Underwater propeller fault detection method and device and computer readable storage medium

Abstract

The invention discloses an underwater propeller fault detection method, an underwater propeller fault detection device and a computer readable storage medium, which are used for improving the accuracy of fault detection. The method comprises the steps that according to the thrust capacity, the thrust direction and the control distribution coefficient of each propeller in the multi-propeller system, the function similarity relation and the control coupling relation between the propellers are established; based on the function similarity relation and the control coupling relation, a combined state prediction model of the multi-propeller system is constructed; collecting current running state data and control input data of each propeller; inputting the control input data into the joint state prediction model to calculate predicted operation state data of each propeller; collecting actual operation state data of each propeller, and calculating a state residual error according to the actual operation state data and the predicted operation state data; and comparing the state residual error with a preset threshold value to judge whether the multi-propeller system has a fault or not.

Description

Technical Field

[0001] This application relates to the technical field of underwater thrusters, and particularly to a method and device for underwater thruster fault detection and a computer-readable storage medium. Background Art

[0002] With the increasing complexity and diversification of underwater operation tasks, as a key power device, the reliable monitoring and fault detection of the operating state of underwater thrusters are particularly important. Timely and accurately detecting the abnormal state of the thruster not only relates to the normal operation of the equipment, but also directly affects the safety and efficiency of the entire underwater operation system.

[0003] Currently, the fault detection technologies commonly used in the market and research fields mainly rely on the acquisition of thruster operation parameters and single-threshold determination, or fault identification based on certain fixed patterns. These methods usually take a single data dimension or a single index as the core and lack the comprehensive utilization of multi-dimensional information. Their detection models are difficult to fully capture the dynamic changes under complex working conditions during the operation of the thruster, resulting in large errors in the detection results.

[0004] Therefore, in the prior art, there are still obvious shortcomings in the accuracy of thruster fault detection technology, which is difficult to meet the requirements of modern underwater propulsion systems for high reliability and high-precision fault diagnosis. Summary of the Invention

[0005] Embodiments of this application provide a method and device for underwater thruster fault detection and a computer-readable storage medium, which can improve the accuracy of fault detection.

[0006] The first aspect of the embodiments of this application provides a method for underwater thruster fault detection, including: Establish a functional similarity relationship and a control coupling relationship between thrusters according to the thrust capacity, thrust direction, and control distribution coefficient of each thruster in a multi-thruster system; Construct a joint state prediction model of the multi-thruster system based on the functional similarity relationship and the control coupling relationship; Collect the current operating state data and control input data of each thruster, where the operating state data includes rotational speed, current, thrust feedback, and attitude response; Input the control input data into the joint state prediction model to calculate the predicted operating state data of each thruster; Collect the actual operating state data of each thruster, and calculate the state residual according to the actual operating state data and the predicted operating state data; where the actual operating state data is the data at the next moment corresponding to the current operating state data; Compare the state residual with a preset threshold to determine whether there is a fault in the multi - thruster system; If there is a fault, determine the faulty thruster and the type of fault according to the distribution characteristics of the state residual and the control coupling relationship.

[0007] Optionally, constructing the joint state prediction model of the multi - thruster system based on the functional similarity relationship and the control coupling relationship includes: Determine the thruster grouping according to the similarity relationship; Determine the joint state transition matrix according to the thruster grouping and the control coupling relationship; Construct the joint state prediction model of the multi - thruster system based on the joint state transition matrix.

[0008] Optionally, constructing the joint state prediction model of the multi - thruster system based on the joint state transition matrix includes: Determine the current joint state vector, input gain matrix, and joint control input matrix according to the parameter information of each thruster; Construct the joint state prediction model of the multi - thruster system according to the current joint state vector, the input gain matrix, the joint control input matrix, and the joint state transition matrix.

[0009] Optionally, the joint state prediction model is specifically designed as:

[0010]

[0011]

[0012]

[0013] Wherein, represents the vector of the predicted next state, represents the current joint state vector, represents the joint control input matrix, represents the input gain matrix, represents the joint state transition matrix.

[0014] Optionally, determining the faulty thruster and the type of fault according to the distribution characteristics of the state residual and the control coupling relationship includes: Construct a residual matrix according to the state residual; Construct a coupling coefficient matrix based on the control coupling relationship; Generate a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix; Input the residual coupling influence matrix into a pre-trained thruster fault identification model to determine the faulty thruster and the type of fault.

[0015] Optionally, the generating the residual coupling influence matrix based on the residual matrix and the coupling coefficient matrix includes: Calculate a normalized coupling weight matrix using the magnitudes of the row residual vectors of the residual matrix and the corresponding elements of the coupling coefficient matrix; Apply a double non-linear transformation to each residual vector in the residual matrix to obtain transformed vectors; Weight and aggregate the transformed vectors through the normalized coupling weight matrix to obtain the residual coupling influence matrix.

[0016] Optionally, the comparing the state residual with a preset threshold to determine whether there is a fault in the multi-thruster system includes: Determine whether each item of the state residual is less than the corresponding preset threshold; If not, determine that there is a fault in the multi-thruster system; if so, determine that there is no fault in the multi-thruster system.

[0017] A second aspect of the embodiments of the present application provides an underwater thruster fault detection device, including: A establishing unit, configured to establish a functional similarity relationship and a control coupling relationship between thrusters according to the thrust capabilities, thrust directions, and control distribution coefficients of the thrusters in the multi-thruster system; A constructing unit, configured to construct a joint state prediction model of the multi-thruster system based on the functional similarity relationship and the control coupling relationship; A first acquisition unit, configured to acquire the current operating state data and control input data of each thruster, where the operating state data includes rotational speed, current, thrust feedback, and attitude response; An input unit, configured to input the control input data into the joint state prediction model to calculate the predicted operating state data of each thruster; A second acquisition unit, configured to acquire the actual operating state data of each thruster, and calculate a state residual according to the actual operating state data and the predicted operating state data; where the actual operating state data is the data at the next moment corresponding to the current operating state data; A judging unit, configured to compare the state residual with a preset threshold to determine whether there is a fault in the multi-thruster system; A determining unit, configured to, if there is a fault, determine the faulty thruster and the type of fault according to the distribution characteristics of the state residual and the control coupling relationship.

[0018] Optionally, the constructing unit includes: The first determination module is configured to determine the thruster grouping according to the similarity relationship; The second determination module is configured to determine the joint state transition matrix according to the thruster grouping and the control coupling relationship; The first construction module is configured to construct a joint state prediction model of the multi-thruster system based on the joint state transition matrix.

[0019] Optionally, the first construction module is specifically configured to: Determine the current joint state vector, the input gain matrix, and the joint control input matrix according to the parameter information of each thruster; Construct a joint state prediction model of the multi-thruster system according to the current joint state vector, the input gain matrix, the joint control input matrix, and the joint state transition matrix.

[0020] Optionally, the joint state prediction model is specifically designed as:

[0021]

[0022]

[0023]

[0024] Wherein, represents the vector of the predicted next state, represents the current joint state vector, represents the joint control input matrix, represents the input gain matrix, represents the joint state transition matrix.

[0025] Optionally, the determination unit includes: The second construction module is configured to construct a residual matrix according to the state residual; The third construction module is configured to construct a coupling coefficient matrix based on the control coupling relationship; The generation module is configured to generate a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix; The input module is configured to input the residual coupling influence matrix into a pre-trained thruster fault identification model to determine the faulty thruster and the fault type.

[0026] Optionally, the generation module is specifically configured to: Calculate a normalized coupling weight matrix by using the magnitude of the row residual vector of the residual matrix and the corresponding elements of the coupling coefficient matrix; Apply a double non - linear transformation to each residual vector in the residual matrix to obtain a transformed vector; Weight - aggregate the transformed vector through the normalized coupling weight matrix to obtain a residual coupling influence matrix.

[0027] Optionally, the determination unit is specifically configured to: Determine whether each item of the state residual is less than the corresponding preset threshold; If not, determine that the multi - thruster system has a fault; if so, determine that the multi - thruster system has no fault.

[0028] The third aspect of the embodiments of the present application provides an underwater thruster fault detection device, including: A processor, a memory, an input - output unit, and a bus; The processor is connected to the memory, the input - output unit, and the bus; The memory stores a program, and the processor calls the program to execute the method in the first aspect and any possible implementation manner of the first aspect.

[0029] The fourth aspect of the embodiments of the present application provides a computer - readable storage medium, on which a program is stored, and when the program is executed on a computer, the computer executes the method in the first aspect and any possible implementation manner of the first aspect.

[0030] From the above technical solutions, the embodiments of the present application have the following advantages: In the embodiments of the present application, the joint state prediction model comprehensively considers the functional similarity and control coupling relationship among multiple thrusters, realizes the accurate prediction of the overall state of the multi - thruster, avoids the problems of information isolation and error accumulation during the prediction of a single thruster, enables the state residual to more truly reflect abnormal situations, and thus greatly improves the accuracy of fault detection and location. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic flowchart of an embodiment of the underwater thruster fault detection method in the embodiments of the present application; Figure 2 It is a schematic flowchart of an embodiment of constructing a joint state prediction model of a multi - thruster system in the embodiments of the present application; Figure 3 It is a schematic flowchart of an embodiment of constructing a joint state prediction model of a multi - thruster system based on a joint state transition matrix in the embodiments of the present application; Figure 4 It is a schematic flowchart of an embodiment of determining a faulty thruster and a fault type in the embodiments of the present application; Figure 5 It is a schematic flowchart of an embodiment for generating a residual coupling influence matrix in an embodiment of the present application; Figure 6 It is a schematic flowchart of an embodiment for determining whether there is a fault in a multi - thruster system in an embodiment of the present application; Figure 7 It is a schematic structural diagram of an embodiment of an underwater thruster fault detection device in an embodiment of the present application; Figure 8 It is a schematic structural diagram of another embodiment of an underwater thruster fault detection device in an embodiment of the present application. Detailed implementation manners

[0032] The embodiments of the present application provide an underwater thruster fault detection method, device and computer - readable storage medium, which are used to improve the accuracy of fault detection.

[0033] The method of the present application can be applied to a server, a terminal or other devices with logical processing capabilities. In this regard, the present application makes no limitation. For the sake of convenience of description, the following description will be made taking the server as the execution subject as an example.

[0034] Next, the embodiments in the present application will be described with reference to the accompanying drawings.

[0035] Please refer to Figure 1 , Figure 1 which is an embodiment of the underwater thruster fault detection method provided by the present application. This embodiment includes: 101. Establish a functional similarity relationship and a control coupling relationship between thrusters according to the thrust capacity, thrust direction and control distribution coefficient of each thruster in the multi - thruster system; The server analyzes according to the thrust capacity, thrust direction and control distribution coefficient of each thruster in the multi - thruster system. First, it extracts the upper limit of the thrust magnitude and the direction vector of each thruster, and combines the control distribution strategy of the system to calculate the substitution ability and response influence degree between thrusters when performing attitude adjustment or position control tasks. The server establishes a functional similarity relationship between thrusters based on the functional similarity degree of thrusters in the control task; then, in combination with the response correlation of each thruster to the same control target, a control coupling relationship is further formed, thereby constituting a quantifiable inter - thruster correlation data structure for subsequent model construction.

[0036] 102. Construct a joint state prediction model of the multi - thruster system based on the functional similarity relationship and the control coupling relationship; Based on the established functional similarity relationships and control coupling relationships above, the server constructs a joint state prediction model. This model treats multiple thrusters as an integrated system with dynamic correlations and uses multivariate modeling methods such as multi-input multi-output system identification (MIMO), neural network prediction models, or extended Kalman filter models, etc., to model the mapping relationship between the control inputs and the changes in the operating states of the entire thruster cluster. This joint model can predict the operating states of multiple thrusters after inputting control signals at any moment, reflecting the collaborative response characteristics between systems. It should be noted that after constructing the joint state prediction model, the server can verify and adjust the joint state prediction model to make it more reliable.

[0037] 103. Collect the current operating state data and control input data of each thruster. The operating state data includes rotational speed, current, thrust feedback, and attitude response. During the operation of the system, the server continuously collects the current operating state data and control input data of each thruster. The operating state data includes the current rotational speed value, current magnitude, thrust feedback (e.g., obtained from a thrust sensor), and attitude response (such as angular velocity or attitude angle change, etc.). These data are collected in real time by sensors and uploaded to the server to reflect the current physical state of the thruster. At the same time, the control input signals issued by the controller, such as voltage or PWM control signals, etc., are collected to drive each thruster to perform actions.

[0038] 104. Input the control input data into the joint state prediction model to calculate the predicted operating state data of each thruster. The server inputs the collected control input data into the joint state prediction model to predict the operating states of each thruster under the current control input. The predicted operating state data corresponds to parameters such as the expected thrust, current, rotational speed, and attitude response in the next moment or within a certain future time window. These predicted values serve as a reference benchmark for subsequent comparison with the actual operating states and are used to evaluate whether the thruster behavior is normal.

[0039] 105. Collect the actual operating state data of each thruster and calculate the state residuals based on the actual operating state data and the predicted operating state data. Among them, the actual operating state data is the data corresponding to the next moment of the current operating state data. After completing the prediction, the server continues to collect the actual operating state data of each thruster at the next moment, that is, the response results actually occurring to the thruster under the action of the control input at the previous moment. The server compares these actual operating state data item by item with the aforementioned predicted operating state data, calculates the difference of each state parameter, and forms state residuals to quantify the deviation between the prediction result and the true response, thereby providing a basis for fault judgment.

[0040] 106. Compare the state residual with a preset threshold to determine whether the multi-thruster system has a fault; The server compares the state residual corresponding to each thruster with the preset threshold set by the system one by one. The preset threshold represents the acceptable prediction error range of the system. If a state residual of a certain thruster exceeds the preset threshold, it is determined that the thruster has abnormal behavior, and then it is determined that there may be a fault in the multi-thruster system. The comparison process can adopt multi-dimensional residual judgment criteria, such as Euclidean distance method or weighted error scoring. It should be noted that the state residual includes information in multiple dimensions, so the preset threshold needs to be set accordingly, and the server needs to compare the residual state with the corresponding preset threshold respectively.

[0041] 107. If there is a fault, the faulty thruster and fault type are determined based on the distribution characteristics of the state residual and the control coupling relationship.

[0042] After determining that there is a fault, the server further analyzes the distribution of state residuals among the thrusters, such as when multiple parameters of a certain thruster exceed the limit at the same time or when a certain type of state parameter has consistency deviation in multiple coupled thrusters. The server analyzes the propagation path of the coupling characteristics in combination with the previously established control coupling relationship and locks the core thruster that caused the fault. Combined with the residual characteristics (such as abnormal thrust, sudden current change, slow attitude response, etc.), the server determines the fault type, such as mechanical jamming, drive failure, or response deviation, thereby identifying the specific faulty thruster and its type.

[0043] In this embodiment, the joint state prediction model comprehensively considers the functional similarities and control coupling relationships among multiple thrusters, realizes the accurate prediction of the overall state of multiple thrusters, avoids the problems of information isolation and error accumulation when predicting a single thruster, and makes the state residual more truly reflect the abnormal situation, thereby greatly improving the accuracy of fault detection and positioning.

[0044] See also Figure 2 In some embodiments of the present application, step 102 of the above embodiment constructs a joint state prediction model of a multi-thruster system based on a functional similarity relationship and a control coupling relationship, and may include the following steps: 201. Determine thruster grouping according to similarity relationship; The server determines the grouping method of thrusters based on the functional similarity relationship between thrusters. The server first reads parameters such as the thrust capacity, response speed, control sensitivity, and thrust direction of each thruster, and calculates the similarity index between thrusters using these parameters. Then, the server uses clustering algorithms (such as K-means, hierarchical clustering, or distance-based partitioning methods) to group thrusters with similar functional characteristics or similar roles in control tasks into the same group. For example, thrusters with the same thrust direction and similar response characteristics can be grouped together. By grouping thrusters, a unified state description can be established for the same type of thrusters in the subsequent modeling process, reducing the modeling complexity while retaining the representativeness of the internal structure of the system.

[0045] 202. Determine the joint state transition matrix based on the thruster grouping and control coupling relationship; The server constructs a joint state transition matrix based on the thruster grouping result in step 201 and combines the control coupling relationship between thrusters. The server first establishes an intra-group state transition model within each thruster group to determine the mapping relationship between the current state and the next state of each thruster in the group. Subsequently, the server combines the control coupling relationship to analyze how the control input or state change in one group affects the response of thrusters in other groups, and converts this mutual influence into a coupling term in the matrix. Finally, the server integrates the local state transition matrices of all groups and the coupling relationship terms into a large-scale joint state transition matrix, which comprehensively reflects the global state evolution characteristics of the multi-thruster system and retains the interaction and interference between each thruster.

[0046] 203. Construct a joint state prediction model for the multi-thruster system based on the joint state transition matrix.

[0047] The server uses the constructed joint state transition matrix to establish a joint state prediction model for the multi-thruster system. The server takes the current operating state and control input of the thrusters as the input of the model, and predicts the operating state of each thruster at the next moment through the state changes and coupling relationships defined in the matrix. The prediction results include state variables such as the rotation speed, current, thrust feedback, and attitude response of each thruster. This model supports unified prediction of the states of all thrusters and can reflect the collaborative response characteristics between thrusters, serving as the basis for subsequent residual analysis and fault detection.

[0048] In this embodiment, the server constructs a joint state transition matrix based on the establishment of thruster grouping and uses this matrix to establish a joint state prediction model, enabling the prediction model to simultaneously consider the functional similarity and control coupling relationship between thrusters, thereby improving the comprehensiveness and accuracy of state prediction, providing a reliable basis for fault detection, and being beneficial to improving the robustness and safety of the multi-thruster system.

[0049] Please refer to Figure 3 In some embodiments of the present application, step 203 of the above embodiments constructs a joint state prediction model for a multi-thruster system based on the joint state transition matrix, which may include the following steps: 301. Determine the current joint state vector, input gain matrix, and joint control input matrix according to the parameter information of each thruster; The server determines the current joint state vector, input gain matrix, and joint control input matrix according to the parameter information of each thruster. The server first collects and integrates the real-time operating parameters of all thrusters, such as rotational speed, thrust feedback, current, and attitude response, etc., and combines these parameters into a joint state vector describing the state of the entire multi-thruster system. At the same time, the server calculates the input gain matrix based on the characteristics and control configurations of the thrusters to reflect the influence amplitude of the control input on the system state. In addition, the server also determines the joint control input matrix according to the current control strategy and control allocation results, indicating the distribution of the control inputs applied to each thruster in the overall system.

[0050] 302. Construct a joint state prediction model for the multi-thruster system according to the current joint state vector, input gain matrix, joint control input matrix, and joint state transition matrix.

[0051] The server uses the joint state vector, input gain matrix, joint control input matrix determined in step 301, and the previously constructed joint state transition matrix to construct a joint state prediction model for the multi-thruster system. The server combines the current joint state with the control input, and through the mapping of the joint state transition matrix and the input gain matrix, predicts the state change of the multi-thruster system at the next moment. This model can simultaneously reflect the dynamic changes of the states of each thruster and the effect of the control input, realize the accurate prediction of the overall operating state of the system, and provide accurate data support for subsequent state residual analysis and fault detection.

[0052] Specifically, the joint state prediction model can be specifically designed as: Formula 1 Formula 2 Formula 3 Formula 4 Wherein, represents the vector of the predicted next state, represents the current joint state vector, represents the joint control input matrix, represents the input gain matrix, represents the joint state transition matrix.

[0053] Represents the state vector of a single thruster, which includes information in four dimensions: rotational speed, current, thrust feedback, and attitude response. Represents the control input of a single thruster, which can include information in multiple dimensions, such as thrust, lateral force, vertical force, moment, rotational speed, angular velocity, control signal strength, phase information, etc. This application does not make any limitations in this regard.

[0054] It should be noted that for the joint state transition matrix The sub - matrix on the diagonal - , thrusters in the same group use the same sub - matrix. For example, if thruster 1 and thruster 2 belong to the same thruster group, then at this time and are equal. For the coupling terms off - diagonal of the joint state transition matrix , they are used to represent the mutual influence terms existing between different thrusters due to the control coupling relationship. These coupling terms are used to reflect the control coupling relationship, that is, the interaction and influence between different thrusters.

[0055] It should be noted that when each thruster is only driven by its own control input, appears as a block - diagonal matrix, in the form of the following formula: Formula 5 However, if there is a certain coupling control channel (such as multiple thrusters sharing a control law), there may also be some non - zero off - diagonal terms. This application does not make any limitations in this regard.

[0056] In this embodiment, the server constructs a joint state prediction model by accurately determining the joint state vector and the input - related matrix, in combination with the joint state transition matrix, which can comprehensively reflect the dynamic characteristics of the multi - thruster system and the influence of control inputs, improve the accuracy and real - time performance of state prediction, and further enhance the reliability and response speed of fault diagnosis, promoting the safe and stable operation of the multi - thruster system.

[0057] Please refer to Figure 4 , in some embodiments of this application, step 107 of the above - mentioned embodiment for determining the faulty thruster and the fault type according to the distribution characteristics of the state residuals and the control coupling relationship may include the following steps: 401. Construct a residual matrix based on the state residuals; The server collects the actual operating status data and corresponding predicted operating status data of each thruster in the multi-thruster system. By comparing the differences between the two at the same moment, the status residuals of each operating parameter of each thruster are calculated. The server classifies and summarizes this residual information according to the thruster number and status type (such as rotational speed, current, thrust feedback, attitude response, etc.) to form a multi-dimensional residual matrix. This residual matrix can reflect in detail the deviation degree between each thruster and the predicted state at the current moment, facilitating subsequent analysis.

[0058] 402. Construct a coupling coefficient matrix based on the control coupling relationship; Based on the design and control allocation scheme of the multi-thruster system, the server analyzes in detail the control coupling relationship between each thruster, quantifies the mutual influence between different thrusters. By collecting and sorting out the control input influence paths and coupling strength information between thrusters, the server establishes a coupling coefficient matrix, and each element in the matrix represents the strength and direction of the control coupling between the corresponding two thrusters. This matrix can systematically describe the mutual coupling effect between thrusters and provide a basis for understanding the residual propagation mechanism.

[0059] 403. Generate a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix; Based on the residual matrix in step 401 and the coupling coefficient matrix in step 402, the server uses mathematical methods such as matrix multiplication or weighted superposition to generate a residual coupling influence matrix. This matrix reflects how the status residual of a single thruster affects the status of other thrusters due to the control coupling relationship between thrusters, thereby revealing the diffusion path and influence range of the residuals in the multi-thruster system. Through this matrix, the server can identify the areas where the residuals are concentrated and the thrusters that may have abnormalities, providing important clues for fault diagnosis.

[0060] 404. Input the residual coupling influence matrix into a pre-trained thruster fault identification model to determine the faulty thruster and the fault type.

[0061] The server inputs the residual coupling influence matrix into a pre-trained thruster fault identification model. This model deeply analyzes the input matrix through machine learning algorithms (such as support vector machines, neural networks, or decision trees, etc.), and identifies the corresponding relationship between the residual pattern and the known fault types. Through the model output, the server determines the specific thruster with a fault and the fault type, such as abnormal thrust, sensor failure, or control malfunction, etc., providing a clear diagnosis result for subsequent fault handling.

[0062] In this embodiment, the server constructs a detailed residual matrix and a coupling coefficient matrix, and uses the residual coupling influence matrix as the input in combination. By leveraging the fault identification model to deeply mine the residual signals of the control coupling between thrusters, the accuracy of fault location and the fineness of fault type discrimination are significantly improved. Furthermore, the fault diagnosis ability and operation safety of the multi-thruster system are enhanced, which helps to take maintenance measures in a timely manner and ensure the stable operation of the system.

[0063] Please refer to Figure 5 , in some embodiments of the present application, step 403 of the above embodiment for generating a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix may include the following steps: 501. Calculate the normalized coupling weight matrix using the amplitude of each row residual vector of the residual matrix and the corresponding elements of the coupling coefficient matrix; The server calculates the normalized coupling weight matrix by using the amplitude of the residual vector corresponding to each row in the residual matrix in combination with the corresponding elements in the coupling coefficient matrix. Specifically, the server first statistically analyzes the residual amplitude of each thruster in the residual matrix to evaluate its influence on the overall state of the system; subsequently, the server combines the control coupling strength between thrusters in the coupling coefficient matrix, and performs a weighted calculation on the residual amplitude and the coupling strength to form a matrix reflecting the residual transfer weights between different thrusters. The server normalizes this matrix to ensure a reasonable weight distribution for subsequent calculations.

[0064] 502. Apply a double non-linear transformation to each residual vector in the residual matrix to obtain a transformed vector; The server applies a double non-linear transformation to each residual vector in the residual matrix to generate a transformed vector. This double non-linear transformation includes non-linear function mappings for the amplitude and direction of the residual vector respectively, which can enhance the key information in the residual signal while suppressing the influence of noise and outliers. Through this transformation, the server highlights the features in the residual vector that may indicate a fault, making the fault information more obvious and reliable in the subsequent weighted aggregation process.

[0065] 503. Weightedly aggregate the transformed vectors through the normalized coupling weight matrix to obtain the residual coupling influence matrix.

[0066] The server uses the normalized coupling weight matrix obtained in step 501 to weightedly aggregate the transformed vectors obtained in step 502 to generate the residual coupling influence matrix. Specifically, the server assigns different weights to the transformed vectors of different thrusters according to the elements in the coupling weight matrix, and performs a linear combination or weighted average of all the transformed vectors according to the weights. The finally obtained residual coupling influence matrix comprehensively reflects the overall propagation and influence of the residuals of each thruster under the control coupling relationship, providing accurate input data for subsequent fault identification.

[0067] In a possible implementation, the server can construct the state residual information of each thruster into a residual matrix: Formula 6 Where, is the state residual vector of the j-th thruster, containing prediction residuals in multiple state dimensions such as rotational speed, current, thrust feedback, and attitude response.

[0068] Then, the server can establish a coupling coefficient matrix based on control coupling information such as the structural layout between thrusters, thrust direction, and control allocation matrix: Formula 7 Where, represents the influence intensity of the state of thruster j on thruster i, thus reflecting the coupling weight in control between thrusters.

[0069] Next, the server can combine the residual matrix R and the coupling coefficient matrix C to calculate the normalized coupling weight matrix: Formula 8 Where Formula 8 introduces the residual magnitude as a regulation factor, so that other thrusters with larger residuals and strong coupling to the current thruster have higher weights in the influence matrix, which is beneficial to highlighting possible fault sources.

[0070] To improve the feature expression ability, the server can input the residual vector of each thruster into a two-channel non-linear mapping function to obtain a transformed vector: Formula 9 Where, is the Sigmoid function; is the hyperbolic tangent function; is the weight matrix; is the bias term; represents element-wise multiplication (i.e., Hadamard product).

[0071] The server enhances the response ability to complex changes in different residual signals by introducing a non-linear gating mechanism, facilitating subsequent processing by the neural network.

[0072] Next, the server uses the normalized coupling weight matrix to perform weighted aggregation on the transformed vector and calculate the residual coupling influence vector . Calculate according to the following formula: Formula 10 Finally, the server can The final residual coupling influence matrix is ​​formed by row combination: Formula 11 The residual coupling influence matrix not only integrates the coupling relationship information and the amplitude change of the residual signal, but also undergoes nonlinear enhancement processing. The final result is It can be used as the input of the downstream thruster fault identification model to effectively improve the fault identification accuracy and positioning precision.

[0073] In this embodiment, the server calculates the normalized coupling weight matrix and performs a double nonlinear transformation on the residual vector, and weights and aggregates the transformation vector to construct a residual coupling influence matrix that reflects the residual propagation and coupling influence. This matrix effectively integrates the residual information and coupling relationship between thrusters, improves the recognition ability of fault signals and the accuracy of overall system fault diagnosis, and enhances the reliability and accuracy of multi-thruster system fault identification.

[0074] See also Figure 6 In some embodiments of the present application, step 106 of the above embodiment compares the state residual with a preset threshold to determine whether the multi-thruster system has a fault, which may include the following steps: 601. Determine whether each item of the state residual is less than the corresponding preset threshold; The server judges the state residual of each thruster in the multi-thruster system one by one, and compares each residual value with its corresponding preset threshold. The server traverses all residual data in the residual matrix in turn, and determines whether each residual is less than the preset safety threshold, so as to evaluate whether the current operating state deviation is within the normal range, ensuring that each thruster and each indicator are fully checked.

[0075] 602. Determine that there is a fault in the multi-thruster system.

[0076] When the server finds that at least one state residual reaches or exceeds the corresponding preset threshold, it determines that the multi-thruster system has a fault. The server marks this abnormal information as a fault signal, triggering the subsequent fault diagnosis and location process, providing a basis for system maintenance and repair, and ensuring the safe and stable operation of the multi-thruster system.

[0077] 603. Determine that there is no fault in the multi-thruster system.

[0078] When the server confirms that all state residuals are less than the corresponding preset thresholds, it is determined that there is no fault in the multi-thruster system. The server records that the current system state is within the normal operating range, and there is no need to start the fault diagnosis program, maintain normal monitoring and control of the system, and ensure the stable operation of the multi-thruster system.

[0079] In this embodiment, the server realizes precise monitoring of the operating state of the multi-propeller system and fault discrimination by strictly comparing the state residuals item by item with the preset thresholds, can detect anomalies in a timely manner and determine the occurrence of faults, or confirm that the system is safe and free of anomalies, thereby improving the reliability and safety of the system operation, reducing potential risks, and ensuring the stable and efficient operation of the multi-propeller system.

[0080] Please refer to Figure 7 , Figure 7 which is an embodiment of the underwater propeller fault detection device provided by this application. This embodiment includes: A building unit 701, configured to establish a functional similarity relationship and a control coupling relationship between the propellers according to the thrust capacity, thrust direction, and control distribution coefficient of each propeller in the multi-propeller system; A constructing unit 702, configured to construct a combined state prediction model of the multi-propeller system based on the functional similarity relationship and the control coupling relationship; A first acquisition unit 703, configured to acquire the current operating state data and control input data of each propeller, where the operating state data includes rotational speed, current, thrust feedback, and attitude response; An input unit 704, configured to input the control input data into the combined state prediction model to calculate the predicted operating state data of each propeller; A second acquisition unit 705, configured to acquire the actual operating state data of each propeller, and calculate the state residual according to the actual operating state data and the predicted operating state data; where the actual operating state data is the data at the next moment corresponding to the current operating state data; A judgment unit 706, configured to compare the state residual with the preset threshold to determine whether there is a fault in the multi-propeller system; A determination unit 707, configured to, if there is a fault, determine the faulty propeller and the fault type according to the distribution characteristics of the state residual and the control coupling relationship.

[0081] In this embodiment, the combined state prediction model comprehensively considers the functional similarity and control coupling relationship among multiple propellers, realizes precise prediction of the overall state of the multi-propeller, avoids the problems of information isolation and error accumulation during the prediction of a single propeller, enables the state residual to more truly reflect abnormal situations, and thus greatly improves the accuracy of fault detection and location.

[0082] Optionally, the constructing unit 702 includes: A first determination module, configured to determine the propeller grouping according to the similarity relationship; A second determination module, configured to determine the combined state transition matrix according to the propeller grouping and the control coupling relationship; The first construction module is used to construct a joint state prediction model of a multi-thruster system based on the joint state transition matrix.

[0083] Optionally, the first construction module is specifically used for: Determine the current joint state vector, input gain matrix, and joint control input matrix according to the parameter information of each thruster; Construct a joint state prediction model of the multi-thruster system based on the current joint state vector, input gain matrix, joint control input matrix, and joint state transition matrix.

[0084] Optionally, the joint state prediction model is specifically designed as:

[0085]

[0086]

[0087]

[0088] Among them, represents the vector of the predicted next state, represents the current joint state vector, represents the joint control input matrix, represents the input gain matrix, represents the joint state transition matrix.

[0089] Optionally, the determination unit 707 includes: The second construction module is used to construct a residual matrix according to the state residual; The third construction module is used to construct a coupling coefficient matrix based on the control coupling relationship; The generation module is used to generate a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix; The input module is used to input the residual coupling influence matrix into a pre-trained thruster fault identification model to determine the faulty thruster and the fault type.

[0090] Optionally, the generation module is specifically used for: Calculate the normalized coupling weight matrix by using the amplitude of each row residual vector of the residual matrix and the corresponding elements of the coupling coefficient matrix; Apply a double non-linear transformation to each residual vector in the residual matrix to obtain a transformed vector; Weight and aggregate the transformed vectors through the normalized coupling weight matrix to obtain a residual coupling influence matrix.

[0091] Optionally, the judgment unit 706 is specifically used for: Determine whether each item of the status residual is less than the corresponding preset threshold; If not, it is determined that there is a fault in the multi-thruster system; if so, it is determined that there is no fault in the multi-thruster system.

[0092] In this embodiment, the functions of each unit and module correspond to the steps in the foregoing Figures 1 to 6 illustrated embodiment, and will not be elaborated here.

[0093] Please refer to Figure 8 , Figure 8 which is another embodiment of the underwater thruster fault detection device provided by this application. This embodiment includes: a processor 801, a memory 802, an input / output unit 803, and a bus 804; The processor 801 is connected to the memory 802, the input / output unit 803, and the bus 804; The memory 802 stores a program, and the processor 801 calls the program to execute Figures 1 to 6 the steps in the illustrated embodiment.

[0094] In this embodiment, the function of the processor 801 corresponds to the steps in the foregoing Figures 1 to 6 illustrated embodiment, and will not be elaborated here.

[0095] This application embodiment also provides a computer-readable storage medium. A program is stored on the computer-readable storage medium, and when the program is executed on a computer, the computer executes the method in any of the foregoing Figures 1 to 6 possible implementation manners.

[0096] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated here.

[0097] In the several embodiments provided by this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0098] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0099] In addition, each functional unit in various embodiments of the present application may be integrated into one processing unit, may exist separately as individual physical units, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0100] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, read-only memory), random access memories (RAM, random access memory), magnetic disks, or optical discs that can store program codes.

Claims

1. Method for detecting underwater thruster faults, characterized in that, Including: Establish functional similarity relationships and control coupling relationships among thrusters according to the thrust capabilities, thrust directions, and control distribution coefficients of the respective thrusters in the multi-thruster system; Construct a joint state prediction model of the multi-thruster system based on the functional similarity relationships and the control coupling relationships; Collect the current operating state data and control input data of each thruster, where the operating state data includes rotational speed, current, thrust feedback, and attitude response; Input the control input data into the joint state prediction model to calculate the predicted operating state data of each thruster; Collect the actual operating state data of each thruster, and calculate the state residual according to the actual operating state data and the predicted operating state data; where the actual operating state data is the data at the next moment corresponding to the current operating state data; Compare the state residual with a preset threshold to determine whether there is a fault in the multi-thruster system; If there is a fault, determine the faulty thruster and the fault type according to the distribution characteristics of the state residual and the control coupling relationship.

2. The underwater thruster fault detection method according to claim 1, wherein The constructing the joint state prediction model of the multi-thruster system based on the functional similarity relationships and the control coupling relationships includes: Determine thruster grouping according to the similarity relationships; Determine the joint state transition matrix according to the thruster grouping and the control coupling relationship; Construct the joint state prediction model of the multi-thruster system based on the joint state transition matrix.

3. The underwater thruster fault detection method according to claim 2, wherein The constructing the joint state prediction model of the multi-thruster system based on the joint state transition matrix includes: Determine the current joint state vector, input gain matrix, and joint control input matrix according to the parameter information of each thruster; Construct the joint state prediction model of the multi-thruster system according to the current joint state vector, the input gain matrix, the joint control input matrix, and the joint state transition matrix.

4. The underwater thruster fault detection method according to claim 3, characterized in that, The joint state prediction model is specifically designed as: Among them, The vector representing the predicted next state, Represents the current joint state vector, Represents the joint control input matrix, Represents the input gain matrix, Represents the joint state transition matrix.

5. The underwater thruster fault detection method according to claim 1, wherein The determining the faulty thruster and the fault type according to the distribution characteristics of the state residual and the control coupling relationship includes: Construct a residual matrix according to the state residual; Construct a coupling coefficient matrix based on the control coupling relationship; Generate a residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix; Input the residual coupling influence matrix into a pre-trained thruster fault identification model to determine the faulty thruster and the fault type.

6. The underwater thruster fault detection method according to claim 5, characterized in that, The generating the residual coupling influence matrix according to the residual matrix and the coupling coefficient matrix includes: Calculate the normalized coupling weight matrix by using the row residual vector amplitudes of the residual matrix and the corresponding elements of the coupling coefficient matrix; Apply a double nonlinear transformation to each residual vector in the residual matrix to obtain a transformed vector; Weight and aggregate the transformed vector through the normalized coupling weight matrix to obtain the residual coupling influence matrix.

7. The underwater thruster fault detection method according to any one of claims 1 to 6, characterized in that, The comparing the state residual with a preset threshold to determine whether there is a fault in the multi-thruster system includes: Judge whether each item of the state residual is less than the corresponding preset threshold; If not, determine that the multi-thruster system has a fault; if so, determine that the multi-thruster system has no fault.

8. Underwater thruster fault detection device, characterized in that, Comprising: A establishing unit, configured to establish a functional similarity relationship and a control coupling relationship between thrusters according to the thrust capabilities, thrust directions, and control distribution coefficients of the thrusters in the multi-thruster system; A constructing unit, configured to construct a joint state prediction model of the multi-thruster system based on the functional similarity relationship and the control coupling relationship; A first acquisition unit, configured to acquire the current operating state data and control input data of each thruster, where the operating state data includes rotational speed, current, thrust feedback, and attitude response; An input unit, configured to input the control input data into the joint state prediction model to calculate the predicted operating state data of each thruster; A second acquisition unit, configured to acquire the actual operating state data of each thruster, and calculate a state residual according to the actual operating state data and the predicted operating state data; wherein, the actual operating state data is the data at the next moment corresponding to the current operating state data; A judging unit, configured to compare the state residual with a preset threshold to judge whether the multi-thruster system has a fault; A determining unit, configured to, if there is a fault, determine the faulty thruster and the type of fault according to the distribution characteristics of the state residual and the control coupling relationship.

9. Underwater thruster fault detection device, characterized in that, Comprising: A processor, a memory, an input-output unit, and a bus; The processor is connected to the memory, the input-output unit, and the bus; The memory stores a program, and the processor calls the program to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A program is stored on the computer-readable storage medium, and when the program is executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images