Photovoltaic and energy storage hybrid power system for cruise ship and energy management method of photovoltaic and energy storage hybrid power system

Abstract

The invention relates to the technical field of ship power systems, and particularly discloses a photovoltaic and energy storage hybrid power system for a cruise ship and an energy management method of the photovoltaic and energy storage hybrid power system. The system comprises a physical energy layer, a sensing and control layer and an intelligent decision-making layer. And the intelligent decision-making layer realizes real-time calculation and performance prediction of the internal state of the equipment through the digital twin cluster of the equipment, and after fusion evaluation is performed by the system-level health situation evaluation module, a real-time energy scheduling strategy is prospectively adjusted by the prospective energy management decision-making module according to a health risk list. According to the invention, active risk avoidance based on equipment health prediction can be realized, and the operation toughness and power supply safety of the system in an offshore or river environment are improved.

Description

Technical Field

[0001] This invention belongs to the field of marine power system technology, specifically relating to photovoltaic and energy storage hybrid power systems for cruise ships and their energy management methods. Background Technology

[0002] In the field of marine propulsion and energy management, hybrid power systems have become an important development direction to meet the power demands and environmental requirements of large offshore or river platforms such as cruise ships, due to their ability to integrate multiple energy sources and improve energy efficiency and reliability. Hybrid power systems combining photovoltaic power generation and energy storage batteries, with their clean and sustainable characteristics, have shown great potential in cruise ship applications.

[0003] The photovoltaic and energy storage hybrid power system for cruise ships aims to provide a stable, efficient, and green power supply to cruise ships through the coordinated operation of solar photovoltaic arrays, energy storage battery packs, and traditional generator sets (such as diesel generators). Its core objective is to intelligently schedule and manage multiple energy sources according to sailing conditions, lighting conditions, and load requirements to achieve optimal energy utilization and economical system operation.

[0004] In existing technologies, energy management strategies for such hybrid power systems are mostly based on preset rules or simple optimization algorithms. Their design often focuses on economic efficiency under steady-state conditions, while lacking in-depth perception and proactive response to the health status and potential faults of key system equipment (such as energy storage batteries, photovoltaic inverters, and generator sets). Due to the complex and variable navigation environment at sea or on rivers, and the extremely limited maintenance and support conditions, traditional preventive maintenance models based on fixed cycles cannot accurately predict sudden equipment failures or performance degradation.

[0005] This results in the system being unable to adaptively adjust through energy management strategies to isolate risks or activate backup plans when faced with latent failures in critical equipment. In the event of a sudden failure, this directly threatens the safety of the cruise ship's power supply and navigation, potentially causing voyage disruptions or even accidents. Therefore, how to construct an intelligent system that deeply integrates equipment health status prediction and thereby achieves proactive energy management and fault risk avoidance has become a pressing technical challenge in the field of photovoltaic and energy storage hybrid power systems for cruise ships. Summary of the Invention

[0006] The purpose of this invention is to provide a photovoltaic and energy storage hybrid power system for cruise ships and its energy management method, so as to solve the technical contradiction in the prior art that the lack of in-depth perception and forward-looking response to the health status of key equipment leads to the system's inability to adaptively adjust when faced with sudden failures, thereby threatening the power supply and navigation safety of cruise ships.

[0007] This invention provides a photovoltaic and energy storage hybrid power system for cruise ships, comprising:

[0008] The physical energy layer includes photovoltaic power generation arrays, energy storage battery packs, diesel generator sets, DC buses, AC buses, bidirectional converters, photovoltaic inverters, and ship propulsion loads and daily loads.

[0009] The photovoltaic power generation array is connected to the AC bus via a photovoltaic inverter;

[0010] The energy storage battery pack is connected to the DC bus and the AC bus respectively via a bidirectional converter;

[0011] The diesel generator set is directly connected to the AC bus; the ship's propulsion load and daily load obtain power from the AC bus.

[0012] The sensing and control layer includes a multi-dimensional sensor cluster, data acquisition unit, and execution control unit deployed on various key devices;

[0013] The multi-dimensional sensor cluster is used to collect real-time operating status data of each key device.

[0014] The data acquisition unit is used to summarize and preprocess the data collected by the multi-dimensional sensor cluster;

[0015] The execution control unit is used to receive control commands and regulate each energy conversion device and power generation device;

[0016] The intelligent decision-making layer includes a cluster of digital twin devices, a system-level health status assessment module, and a forward-looking energy management decision-making module.

[0017] Preferably, the device digital twin cluster is used to build and maintain a high-fidelity virtual mapping model for each key device in the physical energy layer. The virtual mapping model is established based on the physical characteristics, electrical parameters and historical operating data of the device, and is continuously updated synchronously by receiving real-time data streams from the sensing and control layer.

[0018] The system-level health status assessment module is used to receive real-time health indicators and performance prediction data of each device from the device digital twin cluster, and has a built-in multi-source information fusion assessment algorithm.

[0019] The multi-source information fusion evaluation algorithm first defines a set of key health feature vectors for each device type. The feature vectors are composed of core state parameters calculated by the digital twin.

[0020] Furthermore, the evaluation algorithm assigns dynamic weights to each feature vector, and the dynamic weights are dynamically adjusted based on the criticality of the device in the current system energy flow and historical fault statistics.

[0021] The evaluation algorithm outputs a system-level comprehensive health index and a detailed device-level health risk list through weighted fusion calculation. The risk list clearly identifies devices that are in a warning state or are predicted to experience performance degradation, as well as their estimated remaining reliable operating time.

[0022] The forward-looking energy management decision module is used to run a two-layer optimization decision framework, and its decision logic is deeply integrated with the output of the system-level health status assessment module.

[0023] The two-layer optimization decision-making framework includes a first-layer real-time optimization layer and a second-layer forward-looking adjustment layer.

[0024] The real-time optimization layer prioritizes the optimal economic efficiency of system operation. On a time scale of seconds to minutes, it solves for the optimal power allocation command based on real-time load demand, photovoltaic predicted power, and energy storage status.

[0025] The forward-looking adjustment layer operates on an hourly to dayly timescale, and its core decision-making basis is the health risk list provided by the system-level health status assessment module.

[0026] The forward-looking adjustment layer continuously monitors the health risk list. When it is identified that any critical device is predicted to experience a significant performance degradation or failure risk greater than a preset threshold within a specific future time window, it immediately triggers a forward-looking scheduling strategy reconfiguration.

[0027] The strategy reconstruction first calls the corresponding risk mitigation plan from the preset strategy library based on the type of risky equipment and the predicted failure mode.

[0028] The core principle of the risk mitigation plan is to proactively reduce reliance on risky equipment or create maintenance windows for it by adjusting the operating strategies of other healthy equipment before a failure occurs.

[0029] Preferably, in the device digital twin cluster, the digital twin integrated electrochemical-thermal coupling model constructed for the energy storage battery pack is used to calculate the internal state of the battery in real time, including the state of charge, state of health, and state of power.

[0030] A digital twin for photovoltaic arrays integrates a power generation prediction model that considers irradiance, temperature, shading, and aging effects; a digital twin for diesel generator sets integrates a performance degradation prediction model based on vibration, oil analysis, and thermodynamic parameters.

[0031] Preferably, the strategy reconstruction process triggered by the forward adjustment layer specifically includes: converting the risk mitigation plan into adjustment instructions for the objective function or constraints of the real-time optimization layer, thereby intervening in and reconstructing the subsequent real-time optimization decision-making process.

[0032] Preferably, when the at-risk device is an energy storage battery pack and its health status is predicted to decline to a warning threshold within a specific future timeframe, the risk mitigation plan to be invoked includes:

[0033] In subsequent power allocation, the real-time optimization layer gradually reduces the charging and discharging power requirements of the high-risk battery cluster and transfers the power it bears to other healthy battery clusters.

[0034] At the same time, the real-time optimization layer of the instruction appropriately raises the lower limit of the online operating power of the diesel generator set, or adjusts the power output curve of the photovoltaic inverter, in order to partially compensate for the regulation capability lost due to the limitation of the power of high-risk battery clusters.

[0035] Preferably, when the at-risk equipment is a diesel generator set and a specific component is predicted to experience performance abnormalities within a specific future timeframe, the risk mitigation plan to be invoked includes:

[0036] In the economic objective function of the real-time optimization layer, a virtual operating cost term that is positively correlated with the predicted risk level is added to the diesel generator set;

[0037] At the same time, the command system checks in advance and ensures the availability of backup generator sets, and adjusts the charging and discharging strategies of energy storage battery packs to increase the energy storage capacity.

[0038] Preferably, the forward-looking energy management decision module also integrates a global reliability balancing algorithm;

[0039] When formulating medium- and long-term pre-scheduling plans, the global reliability balancing algorithm not only considers economic efficiency but also takes system-level health indices and the cumulative operating load of each device as constraints; the goal is to match the load allocation of each key device with its current health status during the planning period.

[0040] Preferably, the multi-dimensional sensor cluster includes an irradiance sensor and a temperature sensor deployed on the photovoltaic power generation array side, a voltage sensor, a current sensor and a temperature sensor deployed on the energy storage battery pack side, and a vibration sensor, an oil analysis sensor and a thermodynamic sensor deployed on the diesel generator set side.

[0041] Preferably, the execution control unit parses the coordination control instructions from the intelligent decision layer and converts them into low-level control signals that can be recognized by each power electronic device and generator set controller, including issuing active power and reactive power instructions to the bidirectional converter, issuing power limit instructions to the photovoltaic inverter, and issuing start, stop and power setting instructions to the diesel generator set.

[0042] This invention also provides an energy management method for the above-mentioned system, executed by the intelligent decision-making layer, comprising the following steps:

[0043] The sensor and control layer continuously collects real-time operating data from the photovoltaic array, energy storage battery pack, and diesel generator set.

[0044] By utilizing a cluster of digital twin devices, and based on the collected data, we can perform parallel calculations of the real-time internal status and short-term performance predictions of each key device.

[0045] By integrating the outputs of the digital twins of each device through the system-level health status assessment module, a system-level comprehensive health index is calculated, and a device-level health risk list is generated.

[0046] Through the real-time optimization layer of the forward-looking energy management decision module, based on the current load, real-time photovoltaic power and real-time energy storage status, the optimal real-time power allocation command is solved and issued with the goal of minimizing the current operating cost.

[0047] Through the forward-looking adjustment layer of the forward-looking energy management decision-making module, the health risk list is continuously monitored. When a preset risk trigger condition is detected, the corresponding risk mitigation plan is invoked according to the type of risky equipment.

[0048] The risk mitigation plan is transformed into an instruction to adjust the objective function or constraints of the real-time optimization layer, thereby intervening in and reconstructing the subsequent real-time optimization decision-making process;

[0049] The control unit receives and executes the coordinated control commands ultimately generated by the forward-looking energy management decision module to control the operation of each converter, inverter, and generator set.

[0050] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0051] 1. This invention, by constructing a cluster of digital twin devices and a system-level health status assessment module, achieves in-depth perception and quantitative assessment of key equipment in a cruise ship's hybrid power system, from surface operating parameters to internal health status. This changes the limitation of traditional systems that only focus on instantaneous power balance, incorporating long-term reliability factors such as equipment lifespan, performance degradation, and failure risks into the real-time monitoring scope, providing breadth and depth of status information for intelligent decision-making.

[0052] 2. This invention creatively designs a two-layer decision-making framework with a forward-looking adjustment layer at its core, enabling energy management strategies to proactively mitigate risks. When the forward-looking adjustment layer identifies potential failure risks based on health predictions, it can intervene in real-time scheduling strategies in advance. By reducing reliance on risky equipment, adjusting operating modes, or activating backup plans, it can achieve a smooth transition of the system's operating state before the actual failure occurs. This forward-looking "prediction-adjustment" management mechanism transforms failure response from passive emergency handling to proactive preventative scheduling, improving the system's operational resilience and power supply security in harsh and isolated environments at sea or on rivers.

[0053] 3. This invention, through a global reliability balancing algorithm, directly maps equipment health status to constraints or cost factors in optimized scheduling, achieving a dynamic unification and trade-off between economic and reliability objectives at the system operation level. The system no longer simply pursues the lowest fuel consumption, but intelligently seeks the optimal balance between economic costs and equipment wear and tear, as well as failure risks. This achieves optimal overall cost throughout the entire lifecycle while ensuring navigation safety and continuous power supply, extending the service life of critical equipment and reducing overall maintenance costs. Attached Figure Description

[0054] Figure 1 This is a block diagram of the physical energy layer in this invention;

[0055] Figure 2 This is a diagram of the two-layer optimization decision-making framework of the intelligent decision-making layer in this invention;

[0056] Figure 3 This is a block diagram of the system-level health status assessment module in this invention;

[0057] Figure 4 This is a flowchart of the forward adjustment layer risk contingency plan execution process in this invention;

[0058] Figure 5 This is a flowchart of the global reliability balancing algorithm in this invention. Detailed Implementation

[0059] This invention provides a photovoltaic and energy storage hybrid power system for cruise ships. Please refer to the appendix. Figures 1 to 5 The system is physically divided into three tightly coupled layers: the physical energy layer, the sensing and control layer, and the intelligent decision-making layer. These three layers together form a complete closed loop from energy generation, transmission, and consumption to state perception, intelligent analysis, and proactive control.

[0060] The physical energy layer is the physical carrier layer for the system's energy flow. This layer includes photovoltaic power generation arrays, energy storage battery packs, diesel generator sets, DC buses, AC buses, bidirectional converters, photovoltaic inverters, and ship propulsion and daily loads.

[0061] The photovoltaic power generation array consists of multiple photovoltaic modules connected in series and parallel, and is installed in the available deck area of ​​the cruise ship's superstructure.

[0062] The output of the photovoltaic array is connected to the DC input of the photovoltaic inverter.

[0063] A photovoltaic inverter is a power electronic device that converts direct current (DC) to alternating current (AC), with its AC output connected to the AC bus of the system.

[0064] Energy storage battery packs consist of multiple battery modules connected in series or in parallel to achieve the required voltage level and capacity.

[0065] The positive and negative terminals of the energy storage battery pack are connected to the DC bus.

[0066] A bidirectional converter is a converter device with bidirectional energy flow capability, where the DC side is connected to the DC bus and the AC side is connected to the AC bus.

[0067] By controlling the operating mode and power command of the bidirectional converter, the energy storage battery pack can absorb electrical energy from the AC bus for charging or release electrical energy to the AC bus for discharging.

[0068] As a traditional source of power and electricity, diesel generator sets have their generator output directly connected to the AC bus.

[0069] Ship propulsion loads, such as propulsion motors and their drives, as well as day loads, including lighting, air conditioning, galley equipment and other marine facilities, all draw power from the AC bus.

[0070] The AC bus serves as the hub for collecting and distributing electrical energy throughout the system. Its voltage and frequency are supported and regulated by the online diesel generator set or energy storage battery pack through a bidirectional converter.

[0071] The sensing and control layer serves as a bridge connecting the physical energy layer and the intelligent decision-making layer, responsible for state perception and command execution. It comprises a multi-dimensional sensor cluster, data acquisition units, and execution control units deployed on various key devices.

[0072] Multidimensional sensor clusters are distributed, heterogeneous sensor networks. Specifically, on the photovoltaic (PV) power generation array side, the sensor cluster includes irradiance sensors, PV module backsheet temperature sensors, and DC voltage and current sensors at the array output. On the energy storage battery pack side, the sensor cluster includes sensors for measuring the total voltage and current of the battery cluster, as well as individual cell voltage and temperature sensors deployed on representative battery modules. Some advanced configurations may also include battery internal impedance monitoring devices.

[0073] On the diesel generator set side, the sensor cluster includes an engine speed sensor, a cylinder pressure sensor, an exhaust gas temperature sensor, a lubricating oil pressure and temperature sensor, a coolant temperature sensor, and a triaxial accelerometer for monitoring mechanical vibration, deployed in key parts such as the engine block and generator bearing housing.

[0074] At the grid connection point and load side, the sensor cluster includes AC bus voltage sensors, current sensors, frequency sensors, and energy metering devices for measuring the power of each major load branch. The data acquisition unit consists of multiple distributed data acquisition terminals, each responsible for acquiring analog or digital signals from a nearby set of sensors.

[0075] The data acquisition terminal has a built-in signal conditioning circuit that filters, amplifies, and converts the raw signal to digital. The converted digital data is uploaded to the central data aggregation server via an industrial Ethernet or fieldbus network. The central data aggregation server runs a data preprocessing program, including timestamp alignment, data validity verification, outlier removal, and data packaging and formatting according to a preset cycle. The execution control unit is a highly reliable real-time controller that receives final coordinated control commands from the intelligent decision-making layer.

[0076] These instructions are issued in the form of digital signals or standard communication protocol messages. The execution control unit parses the instructions and converts them into low-level control signals that can be recognized by the various power electronic devices and generator set controllers. For example, for a bidirectional converter, the execution control unit issues active power command values, reactive power command values, and operating mode switching commands; for a photovoltaic inverter, it issues maximum power point tracking enable commands or power limit commands; and for a diesel generator set, it issues start, stop, speed setting, and power setting commands.

[0077] The execution control unit is connected to each controlled device via hardwiring or a high-speed communication link to ensure the real-time and deterministic execution of commands.

[0078] The intelligent decision-making layer is the brain of the system, responsible for advanced data analysis, status assessment, and optimization decisions. Deployed on the cruise ship's computing servers or edge computing platform, this layer includes a cluster of digital twin devices, a system-level health status assessment module, and a proactive energy management decision-making module. These three modules are data-driven, forming a decision-making logic chain that combines serial and parallel processing.

[0079] The device digital twin cluster constructs a high-fidelity virtual mapping model for each key device in the physical energy layer. This virtual mapping model is not a simple data mirror, but a dynamic simulation model that integrates the device's physical characteristics, electrical parameters, thermodynamic principles, and historical operating data. The virtual mapping model continuously receives real-time data streams from the central data aggregation server of the sensing and control layer via an application programming interface (API), driving the update of the model's internal state variables and achieving synchronization with the physical entity. For energy storage battery packs, the digital twin is an integrated electrochemical-thermal coupled model.

[0080] This electrochemical-thermal coupled model is based on the equivalent circuit model of the battery, combined with the battery's thermal model. The core inputs to this model are the real-time acquired total current of the battery cluster, representative cell voltages, and temperatures. Internally, the model uses a recursive least squares method to identify parameters such as the battery's equivalent internal resistance, polarization resistance, and polarization capacitance online.

[0081] Based on these parameters and the ampere-hour integration method, this electrochemical-thermal coupled model calculates the battery's state of charge (SOC) in real time, representing the percentage of remaining charge relative to the rated capacity. Simultaneously, by analyzing the battery's internal resistance changes with the number of operating cycles and temperature, and combining this with pre-set battery aging test data, the model calculates the battery's health status, representing the percentage decrease in the battery's current maximum usable capacity relative to its factory rated capacity.

[0082] Furthermore, the model calculates the battery's state of power (SOP) in real time based on the battery's real-time temperature, state of charge (SOC), and internal resistance, combined with the power capability profile provided by the battery manufacturer. This SOP represents the maximum safe power limit that the battery can sustainably input or output under current operating conditions. These internal state parameters, including SOC, health status, and SOP, are output at millisecond to second-level update frequencies, serving as the basis for assessing battery health and formulating power commands.

[0083] For photovoltaic (PV) arrays, a digital twin integrates a power generation prediction model that considers the influence of multiple factors. The core inputs to this model are real-time irradiance, ambient temperature, PV module backsheet temperature, and array output voltage and current. The model first constructs a current-voltage characteristic equation based on standard test conditions for the PV modules. Internally, the model includes a shading analysis submodule. This submodule simulates potential localized shading effects based on the PV array's installation layout and real-time solar azimuth and elevation angle data, and adjusts the characteristic curves of the affected strings accordingly.

[0084] The power generation prediction model also includes an aging degradation submodule. This submodule applies an empirical degradation function to adjust the nominal power of the modules year by year based on the cumulative operating time and historical power generation efficiency data of the photovoltaic array. Taking all these factors into account, this digital twin can predict the maximum available power curve of the photovoltaic array within the next few minutes to hours with a time resolution of minutes, and calculate in real time the ratio of the current actual output power to the ideal maximum power, i.e., the performance ratio, as an indicator for evaluating the health and efficiency of the photovoltaic array.

[0085] For diesel generator sets, a digital twin integrates a performance degradation prediction model based on multi-source signal fusion. This model receives data streams from vibration sensors, oil analysis sensors, and thermodynamic sensors. The vibration analysis submodule performs time-domain and frequency-domain analysis on the acquired vibration acceleration signals, extracting feature values ​​such as root mean square (RMS), peak factor, and kurtosis index, with particular attention paid to the amplitude variation trends of characteristic frequencies related to the main shaft bearing and piston connecting rod mechanism.

[0086] The oil analysis submodule monitors changes in the metal particle content, viscosity, and acid value of lubricating oil.

[0087] The thermodynamics submodule analyzes the uniformity of exhaust gas temperature and combustion pressure in each cylinder.

[0088] The performance degradation prediction model internally establishes a rule base linking these characteristic parameters with known failure modes, such as bearing wear, piston ring wear, and injector clogging. By continuously tracking the rate of change of these characteristic parameters and their deviation from the normal baseline, the model can predict the likelihood of a significant performance degradation or failure in a specific component and estimate its remaining reliable operating time. The prediction results are output in the form of a risk level and a time window.

[0089] The system-level health status assessment module is the hub for transforming device status information into system-level risk perception. This module receives real-time health indicators and performance prediction data streams from each device, output in parallel from the device digital twin cluster. Please refer to the appendix. Figure 3 The system-level health status assessment module incorporates a multi-source information fusion assessment algorithm. The execution process of the multi-source information fusion assessment algorithm consists of three stages: feature extraction, dynamic weight allocation, and fusion calculation.

[0090] During the feature extraction phase, the multi-source information fusion evaluation algorithm predefines a set of key health feature vectors for each equipment type. For energy storage battery packs, the feature vectors include the health status value, the rate of decline in health status, the rate of increase in internal resistance, and the temperature non-uniformity coefficient. For photovoltaic arrays, the feature vectors include the performance ratio value, the performance ratio decline trend, and the series branch current mismatch. For diesel generator sets, the feature vectors include the amplitude of the main bearing vibration characteristic frequency, the rate of increase in lubricating oil metal particle concentration, and the exhaust gas temperature deviation. These feature vectors are all calculated by the corresponding digital twins.

[0091] During the dynamic weight allocation phase, the algorithm does not assign fixed weights to each feature vector, but instead introduces a dynamic weight adjustment mechanism. This dynamic weight adjustment mechanism is based on two core factors.

[0092] The first factor is the criticality of the equipment in the current system energy flow. Criticality is a dynamic variable, fed back in real-time by the forward-looking energy management decision-making module's optimization layer. For example, under the current load conditions and energy dispatch strategy, if the energy storage battery pack is undertaking the main peak-shaving and frequency regulation tasks, its criticality weight will be temporarily increased.

[0093] The second factor is historical fault statistics. The system maintains a historical fault database, recording the fault modes and mean time between failures (MTBF) of various equipment under similar operating conditions. Equipment types with historically high failure rates, or those approaching historical fault thresholds under current operating parameters, receive increased weight. Dynamic weights are recalculated every five minutes.

[0094] During the fusion computing phase, the algorithm performs weighted fusion of the feature vectors of each device with their corresponding dynamic weights.

[0095] First, for each device, a device-level health score is calculated, with a lower score indicating a higher health risk.

[0096] Then, the health scores of all key devices are weighted and summed again, where the weight of each device is related to its redundancy and substitutability in the system architecture, and the final output is a system-level comprehensive health index ranging from 0 to 100.

[0097] A comprehensive health index greater than 80 indicates overall system health, a value between 60 and 80 indicates a need for attention, and a value less than 60 indicates a high level of systemic risk. Simultaneously, the algorithm generates a detailed list of device-level health risks.

[0098] The equipment-level health risk list is presented in a structured table format, with each row corresponding to an instance of the equipment being assessed. Column fields include equipment identifier, equipment type, current health score, key risk characteristics, predicted risk events, estimated remaining reliable uptime, and risk level. Risk levels are categorized into three levels—Attention, Warning, and High Risk—based on the estimated remaining uptime and the severity of the risk events. This list serves as the core input for advanced decision-making in the proactive energy management decision-making module.

[0099] The forward-looking energy management decision-making module is the core decision-making hub of the entire intelligent decision-making layer, and its decision-making logic is deeply integrated with the output of the system-level health status assessment module. Please refer to the appendix. Figure 2 This module operates on a two-layer optimization decision-making framework: a real-time optimization layer and a forward-looking adjustment layer. These two layers are both independent and closely coordinated in terms of time scale, optimization objectives, and decision-making logic.

[0100] The real-time optimization layer prioritizes optimal system economics and operates continuously on a timescale of seconds to minutes. Inputs to the real-time optimization layer include: real-time load power demand from the sensing and control layers, actual photovoltaic output power, real-time state of charge and power status of the energy storage battery pack, real-time operating status and efficiency curves of the diesel generator set, and real-time fuel price information. This layer incorporates a real-time rolling optimization engine.

[0101] This real-time rolling optimization engine solves an optimization problem every control cycle, for example every 10 seconds, with the objective of minimizing the total operating cost of the system in the current moment and the near-term forecast period. The total operating cost mainly includes the fuel consumption cost of the diesel generator set, which has a non-linear relationship with the generator set's output power and is determined by the generator set's fuel consumption rate curve.

[0102] The decision variables for optimization problems include:

[0103] The output power of each online diesel generator set, the charging and discharging power of the energy storage battery pack, and the actual output power setting value of the photovoltaic inverter.

[0104] The constraints of the optimization problem include:

[0105] The real-time power balance constraint of the AC bus is that the sum of the output power of all generating units is equal to the sum of the power consumed by all loads.

[0106] Minimum and maximum technical output constraints for diesel generator sets;

[0107] Constraints on the upper and lower limits of the state of charge of energy storage battery packs, and limits on the charging and discharging power;

[0108] And the maximum available power constraint of the photovoltaic array.

[0109] The real-time optimization engine uses numerical optimization algorithms such as linear programming or quadratic programming to solve problems quickly, and at the end of each control cycle, it sends the optimal power allocation command obtained from the solution to the execution control unit.

[0110] The forward-looking adjustment layer operates on an hourly to daily timescale, with its core decision-making based on a health risk list that is periodically updated by the system-level health status assessment module. As a monitoring and strategy refactoring module, the forward-looking adjustment layer does not directly generate second-level control commands, but rather indirectly and profoundly influences the overall operational trajectory of the system by modifying the operating environment and decision-making rules of the real-time optimization layer.

[0111] Please refer to the attached document. Figure 4The forward-looking adjustment layer continuously monitors the health risk list. Internally, it has risk event triggers that compare the estimated remaining reliable uptime and risk level of each risk record in the list with a preset threshold matrix. When it is identified that any critical equipment is predicted to experience a significant performance degradation or a failure risk level exceeding the preset high-risk threshold within a specific future time window, such as the next 24 or 48 hours, the trigger is immediately activated and sends an event alert to the forward-looking adjustment layer's policy scheduler.

[0112] Upon receiving an alert, the policy scheduler initiates a proactive scheduling policy reconfiguration process. The first step in this process is risk diagnosis and contingency plan matching. Based on the unique identifier of the at-risk device, the policy scheduler retrieves its device type and predicted failure mode. Then, the policy scheduler retrieves the risk mitigation contingency plan from its locally stored pre-defined policy library that perfectly matches the "device type-failure mode" combination.

[0113] The strategy library is a collection of pre-designed and simulated contingency plans. Each plan encapsulates a series of adjustment logics and objectives for a specific risk scenario. The core design principle of the contingency plans is to proactively reduce the system's performance dependence on the equipment at risk before the predicted failure actually occurs, or to create a favorable operating window for preventative maintenance of the equipment at risk, by proactively adjusting the operating strategies of other healthy equipment.

[0114] Let's take a risk scenario involving an energy storage battery pack as an example for a detailed explanation of the extreme cases. Assume that the energy storage battery pack contains four independent battery clusters, numbered A, B, C, and D. The digital twin of battery cluster A, by analyzing its health status decline curve and internal resistance growth trend, predicts that its health status will decrease from the current 85% to 75% within the next 18 hours, reaching the system's preset 75% health status warning threshold.

[0115] Based on predictions and considering the critical role of battery cluster A, which currently handles 30% of the system's power regulation, the system-level health assessment module marks it as high-risk and adds it to the health risk list, estimating a remaining reliable operating time of 18 hours. The risk event trigger in the forward-looking adjustment layer detects this high-risk record and activates immediately.

[0116] The policy scheduler invokes the "Battery Cluster Health Status Approaching Threshold" risk contingency plan. This risk contingency plan contains a series of structured adjustment instructions.

[0117] The first instruction is an adjustment to the power allocation constraints of the real-time optimization layer. The pre-set instruction requires the real-time optimization layer to dynamically limit the upper limit of the available charge / discharge power of battery cluster A in subsequent optimization calculations. Specifically, the initial limit is 80% of its nominal power, and it gradually tightens to 50% or even lower as time approaches the predicted failure point. This limit is implemented by modifying the power inequality constraints corresponding to battery cluster A in the real-time optimization problem.

[0118] The second directive concerns power transfer and compensation strategies. The contingency plan requires that the regulation capacity lost due to power limitations on battery cluster A must be compensated for by other resources within the system. Therefore, the plan will instruct the real-time optimization layer to increase the power regulation capacity weights of other healthy battery clusters (B, C, and D) in the optimization model, encouraging the optimization algorithm to prioritize power allocation to them. Simultaneously, the plan may instruct a moderate increase in the minimum operating power limit of online diesel generator sets, for example, from 30% to 40% of rated power, to allow them to operate in a more efficient and responsive range, providing a more stable base load power and reducing the need for frequent charging and discharging of the energy storage system.

[0119] In addition, the risk contingency plan will adjust the power output curve strategy of the photovoltaic inverter. When there is sufficient sunlight, the photovoltaic power generation will be slightly over-generated, and the excess power will be used to charge the healthy battery clusters first, so as to reserve energy in advance for unforeseen needs.

[0120] The third instruction is the generation of maintenance recommendations. The contingency plan will automatically generate a maintenance work order, alerting the ship's engine room management personnel to a health status warning for battery cluster A. It recommends a focused inspection and maintenance of the battery cluster's connection terminals, cooling system, and battery management system during the cruise ship's next scheduled berthing period, such as 6 hours later. All these adjustment instructions are packaged into a strategy adjustment package.

[0121] The look-ahead adjustment layer sends the strategy adjustment package to the real-time optimization layer. Upon receiving the package, the real-time optimization layer parses and integrates the instructions. For constraint adjustment instructions, the real-time optimization layer directly modifies the corresponding constraint parameters of its internal optimization model. For objective function weight adjustment instructions, the real-time optimization layer may add virtual penalty costs to its economic objective function for operating items involving high-risk equipment.

[0122] For example, in addition to the original fuel cost, an extra virtual cost coefficient is added to the power of high-risk battery cluster A. This virtual cost coefficient is positively correlated with the predicted risk level. Thus, in each subsequent rolling optimization solution, the optimization algorithm, when pursuing economic optimization, will naturally tend to reduce the use of high-risk equipment because it appears "more expensive." In this way, the strategic intent of the forward adjustment layer is seamlessly embedded into the rapid decision-making logic of the real-time optimization layer, achieving a seamless and smooth transition of energy management strategy from forward warning to real-time execution.

[0123] Let's take a diesel generator set risk scenario as an example for further analysis. Assume that the digital twin of diesel generator set No. 1, based on the spectral analysis of the main shaft bearing vibration acceleration signal, finds that the amplitude of the characteristic frequency of the bearing outer ring fault shows a continuous linear increasing trend over the past 12 hours. Based on a comparison of the growth rate with historical fault data, the model predicts that the bearing has a high probability of abnormal wear aggravation within the next 36 to 48 hours, potentially leading to excessive vibration or even shutdown. The system-level health status assessment module marks this prediction as a warning-level risk and adds it to the list. Upon triggering the forward-looking adjustment layer, the "Abnormal Vibration Trend of Generator Set Bearing" risk contingency plan is invoked.

[0124] The core strategy of this risk contingency plan is to reduce the operational stress on high-risk units and prepare backup plans.

[0125] The first instruction of the risk mitigation plan is to modify the objective function of the real-time optimization layer. The plan requires the real-time optimization layer to multiply the output power term of diesel generator set No. 1 by a risk factor, for example, 1.2, when calculating fuel costs. This means that, from the optimization algorithm's perspective, the "cost" of using this generator set to generate electricity has increased by 20%. This will cause the optimization results to prioritize the use of other healthy generator sets, or to make greater use of energy storage and photovoltaics, thereby reducing the number of times diesel generator set No. 1 is called and its load rate, allowing it to operate under more moderate conditions and slowing down the bearing degradation process.

[0126] The second instruction in the risk contingency plan is the verification and pre-configuration of backup resources. The risk contingency plan will send a soft-start self-test instruction to the backup generator set, such as the No. 2 diesel generator set, through the execution control unit to confirm that it is in hot standby status and that the fuel, lubricating oil, and cooling water systems are normal.

[0127] At the same time, the risk contingency plan will instruct adjustments to the energy management strategy of the energy storage battery pack. Specifically, in subsequent real-time optimization, the lower limit of the target state of charge of the energy storage battery pack will be appropriately increased, for example, from 40% to 50%, and the energy storage will be charged to a higher level, such as above 70%, before the predicted risk time window arrives.

[0128] This is equivalent to pre-stocking an extra "energy buffer" for the system. If Unit 1 needs to be taken offline for emergency maintenance, the energy storage can immediately provide considerable power support to ensure uninterrupted power supply for propulsion and critical loads.

[0129] The third instruction in the risk contingency plan is to generate early warning and enhanced monitoring instructions, prompting the engine room crew to increase the frequency of on-site monitoring of vibration and temperature of Unit 1.

[0130] The forward-looking energy management decision-making module also integrates a global reliability balancing algorithm, which operates on a longer timescale to formulate pre-scheduling plans for the next few hours to days. Please refer to the appendix. Figure 5 This global reliability balancing algorithm, when formulating plans, not only considers traditional economic indicators such as predicted load curves, photovoltaic power generation forecasts, and fuel price fluctuations, but more importantly, it incorporates system-level health indices and the cumulative operating load of each key device as core constraints into the optimization model. The goal of the global reliability balancing algorithm is to generate a pre-scheduling plan that, while meeting future energy demands, proactively and intelligently distributes operating loads among devices, ensuring that load allocation matches the current health status of the equipment.

[0131] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0132] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A photovoltaic and energy storage hybrid power system for cruise ships, characterized in that, include: The physical energy layer includes photovoltaic power generation arrays, energy storage battery packs, diesel generator sets, DC buses, AC buses, bidirectional converters, photovoltaic inverters, and ship propulsion loads and daily loads. The photovoltaic power generation array is connected to the AC bus via a photovoltaic inverter; The energy storage battery pack is connected to the DC bus and the AC bus respectively via a bidirectional converter; The diesel generator set is directly connected to the AC bus; the ship's propulsion load and daily load obtain power from the AC bus. The sensing and control layer includes a multi-dimensional sensor cluster, data acquisition unit, and execution control unit deployed on various key devices; The multi-dimensional sensor cluster is used to collect real-time operating status data of each key device. The data acquisition unit is used to summarize and preprocess the data collected by the multi-dimensional sensor cluster; The execution control unit is used to receive control commands and regulate each energy conversion device and power generation device; The intelligent decision-making layer includes a cluster of digital twin devices, a system-level health status assessment module, and a forward-looking energy management decision-making module.

2. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 1, characterized in that, The device digital twin cluster is used to build and maintain a high-fidelity virtual mapping model for each key device in the physical energy layer. The virtual mapping model is built based on the device's physical characteristics, electrical parameters and historical operating data, and is continuously updated synchronously by receiving real-time data streams from the sensing and control layer. The system-level health status assessment module is used to receive real-time health indicators and performance prediction data of each device from the device digital twin cluster, and has a built-in multi-source information fusion assessment algorithm. The multi-source information fusion evaluation algorithm first defines a set of key health feature vectors for each device type. The feature vectors are composed of core state parameters calculated by the digital twin. Furthermore, the evaluation algorithm assigns dynamic weights to each feature vector, and the dynamic weights are dynamically adjusted based on the criticality of the device in the current system energy flow and historical fault statistics. The evaluation algorithm outputs a system-level comprehensive health index and a detailed device-level health risk list through weighted fusion calculation. The risk list clearly identifies devices that are in a warning state or are predicted to experience performance degradation, as well as their estimated remaining reliable operating time. The forward-looking energy management decision module is used to run a two-layer optimization decision framework, and its decision logic is deeply integrated with the output of the system-level health status assessment module. The two-layer optimization decision-making framework includes a first-layer real-time optimization layer and a second-layer forward-looking adjustment layer. The real-time optimization layer prioritizes the optimal economic efficiency of system operation. On a time scale of seconds to minutes, it solves for the optimal power allocation command based on real-time load demand, photovoltaic predicted power, and energy storage status. The forward-looking adjustment layer operates on an hourly to dayly timescale, and its core decision-making basis is the health risk list provided by the system-level health status assessment module. The forward-looking adjustment layer continuously monitors the health risk list. When it is identified that any critical device is predicted to experience a significant performance degradation or failure risk greater than a preset threshold within a specific future time window, it immediately triggers a forward-looking scheduling strategy reconfiguration. The strategy reconstruction first calls the corresponding risk mitigation plan from the preset strategy library based on the type of risky equipment and the predicted failure mode. The core principle of the risk mitigation plan is to proactively reduce reliance on risky equipment or create maintenance windows for it by adjusting the operating strategies of other healthy equipment before a failure occurs.

3. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 2, characterized in that, In the device digital twin cluster, the digital twin integrated electrochemical-thermal coupling model constructed for the energy storage battery pack is used to calculate the internal state of the battery in real time, including the state of charge, state of health, and state of power. A digital twin for photovoltaic arrays integrates a power generation prediction model that considers irradiance, temperature, shading, and aging effects; a digital twin for diesel generator sets integrates a performance degradation prediction model based on vibration, oil analysis, and thermodynamic parameters.

4. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 3, characterized in that, The strategy reconstruction process triggered by the forward adjustment layer specifically includes: transforming risk mitigation plans into adjustment instructions for the objective function or constraints of the real-time optimization layer, thereby intervening in and reconstructing the subsequent real-time optimization decision-making process.

5. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 4, characterized in that, When the at-risk device is an energy storage battery pack and its health status is predicted to decline to a warning threshold within a specific future timeframe, the risk mitigation plans to be invoked include: In subsequent power allocation, the real-time optimization layer gradually reduces the charging and discharging power requirements of the high-risk battery cluster and transfers the power it bears to other healthy battery clusters. At the same time, the real-time optimization layer of the instruction appropriately raises the lower limit of the online operating power of the diesel generator set, or adjusts the power output curve of the photovoltaic inverter, in order to partially compensate for the regulation capability lost due to the limitation of the power of high-risk battery clusters.

6. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 5, characterized in that, When the at-risk equipment is a diesel generator set and a specific component is predicted to experience performance abnormalities within a specific future timeframe, the risk mitigation plans to be invoked include: In the economic objective function of the real-time optimization layer, a virtual operating cost term that is positively correlated with the predicted risk level is added to the diesel generator set; At the same time, the command system checks in advance and ensures the availability of backup generator sets, and adjusts the charging and discharging strategies of energy storage battery packs to increase the energy storage capacity.

7. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 6, characterized in that, The forward-looking energy management decision-making module also integrates a global reliability balancing algorithm; When formulating medium- and long-term pre-scheduling plans, the global reliability balancing algorithm not only considers economic efficiency, but also takes system-level health index and the cumulative operating load of each device as constraints. The goal is to match the load distribution of key equipment during the planning period with its current health status.

8. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 7, characterized in that, The multi-dimensional sensor cluster includes irradiance and temperature sensors deployed on the photovoltaic power generation array side, voltage, current and temperature sensors deployed on the energy storage battery pack side, and vibration, oil analysis and thermodynamic sensors deployed on the diesel generator set side.

9. The photovoltaic and energy storage hybrid power system for cruise ships according to claim 8, characterized in that, The execution control unit parses the coordination control instructions from the intelligent decision-making layer and converts them into low-level control signals that can be recognized by each power electronic device and generator set controller. These include issuing active and reactive power instructions to the bidirectional converter, issuing power limit instructions to the photovoltaic inverter, and issuing start, stop, and power setting instructions to the diesel generator set.

10. An energy management method applied to the system according to any one of claims 1 to 9, characterized in that, Executed by the intelligent decision-making layer, the process includes the following steps: The sensor and control layer continuously collects real-time operating data from the photovoltaic array, energy storage battery pack, and diesel generator set. By utilizing a cluster of digital twin devices, and based on the collected data, we can perform parallel calculations of the real-time internal status and short-term performance predictions of each key device. By integrating the outputs of the digital twins of each device through the system-level health status assessment module, a system-level comprehensive health index is calculated, and a device-level health risk list is generated. Through the real-time optimization layer of the forward-looking energy management decision module, based on the current load, real-time photovoltaic power and real-time energy storage status, the optimal real-time power allocation command is solved and issued with the goal of minimizing the current operating cost. Through the forward-looking adjustment layer of the forward-looking energy management decision-making module, the health risk list is continuously monitored. When a preset risk trigger condition is detected, the corresponding risk mitigation plan is invoked according to the type of risky equipment. The risk mitigation plan is transformed into an instruction to adjust the objective function or constraints of the real-time optimization layer, thereby intervening in and reconstructing the subsequent real-time optimization decision-making process; The control unit receives and executes the coordinated control commands ultimately generated by the forward-looking energy management decision module to control the operation of each converter, inverter, and generator set.

Images