Intelligent sensor-based condition self-diagnosis system and method for marine transmission

Abstract

The present application relates to the technical field of state monitoring, in particular to a state self-diagnosis system and method of a ship transmission device based on intelligent sensing, the system comprising: a torque variation behavior recognition module, a shaft system disturbance chain extraction module, a direction behavior cooperative judgment module, an inertia response feature extraction module and a state excitation node generation module.In the present application, by extracting the consistency relationship between torque and load direction, the time topology structure of the disturbance path is constructed, the cooperative nodes of direction change between displacement and load are recognized, the direction consistency behavior in the continuous region of the structure is extracted in combination with the continuity performance of inertia response in adjacent shaft sections, the response sequence with unified direction evolution trend is formed, the correlation comparison and trend induction analysis of multi-source behaviors in multiple periods are completed, the structure response behavior recognition ability in the disturbance evolution process is enhanced, and the dynamic tracking and correlation analysis of potential abnormal performance in transmission state changes in multiple scenarios are supported.

Description

Technical Field

[0001] This invention relates to the field of condition monitoring technology, and in particular to a condition self-diagnosis system and method for ship transmission devices based on intelligent sensing. Background Technology

[0002] The field of condition monitoring technology involves the real-time perception, identification, and judgment of the operating status of equipment, systems, or structures. Its core aspects include operating feature extraction based on multi-source data acquisition, signal recognition, fault symptom identification, and life prediction. It is widely used in the health management systems of key equipment such as industrial equipment, power systems, rail transit, and marine power. This technology monitors the performance degradation process of key components by constructing operating data models, setting state thresholds, and establishing trend evaluation mechanisms. It also relies on multi-physical quantity sensors and edge computing to achieve comprehensive judgment for fault early warning. Among them, the condition self-diagnosis system of traditional marine transmission devices refers to a device for identifying and diagnosing changes in the operating status of marine propulsion power transmission components. The technical issue addressed by this patent is the difficulty in real-time identification of state changes in transmission devices during ship operation. Traditional condition self-diagnosis systems usually use a single sensor to obtain shaft vibration values ​​or bearing temperature values ​​as the basis for judgment. They analyze the collected data through offline interpretation and combine it with standard criteria to determine the current status of the component.

[0003] Existing technologies rely on fixed-point acquisition of single-variable signals and comparison with preset standards. When multi-source disturbances frequently occur, they cannot establish stable behavioral sequence matching relationships and lack a structured identification mechanism for the continuity and directional changes of disturbance timing. This makes it impossible to clearly extract the dynamic evolution patterns of key nodes. Especially when the disturbance amplitude is not obvious or the direction changes repeatedly, the system state is difficult to accurately judge through isolated signals, which can easily lead to the masking of trend features or the discontinuity of state identification, thereby affecting the accurate assessment of the overall device operating status and the timely capture of abnormal responses. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a self-diagnostic system and method for the condition of a ship's transmission device based on intelligent sensing. The technical solution is as follows: On the one hand, a self-diagnostic system for the condition of ship transmission devices based on intelligent sensing is provided, which includes: The torque change behavior recognition module obtains the torque trend and load direction sequence of the main propulsion bearing housing, determines whether the torque direction and load direction are synchronized, filters out time periods with consistent directions, and obtains torque-in-the-direction trigger segment markers. The shaft disturbance chain extraction module extracts the motion sequence of the main shaft coupling based on the torque-in-the-direction trigger segment marker, identifies the disturbance direction reversal behavior, filters the continuous disturbance paths that occur continuously within the preset time window of the direction change, and obtains the shaft disturbance response chain structure. The directional behavior collaborative judgment module extracts the main shaft displacement and load torque direction sequence based on the shaft system disturbance response chain structure, compares the time position of the changing node, and obtains the directional change cross recognition label. The inertial response feature extraction module extracts the response direction of the inclined inertia surface of the shaft segment based on the direction change cross recognition label, judges the direction continuity, and obtains the inertial lateral deflection response structure segment. The state excitation node generation module, based on the inertial lateral response structure fragment, calls the direction information, analyzes the trend consistency and time sequence, identifies the sequence of behaviors exhibiting common directional changes in continuous evolution, and obtains the transmission state abnormality diagnosis output result.

[0005] As a further embodiment of the present invention, the torque-in-the-south trigger segment marker includes a trend-consistent segment identifier, load response synchronization characteristics, and time evolution direction label; the shaft system disturbance response chain segment structure includes a disturbance direction switching sequence, a disturbance reversal node chain, and a continuous disturbance path; the direction change cross identification label includes a set of direction switching time points, a direction synchronization distribution pattern, and cross change comparison information; the inertial lateral response structure segment includes a tilt inertia direction trend, response continuity between shaft segments, and inertial consistency characteristics of continuous structural regions; and the transmission state abnormality diagnosis output result includes a directional behavior time sequence comparison relationship, multi-trend connection characteristics, and an abnormal behavior co-evolution sequence.

[0006] As a further aspect of the present invention, the torque trend refers to the direction and trajectory of the torque generated by the main propulsion bearing housing during transmission. By analyzing the time series, the continuous segment of the torque direction is identified to determine whether the direction remains unchanged. The load direction sequence refers to the record of the rotational direction change of the load-side coupling. The direction data sequence is formed by collecting the data through an angle encoder to determine the motion direction and change behavior of the load.

[0007] As a further aspect of the present invention, the disturbance direction reversal behavior refers to the phenomenon that the direction reverses between adjacent time points in the motion sequence at both ends of the axis system, and the reversal node is identified to analyze the disturbance process. The inclined inertia surface of the shaft segment refers to the directional plane in which the shaft segment generates an inertial response during operation. It can be used to analyze the continuity and consistency of the inertial direction between shaft segments and identify the dynamic response trend of the structure during disturbance.

[0008] As a further aspect of the present invention, the torque change behavior recognition module includes: The trend extraction submodule obtains the torque change sequence and load response direction sequence formed by the main propulsion bearing housing during the transmission process, analyzes the torque direction trend in adjacent time periods, filters continuous time periods in which the direction has not reversed, extracts the time distribution of the continuous direction sequence, and obtains the torque direction continuous segment data. Based on the continuous torque direction segment data, the direction comparison submodule extracts the load response direction change data within the same segment, corresponds to the direction sequence, compares whether the direction of each time point is consistent, extracts the segments that continuously maintain the same direction, and obtains the direction synchronization relationship distribution sequence. Based on the distribution sequence of the directional synchronization relationship, the segment filtering submodule extracts the start and end times of the directional synchronization segments, removes sequence segments with continuous fluctuations in the directional switching state, and filters segments with continuous directional change trends that are consistent with the load response direction within the time range, thus obtaining torque-in-the-direction trigger segment markers.

[0009] As a further aspect of the present invention, the shaft disturbance chain extraction module includes: The motion sequence extraction submodule extracts the displacement direction at both ends of the main shaft coupling based on the torque-in-the-direction trigger segment marker, sorts the displacement direction by time to form a motion trend sequence, analyzes the direction evolution within the continuous interval, extracts the direction data segment, and obtains the double-end direction change sequence. The perturbation direction identification submodule calls the dual-end direction change sequence, analyzes whether the directions of adjacent nodes have reversed, extracts the direction pairs that show continuous switching behavior, identifies the perturbation direction change segments, and obtains the perturbation reversal change sequence; The reverse path filtering submodule determines the repetition of the reversal direction in time based on the perturbation reversal change sequence, eliminates jump segments that cannot form a continuous path, filters path intervals that continuously cause perturbations and have related directional trends, extracts continuous behavior chain groups, and obtains the axis system perturbation response chain structure.

[0010] As a further aspect of the present invention, the directional behavior collaborative judgment module includes: The direction sequence extraction submodule extracts the sequence of changes in the main shaft displacement direction and the behavior of changes in the load torque direction based on the structure of the shaft system disturbance response chain. It arranges the direction change process in chronological order, filters out data segments with continuity, and obtains a bidirectional change comparison sequence. The switching node identification submodule calls the bidirectional change comparison sequence, compares the change time point positions during the direction change process, identifies whether the direction switch occurs at the same time point, extracts the cross nodes that show synchronous mutation in the sequence, and obtains the bidirectional synchronous switching point set. The synchronous trend analysis submodule, based on the bidirectional synchronous switching point set, determines the number and interval density of nodes in the same period, identifies whether there is a periodic trend, extracts the segments where nodes regularly overlap, and obtains the direction change cross recognition label.

[0011] As a further aspect of the present invention, the inertial response feature extraction module includes: The inertia data extraction submodule extracts inertial response direction data in the inclined inertia surface of the axis segment within the time period covered by the direction change cross recognition label, filters the continuous trend of the inertial direction in the time process, determines whether there is a unilateral offset trend in the inertial change direction in the time period, and obtains the inertial offset change trajectory. Based on the inertial offset change trajectory, the response path analysis submodule extracts the extension path of the shaft segment inertial response in time, identifies whether the inertial response direction between adjacent shaft segments extends continuously, and determines whether the direction extension trend remains consistent during the time process, thereby obtaining the shaft segment inertial transmission path. The directional trend assessment submodule analyzes whether the inertial direction evolves synchronously in the continuous region of the structure based on the inertial transmission path of the axis segment, filters inertial response segments with continuous characteristics, judges the consistency of the response direction in the spatial axis segment, and obtains the inertial lateral response structural segment.

[0012] As a further aspect of the present invention, the state excitation node generation module includes: The trend behavior extraction submodule, based on the direction data of the inertial lateral response structure segment, calls the direction sequence in the torque same direction trigger segment marker and the direction change cross recognition tag, compares the start and end relationship of the sequence direction trend, filters the time period that matches the direction change, and obtains the direction synchronization trend sequence. The temporal sequence connection analysis submodule extracts the time nodes of the direction sequence based on the direction synchronization trend sequence, compares the sequential relationship of the direction changes, identifies the segments where the direction turning points are continuously distributed on the timeline, and obtains the continuous path characteristics of the direction evolution. The direction consistency judgment submodule calls the continuous path feature of the direction evolution, extracts the direction change performance in the disturbance section, judges whether there is a behavior that maintains the same trend in the sequence, extracts the continuous behavior section, and obtains the abnormal transmission state diagnosis output result.

[0013] On the other hand, a self-diagnosis method for the state of a ship's transmission device based on intelligent sensing, wherein the self-diagnosis method for the state of a ship's transmission device based on intelligent sensing is executed based on the aforementioned self-diagnosis system for the state of a ship's transmission device based on intelligent sensing, includes the following steps: S1: Obtain the torque change trend and load response direction data of the main propulsion bearing housing, analyze the change of torque direction in a continuous time period, filter time segments with consistent directions, determine whether the load directions are synchronized, extract the behavioral characteristics of the corresponding time period, and obtain the torque-in-the-direction trigger segment markers. S2: Based on the torque-in-the-direction trigger segment marker, extract the motion direction sequence at both ends of the main shaft coupling, analyze the change of the disturbance direction between adjacent time points, determine whether there is a repeatedly switching motion trend, extract the continuous disturbance performance, and obtain the shaft system disturbance response chain structure. S3: Based on the shaft system disturbance response chain structure, extract the direction change sequence of the main shaft displacement and load torque, compare the change positions of the directions, determine whether they coincide at the time nodes, analyze the distribution of the change points, and obtain the direction change cross recognition label. S4: Based on the direction change cross recognition tag, extract the response direction data of the inclined inertia surface of the shaft segment, analyze the inertial change trend between shaft segments, determine whether the inertial response is continuously transmitted along the same direction, and obtain the inertial lateral response structure segment. S5: Based on the inertial lateral response structure segment, call the directional performance of the torque-in-the-direction behavior segment and the direction-synchronous change node group, compare the time sequence and continuity, determine whether it constitutes a continuous and consistent trend behavior, and obtain the abnormal transmission state diagnosis output result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by extracting the consistency relationship between torque and load direction, a temporal topology of the disturbance path is constructed, cooperating nodes of directional changes between displacement and load are identified, and the continuity of inertial response in adjacent shaft segments is combined to extract directional consistency behavior in continuous structural regions, forming a response sequence with a unified directional evolution trend. This completes the correlation comparison and trend induction analysis of multi-source behaviors in multiple time periods, enhances the ability to identify structural response behavior during disturbance evolution, and supports the dynamic tracking and correlation analysis of potential abnormal performance in transmission state changes under various scenarios. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a system block diagram of the present invention; Figure 3 This is a flowchart of the torque change behavior recognition module in this invention; Figure 4 This is a flowchart of the shaft disturbance chain extraction module in this invention; Figure 5 This is a flowchart of the directional behavior collaborative judgment module in this invention; Figure 6 This is a flowchart of the inertial response feature extraction module in this invention; Figure 7 This is a flowchart of the state excitation node generation module in this invention; Figure 8 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] This invention provides a self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing, such as... Figure 1-2 The diagram shows a self-diagnostic system for the condition of a ship's transmission system based on intelligent sensing. The system includes: The torque change behavior recognition module acquires the torque change trend and load response behavior sequence of the main propulsion bearing housing, analyzes the directional continuity of torque change in multiple adjacent time periods, extracts the synchronous performance of load response direction in segments with consistent change trends, processes the directional data accordingly, identifies consistent behavior features in the time evolution process, extracts the torque change performance of the stage based on trend continuity, and obtains the torque-in-the-direction trigger segment marker. The shaft disturbance chain extraction module extracts the motion sequence of both ends of the main shaft coupling within a time period based on the torque-in-the-south trigger segment marker, identifies the disturbance direction switching relationship between adjacent time points, sorts the behavior nodes that continuously exhibit disturbance reversal into a chain, removes single jump nodes, and filters out related continuous disturbance paths to obtain the shaft disturbance response chain structure. The directional behavior collaborative judgment module obtains the sequence of changes in the main shaft displacement direction and the change in the load-side torque direction within the same segment based on the time period index of the shaft disturbance response chain structure. It compares the time position of the directional change, identifies the situation where the directional change and the time position coincide. It analyzes the synchronous distribution characteristics of the directional change to obtain the directional change cross recognition label. The inertial response feature extraction module extracts the inertial change behavior on the inclined inertial surface of the axis segment based on the period covered by the direction change cross recognition label, identifies whether the inertial response path accumulates in one direction for a long time, judges whether the structural tilting trend continues to appear in adjacent axis segments, and identifies whether there is a consistent inertial response trend in the continuous area of ​​the structure, thus obtaining the inertial lateral response structural segment. The state excitation node generation module extracts the consistent directional behavior from the torque-in-the-south-direction trigger segment marker and the directional change cross recognition label based on the response structure and directional trend of the inertial lateral response structure segment. It performs time-series comparison, analyzes the connection relationship of multiple directional trends in the disturbance behavior process, identifies the behavior sequence that shows a common directional change in continuous evolution, and obtains the transmission state abnormality diagnosis output result.

[0023] The torque-induced trigger segment markers include trend-consistent segment identifiers, load response synchronization characteristics, and time evolution direction labels. The shaft system disturbance response chain segment structure includes disturbance direction switching sequences, disturbance reversal node chains, and continuous disturbance paths. The direction change cross identification labels include a set of direction switching time points, direction synchronization distribution patterns, and cross change comparison information. The inertial lateral response structure segment includes tilt inertia direction trends, response continuity between shaft segments, and inertial consistency characteristics of continuous structural regions. The transmission state abnormality diagnosis output results include directional behavior time sequence comparison relationships, multi-trend connection characteristics, and abnormal behavior co-evolution sequences.

[0024] Specifically, such as Figure 2 , 3 As shown, the torque change behavior recognition module includes: The trend extraction submodule obtains the torque change sequence and load response direction sequence formed by the main propulsion bearing housing during the transmission process, analyzes the torque direction trend in adjacent time periods, filters continuous time periods in which the direction has not reversed, extracts the time distribution of the continuous direction sequence, and obtains the torque direction continuous segment data. To obtain the torque variation sequence and load response direction sequence generated during the transmission process of the main propulsion bearing housing, strain gauge sensors deployed on the surface of the main bearing housing and corresponding load response devices can be used for joint acquisition. The torque variation sequence represents the trajectory of the rotational torque applied to the bearing during continuous shaft movement, while the load response direction sequence represents the force direction switching behavior of the main propulsion system in the downstream segment of operation. Combining the data, bidirectional time series quantities can be obtained. Furthermore, the trend of the torque variation sequence needs to be judged, and the directional state corresponding to each acquisition point needs to be defined. The directional state is divided into directional maintenance or directional reversal state based on the change trend between the torque value of the previous data point and the current point. Then, continuous segments in which the directional state has not reversed are extracted in chronological order as candidate continuous directional segments. Subsequently, all torques within the candidate segments are... The load response direction state corresponding to each direction is compared and judged one by one. The judgment criterion can be classified by whether the load response is in the same direction under each direction state. In this process, if there are points with inconsistent directions, the segment is removed. The continuous segments with strong directional consistency are retained, and their start and end positions are calculated to obtain the segment boundary markers that are continuous in time but do not produce directional reversal. For example, when the torque direction remains positive and the load response direction also remains positive, the direction of all corresponding torque points in the sequence can be defined as "continuously positive". It is then judged whether a continuous sequence is formed in the points of change of direction before and after. If the fluctuation of the value before and after does not produce directional reversal, its directional continuity can be confirmed, and the time range of the corresponding sequence can be extracted. Finally, the torque direction continuous segment data is obtained by filtering according to the two criteria of consistent directional trend and continuous time distribution.

[0025] The direction comparison submodule extracts load response direction change data within the same segment based on the continuous torque direction segment data. It then compares the direction of each time point to see if they are consistent, extracts segments that maintain the same direction, and obtains the direction synchronization relationship distribution sequence. First, extract the load response direction change data that is consistent with its time range. During the acquisition process, the load response direction can be detected by an angle encoder installed on the load-side coupling housing to determine its rotation trend and thus the direction of rotation. The data is a sequence of direction angle changes recorded at the same sampling interval, and converted into a symbolic form for direction determination. For example, clockwise is set as "positive" and counterclockwise as "negative". After extracting the direction data, its time points need to be matched one by one with the time points within the continuous torque direction segment. Each set of direction symbols is matched. If the two direction symbols are consistent at the same time point, it is counted as a direction synchronization point; otherwise, it is considered a direction inconsistency point. This constructs a complete time-direction matching table. Further, the data is scanned in chronological order. The table is used to count the interval length between consecutive directional synchronization points. If the same directional sign is shown in three or more consecutive sampling points, they are classified into the same synchronization segment. Isolated sampling points that show only one reverse direction in the middle should be removed and regarded as interference items. A bidirectional detection method can be used for judgment, that is, after a single point reverses, it is checked whether the preceding and following directions are consistent. If they are consistent, the single point interference is eliminated to ensure that the synchronization segment is complete and continuous. In practice, if the torque and load directions are both "positive" in sample numbers 10 to 20, then all points between numbers 10 and 20 are uniformly marked as directional synchronization segments. Finally, the start and end numbers or time ranges of all synchronization segments that meet the conditions are extracted, and their distribution positions in the total segment are recorded to obtain the directional synchronization relationship distribution sequence.

[0026] The segment filtering submodule extracts the start and end times of the direction synchronization segment based on the direction synchronization relationship distribution sequence, removes the sequence segments with continuous fluctuations in the direction switching state, and filters the segments with continuous direction change trends and consistent with the load response direction within the time range to obtain the torque-in-the-direction trigger segment marker. First, the start and end index positions of each synchronization segment are extracted and converted into a time range consistent with the original data. During extraction, the direction synchronization sequence table needs to be traversed row by row, recording continuous intervals where the sign remains consistent. If the sign remains consistent from row n to row m, this time interval is extracted as a candidate segment. Simultaneously, the torque sampling time and load direction data index corresponding to its start and end are recorded. Within this candidate segment set, the stability of the direction state during time progression is further analyzed. By determining whether the number of sign changes within a segment is continuous, it can be identified whether the direction state is continuously switching. For example, if the number of times the direction indicator changes from positive to negative and back to positive exceeds a preset value within an interval, it is considered that the direction state fluctuates within that interval, and such segments need to be moved. In order to prevent it from entering the subsequent judgment process, the preset value can be extracted from the average change frequency in the sample. For example, after extracting 50 sets of direction change sequences, the change frequency of each segment is counted, and the median is taken as the operation benchmark. Then, the number of internal changes in each candidate segment is compared to see if it exceeds this value. If it does, it is not adopted. For the remaining segments, it is necessary to determine whether the torque direction trend is consistent with the load direction. In the judgment process, the torque direction and load direction within the start and end time range of the synchronization segment are matched point by point, and the proportion of the number of consistent signs is counted. When the number of consistent direction points accounts for more than 90% of the total number of points in the current segment, the segment is considered to have consistency. Finally, all segments that meet this condition are numbered and merged into a unified result set to obtain the torque same direction trigger segment mark.

[0027] Specifically, such as Figure 2 , 4 As shown, the shaft disturbance chain extraction module includes: The motion sequence extraction submodule extracts the displacement direction at both ends of the main shaft coupling based on the torque-in-the-direction trigger segment marker, sorts the displacement direction by time to form a motion trend sequence, analyzes the direction evolution within the continuous interval, extracts the direction data segment, and obtains the double-end direction change sequence. First, determine the start and end range for maintaining consistency in torque direction across all time index segments. Read the axial displacement data of both ends of the spindle coupling within the corresponding time period, extract the position change values ​​at each moment, and calculate the direction indicator using the displacement difference between two consecutive moments. Set a positive difference as positive and a negative difference as negative, applying this criterion to every sampling point pair within the entire segment to generate a direction indicator sequence point by point. Then, store the left and right direction data side-by-side in chronological order to form a dual-end direction trend data group. Next, classify and statistically analyze all direction indicator points, counting the number of positive, negative, and neutral changes to determine whether the directional state at both ends exhibits phased consistency, reversal, or... within a continuous time range. The jump method divides the time series into several continuous trend segments based on the combination relationship of the directional states at both ends. The directional change relationship within each segment maintains constant characteristics. Then, the directional data segments are extracted using the start and end sampling time of each trend segment as boundaries, and the time axis representation is standardized for subsequent comparison and combination. For example, in a certain segment, if the left end of the main axis is continuously positive and the right end is continuously negative, and the fluctuation period of the two is basically the same, then the data segment will be identified as a reverse trend segment. If both ends of a certain segment are positive and continue to maintain more than 10 sampling points, then the data segment is considered a positive trend segment. After the above processing, all directional data segments that meet the trend stability requirements can be extracted to obtain the double-ended directional change sequence.

[0028] The perturbation direction identification submodule calls the double-ended direction change sequence to analyze whether the directions of adjacent nodes have reversed, extracts the direction pairs that show continuous switching behavior, identifies the perturbation direction change segments, and obtains the perturbation reversal change sequence; First, the direction indicator values ​​at the left and right ends of each consecutive time node are read, and the direction data is arranged in chronological order. Changes in direction indicators are used as the basis for identification; for example, if the previous node's direction is positive and the next node's direction is negative or changes from negative to positive, it is considered a direction reversal operation. Each such switch is recorded as a reversal event. During the direction data reading process, a point-by-point sliding window is used, with the window size set to two sampling points. The start and end times of the current window are recorded, and the direction indicators at both ends of the window are compared. If a direction change occurs, the time pair is recorded as a disturbance. Then, all changed node positions are statistically analyzed and classified, and adjacent reversed nodes are grouped... The time interval between reversals is included in the calculation to determine whether the reversal behavior is a continuous event. By setting the time interval range, it is determined whether the disturbance behavior constitutes a continuous reversal process. In practice, for example, if the collected direction sequence is positive-positive-negative-negative-positive-negative, the sliding window will identify three sets of direction reversal node pairs: positive-negative, negative-positive, and positive-negative. If the time intervals all meet the judgment conditions for continuous disturbance, these three sets of nodes will be integrated into the same disturbance reversal segment. The consistency trend and duration of each disturbance direction segment are further judged, and interference segments with insignificant changes in direction state or intervals far exceeding the set threshold are removed. Only direction reversal behaviors with continuity and obvious trends are retained, and finally, the disturbance reversal change sequence is obtained.

[0029] The reverse path filtering submodule determines the temporal repeatability of the reversal direction based on the perturbation reversal change sequence, eliminates jump segments that cannot form a continuous path, filters path intervals that continuously experience perturbations and have related directional trends, extracts continuous behavior chain groups, and obtains the axis system perturbation response chain structure. First, the start and end node positions corresponding to each group of direction switching events are obtained. The time interval data between node pairs is retrieved to determine the continuity of each reversal behavior on the time axis. If there is a sudden change between node pairs and the interval is greater than the continuous disturbance judgment value, the segment is defined as a jump segment and removed. After filtering, the disturbance behavior sequence formed by the remaining node pairs is traversed and compared to detect its continuity and directional trend relationship. Specifically, by reading the direction change identifiers of adjacent disturbance segments, the direction vectors are compared pairwise to extract path combinations whose direction trends are consistent in time, and it is determined whether there is a disturbance within the continuous segment. The trend sequence of intersecting motion and displacement directions, while satisfying continuity and directional consistency, extracts all node positions associated with these paths to form a behavior chain group. For example, if there is a forward rotation disturbance between nodes A and B, and also between nodes B and C, and the time interval between the two segments is within a set range, then AB and BC can be merged into a continuous forward rotation chain. After further obtaining the superposition relationship between displacement and disturbance directions in the path, the path intervals that satisfy the superposition trend are summarized and rearranged. Based on the start and end distribution of nodes in the behavior chain group, the complete interval and direction of each chain group are output, and finally the axis disturbance response chain segment structure is obtained.

[0030] Specifically, such as Figure 2 , 5 As shown, the directional behavior collaborative judgment module includes: The direction sequence extraction submodule is based on the shaft system disturbance response chain structure. It extracts the sequence of changes in the direction of the main shaft displacement and the behavior of changes in the direction of the load torque. The direction change process is arranged in chronological order, and data segments with continuity are selected to obtain a bidirectional change comparison sequence. First, extract the spindle displacement direction change data and load torque direction change data sequentially from the time index covered by the chain segment. After obtaining these two types of data, call the direction code corresponding to each time node to construct a direction change sequence in chronological order. The spindle displacement direction is identified by left and right turns, and the load torque direction is distinguished by increment and decrement signs. In the specific implementation process, first, align the two types of data according to the time label. If there is a spindle direction change at a certain moment but no corresponding torque data, then that moment is marked as missing in the comparison, and vice versa. After completing the basic information, arrange the entire sequence by time and construct a direction combination matrix for the corresponding points. Judge each group of data in the matrix. The existence of continuous directional states is determined by using the consistency of the directions of three adjacent time nodes as the initial criterion for continuity. When a direction change occurs, the corresponding node number is recorded, and all directional segments are traversed in sequence. For each segment, it is calculated whether its internal directional state remains unchanged. For example, if the spindle displacement direction is positive in the encoding and is positive for three consecutive time points, it is considered as a group of directional consistent segments. Further, it is determined whether the direction state of the load torque in the segment is synchronously positive. If the synchronization is established, the segment is retained as valid data. After multiple directional consistent segments are formed, their time ranges are re-labeled, and the valid data segments are merged into a continuous directional matching segment group, finally obtaining a bidirectional change comparison sequence.

[0031] The switching node identification submodule calls the bidirectional change comparison sequence, compares the position of the change time point in the process of directional change, identifies whether the directional switch occurs at the same time point, extracts the cross nodes in the sequence that show synchronous mutation, and obtains the bidirectional synchronous switching point set; First, the time points of change in the state of each direction are obtained from the change data of the spindle displacement direction and the load torque direction. After obtaining the two sequences, the nodes in which the state of the direction changes are identified. That is, when the values ​​of the codes of two adjacent directions change, their corresponding times are recorded as candidate nodes. Then, the candidate node lists in the two direction sequences are cross-processed to determine whether their time points are consistent or occur simultaneously within a set small time deviation range. If this condition is met, the node is considered to be a bidirectional synchronous switching behavior. When performing the comparison operation on each pair of candidate nodes, a synchronization judgment needs to be set. A threshold is set, for example, to allow a time error of one time sampling period. If the spindle direction changes at time t and the load torque direction changes at time t or t+1, it is considered a synchronous switch. For example, if the spindle direction changes from positive to negative in frame 60 and the load torque direction changes from negative to positive in frame 61, then the node meets the synchronization condition and enters the synchronization node set. After filtering out all the cross nodes that meet the conditions, they are arranged into a list in chronological order. Each node is accompanied by its original time index, direction change pair and its previous and subsequent state values ​​as traceable key parameters, and finally a set of bidirectional synchronous switching points is obtained.

[0032] The synchronous trend analysis submodule is based on the bidirectional synchronous switching point set. It determines the number and interval density of nodes in the same period, identifies whether there is a periodic trend, extracts the segments where nodes overlap regularly, and obtains the direction change and cross recognition label. First, the time index is extracted from each node as the main judgment parameter. The node sequence is traversed sequentially, and the time interval between adjacent nodes is compared using the difference. When multiple interval values ​​continuously fall within the set period density range, the group of nodes is considered to have a periodic repetition trend. Further, it is determined whether the distribution of the number of nodes in each group under this periodic trend meets the set minimum node quantity threshold. For example, if a minimum of 5 nodes are set per period, and the number of nodes in the current period exceeds this value, it is retained as a valid period group. For each valid period, the positions of all nodes within the period are statistically analyzed to determine whether there is a high degree of overlap in time and position among the nodes in each period. If three consecutive nodes within a periodic segment appear near a specific time position, and the deviation does not exceed the preset sampling point tolerance range (e.g., if a device has a tolerance of 3 frames, and the nodes appear at the 20th, 21st, and 19th frames in the three periodic segments), they are considered to be overlapping. Then, all periodic segments that meet the above overlapping distribution characteristics are extracted as key intervals where direction changes occur frequently and synchronization is obvious. For example, if a device has concentrated bidirectional synchronization switching points, and the 5th, 25th, and 45th frames appear consecutively at intervals of 20 frames, and each switching is within the range of similar frames, then this sequence segment is extracted and enters the next processing step, ultimately obtaining the direction change cross recognition tag.

[0033] Specifically, such as Figure 2 , 6 As shown, the inertial response feature extraction module includes: The inertia data extraction submodule extracts inertial response direction data in the inclined inertia surface of the axis segment within the time period covered by the direction change cross recognition label, filters the continuous trend of inertial direction in the time process, determines whether there is a unilateral offset trend in the inertial change direction in the time period, and obtains the inertial offset change trajectory. First, extract the inertial data of each frame of the axis segment within the time period. Locate the inertial response direction of the axis segment at each moment in the inclined inertial plane. Using the principal axis center as a reference point, measure the angle change between the current response direction and the direction of the previous frame. Arrange this change according to the frame sequence to obtain a complete sequence of inertial direction evolution. In data processing, call each direction response sequence, filtering out intervals with drastic and irregular continuous trends in direction, retaining only continuous sequences where the angle change of the inertial direction between consecutive frames does not show a reverse switching. Then calculate the response direction within this continuous sequence. The cumulative deflection angle is calculated, and the distribution trend of the offset direction in the entire sequence is statistically analyzed to determine whether the trend always shifts to one side. In this judgment process, a directional offset threshold is set and a continuous test is performed. For example, the condition for judging a unilateral offset trend is that the inertial direction angle change of more than 70% of the frames in the sequence is in the same direction. If this condition is met, the time period is considered to have a unilateral offset trend. To illustrate with actual data, if the main axis of a device shifts in the same direction in 58 out of 80 frames, it is considered to meet the unilateral trend, and the inertial offset change trajectory is finally obtained.

[0034] The response path analysis submodule extracts the extension path of the inertial response of the shaft segment in time based on the trajectory of inertial offset change, identifies whether the inertial response direction between adjacent shaft segments extends continuously, and judges whether the extension trend of the direction remains consistent in the process of time, thus obtaining the inertial transmission path of the shaft segment. First, identify the time marker corresponding to each segment of inertial direction change, extract the response path on the corresponding time axis, and sequentially extract the inertial response direction data of each segment from the starting point of the segment division. Then, compare this direction information with adjacent segments one by one to confirm whether the direction direction is continuous. During the continuity determination process, the difference between the response direction angles of two segments is calculated. If the angle difference between consecutive segments does not exceed 10 degrees, it is considered that the direction extension is consistent. Repeat this process to continuously traverse multiple segments, constructing a response trend sequence for each segment on the time axis. By comparing multiple segments, it can be identified whether the direction continues to extend in the same direction over time. For example, if a certain... The structure is divided into three segments: A, B, and C. The inertial response direction of segment A is northeast, segment B is slightly east, and segment C is southeast. The angle between A and B is calculated to be 45 degrees, and the angle between B and C is 60 degrees. The difference between the two directions does not exceed the set upper limit. Therefore, the entire segment from A to C is determined to be a segment with consistent inertial direction. On the entire time axis, after comparing multiple segments sequentially, sequences with drastic changes in direction are eliminated. The remaining segments with consistent direction are combined to form an inertial response path sequence with both time continuity and direction consistency. Through this sequence, the transmission trajectory of the inertial response from the starting point to the ending point can be identified, and the inertial transmission path of the segment can be obtained.

[0035] The directional trend assessment submodule analyzes whether the inertial direction evolves synchronously in the continuous region of the structure based on the inertial transmission path of the axis segment, filters inertial response segments with continuous characteristics, judges the consistency of the response direction in the spatial axis segment, and obtains the inertial lateral response structural segment. First, extract the inertial response direction data of each axis segment in the path and record its position index within the continuous structural region. Following the spatial distribution order, compare the direction vector differences between adjacent axis segments sequentially. Establish direction correspondences by combining axis segment numbers, and determine whether each segment maintains a consistent offset direction during spatial advancement. In practice, suppose axis segments 1, 2, and 3 are located adjacent positions within the continuous structural region, with axis segment 1 oriented slightly east, axis segment 2 oriented southeast, and axis segment 3 oriented due south. Then, compare the directions of axis segments 1 and 2, and 2 and 3 respectively. The comparison rule is to use the axis segment with the initial direction as the reference, calculate the angle change between the target direction and the reference direction in the plane coordinates. If this change continuously shifts in the same direction, it is considered... To predict the synchronous evolution trend of the inertial direction, an inertial direction advancement sequence is constructed based on the judgment result. The sequence is further grouped according to the judgment results in the whole sequence. Combinations that are continuously judged to advance in the same direction are screened out and defined as response segments with continuous characteristics. Axis segments that reverse direction or change direction exceeding a set threshold angle are removed from the synchronization sequence. Consistency evaluation is then performed on the obtained continuous response segments. In this process, the consistency standard is that the angle between the direction vectors is less than 15 degrees. It is judged whether they maintain unidirectional extension in the entire structural space. The path segments that meet the consistent direction advancement are screened out and marked by comparison results. Finally, a sequence of segments with continuous inertial direction advancement and uniform direction in the spatial structure is formed, resulting in an inertial lateral response structural segment.

[0036] Specifically, such as Figure 2 , 7 As shown, the state excitation node generation module includes: The trend behavior extraction submodule uses the direction data of the inertial lateral response structural segment to call the direction sequence in the torque same direction trigger segment marker and the direction change cross recognition tag, compares the start and end relationship of the sequence direction trend, filters the time period that matches the direction change, and obtains the direction synchronization trend sequence. First, the directional information of each node in the structural segment is extracted point by point and organized according to the spatial distribution number of the axis segments. The direction of each axis segment is set as an angle and recorded. During the processing, a unified reference direction is introduced, such as east as 0 degrees, and other directions are recorded as clockwise angles to ensure that directional information from different sources can be compared. Then, the direction labels in the torque-triggered segment are called, and the start and end points marked therein are extracted as direction change segments. The angle difference between the start and end points of the direction in each segment is read. At the same time, the direction sequence associated with the direction change cross recognition label is extracted. The angle before and after each set of direction switching points in the sequence is calculated, and a complete directional trend structure is constructed. Then, the three sets of direction sequences are time aligned. In the specific operation, the inertial lateral deflection direction is compared one by one. The starting and ending directions of each segment in the potential are compared with the starting and ending directions of the torque segment to determine if the direction change process is a change in the same direction. If the starting and ending angle changes of the two segments are in the same direction, they are marked as a record of direction matching. In the comparison operation, a threshold angle judgment boundary is set. When the direction difference of the two segments is in the same direction and the angle difference is between 10 degrees and 180 degrees, it is considered that there is a consistent directional trend. The starting and ending segments that meet the conditions are recorded with corresponding time indices. Then, the direction pairs in the direction change cross recognition label are compared with the direction pairs of the inertial lateral deviation sequence to determine whether they show the same trend of change within the same time period. The time index segments with three consistent values ​​are output as trend intervals. Finally, all time interval sequences that meet the consistency of directional trend are summarized to obtain the directional synchronization trend sequence.

[0037] The temporal sequence connection analysis submodule extracts the time nodes of the direction sequence based on the direction synchronization trend sequence, compares the sequential relationship of the direction changes, identifies the segments where the direction turning points are continuously distributed on the timeline, and obtains the continuous path characteristics of the direction evolution. First, the time index points of each direction sequence are read one by one. The start and end times of the direction sequence are extracted as interval boundaries. The nodes within all intervals are then organized into time sequences. The time index is set as the main sequence, and the direction change markers are recorded as secondary information at the corresponding positions in the sequence. Then, the positions of two adjacent direction change events in the sequence are compared to obtain the time difference between adjacent events. This further determines whether there is a sequential relationship between the two events. In the specific execution, the continuous judgment interval is set to 1 to 5 units of time. If the time position of the later node is greater than that of the earlier node and the difference is within the interval, the two are determined to be consecutive direction change events and are marked as valid connection nodes in the sequence. Based on this, a combination and sorting operation is performed on all direction change nodes to construct a direction change timeline. The structure is then analyzed, and the change type of each pair of directional change points is read by marking the directional symbols. The directional difference is judged by combining the previous directional change with the next directional change. If the directional change switches from an upward trend to a downward trend or the opposite trend, and the two time points are closely connected, the point is determined to be a directional turning point. Then, all directional change nodes with such turning point judgments are searched in the time series. The spacing between the time index sequences of multiple turning points is statistically analyzed. When all the turning points obtained by the statistics are spaced between 1 and 5 units in the time series and the directional change trends are continuously connected, they are marked as a segment with continuous time distribution. At the same time, the first and last time indices of the segment are extracted as boundaries. Then, the sequence constructed by each directional trend point in the segment is used as the output of the directional trend evolution chain, and finally the continuous path feature of directional evolution is obtained.

[0038] The direction consistency judgment submodule calls the direction evolution continuous path feature to extract the direction change performance in the disturbance section, judges whether there is a behavior that maintains the same trend in the sequence, extracts the continuous behavior section, and obtains the transmission state abnormality diagnosis output result. First, all directional change data in the path features are read and sorted by time. A directional change marker corresponding to each time node is set as the judgment criterion. The directional markers are encoded to form a directional sequence with a time index. Then, the time interval within the disturbance segment is used as a filtering condition to filter out all directional data within that time range as the analysis target. Two adjacent directional markers are read sequentially. If two consecutive directional markers are consistent, it is considered that the directional trend has not changed within that time interval, and the start and end times are recorded as the initial behavior segment. The directional sequence is then traversed. If a situation is encountered where multiple consecutive items have not changed direction again, it is judged as continuous directional maintenance. If there is a directional change point in the sequence, the current segment recording is interrupted, the starting point of the next continuous segment is remarked, and the above process is repeated. Furthermore, all continuously maintaining directional consistent behavior segments are sorted on the time axis, and the process is accumulated segment by segment. The duration of the segment is added and its proportion within the total disturbance time is calculated. By setting a directional consistency ratio benchmark value, it is determined whether the proportion of continuous directional trends in the disturbance time interval exceeds the set value. If the proportion exceeds the benchmark threshold, the segment is extracted as a continuous behavior segment, and the direction code and time start and end points are marked. Finally, all continuous segments that meet the conditions are output as the analysis results. In practical applications, the directional consistency judgment ratio threshold can be set to 0.7, that is, when the continuous directional segment accounts for more than 70% of the disturbance segment time length, it is considered a directional consistent segment. For a specific example, if the total length of the disturbance segment is 100 units of time, and multiple directional consistent sub-segments are identified within it, with a total duration of 75 units of time, then the disturbance segment is identified as a directional consistency abnormal segment. Such continuous segments are considered as potential abnormal areas in transmission state diagnosis, and the final transmission state abnormality diagnosis output result is obtained.

[0039] Please see Figure 8 The self-diagnosis method for the state of a ship's transmission system based on intelligent sensing is executed based on the aforementioned self-diagnosis system for the state of a ship's transmission system based on intelligent sensing, and includes the following steps: S1: Obtain the torque change trend and load response direction data of the main propulsion bearing housing, analyze the change of torque direction in a continuous time period, filter time segments with consistent directions, determine whether the load directions are synchronized, extract the behavioral characteristics of the corresponding time period, and obtain the torque-in-the-direction trigger segment markers. S2: Based on the torque-in-the-direction trigger segment marker, extract the motion direction sequence at both ends of the main shaft coupling, analyze the change of the disturbance direction between adjacent time points, determine whether there is a repeatedly switching motion trend, extract the continuous disturbance performance, and obtain the shaft system disturbance response chain structure. S3: Based on the shaft system disturbance response chain structure, extract the direction change sequence of the main shaft displacement and load torque, compare the change position of the direction, determine whether they coincide at the time node, analyze the distribution of the change point, and obtain the direction change cross recognition label. S4: Based on the direction change cross recognition label, extract the response direction data of the inclined inertia surface of the shaft segment, analyze the inertial change trend between shaft segments, determine whether the inertial response is continuously transmitted along the same direction, and obtain the inertial lateral response structure segment. S5: Based on the inertial lateral response structure segment, the directional performance of the torque-in-the-direction behavior segment and the direction-synchronous change node group is called. The time sequence and continuity are compared to determine whether a continuous and consistent trend behavior is formed, and the abnormal transmission state diagnosis output result is obtained.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A condition self-diagnosis system for ship transmission devices based on intelligent sensing, characterized in that, The system includes: The torque change behavior recognition module obtains the torque trend and load direction sequence of the main propulsion bearing housing, determines whether the torque direction and load direction are synchronized, filters out time periods with consistent directions, and obtains torque-in-the-direction trigger segment markers. The shaft disturbance chain extraction module extracts the motion sequence of the main shaft coupling based on the torque-in-the-direction trigger segment marker, identifies the disturbance direction reversal behavior, filters the continuous disturbance paths that occur continuously within the preset time window of the direction change, and obtains the shaft disturbance response chain structure. The directional behavior collaborative judgment module extracts the main shaft displacement and load torque direction sequence based on the shaft system disturbance response chain structure, compares the time position of the changing node, and obtains the directional change cross recognition label. The inertial response feature extraction module extracts the response direction of the inclined inertia surface of the shaft segment based on the direction change cross recognition label, judges the direction continuity, and obtains the inertial lateral deflection response structure segment. The state excitation node generation module, based on the inertial lateral response structure fragment, calls the direction information, analyzes the trend consistency and time sequence, identifies the sequence of behaviors exhibiting common directional changes in continuous evolution, and obtains the transmission state abnormality diagnosis output result.

2. The self-diagnostic system for the state of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The torque-induced trigger segment marker includes a trend-consistent segment identifier, load response synchronization characteristics, and time evolution direction label. The shaft system disturbance response chain segment structure includes a disturbance direction switching sequence, a disturbance reversal node chain, and a continuous disturbance path. The direction change cross identification label includes a set of direction switching time points, a direction synchronization distribution pattern, and cross change comparison information. The inertial lateral response structure segment includes a tilt inertia direction trend, response continuity between shaft segments, and inertial consistency characteristics of continuous structural regions. The transmission state abnormality diagnosis output results include a directional behavior time sequence comparison relationship, multi-trend connection characteristics, and a co-evolution sequence of abnormal behavior.

3. The self-diagnostic system for the state of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The torque trend refers to the direction and trajectory of the torque generated by the main propulsion bearing housing during transmission. By analyzing the time series, the continuous segment of the torque direction can be identified to determine whether the direction remains unchanged. The load direction sequence refers to the record of the rotational direction change of the load-side coupling. The direction data sequence is formed by collecting the data through an angle encoder to determine the motion direction and change behavior of the load.

4. The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The disturbance direction reversal behavior refers to the phenomenon that the direction flips between adjacent time points in the motion sequence at both ends of the axis system. The reversal node is identified to analyze the disturbance process. The inclined inertia surface of the shaft segment refers to the directional plane in which the shaft segment generates an inertial response during operation. It can be used to analyze the continuity and consistency of the inertial direction between shaft segments and identify the dynamic response trend of the structure during disturbance.

5. The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The torque change behavior recognition module includes: The trend extraction submodule obtains the torque change sequence and load response direction sequence formed by the main propulsion bearing housing during the transmission process, analyzes the torque direction trend in adjacent time periods, filters continuous time periods in which the direction has not reversed, extracts the time distribution of the continuous direction sequence, and obtains the torque direction continuous segment data. Based on the continuous torque direction segment data, the direction comparison submodule extracts the load response direction change data within the same segment, corresponds to the direction sequence, compares whether the direction of each time point is consistent, extracts the segments that continuously maintain the same direction, and obtains the direction synchronization relationship distribution sequence. Based on the distribution sequence of the directional synchronization relationship, the segment filtering submodule extracts the start and end times of the directional synchronization segments, removes sequence segments with continuous fluctuations in the directional switching state, and filters segments with continuous directional change trends that are consistent with the load response direction within the time range, thus obtaining torque-in-the-direction trigger segment markers.

6. The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The shaft disturbance chain extraction module includes: The motion sequence extraction submodule extracts the displacement direction at both ends of the main shaft coupling based on the torque-in-the-direction trigger segment marker, sorts the displacement direction by time to form a motion trend sequence, analyzes the direction evolution within the continuous interval, extracts the direction data segment, and obtains the double-end direction change sequence. The perturbation direction identification submodule calls the dual-end direction change sequence, analyzes whether the directions of adjacent nodes have reversed, extracts the direction pairs that show continuous switching behavior, identifies the perturbation direction change segments, and obtains the perturbation reversal change sequence; The reverse path filtering submodule determines the repetition of the reversal direction in time based on the perturbation reversal change sequence, eliminates jump segments that cannot form a continuous path, filters path intervals that continuously cause perturbations and have related directional trends, extracts continuous behavior chain groups, and obtains the axis system perturbation response chain structure.

7. The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The direction behavior collaborative judgment module includes: The direction sequence extraction submodule extracts the sequence of changes in the main shaft displacement direction and the behavior of changes in the load torque direction based on the structure of the shaft system disturbance response chain. It arranges the direction change process in chronological order, filters out data segments with continuity, and obtains a bidirectional change comparison sequence. The switching node identification submodule calls the bidirectional change comparison sequence, compares the change time point positions during the direction change process, identifies whether the direction switch occurs at the same time point, extracts the cross nodes that show synchronous mutation in the sequence, and obtains the bidirectional synchronous switching point set. The synchronous trend analysis submodule, based on the bidirectional synchronous switching point set, determines the number and interval density of nodes in the same period, identifies whether there is a periodic trend, extracts the segments where nodes regularly overlap, and obtains the direction change cross recognition label.

8. The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The inertial response feature extraction module includes: The inertia data extraction submodule extracts inertial response direction data in the inclined inertia surface of the axis segment within the time period covered by the direction change cross recognition label, filters the continuous trend of the inertial direction in the time process, determines whether there is a unilateral offset trend in the inertial change direction in the time period, and obtains the inertial offset change trajectory. Based on the inertial offset change trajectory, the response path analysis submodule extracts the extension path of the shaft segment inertial response in time, identifies whether the inertial response direction between adjacent shaft segments extends continuously, and determines whether the direction extension trend remains consistent during the time process, thereby obtaining the shaft segment inertial transmission path. The directional trend assessment submodule analyzes whether the inertial direction evolves synchronously in the continuous region of the structure based on the inertial transmission path of the axis segment, filters inertial response segments with continuous characteristics, judges the consistency of the response direction in the spatial axis segment, and obtains the inertial lateral response structural segment.

9. The self-diagnostic system for the state of a ship's transmission device based on intelligent sensing according to claim 1, characterized in that, The state excitation node generation module includes: The trend behavior extraction submodule, based on the direction data of the inertial lateral response structure segment, calls the direction sequence in the torque same direction trigger segment marker and the direction change cross recognition tag, compares the start and end relationship of the sequence direction trend, filters the time period that matches the direction change, and obtains the direction synchronization trend sequence. The temporal sequence connection analysis submodule extracts the time nodes of the direction sequence based on the direction synchronization trend sequence, compares the sequential relationship of the direction changes, identifies the segments where the direction turning points are continuously distributed on the timeline, and obtains the continuous path characteristics of the direction evolution. The direction consistency judgment submodule calls the continuous path feature of the direction evolution, extracts the direction change performance in the disturbance section, judges whether there is a behavior that maintains the same trend in the sequence, extracts the continuous behavior section, and obtains the abnormal transmission state diagnosis output result.

10. A method for self-diagnosing the condition of a ship's transmission system based on intelligent sensing, characterized in that, The self-diagnostic system for the condition of a ship's transmission device based on intelligent sensing, as described in any one of claims 1-9, includes the following steps: S1: Obtain the torque change trend and load response direction data of the main propulsion bearing housing, analyze the change of torque direction in a continuous time period, filter time segments with consistent directions, determine whether the load directions are synchronized, extract the behavioral characteristics of the corresponding time period, and obtain the torque-in-the-direction trigger segment markers. S2: Based on the torque-in-the-direction trigger segment marker, extract the motion direction sequence at both ends of the main shaft coupling, analyze the change of the disturbance direction between adjacent time points, determine whether there is a repeatedly switching motion trend, extract the continuous disturbance performance, and obtain the shaft system disturbance response chain structure. S3: Based on the shaft system disturbance response chain structure, extract the direction change sequence of the main shaft displacement and load torque, compare the change positions of the directions, determine whether they coincide at the time nodes, analyze the distribution of the change points, and obtain the direction change cross recognition label. S4: Based on the direction change cross recognition tag, extract the response direction data of the inclined inertia surface of the shaft segment, analyze the inertial change trend between shaft segments, determine whether the inertial response is continuously transmitted along the same direction, and obtain the inertial lateral response structure segment. S5: Based on the inertial lateral response structure segment, call the directional performance of the torque-in-the-direction behavior segment and the direction-synchronous change node group, compare the time sequence and continuity, determine whether it constitutes a continuous and consistent trend behavior, and obtain the abnormal transmission state diagnosis output result.

Images