A ship start-stop control method based on operating condition prediction

Abstract

The present invention discloses a ship start-stop control method based on operating condition prediction. The method comprises: establishing a dynamic model for operating condition prediction based on ship operation data, and obtaining the predicted operating condition of the ship based on the model; setting the objective function and constraints for power distribution based on the predicted operating condition, and combining a quantum annealing algorithm to find the optimal power distribution path in a complex network to obtain an optimized power distribution result, and controlling the ship's start-stop operation based on the power distribution result. The present invention proposes a ship start-stop control method based on operating condition prediction. The method predicts the future operating condition of the ship through a dynamic model, and constructs an objective function and constraints to optimize global energy consumption and network topology; the method uses a quantum annealing algorithm to find the optimal power path, thereby accurately allocating equipment power, adjusting the ship's start-stop in real time, reducing energy consumption and wear, and improving the overall efficiency of the power system. This method can solve the problem of difficulty in improving the operating efficiency of ship start-stop.

Description

Technical Field

[0001] The present invention relates to the field of automation technology, and in particular to a ship start-stop control method based on working condition prediction. Background Art

[0002] Controlling a ship's start and stop operations is a crucial component of an automation system. Through commands and algorithms, it automatically starts and stops the ship's power system, improving the ship's automation level. A reasonable start-stop control strategy can optimize the performance of a ship's power system, reduce unnecessary energy loss, and improve its operational efficiency. For example, by promptly shutting down the power system when a ship is docked or waiting for an extended period, fuel consumption and emissions can be reduced. Current ship start and stop operations rely primarily on manual judgment, simple sensor feedback, and rule-based control systems. Crew members must manually control the start and stop of the power system, taking into account the ship's operating status, navigation requirements, and environmental conditions. Traditional systems utilize only simple sensors to monitor key parameters such as engine temperature and oil pressure, automatically triggering start and stop operations once these parameters reach preset thresholds. Some ships employ rule-based control systems, determining start and stop timing based on preset logic and safety requirements to achieve intelligent management of the power system.

[0003] However, manual judgment of ship start and stop is time-consuming and experience-dependent, making it difficult to respond swiftly to rapidly changing operating conditions. Simple sensors only monitor limited parameters, making it difficult to fully understand the complex state of the power system. At the same time, rule-based control systems lack real-time learning and adaptive capabilities, and are unable to flexibly adjust strategies based on real-time data. Therefore, current ship start and stop control methods are inadequate to improve operational efficiency, and urgently need innovation to achieve more intelligent and efficient start and stop management to ensure the safety and economy of ship operations. Summary of the Invention

[0004] The present invention provides a ship start-stop control method based on operating condition prediction to solve the problem of difficulty in improving the operating efficiency of ship start-stop.

[0005] To achieve the above objectives, the present application provides a ship start-stop control method based on operating condition prediction, comprising:

[0006] Obtain current ship operation data from the ship's power system;

[0007] Based on the current ship operation data, an operating condition prediction is performed using a dynamic model to obtain a predicted operating condition of the ship; wherein the dynamic model is established by encoding the network state of the power system based on historical ship operation data and capturing the network dynamic characteristics through quantum computing;

[0008] Establishing an objective function and a constraint set for power allocation based on the predicted operating conditions; wherein the objective function is established by minimizing a global energy loss value, maximizing network topology entropy, and suppressing target node overload;

[0009] According to the objective function and the constraint set, combined with a quantum annealing algorithm, an optimal power allocation path is found in the topology of the power system;

[0010] Path decoding and device power mapping are performed on the optimal power allocation path to obtain a power allocation result, and the start and stop operations of the ship are controlled according to the power allocation result.

[0011] This invention uses a dynamic model to accurately predict future operating conditions based on current ship operating data, providing a scientific basis for optimizing start-up and shutdown operations. The dynamic model uses historical ship operating data for learning and encodes the network state of the power system. This process can uncover historical patterns and characteristics of ship operation, providing a solid foundation for operating condition prediction. Furthermore, quantum computing, with its powerful parallel processing capabilities and computational speed, can efficiently capture and analyze the dynamic network characteristics of the power system. The established objective function aims to minimize global energy loss, maximize network topology entropy, and prevent node overload, ensuring efficient and safe operation of the power system. A quantum annealing algorithm is used to find the optimal power allocation path within the power system topology. This algorithm is efficient and capable of handling complex optimization problems, ensuring the optimal solution is found. Through path decoding and device power mapping, the optimal power allocation path is converted into specific device power allocation results. This process ensures that each device receives the power most appropriate for its operating state, avoiding power overload or underload, thereby improving the efficiency of the entire power system. Finally, ship start-up and shutdown operations are adjusted in real time based on the optimal power allocation results, avoiding unnecessary energy consumption and wear, and improving operational efficiency.

[0012] Compared with existing technologies, the present invention predicts the future operating conditions of ships through dynamic models and constructs objective functions and constraints to optimize global energy consumption and network topology. It uses a quantum annealing algorithm to find the optimal power path, thereby accurately allocating equipment power and adjusting the start and stop of ships in real time, reducing energy consumption and wear, and improving the overall efficiency of the power system. Therefore, it can solve the problem of difficulty in improving the operating efficiency of ship start-up and shutdown.

[0013] As a preferred solution, based on the current ship operation data, the operating condition prediction is performed through a dynamic model to obtain the predicted operating condition of the ship, specifically:

[0014] Performing data aggregation and weighted calculation on the current ship operation data according to the dynamic model to obtain a time series enhanced embedding and a dynamic weight matrix;

[0015] Using quantum technology, high-betweenness nodes in the ship operation data are caused to enter a superposition state, and according to the timing enhanced embedding and the dynamic weight matrix, the high-betweenness nodes are controlled to perform quantum transitions and classical transfers to obtain a transfer probability matrix;

[0016] Path sampling is performed according to the transfer probability matrix, and the predicted operating condition of the ship is obtained by combining the data obtained by the path sampling with the time series enhanced embedding calculation.

[0017] The introduction of a dynamic weight matrix in this preferred solution enables the model to capture the temporal evolution of a ship's operating state. By utilizing quantum technology to place high-betweenness nodes into a superposition state, the parallel processing capabilities of quantum computing can be fully utilized to improve computational efficiency. This is crucial for processing large-scale, highly complex ship operation data. Furthermore, the combination of quantum transitions and classical transfers enables the model to explore a wider range of possibilities between ship operating states, thereby more comprehensively reflecting the system's dynamic behavior.

[0018] As a preferred solution, data aggregation and weighted calculation are performed on the current ship operation data according to the dynamic model to obtain a time series enhanced embedding and a dynamic weight matrix, specifically:

[0019] Aggregating adjacent node data of each node in the current ship operation data according to the dynamic model to obtain a node embedding vector;

[0020] Calculate a query vector, a key vector, and a value vector based on the node embedding vector and a preset learnable parameter matrix;

[0021] According to the query vector, the key vector and the value vector, a temporal enhanced embedding of all nodes in the dynamic system is calculated by a temporal attention algorithm;

[0022] The original edge weights are subjected to attention weighting and normalization processing to obtain a dynamic weight matrix; wherein the original edge weights are dynamic weight values ​​assigned to the connection relationships between mechanical components in the power system.

[0023] In this preferred solution, node embedding vectors, as representations of node features, capture the uniqueness of nodes and their position within the system. This ability to express these features enables the model to more accurately understand the roles and functions of nodes, providing precise input for subsequent calculations. The original edge weights are then subjected to attention weighting and normalization to produce a dynamic weight matrix. This process not only preserves the static information about the connections between mechanical components but also reflects how these relationships change over time and across operational states through dynamic weight assignment.

[0024] As a preferred solution, path sampling is performed according to the transition probability matrix, and the predicted operating condition of the ship is obtained by combining the data obtained from the path sampling with the time series enhanced embedding calculation, specifically:

[0025] Perform path sampling according to the transition probability matrix to obtain path distribution and node access frequency;

[0026] According to the time series enhanced embedding and the path distribution, the path similarity and the time series embedding features are fused through Gaussian process regression to obtain the initial predicted operating condition of the ship;

[0027] The nodes of the initial predicted working condition are marked and the paths are bolded according to the node access frequency to obtain a predicted working condition with a dynamic topology graph.

[0028] This preferred solution uses a transition probability matrix for path sampling, capturing the transition patterns between ship operating states and helping to reflect the system's dynamics and uncertainty. Gaussian process regression can integrate path similarity and time series embedding features, which reflect the characteristics of ship operation at the path and time series levels, respectively, helping to improve prediction accuracy and robustness. The dynamic topology map not only provides a comprehensive view of the ship's operating status but also labels nodes and bolds paths based on node access frequency, helping the system quickly identify key states and paths, enabling more informed decisions.

[0029] As a preferred solution, the transition probability matrix is ​​specifically:

[0030]

[0031] in, is the probability value, is the quantum unitary operator; Indicates time When the slave node To Node The classic transfer weight of ; For nodes The sum of the weights of all outgoing edges.

[0032] As a preferred solution, the dynamic model includes an input layer, a wandering layer and an output layer;

[0033] The input layer is established by capturing the temporal enhanced embedding and dynamic weight matrix of ship operation data through temporal graph convolutional coding and attention mechanism;

[0034] The walking layer is established by calculating a mixing probability based on the timing enhanced embedding and the dynamic weight matrix, and performing quantum jumps and classical transfers through the mixing probability to generate a walking path sequence with timing marks;

[0035] The output layer is established based on the wandering path sequence by fusing path similarity and embedded features through Gaussian process regression to predict the probability of future working conditions.

[0036] This preferred solution uses time-series graph convolutional coding, and the input layer can effectively extract time-series features from historical ship operation data. These features are crucial for understanding the dynamic changes in the ship's operating status. The time-series labeled wandering path sequence generated by the wandering layer not only retains the time-series information of the ship's operating status, but also reflects the potential connections between states through the diversity of wandering paths. This is of great significance for understanding the complexity of ship operations and predicting future operating conditions. The output layer uses Gaussian process regression, combining path similarity with embedded features to predict the probability of future operating conditions. This method performs well when dealing with small sample data and nonlinear relationships, and has high prediction accuracy and robustness.

[0037] As a preferred solution, the objective function is established by minimizing the global energy loss value, maximizing the network topology entropy and suppressing the overload of the target node, specifically:

[0038] With the goal of minimizing global energy loss, a first sub-objective function is established according to the node quantum state overlap in the current ship operation data;

[0039] With the goal of maximizing the network topology entropy, the load of high-number nodes is suppressed according to the node degree and betweenness centrality parameters, and the second sub-objective function is established;

[0040] Applying nonlinear penalties to the target risk node according to the predicted working condition to establish a third sub-objective function;

[0041] The objective function is established based on the first sub-objective function, the second sub-objective function and the third sub-objective function.

[0042] In this preferred solution, the first sub-objective function is dedicated to minimizing global energy loss. By accurately calculating the quantum state overlap of each node in the current ship operation data, energy can be more effectively distributed and managed, thereby reducing unnecessary energy waste. The second sub-objective function aims to maximize the network topology entropy, which helps to improve the stability and robustness of the network structure. By reasonably suppressing the load of high-number nodes, the failure of key nodes in the network due to overload can be avoided, thereby ensuring the normal operation of the entire network. The third sub-objective function can more effectively identify and manage potential risks by applying nonlinear penalties to target risk nodes in the predicted working conditions.

[0043] As a preferred solution, the first sub-objective function is specifically:

[0044]

[0045] in, For the edge The power flow, the edge represents the channel of power flow or energy transmission; is the capacity of the edge; For nodes The node represents the starting point or end point of power flow or energy transfer; E is the set of edges.

[0046] As a preferred solution, the constraint set consists of hard constraints and soft constraints;

[0047] Wherein, the hard constraint is established based on the load power limit power change value and based on the node power limit device capacity;

[0048] The soft constraint is established by limiting the temporal entropy change value of the power system according to a preset topological entropy change threshold.

[0049] This preferred solution prevents system instability due to excessive power fluctuations by limiting the power variation range. Ensuring that each device's power output does not exceed its designed capacity prevents device overload, extends its service life, and improves the overall reliability of the system. Therefore, setting hard constraints helps prevent the system from entering a dangerous state. By limiting the variation range of temporal entropy, the system maintains a certain degree of stability and predictability during the optimization process. Soft constraints are more flexible than hard constraints and can, to a certain extent, allow for minor changes in system state to adapt to fluctuations in external conditions.

[0050] As a preferred solution, path decoding and device power mapping are performed on the optimal power allocation path to obtain a power allocation result, specifically:

[0051] performing binary decoding and power mapping on the optimal power allocation path to obtain an initial power allocation result;

[0052] According to a preset method, the power distribution of the conflicting nodes in the initial power distribution result is adjusted through topology dependency verification, and the initial power distribution result is scaled or reduced according to the current total load of the power system to obtain the power distribution result.

[0053] This preferred solution's topology dependency check ensures the rationality of the power allocation results, avoids power allocation errors caused by node conflicts, and guarantees data accuracy. The initial power allocation results are scaled or reduced based on the current total load of the power system, allowing the power allocation results to dynamically adapt to the actual load requirements of the system. This flexibility helps maintain system stability and efficiency as loads change, improving system adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of a ship start-stop control method based on operating condition prediction provided by an embodiment of the present application;

[0055] Figure 2 This is a structural diagram of a ship start-stop control device based on operating condition prediction provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0057] In the description of this application, it should be understood that the terms "first," "second," and "third" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature defined as "first," "second," and "third" may explicitly or implicitly include one or more of such features. In the description of this application, unless otherwise specified, "several" means two or more.

[0058] Example 1:

[0059] See also Figure 1 The embodiment of the present application provides a ship start-stop control method based on working condition prediction, including S1 to S5, and the specific implementation steps are as follows:

[0060] S1. Obtain current ship operation data from the ship's power system;

[0061] Step S1 of the embodiment of the present application is specifically as follows:

[0062] In the ship's power system, a supervisory control and data acquisition system (SCADA) or other preset methods are used to collect current ship operation data in real time, including equipment operation data, time-varying network snapshots, node characteristics, and time series data, and store the current ship operation data in a database.

[0063] Among them, the current ship operation data includes the speed, temperature, pressure of physical equipment, power transmission efficiency and loss, as well as energy and information flow records between equipment, which are used to reflect the equipment status.

[0064] The network snapshot includes the topological structure of the dynamic system over time, including nodes and connections, which is used to reflect state changes.

[0065] Node characteristics include node type, capacity, and performance parameters, which are used to help the model understand behavior.

[0066] Time series include system, equipment and functional module status data arranged in chronological order, which are used to adapt the model.

[0067] S2. Based on the current ship operating data, the operating condition is predicted through a dynamic model to obtain the predicted operating condition of the ship. The dynamic model is established by encoding the network status of the power system based on historical ship operating data and capturing the dynamic characteristics of the network through quantum computing.

[0068] Step S2 of the embodiment of the present application includes S2.1 to S2.4, specifically:

[0069] S2.1. Collect historical ship operation data from the ship's power system, such as the operating status of physical equipment (e.g., main engine, gearbox), the working conditions of functional modules (e.g., power transmission), and the energy and information flows between them;

[0070] A dynamic model consisting of input layer, wandering layer and output layer is established based on historical ship operation data;

[0071] The input layer is built by capturing the time-enhanced embedding and dynamic weight matrix of ship operation data through temporal graph convolutional coding and an attention mechanism. Specifically, in the input layer, a temporal graph convolutional network (TGCN) is used to encode the snapshot sequence of the time-varying complex network, and a node embedding vector is generated by aggregating the neighbor information of each node (taking into account time dependencies). At the same time, a temporal attention mechanism is used to weight the historical embedding sequence to obtain the time-enhanced embedding and dynamic weight matrix. These results are directly used as the input of the subsequent walk layer. In addition, "time-varying complex network" refers to the topological structure that changes dynamically over time in the ship's power system.

[0072] The walk layer is built by calculating the mixed probability based on the time-enhanced embedding and the dynamic weight matrix, and performing quantum jumps and classical transfers through the mixed probability to generate a sequence of walk paths with time series marks. Specifically, in the walk layer, quantum computing and classical random walk strategies are integrated to implement quantum-classical hybrid walks. That is, high-betweenness nodes are first placed in a quantum superposition state. Then, each step of the walk executes a quantum jump controlled by the dynamic weight matrix with probability p, or a classical transfer controlled by the time-enhanced embedding with probability 1-p. By alternating between these two transfer methods, walk paths are generated and the node visit sequence is recorded to provide input for the output layer.

[0073] The output layer predicts the probability of future operating conditions based on the travel path sequence using Gaussian process regression, combining path similarity with embedded features. Specifically, the output layer uses Gaussian process regression (GPR) technology, combined with path similarity and node embedding features, to predict the probability of operating conditions in the next T steps. Ultimately, this layer outputs the probability distribution of the operating conditions and visualization results, such as a dynamic topology map.

[0074] In this embodiment S2.1, through time-series graph convolutional coding, the input layer can effectively extract time-series features from historical ship operation data. These features are crucial for understanding the dynamic changes in the ship's operating status. The time-series marked wandering path sequence generated by the wandering layer not only retains the time-series information of the ship's operating status, but also reflects the potential connections between states through the diversity of wandering paths. This is of great significance for understanding the complexity of ship operation and predicting future operating conditions. The output layer uses Gaussian process regression, combining path similarity with embedded features to predict the probability of future operating conditions. This method performs well when processing small sample data and nonlinear relationships, and has high prediction accuracy and robustness.

[0075] S2.2. In the input layer of the dynamic model, based on the current ship operation data, the adjacent node data of each node are aggregated to obtain the node embedding vector , such as the host node embedded as:

[0076] ;

[0077] According to the node embedding vector, combined with the preset learnable parameter matrix, the query vector q and the key vector are calculated. Sum value vector ;

[0078] According to the query vector q, key vector Sum value vector , the temporal enhanced embedding of all nodes in the dynamic system is calculated by the temporal attention algorithm ;

[0079] Perform attention weighting and normalization on the original edge weights to obtain a dynamic weight matrix ; Among them, the original edge weight is the dynamic weight value assigned to the connection relationship between mechanical components in the power system, such as the transmission efficiency of the main engine → gearbox.

[0080] Among them, the node embedding vector is the final output of the multi-layer temporal graph convolutional network, which is generated by progressively aggregating neighbor information; In the process, there is a single layer output :

[0081]

[0082] Among them, the query vector q, the key vector Sum value vector They are:

[0083] ;

[0084] ;

[0085]

[0086] in, is the set of direct neighbor nodes of node v, is the direct neighbor node of node v, For the The edge weight matrix of the layer, is the learnable parameter matrix; An activation function (such as ReLU, which stands for Rectified Linear Unit and is a widely used activation function in artificial neural networks). Embedding vector for nodes in the previous layer;

[0087] 、 and is the learnable parameter matrix; Compute the key vector for representation Sum value vector When the referenced historical time step is the same as the current time step interval.

[0088] In this example, S2.2, the node embedding vector, as a representation of node features, captures the uniqueness of the node and its position in the system. This feature representation capability enables the model to more accurately understand the role and function of the node, providing accurate input for subsequent calculations. The original edge weights are then subjected to attention weighting and normalization to produce a dynamic weight matrix. This process not only preserves the static information about the connection relationships between mechanical components but also reflects the changes in these relationships over time and operating status through dynamic weight assignment.

[0089] S2.3. In the wandering layer of the dynamic model, quantum technology is used to make high-betweenness nodes (such as the main engine) in the ship operation data enter the quantum superposition state, and the embedded and dynamic weight matrix Control high-betweenness nodes to perform quantum jumps and classical transfers to obtain the transfer probability matrix ; Each step is based on the probability Performing a quantum jump (affected by Regulation), with 1- Perform a classical transfer (affected by Regulation);

[0090] According to the transition probability matrix Perform path sampling to obtain path distribution and node access frequency .

[0091] Among them, the transition probability matrix is:

[0092]

[0093] in, is the probability value, is the quantum unitary operator; Indicates time When the slave node To Node The classic transfer weight of ; For nodes The sum of the weights of all outgoing edges.

[0094] In this embodiment S2.3, the introduction of a dynamic weight matrix enables the model to capture the temporal changes in the ship's operating state. Using quantum technology to place high-betweenness nodes into a superposition state fully exploits the parallel processing capabilities of quantum computing and improves computational efficiency, which is crucial for processing large-scale, highly complex ship operation data. Furthermore, the combination of quantum transitions and classical transfers enables the model to explore more possibilities between ship operating states, thereby more comprehensively reflecting the system's dynamic behavior.

[0095] In addition, path sampling through the transition probability matrix can capture the transition rules between ship operating states, which helps to reflect the dynamics and uncertainty of the system.

[0096] S2.4. In the output layer of the dynamic model, embedding is enhanced according to the time sequence and path distribution ,The path similarity (i.e., path similarity) and time series embedding features are fused through Gaussian process regression (GPR) to obtain the initial predicted operating condition of the ship;

[0097] Based on node access frequency The nodes of the initial predicted working condition are marked and the paths are bolded to obtain the predicted working condition with a dynamic topological graph.

[0098] Among them, the initial prediction condition is:

[0099]

[0100] in, Indicates the distribution on a given path and node access frequency Under the conditions of The posterior probability distribution of ; Represents a Gaussian process, represented by the mean function and covariance function Common definition.

[0101] In this example, S2.4, Gaussian process regression integrates path similarity and time series embedding features. These two features reflect the characteristics of ship operations at the path and time series levels, respectively, helping to improve prediction accuracy and robustness. The dynamic topology map not only provides a comprehensive view of the ship's operating status but also labels nodes and bolds paths based on node access frequency. This helps the system quickly identify critical states and paths, enabling more informed decisions.

[0102] S3. Establish an objective function and constraint set for power allocation based on the predicted operating conditions; wherein the objective function is established by minimizing the global energy loss value, maximizing the network topology entropy, and suppressing target node overload.

[0103] Step S3 of the embodiment of the present application includes S3.1 to S3.2, specifically:

[0104] S3.1. The objective function is established by minimizing the global energy loss value, maximizing the network topology entropy and suppressing the overload of the target node. Specifically:

[0105] With the goal of minimizing global energy loss, the first sub-objective function is established based on the node quantum state overlap in the current ship operation data. In addition, the classical optimization in the traditional method is limited to only considering explicit parameters (such as resistance, distance), ignoring the implicit coupling relationship, and the first sub-objective function By introducing "soft" topological constraints through quantum state overlap, the optimizer can actively explore efficient paths that are not explicitly modeled.

[0106] With the goal of maximizing the network topology entropy, the load of high-number nodes is suppressed according to the node degree and betweenness centrality parameters, and the second sub-objective function is established. In addition, traditional invulnerability optimization only considers connectivity and ignores the impact of subgraph structure complexity on fault diffusion, while the second sub-objective function Combining topological entropy with betweenness centrality simultaneously optimizes global connectivity and local robustness.

[0107] According to the predicted working conditions, nonlinear penalties are imposed on the target risk nodes to establish the third sub-objective function In addition, traditional optimization is based on historical data and cannot cope with sudden operating conditions such as equipment failure. By reducing the power of high-risk nodes in advance, the passive response mode of "fail-safe" can be avoided.

[0108] According to the first sub-objective function , the second sub-objective function And the third sub-objective function Establishing the objective function .

[0109] The first sub-objective function for:

[0110]

[0111] The second sub-objective function for:

[0112]

[0113] The third sub-objective function for:

[0114] ;

[0115]

[0116] Objective function for

[0117]

[0118] in, For the edge The power flow, the edge represents the channel of power flow or energy transmission; is the capacity of the edge; For nodes The quantum state overlap degree of , the node represents the starting point or end point of power flow or energy transfer; E is the set of edges;

[0119] is a collection of nodes, For nodes The degree, is the betweenness centrality, For The topological entropy of the subgraph centered on

[0120] is the risk sensitivity coefficient, is the quadratic penalty term, Nodes in the predicted working condition At the moment The failure prediction probability, is the risk sensitivity coefficient;

[0121] 、 and As dynamic weights, they can be adjusted online through reinforcement learning.

[0122] In Example S3.1, the first sub-objective function is dedicated to minimizing global energy loss. By accurately calculating the quantum state overlap of each node in the current ship operation data, energy can be more effectively allocated and managed, thereby reducing unnecessary energy waste. The second sub-objective function aims to maximize the network topology entropy, which helps to improve the stability and robustness of the network structure. By reasonably suppressing the load of high-number nodes, key nodes in the network can be prevented from failing due to overload, thereby ensuring the normal operation of the entire network. The third sub-objective function can more effectively identify and manage potential risks by applying nonlinear penalties to target risk nodes in the predicted working conditions.

[0123] S3.2. Establish a constraint set based on hard constraints and soft constraints;

[0124] Among them, hard constraints are established based on the predicted working conditions, limiting the power change value according to the load power and limiting the equipment capacity according to the node power, including power balance constraints, equipment capacity constraints and topology connectivity constraints;

[0125] Soft constraints are established based on the predicted working conditions and according to the preset topological entropy change threshold to limit the temporal entropy change value of the power system, including quantum path probability constraints and temporal entropy smoothing constraints.

[0126] The power balance constraint is:

[0127]

[0128] The equipment capacity constraint is:

[0129]

[0130] The topological connectivity constraints are:

[0131]

[0132] The quantum path probability constraint is:

[0133]

[0134] The time series entropy smoothing constraint is:

[0135]

[0136] in, For nodes The power generation capacity, is the load power, is a collection of nodes;

[0137] and are the upper and lower limits of node power, respectively. For nodes Power;

[0138] is a set of critical edges, such as the power transmission path of the main engine gearbox; For nodes To the node The transmitted power value, such as the mechanical power output from the main engine to the gearbox;

[0139] is the set of all possible power paths, For path The probability weight of ; I(-) is a binary function, which takes the value of 1 when the condition is met, otherwise it takes the value of 0; is the safety subgraph under the predicted working condition, is the minimum safe path probability threshold;

[0140] is the time step Time system topology diagram The topological entropy of The previous time step The topological entropy of is the total number of time steps, is the topological entropy change threshold.

[0141] In this embodiment, S3.2, by limiting the range of power variation, can prevent the system from becoming unstable due to excessive power fluctuations; ensuring that the power output of each device does not exceed its design capacity can prevent device overload, extend the service life of the device, and improve the overall reliability of the system. Therefore, the setting of hard constraints helps prevent the system from entering a dangerous state. By limiting the range of variation of the temporal entropy, it can ensure that the system maintains a certain degree of stability and predictability during the optimization process. Soft constraints are more flexible than hard constraints and can allow for slight changes in the system state to a certain extent to adapt to fluctuations in external conditions.

[0142] S4. According to the objective function and constraint sets, combined with the quantum annealing algorithm, to find the optimal power allocation path in the topology of the power system.

[0143] Step S4 of the embodiment of the present application is specifically as follows:

[0144] By quadratic expansion, the objective function Convert to Ising form, convert the constraint set into a quadratic penalty term; convert the converted objective function Added to the constraint set, we get the total Hamiltonian ;

[0145] According to the total Hamiltonian ,finding the optimal power allocation path in time-varying complex networks via quantum annealing algorithm.

[0146] S5. Perform path decoding and equipment power mapping on the optimal power allocation path to obtain a power allocation result, and control the start and stop operations of the ship according to the power allocation result.

[0147] Step S5 of the embodiment of the present application is specifically as follows:

[0148] The optimal power allocation path is binary decoded and power mapped to obtain the initial power allocation result. Specifically: key equipment achieves continuous adjustment through 8-bit binary code, auxiliary equipment supports 16 levels of discrete power with 4-bit code, combined with floating-point percentage mapping and equipment to correct nonlinear characteristics of data; built-in 2-bit checksum and dynamic proportional allocation mechanism ensure real-time power balance, force adaptation to minimum starting power and compatibility with preset hardware interfaces; finally, through load prediction, dynamic optimization of coding accuracy, combined with reinforcement learning to achieve high-precision power allocation that adapts to working conditions, and obtain the initial power allocation result; and the initial power allocation result is specifically manifested as a device power allocation table;

[0149] The power allocation of conflicting nodes in the initial power allocation result is adjusted through topology dependency verification, and the initial power allocation result is proportionally scaled or reduced according to the current total load of the power system to obtain the power allocation result. Specifically: verify whether the power allocation violates physical dependencies (such as allocating power to the cooling system when the host is not started). If a conflict occurs, the core device power is forcibly corrected according to priority to cover the conflicting node, and the power allocation is recalculated within the conflicting subnet for local reoptimization to obtain the corrected power allocation result; if the current total load of the power system exceeds the limit, it is proportionally scaled (such as: all devices are multiplied by a preset coefficient) or selectively reduced (such as: only reducing the power of non-critical devices); if there is power redundancy, the power of the scalable device is increased as needed to obtain the optimized power allocation result.

[0150] The start and stop operations of the ship are controlled according to the optimized power distribution results.

[0151] The topology dependency check in S5 of this embodiment ensures the rationality of the power allocation results, avoids power allocation errors caused by node conflicts, and guarantees data accuracy. The initial power allocation results are scaled or reduced based on the current total load of the power system, allowing the power allocation results to dynamically adapt to the actual load requirements of the system. This flexibility helps maintain system stability and efficiency as loads change, improving system adaptability.

[0152] Overall, this embodiment has the following beneficial effects:

[0153] This application uses a dynamic model to accurately predict future operating conditions based on current ship operating data, providing a scientific basis for optimizing start-up and shutdown operations. The dynamic model uses historical ship operating data for learning and encodes the network state of the power system. This process can uncover historical patterns and characteristics of ship operation, providing a solid foundation for operating condition prediction. Furthermore, quantum computing, with its powerful parallel processing capabilities and computational speed, can efficiently capture and analyze the dynamic network characteristics of the power system. The established objective function aims to minimize global energy loss, maximize network topology entropy, and prevent node overload, ensuring efficient and safe operation of the power system. A quantum annealing algorithm is used to find the optimal power allocation path within the power system topology. This algorithm is efficient and capable of handling complex optimization problems, ensuring the optimal solution is found. Through path decoding and device power mapping, the optimal power allocation path is converted into specific device power allocation results. This process ensures that each device receives the power most appropriate for its operating state, avoiding power overload or underload, thereby improving the efficiency of the entire power system. Ultimately, ship start-up and shutdown operations are adjusted in real time based on the optimal power allocation results, avoiding unnecessary energy consumption and wear, and improving operational efficiency.

[0154] Example 2:

[0155] See also Figure 2 , an embodiment of the present application provides a ship start-stop control device based on working condition prediction, comprising a data module 10, a prediction module 20, a calculation module 30, a solution module 40 and a control module 50;

[0156] The data module 10 is used to obtain the current ship operation data from the ship's power system;

[0157] The prediction module 20 is used to predict the operating condition of the ship using a dynamic model based on the current ship operating data to obtain the predicted operating condition of the ship. The dynamic model is established by encoding the network state of the power system based on historical ship operating data and capturing the network dynamic characteristics through quantum computing;

[0158] A calculation module 30 is configured to establish an objective function and a constraint set for power allocation based on the predicted operating conditions; wherein the objective function is established by minimizing the global energy loss value, maximizing the network topology entropy, and suppressing target node overload;

[0159] A solution module 40 is used to find an optimal power allocation path in the topology of the power system based on the objective function and the constraint set in combination with a quantum annealing algorithm;

[0160] The control module 50 is used to perform path decoding and equipment power mapping on the optimal power allocation path to obtain a power allocation result, and control the start and stop operations of the ship according to the power allocation result.

[0161] In one embodiment, the data module 10 is specifically:

[0162] In the ship's power system, a supervisory control and data acquisition system (SCADA) or other preset methods are used to collect current ship operation data in real time, including equipment operation data, time-varying network snapshots, node characteristics, and time series data, and store the current ship operation data in a database.

[0163] Among them, the current ship operation data includes the speed, temperature, pressure of physical equipment, power transmission efficiency and loss, as well as energy and information flow records between equipment, which are used to reflect the equipment status.

[0164] The network snapshot includes the topological structure of the dynamic system over time, including nodes and connections, which is used to reflect state changes.

[0165] Node characteristics include node type, capacity, and performance parameters, which are used to help the model understand behavior.

[0166] Time series include system, equipment and functional module status data arranged in chronological order, which are used to adapt the model.

[0167] In one embodiment, the prediction module 20 includes a model unit, an input unit, a walk unit, and an output unit;

[0168] The model unit is used to collect historical ship operation data from the ship's power system, such as the operating status of physical equipment (such as the main engine and gearbox), the working conditions of functional modules (such as power transmission), and the energy flow and information flow between them;

[0169] The model unit is also used to establish a dynamic model including an input layer, a wandering layer and an output layer based on historical ship operation data;

[0170] The input layer is built by capturing the time-enhanced embedding and dynamic weight matrix of ship operation data through temporal graph convolutional coding and an attention mechanism. Specifically, in the input layer, a temporal graph convolutional network (TGCN) is used to encode the snapshot sequence of the time-varying complex network, and a node embedding vector is generated by aggregating the neighbor information of each node (taking into account time dependencies). At the same time, a temporal attention mechanism is used to weight the historical embedding sequence to obtain the time-enhanced embedding and dynamic weight matrix. These results are directly used as the input of the subsequent walk layer. In addition, "time-varying complex network" refers to the topological structure that changes dynamically over time in the ship's power system.

[0171] The walk layer is built by calculating the mixed probability based on the time-enhanced embedding and the dynamic weight matrix, and performing quantum jumps and classical transfers through the mixed probability to generate a sequence of walk paths with time series marks. Specifically, in the walk layer, quantum computing and classical random walk strategies are integrated to implement quantum-classical hybrid walks. That is, high-betweenness nodes are first placed in a quantum superposition state. Then, each step of the walk executes a quantum jump controlled by the dynamic weight matrix with probability p, or a classical transfer controlled by the time-enhanced embedding with probability 1-p. By alternating between these two transfer methods, walk paths are generated and the node visit sequence is recorded to provide input for the output layer.

[0172] The output layer predicts the probability of future operating conditions based on the travel path sequence using Gaussian process regression, combining path similarity with embedded features. Specifically, the output layer uses Gaussian process regression (GPR) technology, combined with path similarity and node embedding features, to predict the probability of operating conditions in the next T steps. Ultimately, this layer outputs the probability distribution of the operating conditions and visualization results, such as a dynamic topology map.

[0173] The model unit in this embodiment uses time-series graph convolutional coding, and the input layer can effectively extract time-series features from historical ship operation data. These features are crucial for understanding the dynamic changes in the ship's operating status. The time-series labeled wandering path sequence generated by the wandering layer not only retains the time-series information of the ship's operating status, but also reflects the potential connections between states through the diversity of wandering paths. This is of great significance for understanding the complexity of ship operations and predicting future operating conditions. The output layer uses Gaussian process regression, combining path similarity with embedded features to predict the probability of future operating conditions. This method performs well when dealing with small sample data and nonlinear relationships, and has high prediction accuracy and robustness.

[0174] The input unit is used to aggregate the adjacent node data of each node based on the current ship operation data in the input layer of the dynamic model to obtain the node embedding vector , such as the host node embedded as:

[0175] ;

[0176] The input unit is also used to calculate the query vector q and the key vector according to the node embedding vector and the preset learnable parameter matrix. Sum value vector ;

[0177] The input unit is also used to calculate the query vector q and the key vector Sum value vector , the temporal enhanced embedding of all nodes in the dynamic system is calculated by the temporal attention algorithm ;

[0178] The input unit is also used to perform attention weighting and normalization on the original edge weights to obtain the dynamic weight matrix ; Among them, the original edge weight is the dynamic weight value assigned to the connection relationship between mechanical components in the power system, such as the transmission efficiency of the main engine → gearbox.

[0179] Among them, the node embedding vector is the final output of the multi-layer temporal graph convolutional network, which is generated by progressively aggregating neighbor information; In the process, there is a single layer output :

[0180]

[0181] Among them, the query vector q, the key vector Sum value vector They are:

[0182] ;

[0183] ;

[0184]

[0185] in, is the set of direct neighbor nodes of node v, is the direct neighbor node of node v, For the The edge weight matrix of the layer, is the learnable parameter matrix; An activation function (such as ReLU, which stands for Rectified Linear Unit and is a widely used activation function in artificial neural networks). Embedding vector for nodes in the previous layer;

[0186] 、 and is the learnable parameter matrix; Compute the key vector for representation Sum value vector When the referenced historical time step is the same as the current time step interval.

[0187] In the input unit of this embodiment, the node embedding vector, as a representation of node features, captures the uniqueness of the node and its position in the system. This feature representation capability enables the model to more accurately understand the role and function of the node, providing accurate input for subsequent calculations. The original edge weights are then subjected to attention weighting and normalization to produce a dynamic weight matrix. This process not only preserves the static information about the connection relationships between mechanical components but also reflects the changes in these relationships over time and operating status through dynamic weight assignment.

[0188] The walking unit is used to make the high-betweenness nodes (such as the host) in the ship operation data enter the quantum superposition state through quantum technology in the walking layer of the dynamic model, and enhance the embedding according to the time sequence and dynamic weight matrix Control high-betweenness nodes to perform quantum jumps and classical transfers to obtain the transfer probability matrix ; Each step is based on the probability Performing a quantum jump (affected by Regulation), with 1- Perform a classical transfer (affected by Regulation);

[0189] The walk unit is also used to calculate the transition probability matrix Perform path sampling to obtain path distribution and node access frequency .

[0190] Among them, the transition probability matrix is:

[0191]

[0192] in, is the probability value, is the quantum unitary operator; Indicates time When the slave node To Node The classic transfer weight of ; For nodes The sum of the weights of all outgoing edges.

[0193] In this embodiment, the introduction of a dynamic weight matrix in the wandering unit enables the model to capture the changes in the ship's operating state over time. Using quantum technology to place high-betweenness nodes into a superposition state fully utilizes the parallel processing capabilities of quantum computing and improves computational efficiency, which is of great significance for processing large-scale, highly complex ship operation data. Furthermore, the combination of quantum transitions and classical transfers enables the model to explore more possibilities between ship operating states, thereby more comprehensively reflecting the dynamic behavior of the system.

[0194] In addition, path sampling through the transition probability matrix can capture the transition rules between ship operating states, which helps to reflect the dynamics and uncertainty of the system.

[0195] Output unit, used to enhance the embedding according to the temporal sequence in the output layer of the dynamic model and path distribution ,The path similarity (i.e., path similarity) and time series embedding features are fused through Gaussian process regression (GPR) to obtain the initial predicted operating condition of the ship;

[0196] The output unit is also used to calculate the node access frequency The nodes of the initial predicted working condition are marked and the paths are bolded to obtain the predicted working condition with a dynamic topological graph.

[0197] Among them, the initial prediction condition is:

[0198]

[0199] in, Indicates the distribution on a given path and node access frequency Under the conditions of The posterior probability distribution of ; Represents a Gaussian process, represented by the mean function and covariance function Common definition.

[0200] In this embodiment, the output unit uses Gaussian process regression to integrate path similarity and time series embedding features. These two features reflect the characteristics of ship operations at the path and time series levels, respectively, helping to improve prediction accuracy and robustness. The dynamic topology map not only provides a comprehensive view of the ship's operating status but also labels nodes and bolds paths based on node access frequency. This helps the system quickly identify critical states and paths, enabling more informed decisions.

[0201] In one embodiment, the calculation module 30 includes a function unit and a constraint unit;

[0202] Among them, the function unit is used to establish the objective function by minimizing the global energy loss value, maximizing the network topology entropy and suppressing the overload of the target node, specifically:

[0203] With the goal of minimizing global energy loss, the first sub-objective function is established based on the node quantum state overlap in the current ship operation data. In addition, the classical optimization in the traditional method is limited to only considering explicit parameters (such as resistance, distance), ignoring the implicit coupling relationship, and the first sub-objective function By introducing "soft" topological constraints through quantum state overlap, the optimizer can actively explore efficient paths that are not explicitly modeled.

[0204] With the goal of maximizing the network topology entropy, the load of high-number nodes is suppressed according to the node degree and betweenness centrality parameters, and the second sub-objective function is established. In addition, traditional invulnerability optimization only considers connectivity and ignores the impact of subgraph structure complexity on fault diffusion, while the second sub-objective function Combining topological entropy with betweenness centrality simultaneously optimizes global connectivity and local robustness.

[0205] According to the predicted working conditions, nonlinear penalties are imposed on the target risk nodes to establish the third sub-objective function In addition, traditional optimization is based on historical data and cannot cope with sudden operating conditions such as equipment failure. By reducing the power of high-risk nodes in advance, the passive response mode of "fail-safe" can be avoided.

[0206] According to the first sub-objective function , the second sub-objective function And the third sub-objective function Establishing the objective function .

[0207] The first sub-objective function for:

[0208]

[0209] The second sub-objective function for:

[0210]

[0211] The third sub-objective function for:

[0212] ;

[0213]

[0214] Objective function for

[0215]

[0216] in, For the edge The power flow, the edge represents the channel of power flow or energy transmission; is the capacity of the edge; For nodes The quantum state overlap degree of , the node represents the starting point or end point of power flow or energy transfer; E is the set of edges;

[0217] is a collection of nodes, For nodes The degree, is the betweenness centrality, For The topological entropy of the subgraph centered on

[0218] is the risk sensitivity coefficient, is the quadratic penalty term, Nodes in the predicted working condition At the moment The failure prediction probability, is the risk sensitivity coefficient;

[0219] 、 and As dynamic weights, they can be adjusted online through reinforcement learning.

[0220] In the functional unit of the embodiment, the first sub-objective function is dedicated to minimizing global energy loss. By accurately calculating the quantum state overlap of each node in the current ship operation data, energy can be more effectively distributed and managed, thereby reducing unnecessary energy waste. The second sub-objective function aims to maximize the network topology entropy, which helps to improve the stability and robustness of the network structure; by reasonably suppressing the load of high-number nodes, the failure of key nodes in the network due to overload can be avoided, thereby ensuring the normal operation of the entire network. The third sub-objective function can more effectively identify and manage potential risks by applying nonlinear penalties to target risk nodes in the predicted working conditions.

[0221] The constraint unit is used to establish a constraint set based on hard constraints and soft constraints;

[0222] Among them, hard constraints are established based on the predicted working conditions, limiting the power change value according to the load power and limiting the equipment capacity according to the node power, including power balance constraints, equipment capacity constraints and topology connectivity constraints;

[0223] Soft constraints are established based on the predicted working conditions and according to the preset topological entropy change threshold to limit the temporal entropy change value of the power system, including quantum path probability constraints and temporal entropy smoothing constraints.

[0224] The power balance constraint is:

[0225]

[0226] The equipment capacity constraint is:

[0227]

[0228] The topological connectivity constraints are:

[0229]

[0230] The quantum path probability constraint is:

[0231]

[0232] The time series entropy smoothing constraint is:

[0233]

[0234] in, For nodes The power generation capacity, is the load power, is a collection of nodes;

[0235] and are the upper and lower limits of node power, respectively. For nodes Power;

[0236] is a set of critical edges, such as the power transmission path of the main engine gearbox; For nodes To the node The transmitted power value, such as the mechanical power output from the main engine to the gearbox;

[0237] is the set of all possible power paths, For path The probability weight of ; I(-) is a binary function, which takes the value of 1 when the condition is met, otherwise it takes the value of 0; is the safety subgraph under the predicted working condition, is the minimum safe path probability threshold;

[0238] is the time step Time system topology diagram The topological entropy of The previous time step The topological entropy of is the total number of time steps, is the topological entropy change threshold.

[0239] The constraint unit in this embodiment can prevent the system from becoming unstable due to excessive power fluctuations by limiting the range of power variation. Ensuring that the power output of each device does not exceed its design capacity can prevent device overload, extend the device's service life, and improve the overall reliability of the system. Therefore, the setting of hard constraints helps prevent the system from entering a dangerous state. By limiting the range of variation in time series entropy, it can ensure that the system maintains a certain degree of stability and predictability during the optimization process. Soft constraints are more flexible than hard constraints and can allow for slight changes in system state to a certain extent to adapt to fluctuations in external conditions.

[0240] In one embodiment, the solution module 40 is specifically:

[0241] By quadratic expansion, the objective function Convert to Ising form, convert the constraint set into a quadratic penalty term; convert the converted objective function Added to the constraint set, we get the total Hamiltonian ;

[0242] According to the total Hamiltonian ,finding the optimal power allocation path in time-varying complex networks via quantum annealing algorithm.

[0243] In one embodiment, the control module 50 is specifically:

[0244] The optimal power allocation path is binary decoded and power mapped to obtain the initial power allocation result. Specifically: key equipment achieves continuous adjustment through 8-bit binary code, auxiliary equipment supports 16 levels of discrete power with 4-bit code, combined with floating-point percentage mapping and equipment to correct nonlinear characteristics of data; built-in 2-bit checksum and dynamic proportional allocation mechanism ensure real-time power balance, force adaptation to minimum starting power and compatibility with preset hardware interfaces; finally, through load prediction, dynamic optimization of coding accuracy, combined with reinforcement learning to achieve high-precision power allocation that adapts to working conditions, and obtain the initial power allocation result; and the initial power allocation result is specifically manifested as a device power allocation table;

[0245] The power allocation of conflicting nodes in the initial power allocation result is adjusted through topology dependency verification, and the initial power allocation result is proportionally scaled or reduced according to the current total load of the power system to obtain the power allocation result. Specifically: verify whether the power allocation violates physical dependencies (such as allocating power to the cooling system when the host is not started). If a conflict occurs, the core device power is forcibly corrected according to priority to cover the conflicting node, and the power allocation is recalculated within the conflicting subnet for local reoptimization to obtain the corrected power allocation result; if the current total load of the power system exceeds the limit, it is proportionally scaled (such as: all devices are multiplied by a preset coefficient) or selectively reduced (such as: only reducing the power of non-critical devices); if there is power redundancy, the power of the scalable device is increased as needed to obtain the optimized power allocation result.

[0246] The start and stop operations of the ship are controlled according to the optimized power distribution results.

[0247] The topology dependency check in S5 of this embodiment ensures the rationality of the power allocation results, avoids power allocation errors caused by node conflicts, and guarantees data accuracy. The initial power allocation results are scaled or reduced based on the current total load of the power system, allowing the power allocation results to dynamically adapt to the actual load requirements of the system. This flexibility helps maintain system stability and efficiency as loads change, improving system adaptability.

[0248] Overall, this embodiment has the following beneficial effects:

[0249] This application uses a dynamic model to accurately predict future operating conditions based on current ship operating data, providing a scientific basis for optimizing start-up and shutdown operations. The dynamic model uses historical ship operating data for learning and encodes the network state of the power system. This process can uncover historical patterns and characteristics of ship operation, providing a solid foundation for operating condition prediction. Furthermore, quantum computing, with its powerful parallel processing capabilities and computational speed, can efficiently capture and analyze the dynamic network characteristics of the power system. The established objective function aims to minimize global energy loss, maximize network topology entropy, and prevent node overload, ensuring efficient and safe operation of the power system. A quantum annealing algorithm is used to find the optimal power allocation path within the power system topology. This algorithm is efficient and capable of handling complex optimization problems, ensuring the optimal solution is found. Through path decoding and device power mapping, the optimal power allocation path is converted into specific device power allocation results. This process ensures that each device receives the power most appropriate for its operating state, avoiding power overload or underload, thereby improving the efficiency of the entire power system. Ultimately, ship start-up and shutdown operations are adjusted in real time based on the optimal power allocation results, avoiding unnecessary energy consumption and wear, and improving operational efficiency.

[0250] Example 3:

[0251] An embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the ship start-stop control method based on operating condition prediction;

[0252] The ship start-stop control method based on operating condition prediction, if implemented as a software functional unit and used as a standalone product, can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunications signal, and a software distribution medium.

[0253] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A ship start-stop control method based on operating condition prediction, characterized in that: include: Obtain current ship operation data from the ship's power system; Based on the current ship operation data, an operating condition prediction is performed using a dynamic model to obtain a predicted operating condition of the ship; wherein the dynamic model is established by encoding the network state of the power system based on historical ship operation data and capturing the network dynamic characteristics through quantum computing; Establishing an objective function and a constraint set for power allocation according to the predicted operating conditions; According to the objective function and the constraint set, combined with a quantum annealing algorithm, an optimal power allocation path is found in the topology of the power system; performing path decoding and device power mapping on the optimal power allocation path to obtain a power allocation result, and controlling the start and stop operations of the ship according to the power allocation result; The objective function is established by minimizing the global energy loss value, maximizing the network topology entropy and suppressing the overload of the target node, specifically: With the goal of minimizing global energy loss, a first sub-objective function is established according to the node quantum state overlap in the current ship operation data; With the goal of maximizing the network topology entropy, the load of high-number nodes is suppressed according to the node degree and betweenness centrality parameters, and the second sub-objective function is established; Applying nonlinear penalties to the target risk node according to the predicted working condition to establish a third sub-objective function; Establishing the objective function according to the first sub-objective function, the second sub-objective function and the third sub-objective function; The first sub-objective function is specifically: ; in, For the edge The power flow, the edge represents the channel of power flow or energy transmission; is the capacity of the edge; For nodes The node represents the starting point or end point of power flow or energy transfer; E is the set of edges.

2. A ship start-stop control method based on operating condition prediction according to claim 1, characterized in that: Based on the current ship operation data, the operating condition prediction is performed through the dynamic model to obtain the predicted operating condition of the ship, specifically: Performing data aggregation and weighted calculation on the current ship operation data according to the dynamic model to obtain a time series enhanced embedding and a dynamic weight matrix; Using quantum technology, high-betweenness nodes in the ship operation data are caused to enter a superposition state, and according to the timing enhanced embedding and the dynamic weight matrix, the high-betweenness nodes are controlled to perform quantum transitions and classical transfers to obtain a transfer probability matrix; Path sampling is performed according to the transfer probability matrix, and the predicted operating condition of the ship is obtained by combining the data obtained by the path sampling with the time series enhanced embedding calculation.

3. A ship start-stop control method based on operating condition prediction according to claim 2, characterized in that: According to the dynamic model, data aggregation and weighted calculation are performed on the current ship operation data to obtain a time series enhanced embedding and a dynamic weight matrix, specifically: Aggregating adjacent node data of each node in the current ship operation data according to the dynamic model to obtain a node embedding vector; Calculate a query vector, a key vector, and a value vector based on the node embedding vector and a preset learnable parameter matrix; According to the query vector, the key vector and the value vector, a temporal enhanced embedding of all nodes in the dynamic system is calculated by a temporal attention algorithm; The original edge weights are subjected to attention weighting and normalization processing to obtain a dynamic weight matrix; wherein the original edge weights are dynamic weight values ​​assigned to the connection relationships between mechanical components in the power system.

4. A ship start-stop control method based on operating condition prediction according to claim 2, characterized in that: Path sampling is performed according to the transition probability matrix, and the predicted operating condition of the ship is obtained by combining the data obtained from the path sampling with the time series enhanced embedding calculation, specifically: Perform path sampling according to the transition probability matrix to obtain path distribution and node access frequency; According to the time series enhanced embedding and the path distribution, the path similarity and the time series embedding features are fused through Gaussian process regression to obtain the initial predicted operating condition of the ship; The nodes of the initial predicted working condition are marked and the paths are bolded according to the node access frequency to obtain a predicted working condition with a dynamic topology graph.

5. A ship start-stop control method based on operating condition prediction according to claim 2, characterized in that: The transition probability matrix is ​​specifically: ; in, is the probability value, is the quantum unitary operator; Indicates time When the slave node To Node The classic transfer weight of ; For nodes The sum of the weights of all outgoing edges.

6. A ship start-stop control method based on operating condition prediction according to claim 1, characterized in that: The dynamic model includes an input layer, a wandering layer and an output layer; The input layer is established by capturing the temporal enhanced embedding and dynamic weight matrix of ship operation data through temporal graph convolutional coding and attention mechanism; The walking layer is established by calculating a mixing probability based on the timing enhanced embedding and the dynamic weight matrix, and performing quantum jumps and classical transfers through the mixing probability to generate a walking path sequence with timing marks; The output layer is established based on the wandering path sequence by fusing path similarity and embedded features through Gaussian process regression to predict the probability of future working conditions.

7. A ship start-stop control method based on operating condition prediction according to claim 1, characterized in that: The constraint set consists of hard constraints and soft constraints; Wherein, the hard constraint is established based on the load power limit power change value and based on the node power limit device capacity; The soft constraint is established by limiting the temporal entropy change value of the power system according to a preset topological entropy change threshold.

8. A ship start-stop control method based on operating condition prediction according to claim 1, characterized in that: Perform path decoding and device power mapping on the optimal power allocation path to obtain a power allocation result, specifically: performing binary decoding and power mapping on the optimal power allocation path to obtain an initial power allocation result; According to a preset method, the power distribution of the conflicting nodes in the initial power distribution result is adjusted through topology dependency verification, and the initial power distribution result is scaled or reduced according to the current total load of the power system to obtain the power distribution result.

Images