A Method and System for Measuring Fuel Consumption of Marine Diesel Engines Based on Dynamic Pressure Analysis
By using dynamic pressure analysis and deep learning networks, the problem of distinguishing between the amount of oil flowing through and the amount of oil used for power generation in the measurement of fuel consumption of marine diesel engines has been solved, enabling accurate assessment of combustion efficiency and fault early warning, and improving measurement accuracy and anti-interference capability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CCCC SHANGHAI THIRD HARBOR SCI RES INST CO LTD
- Filing Date
- 2026-06-04
- Publication Date
- 2026-06-30
AI Technical Summary
Existing methods for measuring fuel consumption in marine diesel engines cannot distinguish between the amount of fuel flowing through and the amount of fuel used for power generation. They also cannot utilize high-frequency pressure information for energy efficiency assessment and have weak anti-interference capabilities, resulting in low measurement accuracy.
A dynamic pressure analysis-based approach is adopted. By collecting low-frequency and high-frequency dynamic pressure signals, the Hilbert-Huang transform is used to extract the injection duration, pressure rise slope, and high-frequency energy entropy. Combined with adaptive Kalman filtering and deep learning networks, a multi-scale pressure fusion model is constructed to generate correction factors for fuel consumption calculation and fault warning.
It enables precise quantification of fuel consumption of marine diesel engines, improves measurement accuracy under transient conditions, overcomes the limitations of traditional methods, enhances the accuracy of combustion efficiency assessment, and effectively eliminates mechanical noise interference under harsh sea conditions.
Smart Images
Figure CN122306423A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of marine fuel consumption measurement technology, and in particular to a method and system for measuring marine diesel engine fuel consumption based on dynamic pressure analysis. Background Technology
[0002] With the International Maritime Organization imposing increasingly stringent requirements on ship energy efficiency index and carbon intensity indicators, ship operators are placing extremely high demands on the accuracy of fuel consumption rate monitoring. Fuel costs account for 40%-60% of ship operating costs; therefore, accurate fuel consumption measurement is crucial for achieving economical navigation.
[0003] The existing methods for measuring fuel consumption of marine diesel engines mainly have the following technical problems:
[0004] Currently, the mainstream method for measuring marine fuel consumption relies on Coriolis mass flow meters or volumetric flow meters. While these devices offer high accuracy, they suffer from significant pressure loss, susceptibility to wear, and high maintenance costs. More importantly, they are flow-through meters, measuring only the mass of fuel passing through and failing to detect the actual combustion efficiency within the cylinder. For instance, when fuel injector wear leads to poor atomization, even with the same injection volume, the actual fuel efficiency decreases, and traditional flow meters cannot reflect this ineffective fuel consumption.
[0005] Some existing technologies attempt to use mechanical sensors for density compensation. However, these methods typically treat pressure as a static or quasi-static physical quantity (used only for looking up density in tables), ignoring the frequency characteristics of the pressure signal. In actual ship navigation, the pressure signal contains rich dynamic information: the low-frequency range (<10Hz) reflects pipeline back pressure and hydrostatic pressure, while the high-frequency range (>1kHz) contains microscopic information about injector needle valve movement, fuel atomization, and in-cylinder combustion. Existing technologies have failed to utilize high-frequency information for energy efficiency correction.
[0006] Marine diesel engines operate under unsteady conditions (acceleration, deceleration, variable load) and are affected by wave impacts, resulting in strong low-frequency mechanical noise (0.1-1Hz) in the pipeline pressure. Traditional filtering methods are prone to misinterpreting the true pressure pulsation signal, leading to significant deviations in transient fuel consumption measurements. Summary of the Invention
[0007] The purpose of this invention is to provide a method and system for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis, so as to solve the problems in the prior art that cannot distinguish between the amount of flowing oil and the amount of oil used for power generation, cannot use high-frequency pressure information for energy efficiency assessment, and have weak anti-interference ability.
[0008] To achieve the above objectives, the present invention provides a method for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis, comprising: Step S10: Collect low-frequency dynamic pressure signals with a frequency of less than 10Hz on the fuel line of the marine diesel engine. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics. Step S20, for low-frequency dynamic pressure signal Preprocessing is performed, and the real-time density and pipeline deformation coefficient are calculated based on fuel temperature to generate a first correction factor for correcting fluid volume errors. ; Step S30, processing the high-frequency dynamic pressure signal The Hilbert-Huang Transform (HHT) is performed to decompose the combustion into several intrinsic mode functions, from which the injection duration, pressure rise slope, and high-frequency energy entropy are extracted to generate a second correction factor for evaluating combustion efficiency. ; Step S40: Construct a multi-scale pressure fusion model, and incorporate the first correction factor. Second correction factor Input the model and combine it with the theoretical reference flow rate of the marine diesel engine. Instantaneous effective fuel consumption is calculated through nonlinear mapping. ; Step S50, instantaneous effective fuel consumption Integrate over time to obtain the cumulative fuel consumption. Simultaneously, the injection duration, pressure rise slope, or high-frequency energy entropy extracted in step S30 are compared with the benchmark values in the preset standard feature database to output a fault warning for the injection system.
[0009] Furthermore, in step S20, the low-frequency dynamic pressure signal... Preprocessing includes: introducing six-degree-of-freedom motion data acquired by the ship's inertial measurement unit (IMU), and then filtering it using an adaptive Kalman filter. The mechanical noise caused by wave impact and hull rolling is eliminated.
[0010] First correction factor It is calculated using the following formula:
[0011] in, To use low-frequency dynamic pressure signals The calibrated pipeline deformation coefficient For the current fuel density relative to The change The standard fuel density; the current fuel density is obtained after pretreatment. The data was obtained by referring to a table based on real-time fuel temperature.
[0012] Furthermore, in step S30, the high-frequency dynamic pressure signal... Perform Hilbert-Huang transform, including: for high-frequency dynamic pressure signals The signal is subjected to empirical mode decomposition to obtain several intrinsic mode function components arranged in descending order of frequency; the intrinsic mode function components containing fuel injection atomization information are selected, and the fuel injection duration, pressure rise slope and high-frequency energy entropy are extracted based on the components; The injection duration is determined by identifying high-frequency dynamic pressure signals. The time difference between the start and end points is determined; the slope of the pressure rise edge is determined by calculating the ratio of the peak pressure of the injection pulse to the rise time; the high-frequency energy entropy is determined by performing energy statistics on the intrinsic mode function components containing injection atomization information and calculating the energy distribution probability of the component within a preset frequency band.
[0013] Second correction factor It is calculated using the following formula:
[0014] in, The combustion efficiency attenuation coefficient, For the duration of fuel injection, As the baseline injection duration, As a reference high-frequency energy entropy, For high-frequency energy entropy, The slope of the pressure rise, The slope of the rising edge of the reference pressure; , and All were obtained by collecting high-frequency dynamic pressure signals under healthy conditions at standard ambient temperature and standard fuel grade during the diesel engine bench test phase, and pre-calibrated by the same Hilbert-Huang transform decomposition and feature extraction steps.
[0015] Furthermore, in step S40, the multi-scale pressure fusion model is constructed using a pre-trained Long Short-Term Memory (LSTM) network, the input layer of which receives a first correction factor. Second correction factor Real-time load signal of marine diesel engine and theoretical reference flow rate The instantaneous effective fuel consumption is output through the nonlinear mapping of the hidden layer. This can be expressed as a formula:
[0016] in, This indicates the output predicted by the pre-trained LSTM. , and The input features for LSTM processing, This represents network parameters.
[0017] The Long Short-Term Memory (LSTM) network model dynamically adjusts the first correction factor based on changes in diesel engine load. With the second correction factor Hidden layer weights: Increased when the load is below a preset threshold. The corresponding weights are used to focus on the combustion stability of marine diesel engines, and are increased under operating conditions where the load exceeds a preset threshold. The weighting is used to focus on the accuracy of flow measurement in marine diesel engines.
[0018] Furthermore, in step S50, the fault warning includes: when the high-frequency energy entropy is lower than the preset standard value and the slope of the pressure rise decreases, it is determined to be a fuel injector carbon buildup or wear fault; when the fuel injection duration is abnormally prolonged and the high-frequency energy entropy oscillates at a frequency higher than the preset threshold, it is determined to be a fuel injector needle valve sticking fault.
[0019] The present invention also provides a marine diesel engine fuel consumption measurement system based on dynamic pressure analysis, comprising: Pressure signal acquisition module: used to acquire low-frequency dynamic pressure signals with a frequency of less than 10Hz from the fuel lines of marine diesel engines. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics. Low-frequency signal processing module: used for processing low-frequency dynamic pressure signals Preprocessing is performed, and the real-time density and pipeline deformation coefficient are calculated based on fuel temperature to generate a first correction factor for correcting fluid volume errors. ; High-frequency signal processing module: used for processing high-frequency dynamic pressure signals. The Hilbert-Huang transform is performed to decompose the combustion into several intrinsic mode functions, from which the injection duration, pressure rise slope, and high-frequency energy entropy are extracted to generate a second correction factor for evaluating combustion efficiency. ; Fuel consumption measurement module: used to construct a multi-scale pressure fusion model, including the first correction factor. Second correction factor Input the model and combine it with the theoretical reference flow rate of the marine diesel engine. Instantaneous effective fuel consumption is calculated through nonlinear mapping. ; Anomaly warning module: used for monitoring instantaneous effective fuel consumption Integrate over time to obtain the cumulative fuel consumption. Simultaneously, the injection duration, pressure rise slope, or high-frequency energy entropy extracted in step S30 are compared with the benchmark values in the preset standard feature database to output a fault warning for the injection system.
[0020] The present invention discloses the following technical effects: This invention provides a method and system for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis. It employs a dynamic pressure analysis strategy, using mechanical sensors to acquire low-frequency dynamic pressure signals reflecting pipeline back pressure and high-frequency dynamic pressure signals reflecting the micro-frequency characteristics of fuel injection. Two correction factors are generated, and multiple high-frequency features are extracted using the Hilbert-Huang transform. This method overcomes the limitation of traditional flowmeters that only measure the amount of fuel flowing through, and accurately quantifies combustion efficiency through high-frequency features, achieving a shift from flow accumulation to effective energy efficiency measurement, thus improving the accuracy of fuel consumption measurement under transient operating conditions. Furthermore, this method constructs a multi-scale pressure fusion model based on a deep learning network, dynamically adjusting the weight ratio of the correction factor according to the load changes of the ship's diesel engine; it solves the defect that the traditional linear fuel consumption calculation formula is difficult to fit the nonlinear characteristics of the diesel engine, and realizes adaptive intelligent sensing of heavy combustion under low load and heavy flow under high load; when preprocessing the low-frequency dynamic pressure signal, this method introduces an IMU and an adaptive Kalman filter to construct a state-space model, and removes mechanical noise caused by wave impact and hull rolling from the signal, effectively overcoming the interference of violent hull movement on pressure readings under severe sea conditions, ensuring the smoothness and stability of the measurement data, and solving the problem that traditional filtering methods are prone to signal distortion or lag. Attached Figure Description
[0021] Figure 1 This is a flowchart illustrating a method for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis, provided as an embodiment of the present invention.
[0022] Figure 2 This is a schematic diagram of a marine diesel engine fuel consumption measurement system based on dynamic pressure analysis, provided as an embodiment of the present invention. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on the present invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0024] Example 1: This embodiment of the invention provides a method for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis, such as... Figure 1 As shown, the method includes: Step S10: Collect low-frequency dynamic pressure signals with a frequency of less than 10Hz on the fuel line of the marine diesel engine. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics.
[0025] In this embodiment, sensors are deployed for the main engine fuel system inside the cabin: Two Rosemount 3051S diffused silicon pressure transmitters (range 0-1.0 MPa, response frequency less than 10Hz) are installed on the main fuel inlet manifold (low-pressure side) to acquire low-frequency dynamic pressure signals in real time. This signal mainly reflects the fluctuation of pipeline back pressure and the slow change in fuel density.
[0026] A Kistler 6125C piezoelectric dynamic pressure sensor is installed near the injector on the high-pressure fuel line of each of the six cylinders. This sensor is configured to acquire pressure pulses with a frequency greater than 1kHz to capture the lift of the injector needle valve, fuel atomization, and combustion shock waves, thereby obtaining a high-frequency dynamic pressure signal. .
[0027] Step S20, for low-frequency dynamic pressure signal Preprocessing is performed, and the real-time density and pipeline deformation coefficient are calculated based on fuel temperature to generate a first correction factor for correcting fluid volume errors. .
[0028] In this embodiment, due to the ship's rolling motion (approximately ±25 degrees) caused by waves during navigation, the IMU (Inertial Measurement Unit) collects the ship's acceleration data and uses an adaptive Kalman filter to construct a state-space model from the IMU data. To remove low-frequency mechanical noise in the 0.1Hz-1Hz range and obtain the pure fluid pressure component, the following detailed steps are involved: Receive real-time low-frequency dynamic pressure signal from low-frequency pressure transmitter And six-degree-of-freedom data from the IMU, including three-axis accelerations and three-axis angular velocities of the hull motion; Since the pressure sensor is installed on the fuel line of the marine diesel engine, the IMU coordinate system needs to be aligned with the pipeline installation coordinate system through a coordinate transformation matrix to ensure that the acceleration direction is consistent with the sensing direction of the pressure sensor. Construct a discrete-time state-space model and define State vector at time step Includes real fluid pressure components and hull motion disturbance components : , This represents the transpose operation; the state equation is expressed as: ,in Here is the state transition matrix. and Let these represent the state vector and process noise at the previous time step, respectively; the observation equation is expressed as: ,in The original low-frequency dynamic pressure signal containing noise , The observation matrix (the transmitter observations are the superposition of the actual pressure and motion disturbances). To observe noise; Unlike traditional Kalman filtering, this invention employs an adaptive algorithm to estimate the noise covariance online to address sudden changes in sea state, specifically including: Predict the current state based on the state estimate from the previous time step. And error covariance matrix Using real-time acceleration data collected by the IMU, the theoretical value of the additional pressure caused by ship rolling and heave is calculated. ,in For fuel density, It is the acceleration due to gravity. Obtained from IMU inulin; the difference between theoretical and predicted interference is calculated, and the observation noise covariance matrix is dynamically adjusted to increase its weight under severe sea conditions, thus preventing filter divergence; when new pressure data... Calculate the Kalman gain upon arrival. And update the state estimate: ,in and This represents the updated current state and the predicted state. Kalman gain; After the above iterative calculations, from the state vector Extract pure real fluid pressure components And the ship motion disturbance component Discard it.
[0029] Preprocessed low-frequency dynamic pressure signal To remove the mechanical noise in the 0.1Hz to 1Hz frequency band (corresponding to the wave impact frequency) from the pure low-frequency pressure signal, the first correction factor is calculated using the following formula:
[0030] in, To use low-frequency dynamic pressure signals The calibrated pipeline deformation coefficient For the current fuel density relative to The change The standard fuel density; the current fuel density is obtained after pretreatment. The data was obtained by referring to a table based on real-time fuel temperature.
[0031] For example, the current fuel temperature is read as 145°C. Combined with the preprocessed current pressure Consult the ISO 8217 fuel property table to calculate the current real-time fuel density. Based on the physical parameters of the fuel line (outer diameter 127mm, wall thickness 10mm), the deformation coefficient was calculated in real time using Hooke's law and Lamé's formula for thin-walled cylinders. Substituting into the formula, the first correction factor is obtained. :
[0032] Here A negative value indicates an artificial increase in volumetric flow rate due to a decrease in fuel density, which requires negative correction.
[0033] Step S30, processing the high-frequency dynamic pressure signal The Hilbert-Huang transform is performed to decompose the combustion into several intrinsic mode functions, from which the injection duration, pressure rise slope, and high-frequency energy entropy are extracted to generate a second correction factor for evaluating combustion efficiency. .
[0034] In this embodiment, a single injection cycle signal from the third cylinder is extracted as the real-time high-frequency dynamic pressure signal. To process the data, first perform a Hilbert-Huang transform, including the following detailed steps: First, empirical mode decomposition is performed on the signal to obtain several intrinsic mode functions (IMFs): Filtering process: Identify all local maxima and minima in the signal, generate upper and lower envelopes for each, and calculate the mean of the upper and lower envelopes. And subtract from the original signal: ; Iteration stops: Repeat the above filtering process until... The two conditions for IMF are met: the number of extrema and the number of zero crossings differ by no more than 1, and the upper and lower envelopes are symmetrical about the time axis; at this point, the first high-frequency IMF component is obtained. ; Residual calculation: from the original signal Subtraction Repeat the above steps for the remaining residuals to obtain... , ...until the residual becomes a monotonic function; Component filtering: The highest frequency IMF component was identified, which includes injector needle valve impact, fuel atomization, and cavitation noise, and was denoted as... .
[0035] Secondly, for the selected By performing a Hilbert transform to construct an analytic signal, the instantaneous frequency and instantaneous amplitude are obtained. Based on this, the following three core features are extracted: Injection duration :set up The detection threshold is set at 10% of the peak amplitude; the moment when the signal amplitude first exceeds the threshold is identified. (At the start of fuel injection) and the last time the fuel level falls below the threshold. (Fuel injection complete), calculation and The difference is used to obtain the fuel injection duration. ; Pressure rise slope :exist Find the pressure peak during the first 10% of the subsequent injection cycles. Calculate rise time Starting from fuel injection When the pressure reaches The time difference further yields the pressure rise along the slope. : ,in This indicates the initial static pressure in the high-pressure fuel line before fuel injection begins; High-frequency energy entropy :Will The frequency band is divided into Each sub-band has equal width, and the signal energy within each sub-band is calculated. : ,in Indicates the first One equal-width sub-band, Indicates frequency, Indicate the square calculation; further calculate the total energy: To obtain the probability of each sub-band High-frequency energy entropy Represented as: .
[0036] Finally, the preset standard feature database (which stores the benchmark values for bench calibration) is invoked. , and Substitute the extracted real-time features into the following formula to generate the second correction factor. :
[0037] in, The combustion efficiency attenuation coefficient is set empirically. As the baseline injection duration, As a reference high-frequency energy entropy, The slope of the rising edge of the reference pressure; For example, the selected high-frequency components The center frequency is 15kHz. Hilbert spectrum analysis was performed on it to identify the time difference between the pulse waveform rising from a 10% peak to 90% and then falling back to 10%. This was measured in practice. Calculate the ratio of peak pressure to rise time to obtain... Calculate the energy distribution probability in the 1kHz-30kHz frequency band. , and thus The baseline values for the database were calibrated during the bench testing phase. , , Substituting into the formula yields the second correction factor. :
[0038] This indicates that approximately 6.3% of the fuel was not fully utilized due to decreased combustion efficiency.
[0039] Step S40: Construct a multi-scale pressure fusion model, and incorporate the first correction factor. Second correction factor Input the model and combine it with the theoretical reference flow rate of the marine diesel engine. Instantaneous effective fuel consumption is calculated through nonlinear mapping. .
[0040] In this embodiment, the theoretical reference flow rate of the marine diesel engine It is calculated based on the control parameters of the diesel engine itself, and includes the following detailed steps: The current main engine speed is obtained by reading real-time data from the diesel engine's ECS (Electronic Control System) via the ship's local area network. And throttle lever position; call the diesel engine injection pulse spectrum diagram pre-stored in ROM, which is provided by the diesel engine manufacturer and describes the theoretical cyclic injection quantity at a specific speed and throttle opening; Combined with the current diesel engine speed Convert the fuel injection amount per cycle to hourly flow rate:
[0041] in, This indicates the amount of fuel injected per cylinder per cycle. The number of cylinders; in this embodiment, the number of cylinders is 6, calculated as follows: , Indicates hours.
[0042] The multi-scale pressure fusion model employs a pre-trained two-layer stacked Long Short-Term Memory (LSTM) network as its core algorithm. This LSTM model has been trained on a large amount of historical fuel consumption measurement data. At each time step, an input tensor is constructed as the input to the LSTM model, containing the following four dimensions: theoretical baseline flow rate. First correction factor Second correction factor and normalized load factor ,in The dynamic weight mechanism from ECS is used to activate the model. During model training, to prevent gradient explosion, the input tensor is min-max scaled before entering the LSTM and mapped to the [-1,1] interval.
[0043] The LSTM model learns the feature importance under different operating conditions through gating units. During the inference phase, the model exhibits a clear dynamic weight preference: increasing the weight when the load is below a preset threshold. The corresponding weights are used to focus on the combustion stability of marine diesel engines, and are increased under operating conditions where the load exceeds a preset threshold. The weighting is used to focus on the accuracy of flow measurement in marine diesel engines.
[0044] The LSTM model's memory cells store the flow rate state from the previous time step, ensuring the temporal continuity of the output instantaneous fuel consumption and avoiding sawtooth fluctuations caused by cylinder firing intervals. The LSTM output passes through a fully connected layer to map the high-dimensional features back to a one-dimensional flow rate value; the instantaneous effective fuel consumption is obtained based on the nonlinear mapping of the LSTM. The process can be expressed by the formula:
[0045] in, This indicates the output predicted by the pre-trained LSTM. , and The input features for LSTM processing, This represents network parameters.
[0046] To verify the effectiveness of the LSTM model, the system background simultaneously ran the linear formula as a comparison. The results showed that the output of LSTM was higher than that of the linear formula. This is because LSTM captured nonlinear losses: under high load, due to high-pressure leakage and secondary atomization effect caused by carbon deposits on the fuel injectors, the actual fuel consumption loss is greater than that predicted by the linear model. LSTM compensated for this hidden loss by learning from historical data, demonstrating the advantages of intelligent perception.
[0047] Step S50, instantaneous effective fuel consumption Integrate over time to obtain the cumulative fuel consumption. Simultaneously, the injection duration, pressure rise slope, or high-frequency energy entropy extracted in step S30 are compared with the benchmark values in the preset standard feature database to output a fault warning for the injection system.
[0048] In this embodiment, the fault warning includes: when the high-frequency energy entropy is lower than the preset standard value and the slope of the pressure rise decreases, it is determined to be a fuel injector carbon buildup or wear fault; when the fuel injection duration is abnormally prolonged and the high-frequency energy entropy shows oscillations with a frequency higher than the preset threshold, it is determined to be a fuel injector needle valve sticking fault.
[0049] For example, for Perform millisecond-level integration and update cumulative fuel consumption in real time. The application layer interface displays the current cumulative fuel consumption as 125.4 tons and continuously compares real-time characteristics with the standard database. On the fifth day of sailing, high-frequency energy entropy was detected in the third cylinder. The pressure dropped sharply from 0.82 to 0.60, and the pressure increased along the slope. The energy level decreased from 25 to 18. Based on the preset logic (energy entropy is lower than the standard value and the slope becomes slower), the system determined that "Cylinder No. 3 injector carbon buildup or wear failure" occurred. The application layer immediately popped up a red warning window with the following message: "Warning: Cylinder No. 3 Injector Efficiency Low. Check for Carbon Deposits." This guided the crew to carry out maintenance at the next port, avoiding fuel waste and piston burning risks caused by single-cylinder failure.
[0050] Example 2: The marine diesel engine fuel consumption measurement system based on dynamic pressure analysis provided in this embodiment of the invention can execute the marine diesel engine fuel consumption measurement method based on dynamic pressure analysis provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method, such as... Figure 2 As shown, it includes: Pressure signal acquisition module: used to acquire low-frequency dynamic pressure signals with a frequency of less than 10Hz from the fuel lines of marine diesel engines. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics. Low-frequency signal processing module: used for processing low-frequency dynamic pressure signals Preprocessing is performed, and the real-time density and pipeline deformation coefficient are calculated based on fuel temperature to generate a first correction factor for correcting fluid volume errors. ; High-frequency signal processing module: used for processing high-frequency dynamic pressure signals. The Hilbert-Huang transform is performed to decompose the combustion into several intrinsic mode functions, from which the injection duration, pressure rise slope, and high-frequency energy entropy are extracted to generate a second correction factor for evaluating combustion efficiency. ; Fuel consumption measurement module: used to construct a multi-scale pressure fusion model, including the first correction factor. Second correction factor Input the model and combine it with the theoretical reference flow rate of the marine diesel engine. Instantaneous effective fuel consumption is calculated through nonlinear mapping. ; Anomaly warning module: used for monitoring instantaneous effective fuel consumption Integrate over time to obtain the cumulative fuel consumption. Simultaneously, the injection duration, pressure rise slope, or high-frequency energy entropy extracted in step S30 are compared with the benchmark values in the preset standard feature database to output a fault warning for the injection system.
[0051] Although this invention makes various references to certain modules in the system according to embodiments of the present invention, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of the present invention.
[0052] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention. In some cases, the actions or steps described in this invention can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
Claims
1. A method for measuring the fuel consumption of a marine diesel engine based on dynamic pressure analysis, characterized in that, The method includes: Step S10: Collect low-frequency dynamic pressure signals with a frequency of less than 10Hz on the fuel line of the marine diesel engine. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics. Step S20, pre-process the low-frequency dynamic pressure signal Calculate real-time density and pipeline deformation coefficient combined with fuel temperature to generate the first correction factor for correcting fluid volume error ; Step S30, the high-frequency dynamic pressure signal is subjected to Hilbert-Huang transform, which is decomposed into several intrinsic mode functions, and the injection duration, pressure rising slope and high-frequency energy entropy are extracted therefrom to generate a second correction factor for evaluating the combustion efficiency ; Step S40, construct a multi-scale pressure fusion model, input the first correction factor and the second correction factor to the model, and combine the theoretical reference flow of the ship diesel engine to calculate the instantaneous effective fuel consumption through nonlinear mapping ; Step S50, time-integrating the instantaneous effective fuel consumption to obtain the cumulative fuel consumption ; meanwhile, the injection duration, the pressure rise slope or the high-frequency energy entropy extracted in step S30 are compared with the reference values in the preset standard feature database, and an injection system fault early warning is output.
2. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 1, characterized in that, In step S20, the low-frequency dynamic pressure signal Pre-processing is performed, including: introducing six-degree-of-freedom motion data collected by a ship inertial measurement unit IMU, removing mechanical noise caused by sea wave beating and ship body swinging from the low-frequency dynamic pressure signal by an adaptive Kalman filter.
3. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 1, characterized in that, In step S20, the first correction factor is calculated by the following equation: in, To use low-frequency dynamic pressure signals The calibrated pipeline deformation coefficient For the current fuel density relative to The change The standard fuel density; the current fuel density is obtained after pretreatment. The data was obtained by referring to a table based on real-time fuel temperature.
4. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 1, characterized in that, In step S30, the high-frequency dynamic pressure signal is subjected to a Hilbert-Huang transform, including: performing empirical mode decomposition on the high-frequency dynamic pressure signal to obtain a plurality of intrinsic mode function components arranged in descending order of frequency; screening out an intrinsic mode function component containing fuel injection atomization information, and extracting the fuel injection duration, pressure rising edge slope, and high-frequency energy entropy based on the component.
5. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 4, characterized in that, The injection duration is determined by identifying high-frequency dynamic pressure signals. The time difference between the start and end points is determined; the slope of the pressure rise edge is determined by calculating the ratio of the peak pressure of the injection pulse to the rise time; the high-frequency energy entropy is determined by performing energy statistics on the intrinsic mode function components containing injection atomization information and calculating the energy distribution probability of the component within a preset frequency band.
6. The method for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis as described in claim 1, characterized in that, In step S30, the second correction factor It is calculated using the following formula: in, The combustion efficiency attenuation coefficient, For the duration of fuel injection, As the baseline injection duration, As a reference high-frequency energy entropy, For high-frequency energy entropy, The slope of the pressure rise, The slope of the rising edge of the reference pressure; , and All were obtained by collecting high-frequency dynamic pressure signals under healthy conditions at standard ambient temperature and standard fuel grade during the diesel engine bench test phase, and pre-calibrated by the same Hilbert-Huang transform decomposition and feature extraction steps.
7. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 1, characterized in that, In step S40, the multi-scale pressure fusion model is constructed using a pre-trained Long Short-Term Memory (LSTM) network, whose input layer receives a first correction factor. Second correction factor Real-time load signal of marine diesel engine and theoretical reference flow rate The instantaneous effective fuel consumption is output through the nonlinear mapping of the hidden layer. This can be expressed as a formula: in, This indicates the output predicted by the pre-trained LSTM. , and The input features for LSTM processing, This represents network parameters.
8. The method for measuring fuel consumption of marine diesel engines based on dynamic pressure analysis as described in claim 7, characterized in that, The Long Short-Term Memory (LSTM) model dynamically adjusts the first correction factor based on changes in diesel engine load. With the second correction factor Hidden layer weights: Increased when the load is below a preset threshold. The corresponding weights are used to focus on the combustion stability of marine diesel engines, and are increased under operating conditions where the load exceeds a preset threshold. The weighting is used to focus on the accuracy of flow measurement in marine diesel engines.
9. The method for measuring marine diesel engine fuel consumption based on dynamic pressure analysis as described in claim 1, characterized in that, In step S50, the fault warning includes: when the high-frequency energy entropy is lower than the preset standard value and the slope of the pressure rise decreases, it is determined to be a fuel injector carbon buildup or wear fault; when the fuel injection duration is abnormally prolonged and the high-frequency energy entropy oscillates at a frequency higher than the preset threshold, it is determined to be a fuel injector needle valve sticking fault.
10. A marine diesel engine fuel consumption measurement system based on dynamic pressure analysis, characterized in that, The system is used to implement the marine diesel engine fuel consumption measurement method based on dynamic pressure analysis as described in any one of claims 1-9, and the system comprises: Pressure signal acquisition module: used to acquire low-frequency dynamic pressure signals with a frequency of less than 10Hz on the fuel lines of marine diesel engines. and high-frequency dynamic pressure signals with frequencies greater than 1kHz The low-frequency dynamic pressure signal reflects the changes in pipeline back pressure and fuel density, while the high-frequency dynamic pressure signal reflects the injection pulse and combustion characteristics. Low-frequency signal processing module: used for processing low-frequency dynamic pressure signals Preprocessing is performed, and the real-time density and pipeline deformation coefficient are calculated based on fuel temperature to generate a first correction factor for correcting fluid volume errors. ; High-frequency signal processing module: used for processing high-frequency dynamic pressure signals. The Hilbert-Huang transform is performed to decompose the combustion into several intrinsic mode functions, from which the injection duration, pressure rise slope, and high-frequency energy entropy are extracted to generate a second correction factor for evaluating combustion efficiency. ; Fuel consumption measurement module: used to construct a multi-scale pressure fusion model, including the first correction factor. Second correction factor Input the model and combine it with the theoretical reference flow rate of the marine diesel engine. Instantaneous effective fuel consumption is calculated through nonlinear mapping. ; Anomaly warning module: used for monitoring instantaneous effective fuel consumption Integrate over time to obtain the cumulative fuel consumption. Simultaneously, the injection duration, pressure rise slope, or high-frequency energy entropy extracted in step S30 are compared with the benchmark values in the preset standard feature database to output a fault warning for the injection system.