A fishery equipment group distributed cooperative operation process modeling and dynamic control method

Abstract

This invention discloses a method for modeling and dynamically controlling a distributed collaborative operation process for a group of fishery equipment, belonging to the field of smart fisheries and multi-agent collaborative control technology. Addressing the challenges of significant monsoon climate, variable sea conditions, and difficulties in coordinating heterogeneous equipment in deep-sea aquaculture, this invention constructs an environmentally coupled generalized stochastic Petri net model that includes an environmental constraint library and a sea state inhibition arc, embedding real-time sea state data into the model's topology. Color token coloring and guard functions are used to achieve differentiated equipment management based on wavekeeping levels. A dynamic nonlinear function for transition triggering rates is established based on wave spectrum parameters. Environmental deadlocks are identified and trigger risk avoidance and retreat transitions through a state reachability graph. The collaborative process is modeled as a semi-Markov decision process to achieve multi-objective optimization. This invention significantly improves the operational safety and continuity of heterogeneous equipment groups under complex sea conditions.

Description

Technical Field

[0001] This invention belongs to the field of smart fisheries, marine engineering equipment automation and multi-agent collaborative control technology, and in particular relates to a method for modeling and dynamic control of distributed collaborative operation processes of fishery equipment groups. Background Technology

[0002] Offshore large-scale aquaculture facilities are typically deployed in open sea areas with deep waters and strong currents. The production operations of these facilities rely on the coordinated operation of a series of intelligent equipment. However, existing collaborative operation control technologies for fishery equipment groups face the following significant technical bottlenecks when dealing with complex sea conditions: 1. Low environmental coupling and lack of sea state adaptability: Traditional collaborative control methods are mostly based on deterministic environmental assumptions, using static timetables or simple logical interlocking mechanisms to plan operational processes. However, the impact of actual sea states on equipment operational capabilities exhibits significant nonlinear characteristics: small unmanned vessels (such as 3-5m class transfer unmanned vessels) experience increased hull sway and significantly reduced maneuverability when wave heights exceed 1.5 meters, potentially leading to a complete loss of operational capability; while large fishing vessels (such as those over 200 tons), with their greater draft and stable hull structure, can still maintain a certain level of operational efficiency at wave heights of 2.5 meters. Existing models, such as traditional Petri net modeling methods, while introducing random delays, do not directly embed physical sea state parameters such as wave height, period, wind speed, and current velocity as process constraints into the model structure, relying solely on empirically set fixed thresholds for simple judgment. This leads to situations where, in actual operations, when sea states change rapidly beyond the equipment's limits, the scheduling system may still issue operational commands, easily causing equipment capsizing, collisions, or operational failures.

[0003] 2. The "weakest link" effect is significant in the collaboration of heterogeneous equipment: Deep-sea aquaculture operations have long chains and many links, involving multiple collaborative scenarios such as "harvesting-grading", "harvesting-transfer", "transfer-landing", "feeding-monitoring", and "net cleaning-inspection". The equipment involved in the collaboration is diverse in type and function, including high-efficiency harvesting equipment (fish suction pump boats, purse seine harvesting boats), multimodal transfer equipment (unmanned transfer boats, refrigerated transport boats), relay landing equipment (underwater lifting devices, docking platforms), underwater inspection robots (ROV / AUV), and intelligent feeding equipment (fixed-point feeding boats, drone feeding systems). These equipment vary greatly in physical size, dynamic characteristics, wave resistance, and endurance: large aquaculture vessels can withstand waves of more than 4 meters, while small unmanned surface vessels (USVs) typically have a wave resistance of only 1.5-2.0 meters; underwater inspection robots can operate stably in complex flow fields, while aerial feeding drones are significantly affected by wind speed. The existing unified scheduling model lacks fine-grained differentiation of the seakeeping level and environmental adaptability of individual equipment. The "one-size-fits-all" scheduling strategy is prone to causing the entire collaborative process to be blocked or deadlocked due to the inability of individual "weak" equipment to operate due to sea conditions, which seriously affects the efficiency of operations.

[0004] 3. Lack of Dynamic Replanning and Risk Avoidance Mechanisms: Traditional distributed auction / negotiation mechanisms (such as contract network protocols) focus on instantaneous optimal task allocation, achieving distributed decision-making through local communication between equipment for task bidding and negotiation. However, they lack long-term considerations for the "safety" of the entire process. In rapidly changing weather scenarios (such as sudden cold waves, severe convective weather outbreaks, and storm surge precursors), the system lacks a process circuit breaker mechanism to instantly switch from "operation mode" to "risk avoidance mode." When sudden severe sea conditions occur, traditional methods often only allow for manual intervention to terminate operations, failing to achieve a smooth transition and dynamic reconfiguration of the operational process. This can not only lead to operational failures but also safety accidents due to untimely equipment evacuation. Furthermore, existing methods struggle to balance operational efficiency and safety risks. During the edge window of sea state fluctuations, either excessive conservatism leads to wasted operational time, or reckless advancement causes safety hazards.

[0005] 4. Insufficient adaptation of multi-objective optimization to actual scenarios: The optimization objectives of collaborative operation of deep-sea aquaculture equipment groups are multi-dimensional, including the shortest operation time, the lowest energy consumption, the lowest safety risk, and the least equipment wear and tear. However, existing methods often only focus on the optimization of a single objective (such as pursuing only operation efficiency), ignoring the coupling and trade-offs between multiple objectives.

[0006] In summary, existing technologies are insufficient to meet the actual needs of collaborative operations of fishery equipment groups under complex sea conditions. There is an urgent need to develop a collaborative operation modeling and control method that deeply integrates environmental perception, has dynamic reconstruction capabilities, and supports multi-objective optimization, so as to provide core technical support for the stable operation of "unmanned fishing grounds" in deep sea. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method for modeling and dynamically controlling a distributed collaborative operation process of a fishery equipment group, comprising the following steps: An environmentally coupled generalized stochastic Petri net model is constructed based on real-time sea state data, which includes an environmental constraint library and a sea state suppression arc. The environmental constraint library is used to map sea state levels, and the sea state suppression arc is used to suppress operational change triggering based on the sea state level. The tokens are colored according to the wave-keeping properties of the heterogeneous equipment, and the matching relationship between the wave-keeping properties and the current sea state level is determined by the guard function to determine the currently enabled set of operational transitions. For the enabled operational transition set, a dynamic nonlinear function of the transition trigger rate is established based on the wave spectrum parameters. The real-time sea state data is mapped to random time delay parameters, and the random time delay parameters are used to control the random triggering process of the enabled operational transition. Based on the state changes generated by the random triggering process, a state reachability graph is dynamically constructed to identify environmental deadlock states and trigger risk avoidance and retreat transitions to resolve the environmental deadlock. The state change process is modeled as a semi-Markov decision process. A state space is constructed based on the state reachability graph, and an action space is constructed based on the enabling transition set. The optimal transition trigger sequence is searched in the state space and action space based on the comprehensive reward function. The real-time identifiers of the Petri net model are mapped to physical equipment control commands based on the optimal transition trigger sequence, and a rolling time-domain replanning is triggered when a step change in sea state is detected.

[0008] Optionally, the construction of an environmentally coupled generalized stochastic Petri net model, which includes an environmental constraint repository and a sea state suppression arc, based on real-time sea state data includes: Perform multi-source data fusion based on shipborne sensor data and public meteorological service data; The combined sea state index for the current time and the future forecast time domain is calculated based on the fused data; The comprehensive sea state index is quantified into a number of tokens and injected into the environmental constraint library.

[0009] Optionally, determining the matching relationship between the seakeeping attribute and the current sea state level through the guard function includes: Determine the conditions for switching operating modes based on the effective wave height threshold; The state of the mutex semaphore is controlled according to the job mode switching conditions to manage serial or parallel access permissions to the shared job space. Adjust the trigger frequency of feeding changes based on the feeding intensity feedback factor; Based on the relationship between the current sea state level and the fish feeding activity threshold, the triggering of feeding changes is paused through a sea state inhibition arc.

[0010] Optionally, the step of establishing the dynamic nonlinear function for the transition trigger rate based on wave spectrum parameters includes: The operational efficiency attenuation coefficient is calculated based on the significant wave height, spectral peak period, and wind speed. The transition trigger rate is determined based on the operation efficiency attenuation coefficient.

[0011] Optionally, the step of identifying the environmental deadlock state and triggering a risk avoidance and rollback transition to resolve the environmental deadlock includes: The risk avoidance retreat transition is triggered based on the storm warning status in the environmental constraint database. Based on the triggering of the aforementioned risk avoidance and retreat transition, a mission abort command is broadcast to the relevant equipment; Based on the current location of each piece of equipment, a path plan is generated for sheltered anchorage or deep-sea diving to avoid waves; Based on the triggering of the risk avoidance and withdrawal transition, the tokens of the relevant operation warehouses are cleared using the reset arc; The system is guided to enter a safe standby state based on the status of the emptied warehouse.

[0012] Optionally, the step of searching for the optimal transition trigger sequence based on the comprehensive reward function includes: The state space is refined based on the Cartesian product of the Petri net reachability set and the environment state; The comprehensive reward function is constructed based on operation time, energy consumption, and sea state risk penalties. The optimal transition trigger sequence is searched based on the state space, action space, and comprehensive reward function.

[0013] Optionally, triggering rolling temporal replanning upon detecting a step change in sea state includes: The rolling time-domain replanning is triggered based on a step change in physical sea state monitoring data or a predicted window closing event. The workflow is reconstructed based on the results of the rolling time-domain replanning, enabling dynamic switching from production mode to risk-avoidance mode.

[0014] Optionally, mapping the real-time identifiers of the Petri net model to physical equipment control commands based on the optimal transition trigger sequence includes: The real-time identifiers of the Petri net model are mapped to equipment control commands based on the digital twin runtime engine. The status markers of the environmental constraint library are updated in real time based on the data fusion results of shipborne sensors and shore-based radar.

[0015] On the other hand, the present invention also provides an electronic device including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0016] On the other hand, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0017] Compared with the prior art, the present invention has the following advantages and technical effects: This invention fundamentally solves the problems of low environmental coupling and inability to respond to sea state changes in traditional methods by constructing an environment-coupled generalized stochastic Petri net model, embedding sea state parameters such as effective wave height and spectral peak period into environmental constraint repositories and sea state suppression arcs. It achieves differentiated access control based on equipment seakeeping level through color token coloring and guard functions, overcoming the "barrel effect" of process blockage caused by the limitation of "weakest link" equipment in heterogeneous equipment collaboration. By using a dynamic reachability graph to predict environmental deadlock in real time and trigger the highest priority risk avoidance retreat transition across the entire network, combined with reset arc clearing operation tokens and windbreak anchorage path planning, it achieves millisecond-level dynamic reconstruction from production mode to risk avoidance mode, compensating for the lack of long-term safety considerations and process circuit breaker mechanisms in traditional methods. Through multi-objective optimization using a semi-Markov decision process and risk-sensitive reward functions, combined with precise JONSWAP wave spectrum modeling and embedded fisheries safety regulations, the optimization strategy is deeply adapted to the actual scenario. Compared with existing technologies, this invention effectively prevents safety accidents such as equipment overturning and collisions in harsh sea conditions, and significantly improves the utilization rate of effective operating windows in medium and high sea conditions, providing core technical support for the stable operation of deep-sea "unmanned fishing grounds". Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a diagram illustrating the overall modeling architecture of the Environment-Coupled Generalized Stochastic Petri Net (EC-GSPN) according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the EC-GSPN architecture according to an embodiment of the present invention.

[0019] Figure 3This is a dynamic adjustment curve of the transition trigger rate based on JONSWAP wave spectrum parameters according to an embodiment of the present invention. Figure 4 This is a flowchart of the dynamic control logic of an embodiment of the present invention; Figure 5 The following is a comparison chart of simulation results from an embodiment of the present invention. Detailed Implementation

[0020] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0022] Example 1 This embodiment provides a method for modeling and dynamically controlling a distributed collaborative operation process of a fishery equipment group, including the following steps: Based on real-time sea state data, an environmentally coupled generalized stochastic Petri net model is constructed, which includes an environmental constraint library and a sea state suppression arc. The environmental constraint library is used to map the sea state level, and the sea state suppression arc is used to suppress the triggering of operational changes according to the sea state level. The tokens are colored according to the seakeeping properties of the heterogeneous equipment, and the matching relationship between the seakeeping properties and the current sea state level is determined by the guard function to determine the set of currently enabled operational transitions. For the enabled operational transition set, a dynamic nonlinear function of the transition trigger rate is established based on the wave spectrum parameters. Real-time sea state data is mapped to random time delay parameters, and the random time delay parameters are used to control the random triggering process of the enabled operational transition. Based on the state changes generated by the random triggering process, a state reachability graph is dynamically constructed to identify environmental deadlock states and trigger risk avoidance and retreat transitions to resolve environmental deadlocks. The state change process is modeled as a semi-Markov decision process. The state space is constructed based on the state reachability graph, the action space is constructed based on the enabling transition set, and the optimal transition trigger sequence is searched in the state space and action space based on the comprehensive reward function. The real-time identifiers of the Petri net model are mapped to physical equipment control commands based on the optimal transition trigger sequence, and a rolling time-domain replanning is triggered when a step change in sea state is detected.

[0023] Furthermore, based on real-time sea state data, an environmentally coupled generalized stochastic Petri net model is constructed, including an environmental constraint repository and a sea state suppression arc. Perform multi-source data fusion based on shipborne sensor data and public meteorological service data; The combined sea state index for the current time and the future forecast time domain is calculated based on the fused data; The comprehensive sea state index is quantified into tokens and injected into the environmental constraint library.

[0024] Specifically, a distributed collaborative operation model for fishery equipment clusters is constructed based on generalized stochastic Petri nets. The model introduces "environmental constraint repository" and "sea state suppression arc" on the basis of traditional GSPN. ECP is used to map the effective wave height (SWH), average wind speed and current speed data of the operation sea area in real time, and the number of tokens in its internal tokens corresponds to the discretized sea state level. SSIA connects ECP with specific operation transitions. When environmental parameters exceed the preset threshold, the triggering of the corresponding transition is prohibited through SSIA, thereby constraining the unsafe operation behavior of equipment under high sea states at the physical and logical level and preventing the issuance of operation commands.

[0025] The data update of the "Environmental Constraints Library" adopts a multi-source data fusion and sliding window filtering mechanism; combined with measured data from shipborne wave buoys ( The gridded forecast data from public meteorological services (such as ERA5 reanalysis data or NMEFC forecasts) are used to calculate the comprehensive sea state index for the current time and the future forecast time domain (1-3 hours). The index is then rounded down and quantified into tokens and injected into the ECP, thereby realizing the transformation from reactive control to predictive control.

[0026] like Figure 1 As shown, this embodiment constructs an environment-sensitive formal model structure (EC-GSPN): Environmental Constraint Places (ECP): Building upon the traditional GSPN, a special type of "environmental place" is added. The number of tokens within these places does not represent the quantity of physical resources, but rather maps to the current sea state level through a quantification algorithm. Using the Douglas sea state level standard, sea states 0-9 are mapped to 0-9 tokens, achieving a discretized representation of continuous environmental parameters. Specifically, ECP includes wave height (Pwave), wind speed (Pwind), and current speed (Pcurrent), corresponding to the quantified values ​​of the three core environmental parameters—significant wave height, average wind speed, and average current speed—of the operating sea area, respectively.

[0027] Sea State Inhibitor Arc (SSIA): An inhibitor arc is introduced to connect the ECP (Extended Control Plan) and the operational transition. The inhibitor arc has a weight attribute, representing the maximum sea state level allowed for that operational transition. When the number of tokens in the ECP is greater than or equal to the inhibitor arc weight, the inhibitor arc is activated, forcibly preventing the transition from occurring. For example, the "unmanned vessel departure" transition is connected to an inhibitor arc with a weight of 3 (corresponding to sea state 3, significant wave height 1.25-2.5m). Therefore, when sea state ≥ 3, this action is physically "locked" and cannot be triggered. Through this mechanism, the environment's rigid constraints on the operational process are implemented at the model topology level.

[0028] Wave spectrum parameterization of transition rates: The JONSWAP spectrum (Joint North Sea Wave Project spectrum) is used to describe the wind and wave characteristics of the Yellow and Bohai Seas. This spectrum accurately reflects the concentrated wave energy and steep spectral peaks under the influence of the monsoon in this sea area. A nonlinear regression model is established for operation time and significant wave height (Hs), spectral peak period (Tp), and wind speed (Uwind). The random triggering rate λ of timed transitions is defined as a function of real-time sea state λ(t) = f(Hs(t), Tp(t), Uwind(t)), reflecting the nonlinear decay of operation efficiency under severe sea conditions. For example, the rate of navigation transition decreases with increasing wave height and increases with increasing spectral peak period, accurately simulating the speed reduction effect caused by wave drag.

[0029] Multi-source data fusion and environmental prediction are as follows: The Environmental Constraints Database (ECP) employs a multi-source data fusion and sliding window filtering mechanism for data updates, integrating three types of data sources: shipborne sensor data (measured data collected by wave buoys, anemometers, and current meters), shore-based radar monitoring data (long-range sea state monitoring data), and public meteorological service data (ERA5 reanalysis data and NMEFC refined grid forecast data). Kalman filtering is used to reduce noise in the multi-source data, removing outliers and improving data reliability. A sliding window (with a window size of 5 minutes) is then used to calculate the mean, yielding the current comprehensive sea state parameters. Finally, a time-series prediction model (such as LSTM) is used to predict sea states for the next 1-3 hours, generating a sea state prediction sequence to provide data support for predictive control.

[0030] The quantification of sea state levels adopts an adaptive threshold adjustment mechanism, which dynamically adjusts the environmental parameter thresholds corresponding to each sea state level based on the sea state characteristics of the Yellow and Bohai Seas in different seasons (such as frequent cold waves in winter and occasional typhoons in summer). For example, the threshold for sea state level 3 with significant wave height in winter can be adjusted to 1.2m-2.4m, and in summer it can be adjusted to 1.3m-2.6m, ensuring that the sea state level classification matches the actual marine environmental characteristics.

[0031] Furthermore, determining the matching relationship between seakeeping properties and the current sea state level through the guard function includes: Determine the conditions for switching operating modes based on the effective wave height threshold; The state of the mutex semaphore is controlled according to the job mode switching conditions to manage serial or parallel access permissions for the shared job space. Adjust the trigger frequency of feeding changes based on the feeding intensity feedback factor; Based on the relationship between the current sea state level and the fish feeding activity threshold, the triggering of feeding changes is paused through a sea state inhibition arc.

[0032] Specifically, using the Colored Petri Net (CPN) theory, a token attribute set is defined: ; in, The equipment types are identified, including fishing vessels, multimodal transport equipment, relay landing equipment, underwater unmanned monitoring equipment, and underwater inspection robots. Define the maximum seakeeping rating of this single piece of equipment; the transition triggering conditions in the model include a guard function, which is invoked only when the token... Returns the true value when the sea state is greater than or equal to the real-time sea state level in the current ECP, thus enabling differentiated access control based on equipment capabilities.

[0033] When this method is applied to a collaborative "capture-grading-transfer" scenario, a mutually exclusive semaphore place (SemaphorePlace) is constructed to manage the shared workspace; when the effective wave height... When the model automatically disables parallel operation transitions and forces a switch to serial operation mode, the transfer vessel is only allowed to enter for transfer after the catching vessel has completed the catching and left the operation area, in order to reduce the risk of collision in bad sea conditions.

[0034] The modeling method also includes special control logic for the "intelligent feeding equipment": a "feeding intensity feedback" factor is introduced into the ECP, and the trigger frequency of feeding changes is dynamically adjusted in combination with the output of the feeding intensity judgment software; when the sea state level exceeds the fish feeding activity threshold, feeding is suspended by the inhibition arc to prevent feed loss and water pollution.

[0035] In this embodiment, the refined management and control of heterogeneous equipment (based on CPN) is as follows: Using the concept of Color Petri Net (CPN), the token representing the equipment is assigned a color attribute (Color Set). The attribute set is defined as follows: ; in: Used as a unique identifier for equipment; Equipment type; To equip the equipment with the highest seaworthiness rating; Real-time equipment status (standby, underway, in operation, malfunction); Remaining battery level. Set a guard function on the input arc of the transition. For example, transition... The guard function is: ; This function comprehensively assesses whether the equipment's seakeeping capability, remaining battery power, and real-time status meet the current operational requirements. Only when all conditions are met can the transition be enabled. This mechanism ensures that equipment with poor seakeeping capability (such as small unmanned baiting vessels) automatically ceases operation in adverse sea conditions, while large work vessels with strong seakeeping capability can continue to perform their tasks. This achieves differentiated scheduling based on equipment capabilities and solves the "weakest link" problem in the collaboration of heterogeneous equipment.

[0036] Furthermore, the dynamic nonlinear function for the transition trigger rate is established based on the wave spectrum parameters, including: The operational efficiency attenuation coefficient is calculated based on the significant wave height, spectral peak period, and wind speed. The transition trigger rate is determined based on the operation efficiency attenuation coefficient.

[0037] Specifically, the trigger rate of the timed transition. Defined as a dynamic nonlinear function of environmental parameters ,in For the effective wave height, For the spectral peak period, The wind speed was used as the reference value; the operational efficiency decay was fitted using the parameters of the characteristic wave spectrum (JONSWAP spectrum) of the Yellow and Bohai Seas, so that the time evolution of the model can truly reflect the risk of operational delays or interruptions caused by severe sea conditions.

[0038] Dynamic rate function The specific format is as follows: in, The standard operating rate in a still water environment. and These are the extreme operating wave height and wind speed thresholds for the equipment, respectively. This is the sensitivity coefficient for the operation; it is for precision docking operations (such as "capture-transfer"). The value is greater than 1 to reflect the characteristic that the difficulty of operations increases exponentially under high sea states.

[0039] Furthermore, identifying environmental deadlock states and triggering risk avoidance and rollback transitions to resolve environmental deadlocks includes: The risk avoidance retreat transition is triggered based on the storm warning status in the environmental constraint database. Based on the triggering of the risk avoidance and withdrawal change, a mission abort instruction is broadcast to the relevant equipment; Based on the current location of each piece of equipment, a path plan is generated for sheltered anchorage or deep-sea diving to avoid waves; Based on the triggering of the risk avoidance and pullback transition, the tokens of the relevant operation warehouses are cleared using the reset arc; The system is guided to enter a safe standby state based on the status of the emptied warehouse.

[0040] Specifically, based on the real-time updated ECP status, a state reachability graph is dynamically constructed to identify "environmental deadlock" states caused by sudden changes in sea state. The environmental deadlock refers to a state in which the system has no resource conflicts but the critical path is blocked due to environmental constraints. A high-priority transition of "risk avoidance retreat" is introduced, which is forcibly triggered when deadlock risk is detected or a weather warning is received. The relevant work storage tokens are cleared using the Reset Arc, guiding the system into a safe standby state.

[0041] The "Distress Retreat" transition has the highest trigger priority across the entire network; its input arc is connected to the "Storm Warning" status bit in the ECP; when this transition is triggered, the system immediately interrupts all current non-atomic operations, broadcasts the "Abort" command to all relevant equipment, and generates the nearest safe anchorage or deep-sea diving shelter path based on the current location of each piece of equipment.

[0042] In this embodiment, deadlock detection and intelligent resolution are specifically as follows: Dynamic Reachability Graph Construction: The system periodically (every 5 minutes) generates a time-varying reachability graph based on the current ECP status and equipment status combination as the initial identifier, combined with future sea state prediction sequences. The nodes of the reachability graph represent the system status (including ECP Token distribution and equipment Token distribution), and the edges represent transition trigger events, comprehensively covering the possible future operating paths of the system.

[0043] Dual-type deadlock identification: Simultaneously searching for traditional resource deadlocks and environmental deadlocks in the reachability graph: Traditional resource deadlocks refer to deadlocks caused by competition for resources such as equipment and workspace; environmental deadlocks refer to deadlocks caused by sea state exceeding limits, which inhibits critical changes and prevents the operation from proceeding. By traversing all non-terminal nodes with an out-degree of 0 in the reachability graph, the deadlock state is accurately identified, and the critical path that led to the deadlock is traced back.

[0044] Tiered resolution strategy: Differentiated resolution solutions are adopted for different types of deadlocks: For traditional resource deadlocks, resolution is achieved by adjusting equipment task priorities and releasing non-critical resources; for environmental deadlocks, a high-priority "risk avoidance retreat" transition is triggered, and the tokens of related warehouses such as "operating" and "awaiting docking" are cleared using a reset arc, forcibly guiding the equipment to a "safe standby" state. Simultaneously, based on the equipment's current location, remaining endurance, and distribution of safe anchorages, A... The path planning algorithm generates the optimal risk avoidance path to ensure the safe evacuation of equipment.

[0045] Furthermore, the search for the optimal transition trigger sequence based on the comprehensive reward function includes: The state space is refined based on the Cartesian product of the Petri net reachability set and the environmental state; A comprehensive reward function is constructed based on operation time, energy consumption, and sea state risk penalties; The optimal transition trigger sequence is searched based on the state space, action space, and comprehensive reward function.

[0046] Specifically, the collaborative operation process is modeled as a semi-Markov decision process (SMDP), defining the state space as the Cartesian product of the reachable identifier set of Petri net and the environmental state, and the action space as the set of enabling transitions; a comprehensive reward function is constructed that includes operation time, energy consumption, and sea state risk penalties; and deep reinforcement learning algorithms (such as DQN or PPO) are used to search for the optimal transition trigger sequence in the state space to generate a disturbance-resistant dynamic scheduling strategy.

[0047] The specific optimizations of dynamic reprogramming and reinforcement learning are as follows: The collaborative work process is modeled as a semi-Markov decision process (SMDP). The state space consists of Petri net markings and environmental states.

[0048] Design a risk-sensitive reward function: ; in It is a nonlinear risk penalty term based on wave height.

[0049] The algorithm uses a Deep Q-Network (DQN) or PPO algorithm to search the state reachability graph of the Petri net to find the optimal operation path under the current sea state forecast. For example, if strong winds are predicted in the next 2 hours, the algorithm will automatically choose the strategy of "accelerating the completion of the current atomic task and immediately retreating" instead of "starting a new long-running task".

[0050] Implantation of constraints in accordance with Chinese fisheries safety regulations: The model incorporates hard constraints that comply with the "Maritime Traffic Safety Law of the People's Republic of China" and the wind resistance standards for fishing vessels.18 For example, it sets legally mandated sea state thresholds for departure and operation for fishing vessels of different tonnages to ensure the legality and compliance of scheduling strategies.

[0051] Furthermore, triggering rolling temporal replanning upon detecting a step change in sea state includes: Rolling time-domain replanning is triggered by step changes in physical sea state monitoring data or by predicted window closing events. The workflow is restructured based on the results of rolling time-domain replanning to achieve dynamic switching from production mode to risk avoidance mode.

[0052] Furthermore, mapping the real-time identifiers of the Petri net model to physical equipment control commands based on the optimal transition trigger sequence includes: The real-time identifiers of the Petri net model are mapped to equipment control commands based on the digital twin runtime engine. The status markers of the environmental constraint library are updated in real time based on the data fusion results of shipborne sensors and shore-based radar.

[0053] Specifically, a digital twin-driven runtime engine is established to map the real-time identifiers of the Petri net model to control commands for the physical equipment group, such as navigation, net deployment, and docking; the ECP markers in the model are updated in real time through the fusion of shipborne sensor and shore-based radar data; and the model's dynamic replanning mechanism is triggered when physical sea state monitoring data undergoes a step change or when the predicted future window closes.

[0054] On the other hand, this embodiment also provides an electronic device, including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0055] On the other hand, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0056] Example 2 This embodiment is designed for a deep-sea aquaculture scenario in the Yellow and Bohai Seas as follows: A specific EC-GSPN model was constructed for deep-sea aquaculture scenarios in the Yellow and Bohai Seas, as follows: 1. Formal definition of the model: The EC-GSPN defined in this invention is a nine-tuple: ; The elements are defined as follows: (Placement Set): Divided into two types of subsets, ; For resource and status databases, for example: "Fishing vessel idle" ( "Net cage ready" "In transit" ).

[0057] As an environmental constraint repository, this embodiment includes (High wave level) and (Wind speed rating). Number of Tokens It corresponds to the current effective wave height level in real time. For example, if $H_s \in.

[0058] (Guardian Function Set): Performs logical judgments on the attributes of colored tokens. For each type of equipment token... Define the guard function This ensures that only equipment whose seakeeping meets the requirements of the current sea state can be assigned tasks.

[0059] 2. Yellow and Bohai Sea Wave Spectrum Parameter Mapping: Considering the characteristics of the Yellow and Bohai Sea waves, which are mainly influenced by monsoons and have a sharp, narrow spectral shape, this embodiment uses the JONSWAP spectrum to map timing variations. The delay parameters are adjusted. Let the standard operation time be... Actual working time Obtain the parameter as The exponential distribution. Rate The calculation formula is as follows: ; in, To monitor the effective wave height in real time, For the spectral peak period, This is the "wave damping coefficient" obtained by fitting a large amount of historical operational data. This formula reflects the physical fact that the higher the wave height and the shorter the period (the steeper the wave), the more violent the ship's rolling, resulting in lower operational efficiency (such as lifting speed and sailing speed).

[0060] like Figure 3 As shown in the figure, the formula visually illustrates the nonlinear decay trend of normalized operational efficiency (i.e., transition trigger rate) with increasing significant wave height under different spectral peak periods using a graph. The figure clearly reflects the physical law that the operational efficiency decreases more drastically when the waves are steeper (shorter period), verifying the accuracy of this model in simulating operational efficiency decay under complex sea conditions in the Yellow and Bohai Seas.

[0061] like Figure 2 The "capture-transfer" collaborative operation process control shown is as follows: This embodiment describes how to use the method of the present invention for dynamic control in a "capture-transfer" collaborative scenario (control flow as follows). Figure 4 (As shown).

[0062] Scene description: Equipment: 1 high-efficiency fishing vessel (Type A, 3.5m seakeeping), 2 multi-modal unmanned transfer vessels (Type B, 2.0m seakeeping), and 1 large truss net cage (fixed facility).

[0063] Environment: Cold water mass area in the northern Yellow Sea. Current sea state: SWH=1.8m, predicted to increase to 2.8m in 1 hour due to strong winds.

[0064] Control process: Initialization: ECP library The current token count is 3 (representing sea state 3, corresponding to 1.25m-2.5m). Fishing vessel token. and transshipment ship token Located in the "standby" warehouse.

[0065] Task distribution: The system generates a "capture task".

[0066] Changes in fishing boats The suppression arc threshold is 5 (corresponding to >3.5m). Currently, Token = 3 < 5, suppression is ineffective, and the guard function... If true, change enables, and fishing boats begin operations.

[0067] Heterogeneous Collaborative Detection: The Evolution of Subsequent "Transfer and Docking" A transfer vessel is required. The suppression arc threshold for this transition is set to 3 (corresponding to 2.0m).

[0068] Critical state: The current sea state of 1.8m is close to the limit of the transfer vessel (2.0m).

[0069] Dynamic Sea State Response and Replanning: Thirty minutes into the operation, the marine sensors detected that the SWH had risen to 2.2m. The system updated the ECP in real time and sent it to... Inject a Token to make its quantity 4 (representing sea state 4, >2.0m).

[0070] Triggered inhibition: At this time, transport transition Suppression conditions (sea state rating) If the condition is met, the system immediately freezes the enabled state of the transition, even if the transport ship has already reached the designated location.

[0071] Deadlock prediction and resolution: The runtime engine detects the transfer ship. The system is in a "waiting to dock" state, but the target transition is suppressed, which is determined to be an "environmental deadlock". Based on the strategy in step S4, the system triggers a high-priority transition. (Category B equipment abort / return). The transfer vessel receives the "Sea state exceeds limits, return immediately" instruction, abandons the docking mission, and returns to the mother ship or a sheltered spot.

[0072] Robustness is demonstrated by the operational changes of the fishing vessel (3.5m seakeeping depth). The suppression threshold is 5, the current level 4 is less than 5, and its own seakeeping meets the requirements. Therefore, the fishing vessel is not affected and continues to complete the fishing and net-hauling operations, temporarily storing the catch in the live fish holding compartments attached to the net cages, awaiting the subsequent transfer window, thus avoiding a complete paralysis of the entire process.

[0073] Deadlock prediction and resolution based on digital twins are as follows: This method integrates deadlock prediction functionality for "intelligent operation and maintenance" and "production demonstration".

[0074] State space construction: The system periodically (every 5 minutes) uses the current state to construct the state space. As the root node, it combines the sea state forecast sequence for the next 3 hours (provided by ERA5 data, such as...) ), generating a time-varying state reachability graph.

[0075] Deadlock search: Search for all non-terminal nodes with an out-degree of 0 in the reachable graph (i.e., deadlock state).

[0076] Case: The predicted wave height will reach 4.0m in T+2 hours. Analysis revealed that if the fishing vessel is fully loaded with fish and waiting for the transfer vessel (wave resistance 2.0m) at this time, but the transfer vessel is unable to leave port due to the large waves, the system will enter a dangerous deadlock state of "fishing vessel stranded at sea" (the wind resistance capability is reduced when fully loaded).

[0077] Solution: The system traces back at time T (now) and finds that the critical path leading to the deadlock is "fishing vessels waiting at full load". Therefore, the system automatically modifies the scheduling strategy at time T: suspending fishing operations, or forcing fishing vessels to complete unloading or enter a sheltered anchorage before time T+1.

[0078] The data sources and hardware implementation are as follows: The hardware implementation of this method relies on the integration of the following components, which meets the requirements for the development of "intelligent equipment": Sensing Layer: Deploys multiple types of sensors, including: ① Shipborne sensors: Each vessel is equipped with wave buoys, anemometers, current meters, IMU inertial measurement units, and GPS positioning modules; ② Shore-based sensors: Deploy shore-based radars around aquaculture areas to monitor distant sea conditions and weather stations to collect nearshore wind speed and air pressure data; ③ Third-party data interfaces: Connect to the API interface of the National Marine Environmental Forecasting Center (NMEFC) to obtain refined grid forecast data and ERA5 reanalysis data for the Yellow and Bohai Seas.

[0079] Computation layer: (1) Petri Net Engine: Runs on industrial PCs or edge computing nodes, written in C++, and integrates the kernel algorithm of Pipe or CPNTools to achieve millisecond-level simulation.

[0080] (2) Environmental database: Access the API interface of the National Marine Environmental Forecasting Center (NMEFC) to obtain refined grid forecast data for the Yellow and Bohai Seas.

[0081] Execution layer: Petri net transitions are mapped to underlying PLC control commands via the MQTT / DDS protocol and sent to the controllers of each unmanned equipment.

[0082] This embodiment obtained the following through simulation: Figure 5 The simulation results comparison chart shown below demonstrates how simulation results can be compared to simulation results. Figure 5 It can be seen that in calm sea states (SWH<1m), the two methods are equally efficient; in medium to high sea states (SWH1.5-2.5m), EC-GSPN significantly improves the effective operating time through dynamic scheduling; in severe sea states (SWH>3m), EC-GSPN successfully avoids all potential safety accidents (the accident rate is reduced to 0).

[0083] Compared with the prior art, the present invention has the following significant advantages: Significantly enhanced safety: Through a dual constraint mechanism of "environmental constraint library + sea state suppression arc," the possibility of forced operation of equipment under extreme sea conditions is eliminated from the model's underlying logic, effectively avoiding safety accidents such as equipment capsizing and collisions caused by strong winds and waves in the Yellow and Bohai Seas. Under severe sea conditions (SWH>3m), the accident rate is reduced to 0, a significant improvement in safety compared to traditional methods (accident rate 45%). Simultaneously, the dynamic deadlock detection and hierarchical resolution mechanism ensures that the system can quickly self-heal in the event of sudden environmental changes or resource competition, preventing operational process paralysis and greatly enhancing system robustness.

[0084] Maximizing the utilization of the operational window: Based on a differentiated scheduling mechanism using a colored Petri net, precise control of heterogeneous equipment is achieved, avoiding a "one-size-fits-all" shutdown strategy. In medium to high sea states (SWH=1.5-2.5m), large equipment with strong seakeeping capabilities (such as 3.5m seakeeping class fishing vessels) are allowed to continue operating, while only equipment with low seakeeping capabilities (such as 2.0m seakeeping class unmanned transfer vessels) is restricted. The effective operating time is increased by more than 22.5% compared to traditional methods, significantly improving the return on investment of deep-sea aquaculture facilities.

[0085] Highly adaptable: Integrating JONSWAP wave spectrum and multi-source data fusion technology, the model can accurately simulate the nonlinear impact of sea state on operational efficiency, and the operational plan is highly matched with the actual marine environmental characteristics. Combining sea state forecasts for the next 1-3 hours with a rolling time-domain replanning strategy, the system can anticipate the risks of severe sea state in advance and proactively adjust the operational process, realizing the transformation from "reactive control" to "predictive control," and exhibiting extremely strong adaptability in the rapidly changing sea state areas of the Yellow and Bohai Seas.

[0086] Deadlock prevention and self-healing: Through state space analysis, it is possible to identify process deadlocks that may be caused by environmental deterioration in advance (such as "ship A is waiting for ship B, but ship B cannot arrive due to large waves"), and automatically trigger replanning or degraded operation mode.

[0087] High engineering practicality and compliance: Specialized adaptation logic is designed for typical scenarios such as "harvesting and transfer" and "intelligent feeding," and hard constraints of fishery safety regulations are embedded within, allowing scheduling strategies to be directly applied to engineering practice. The dual-mode communication mechanism solves the problem of limited communication in deep-sea areas, and the millisecond-level response capability of the runtime engine meets real-time control requirements, providing reliable technical support for realizing "unmanned fishing grounds" in deep-sea areas.

[0088] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for modeling and dynamically controlling a distributed collaborative operation process of a fishery equipment group, characterized in that, Includes the following steps: An environmentally coupled generalized stochastic Petri net model is constructed based on real-time sea state data, which includes an environmental constraint library and a sea state suppression arc. The environmental constraint library is used to map sea state levels, and the sea state suppression arc is used to suppress operational change triggering based on the sea state level. The tokens are colored according to the wave-keeping properties of the heterogeneous equipment, and the matching relationship between the wave-keeping properties and the current sea state level is determined by the guard function to determine the currently enabled set of operational transitions. For the enabled operational transition set, a dynamic nonlinear function of the transition trigger rate is established based on the wave spectrum parameters. The real-time sea state data is mapped to random time delay parameters, and the random time delay parameters are used to control the random triggering process of the enabled operational transition. Based on the state changes generated by the random triggering process, a state reachability graph is dynamically constructed to identify environmental deadlock states and trigger risk avoidance and retreat transitions to resolve the environmental deadlock. The state change process is modeled as a semi-Markov decision process. A state space is constructed based on the state reachability graph, and an action space is constructed based on the enabling transition set. The optimal transition trigger sequence is searched in the state space and action space based on the comprehensive reward function. The real-time identifiers of the Petri net model are mapped to physical equipment control commands based on the optimal transition trigger sequence, and a rolling time-domain replanning is triggered when a step change in sea state is detected.

2. The method according to claim 1, characterized in that, The construction of an environmentally coupled generalized stochastic Petri net model based on real-time sea state data, which includes an environmental constraint library and a sea state inhibition arc, includes: Perform multi-source data fusion based on shipborne sensor data and public meteorological service data; The combined sea state index for the current time and the future forecast time domain is calculated based on the fused data; The comprehensive sea state index is quantified into a number of tokens and injected into the environmental constraint library.

3. The method according to claim 1, characterized in that, The step of determining the matching relationship between the seakeeping attribute and the current sea state level through the guard function includes: Determine the conditions for switching operating modes based on the effective wave height threshold; The state of the mutex semaphore is controlled according to the job mode switching conditions to manage serial or parallel access permissions to the shared job space. Adjust the trigger frequency of feeding changes based on the feeding intensity feedback factor; Based on the relationship between the current sea state level and the fish feeding activity threshold, the triggering of feeding changes is paused through a sea state inhibition arc.

4. The method according to claim 1, characterized in that, The dynamic nonlinear function for establishing the transition trigger rate based on wave spectrum parameters includes: The operational efficiency attenuation coefficient is calculated based on the significant wave height, spectral peak period, and wind speed. The transition trigger rate is determined based on the operation efficiency attenuation coefficient.

5. The method according to claim 1, characterized in that, The process of identifying environmental deadlock states and triggering risk avoidance and rollback transitions to resolve environmental deadlocks includes: The risk avoidance retreat transition is triggered based on the storm warning status in the environmental constraint database. Based on the triggering of the aforementioned risk avoidance and retreat transition, a mission abort command is broadcast to the relevant equipment; Based on the current location of each piece of equipment, a path plan is generated for sheltered anchorage or deep-sea diving to avoid waves; Based on the triggering of the risk avoidance and withdrawal transition, the tokens of the relevant operation warehouses are cleared using the reset arc; The system is guided to enter a safe standby state based on the status of the emptied warehouse.

6. The method according to claim 1, characterized in that, The search for the optimal transition trigger sequence based on the comprehensive reward function includes: The state space is refined based on the Cartesian product of the Petri net reachability set and the environment state; The comprehensive reward function is constructed based on operation time, energy consumption, and sea state risk penalties. The optimal transition trigger sequence is searched based on the state space, action space, and comprehensive reward function.

7. The method according to claim 1, characterized in that, The triggering of rolling temporal replanning upon detecting a step change in sea state includes: The rolling time-domain replanning is triggered based on a step change in physical sea state monitoring data or a predicted window closing event. The workflow is reconstructed based on the results of the rolling time-domain replanning, enabling dynamic switching from production mode to risk-avoidance mode.

8. The method according to claim 1, characterized in that, The step of mapping the real-time identifiers of the Petri net model to physical equipment control commands based on the optimal transition trigger sequence includes: The real-time identifiers of the Petri net model are mapped to equipment control commands based on the digital twin runtime engine. The status markers of the environmental constraint library are updated in real time based on the data fusion results of shipborne sensors and shore-based radar.

9. An electronic device comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, characterized in that, When the processor executes the computing program, it implements the method of any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-8.

Images