Fault prediction-oriented fault-tolerant control method for ship key equipment

Abstract

The invention discloses a fault prediction-oriented fault-tolerant control method for ship key equipment, and relates to the technical field of fault-tolerant control. Comprising the following steps: acquiring multi-source sensing data, continuously coding the running state of each structural unit to form a time period feature coding matrix, and establishing an equipment control topological coding diagram; extracting a time period feature coding sequence with abnormality, positioning a structural unit with an abnormal state in combination with an equipment control topological coding graph, and delimiting a frozen granularity boundary; freezing the data interface and the control signal path according to the freezing granularity boundary; meanwhile, a structural unit access channel is opened, and a granularity partition limiting and controlling mode is formed; according to the path nodes corresponding to the frozen structural units, corresponding channels are removed from the equipment control topological coding graph, and control path reconstruction operation is executed among the remaining structural units. According to the invention, flexible connection and multi-path scheduling of equipment control paths are realized, and the control continuity and fault resistance of key tasks are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault-tolerant control, in particular to a fault-tolerant control method for key ship equipment oriented to fault prediction. Background Art

[0002] In the field of intelligent ship control and fault-tolerant management, with the continuous advancement of digitalization and automation technologies, the demand for critical equipment status awareness and intelligent fault-tolerant control during ship operation is increasing. Modern ship systems integrate a variety of sensors, communication, and control modules, forming complex control topologies that rely on multi-source status data for efficient navigation and maneuvering. Against this backdrop, fault-tolerant control methods have continuously evolved, from early redundant hardware designs to intelligent fault-tolerant approaches centered on model-driven and data-driven approaches. Control strategies based on fault identification, adaptive compensation, and behavioral learning mechanisms have been widely studied to address risk factors such as equipment aging, sudden failures, and parameter uncertainty. However, current technologies still primarily rely on the response and adjustment of single control objects and lack mechanisms for collaboratively identifying and dynamically adjusting the interconnected paths between multiple structural units in the control network. This makes it difficult to promptly delineate the affected area and accurately execute fault-tolerant operations at the initial stage of a fault.

[0003] For example, the comparative document CN116088309B discloses a composite learning fault-tolerant control method for surface ships based on fault identification. It achieves robust compensation control for model uncertainty and propeller faults by constructing a ship dynamics model and combining error definition, adaptive design and other means. This solution focuses on improving the control accuracy of the propeller, but its fault-tolerant control granularity is mainly concentrated at the control law level. It fails to achieve risk partition identification and local freezing strategy at the equipment structure level. The control response does not yet have the ability to structurally adjust, making it difficult to adapt to the path-dependent complexity brought about by the network evolution of key ship equipment. In addition, the lack of a mechanism for encoding and modeling the time-continuous mutation characteristics in multi-source data causes the system to have a response lag problem in identifying the early state of the fault, which limits the initiative of the fault-tolerant control.

[0004] Comparative document CN116700271A proposes a ship fault-tolerant control method, system, device, and storage medium. This method utilizes target ship state variable information and rudder angle control inputs to achieve dynamic control adjustments within the deviation tolerance range through waypoint prediction and deviation judgment mechanisms. Although this method achieves energy conservation and response optimization in heading control, its control object is a single navigation behavior. It lacks overall modeling and adjustment of the internal equipment control topology, making it difficult to support partitioned control strategies when key equipment fails. Furthermore, this method does not introduce technical paths for fault-tolerant management of key equipment, such as multi-period coding, structural unit freezing, and path reconstruction, making it difficult to meet the requirements for coordinated control of multi-node interactions and path-dependency risks in complex systems. Summary of the Invention

[0005] In view of the problems existing in the existing ship fault-tolerant control technology, the present invention is proposed.

[0006] Therefore, the problem to be solved by the present invention is how to improve the dynamic adaptability of the ship system in failure risk scenarios.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In the first aspect, the present invention provides a fault-tolerant control method for key ship equipment for fault prediction, which includes: obtaining multi-source sensor data of structural units of key ship equipment, continuously encoding the operating status of each structural unit to form a time period feature coding matrix, and synchronously establishing an equipment control topology coding map; based on the time period feature coding matrix, extracting time period feature coding sequences with continuous mutations or fluctuation anomalies, and combining with the equipment control topology coding map, locating structural units with abnormal states, and delineating the frozen granularity boundaries of the structural units; when any structural unit is identified as an abnormal state, the corresponding data interface and control signal path are frozen according to the frozen granularity boundaries; at the same time, the structural unit access channel is opened to form a granular partitioned control mode; according to the path node corresponding to the frozen structural unit, the corresponding channel is removed from the equipment control topology coding map, and a control path reconstruction operation is performed between the remaining structural units.

[0009] As an optimal solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the continuous coding includes: selecting multi-source observation data indicators of the structural unit, constructing a parameter index group, and binding a unified observation time period interval to form a two-dimensional observation data table of the structural unit-time period; according to the sliding time window, the parameter change trend of the structural unit in each window is symbolically encoded, and a state aggregation symbol vector is constructed.

[0010] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the establishment of the equipment control topology coding diagram includes: extracting the control signal trigger relationship between structural units based on the electrical control logic diagram and equipment communication constraints of the key ship equipment, and representing all pairs of structural units with signal transmission or control dependencies in the form of directed edges to form a set of structural unit dependency edges; performing path traversal analysis on the set of structural unit dependency edges, generating multiple directed control paths based on the type and task role of the starting structural unit, and adding a path identification sequence to each path to construct a traceable equipment control topology coding diagram.

[0011] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the method further comprises: selecting a time period feature code row corresponding to each structural unit in a time period feature code matrix, counting the cumulative number and time distribution density of mutation marks in a continuous observation time period, and constructing a structural unit mutation weight vector; using the environmental disturbance data recorded in the ship operation status table to identify non-structural mutations caused by external disturbances in the synchronization time period, and eliminating the corresponding positions in the structural unit mutation weight vector to generate a mutation net value vector after interference cleaning; centrally counting the elements of the mutation net value vector corresponding to all structural units, and calculating the median value and the upper and lower quantile difference; comparing the mutation net value of each structural unit with the median value and the upper and lower quantile difference: if the mutation net value is less than the median value minus the upper and lower quantile difference, it is marked as stable; if the mutation net value is greater than or equal to the median value plus the lower quantile difference, it is marked as abnormal; if the mutation net value is between the median value minus the upper and lower quantile difference and the median value plus the lower quantile difference, it is marked as alert.

[0012] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the delineation of the frozen granularity boundary of the structural unit includes: retrieving the node where the structural unit marked as abnormal state is located from the equipment control topology coding diagram, and identifying the upstream control input unit and the downstream control action unit in the control path to form a path connection node set of the structural unit; analyzing each structural unit in the path connection node set, extracting the edge structural unit with weak coupling or path isolation characteristics, and marking it as the minimum freezable boundary unit; taking the nodes on both sides of the minimum freezable boundary unit as the endpoints of the frozen boundary segment, constructing the frozen granularity boundary of the structural unit, and adding boundary identification attributes in the equipment control topology coding diagram.

[0013] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the analysis of each structural unit in the path connection node set includes: extracting the start and end nodes and the intermediate structural unit node set from all path segments in the path connection node set, counting the total number of times each structural unit appears as an intermediate node in the path as the adjacency frequency, sorting and mean analyzing the adjacency frequency, and marking structural units with an adjacency frequency below the average level; counting the proportion of structural units appearing in all paths, and classifying structural units that simultaneously meet the conditions of an adjacency frequency below the average level and a proportion below a proportion threshold into a weak coupling candidate set; for structural units that have been classified into the weak coupling candidate set, judging one by one whether freezing will block the original main control path, and if the path is still passable, they are selected as candidate freezing units; identifying the functional position of the candidate freezing units in the control path, judging whether they have a critical jump function, and if not, listing them as minimum freezable boundary units.

[0014] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the method comprises: opening the structural unit access channel to form a granular partitioned control mode, including: screening all structural units marked as stable, and using the numbers as the candidate set of structural units currently under consideration for restoring access; locating the candidate structural units in the control topology coding diagram, confirming whether there is a controllable control signal channel in the path segment between the frozen granularity boundary, and eliminating the structural units that have a direct coupling relationship with the frozen path; limiting the control interface of the structural unit that meets the boundary connectivity conditions to only receive the main control task data, and shielding the horizontal data forwarding function between the structural units that have not been verified.

[0015] As a preferred solution of the fault-tolerant control method for key ship equipment for fault prediction described in the present invention, the control path reconstruction operation performed between the remaining structural units includes: based on the frozen granularity boundary of the structural unit and the control interface status, all nodes and directed edge sets connected to the frozen structural units in the control topology coding graph are deleted to form a path residual graph structure; in the path residual graph, all remaining paths between each pair of main control structural units and task target structural units are depth-first traversed to screen out reachable paths and calculate the link reachability index; according to the link reachability and the structural unit control interface response delay data, the reachable path segments are arranged in order of priority, and the first k path segments are selected to form a priority path set, where k is a constant; each path segment in the priority path set is connected in the path residual graph structure according to the starting point and end point node numbers to form a schedulable task control path chain, and output as a reconstructed path structure.

[0016] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the fault-tolerant control method for key ship equipment for fault prediction as described in the first aspect of the present invention are implemented.

[0017] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the fault-tolerant control method for key ship equipment oriented to fault prediction as described in the first aspect of the present invention are implemented.

[0018] The beneficial effects of the present invention are as follows: the present invention can realize real-time perception, refined discrimination and dynamic control of the operating status of complex electrical equipment systems without relying on traditional whole-machine shutdown detection. By collecting multi-source sensor data of structural units and constructing a structural unit period feature coding matrix, the state evolution process of key equipment in different operating stages can be fully characterized, improving the foresight and accuracy of abnormal trend identification; with the help of the equipment control topology coding diagram, the path dependency modeling between structural units is realized, so that when identifying unit anomalies, the scope of influence is reasonably delineated, and the freezing boundary is determined based on weak coupling and path isolation characteristics, thereby avoiding the problems of false freezing and excessive system intervention.

[0019] On this basis, the present invention realizes the flexible connection and multi-path scheduling of equipment control paths by constructing a granular partition control mode and an adaptive reconstruction mechanism of the control path, which greatly improves the control continuity and fault resistance of key tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 Flowchart of the fault-tolerant control method for key ship equipment oriented to fault prediction. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0025] As mentioned in the background technology above, fault-tolerant control methods are constantly evolving, from early redundant hardware designs to intelligent fault-tolerant methods with model-driven and data-driven approaches as their core. Among them, control strategies based on fault identification, adaptive compensation, and behavioral learning mechanisms have been widely studied to address risk factors such as equipment aging, sudden failures, and parameter uncertainty. However, current technologies still rely mainly on response adjustments to a single control object, lacking a collaborative identification and dynamic adjustment mechanism for the associated paths between multiple structural units in the control network, making it difficult to promptly define the affected area and accurately perform fault-tolerant operations at the initial stage of a fault.

[0026] Figure 1 FIG. 1 is a flow chart of a fault-tolerant control method for key ship equipment oriented to fault prediction according to an embodiment of the present invention. Figure 1 As shown in FIG, the fault-tolerant control method for key ship equipment oriented to fault prediction includes:

[0027] S1: Acquire multi-source sensor data of the structural units of key ship equipment, continuously encode the operating status of each structural unit to form a time period feature coding matrix, and simultaneously establish an equipment control topology coding diagram.

[0028] In an embodiment of the present invention, the generation of the time period feature coding matrix includes the following steps:

[0029] S1.1: Select multi-source observation data indicators of the structural unit, construct a parameter index group, and bind a unified observation time period interval to form a two-dimensional observation data table of the structural unit-time period.

[0030] It should be noted that key ship equipment usually includes multiple structural units such as propulsion systems, power distribution units, cooling components, hydraulic actuators, and sensing and feedback loops. Each unit is accompanied by a series of observable physical or logical variables during normal operation, such as changes in voltage, current, temperature, speed, pressure, displacement and other parameters.

[0031] Conventional technology usually adopts a periodic sampling method during the monitoring process, and the limit value judgment is made separately based on the data of each sampling point. However, this method cannot effectively reflect the continuous evolution trend of the parameters over time, especially the lack of early warning capabilities before the equipment is about to degrade or mutate. Therefore, in the present invention, a unified observation parameter index group is first defined, and the multiple physical sensor parameters and control variables associated with each structural unit are named and numbered; at the same time, a standard observation time period interval is defined for each structural unit, and a sliding time window of a fixed length, such as 5 minutes or 10 minutes, is selected as the unit analysis period, so that the collected raw data constitutes a two-dimensional observation data table with time consistency. The row vectors of the two-dimensional data table correspond to different structural units, and the column vectors represent continuous time periods. Each cell records all the observation parameter values ​​of the structural unit within a specific time period.

[0032] S1.2: According to the sliding time window, the parameter change trend of the structural unit in each window is symbolically encoded, and the state aggregation symbol vector is constructed based on the abnormal density, change direction and synchronization deviation.

[0033] Specifically, after constructing the two-dimensional observation data table, in order to better capture the changing patterns of device states, parameter change trends must be mapped from the original numerical space into recognizable state labels. To this end, the present invention uses a state symbolic encoding method under a sliding window mechanism to encode the structural unit parameter change trends.

[0034] Specifically, the original observations within the current time window are used as input, and the first-order difference operation is used to calculate the incremental value between two consecutive time points. The trend direction symbol is defined based on the positive and negative results of each differential value. For example, if the incremental value is greater than the minimum significant threshold for trend determination, it is encoded as +1 (continuous rise); if the incremental value is less than the negative value of the minimum significant threshold for trend determination, it is encoded as -1 (continuous decline); if it is in the interval, it is encoded as 0 (oscillation or stability); the differential symbol sequence is statistically analyzed within the entire sliding time window. If a certain symbol (+1, -1 or 0) continuously occupies a proportion exceeding a set proportion threshold (such as 70%), the overall trend of the parameter within the sliding time window is determined, and the trend symbol is assigned to the directional symbol code of the parameter in the current time window, forming a preliminary trend classification. Finally, for each structural unit, the directional symbol codes of all parameters in the window are combined into a set of symbols.

[0035] Furthermore, to avoid misjudgments, within each sliding time window, for each parameter's observation sequence, we calculate whether it deviates from its respective statistical normal range. The specific process is as follows: Calculate the mean μ and standard deviation σ of the parameter in the corresponding time period in the historical sample, set the anomaly determination threshold range to [μ-ασ, μ+ασ], where α is an empirical value that can be set based on actual conditions; Determine whether the observation value at each moment in the current time window falls outside this threshold range; if it does, it is recorded as an anomaly; Divide the number of anomalies by the window length to obtain the anomaly density (i.e., the frequency of deviations from the normal state) of the corresponding parameter in the current time window.

[0036] Synchronization deviation is used to measure whether there is a lack of consistency in the trend direction of multiple parameters. The specific operation is as follows: First, the trend symbols (+1, 0, -1) of each parameter in the current time window are combined into a trend direction comparison matrix according to the parameter sequence order. Secondly, the trend consistency score calculation method is adopted: the trend code of each parameter is element-by-element multiplied with the trend code of the master parameter, and the average value of the product vector is calculated; the average value represents the degree of consistency of the directions of the two vectors, and the range is [-1, 1]. Finally, the consistency score is converted into a synchronization deviation value, that is, synchronization deviation = 1-average direction consistency. The larger the value, the more inconsistent the parameter is with the master trend direction in the window. By combining the three indicators of trend direction, anomaly density and synchronization deviation, a three-dimensional aggregation vector is formed, and the state aggregation symbol is defined on this basis. For example, "↑△" indicates that the parameter rises accompanied by an increase in anomaly and a slight synchronization disorder.

[0037] S1.3: Fill the symbol vectors of all structural units into the coding matrix in time sequence to form a two-dimensional coding matrix consisting of structural units and time periods.

[0038] The state aggregation symbol vector is further embedded in the unified symbol space and arranged in a matrix in the order of structural units and time periods to form a structural unit time period feature coding matrix. The horizontal dimension represents the structural unit, and the vertical dimension is the continuous time period. Each element in the matrix is ​​a state aggregation symbol.

[0039] In an embodiment of the present invention, establishing a device control topology coding map includes the following steps:

[0040] S1.4: Based on the electrical control logic diagram of key ship equipment and the equipment communication constraints, extract the control signal trigger relationship between structural units, and represent all structural unit pairs with signal transmission or control dependencies in the form of directed edges to form a structural unit dependency edge set.

[0041] It should be noted that, when describing control path relationships, existing technologies are often limited to logic block diagrams or static control chain descriptions, and fail to dynamically map control transmission and signal response dependencies between structural units.

[0042] To solve this problem, the present invention analyzes documents such as equipment layout diagrams, control circuit diagrams, and software communication protocols, and extracts the boundary conditions and constraint logic for control signal transmission between key structural units one by one, clarifying which unit will send start-stop signals, locking conditions, or parameter feedback control commands to another unit.

[0043] Structural unit pairs with signal transmission or logical control relationships are constructed as a set of directed edges, where each directed edge uses the source node to represent the signal sending structural unit and the target node to represent the signal receiving structural unit. The edge attributes record the trigger condition type, such as switching signal, voltage control, threshold feedback, etc.

[0044] Finally, the structural units are combined as the node set of the graph, and the edge set is used as the edge set of the graph to generate an initial topological graph model containing node and edge attributes and direction identifiers.

[0045] S1.5: Perform path traversal analysis on the set of structural unit dependency edges, generate multiple directed control paths based on the type and task role of the starting structural unit, and attach a path identification sequence to each path to construct a traceable equipment control topology coding diagram.

[0046] The generation of directed control paths is based on the task roles and operation sequences of the structural units. For example, the control instructions issued by the main controller are sequentially transmitted to the intermediate drive unit and then to the execution component, forming a typical main control-intermediate control-execution chain path.

[0047] In each chain path, to ensure uniqueness and traceability between paths, a path identification sequence is added, for example, P1-P2-P3, which means starting from the structural unit P1, P2 and P3 are controlled in sequence.

[0048] By traversing and numbering the graph of all path sets, the construction of a control topology coding graph with task tracking capability can be completed.

[0049] S1.6: Based on the mutation mark bits in the structural unit period feature coding matrix, the marks are mapped and connected with the corresponding path nodes in the topological coding diagram, so that the continuous mutation structural units in the path have the ability of linkage judgment, forming a path cascade sensitivity section.

[0050] Specifically, all units with mutation marker bits in the structural unit period feature coding matrix constructed above are mapped one-to-one to the path nodes in the device control topology coding graph.

[0051] Mutation markers usually indicate abnormal density increases or sudden changes in state direction in the coding sequence, indicating that the structural unit may enter an unstable working area. By projecting the mutation node into the control path, it is possible to identify whether there are continuous mutation structural units in the path. If the continuity meets the preset length (such as more than 3 nodes), the corresponding path is identified as a cascade sensitive section.

[0052] Cascade sensitive sections are used to trigger subsequent control interruptions, granular freezing, and risk propagation modeling, ensuring that after a structural unit mutates, the downstream units that may be affected can be quickly located, thereby improving the timeliness and accuracy of fault response.

[0053] S2: Based on the period feature coding matrix, extract the period feature coding sequence with continuous mutation or fluctuation anomalies, and combine it with the equipment control topology coding diagram to locate the structural units with abnormal states and delineate the frozen granularity boundaries of the structural units.

[0054] S2.1: Locate structural units with abnormal conditions.

[0055] In the time period feature coding matrix, the time period feature coding row corresponding to each structural unit is selected to count the cumulative number of mutation marks and the time distribution density within the continuous observation time period, and then construct the structural unit mutation weight vector. Specifically, for each structural unit, each row in the time period feature coding matrix represents an observation window within a continuous time period. Within these windows, if the trend direction dimension of the aggregate symbol vector suddenly changes from +1 to -1 or jumps from a stable state to an abnormal state, the corresponding matrix position is marked as a mutation state. Each mutation event is accumulated to obtain the cumulative number of mutations for the structural unit. A sliding window is used to perform local statistics on the time index set of mutation events; within each sliding window, the frequency of mutation occurrence is calculated, and the maximum frequency value is recorded as the time distribution density. The cumulative number of mutations and the time distribution density are weighted to form the mutation weight vector for the structural unit.

[0056] In order to prevent non-structural factors from interfering with mutation identification, the environmental disturbance data recorded in the ship operation status table are used to identify non-structural mutations caused by external disturbances during the synchronization period, and the corresponding positions are eliminated from the structural unit mutation weight vector to generate the net mutation value vector after interference cleaning.

[0057] The net mutation value vector elements corresponding to all structural units are centrally counted, and the median value and the upper and lower quantile difference (i.e., the difference between the 25th and 75th percentiles) are calculated; the net mutation value of each structural unit is compared with the median value and the upper and lower quantile difference: if the net mutation value is less than the median value minus the upper and lower quantile difference, it is marked as stable; if the net mutation value is greater than or equal to the median value plus the lower quantile difference, it is marked as abnormal; if the net mutation value is between the median value minus the upper and lower quantile difference and the median value plus the lower quantile difference, it is marked as alert and requires continuous monitoring.

[0058] S2.2: Delineate the frozen grain boundaries of the structural units.

[0059] S2.2.1: From the device control topology coding diagram, retrieve the node where the structural unit marked as abnormal is located, and identify the upstream control input unit and downstream control action unit in the control path (that is, identify other structural units with direct signal inflow or signal output relationships with the node), forming a path connection node set for the structural unit. This connection node set not only reflects the direction of signal flow but also determines whether the subsequent control chain will be interrupted. By identifying these path connection node sets, a local control subgraph with the abnormal unit as the core and upstream and downstream nodes as extensions can be formed, providing a structural basis for frozen granularity analysis.

[0060] S2.2.2: Analyze each structural unit in the path connection node set, extract the edge structural units with weak coupling or path isolation characteristics, and mark them as the minimum freezable boundary units.

[0061] a. For all path segments in the path connection node set, extract the start and end nodes and the set of intermediate structural unit nodes. For each structural unit, count the total number of times it appears as an intermediate node in the path. This number is defined as the adjacency frequency of the structural unit. The adjacency frequencies are sorted and averaged. Structural units with adjacency frequencies significantly below the average level are marked and preliminarily determined to be marginal structural units with low participation. At the same time, count the proportion of these structural units appearing in all paths. If a structural unit appears in a low proportion of the path, for example, less than 30%, it is further determined not to have a critical bridging function in the control path. Finally, structural units that meet both the low adjacency frequency and low path appearance ratio conditions are classified as weakly coupled candidates.

[0062] b. For each structural unit that has been classified as a weakly coupled candidate, determine whether freezing it will disrupt the original primary control path. Specifically, temporarily remove the structural unit and its directly connected edges from the control topology diagram to see if the path from the primary control starting point to the critical task endpoint is still maintained. If the path remains accessible, it is preliminarily determined that freezing the structural unit will not disrupt the continuity of the primary control logic, and the unit can continue to be a candidate for freezing.

[0063] To ensure structural stability, the status marks of the structural units are checked synchronously during the above process. If a structural unit has been identified as abnormal before, then even if the connectivity conditions are met, the structural unit shall not be included in the frozen boundary candidate range and shall be eliminated.

[0064] c. For the candidate cells mentioned above, further identify their functional location within the control path. During operation, the path structure in the control topology graph is used to determine whether the structural unit is a path starting point, a path ending point, or simply a path relay point. If the structural unit serves only as the end point of certain edge paths, or if there are alternative paths between its upstream and downstream paths, it is determined that it does not have a critical jump function in the control logic and can be further classified as a minimum freezable boundary unit.

[0065] In practical applications, it is important to avoid freezing structural units located at critical jump locations in the main control path to ensure that the freezing operation does not interfere with the basic control flow of the main path. Ultimately, all weakly coupled structural units that meet the above structural judgment conditions and are not marked as abnormal constitute the frozen boundary set.

[0066] If after freezing a certain stable weakly coupled structural unit, more than P% (for example, 80%) of the paths in the master path set remain reachable, and the average link reachability value of the path is not lower than Q% (such as 90%) of the average link value before freezing, then it is considered that freezing the structural unit does not significantly weaken the connectivity of the master path. In addition, the path reachability is obtained by weighted average of the inverse of the response delay of the control interface of each node in the path, reflecting the path response quality. Finally, the structural unit that meets the conditions that the path coverage rate is lower than the set threshold, the path connectivity rate after freezing is higher than P%, and the link response quality remains above Q% is marked as the minimum freezable boundary unit and serves as the valid endpoint for the subsequent construction of the frozen boundary segment.

[0067] S2.2.3: Use the nodes on both sides of the minimum freezable boundary unit as the endpoints of the frozen boundary segment to construct the frozen granularity boundary of the structural unit and add boundary identification attributes in the device control topology coding diagram.

[0068] The specific operation is as follows: take each minimum freezable boundary unit as the boundary starting point and end point, identify the nodes on both sides, and establish a frozen boundary segment in the device control topology coding map, and mark all structural units in the segment as controllable frozen areas; at the same time, add boundary identification attributes to the corresponding edges and nodes of the topology map.

[0069] It should be noted that after the freeze granularity boundary is established, the set of all paths marked as cascade-sensitive segments in the control topology encoding diagram is retrieved, and the paths in the set whose starting nodes are located in the set of mutation-state structural units are assigned a freeze priority flag. All structural unit control interfaces in this type of path segment are set to the first-level freeze state, and a cascade risk tag field is added to the control instruction scheduling table to facilitate path exclusion or priority avoidance during subsequent path reconstruction.

[0070] S3: When any structural unit is identified as an abnormal state, the corresponding data interface and control signal path are frozen according to the freezing granularity boundary; at the same time, the structural unit access channel is opened to form a granular partition control mode.

[0071] The following steps are involved in opening the access channel of the structural unit and forming a granular partition control mode:

[0072] S3.1: Filter all structural units marked as stable and use their numbers as the candidate set of structural units currently being considered for restoration access.

[0073] S3.2: Position the candidate structural units as nodes in the control topology coding graph, confirm whether there is a controllable control signal channel in the path segment between the frozen granularity boundary, and eliminate the structural units that have a direct coupling relationship with the frozen path.

[0074] After the set of candidate structural units is determined, in order to prevent the control instructions from mistakenly crossing the frozen boundary, the candidate structural units should be screened based on the connectivity of the control path.

[0075] In the specific operation, for each structural unit number in the candidate structural unit set, the corresponding node number in the topology graph is located according to the graph node mapping table established in the device control topology coding graph; then, starting from this node, a depth-first path traversal algorithm is used to find all reachable paths between any frozen granularity boundary unit.

[0076] For each candidate path, check whether it contains any structural unit nodes in an abnormal or alert state. If such structural unit nodes exist, the corresponding path is considered a potential risk channel and the starting point number should be immediately removed from the candidate structural unit set.

[0077] In addition, any structural units in the path that have direct control signal dependencies (i.e., formed directed edge associations) with frozen path nodes should also be removed from the candidate set. The above path dependencies are determined based on the edge attributes in the topology graph established in step S1. The edge attributes include clear information such as trigger signal type, direction, and threshold conditions.

[0078] After the above screening process is completed, a set of structural unit numbers with path connectivity security is obtained. Each structural unit in the set maintains boundary isolation from the frozen area in the current topology map and does not form a reachable channel with any risk node, meeting the prerequisites for restoring access.

[0079] S3.3: Limit the control interface of the structural unit that meets the boundary connectivity conditions to only receive the main control task data, and shield the horizontal data forwarding function between unverified structural units.

[0080] It should be noted that, in order to further ensure that the structural unit that can recover access does not cause system interference on the control path, it is necessary to limit the configuration of the control interface state of the structural unit that can recover access.

[0081] During operation, for each structural unit number in the structural unit set obtained by the above screening, only the input channel for receiving the main control task instructions is retained at the hardware control interface level. Such task instructions include start and stop commands, status query requests or data synchronization trigger instructions directly assigned by the main control controller or scheduling module, and do not include horizontal forwarding data between other structural units.

[0082] During operation, the filtered set of structural units only retains the communication channel to the incoming main control task instructions. Such instructions are limited to task control instructions directly issued by the equipment master controller or central scheduling module, including start and stop control commands, status feedback requests and main path data synchronization trigger signals. They do not include unauthorized instructions from other structural units or secondary modules for horizontal interaction, nor do they include auxiliary control instructions on non-critical links.

[0083] In order to limit the data propagation range of the structural unit in the topology diagram, the present invention logically masks all directed edges of the structural unit node in the non-master control direction in the device control topology diagram, wherein the logical masking operation refers to changing the edge status from "activatable" to "pending verification" and marking it as a "non-task path channel". It can only be restored for use when the path security is subsequently confirmed or the path integrity is re-evaluated. The masking information will be recorded as an additional field of the edge attribute to facilitate the unified call of subsequent recovery management operations.

[0084] Through the above interface limitation method, the risk that unverified structural units may interfere with stable structural units can be significantly reduced, and the security level and access management accuracy of the system under fault-tolerant control state can be improved.

[0085] S4: According to the path nodes corresponding to the frozen structural units, the corresponding channels are removed from the device control topology coding diagram, and a control path reconstruction operation is performed between the remaining structural units.

[0086] In an embodiment of the present invention, performing a control path reconstruction operation between the remaining structural units includes the following steps:

[0087] S4.1: Based on the frozen granularity boundary of the structural unit and the control interface status, delete all nodes and directed edge sets connected to the frozen structural units in the control topology coding graph to form a path residual graph structure.

[0088] In the specific operation, first, based on the determined structural unit freezing granularity boundary, all included structural unit number sets are identified; in the topological coding graph, the graph nodes corresponding to the number set are retrieved one by one, and all outgoing and incoming edge sets are recorded; then, the structural unit node and all directed edges connected to it are deleted from the topological graph structure to generate a path residual graph; in addition, in the process of constructing the path residual graph structure, all path segments that have a connection relationship with any node in the cascade sensitive segment must be automatically excluded.

[0089] The path residual graph is a directed graph subset, the vertex set is the structural unit nodes in the original graph except the frozen units, and the edge set is the control dependency edges that do not involve the frozen path.

[0090] By constructing a residual graph, we can provide a set of topological foundations that have undergone preliminary security screening for subsequent path reconstruction, reducing the probability of link conflicts during path reconstruction.

[0091] S4.2: In the path residual graph, perform depth-first traversal on all remaining paths between each pair of main control structure units and task target structure units, screen out reachable paths, and calculate the link reachability index.

[0092] To ensure that the remaining structural units can be normally scheduled to continue executing critical tasks after freezing, the present invention searches for all possible paths from the main control structural unit to the task target structural unit in the path residual graph structure.

[0093] In the specific operation, first determine the target node number set based on the task target node definition in the task scheduling table; then, for each main control structure unit number, execute the depth-first path search algorithm to find all valid paths to the target node.

[0094] For each reachable path, control path continuity, path length, and node responsiveness are used as reference indicators of link reachability. Control path continuity is the ratio of nodes with continuous control dependencies in the path; path length is the number of structural units in the path; and node responsiveness is the average of each node's control interface response delay data within the most recent time window.

[0095] After the three are normalized and weighted (the specific weight coefficient can be set according to actual conditions), a link reachability index value is formed, which is used to indicate the degree of path availability.

[0096] S4.3: Based on the link reachability and the structural unit control interface response delay data, the reachable path segments are arranged in order of priority, and the first k path segments (k is set to 1.5 times the number of nodes within the structural unit frozen granularity boundary, rounded up to ensure that a sufficient number of high-scoring control paths can still be retained under the frozen area constraints) are selected to form a priority path set, where k is a constant.

[0097] S4.4: Connect each path segment in the priority path set to the path residual graph structure according to the start and end node numbers to form a schedulable task control path chain, and output it as a reconstructed path structure.

[0098] This embodiment also provides a computer device, which is suitable for the fault-tolerant control method of key ship equipment for fault prediction, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the fault-tolerant control method of key ship equipment for fault prediction as proposed in the above embodiment.

[0099] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0100] This embodiment further provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the method for realizing fault-tolerant control of key ship equipment for fault prediction proposed in the above embodiment is implemented.

[0101] In summary, the present invention can achieve real-time perception, refined discrimination and dynamic control of the operating status of complex electrical equipment systems without relying on traditional whole-machine shutdown detection. By collecting multi-source sensor data of structural units and constructing a structural unit period feature coding matrix, it can comprehensively characterize the state evolution process of key equipment in different operating stages, improving the foresight and accuracy of abnormal trend identification; with the help of the equipment control topology coding diagram, the path dependency modeling between structural units is realized, so that when identifying unit anomalies, the impact range is reasonably delineated, and the freezing boundary is determined based on weak coupling and path isolation characteristics, thereby avoiding the problems of false freezing and excessive system intervention.

[0102] On this basis, the present invention realizes the flexible connection and multi-path scheduling of equipment control paths by constructing a granular partition control mode and an adaptive reconstruction mechanism of the control path, which greatly improves the control continuity and fault resistance of key tasks.

[0103] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fault-tolerant control method for key ship equipment based on fault prediction, characterized by: include: Acquire multi-source sensor data of the structural units of key ship equipment, continuously encode the operating status of each structural unit, form a time period feature coding matrix, and simultaneously establish an equipment control topology coding diagram; Based on the period feature coding matrix, the period feature coding sequence with continuous mutation or fluctuation anomalies is extracted. Combined with the equipment control topology coding diagram, the structural units with abnormal states are located and the frozen granularity boundaries of the structural units are delineated. When any structural unit is identified as abnormal, the corresponding data interface and control signal path are frozen according to the freezing granularity boundary; at the same time, the structural unit access channel is opened to form a granular partition control mode; According to the path nodes corresponding to the frozen structural units, the corresponding channels are removed from the device control topology coding graph, and the control path reconstruction operation is performed between the remaining structural units.

2. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 1, characterized in that: The continuous encoding includes: Select the multi-source observation data indicators of the structural unit, build a parameter index group, and bind a unified observation time period interval to form a two-dimensional observation data table of structural unit-time period; According to the sliding time window, the parameter change trend of the structural unit within each sliding time window is symbolically encoded, and a state aggregation symbol vector is constructed.

3. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 2, characterized in that: The establishment of the device control topology coding map includes: According to the electrical control logic diagram and equipment communication constraints of key ship equipment, the control signal trigger relationship between structural units is extracted, and all structural unit pairs with signal transmission or control dependencies are represented in the form of directed edges to form a structural unit dependency edge set; A path traversal analysis is performed on the set of structural unit dependency edges. Based on the type and task role of the starting structural unit, multiple directed control paths are generated. A path identification sequence is attached to each path to construct a traceable equipment control topology coding diagram.

4. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 1, characterized in that: The structural unit for locating an abnormal state includes: In the period feature coding matrix, the period feature coding row corresponding to each structural unit is selected, the cumulative number and time distribution density of mutation marks in the continuous observation period are counted, and the structural unit mutation weight vector is constructed; Using the environmental disturbance data recorded in the ship operation status table, the non-structural mutations caused by external disturbances during the synchronization period are identified, and the corresponding positions are eliminated from the structural unit mutation weight vector to generate the net mutation value vector after interference cleaning; Collect statistics on the net mutation value vector elements corresponding to all structural units, and calculate the median value and the upper and lower quantile difference; The net mutation value of each structural unit is compared with the median value and the upper and lower quantile difference values: If the net mutation value is less than the median value minus the difference between the upper and lower quantiles, it is marked as stable; If the net mutation value is greater than or equal to the median value plus the lower quantile difference, it is marked as abnormal; If the net mutation value is between the median value minus the upper and lower quantile difference and the median value plus the lower quantile difference, it is marked as a warning.

5. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 4, characterized in that: The frozen granularity boundaries of the structural unit are defined as follows: Retrieve the node where the structural unit marked as abnormal is located from the device control topology coding diagram, and identify the upstream control input unit and the downstream control action unit in the control path to form a path connection node set of the structural unit; Analyze each structural unit in the path connection node set, extract the edge structural units with weak coupling or path isolation characteristics, and mark them as the minimum freezable boundary units; The nodes on both sides of the minimum freezable boundary unit are used as the endpoints of the frozen boundary segment to construct the frozen granularity boundary of the structural unit, and the boundary identification attribute is added to the device control topology coding diagram.

6. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 5, characterized in that: The analyzing of each structural unit in the path connection node set includes: For all path segments in the path connection node set, the start and end nodes and the intermediate structural unit node sets are extracted. The total number of times each structural unit appears as an intermediate node in the path is counted as the adjacency frequency. The adjacency frequencies are sorted and averaged, and structural units with adjacency frequencies below the average level are marked. The proportion of structural units appearing in all paths is counted, and structural units that simultaneously meet the conditions of having an adjacency frequency lower than the average level and a proportion lower than the threshold are classified into the weak coupling candidate set; For the structural units that have been classified into the weak coupling candidate set, determine whether freezing will block the original main control path one by one. If the path is still accessible, it will be regarded as a candidate freezing unit; The functional position of the candidate freezing unit in the control path is identified to determine whether it has the critical jump function. If not, it is included in the minimum freezable boundary unit.

7. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 1, characterized in that: The open structural unit access channel forms a granular partition control mode including: Filter all structural units marked as stable and use their numbers as candidate sets of structural units currently under consideration for restoration of access; The candidate structural units are located as nodes in the control topology coding graph to confirm whether there is a controllable control signal channel in the path segment between the frozen granularity boundary and to eliminate the structural units that have a direct coupling relationship with the frozen path; The control interface of the structural unit that meets the boundary connectivity conditions is limited to only receiving the main control task data, and the horizontal data forwarding function between the structural units that have not been verified is shielded.

8. The method for fault-tolerant control of key ship equipment for fault prediction according to claim 1, characterized in that: The performing of the control path reconstruction operation between the remaining structural units includes: Based on the frozen granularity boundary of the structural unit and the control interface status, all nodes and directed edge sets connected to the frozen structural units in the control topology coding graph are deleted to form a path residual graph structure; In the path residual graph, all remaining paths between each pair of main control structure units and task target structure units are traversed in depth first, the reachable paths are screened out, and the link reachability index is calculated; According to the link reachability and the structural unit control interface response delay data, the reachable path segments are sorted in priority order, and the top k path segments are selected to form a priority path set, where k is a constant; Each path segment in the priority path set is connected to the path residual graph structure according to the starting and ending node numbers to form a schedulable task control path chain, which is output as a reconstructed path structure.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the fault-tolerant control method for key ship equipment oriented to fault prediction according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fault-tolerant control method for key ship equipment oriented to fault prediction according to any one of claims 1 to 8 are implemented.

Images