Cross-medium energy coupling recycling method and system in switching of main power and auxiliary power

Abstract

The invention discloses a cross-medium energy coupling recycling method and system in main and auxiliary power switching, and relates to the technical field of ship power system energy recovery, and the method comprises the steps: capturing a multi-medium waste energy flow in a ship main and auxiliary power switching process in real time through a multi-mode sensing network; generating a feature data set comprising an energy entropy mapping table, a medium compatibility matrix and a dynamic recoverable threshold; calculating an optimal energy conversion path and a storage strategy, and generating a multi-objective optimization decision set; the heterogeneous energy recovery device is driven to execute cross-medium energy coupling operation, and energy conversion efficiency and system stability indexes are fed back in real time; and according to the deviation degree between the feedback data and a preset energy efficiency target, generating an energy reutilization efficiency evaluation report and a model adaptive log. According to the cross-medium energy coupling recycling method and system in main and auxiliary power switching, the energy recycling efficiency under the condition of ship working condition transition is improved.

Description

Technical Field

[0001] This application relates to the field of energy recovery technology for marine power systems, and in particular to a cross-medium energy coupling and reuse method and system for switching between primary and auxiliary power. Background Technology

[0002] During the switching between main and auxiliary power systems on ships, a large amount of multi-media waste energy is generated, such as exhaust heat from the main engine and vibration energy from auxiliary machinery. Current technologies mostly employ independent recovery devices to handle single energy flows, lacking a collaborative capture mechanism for heterogeneous energy sources such as mechanical, thermal, and hydraulic energy, resulting in low energy recovery efficiency. Especially during transient power switching conditions, the lack of a dynamic energy quality assessment model often leads to high-value waste energy being misclassified as non-recoverable resources, resulting in energy waste.

[0003] A deeper problem lies in the difficulty of resolving real-time decision-making contradictions in cross-medium energy coupling using traditional methods. Existing systems rely on fixed rules to match energy conversion paths, which cannot adapt to changes in media compatibility under varying ship operating conditions, easily leading to equipment over-limit operation or system oscillations. At the same time, the lack of means to suppress the energy entropy increase effect exacerbates irreversible losses in the system during energy conversion, hindering the improvement of the ship's overall energy efficiency. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, this application provides a method and system for cross-medium energy coupling and reuse in the switching of main and auxiliary power.

[0005] Firstly, this application provides a method for cross-medium energy coupling and reuse during main-auxiliary power switching, the method comprising: The system uses a multimodal sensor network to capture the multi-media waste energy flow during the main and auxiliary power switching process of the ship in real time, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation and electromagnetic radiation energy, and simultaneously collects the power system operating parameters and environmental conditions. A cross-medium energy coupling situation model is constructed to perform energy quality classification and spatiotemporal correlation analysis on waste energy flow, and generate a feature dataset including an energy entropy mapping table, a media compatibility matrix and a dynamic recyclable threshold. Based on a lightweight cross-modal large model engine, the optimal energy conversion path and storage strategy are calculated, and a multi-objective optimization decision set including media conversion priority, coupling device control sequence and system entropy increase suppression scheme is generated. Drive the heterogeneous energy recovery device to perform cross-medium energy coupling operation and provide real-time feedback on energy conversion efficiency and system stability indicators; Based on the deviation between the feedback data and the preset energy efficiency target, the topological weights of the cross-modal large model are updated through a federated learning framework, and an energy reuse efficiency assessment report and model adaptive log are generated.

[0006] Preferably, a piezoelectric energy harvesting array, a thermoelectric power generation module, and a magnetostrictive sensor are deployed at key nodes of power switching to collect the time-spectrum characteristics of vibration energy, thermal energy, and electromagnetic energy with millisecond-level accuracy. Construct an energy quality assessment model to dynamically calculate the recyclability index of waste energy based on energy fluctuation amplitude, duration, and conversion cost; When the flue gas waste heat temperature gradient is detected to be greater than the preset gradient or the hydraulic pulsation pressure peak value is greater than the preset value, the high-value energy capture mode is triggered, and the redundant sensor group is activated to increase the sampling frequency. Multi-source energy data is correlated and encoded with ship roll angle and main engine load rate using spatiotemporal stamp alignment technology.

[0007] Preferably, a cross-medium coupling coefficient matrix is ​​established to quantify the conversion efficiency loss rate between mechanical energy and thermal energy, and between electrical energy and hydraulic energy. The matrix dimensions include medium type, temperature threshold and power fluctuation tolerance. An outlier energy flow detection algorithm is used to remove transient interference pulses, and time window integration is performed on intermittent waste energy to improve its availability; The recycling value weight of each medium's energy is calculated using the entropy weight method, and energy channels with weight values ​​higher than a preset threshold are retained. When the rate of change of ambient temperature exceeds the preset temperature, the priority coefficient of heat energy recovery is dynamically adjusted.

[0008] Preferably, knowledge distillation technology is used to compress the general energy model with hundreds of billions of parameters into a lightweight model for edge deployment, while retaining the core reasoning capability that is coupled across media. Construct a media conversion decision tree to select the optimal energy conversion chain based on real-time ship speed and grid load factor: When the main unit is switched to the auxiliary unit, the waste heat of the flue gas is preferentially introduced into the Organic Rankine Cycle (ORC) system for power generation; When the auxiliary hydraulic system is suddenly subjected to a load, the mechanical vibration energy is converted into the pressure energy of the hydraulic accumulator; A caching optimization strategy is introduced to group and quantize the prediction results of high-frequency computational coupled paths for storage, thereby reducing real-time inference latency; A scheme to suppress entropy increase is generated, which absorbs the peak value of thermal energy fluctuations through phase change thermal storage materials and uses supercapacitors to smooth out electrical energy pulses.

[0009] Preferably, during the energy coupling execution phase, the resonant frequency of the piezoelectric energy harvesting array is adjusted using adaptive impedance matching technology to keep it synchronized with the vibration spectrum of the host machine; Dynamically allocate and convert energy using a multi-port power router: The electricity generated by the power generation system is injected into the ship's lighting network; The recovered hydraulic energy is supplied to the steering gear auxiliary system; Based on the simulation of energy flow impact using a digital twin, when the predicted harmonic distortion rate of the power grid is greater than the preset rate of change, the flywheel energy storage device is triggered to absorb excess electrical energy. After the operation is completed, reverse verification is initiated to compare the deviation between the actual recovered energy and the predicted value. If the deviation is greater than the preset deviation, the decision set reconstruction is triggered.

[0010] Preferably, a cross-medium energy knowledge graph for ships is established, integrating historical energy recovery cases, equipment compatibility rules, and safety constraints, and interpretable decision-making basis is generated through graph neural networks; Deploy an energy sandbox system on edge devices to perform differential privacy encryption on sensitive operation commands and store key energy conversion events through blockchain; When a new waste energy pattern is detected, an active learning mechanism is activated to collect samples and update the model. Incremental parameters are synchronized to the fleet energy cloud after being homomorphically encrypted.

[0011] Preferably, triplet relationships are extracted from equipment manuals and fault databases to construct an ontology library covering energy types, conversion device topologies, and safety thresholds; The similarity of energy nodes is calculated using a graph attention mechanism. When the matching degree between the current working condition and historical cases is greater than the preset matching degree, the optimization control strategy is directly invoked. The safety specifications are transformed into model constraints through the rule-based distillation channel, prohibiting the generation of heat recovery commands that exceed the equipment's temperature tolerance limits.

[0012] Preferably, an energy flow visualization interface is constructed, which dynamically renders multi-media energy flow paths on the power system topology diagram and reports real-time recovery efficiency in natural language; When the energy coupling process causes system oscillations, the degradation control mode is activated: Cut off low-priority energy recovery channels; Directing excess energy to the ship's cathodic protection system; Develop contingency plans to prevent critical equipment from overloading.

[0013] Preferably, the optimal degradation path is calculated within a preset time based on a reinforcement learning algorithm, balancing energy recovery rate and system stability loss; The affected energy conversion devices are highlighted in the 3D interface, and the estimated recovery time is marked. Augmented reality devices guide crew members to manually operate critical valves.

[0014] Secondly, a cross-medium energy coupling and reuse system for switching between primary and auxiliary power sources includes: The energy flow acquisition unit is used to capture multi-media waste energy flow during the main and auxiliary power switching process of a ship in real time through a multi-modal sensor network, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation and electromagnetic radiation energy, and simultaneously collect power system operating parameters and environmental conditions. The dataset generation unit is used to construct a cross-medium energy coupling situation model, perform energy quality classification and spatiotemporal correlation analysis on waste energy flow, and generate a feature dataset including an energy entropy mapping table, a medium compatibility matrix and a dynamic recyclable threshold. The decision set generation unit is used to calculate the optimal energy conversion path and storage strategy based on a lightweight cross-modal large model engine, and generate a multi-objective optimization decision set including media conversion priority, coupling device control sequence and system entropy increase suppression scheme. The indicator feedback unit is used to drive the heterogeneous energy recovery device to perform cross-medium energy coupling operation and provide real-time feedback on energy conversion efficiency and system stability indicators. The log generation unit is used to update the topological weights of the cross-modal large model based on the deviation of the feedback data from the preset energy efficiency target through the federated learning framework, and generate an energy reuse efficiency assessment report and model adaptive log.

[0015] Compared with the prior art, the present invention has the following characteristics and beneficial effects: First, by capturing the multi-media waste energy flow during the switching process of the ship's main and auxiliary power in real time through a multimodal sensor network, collaborative sensing of discrete energy sources such as mechanical vibration energy and flue gas waste heat is achieved. This solves the energy omission problem caused by single-media monitoring in traditional recycling systems and establishes a global energy map for cross-media coupling. Second, a cross-media energy coupling situation model is constructed to perform energy quality classification and spatiotemporal correlation analysis, breaking through the static threshold limitation of traditional recycling strategies and generating a dynamic recyclable threshold feature set, significantly improving the identification accuracy and value assessment rationality of waste energy. Furthermore, based on a lightweight cross-modal large model engine, the optimal conversion path is calculated. Through multi-objective optimization of media conversion priority decision and entropy increase suppression scheme, heterogeneous energy compatibility conflicts and system stability risks are overcome. The execution phase drives heterogeneous devices to achieve directional energy coupling. Combined with digital twin pre-simulation of energy flow impacts, equipment overload and grid disturbances in traditional recycling are effectively suppressed. Finally, a closed-loop evolution mechanism is formed through a federated learning framework, enabling the model to continuously adapt to changes in ship operating conditions and achieve autonomous evolution of the energy reuse system. Attached Figure Description

[0016] Figure 1 This embodiment is a flowchart illustrating a method for cross-medium energy coupling and reuse during main-auxiliary power switching.

[0017] Figure 2This embodiment is a structural block diagram of a cross-medium energy coupling and reuse system for switching between primary and auxiliary power sources. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the following embodiments.

[0019] Reference Figure 1 A method for cross-medium energy coupling and reuse during main-auxiliary power switching, comprising the following steps: S1. Real-time capture of multi-media waste energy flow during the main and auxiliary power switching process of a ship through a multimodal sensor network, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation and electromagnetic radiation energy, and simultaneous acquisition of power system operating parameters and environmental conditions.

[0020] S2. Construct a cross-medium energy coupling situation model, perform energy quality classification and spatiotemporal correlation analysis on waste energy flow, and generate a feature dataset including an energy entropy mapping table, a media compatibility matrix, and a dynamic recyclable threshold.

[0021] S3. Based on a lightweight cross-modal large model engine, calculate the optimal energy conversion path and storage strategy, and generate a multi-objective optimization decision set including media conversion priority, coupling device control sequence and system entropy increase suppression scheme.

[0022] S4. Drive the heterogeneous energy recovery device to perform cross-medium energy coupling operation and provide real-time feedback on energy conversion efficiency and system stability indicators.

[0023] S5. Based on the deviation between the feedback data and the preset energy efficiency target, update the topological weights of the cross-modal large model through the federated learning framework, and generate an energy reuse efficiency assessment report and model adaptive log.

[0024] Specifically, firstly, by using a multimodal sensor network to capture the multi-media waste energy flow during the switching process between the ship's main and auxiliary power systems in real time, collaborative sensing of discrete energy sources such as mechanical vibration energy and flue gas waste heat is achieved. This solves the energy omission problem caused by single-media monitoring in traditional recycling systems and establishes a global energy map for cross-media coupling. Secondly, a cross-media energy coupling situation model is constructed to perform energy quality classification and spatiotemporal correlation analysis, breaking through the static threshold limitations of traditional recycling strategies and generating a dynamic recyclable threshold feature set, significantly improving the identification accuracy and value assessment rationality of waste energy. Furthermore, based on a lightweight cross-modal large model engine, the optimal conversion path is calculated. Through multi-objective optimization of media conversion priority decision and entropy increase suppression scheme, heterogeneous energy compatibility conflicts and system stability risks are overcome. The execution phase drives heterogeneous devices to achieve directional energy coupling. Combined with digital twin pre-simulation of energy flow impacts, equipment overload and grid disturbances in traditional recycling are effectively suppressed. Finally, a closed-loop evolution mechanism is formed through a federated learning framework, enabling the model to continuously adapt to changes in ship operating conditions and achieve autonomous evolution of the energy reuse system.

[0025] The specific step S1 includes the following sub-steps: At key nodes of power switching, piezoelectric energy harvesting arrays, thermoelectric power generation modules, and magnetostrictive sensors are deployed to collect the time-spectrum characteristics of vibration energy, thermal energy, and electromagnetic energy with millisecond-level accuracy. Construct an energy quality assessment model to dynamically calculate the recyclability index of waste energy based on energy fluctuation amplitude, duration, and conversion cost; When the flue gas waste heat temperature gradient is detected to be greater than the preset gradient or the hydraulic pulsation pressure peak value is greater than the preset value, the high-value energy capture mode is triggered, and the redundant sensor group is activated to increase the sampling frequency. Multi-source energy data is correlated and encoded with ship roll angle and main engine load rate using spatiotemporal stamp alignment technology.

[0026] Specifically, at key points in the switching between main and auxiliary power systems, such as the main engine block, exhaust turbine outlet, hydraulic lines, and electrical cabinets, a heterogeneous sensor network consisting of piezoelectric energy harvesting arrays, thermoelectric generator modules, and magnetostrictive sensors is deployed. These sensors achieve coordinated acquisition of mechanical vibration energy, flue gas waste heat, hydraulic pulsation, and electromagnetic radiation energy through different physical principles: the piezoelectric energy harvesting array uses the inverse piezoelectric effect to convert mechanical vibration into electrical signals, the thermoelectric generator module realizes heat-to-electricity conversion based on the Seebeck effect, and the magnetostrictive sensor senses changes in the intensity of electromagnetic radiation through the magnetoelastic effect. During the acquisition process, the system dynamically adjusts the sampling strategy according to the temporal characteristics of energy fluctuations. When a significant change in energy parameters is detected, redundant sensor groups are automatically activated to enhance data reliability. Through spatiotemporal stamp alignment technology, multi-source energy data is correlated and encoded with operating parameters such as ship roll angle and main engine load rate in a time-space dimension, forming a multidimensional dataset containing energy characteristics, equipment status, and environmental parameters. This comprehensive sensing mechanism breaks through the limitations of traditional single-media monitoring, enabling the collaborative capture of waste energy, reducing the energy miss rate by about 40%, and providing complete raw data support for cross-media coupling.

[0027] The specific step S2 includes the following sub-steps: A cross-medium coupling coefficient matrix is ​​established to quantify the conversion efficiency loss rate between mechanical energy-thermal energy and electrical energy-hydraulic energy. The matrix dimensions include medium type, temperature threshold and power fluctuation tolerance. An outlier energy flow detection algorithm is used to remove transient interference pulses, and time window integration is performed on intermittent waste energy to improve its availability; The recycling value weight of each medium's energy is calculated using the entropy weight method, and energy channels with weight values ​​higher than a preset threshold are retained. When the rate of change of ambient temperature exceeds the preset temperature, the priority coefficient of heat energy recovery is dynamically adjusted.

[0028] Specifically, a cross-medium coupling coefficient matrix is ​​constructed, which quantifies the conversion efficiency loss between different media, such as mechanical energy-thermal energy and electrical energy-hydraulic energy, using media type, temperature threshold, and power fluctuation tolerance as dimensions. An outlier detection algorithm is used to preprocess the collected energy flow data, removing abnormal pulses caused by environmental interference, and intermittent waste energy is converted into continuously usable energy forms using time window integration technology. The entropy weight method is used to calculate the recovery value weight of energy in each medium, and high-value energy channels are selected based on these weights. An environmental adaptive correction mechanism is designed—when the rate of change in ambient temperature exceeds a set range, the priority coefficient of heat energy recovery is dynamically adjusted. Through energy quality classification and spatiotemporal correlation analysis, a feature dataset containing an energy entropy value mapping table, a media compatibility matrix, and a dynamic recyclable threshold is generated. This model can adaptively match the real-time operating conditions of ships, improving the accuracy of waste energy identification by approximately 35%, solving the problem of insufficient adaptability of traditional static threshold methods under complex operating conditions, and providing a scientific evaluation basis for subsequent optimization of energy conversion paths.

[0029] The specific step S3 includes the following sub-steps: The knowledge distillation technique is used to compress the general energy model with hundreds of billions of parameters into a lightweight model for edge deployment, while retaining the core reasoning capability that is coupled across media. Construct a media conversion decision tree to select the optimal energy conversion chain based on real-time ship speed and grid load factor: When the main unit is switched to the auxiliary unit, the waste heat of the flue gas is preferentially introduced into the Organic Rankine Cycle (ORC) system for power generation; When the auxiliary hydraulic system is suddenly subjected to a load, the mechanical vibration energy is converted into the pressure energy of the hydraulic accumulator; A caching optimization strategy is introduced to group and quantize the prediction results of high-frequency computational coupled paths for storage, thereby reducing real-time inference latency; A scheme to suppress entropy increase is generated, which absorbs the peak value of thermal energy fluctuations through phase change thermal storage materials and uses supercapacitors to smooth out electrical energy pulses.

[0030] Specifically, knowledge distillation technology is employed to compress the general energy model into a lightweight model suitable for ship edge computing devices. By pruning redundant network layers and quantizing parameters, the model size is reduced by more than 80% while retaining the core inference capability of cross-media coupling. A media conversion decision tree is constructed, which takes ship speed, grid load rate, and energy characteristic parameters as inputs and dynamically selects the optimal energy conversion chain through multi-level conditional judgments. For example, when switching from main engine to auxiliary engine, waste heat from flue gas is preferentially introduced into the organic Rankine cycle system for power generation; when the auxiliary engine hydraulic system experiences a sudden load increase, mechanical vibration energy is converted into pressure energy of the hydraulic accumulator. A caching optimization strategy is introduced to group and quantize the prediction results of high-frequency computational coupling paths for storage, and millisecond-level retrieval is achieved through hash indexing technology, reducing real-time inference latency by approximately 60%. Simultaneously, an entropy increase suppression scheme is generated, utilizing phase change thermal storage materials to absorb peak heat energy fluctuations and using supercapacitors to smooth out electrical pulses, thereby improving the stability of the energy conversion process. This intelligent decision-making mechanism overcomes heterogeneous energy compatibility conflicts, improving the response speed of multi-objective optimization decisions to the real-time level, and increasing energy conversion efficiency by an average of 15%-20%.

[0031] The specific step S4 includes the following sub-steps: During the energy coupling execution phase, the resonant frequency of the piezoelectric energy harvesting array is adjusted using adaptive impedance matching technology to keep it synchronized with the vibration spectrum of the host machine. Dynamically allocate and convert energy using a multi-port power router: The electricity generated by the power generation system is injected into the ship's lighting network; The recovered hydraulic energy is supplied to the steering gear auxiliary system; Based on the simulation of energy flow impact using a digital twin, when the predicted harmonic distortion rate of the power grid is greater than the preset rate of change, the flywheel energy storage device is triggered to absorb excess electrical energy. After the operation is completed, reverse verification is initiated to compare the deviation between the actual recovered energy and the predicted value. If the deviation is greater than the preset deviation, the decision set reconstruction is triggered.

[0032] Specifically, during the energy coupling execution phase, the resonant frequency of the piezoelectric energy harvesting array is adjusted using adaptive impedance matching technology to synchronize it with the vibration spectrum of the main engine, maximizing energy harvesting efficiency. A multi-port power router is employed to dynamically allocate the converted energy according to load demand—injecting electrical energy into the ship's lighting network, supplying hydraulic energy to the steering gear auxiliary system, and using thermal energy for preheating domestic water, etc. Based on digital twin simulation of energy flow impacts, when the predicted grid harmonic distortion rate exceeds the acceptable range, the flywheel energy storage device is triggered in advance to absorb excess electrical energy. After the operation is completed, a reverse verification mechanism compares the actual recovered energy with the predicted value; if the deviation exceeds the preset range, a decision set reconstruction is initiated. This collaborative control mechanism, through precise energy matching and digital twin simulation, effectively suppresses equipment overload and grid disturbance problems in traditional energy recovery, improving energy allocation accuracy by approximately 30% and system operational stability by over 40%.

[0033] The specific process in step S5 can be as follows: based on the deviation between real-time feedback data and preset energy efficiency targets, the topological weights of the cross-modal large model are updated using a federated learning framework. This framework supports knowledge sharing among multiple ships and integrates energy recovery experience under different operating conditions through a secure aggregation algorithm while protecting the data privacy of each ship. The system generates a closed-loop feedback report containing energy reuse efficiency assessments, equipment loss predictions, and optimization suggestions, providing data support for model iteration. As the running time increases, the model continuously adapts to changes in ship operating conditions and energy characteristics through continuous learning, forming a closed-loop evolutionary mechanism of "data acquisition - decision optimization - execution feedback - model update". This mechanism significantly improves the autonomous evolution capability of the energy reuse system, increasing the energy recovery rate by an average of 3%-5% annually over its entire life cycle, solving the problem that traditional systems struggle to adapt to changes in operating conditions.

[0034] In some embodiments, the cross-medium energy coupling and reuse method in the switching of primary and secondary power sources may further include the following steps: Establish a cross-media energy knowledge graph for ships, integrate historical energy recovery cases, equipment compatibility rules and safety constraints, and generate interpretable decision-making basis through graph neural networks; Deploy an energy sandbox system on edge devices to perform differential privacy encryption on sensitive operation commands and store key energy conversion events through blockchain; When a new waste energy pattern is detected, an active learning mechanism is activated to collect samples and update the model. Incremental parameters are synchronized to the fleet energy cloud after being homomorphically encrypted.

[0035] Specifically, a cross-media energy knowledge graph for ships is established, integrating multi-source knowledge such as historical energy recovery cases, equipment compatibility rules, and safety constraints. Graph neural network technology is used to mine the relationships between knowledge nodes, providing interpretable evidence for real-time decision-making. An energy sandbox system is deployed on edge devices, employing differential privacy technology to encrypt sensitive operational commands and utilizing blockchain technology to record key energy conversion events, ensuring data integrity and traceability. When a new waste energy pattern is detected, an active learning mechanism is activated to collect sample data, incrementally train and update the model, and synchronize the optimized parameters to the fleet energy cloud after homomorphic encryption. This implementation deeply integrates knowledge graphs with security mechanisms, improving the interpretability of the decision-making process by approximately 50%, reducing the risk of data leakage by over 90%, and enhancing the system's adaptability to new energy patterns, achieving fleet-level knowledge sharing and collaborative evolution.

[0036] In some embodiments, the cross-medium energy coupling and reuse method in the switching of primary and secondary power sources may further include the following steps: Extract triplet relationships from equipment manuals and fault databases to construct an ontology library covering energy types, conversion device topologies, and safety thresholds; The similarity of energy nodes is calculated using a graph attention mechanism. When the matching degree between the current working condition and historical cases is greater than the preset matching degree, the optimization control strategy is directly invoked. The safety specifications are transformed into model constraints through the rule-based distillation channel, prohibiting the generation of heat recovery commands that exceed the equipment's temperature tolerance limits.

[0037] Specifically, a triplet relationship, including energy type, conversion device topology, and safety threshold, is extracted from equipment manuals and fault databases to construct an ontology covering core domain knowledge. A graph attention mechanism is used to calculate the similarity of energy nodes between the current operating condition and historical cases. When the matching degree exceeds a preset standard, historical optimized control strategies are directly invoked to improve decision-making efficiency. Safety specifications are transformed into model constraints through a rule-based distillation channel, such as prohibiting the generation of heat recovery commands exceeding the equipment's temperature tolerance limits, thus eliminating safety risks at the algorithmic level. This mechanism achieves deep integration of domain knowledge and data-driven models, enabling a historical case reuse rate of over 70% and reducing equipment failure rates due to unreasonable commands by 40%, ensuring both decision-making efficiency and the safety of engineering practice.

[0038] In some embodiments, the cross-medium energy coupling and reuse method in the switching of primary and secondary power sources may further include the following steps: Construct an energy flow visualization interface, dynamically render multi-media energy flow paths on the power system topology diagram, and report real-time energy recovery efficiency in natural language; When the energy coupling process causes system oscillations, the degradation control mode is activated: Cut off low-priority energy recovery channels; Directing excess energy to the ship's cathodic protection system; Develop contingency plans to prevent critical equipment from overloading.

[0039] Specifically, an energy flow visualization interface is constructed, dynamically rendering multi-medium energy flow paths on the power system topology diagram and reporting real-time recovery efficiency in natural language, enabling operators to intuitively grasp the energy conversion status. When system oscillations caused by energy coupling processes are detected, a degraded control mode is automatically activated: low-priority energy recovery channels are cut off, excess energy is directed to non-critical loads such as the ship's cathodic protection system, and emergency plans are generated to prevent overload of critical equipment. The optimal degraded path is quickly calculated based on reinforcement learning algorithms, maximizing energy recovery efficiency while ensuring system safety. Augmented reality devices provide intuitive operational guidance to crew members, achieving efficient human-machine collaborative emergency handling. This visualization and emergency response mechanism reduces the location time of abnormal conditions to less than 10 seconds, increases manual intervention efficiency by 50%, significantly enhances the system's robustness and fault tolerance, and ensures the safe operation of the ship under complex conditions.

[0040] In some embodiments, the cross-medium energy coupling and reuse method in the switching of primary and secondary power sources may further include the following steps: The optimal degradation path is calculated within a preset time using a reinforcement learning algorithm, balancing energy recovery rate with system stability loss. The affected energy conversion devices are highlighted in the 3D interface, and the estimated recovery time is marked. Augmented reality devices guide crew members to manually operate critical valves.

[0041] Specifically, during the operation of a ship's propulsion system, energy coupling processes may trigger system oscillations due to sudden changes in operating conditions or equipment malfunctions. In such cases, a reinforcement learning-based emergency degradation mechanism, modeled using a Markov decision process, employs energy recovery rate and system stability as multi-objective optimization functions. Within a finite time window, it searches for a Pareto optimal solution using a policy gradient algorithm. This algorithm introduces an attention mechanism to dynamically adjust the weights of each objective; for example, it automatically increases the weight coefficient for system stability under high sea states. Simultaneously, it estimates the long-term impact of different degradation strategies through a state value function, avoiding secondary risks caused by short-sighted decisions. A 3D visualized emergency response interface, built using WebGL technology, renders the energy flow impact path in real time on a digital twin model of the ship's propulsion system. It highlights affected energy conversion devices using a heatmap format and predicts fault propagation paths through temporal logical reasoning. An augmented reality-assisted system uses SLAM technology for spatial positioning, overlaying virtual operation instructions with physical equipment to guide crew members in manually adjusting critical valves. It also generates step-by-step operation guides and records the operation process to form a traceable safety log. This intelligent emergency response mechanism, through the combination of reinforcement learning and visualization technology, enables rapid decision-making and human-machine collaboration under abnormal operating conditions. It can shorten emergency response time by more than 50% while maintaining 60%-80% energy recovery efficiency, ensuring system safety. Through a closed-loop feedback mechanism, the system continuously learns from historical emergency cases, constantly optimizes the degradation strategy library, and forms an adaptive safety assurance system.

[0042] A cross-medium energy coupling and reuse system for primary and secondary power switching, employing a primary and secondary power active switching method for fault prediction as described above, includes an energy flow acquisition unit, a dataset generation unit, a decision set generation unit, an indicator feedback unit, and a log generation unit, as referenced. Figure 2The system employs a multi-modal sensor network to capture multi-media waste energy flows during the switching process between main and auxiliary power systems in real time. These flows include mechanical vibration energy, flue gas waste heat, hydraulic pulsation, and electromagnetic radiation energy. Simultaneously, it collects power system operating parameters and environmental conditions. A dataset generation unit constructs a cross-media energy coupling situation model, performs energy quality classification and spatiotemporal correlation analysis on the waste energy flows, and generates a feature dataset including an energy entropy mapping table, a media compatibility matrix, and a dynamic recyclable threshold. A decision set generation unit, based on a lightweight cross-modal large model engine, calculates the optimal energy conversion path and storage strategy, generating a multi-objective optimization decision set including media conversion priority, coupling device control sequence, and system entropy increase suppression scheme. An index feedback unit drives heterogeneous energy recovery devices to perform cross-media energy coupling operations, providing real-time feedback on energy conversion efficiency and system stability indicators. Finally, a log generation unit updates the topological weights of the cross-modal large model using a federated learning framework based on the deviation between the feedback data and preset energy efficiency targets, generating an energy reuse efficiency assessment report and a model adaptive log.

[0043] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for cross-medium energy coupling and reuse during main-auxiliary power switching, characterized in that, Includes the following steps: The system uses a multimodal sensor network to capture the multi-media waste energy flow during the main and auxiliary power switching process of the ship in real time, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation and electromagnetic radiation energy, and simultaneously collects the power system operating parameters and environmental conditions. A cross-medium energy coupling situation model is constructed, and the waste energy flow is subjected to energy quality classification and spatiotemporal correlation analysis to generate a feature dataset including an energy entropy mapping table, a media compatibility matrix and a dynamic recyclable threshold. Based on a lightweight cross-modal large model engine, the optimal energy conversion path and storage strategy are calculated, and a multi-objective optimization decision set including media conversion priority, coupling device control sequence and system entropy increase suppression scheme is generated. Drive the heterogeneous energy recovery device to perform cross-medium energy coupling operation and provide real-time feedback on energy conversion efficiency and system stability indicators; Based on the deviation between the feedback data and the preset energy efficiency target, the topological weights of the cross-modal large model are updated through a federated learning framework, and an energy reuse efficiency assessment report and model adaptive log are generated.

2. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 1, characterized in that, The steps involved in capturing multi-media waste energy flows during the real-time switching process of a ship's main and auxiliary power systems using a multimodal sensor network, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation, and electromagnetic radiation energy, while simultaneously collecting power system operating parameters and environmental conditions, are as follows: At key nodes of power switching, piezoelectric energy harvesting arrays, thermoelectric power generation modules, and magnetostrictive sensors are deployed to collect the time-spectrum characteristics of vibration energy, thermal energy, and electromagnetic energy with millisecond-level accuracy. Construct an energy quality assessment model to dynamically calculate the recyclability index of waste energy based on energy fluctuation amplitude, duration, and conversion cost; When the flue gas waste heat temperature gradient is detected to be greater than the preset gradient or the hydraulic pulsation pressure peak value is greater than the preset value, the high-value energy capture mode is triggered, and the redundant sensor group is activated to increase the sampling frequency. Multi-source energy data is correlated and encoded with ship roll angle and main engine load rate using spatiotemporal stamp alignment technology.

3. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 2, characterized in that, The steps for constructing a cross-medium energy coupling situation model, performing energy quality classification and spatiotemporal correlation analysis on the waste energy flow, and generating a feature dataset including an energy entropy mapping table, a media compatibility matrix, and a dynamic recyclability threshold are as follows: A cross-medium coupling coefficient matrix is ​​established to quantify the conversion efficiency loss rate between mechanical energy-thermal energy and electrical energy-hydraulic energy. The matrix dimensions include medium type, temperature threshold and power fluctuation tolerance. An outlier energy flow detection algorithm is used to eliminate transient interference pulses, and time window integration is performed on intermittent waste energy to improve its availability; The recycling value weight of each medium's energy is calculated using the entropy weight method, and energy channels with weight values ​​higher than a preset threshold are retained. When the rate of change of ambient temperature exceeds the preset temperature, the priority coefficient of heat energy recovery is dynamically adjusted.

4. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 3, characterized in that, Based on a lightweight cross-modal large model engine, the steps for calculating the optimal energy conversion path and storage strategy, and generating a multi-objective optimization decision set including media conversion priority, coupling device control sequence, and system entropy increase suppression scheme are as follows: The knowledge distillation technique is used to compress the general energy model with hundreds of billions of parameters into a lightweight model for edge deployment, while retaining the core reasoning capability that is coupled across media. Construct a media conversion decision tree to select the optimal energy conversion chain based on real-time ship speed and grid load factor: When the main unit is switched to the auxiliary unit, the waste heat of the flue gas is preferentially introduced into the Organic Rankine Cycle (ORC) system for power generation; When the auxiliary hydraulic system is suddenly subjected to a load, the mechanical vibration energy is converted into the pressure energy of the hydraulic accumulator; A caching optimization strategy is introduced to group and quantize the prediction results of high-frequency computational coupled paths for storage, thereby reducing real-time inference latency; A scheme to suppress entropy increase is generated, which absorbs the peak value of thermal energy fluctuations through phase change thermal storage materials and uses supercapacitors to smooth out electrical energy pulses.

5. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 4, characterized in that, The steps for driving the heterogeneous energy recovery device to perform cross-medium energy coupling operation and providing real-time feedback on energy conversion efficiency and system stability indicators are as follows: During the energy coupling execution phase, the resonant frequency of the piezoelectric energy harvesting array is adjusted by adaptive impedance matching technology to keep it synchronized with the vibration spectrum of the host machine; Dynamically allocate and convert energy using a multi-port power router: The electricity generated by the power generation system is injected into the ship's lighting network; The recovered hydraulic energy is supplied to the steering gear auxiliary system; Based on the simulation of energy flow impact using a digital twin, when the predicted harmonic distortion rate of the power grid is greater than the preset rate of change, the flywheel energy storage device is triggered to absorb excess electrical energy. After the operation is completed, reverse verification is initiated to compare the deviation between the actual recovered energy and the predicted value. If the deviation is greater than the preset deviation, the decision set reconstruction is triggered.

6. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 1, characterized in that, The method further includes: Establish a cross-media energy knowledge graph for ships, integrate historical energy recovery cases, equipment compatibility rules and safety constraints, and generate interpretable decision-making basis through graph neural networks; Deploy an energy sandbox system on edge devices to perform differential privacy encryption on sensitive operation commands and store key energy conversion events through blockchain; When a new waste energy pattern is detected, an active learning mechanism is activated to collect samples and update the model. Incremental parameters are synchronized to the fleet energy cloud after being homomorphically encrypted.

7. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 6, characterized in that, The specific steps for establishing a cross-medium energy knowledge graph for ships are as follows: Extract triplet relationships from equipment manuals and fault databases to construct an ontology library covering energy types, conversion device topologies, and safety thresholds; The similarity of energy nodes is calculated using a graph attention mechanism. When the matching degree between the current working condition and historical cases is greater than the preset matching degree, the optimization control strategy is directly invoked. The safety specifications are transformed into model constraints through the rule-based distillation channel, prohibiting the generation of heat recovery commands that exceed the equipment's temperature tolerance limits.

8. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 1, characterized in that, The method further includes: Construct an energy flow visualization interface, dynamically render multi-media energy flow paths on the power system topology diagram, and report real-time energy recovery efficiency in natural language; When the energy coupling process causes system oscillations, the degradation control mode is activated: Cut off low-priority energy recovery channels; Directing excess energy to the ship's cathodic protection system; Develop contingency plans to prevent critical equipment from overloading.

9. The method for cross-medium energy coupling and reuse in main-auxiliary power switching according to claim 8, characterized in that, When the energy coupling process causes system oscillation, the steps to initiate the degraded control mode are as follows: The optimal degradation path is calculated within a preset time using a reinforcement learning algorithm, balancing energy recovery rate with system stability loss. The affected energy conversion devices are highlighted in the 3D interface, and the estimated recovery time is marked. Augmented reality devices guide crew members to manually operate critical valves.

10. A cross-medium energy coupling and reuse system for main and auxiliary power switching, characterized in that, The system is used to implement the cross-medium energy coupling and reuse method in the main-auxiliary power switching as described in any one of claims 1-9, including: The energy flow acquisition unit is used to capture multi-media waste energy flow during the main and auxiliary power switching process of a ship in real time through a multi-modal sensor network, including mechanical vibration energy, flue gas waste heat, hydraulic pulsation and electromagnetic radiation energy, and simultaneously collect power system operating parameters and environmental conditions. The dataset generation unit is used to construct a cross-medium energy coupling situation model, perform energy quality classification and spatiotemporal correlation analysis on the waste energy flow, and generate a feature dataset including an energy entropy value mapping table, a medium compatibility matrix and a dynamic recyclable threshold. The decision set generation unit is used to calculate the optimal energy conversion path and storage strategy based on a lightweight cross-modal large model engine, and generate a multi-objective optimization decision set including media conversion priority, coupling device control sequence and system entropy increase suppression scheme. The indicator feedback unit is used to drive the heterogeneous energy recovery device to perform cross-medium energy coupling operation and provide real-time feedback on energy conversion efficiency and system stability indicators. The log generation unit is used to update the topological weights of the cross-modal large model through a federated learning framework based on the deviation between the feedback data and the preset energy efficiency target, and to generate an energy reuse efficiency assessment report and model adaptive logs.

Images