Multi-mode fault diagnosis method and system for controllable-pitch propeller hydraulic system based on GAF
Through GAF, the deviation of pitch command and feedback is converted into two-dimensional images, combined with ResNet and GRU to extract features, and the feature fusion is used to solve the complexity of fault diagnosis of distance-adjusting paddle hydraulic system and improve the accuracy of diagnosis Sex and efficiency.
Patent Information
- Application Number
- CN202411465966.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The distance adjustment paddle hydraulic system is difficult to diagnose faults in complex marine environments, and the existing technology is difficult to effectively improve diagnostic accuracy and efficiency.
Using a multimodal fault diagnosis method based on GAF, the deviations of pitch instructions and pitch feedback are converted into two-dimensional images through the Gram matrix, the image features are extracted using the residual neural network, and the time features are extracted in combination with the gating recursive unit, and the multi-header Attention mechanisms perform feature fusion and fault classification.
It improves the accuracy and efficiency of fault diagnosis, can more effectively identify the fault types of distance adjustment paddle hydraulic system, reduces the limitations of singular feature types, and improves classification accuracy.
Smart Images

Figure CN119004215A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of ship fault diagnosis, and in particular to a GAF-based multi-modal fault diagnosis method and system for a controllable pitch propeller hydraulic system. Background Art
[0002] The controllable pitch propeller (CPP) system is a main device in the ship propulsion system, which is able to change the thrust size and direction of the propeller by adjusting the pitch of the blades, while maintaining a constant engine direction and speed. This adjustment facilitates the change of the ship's cruising speed, or realizes the conversion between forward thrust and reverse thrust. Compared with the standard fixed pitch propeller, the CPP can dynamically adjust the pitch according to different operating environments to achieve the best alignment between cruising speed, main engine power, speed and propeller efficiency. This improves the maneuverability of the ship and the efficiency of the ship's engine-propeller integration. Therefore, due to its enhanced adaptability, maneuverability and economic efficiency, CPP is increasingly used in ships operating under various conditions.
[0003] The CPP system is a complex integrated system that includes multiple subcomponents, including propellers, hubs, blades, distributors, in-shaft oil pipes, hydraulic systems, and electronic control systems. As the frequency and intensity of CPP use increase significantly, the probability and risk of component failure are also rising. In addition, due to the long time ships spend at sea and the harsh operating environment of CPP, it is difficult to diagnose the fault when it occurs, which complicates the implementation of appropriate maintenance measures. Summary of the invention
[0004] The present invention aims to solve at least one of the technical problems existing in the related art. To this end, the present invention provides a multimodal fault diagnosis method and system for a controllable pitch propeller hydraulic system based on GAF (Gram matrix). The present invention uses the deviation value of pitch feedback and pitch command as the key variable, converts it into a two-dimensional image using GAF, and uses a residual neural network (ResNet) to extract image features, thereby obtaining more effective feature information. At the same time, considering the relative relationship between time points, a gated recurrent unit (GRU) is used to extract additional time features, and feature fusion is performed through a multi-head attention mechanism, thereby improving the accuracy of fault diagnosis classification.
[0005] The present invention provides a GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system, comprising: S1: obtaining the pitch command and pitch feedback of the controllable pitch propeller hydraulic system in real time through the first sensor, and calculating the deviation between the pitch command and the pitch feedback; S2: The deviation between pitch command and pitch feedback is converted into a two-dimensional image through the Gram matrix; S3: Extracting image features of 2D images through residual neural network; S4: obtaining an adjustment instruction and an adjustment feedback of the controllable pitch propeller hydraulic system through a second sensor, and calculating a difference between the adjustment instruction and the adjustment feedback; S5: extracting the temporal characteristics of the deviation between the adjustment instruction and the adjustment feedback through the gated recurrent unit; S6: The image features and time features are fused through the multi-head attention module to obtain the fused features; the faults are classified through the fused features to obtain the fault type of the controllable pitch propeller hydraulic system.
[0006] According to a GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system provided by the present invention, the method also includes, in the step S1, forming a deviation time series data with time as the horizontal axis and deviation as the vertical axis, from the deviation between the pitch command and the pitch feedback obtained in real time by the first sensor.
[0007] According to a GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system provided by the present invention, the method further includes the following steps: Normalize the deviation time series data, and the calculation expression is: in, is the deviation time series data, is the first data points, is the first in the normalized time series data points, is the largest value in the time series. is the smallest value in the time series; Map the normalized deviation time series data into polar coordinates, convert the pitch angle feedback value into the cosine of the polar angle, and the calculation expression is: in, For the Data points mapping angle; The obtained Gram matrix calculation expression is: in, is the Gram matrix, Mapping angle for the first data point, Mapping angle for the second data point, for The data point mapping angle, is the length of the time series; The generated Gram matrix is directly converted into a two-dimensional grayscale image or mapped into a two-dimensional pseudo-color image to obtain a two-dimensional image.
[0008] According to a GAF-based multimodal fault diagnosis method for a controllable pitch propeller hydraulic system provided by the present invention, the method also includes calculating the attention weight of each head and aggregating the weights in the step S6, connecting the outputs of all multiple heads together, and then fusing them through a linear layer to obtain a fusion feature.
[0009] According to a GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system provided by the present invention, the method further includes classifying the fault according to the fusion features through a classifier to obtain the fault type, and the classifier includes a naive Bayes classifier and a support vector machine.
[0010] The present invention also provides a multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system based on GAF, which uses the multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system based on GAF and communicates based on a DSP Ruihua operating system, including a data input module and an internal data analysis and processing module; The data input module is used to obtain the pitch command and pitch feedback of the adjustable pitch propeller hydraulic system in real time through the first sensor, and obtain the adjustment command and adjustment feedback of the adjustable pitch propeller hydraulic system through the second sensor; The internal data analysis and processing module is used to calculate the deviation between the pitch command and the pitch feedback; convert the deviation between the pitch command and the pitch feedback into a two-dimensional image through a Gram matrix; extract the image features of the two-dimensional image through a residual neural network; calculate the difference between the adjustment command and the adjustment feedback; extract the time features of the deviation between the adjustment command and the adjustment feedback through a gated recursive unit; perform feature fusion of the image features and the time features through a multi-head attention module to obtain fused features; classify faults through the fused features to obtain the fault type of the adjustable pitch propeller hydraulic system.
[0011] A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system provided by the present invention also includes the DSP Ruihua operating system integrating a real-time clock.
[0012] A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system provided according to the present invention also includes a data output module, and the data output module uploads the fault type in real time based on a real-time clock.
[0013] A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system provided according to the present invention also includes a real-time alarm module, which is used to receive the fault type and issue a warning message.
[0014] A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system provided according to the present invention also includes a real-time control module, and the real-time control module is used to adjust the state according to the fault type.
[0015] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: The present invention provides a GAF-based multimodal fault diagnosis method and system for a controllable pitch propeller hydraulic system. The method uses the deviation value of pitch feedback and pitch command as the key variable, converts it into a two-dimensional image using GAF, and uses ResNet to extract image features, thereby obtaining more effective feature information. At the same time, the relative relationship between time points is considered, and GRU is used to extract additional time features. Feature fusion is performed through a multi-head attention mechanism, thereby improving the accuracy of fault diagnosis classification.
[0016] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0018] Figure 1 It is a flow chart of a multi-modal fault diagnosis method of a controllable pitch propeller hydraulic system based on GAF provided by the present invention.
[0019] Figure 2 It is a schematic diagram of a model framework of a multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system based on GAF provided by the present invention.
[0020] Figure 3 It is a comparison chart of classification accuracy of a multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system based on GAF provided by the present invention.
[0021] Figure 4 The present invention provides a GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system.
[0022] Reference numerals: 101. DSP Ruihua operating system; 102. Data input module; 103. Internal data analysis and processing module; 104. Data output module; 105. Real-time alarm module; 106. Real-time control module. DETAILED DESCRIPTION
[0023] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme in the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.
[0024] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0025] Combine the following Figures 1 to 4 Description The present invention provides a GAF-based multi-modal fault diagnosis method and system for a controllable pitch propeller hydraulic system.
[0026] like Figure 1 and Figure 2 As shown, a multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system based on GAF includes: S1: obtaining the pitch command and pitch feedback of the controllable pitch propeller hydraulic system in real time through the first sensor, and calculating the deviation between the pitch command and the pitch feedback; The hydraulic system of CPP is a key component in the CPP assembly and is the power source for adjusting the blade pitch. The hydraulic system of CPP is configured as an integrated layout, mainly including the main oil tank located at the bottom of the ship, pump assembly, valve group and related hydraulic accessories. Its main function is to execute the pitch change command issued by the console in the electronic control system by delivering hydraulic oil to the piston chamber of the propeller hub hydraulic cylinder. The movement of the piston rotates the blade to the specified position through the crank slider mechanism, thereby changing the pitch of the blade. Other functions include delivering hydraulic oil to the propeller hub cavity to maintain a certain oil pressure, prevent seawater infiltration, provide lubrication for the distributor and other components, and ensure blade pitch change in an emergency.
[0027] The CPP hydraulic system is also equipped with pitch angle sensors, pressure sensors, flow sensors and temperature sensors to obtain relevant characteristics. Pitch command and pitch feedback are characteristic quantities inherent in each CPP system. Pitch command and pitch feedback are obtained through pitch feedback and pitch command. Pitch feedback and pitch command can be obtained through sensors, and the deviation between pitch feedback and pitch command can be calculated.
[0028] In some embodiments of the present invention, the pitch command and pitch feedback of the adjustable pitch propeller hydraulic system are obtained in real time through a first sensor, and the first sensor includes a pitch angle sensor, a pressure sensor, a flow sensor, a temperature sensor, etc. The deviation between the pitch command and the pitch feedback obtained in real time by the first sensor is organized into time series data with time as the horizontal axis and the deviation as the vertical axis.
[0029] S2: The deviation between pitch command and pitch feedback is converted into a two-dimensional image through the Gram matrix; The Gram Array Matrix (GAF) is a technique used in the field of time series analysis to encode one-dimensional data into a visually interpretable two-dimensional image; The present invention converts the deviation between the pitch instruction and the pitch feedback into a two-dimensional image through a Gram matrix.
[0030] Normalize the deviation time series data, and the calculation expression is: in, is the deviation time series data, is the first data points, is the first in the normalized time series data points, is the largest value in the time series. is the smallest value in the time series; Map the normalized deviation time series data into polar coordinates, convert the pitch angle feedback value into the cosine of the polar angle, and convert the time into the radius. The calculation expression is: in, is the length of the time series, is the polar coordinate The radius of the data points, For the Data points mapping angle; The obtained Gram matrix calculation expression is in, is the Gram matrix, Mapping angle for the first data point, Mapping angle for the second data point, for The mapping angle of the data point; Represents the cosine of the angle between the vector corresponding to the first data point and the vector corresponding to the first data point, Represents the cosine of the angle between the vector corresponding to the first data point and the vector corresponding to the second data point, Indicates The vector corresponding to the data point is The cosine of the angle between the vectors corresponding to the data points.
[0031] The generated Gram matrix is directly converted into a two-dimensional grayscale image or mapped into a two-dimensional pseudo-color image to obtain a two-dimensional image, thereby realizing the conversion of a one-dimensional time series signal to a two-dimensional image.
[0032] S3: Extracting image features of 2D images through residual neural network; Residual neural network (ResNet) extracts image features through skip connections and residual learning; residual neural network adopts skip connections that bypass one or more layers, allowing the output of a layer to skip one or more layers and directly connect to the input of subsequent layers. The skip connection enables the network to learn the residual function, which represents the difference between the input and output of a given layer. By utilizing skip connections and residual learning, ResNet has achieved state-of-the-art performance on various image classification tasks.
[0033] In some embodiments of the present invention, ResNet is used to extract image features of a two-dimensional image.
[0034] S4: obtaining an adjustment instruction and an adjustment feedback of the controllable pitch propeller hydraulic system through a second sensor, and calculating a difference between the adjustment instruction and the adjustment feedback; In some embodiments of the present invention, the adjustment instruction and adjustment feedback of the controllable pitch propeller hydraulic system are obtained through a second sensor, and the second sensor includes a pressure sensor, a flow sensor, a temperature sensor, and the like.
[0035] S5: extracting the temporal characteristics of the deviation between the adjustment instruction and the adjustment feedback through the gated recurrent unit; The Gated Recurrent Unit (GRU) is used to capture long-range dependencies in time series data. The GRU includes a reset gate and an update gate that control the flow of information within the network. The reset gate determines how much past information to forget, while the update gate regulates the amount of new information to retain. Temporal features are extracted through the output of the hidden state.
[0036] In some embodiments of the present invention, the temporal features of the deviation between the adjustment instruction and the adjustment feedback are extracted by GRU.
[0037] S6: The image features and time features are fused through the multi-head attention module to obtain the fused features; the faults are classified through the fused features to obtain the fault type of the controllable pitch propeller hydraulic system.
[0038] Calculate the attention weight of each head and aggregate the weights, connect the outputs of all multiple heads together, and then fuse them through the linear layer to obtain the fusion feature. The calculation expression is: in, is the key matrix, is the value matrix, For query, For the dimension, is the activation function, It is a multi-head attention feature; is the output of attention, is the attention weight; Through the classifier, the fault is classified according to the fusion features, the fault type is obtained, and adjustments are made according to the fault type to ensure the normal navigation of the ship. The classifier includes the naive Bayes classifier and the support vector machine.
[0039] The multi-head attention module can simultaneously pay attention to the input sequence from different angles, effectively evaluate the extracted features, and improve the classification accuracy of fault diagnosis.
[0040] like Figure 4 As shown, a GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system communicates based on a DSP Ruihua operating system 101, and the DSP is equipped with a Ruihua system to achieve real-time control of a ship's CPP hydraulic system, including a data input module 102 and an internal data analysis and processing module 103; The data input module 102 is used to obtain the pitch command and pitch feedback of the controllable pitch propeller hydraulic system in real time through the first sensor, and obtain the adjustment command and adjustment feedback of the controllable pitch propeller hydraulic system through the second sensor; The internal data analysis and processing module 103 has a storage unit greater than or equal to 512G, which is used to calculate the deviation between the pitch command and the pitch feedback; convert the deviation between the pitch command and the pitch feedback into a two-dimensional image through a Gram matrix; extract the image features of the two-dimensional image through a residual neural network; calculate the difference between the adjustment command and the adjustment feedback; extract the time feature of the deviation between the adjustment command and the adjustment feedback through a gated recursive unit; fuse the image features and the time features through a multi-head attention module to obtain a fused feature; classify the fault through the fused feature to obtain the fault type of the adjustable pitch propeller hydraulic system.
[0041] DSP Ruihua operating system is an operating system product with independent intellectual property rights developed by the 32nd Institute of China Electronics Technology Group Corporation since the 9th Five-Year Plan period. Compared with the mainstream foreign commercial multi-core embedded operating systems VxWorks, SYS / BIOS and domestic embedded multi-core operating system DeltaOS6.0, Ruihua embedded real-time operating system ReWorks has complete support for domestic CPU / DSP processor architecture, wide application, and strong adaptation and optimization capabilities with domestic multi-core processors.
[0042] Compared with other DSP operating systems (such as Linux, SYSBIOS, etc.), Ruihua DSP operating system has certain advantages in system tailoring, real-time performance, ease of use, and standardization. Ruihua DSP operating system has strong real-time and high reliability characteristics. Based on micro-kernel and component technology, it can configure, tailor, expand and customize the operating system kernel according to the needs of DSP applications, and has strong system scalability; strong real-time scheduling based on priority preemption ensures the strong real-time response capability of the system, and hardware-based UDP acceleration improves network performance; the system installation and deployment can use a unified image, and each core executes different branch codes; it supports dynamic unloading of modules, which is easy for system reconstruction and application upgrades; the easy-to-operate system console realizes fast switching between multiple cores; it supports OpenMP3.0 parallel computing specifications, and provides application programming interfaces that comply with POSIX and GJB7714-2012 standards. Ruihua DSP operating system is equipped with a dedicated integrated development environment, which solves the problem that multi-core DSP is difficult to use, and has powerful network, file system, middleware and other processing capabilities, which changes the original DSP processor programming mode and improves the user's software development efficiency. The present invention uses Ruihua DSP operating system for ship systems for the first time.
[0043] DSP Ruihua operating system integrates a real-time clock.
[0044] In a specific embodiment of the present invention, the ship data can be stored and analyzed in real time through a large-capacity storage unit.
[0045] The system further includes a data output module 104 , which uploads the fault type in real time based on a real-time clock (RTC).
[0046] In a specific embodiment of the present invention, the system has high punctuality based on RTC, and can realize the absolute and accurate benchmarking relationship between the behavior of ship navigation and time, and can also benchmark information and time with monitoring equipment such as satellites through the network and other means. The fault type is uploaded in real time, so that the fault and fault type of the CPP hydraulic system can be discovered earlier.
[0047] The system also includes a real-time alarm module 105, which is used to receive the fault type and issue a warning message.
[0048] In a specific embodiment of the present invention, when a CPP hydraulic system fails, the real-time alarm module 105 sends out a warning message, and engineers can locate the source and position of the failure more timely and accurately, thereby improving the safety performance of the ship's navigation.
[0049] The system further includes a real-time control module 106, which is used to adjust the state according to the fault type.
[0050] In a specific embodiment of the present invention, the real-time control module 106 adjusts the pitch command in real time according to the fault type, thereby stabilizing the navigation state.
[0051] In a specific embodiment of the present invention, the deviation value between the pitch instruction and pitch feedback of the adjustable pitch propeller is obtained by the first sensor when the flow rate of the main pump is reduced, the pressure of the relief valve is reduced by 20%, the pressure of the relief valve is reduced by 50%, and the electromagnetic reversing valve is closed, and these values are converted into a two-dimensional image by GAF. The image features of the two-dimensional image are extracted by ResNet using the obtained two-dimensional image, and the adjustment instructions and adjustment feedback of the adjustable pitch propeller of the main pump when the flow rate of the main pump is reduced, when the pressure of the relief valve is reduced by 20%, the pressure of the relief valve is reduced by 50%, and the electromagnetic reversing valve is closed are obtained by the second sensor, and the time feature of the difference between the adjustment instruction and the adjustment feedback is extracted by GRU. By fusing the image feature with the time feature, the fault is classified according to the fusion feature through a naive Bayes classifier or a support vector machine, and the main pump flow rate reduction fault, the overflow valve pressure reduction fault of 20%, the overflow valve pressure reduction fault of 50%, and the electromagnetic reversing valve closing fault are obtained, and the classification accuracy can reach 100%. The results are compared with those of directly classifying faults on images using GAF and ResNet and directly using GRU for fault classification. The fusion of the two features establishes a more robust correlation between instructions and feedback, thereby improving the efficiency of fault diagnosis over the other two methods, demonstrating the effectiveness of integrating multiple information modes. By merging image features with temporal features, this method alleviates the inherent limitations of singular feature types, thereby improving classification accuracy. The GAF-based multimodal fault diagnosis system for controllable pitch propeller hydraulic systems can produce better classification results, especially when image features provide more informative hints. The comparison results are shown in Figure 2. Figure 3 shown.
[0052] The present invention is mainly applied to the CPP hydraulic system of a ship. The data processing is based on the signal processor (DSP) main control, equipped with the Ruihua system for control, and includes a real-time clock (RTC), so that the system is more accurate in time, and the data analysis and logic control of the entire ship system are more timely. The ship's CPP hydraulic system is monitored in real time to overcome the problem of the lack of timeliness in the data processing of the ship's CPP hydraulic system and some status monitoring in the prior art.
[0053] The present invention takes the deviation value of pitch feedback and pitch command as the key variable, converts it into a two-dimensional image using GAF, and uses ResNet to extract image features to obtain more effective feature information. At the same time, considering the relative relationship between time points, GRU is used to extract additional time features, and feature fusion is performed through a multi-head attention mechanism to improve the accuracy of fault diagnosis classification.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system based on GAF, characterized in that: include: S1: obtaining the pitch command and pitch feedback of the controllable pitch propeller hydraulic system in real time through the first sensor, and calculating the deviation between the pitch command and the pitch feedback; S2: The deviation between pitch command and pitch feedback is converted into a two-dimensional image through the Gram matrix; S3: Extracting image features of 2D images through residual neural network; S4: obtaining an adjustment instruction and an adjustment feedback of the controllable pitch propeller hydraulic system through a second sensor, and calculating a difference between the adjustment instruction and the adjustment feedback; S5: extracting the temporal characteristics of the deviation between the adjustment instruction and the adjustment feedback through the gated recurrent unit; S6: The image features and time features are fused through the multi-head attention module to obtain fused features; The faults are classified by fusion features and the fault types of the controllable pitch propeller hydraulic system are obtained.
2. A GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system according to claim 1, characterized in that: In the step S1, the deviation between the pitch command and the pitch feedback obtained in real time by the first sensor is formed into deviation time series data with time as the horizontal axis and the deviation as the vertical axis.
3. A GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system according to claim 2, characterized in that: In the step S2, Normalize the deviation time series data, and the calculation expression is: in, is the deviation time series data, is the first data points, is the first in the normalized time series data points, is the largest value in the time series. is the smallest value in the time series; Map the normalized deviation time series data into polar coordinates, convert the pitch angle feedback value into the cosine of the polar angle, and the calculation expression is: in, For the Data points mapping angle; The obtained Gram matrix calculation expression is: in, is the Gram matrix, Mapping angle for the first data point, Mapping angle for the second data point, for The data point mapping angle, is the length of the time series; The generated Gram matrix is directly converted into a two-dimensional grayscale image or mapped into a two-dimensional pseudo-color image to obtain a two-dimensional image.
4. The GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system according to claim 1, characterized in that: In the step S6, the attention weight of each head is calculated and weighted, the outputs of all multiple heads are connected together, and then fused through a linear layer to obtain a fused feature.
5. A GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system according to claim 4, characterized in that: The fault is classified according to the fusion features through a classifier to obtain the fault type. The classifier includes a naive Bayes classifier and a support vector machine.
6. A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system, using a GAF-based multi-modal fault diagnosis method for a controllable pitch propeller hydraulic system as claimed in any one of claims 1 to 5, characterized in that: Communicate based on DSP Ruihua operating system, including data input module and internal data analysis and processing module; The data input module is used to obtain the pitch command and pitch feedback of the adjustable pitch propeller hydraulic system in real time through the first sensor, and obtain the adjustment command and adjustment feedback of the adjustable pitch propeller hydraulic system through the second sensor; The internal data analysis and processing module is used to calculate the deviation between the pitch command and the pitch feedback; convert the deviation between the pitch command and the pitch feedback into a two-dimensional image through a Gram matrix; extract the image features of the two-dimensional image through a residual neural network; calculate the difference between the adjustment command and the adjustment feedback; extract the time feature of the deviation between the adjustment command and the adjustment feedback through a gated recursive unit; and perform feature fusion of the image feature and the time feature through a multi-head attention module to obtain a fusion feature; The faults are classified by fusion features and the fault types of the controllable pitch propeller hydraulic system are obtained.
7. A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system according to claim 6, characterized in that: The DSP Ruihua operating system integrates a real-time clock.
8. A GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system according to claim 7, characterized in that: It also includes a data output module, which uploads the fault type in real time based on a real-time clock.
9. The GAF-based multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system according to claim 7, characterized in that: It also includes a real-time alarm module, which is used to receive the fault type and issue a warning message.
10. The multi-modal fault diagnosis system for a controllable pitch propeller hydraulic system based on GAF according to claim 7, characterized in that: It also includes a real-time control module, which is used to adjust the state according to the fault type.
Citation Information
Patent Citations
Propulsion system and navigation control method for ship
CN101342937A
Parallel computing method and system based on AMP framework DSP operating system, and medium
CN110297661A
Transformer fault diagnosis method and system based on multi-sensor information fusion
CN115641283A
Cited By
Real-time control and monitoring system for controllable-pitch propeller of ship
CN121929283A
A real-time control and monitoring system for a controllable pitch propeller of a ship
CN121929283B
Satellite orbit maneuver detection method based on multichannel residual imaging
CN122049482A
