Marine main engine cylinder diameter defect real-time positioning and loss analysis method

Abstract

The invention relates to the technical field of ship power machinery monitoring, in particular to a ship main engine cylinder diameter defect real-time positioning and loss analysis method, which is characterized by comprising the following steps: data acquisition: acquiring data through a multi-modal sensor array deployed on a preset monitoring point on the inner wall of a cylinder diameter; vibration signals, temperature field distribution data and ultrasonic echo signals of the cylinder diameter surface are collected in real time, and the sampling frequency is not lower than 1 kHz; performing data preprocessing: performing noise reduction filtering, abnormal value elimination and time-space registration on the collected original signals to generate a standardized data set; the method has real-time performance: the data acquisition and processing delay is less than 1 second, and real-time tracking of defect dynamic change is realized; the positioning error is smaller than or equal to 0.5 mm, the loss parameter quantization error is smaller than or equal to 3%, and the maintenance decision requirement is met; comprehensiveness: main defect types such as cracks, abrasion and corrosion are covered, and multi-dimensional loss analysis is provided; and intelligentization: through adaptive calibration and trend prediction, the manual intervention demand is reduced, and the early warning timeliness is improved.

Description

Technical Field

[0001] This application relates to the field of marine power machinery monitoring technology, specifically to a method for real-time location and quantitative analysis of wear of defects in marine main engine cylinder diameter based on multi-sensor fusion. Background Technology

[0002] As the core power plant of a ship, the operating condition of the cylinder bore (the inner wall of the cylinder liner) directly affects the reliability and safety of the main engine. Under long-term high-frequency, high-temperature, and high-pressure operating environments, the cylinder bore is prone to defects such as cracks, wear, and corrosion. If these defects are not detected and addressed in time, they may lead to cylinder liner breakage, fuel leakage, or even main engine failure, resulting in huge economic losses and safety hazards.

[0003] Existing cylinder bore defect detection technologies are mainly divided into two categories: one is offline detection methods, such as manual inspection using endoscopes and ultrasonic flaw detectors after the machine is shut down. This method requires interrupting the main unit's operation, has a long inspection cycle (usually once every 3-6 months), and cannot reflect the dynamic changes of defects in real time; the other is online monitoring methods, such as condition monitoring using a single vibration sensor or temperature sensor, but it has the following shortcomings:

[0004] Low defect location accuracy: It can only determine the approximate area of ​​the defect (error is often >5mm), which cannot meet the maintenance needs;

[0005] The loss analysis is one-sided: it focuses too much on a single defect type (such as wear) and lacks a comprehensive assessment of key parameters such as crack propagation and corrosion rate;

[0006] Poor real-time performance: Data processing delay > 10 seconds, making it difficult to achieve dynamic early warning.

[0007] Therefore, there is an urgent need for a method that can achieve real-time and accurate defect location, multi-dimensional loss analysis, and dynamic early warning to overcome the limitations of existing technologies. Summary of the Invention

[0008] This application provides a method for real-time location and loss analysis of defects in the bore of a marine main engine. Through multimodal sensing, intelligent algorithm fusion and dynamic modeling, it realizes real-time monitoring, accurate location, quantitative analysis and early warning of defects in the bore, and solves the problems of offline detection lag, low accuracy of online monitoring and single analysis dimension in the prior art.

[0009] To achieve the above objectives, this application provides the following technical solution: a method for real-time location and loss analysis of defects in the bore of a marine main engine, comprising the following steps:

[0010] Data acquisition: A multimodal sensor array deployed at preset monitoring points on the inner wall of the cylinder bore is used to collect vibration signals, temperature field distribution data and ultrasonic echo signals on the cylinder bore surface in real time, with a sampling frequency of not less than 1kHz;

[0011] Data preprocessing: The acquired raw signals are subjected to noise reduction filtering, outlier removal, and spatiotemporal registration to generate a standardized dataset;

[0012] Defect identification and localization: The preprocessed data is input into the trained defect identification model. The defect type is identified through feature extraction, and the precise location of the defect in the cylinder bore circumference and axial direction is output by combining the spatial coordinate mapping of the sensor array.

[0013] Loss Quantification Analysis: Based on the positioning results, loss parameters are calculated through the following sub-steps:

[0014] For crack-like defects, fracture mechanics models are used to calculate crack depth and propagation rate;

[0015] For wear-related defects, the radial wear of the cylinder diameter is measured with the assistance of a laser displacement sensor, and the wear rate is calculated in combination with the material hardness parameters;

[0016] For corrosion defects, the corrosion depth and area are calculated using data from electrochemical sensors;

[0017] Results output and early warning: Defect location information, loss quantification data and trend prediction results are displayed in real time through a visual interface, and an audible and visual early warning is triggered when the loss parameter exceeds the preset threshold.

[0018] Preferably, the multimodal sensor array in step 1 includes:

[0019] 8-12 piezoelectric vibration sensors are arranged at equal intervals along the circumference of the cylinder liner, with a sampling range of 50Hz-10kHz.

[0020] Four infrared temperature sensors are deployed at the midpoint and both ends of the cylinder bore axial direction, with a temperature measurement range of -20℃ to 300℃ and an accuracy of ±0.5℃.

[0021] Two sets of ultrasonic sensors are arranged axially at the top and bottom of the cylinder bore, with a detection frequency of 5MHz-10MHz.

[0022] Preferably, the optimization of the improved YOLOv5 algorithm in step 3 includes:

[0023] Introducing an attention mechanism to enhance the weighting of defect features;

[0024] The FocalLoss function is used to address the imbalance problem of small defect samples.

[0025] The output layer adds a defect confidence score (0-100%), and a defect is considered valid when the score is ≥85%.

[0026] Preferably, it also includes a data storage and traceability module, which stores the original signals and analysis results by timestamp, and supports historical data query and trend review for at least 180 days.

[0027] Preferably, the warning threshold in step 5 includes:

[0028] Crack depth ≥ 0.3 mm or propagation rate ≥ 0.01 mm / day;

[0029] Radial wear ≥ 0.5% of the nominal cylinder diameter;

[0030] The corrosion area accounts for more than 3% of the total inner wall area of ​​the cylinder diameter.

[0031] Preferably, the piezoelectric vibration sensor is connected to the outer wall of the cylinder liner via a magnetic adsorption mounting base, and the mounting base has a built-in temperature compensation module.

[0032] Preferably, the spatiotemporal registration in step 2 achieves time alignment of multi-sensor data through GPS time synchronization and spatial alignment through coordinate transformation of the cylinder bore 3D model.

[0033] Preferably, the method further includes an adaptive calibration step: every 24 hours of operation, the standard defect sample library is automatically called to fine-tune the identification model online, and the calibration positioning error is ≤0.1mm.

[0034] Preferably, the visualization interface in step 5 supports dynamic display of the cylinder diameter 3D model, and detailed parameters can be displayed in a pop-up window by clicking on the defect location.

[0035] Compared with the prior art, the beneficial effects of this application are:

[0036] Real-time performance: Data acquisition and processing latency < 1 second, enabling real-time tracking of dynamic changes in defects;

[0037] High precision: Positioning error ≤0.5mm, loss parameter quantification error ≤3%, meeting maintenance decision-making requirements;

[0038] Comprehensive: Covers major defect types such as cracks, wear, and corrosion, providing multi-dimensional loss analysis;

[0039] Intelligentization: By using adaptive calibration and trend prediction, the need for manual intervention is reduced and the timeliness of early warning is improved. Attached Figure Description

[0040] Figure 1 This is a flowchart of the application process. Detailed Implementation

[0041] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0042] In the description of this application, if directional descriptions are involved, such as "up," "down," "front," "back," "left," "right," etc., indicating directional or positional relationships, they are based on the appendix. Figure 1 The orientations or positional relationships shown are for the convenience of describing this application and simplifying the description only, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. When a feature is referred to as "set", "fixed", or "connected" to another feature, it can be directly set, fixed, or connected to the other feature, or it can be indirectly set, fixed, or connected to the other feature.

[0043] Please see Figure 1 This application provides the following technical solution: a method for real-time location and loss analysis of defects in the cylinder diameter of a marine main engine, comprising the following steps:

[0044] Data acquisition: A multimodal sensor array deployed at preset monitoring points on the inner wall of the cylinder bore is used to collect vibration signals, temperature field distribution data and ultrasonic echo signals on the cylinder bore surface in real time, with a sampling frequency of not less than 1kHz;

[0045] Data preprocessing: The acquired raw signals are subjected to noise reduction filtering, outlier removal, and spatiotemporal registration to generate a standardized dataset;

[0046] Defect identification and localization: The preprocessed data is input into the trained defect identification model. The defect type is identified through feature extraction, and the precise location of the defect in the cylinder bore circumference and axial direction is output by combining the spatial coordinate mapping of the sensor array.

[0047] Loss Quantification Analysis: Based on the positioning results, loss parameters are calculated through the following sub-steps:

[0048] For crack-like defects, fracture mechanics models are used to calculate crack depth and propagation rate;

[0049] For wear-related defects, the radial wear of the cylinder diameter is measured with the assistance of a laser displacement sensor, and the wear rate is calculated in combination with the material hardness parameters;

[0050] For corrosion defects, the corrosion depth and area are calculated using data from electrochemical sensors;

[0051] Results output and early warning: Defect location information, loss quantification data and trend prediction results are displayed in real time through a visual interface, and an audible and visual early warning is triggered when the loss parameter exceeds the preset threshold.

[0052] Furthermore, the multimodal sensor array mentioned in step 1 includes:

[0053] 8-12 piezoelectric vibration sensors are arranged at equal intervals along the circumference of the cylinder liner, with a sampling range of 50Hz-10kHz.

[0054] Four infrared temperature sensors are deployed at the midpoint and both ends of the cylinder bore axial direction, with a temperature measurement range of -20℃ to 300℃ and an accuracy of ±0.5℃.

[0055] Two sets of ultrasonic sensors are arranged axially at the top and bottom of the cylinder bore, with a detection frequency of 5MHz-10MHz.

[0056] Preferably, the optimization of the improved YOLOv5 algorithm in step 3 includes:

[0057] Introducing an attention mechanism to enhance the weighting of defect features;

[0058] The FocalLoss function is used to address the imbalance problem of small defect samples.

[0059] The output layer adds a defect confidence score (0-100%), and a defect is considered valid when the score is ≥85%.

[0060] Furthermore, it also includes a data storage and traceability module, which stores the original signals and analysis results by timestamp, and supports historical data query and trend review for at least 180 days.

[0061] Furthermore, the warning threshold mentioned in step 5 includes:

[0062] Crack depth ≥ 0.3 mm or propagation rate ≥ 0.01 mm / day;

[0063] Radial wear ≥ 0.5% of the nominal cylinder diameter;

[0064] The corrosion area accounts for more than 3% of the total inner wall area of ​​the cylinder diameter.

[0065] Furthermore, the piezoelectric vibration sensor is connected to the outer wall of the cylinder liner via a magnetic adsorption mounting base, which has a built-in temperature compensation module.

[0066] Furthermore, the spatiotemporal registration described in step 2 achieves time alignment of multi-sensor data through GPS time synchronization and spatial alignment through coordinate transformation of the cylinder bore 3D model.

[0067] Furthermore, the method also includes an adaptive calibration step: every 24 hours of operation, the standard defect sample library is automatically called to fine-tune the identification model online, and the calibration positioning error is ≤0.1mm.

[0068] The visualization interface in step 5 supports dynamic display of the cylinder diameter 3D model, and detailed parameters can be displayed in a pop-up window by clicking on the defect location.

[0069] During use: Data acquisition phase:

[0070] The vibration sensor sampling frequency is set to 2kHz to collect cylinder bore vibration acceleration signals (range: ±50g).

[0071] The temperature sensor collects data every 100ms and records the temperature distribution on the inner wall of the cylinder bore.

[0072] The ultrasonic sensor emits a 5MHz pulse wave at a frequency of 10Hz and receives the reflected echo signal.

[0073] Data preprocessing stage:

[0074] Wavelet thresholding denoising uses the db4 wavelet basis, with a decomposition level of 5, and the thresholding adopts an adaptive Birgé-Massart strategy.

[0075] Spatiotemporal registration is achieved through time synchronization via a GPS timing module (1ms accuracy) and spatial alignment via coordinate transformation of the cylinder bore 3D model (built on SolidWorks) (mapping the physical position of the sensor to the model coordinate system).

[0076] Defect identification and localization stage:

[0077] The training sample library for the improved YOLOv5 algorithm contains 10,000+ cylinder diameter defect images (including three categories: cracks, wear, and corrosion, with 3,000+ samples in each category).

[0078] The model is deployed on a GPU (NVIDIA RTX 3060), and the inference speed is ≥20 frames / second. The defect confidence is ≥85% when the model is considered valid.

[0079] Loss analysis phase:

[0080] Crack depth calculation: Based on the ultrasonic echo time difference Δt, using the formula...

[0081] H = 0.5 × c × Δt

[0082] (c is the sound velocity of the cylinder liner material, taken as 5900m / s);

[0083] Wear rate calculation: The formula described in claim 4 is used, where the density ρ of the cylinder diameter material is taken as 7.85 g / cm³ (gray cast iron).

[0084] Corrosion area: The percentage of pixels in the corroded area is calculated by processing the ultrasonic echo imaging data using an image segmentation algorithm (U-Net model).

[0085] Results output and alerts:

[0086] The visualization interface refreshes the data every 2 seconds, and the 3D model uses different colors to mark the defect types (red: crack, yellow: wear, blue: corrosion).

[0087] When the defect parameter exceeds the threshold described in claim 6, a three-level warning is triggered (yellow: attention, orange: maintenance, red: shutdown).

[0088] Adaptive calibration

[0089] The calibration starts automatically at 3:00 AM every day (during low load periods for the host): it calls up 100 sets of built-in standard defect samples (including artificial defect data with known size and location), and fine-tunes the model parameters through error backpropagation to ensure stable positioning accuracy.

[0090] This method has been tested and verified on a marine diesel engine with a cylinder diameter of 170 mm. The results show that the crack location error is ≤0.3 mm, the wear measurement error is ≤2%, the corrosion area calculation error is ≤5%, and the early warning response time is ≤0.5 seconds, which meets the actual needs of monitoring the cylinder diameter of marine main engines.

[0091] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for real-time location and loss analysis of defects in the bore of a marine main engine, characterized in that, Includes the following steps: Data acquisition: A multimodal sensor array deployed at preset monitoring points on the inner wall of the cylinder bore is used to collect vibration signals, temperature field distribution data and ultrasonic echo signals on the cylinder bore surface in real time, with a sampling frequency of not less than 1kHz; Data preprocessing: The acquired raw signals are subjected to noise reduction filtering, outlier removal, and spatiotemporal registration to generate a standardized dataset; Defect identification and localization: The preprocessed data is input into the trained defect identification model. The defect type is identified through feature extraction, and the precise location of the defect in the cylinder bore circumference and axial direction is output by combining the spatial coordinate mapping of the sensor array. Loss Quantification Analysis: Based on the positioning results, loss parameters are calculated through the following sub-steps: For crack-like defects, fracture mechanics models are used to calculate crack depth and propagation rate; For wear-related defects, the radial wear of the cylinder diameter is measured with the assistance of a laser displacement sensor, and the wear rate is calculated in combination with the material hardness parameters; For corrosion defects, the corrosion depth and area are calculated using data from electrochemical sensors; Results output and early warning: Defect location information, loss quantification data and trend prediction results are displayed in real time through a visual interface, and an audible and visual early warning is triggered when the loss parameter exceeds the preset threshold.

2. The method according to claim 1, characterized in that, The multimodal sensor array mentioned in step 1 includes: 8-12 piezoelectric vibration sensors are arranged at equal intervals along the circumference of the cylinder liner, with a sampling range of 50Hz-10kHz. Four infrared temperature sensors are deployed at the midpoint and both ends of the cylinder bore axial direction, with a temperature measurement range of -20℃ to 300℃ and an accuracy of ±0.5℃. Two sets of ultrasonic sensors are arranged axially at the top and bottom of the cylinder bore, with a detection frequency of 5MHz-10MHz.

3. The method according to claim 1, characterized in that, The optimizations to the improved YOLOv5 algorithm described in step 3 include: Introducing an attention mechanism to enhance the weighting of defect features; The FocalLoss function is used to address the imbalance problem of small defect samples. The output layer adds a defect confidence score (0-100%), and a defect is considered valid when the score is ≥85%.

4. The method according to claim 1, characterized in that, It also includes a data storage and traceability module, which stores the original signals and analysis results by timestamp, and supports historical data query and trend review for at least 180 days.

5. The method according to claim 1, characterized in that, The warning thresholds mentioned in step 5 include: Crack depth ≥ 0.3 mm or propagation rate ≥ 0.01 mm / day; Radial wear ≥ 0.5% of the nominal cylinder diameter; The corrosion area accounts for more than 3% of the total inner wall area of ​​the cylinder diameter.

6. The method according to claim 2, characterized in that, The piezoelectric vibration sensor is connected to the outer wall of the cylinder liner via a magnetic adsorption mounting base, which has a built-in temperature compensation module.

7. The method according to claim 1, characterized in that, The spatiotemporal registration described in step 2 achieves time alignment of multi-sensor data through GPS time synchronization and spatial alignment through coordinate transformation of the cylinder bore 3D model.

8. The method according to claim 1, characterized in that, The method also includes an adaptive calibration step: every 24 hours of operation, the standard defect sample library is automatically called to fine-tune the identification model online, and the calibration positioning error is ≤0.1mm.

9. The method according to claim 1, characterized in that, The visualization interface in step 5 supports dynamic display of the cylinder diameter 3D model, and detailed parameters can be displayed in a pop-up window by clicking on the defect location.

Images