for each smart control cabinet includes:

[0044] Constructing an optimization function with the goals of maximizing the overall operating efficiency of the system and minimizing the conflict occurrence rate; iteratively optimizing the candidate control task sequence using a genetic algorithm to obtain a global optimal solution that satisfies the optimization function;

[0045] The candidate control task sequence corresponding to the global optimal solution is used as the optimized control task sequence of each smart control cabinet.

[0046] An AI smart control cabinet distributed control system, used to implement the AI ​​smart control cabinet distributed control method, comprising: a distributed management center, the distributed management center connected to the smart control cabinet via a distributed communication network to achieve real-time transmission of control instructions and status data;

[0047] The distributed management center includes:

[0048] Data acquisition module, used to obtain the operating status data and control task data of each smart control cabinet in each control cycle;

[0049] A task priority analysis module, configured to construct a task priority vector for each smart control cabinet in each control cycle based on the operating status data and the control task data;

[0050] A conflict graph building module, configured to generate a global task conflict graph based on the task priority vectors of all smart control cabinets in the same control cycle;

[0051] A conflict analysis module is used to obtain the task conflict probability and task execution order of each smart control cabinet in the same control cycle according to the global task conflict graph;

[0052] A task scheduling module, configured to adjust the control task execution strategy of each smart control cabinet based on the task conflict probability and the task execution order, to obtain an optimized control task sequence for each smart control cabinet;

[0053] A collaborative control module, configured to control the distributed collaborative operation of each smart control cabinet in the next control cycle according to the optimized control task sequence;

[0054] The modules are connected to each other via wired and / or wireless means.

[0055] The technical effects and advantages of the AI ​​intelligent control cabinet distributed control system and method of the present invention are as follows:

[0056] The present invention improves the overall operational efficiency of the intelligent power system, and effectively avoids the common resource competition problems in traditional control systems through conflict prediction and coordination mechanisms. It achieves precise dynamic adjustment of task priorities, significantly reduces the system task conflict rate, and can still maintain relatively good response performance under high-load environments. The system resource utilization rate is improved, the energy-saving and consumption-reducing effects are obvious, and the system's ability to respond quickly to emergencies is enhanced. In addition, the present invention also has fault prediction and risk prevention and control capabilities. Through intelligent analysis and early warning, potential faults can be eliminated in the bud, thereby improving the safety and reliability of the power system. In addition, the adaptive learning characteristics of the present invention enable the system to continuously optimize the operating strategy, adapt to complex and changing operating environments, and achieve long-term stable and efficient operation. In practical applications, the need for manual intervention is significantly reduced, the operation and maintenance costs are reduced, and the service life of the equipment is increased. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a schematic diagram of a distributed control method for an AI smart control cabinet of the present invention;

[0058] Figure 2 This is a schematic diagram of an AI smart control cabinet distributed control system of the present invention. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] This invention proposes an AI intelligent control cabinet distributed control system and method, please refer to Figure 1 , which shows a schematic diagram of a distributed control method for an AI intelligent control cabinet provided by one embodiment of the present invention, the method comprising:

[0061] Step S1: Obtain the operating status data and control task data of each smart control cabinet in each control cycle.

[0062] Multiple smart control cabinets are deployed in a distributed system, each responsible for managing the power equipment, industrial equipment, or other intelligent control objects within its area. Through a built-in sensor network, the smart control cabinet collects real-time operating status data for itself and the equipment it manages, including but not limited to equipment load rate, operational stability indicators, and fault risk. It also receives control task data issued by the system dispatch center or upper-level control units, including but not limited to task urgency, task complexity, and task dependencies.

[0063] It should be noted that each smart control cabinet is equipped with an edge computing unit that can perform preliminary processing and analysis on the collected data and transmit the data to the central control system or other smart control cabinets via a distributed network. Smart control cabinets can communicate directly with each other, forming a distributed collaborative network.

[0064] In one implementation of the embodiment of the present invention, the control period is set to once every 5 seconds.

[0065] In one implementation of this embodiment, operational status data is standardized, using the Z-score method to map various indicators to the same scale for subsequent analysis. The urgency of tasks in the control task data is represented using a numerical scale of 1-10, with higher values ​​indicating more urgent tasks. Task complexity is represented using a numerical scale of 1-5, with higher values ​​indicating more complex tasks. Task dependencies are represented using a directed graph, with dependent tasks being executed before dependent tasks.

[0066] The following steps all use the processed operating status data and control task data for analysis.

[0067] Step S2: constructing a task priority vector for each smart control cabinet in each control cycle according to the operating status data and the control task data.

[0068] The task priority vector is a quantitative representation of the relative importance of each control task in a smart control cabinet. By extracting and fusion-analyzing features from operational status data and control task data, it compresses multi-dimensional task characteristics into a single vector representation, facilitating subsequent task coordination and conflict detection in a distributed environment. The dimensionality of the task priority vector depends on the type and number of features of the control task, typically encompassing multiple dimensions such as task urgency, impact on system stability, resource consumption, and dependency strength.

[0069] Preferably, in some possible implementation methods of the embodiments of the present invention, a task priority vector of each smart control cabinet in each control cycle is constructed, including: extracting the equipment load rate, operation stability index and fault risk value in the operating status data; extracting the task urgency, task complexity and task dependency in the control task data; and constructing a task feature matrix of each smart control cabinet in each control cycle based on the equipment load rate, operation stability index, fault risk value, task urgency, task complexity and task dependency.

[0070] The device load rate reflects the current load on the devices managed by the smart control cabinet, typically expressed as a percentage. When the device load rate is too high, adding new control tasks may cause device overload, affecting system stability. The operational stability index indicates the stability of the system in its current operating state. Higher stability indicates a greater ability to withstand external interference. The fault risk value predicts the likelihood of system failure in its current state. Higher risk values ​​indicate lower system security and require prioritization of related tasks.

[0071] The method for obtaining an operation stability index includes: obtaining a time series of the operation status data of each smart control cabinet in the current control cycle; performing Fourier transform on the time series to extract the frequency domain characteristics of the time series; calculating the fluctuation amplitude and periodic change rate of the time series based on the frequency domain characteristics; and taking the weighted sum of the fluctuation amplitude and the periodic change rate as the operation stability index.

[0072] The time series of operating status data reflects the evolution of the operating status of the smart control cabinet within the control cycle. By performing a Fourier transform on the time series, the time domain signal can be converted into a frequency domain signal, extracting implicit periodic features. Frequency domain features include power spectral density, main frequency components, and harmonic ratios. These features can reflect the stability and fluctuations of system operation. The fluctuation amplitude indicates the magnitude of the change in the system's operating status; greater fluctuations indicate poorer system stability. The periodic rate of change indicates the regularity of changes in the system's operating status; a higher rate of change indicates more regular system operation and better stability.

[0073] In one implementation of this embodiment of the present invention, the fluctuation amplitude and the periodic change rate are weighted as follows: Operational Stability Index = 0.6 × (1 - Normalized Fluctuation Amplitude) + 0.4 × Normalized Periodic Change Rate. Normalization maps the value to the interval [0, 1]. Smaller fluctuation amplitudes indicate higher stability, while larger periodic change rates indicate higher stability.

[0074] Task urgency reflects the time sensitivity of the control task. The more urgent the task, the faster it needs to be executed. Task complexity indicates the complexity and resource requirements of the control task. The more complex the task, the more difficult it is to execute. Task dependency describes the forward and backward execution constraints between control tasks. The more complex the dependency, the more difficult the task coordination.

[0075] In this embodiment of the present invention, a task feature matrix is ​​constructed by combining the equipment load rate, operational stability index, fault risk value, task urgency, task complexity, and task dependencies. Each row of the task feature matrix corresponds to a control task, and each column corresponds to a feature dimension. Since task dependencies are typically represented as graphs, they need to be converted into numerical features. The number of predecessor and successor dependent tasks for each task can be calculated as a numerical representation of the task dependencies.

[0076] Performing principal component analysis (PCA) on the task feature matrix can reduce feature dimensionality, extract the most representative feature combinations, and form a task priority vector. PCA preserves the primary variation in the data while reducing redundancy and noise, making subsequent task priority calculations more efficient and accurate.

[0077] In one implementation of the embodiment of the present invention, the first several principal components whose explained variance ratio reaches 85% in the principal component analysis result are retained to form a task priority vector.

[0078] Step S3: Generate a global task conflict graph of the distributed system based on the task priority vectors of all smart control cabinets in the same control cycle.

[0079] The global task conflict graph is a directed graph structure used to represent potential task execution conflicts between smart control cabinets in a distributed system. In a distributed system, control tasks in different smart control cabinets may impact each other, such as competing for shared resources or having execution dependencies. These potential conflicts can affect the overall stability and efficiency of the system. By constructing a global task conflict graph, task conflicts in the system can be intuitively represented and analyzed, providing a basis for subsequent task scheduling and optimization.

[0080] Preferably, in some possible implementation methods of the embodiments of the present invention, a global task conflict graph of a distributed system is generated, including: calculating the Euclidean distance of the task priority vectors between any two smart control cabinets according to the task priority vectors of all smart control cabinets in the same control cycle as the task competitiveness of the corresponding two smart control cabinets; constructing a task conflict directed graph between all smart control cabinets in the distributed system according to the task competitiveness and the task dependency in the control task data; topologically sorting the task conflict directed graph to obtain the task conflict path of each smart control cabinet in the task conflict directed graph; weightedly fusing the task conflict path with the task competitiveness to generate a global task conflict graph of the distributed system.

[0081] The task priority vector reflects the importance and urgency of the smart control cabinet's current control task. When two control cabinets have similar task priority vectors, they may need to execute tasks of similar importance at the same time, increasing the likelihood of task competition and conflict. Euclidean distance is a common method for measuring the similarity between two vectors. The smaller the distance, the higher the similarity, and the corresponding task competition is greater.

[0082] In this embodiment of the present invention, the Euclidean distance d(A, B) between the task priority vectors V_A and V_B of any two smart control cabinets A and B is calculated using the following formula:

[0083] Where V_A_i and V_B_i are the i-th components of vectors V_A and V_B respectively.

[0084] Since the larger the Euclidean distance, the greater the difference between the two vectors, and the task competitiveness is proportional to the vector similarity, it is necessary to convert the Euclidean distance. In this embodiment of the present invention, the calculation formula for task competitiveness C(A, B) is:

[0085] Through this conversion, the smaller the Euclidean distance d(A, B), the closer the task competitiveness C(A, B) is to 1; the larger the Euclidean distance d(A, B), the closer the task competitiveness C(A, B) is to 0.

[0086] When constructing a directed graph of task conflicts, in addition to considering task competition, we also need to consider the task dependencies within the control task data. Task dependencies reflect the constraints between the execution of tasks in different control cabinets. When task A depends on task B, task B must be completed before task A. This dependency also increases the likelihood of task conflicts.

[0087] In an embodiment of the present invention, the task conflict directed graph is a weighted directed graph, in which nodes represent intelligent control cabinets and edges represent task conflict relationships between control cabinets. The edge weights are determined by both the task competition and the task dependency strength. When a task dependency exists between two control cabinets, a directed edge is added between the corresponding nodes, with the direction of the edge pointing from the dependent task to the dependent task. When the task competition between the two control cabinets exceeds a preset threshold, an undirected edge (or a bidirectional edge) is also added between the corresponding nodes.

[0088] In one implementation of the embodiment of the present invention, the preset threshold of task competitiveness is set to 0.6.

[0089] Topologically sorting the task conflict directed graph yields a node sequence that satisfies all dependency constraints. While this topological sorting reflects the ideal execution order for smart control cabinet tasks, it may not completely avoid conflicts caused by task competition. Therefore, further analysis is needed to determine the position and connectivity of each smart control cabinet within the task conflict directed graph and identify its task conflict path.

[0090] Task conflict paths refer to all possible paths from a source node to a target node in the task conflict directed graph. They reflect the propagation and impact of task conflicts. For each smart control cabinet, all conflict paths where it serves as both a source and a target node must be considered. The more and longer conflict paths a smart control cabinet has, the more severe the conflict between that control cabinet's tasks and those of other control cabinets in the system, requiring more careful scheduling and optimization.

[0091] By weighting and fusing the task conflict path and the task competitiveness, a more comprehensive conflict assessment can be obtained. In an embodiment of the present invention, the global task conflict graph is based on the task conflict directed graph, and the weight of each edge is adjusted. The weight adjustment formula is:

[0092] W_new(A,B)=h×W_old(A,B)+(1-h)×P(A,B);

[0093] Where W_new(A,B) is the adjusted weight, W_old(A,B) is the original edge weight (task competitiveness), P(A,B) is the evaluation value of all conflicting paths from node A to node B, and h is the weight coefficient, which ranges from [0,1].

[0094] In one implementation of the embodiment of the present invention, h is set to 0.7. P(A, B) is calculated as a normalized value of the weighted sum of the number of paths and the path length.

[0095] Step S4: According to the global task conflict graph, the task conflict probability and task execution order of each smart control cabinet in the same control cycle are obtained.

[0096] By analyzing the topology and conflict intensity of the global task conflict graph, the probability of conflict between each smart control cabinet's tasks and those of other control cabinets is calculated, and a task execution order that minimizes conflict is determined. The task conflict probability reflects the likelihood of a task encountering resource contention or dependency conflicts during execution. The task execution order is the optimal execution path based on the conflict graph and the overall system efficiency.

[0097] Preferably, in some possible implementation methods of the embodiments of the present invention, the task conflict probability and task execution order of each smart control cabinet in the same control cycle are obtained, including: extracting the number of conflict nodes with other smart control cabinets in the task conflict path of each smart control cabinet according to the global task conflict graph; calculating the task conflict probability of each smart control cabinet in the same control cycle according to the number of conflict nodes and the task competitiveness; and determining the task execution order of each smart control cabinet in the same control cycle according to the task conflict path in the global task conflict graph.

[0098] Conflict nodes refer to other smart control cabinets that are directly connected (i.e., in a conflict relationship) to the target smart control cabinet in the global task conflict graph. A greater number of conflict nodes means that the tasks of that smart control cabinet conflict with more other control cabinets in the system, making task coordination more difficult and increasing the probability of task conflicts.

[0099] In this embodiment of the present invention, conflicting nodes include not only those directly connected to the target control cabinet, but also those indirectly connected via task conflict paths. By analyzing task conflict paths, it is possible to identify control cabinets that appear unrelated but may actually lead to conflicts through chain reactions. This analysis method considers the complex topology of the system and enables a more comprehensive assessment of potential task conflicts.

[0100] Based on the number of conflicting nodes and the task competition degree, the task conflict probability of each smart control cabinet is calculated. The number of conflicting nodes reflects the scope of the conflict, and the task competition degree reflects the intensity of the conflict. The combination of the two can more accurately assess the possibility of task conflict. In this embodiment of the present invention, the task conflict probability P is calculated as follows:

[0101] P=1-exp(-(z×N+(1-z)×C_avg));

[0102] Where N is the normalized number of conflicting nodes, C_avg is the average task competition degree of all edges connected to the control cabinet, z is the weight coefficient with a value range of [0,1], and exp represents the natural exponential function.

[0103] In one implementation of the present invention, z is set to 0.65. This calculation method ensures that when the number of conflicting nodes or the average task competition increases, the task conflict probability will also increase, but the growth rate will gradually slow down and approach but not exceed 1, which is consistent with the definition of probability.

[0104] Based on the task conflict paths in the global task conflict graph, the task execution order of each smart control cabinet within the same control cycle is determined. The ideal task execution order should satisfy all dependency constraints and minimize conflicts. In an embodiment of the present invention, a method based on critical path analysis is used to determine the task execution order: first, the global task conflict graph is converted into a task dependency directed acyclic graph (DAG); then, the earliest start time and latest end time of each node in the DAG are calculated; finally, based on the node's time constraints and conflict probability, the final task execution order is determined.

[0105] Preferably, in some possible implementation methods of the embodiments of the present invention, the task execution order of each smart control cabinet within the same control cycle is determined according to the task conflict path in the global task conflict graph, including: performing depth-first traversal on each task conflict path in the global task conflict graph to generate a directed acyclic graph of task dependencies between the smart control cabinets; calculating the task criticality indicators of each smart control cabinet based on the task dependency directed acyclic graph, the task criticality indicators include in-degree value, out-degree value and betweenness centrality; constructing a task execution priority matrix according to the task criticality indicators and the task priority vector; applying the critical path algorithm to the task execution priority matrix to identify the task execution critical chain in the distributed system; allocating a task execution time window to each smart control cabinet according to the task execution critical chain and the time constraints in the control task data of each smart control cabinet; and adopting an adaptive time slicing strategy based on the task execution time window and the task conflict probability to determine the final task execution order of each smart control cabinet within the same control cycle.

[0106] Depth-first traversal is a graph search algorithm that starts at a vertex in the graph, follows a path to the bottom, then backtracks to the previous node and continues along another path to the bottom until all vertices have been visited. By performing a depth-first traversal on each task conflict path in the global task conflict graph, all potential conflicting links can be identified and this information can be integrated into a directed acyclic graph (DAG) of task dependencies.

[0107] In an embodiment of the present invention, a depth-first traversal is implemented recursively. First, all edges in the global task conflict graph are sorted by weight (conflict intensity) from largest to smallest. Then, the traversal begins with the edge with the largest weight, ensuring that the path with the most severe conflict is prioritized. During the traversal, if a loop (circular dependency) is discovered, it is necessary to break the loop to ensure that the resulting graph is a directed acyclic graph. The principle of breaking the loop is to remove the edge with the smallest weight.

[0108] Based on a directed acyclic graph of task dependencies, the task criticality index of each smart control cabinet is calculated. This index reflects the importance and influence of the smart control cabinet in the distributed system task network and consists of three metrics: in-degree, out-degree, and betweenness centrality. The in-degree indicates the number of other control cabinets that depend on it, reflecting the leading influence of the control cabinet's tasks; the out-degree indicates the number of other control cabinets that it depends on, reflecting the subsequent dependencies of its tasks; and the betweenness centrality indicates the degree to which the control cabinet acts as a "bridge" in the entire network. The higher the betweenness centrality, the more critical the control cabinet's role in task coordination.

[0109] In the embodiment of the present invention, the calculation formula of betweenness centrality BC(v) is:

[0110] Where sig(s,t) represents the number of shortest paths from node s to node t, and sig(s,t|v) represents the number of shortest paths from node s to node t that must pass through node v.

[0111] Based on the task criticality indicators and task priority vectors, a task execution priority matrix is ​​constructed. The task execution priority matrix is ​​a multidimensional matrix that comprehensively considers the priority of the task itself and the impact of the system network, and is used to guide the determination of the task execution order.

[0112] Apply the critical path algorithm to the task execution priority matrix to identify the critical chain of task execution in a distributed system. The critical path is the longest path from the starting node to the ending node. The tasks along this path determine the shortest completion time for the entire project. In a distributed system, the critical chain of task execution is the sequence of key tasks that affects the overall execution efficiency of the system. Delays in tasks along the critical chain directly lead to delays in the entire system.

[0113] In this embodiment of the present invention, the critical path algorithm is implemented based on a task execution priority matrix and a directed acyclic graph of task dependencies. First, the earliest start time (ES) and earliest finish time (EF) of each node are calculated. Then, the latest start time (LS) and latest finish time (LF) of each node are calculated. Finally, the time margin (Slack = LS - ES) of each node is calculated. Nodes with a time margin of 0 constitute the critical path.

[0114] Based on the task execution critical chain and the time constraints in the control task data of each smart control cabinet, a task execution time window is assigned to each smart control cabinet. The time window refers to the time range within which a task is allowed to begin execution, including the earliest and latest start times. For tasks on the critical chain, the time window is more stringent, typically requiring execution near the earliest start time. For tasks outside the critical chain, the time window is more flexible and can be arranged flexibly within the allowed range.

[0115] In an embodiment of the present invention, the time window allocation strategy is based on the following principles: first, ensuring that tasks on the critical chain can be executed as planned; second, rationally utilizing the time margin of non-critical chain tasks to reduce resource conflicts; and finally, considering the dependencies between tasks to ensure that the order of task execution is correct.

[0116] Based on the task execution time window and task conflict probability, an adaptive time slicing strategy is used to determine the final task execution order for each smart control cabinet within the same control cycle. The adaptive time slicing strategy dynamically adjusts task execution time, automatically adjusting the size and allocation of task execution time slices based on the task conflict probability and system load.

[0117] In an embodiment of the present invention, the method for implementing the adaptive time slicing strategy is: first, the control cycle is divided into multiple time slices; then, based on the task conflict probability, a larger time slice or an exclusive time slice is allocated to the task with a high conflict probability, and a shared time slice is allocated to the task with a low conflict probability; finally, based on the constraints of the task execution time window, the specific execution time point and order of each task are determined.

[0118] In one implementation of the present invention, the time slice size is set to 1 / 20 of the control period, that is, 0.25 seconds (assuming the control period is 5 seconds). When the task conflict probability is greater than 0.8, an exclusive time slice is allocated; when the task conflict probability is between 0.5 and 0.8, a 50% overlapping time slice is allocated; when the task conflict probability is less than 0.5, a fully shared time slice is allocated.

[0119] Step S5: Based on the task conflict probability and the task execution order, the control task execution strategy of each smart control cabinet is adjusted to obtain an optimized control task sequence of each smart control cabinet.

[0120] Based on the probability of task conflicts and the order of task execution, the smart control cabinet's existing control task execution strategy is adjusted, including task priority rescheduling, execution time window allocation, and resource reservation, resulting in an optimized control task sequence. This optimized control task sequence considers not only the importance and urgency of the tasks themselves, but also the overall coordination of the system. This ensures correct task execution while minimizing task conflicts within the system.

[0121] Preferably, in some possible implementation methods of the embodiments of the present invention, the control task execution strategy of each smart control cabinet is adjusted to obtain an optimized control task sequence for each smart control cabinet, including: determining the task adjustment weight of each smart control cabinet in the same control cycle based on the task conflict probability; and rearranging the control task data of each smart control cabinet based on the task execution order and the task adjustment weight to obtain a candidate control task sequence for each smart control cabinet.

[0122] The task adjustment weight is an indicator that measures the adjustment range of the smart control cabinet task. The higher the task conflict probability, the greater the necessity of task adjustment, and the corresponding task adjustment weight is also greater. In this embodiment of the present invention, the calculation formula of the task adjustment weight W is:

[0123] W=k×P n ; Where P is the task conflict probability, k and n are adjustment parameters used to control the growth rate and range of the weight.

[0124] In one implementation of the present invention, k is set to 2 and n is set to 0.5. This setting allows the task adjustment weight to grow slowly when the probability of conflict is low, and to grow rapidly when the probability of conflict is high, thereby making larger adjustments to tasks with high conflict risks.

[0125] Simulate the execution of candidate control task sequences to obtain the overall system operational efficiency and conflict rate from the simulation results. Simulation is an important means of evaluating the effectiveness of task sequences before actual deployment. By building a digital twin model of the distributed system, the execution of the task sequence is simulated in a virtual environment, and key performance indicators are obtained to provide data support for subsequent optimization.

[0126] In this embodiment of the present invention, simulation execution utilizes a discrete event simulation approach, discretizing time within a control cycle into multiple event points, each corresponding to the start or end of one or more tasks. During the simulation, system state changes, including resource usage, task execution progress, and control instruction transmission delays, are tracked in real time to evaluate the performance of candidate control task sequences.

[0127] The overall system operating efficiency is a comprehensive indicator that measures the performance of distributed system tasks and reflects the system's ability to complete effective work within a unit of time. In this embodiment of the present invention, the calculation formula for the overall system operating efficiency η is:

[0128] Where Wi is the importance weight of the i-th task, Ci is the completion degree of the i-th task, T is the total execution time, and R is the total system resources. This formula takes into account three aspects: task completion, time utilization, and resource utilization, and can comprehensively reflect the operating efficiency of the system.

[0129] The conflict rate refers to the ratio of the number of actual task conflicts to the theoretical number of possible conflicts during the simulation execution process. It is an important indicator for evaluating the coordination of task sequences. In this embodiment of the present invention, the calculation formula of the conflict rate CR is:

[0130] Where Nc is the actual number of conflicts that occurred, and Np is the theoretical number of possible conflicts. The conflict count is based on three types of conflicts: resource contention, timing violation, and data consistency.

[0131] Preferably, random perturbations are introduced during the simulation to simulate communication delays, computational biases, and external interference that may occur in a real environment, enhancing the reliability and robustness of the simulation results. The magnitude of the random perturbations is determined based on the statistical characteristics of fluctuations in historical operational data to ensure the similarity between the simulation environment and the actual environment.

[0132] In one implementation of an embodiment of the present invention, the simulation is performed using the Monte Carlo method, and each candidate control task sequence is simulated 100 times. Different random disturbances are introduced in each simulation, and the average value is finally taken as the evaluation result of the overall operating efficiency and conflict occurrence rate of the system.

[0133] Based on the overall system operating efficiency and conflict rate, the candidate control task sequence is optimized to obtain the optimal control task sequence for each smart control cabinet. The optimization process aims to find a balance that ensures efficient system operation while minimizing task conflicts.

[0134] Preferably, in some possible implementation methods of the embodiments of the present invention, the candidate control task sequence is optimized to obtain the optimized control task sequence of each smart control cabinet, including: constructing an optimization function with the goal of maximizing the overall operating efficiency of the system and minimizing the conflict occurrence rate; using a genetic algorithm to iteratively optimize the candidate control task sequence to obtain a global optimal solution that satisfies the optimization function; and using the candidate control task sequence corresponding to the global optimal solution as the optimized control task sequence of each smart control cabinet.

[0135] An optimization function is a mathematical expression that evaluates the quality of candidate control task sequences and guides the optimization algorithm in its search for the optimal solution. In this invention, the optimization objectives include two aspects: maximizing the overall system operating efficiency and minimizing the occurrence of conflicts. These two objectives may conflict. For example, improving operating efficiency may increase the risk of conflicts, while reducing conflicts may reduce efficiency. Therefore, it is necessary to find a balance between the two.

[0136] In the embodiment of the present invention, the optimization function F is defined as:

[0137] F = a × E - (1 - a) × C;

[0138] Among them, E is the overall operating efficiency of the system (the value range is [0,1]), C is the conflict occurrence rate (the value range is [0,1]), and a is the weight coefficient (the value range is [0,1]), which is used to balance the relative importance of the two goals.

[0139] In one implementation of this embodiment, a is set to 0.7, indicating that the system prioritizes improving operational efficiency while also considering reducing conflicts. When a is set to 1, the system prioritizes operational efficiency; when a is set to 0, the system prioritizes reducing conflicts. By adjusting the value of a, the optimization objective can be flexibly adjusted based on actual needs.

[0140] A genetic algorithm is an optimization algorithm based on natural selection and genetic mechanisms, and is applicable to complex combinatorial optimization problems. In an embodiment of the present invention, a genetic algorithm is used to iteratively optimize a sequence of candidate control tasks, searching for a global optimal solution that satisfies an optimization function. The basic principle of a genetic algorithm is to simulate evolutionary operations such as natural selection, crossover, and mutation, starting from a set of initial solutions (population), gradually generate new solutions, and evaluate the quality of the solutions based on a fitness function (i.e., the optimization function). Solutions with high fitness are retained, and solutions with low fitness are eliminated. After multiple generations of evolution, the optimal solution in the population gradually approaches the global optimal solution.

[0141] In an embodiment of the present invention, the specific implementation of the genetic algorithm includes the following steps:

[0142] 1) Initialize the population: Generate a set of random candidate control task sequences as the initial population;

[0143] 2) Fitness evaluation: Calculate the optimization function value of each candidate solution as its fitness;

[0144] 3) Selection operation: Based on the size of fitness, the parent individuals are selected using methods such as roulette or tournament;

[0145] 4) Crossover operation: Pair the selected parent individuals with each other, exchange some of their task sequences, and generate offspring individuals;

[0146] 5) Mutation operation: Randomly mutate the task sequence of the offspring individuals with a certain probability to increase the diversity of the population;

[0147] 6) Update the population: According to fitness, select outstanding individuals from the parent and offspring generations to form a new generation of population;

[0148] 7) Termination condition: The algorithm terminates when the maximum number of iterations is reached or the fitness no longer improves significantly;

[0149] 8) Output result: Return the individual with the highest fitness in the final population as the global optimal solution.

[0150] In one implementation of the embodiment of the present invention, the population size is set to 100, the maximum number of iterations is set to 200, the crossover probability is set to 0.8, and the mutation probability is set to 0.1.

[0151] In the implementation of a genetic algorithm, special attention must be paid to the encoding and decoding of task sequences, as well as how task dependency constraints are handled. In this embodiment of the present invention, a task priority list is used for encoding, where tasks are sorted by priority. During decoding, tasks are scheduled for execution in this order, while also taking dependency constraints into account. For candidate solutions that violate dependency constraints, repair operations can be performed to bring them into compliance, or a larger penalty term can be imposed to reduce their fitness.

[0152] The candidate control task sequence corresponding to the global optimal solution is used as the optimized control task sequence for each smart control cabinet. The global optimal solution is the best task scheduling solution found under current conditions after iterative optimization using a genetic algorithm. It strikes the best balance between overall system efficiency and conflict rate. The optimized control task sequence includes information such as the tasks that each smart control cabinet should perform in the next control cycle, their execution order, and their start and completion times. This provides specific execution guidance for the distributed collaborative operation of smart control cabinets.

[0153] Preferably, in some possible implementation methods of the embodiments of the present invention, the control task data of each smart control cabinet are rearranged, including: constructing a task description tuple for the control task data of each smart control cabinet, the task description tuple containing a task identifier, original priority, execution time and resource requirement; applying the task adjustment weight to the original priority in the task description tuple to calculate a weighted priority value.

[0154] A sequence constraint matrix is ​​constructed according to the task execution order to characterize the forward and backward execution relationship between tasks; a priority-weighted directed acyclic graph is constructed based on the weighted priority values ​​and the sequence constraint matrix; a topological sorting algorithm is applied to the priority-weighted directed acyclic graph to generate an initial sorting scheme that satisfies the dependency constraints; resource conflict detection is performed on the initial sorting scheme to identify task pairs that compete for resources in the time overlapping interval; for the identified task pairs, the weights are adjusted according to their respective tasks to resolve conflicts and adjust the relative execution positions to obtain a sorting scheme after conflict resolution; the sorting scheme after conflict resolution is mapped to the time axis to generate a candidate control task sequence containing a precise execution time window; the candidate control task sequence is optimized for compactness to reduce the idle waiting time between tasks while keeping all sequence constraints unchanged.

[0155] A task description tuple is a structured representation of control task data, containing key attribute information about the task. A task identifier uniquely identifies a task, its original priority indicates its initial importance, its execution duration indicates the time required to complete the task, and its resource requirements indicate the system resources required during task execution. By constructing task description tuples, complex control task data can be converted into a unified structured format, facilitating subsequent processing and analysis.

[0156] In an embodiment of the present invention, the task description tuple is in the form of: Task = (ID, Pr, Dur, Res), where ID is the task identifier, Pr is the original priority (ranging from 1 to 10), Dur is the execution duration (in seconds), and Res is a vector of resource requirements, indicating the task's requirements for various types of resources.

[0157] The task adjustment weight is applied to the original priority in the task description tuple to calculate the weighted priority value. The weighted priority value is a modified value based on the original priority after taking into account the task adjustment weight, reflecting the actual importance of the task under the current system state. In this embodiment of the present invention, the calculation formula of the weighted priority value WP is:

[0158] WP=Pr×(1+W×f(P));

[0159] Where Pr is the original priority, W is the task adjustment weight, P is the task conflict probability, and f(P) is the adjustment function of the conflict probability, which is used to control the impact of the conflict probability on the priority.

[0160] In one implementation of the embodiment of the present invention, f(P)=P 2 ,This setting makes the impact of the conflict probability on the priority grow like a quadratic function, which has a small impact when the conflict probability is low, and significantly increases when the conflict probability is high.

[0161] A sequence constraint matrix is ​​constructed based on the task execution order to characterize the execution relationship between tasks. The sequence constraint matrix is ​​a two-dimensional matrix, and the elements in the matrix represent the execution relationship between tasks. In this embodiment of the present invention, the sequence constraint matrix S is defined as:

[0162] S[i,j]=1, if task i must be executed before task j;

[0163] S[i,j]=0, ​​if there is no direct constraint relationship between task i and task j;

[0164] S[i,j]=-1, if task j must be executed before task i;

[0165] Based on the weighted priority values ​​and the sequence constraint matrix, a priority-weighted directed acyclic graph (DAG) is constructed. A priority-weighted DAG is a special directed graph in which nodes represent tasks, edges represent dependencies between tasks, and node weights are the weighted priorities of the tasks. This graph structure preserves the dependencies between tasks while also taking into account their actual importance, providing a foundation for subsequent topological sorting.

[0166] A topological sorting algorithm is applied to a priority-weighted directed acyclic graph to generate an initial sorting scheme that satisfies dependency constraints. Topological sorting is an algorithm that arranges all nodes in a directed acyclic graph into a linear sequence, such that for each edge (u, v) in the graph, node u appears before node v in the sequence. In an embodiment of the present invention, a topological sorting algorithm based on depth-first search (DFS) is used, and the specific steps are as follows:

[0167] 1) Perform a depth-first search on all nodes in the graph;

[0168] 2) During the search process, when all adjacent nodes of a node have been visited, the node is added to the result sequence;

[0169] 3) The final sequence obtained after reversal is the result of topological sorting.

[0170] The initial sorting scheme satisfies all dependency constraints, but may still have resource conflicts. Therefore, we need to perform resource conflict detection on the initial sorting scheme to identify task pairs that compete for resources within the time overlap interval. The basic principle of resource conflict detection is that if two tasks have overlapping execution times and require the same resources, then there is a resource conflict between the two tasks.

[0171] In an embodiment of the present invention, the specific steps of resource conflict detection include: first, calculating the expected start time and end time of each task based on the initial sorting scheme and the task execution duration; then, for each pair of tasks that may be executed simultaneously, checking whether their resource requirements have an intersection; finally, marking the task pairs with overlapping resource requirements and overlapping execution times as conflicting task pairs.

[0172] For each identified task pair, conflict resolution is performed by adjusting the weights of the respective tasks and adjusting their relative execution positions to obtain a conflict-resolved sequencing solution. The goal of conflict resolution is to eliminate resource conflicts by adjusting the execution order or execution time of tasks while maintaining the original task priorities and dependencies as much as possible.

[0173] In an embodiment of the present invention, the conflict resolution strategy is based on task adjustment weights: for a pair of conflicting tasks, the task with the smaller task adjustment weight is given priority, that is, its execution time is postponed; if the adjustment weights of two tasks are similar, their weighted priority values ​​are considered, and the task with the lower priority is given priority; if the conflict still cannot be resolved, the overall impact is considered and an adjustment plan with less impact on the system is selected.

[0174] The conflict-resolved sequencing scheme is then time-mapped to generate a candidate control task sequence containing precise execution time windows. Time-mapped tasks are converted from their relative execution order to specific time points, including the calculation of each task's specific start and end times. In this embodiment of the present invention, time-mapped tasks utilize a forward scheduling approach, starting from the start of the control cycle and sequentially scheduling the execution times of each task according to their dependencies and resource constraints.

[0175] In one implementation of an embodiment of the present invention, the specific steps of timeline mapping include: first, setting the task with in-degree 0 (no predecessor dependency) as the earliest startable time; then, calculating the start time of other tasks one by one in the order of topological sorting, and the start time of a task is equal to the maximum end time of all its predecessor tasks; finally, the end time of the task is equal to the start time plus the execution duration.

[0176] Compactness optimization is performed on candidate control task sequences to reduce idle waiting time between tasks while maintaining all sequence constraints unchanged. The goal of compactness optimization is to improve system resource utilization and reduce idle time during task execution, thereby improving the overall execution efficiency of the system. In this embodiment of the present invention, compactness optimization adopts a forward shift strategy, that is, tasks are moved forward as much as possible without violating dependency constraints and resource constraints, reducing the gaps between tasks.

[0177] In one implementation of this invention, the specific steps for compactness optimization include: first, calculating the earliest possible start time and the latest required start time for each task; then, within this time window, attempting to move the task forward until it can no longer be moved; and finally, checking resource constraints to ensure that the move will not cause new resource conflicts. This approach minimizes idle waiting time between tasks, improving system compactness and execution efficiency.

[0178] Preferably, in some possible implementation methods of the embodiments of the present invention, the method for calculating the task competitiveness includes: obtaining the task priority vectors of any two smart control cabinets in the same control cycle; calculating the Euclidean distance between the task priority vectors as the initial competitiveness; determining whether there is a direct dependency between the two smart control cabinets based on the task dependency in the control task data; if there is a direct dependency, performing a positive weighted adjustment on the initial competitiveness; if there is no direct dependency, performing a negative weighted adjustment on the initial competitiveness to obtain the task competitiveness.

[0179] The task priority vector is a quantitative representation of the current control task priority of the smart control cabinet. The dimension and specific meaning of the vector are determined by the results of principal component analysis. By comparing the task priority vectors of different control cabinets, we can assess the similarity and degree of competition between them.

[0180] In this embodiment of the present invention, the task priority vectors V_A and V_B of any two smart control cabinets A and B in the same control cycle are obtained, where the dimension of the vector is n. The initial competition degree can be obtained by calculating the Euclidean distance d(V_A, V_B) between the two vectors.

[0181] The Euclidean distance is the straight-line distance between two points in n-dimensional space. The smaller the Euclidean distance, the more similar the two vectors are, and the corresponding initial competitiveness is greater; the larger the Euclidean distance, the less similar the two vectors are, and the corresponding initial competitiveness is smaller.

[0182] In the embodiment of the present invention, the formula for converting the Euclidean distance into the initial competitiveness is:

[0183]

[0184] Thus, when the Euclidean distance is 0, the initial competitiveness is 1 (maximum value); when the Euclidean distance approaches infinity, the initial competitiveness approaches 0 (minimum value).

[0185] Based on the task dependencies in the control task data, determine whether there is a direct dependency between two smart control cabinets. A direct dependency means that the execution of a task in one control cabinet depends on the completion of a task in another control cabinet. When a direct dependency exists between two control cabinets, their task execution must be coordinated; otherwise, system performance may degrade or task failure may occur.

[0186] In an embodiment of the present invention, by analyzing the dependency graph in the control task data, it is determined whether there is a directly connected edge between two smart control cabinets. If so, it is considered that there is a direct dependency relationship between them.

[0187] If a direct dependency exists, the initial competitiveness is adjusted positively; if not, a negative adjustment is made to the initial competitiveness to obtain the final task competitiveness. A positive adjustment increases the initial competitiveness, reflecting that dependencies increase the need for task coordination and the potential for conflict. A negative adjustment decreases the initial competitiveness, reflecting that conflicts between tasks without dependencies are less likely.

[0188] In the embodiment of the present invention, the weighted adjustment formula is:

[0189] If there is a direct dependency: C = In_C × (1 + d);

[0190] If there is no direct dependency: C = In_C × (1-d);

[0191] Where d is the adjustment coefficient, and its value range is [0,1].

[0192] In one implementation of the present invention, d is set to 0.3, which means that when a direct dependency exists, the task competitiveness increases by 30% compared to the initial competitiveness; when no direct dependency exists, the task competitiveness decreases by 30% compared to the initial competitiveness.

[0193] Step S6: According to the optimized control task sequence, each smart control cabinet is controlled to operate in a distributed collaborative manner in the next control cycle to avoid control conflicts and execution sequence errors.

[0194] The optimized control task sequence is distributed to each smart control cabinet to guide its operation during the next control cycle. Through the distributed system's communication network, each smart control cabinet can perceive the execution status of other control cabinets in real time and coordinate its own task execution according to the optimized control task sequence, thus achieving distributed collaborative operation of the entire system. At the same time, each smart control cabinet continuously collects new operating status data, providing basic data for optimization of the next control cycle.

[0195] Preferably, in some possible implementation methods of the embodiments of the present invention, controlling the distributed collaborative operation of each smart control cabinet in the next control cycle includes: generating a control instruction sequence for each smart control cabinet in the next control cycle based on the optimized control task sequence; sending the control instruction sequence to the corresponding smart control cabinet through the communication network of the distributed system; each smart control cabinet executes the corresponding control task according to the received control instruction sequence, and feeds back the operating status data to the distributed system in real time to update the global task conflict graph of the next control cycle.

[0196] A control instruction sequence is a specific execution form of an optimized control task sequence, containing detailed operational commands that the smart control cabinet needs to execute. In an embodiment of the present invention, a control instruction sequence consists of the following components: an instruction identifier, an execution timestamp, an operation type, operation parameters, and a completion condition. The instruction identifier uniquely identifies an instruction, the execution timestamp specifies the execution time of the instruction, the operation type defines the specific behavior of the instruction (such as starting a device, adjusting parameters, shutting down a device, etc.), the operation parameters provide the specific parameter values ​​required for instruction execution, and the completion condition defines the criteria for determining whether the instruction has been executed.

[0197] In an embodiment of the present invention, the process of generating a control instruction sequence includes: first, extracting the execution time and task attributes of each task from the optimized control task sequence; then, querying the instruction template library based on the task attributes to obtain the corresponding operation type and operation parameter format; finally, filling in specific parameter values ​​and execution timestamps to generate a complete control instruction sequence.

[0198] In one implementation of the embodiment of the present invention, the control instruction is expressed in JSON format.

[0199] The control command sequence is sent to the corresponding smart control cabinet via the distributed system's communication network. The distributed system's communication network serves as the data transmission channel connecting each smart control cabinet to the central control unit, typically using industrial Ethernet, wireless networks, or proprietary communication protocols. In this embodiment of the present invention, the communication network utilizes a secure, low-latency, and highly reliable communication mechanism to ensure that control commands are accurately and promptly transmitted to each smart control cabinet.

[0200] In this embodiment of the present invention, the communication protocol uses standard Industrial Internet of Things protocols such as MQTT or OPCUA, supporting both publish / subscribe and request / response models, adapting to communication needs in different scenarios. Data encryption and identity authentication mechanisms are used during the communication process to ensure security and reliability.

[0201] Each smart control cabinet executes the corresponding control task based on the sequence of control commands it receives. The smart control cabinet has a built-in control command parsing engine and task execution manager, which parses received control commands and executes the corresponding operations at the specified time and sequence. During execution, the smart control cabinet monitors the execution status of tasks to ensure they proceed as planned and handle any anomalies that may arise.

[0202] In this embodiment of the present invention, the smart control cabinet's task execution manager utilizes a real-time operating system and a priority queue management mechanism to handle concurrent multi-task execution and prioritized task scheduling. For high-priority or time-critical tasks, the system allocates more computing resources and a higher execution priority to ensure timely completion.

[0203] While executing control tasks, the smart control cabinet also provides real-time operational status data to the distributed system to update the global task conflict graph for the next control cycle. This operational status data includes task execution progress, resource usage, and device operating parameters. This data is crucial for evaluating system performance and optimizing task scheduling.

[0204] In this embodiment of the present invention, operational status data is fed back using a combination of scheduled reporting and event-triggered reporting. Scheduled reporting refers to reporting basic status data at regular intervals (e.g., 1 second); event-triggered reporting refers to immediate reporting of relevant data upon the occurrence of important events (e.g., task completion, resource shortages, abnormal conditions, etc.). This approach ensures real-time status data while reducing unnecessary communication overhead.

[0205] The distributed system updates the global task conflict graph based on received operational status data, providing a basis for decision-making regarding task scheduling and optimization for the next control cycle. This update process is a dynamic adjustment process. Based on the latest status data and execution results, the system adjusts edge weights and node attributes in the conflict graph to ensure that the graph accurately reflects the current system's task conflict status.

[0206] In this embodiment of the present invention, the global task conflict graph is updated using an incremental update strategy, updating only the changed parts rather than rebuilding the entire conflict graph, thereby improving the system's responsiveness and efficiency. During the update process, the system will focus on tasks whose execution does not meet expectations, analyze the causes of the conflicts, and adjust them in the next optimization.

[0207] The above describes an AI smart control cabinet distributed control method in the embodiment of the present application. The following describes an AI smart control cabinet distributed control system in the embodiment of the present application. The system is used to implement the above AI smart control cabinet distributed control method; please refer to Figure 2, an embodiment of an AI smart control cabinet distributed control system in the embodiment of the present application includes: a distributed management center, the distributed management center is connected to the smart control cabinet via a distributed communication network to achieve real-time transmission of control instructions and status data;

[0208] The distributed management center includes:

[0209] Data acquisition module, used to obtain the operating status data and control task data of each smart control cabinet in each control cycle;

[0210] The task priority analysis module is used to construct the task priority vector of each smart control cabinet in each control cycle based on the operating status data and control task data;

[0211] A conflict graph construction module is used to generate a global task conflict graph for the distributed system based on the task priority vectors of all smart control cabinets in the same control cycle;

[0212] The conflict analysis module is used to obtain the task conflict probability and task execution order of each smart control cabinet in the same control cycle based on the global task conflict graph;

[0213] A task scheduling module is used to adjust the control task execution strategy of each smart control cabinet based on the task conflict probability and the task execution order to obtain an optimized control task sequence for each smart control cabinet;

[0214] The collaborative control module is used to control the distributed collaborative operation of each smart control cabinet in the next control cycle according to the optimized control task sequence;

[0215] The modules in the distributed management center are connected via wired and / or wireless means to achieve data transmission between modules.

[0216] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will be able to modify the technical solutions described in the foregoing embodiments or to substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0217] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0218] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0219] In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0220] In the description of the present invention, “several” means one or more, and “a large number” means two or more.

[0221] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0222] The formulas in this manual are all dimensionless and calculated using numerical values. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field based on actual conditions.

[0223] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A distributed control method for AI intelligent control cabinet, characterized in that: include: Obtain the operating status data and control task data of each smart control cabinet in each control cycle; Constructing a task priority vector for each smart control cabinet in each control cycle according to the operating status data and the control task data; Generate a global task conflict graph based on the task priority vectors of all smart control cabinets in the same control cycle; According to the global task conflict graph, the task conflict probability and task execution order of each smart control cabinet in the same control cycle are obtained; Based on the task conflict probability and the task execution order, adjusting the control task execution strategy of each smart control cabinet to obtain an optimized control task sequence for each smart control cabinet; According to the optimized control task sequence, each smart control cabinet is controlled to operate in a distributed collaborative manner in the next control cycle.

2. The distributed control method of an AI intelligent control cabinet according to claim 1 is characterized in that: The constructing of the task priority vector of each smart control cabinet in each control cycle includes: extracting the equipment load rate, operation stability index and fault risk value from the operation status data; Extracting the task urgency, task complexity, and task dependency from the control task data; constructing a task feature matrix for each smart control cabinet in each control cycle based on the equipment load rate, the operation stability index, the fault risk value, the task urgency, the task complexity, and the task dependency; The method for obtaining the operation stability index includes: obtaining a time series of operation status data of each smart control cabinet in the current control cycle; performing Fourier transform on the time series to extract the frequency domain characteristics of the time series; calculating the fluctuation amplitude and periodic change rate of the time series based on the frequency domain characteristics; and taking the weighted sum of the fluctuation amplitude and the periodic change rate as the operation stability index; A principal component analysis is performed on the task feature matrix to obtain the task priority vector of each smart control cabinet in each control cycle.

3. The distributed control method of an AI intelligent control cabinet according to claim 1 is characterized in that: Generating a global task conflict graph includes: Calculate the Euclidean distance between the task priority vectors of any two smart control cabinets according to the task priority vectors of all smart control cabinets in the same control cycle as the task competition degree of the corresponding two smart control cabinets; Constructing a task conflict directed graph among all smart control cabinets according to the task competitiveness and the task dependency in the control task data; The task conflict directed graph is topologically sorted to obtain the task conflict path of each smart control cabinet in the task conflict directed graph; the task conflict path is weightedly fused with the task competitiveness to generate a global task conflict graph.

4. The distributed control method of an AI intelligent control cabinet according to claim 1 is characterized in that: The obtaining of the task conflict probability and task execution order of each smart control cabinet in the same control cycle includes: According to the global task conflict graph, extract the number of conflicting nodes with other smart control cabinets in the task conflict path of each smart control cabinet; Calculate the task conflict probability of each smart control cabinet in the same control cycle according to the number of conflicting nodes and the task competition degree; According to the task conflict path in the global task conflict graph, the task execution order of each smart control cabinet in the same control cycle is determined.

5. The distributed control method of an AI intelligent control cabinet according to claim 4 is characterized in that: Determining the task execution order of each smart control cabinet within the same control cycle according to the task conflict path in the global task conflict graph includes: Performing a depth-first traversal on each task conflict path in the global task conflict graph to generate a directed acyclic graph of task dependency relationships between the smart control cabinets; Based on the task dependency directed acyclic graph, calculating the task criticality index of each smart control cabinet, wherein the task criticality index includes in-degree value, out-degree value and betweenness centrality; Constructing a task execution priority matrix according to the task criticality indicator and the task priority vector; applying a critical path algorithm to the task execution priority matrix to identify a task execution critical chain; Allocate a task execution time window for each smart control cabinet according to the task execution critical chain and the time constraints in the control task data of each smart control cabinet; Based on the task execution time window and the task conflict probability, an adaptive time slicing strategy is adopted to determine the final task execution order of each smart control cabinet within the same control cycle.

6. The distributed control method of an AI intelligent control cabinet according to claim 1 is characterized in that: The control task execution strategy of each smart control cabinet is adjusted to obtain an optimized control task sequence for each smart control cabinet, including: Determine the task adjustment weight of each smart control cabinet within the same control cycle based on the task conflict probability; rearrange the control task data of each smart control cabinet based on the task execution order and the task adjustment weight to obtain a candidate control task sequence for each smart control cabinet; simulate the execution of the candidate control task sequence to obtain the overall system operation efficiency and conflict occurrence rate from the simulation execution results; Based on the overall operating efficiency of the system and the conflict occurrence rate, the candidate control task sequence is optimized to obtain an optimized control task sequence for each smart control cabinet.

7. The distributed control method of an AI intelligent control cabinet according to claim 5 is characterized in that: The task rearrangement of the control task data of each smart control cabinet includes: Constructing a task description tuple for the control task data of each smart control cabinet, wherein the task description tuple includes a task identifier, an original priority, an execution time, and a resource requirement; Applying the task adjustment weight to the original priority in the task description tuple to calculate a weighted priority value; Constructing a sequence constraint matrix according to the task execution order to characterize the front-to-back execution relationship between tasks; constructing a priority weighted directed acyclic graph based on the weighted priority value and the sequence constraint matrix; Applying a topological sorting algorithm on the priority-weighted directed acyclic graph to generate an initial sorting scheme; performing resource conflict detection on the initial sorting scheme to identify task pairs that compete for resources in time-overlapping intervals; For the identified task pairs, conflict resolution is performed according to the weights of the respective tasks, and the relative execution positions are adjusted to obtain a sorting scheme after conflict resolution; the sorting scheme after conflict resolution is mapped to the time axis to generate a candidate control task sequence; The candidate control task sequence is optimized for compactness to reduce idle waiting time between tasks while keeping all sequence constraints unchanged.

8. The distributed control method of an AI intelligent control cabinet according to claim 3 is characterized in that: The method for calculating the task competitiveness includes: Obtaining the task priority vectors of any two smart control cabinets in the same control cycle; calculating the Euclidean distance between the task priority vectors as the initial competition degree; According to the task dependency in the control task data, determine whether there is a direct dependency between the two smart control cabinets; if there is a direct dependency, perform a positive weighted adjustment on the initial competitiveness; if there is no direct dependency, perform a negative weighted adjustment on the initial competitiveness to obtain the task competitiveness.

9. The distributed control method of an AI intelligent control cabinet according to claim 5 is characterized in that: The optimizing the candidate control task sequence to obtain the optimized control task sequence for each smart control cabinet includes: Constructing an optimization function with the goals of maximizing the overall operating efficiency of the system and minimizing the conflict occurrence rate; iteratively optimizing the candidate control task sequence using a genetic algorithm to obtain a global optimal solution that satisfies the optimization function; The candidate control task sequence corresponding to the global optimal solution is used as the optimized control task sequence of each smart control cabinet.

10. An AI intelligent control cabinet distributed control system, characterized in that: A distributed control method for an AI smart control cabinet for implementing any one of claims 1 to 9, comprising: a distributed management center, the distributed management center being connected to the smart control cabinet via a distributed communication network to achieve real-time transmission of control instructions and status data; The distributed management center includes: Data acquisition module, used to obtain the operating status data and control task data of each smart control cabinet in each control cycle; A task priority analysis module, configured to construct a task priority vector for each smart control cabinet in each control cycle based on the operating status data and the control task data; A conflict graph building module, configured to generate a global task conflict graph based on the task priority vectors of all smart control cabinets in the same control cycle; A conflict analysis module is used to obtain the task conflict probability and task execution order of each smart control cabinet in the same control cycle according to the global task conflict graph; A task scheduling module, configured to adjust the control task execution strategy of each smart control cabinet based on the task conflict probability and the task execution order, to obtain an optimized control task sequence for each smart control cabinet; A collaborative control module, configured to control the distributed collaborative operation of each smart control cabinet in the next control cycle according to the optimized control task sequence; The modules are connected to each other via wired and / or wireless means.

Images