Method and system for wear diagnosis of a fuel system of a diesel engine

Abstract

The present disclosure provides a method and system for diagnosing wear of a fuel system of a diesel engine. The method comprises: acquiring injection parameter data of the fuel system in an injection combustion process, and pressure signal data in an oil pipe; collecting particulate matters in exhaust gas in the injection combustion process by using an electrostatic deposition sampler, and shooting images of the particulate matters by using a microscopic camera device to extract morphology data of the particulate matters from the images; emitting pulsed laser to the particulate matters by using a laser-induced breakdown spectrometer to excite plasma emission characteristic spectrum in the particulate matters to obtain composition data of the particulate matters according to the characteristic spectrum; extracting features of the injection parameter data, the morphology data of the particulate matters and the composition data of the particulate matters respectively to obtain injection features, morphology features of the particulate matters and composition features; performing time-frequency analysis on the pressure signal data to obtain pressure features; and inputting the injection features, the pressure features, the morphology features and the composition features into a wear diagnosis model to obtain a wear diagnosis result.

Description

Technical Field

[0001] This disclosure relates to the field of diesel engine fault diagnosis technology, and specifically to a method and system for diagnosing wear in the fuel system of a diesel engine. Background Technology

[0002] Diesel engines, with their advantages of high thermal efficiency and strong power, are widely used in industry, agriculture, transportation and other fields. As the core component of a diesel engine, the fuel system is responsible for the storage, transportation, injection and atomization of fuel, and its performance directly affects the power output, fuel economy and emission standards of the diesel engine.

[0003] However, diesel engine fuel systems operate under high temperature, high pressure, and complex mechanical vibration environments for extended periods, making them prone to wear. Therefore, the development of diagnostic methods for fuel systems is particularly important. Current diesel engine fault diagnosis methods include vibration analysis, acoustic emission analysis, pressure analysis, and thermodynamic parameter analysis. However, these methods rely on single indicators and cannot comprehensively reflect the wear condition; thus, the accuracy of diagnosis needs improvement. Summary of the Invention

[0004] In view of this, the present disclosure provides a method and system for diagnosing wear in the fuel system of a diesel engine.

[0005] One aspect of this disclosure provides a method for diagnosing wear in a diesel engine's fuel system, comprising: acquiring fuel injection parameter data and pressure signal data within the fuel line during the fuel injection combustion process; collecting particulate matter from the exhaust gas during the fuel injection combustion process using an electrostatic deposition sampler and capturing images of the particulate matter using a microscopic imaging device to extract particulate morphology data from the images; emitting pulsed laser light onto the particulate matter using a laser-induced breakdown spectrometer to excite plasma emission characteristic spectra in the particulate matter, thereby obtaining particulate composition data based on the characteristic spectra; performing feature extraction on the fuel injection parameter data, particulate morphology data, and composition data respectively to obtain fuel injection features, particulate morphology features, and composition features; performing time-frequency analysis on the pressure signal data to obtain pressure features; and inputting the fuel injection features, pressure features, particulate morphology features, and composition features into a wear diagnosis model to obtain wear diagnosis results, the wear diagnosis results including worn components and wear values ​​of the worn components.

[0006] According to embodiments of this disclosure, the morphological data of particulate matter includes geometric morphological data and particle size distribution data; feature extraction is performed on the morphological data of particulate matter to obtain morphological features of particulate matter, including: feature extraction of the geometric morphological data of particulate matter to obtain the roundness and aspect ratio of particulate matter; feature extraction of the particle size distribution data of particulate matter to obtain the fractal dimension of particulate matter; and obtaining the morphological features of particulate matter based on roundness, aspect ratio and fractal dimension.

[0007] According to embodiments of this disclosure, feature extraction is performed on the composition data of particulate matter to obtain the composition features of particulate matter, including: extracting the features of metal elements from the composition data of particulate matter, determining the content of each metal element and the target content ratio, wherein the target content ratio represents the ratio of the contents of two target metal elements; and obtaining the composition features of particulate matter based on the content of each metal element and the target content ratio.

[0008] According to embodiments of this disclosure, time-frequency analysis is performed on pressure signal data to obtain pressure characteristics, including: using wavelet packets to perform multiple iterative decompositions on the pressure signal data to obtain multiple sub-band signals in different frequency ranges; determining the energy proportion of each sub-band signal based on the energy of each sub-band signal; determining the energy entropy and the target sub-band signal representing the main frequency based on the energy proportion of each sub-band signal; and obtaining the pressure characteristics based on the energy entropy and the target sub-band signal.

[0009] According to embodiments of this disclosure, the fuel system includes multiple cylinder blocks; feature extraction of injection parameter data to obtain injection features includes: for each cylinder block, performing statistical feature extraction on the injection parameter data of the cylinder block to obtain injection statistical features of the cylinder block; performing feature extraction on the differences in injection parameter data of the cylinder block between multiple injections to obtain injection stability features of the cylinder block; performing feature extraction on the differences in injection parameter data between the multiple cylinder blocks to obtain injection difference features between the multiple cylinder blocks; and obtaining injection features based on the injection statistical features of any cylinder block, the injection difference features between the multiple cylinder blocks, and the injection stability features of any cylinder block.

[0010] According to embodiments of this disclosure, the injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics are input into a wear diagnosis model to obtain wear diagnosis results. This includes: inputting the injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics into a feature fusion module of the wear diagnosis model to obtain fused features; inputting the fused features into a wear classification module of the wear diagnosis model to obtain worn components; and inputting the fused features into a wear value determination module of the wear diagnosis model to obtain the wear value of the worn components.

[0011] According to an embodiment of this disclosure, the wear value determination module includes a wear value determination submodule, and the wear components include a fuel injector and a high-pressure fuel pump; the module that inputs the fused features into the wear diagnosis model to obtain the wear value includes: when the wear component is a fuel injector or a high-pressure fuel pump, inputting the fused features into the wear value determination submodule to obtain the wear value of the fuel injector or the wear value of the high-pressure fuel pump.

[0012] According to embodiments of this disclosure, the wear value determination module includes a wear ratio determination submodule, and the wear components include the combined wear of the injector and the high-pressure pump. The module inputs fusion features into the wear diagnosis model to obtain wear values, including: in the case of combined wear of the injector and the high-pressure pump, inputting the content of each metal element in the particulate matter and the target content ratio into the wear ratio determination submodule to obtain the wear ratio of the injector and the high-pressure pump; and inputting the wear ratio and fusion features into the wear value determination submodule to obtain the wear values ​​of the injector and the high-pressure pump respectively.

[0013] According to embodiments of this disclosure, the method further includes: inputting the fusion features, the worn component, and the wear value of the worn component into the remaining time prediction module of the wear diagnosis model to obtain the remaining service time of the high-pressure oil pump and the injector respectively.

[0014] Another aspect of this disclosure provides a system for diagnosing wear in a diesel engine's fuel system, comprising: an acquisition module for acquiring fuel injection parameter data and pressure signal data in the fuel line during the fuel injection combustion process; a first data extraction module for collecting particulate matter from the exhaust gas during the fuel injection combustion process using an electrostatic deposition sampler and capturing images of the collected particulate matter using a microscopic imaging device to extract particulate matter morphology data from the images; a second data extraction module for emitting pulsed laser light onto the collected particulate matter using a laser-induced breakdown spectrometer to excite plasma emission characteristic spectra in the particulate matter, thereby obtaining particulate matter composition data based on the characteristic spectra; a feature extraction module for extracting features from the fuel injection parameter data, particulate matter morphology data, and composition data respectively, to obtain fuel injection features, particulate matter morphology features, and composition features; an analysis module for performing time-frequency analysis on the pressure signal data to obtain pressure features; and a wear diagnosis module for inputting the fuel injection features, pressure features, particulate matter morphology features, and composition features into a wear diagnosis model to obtain wear diagnosis results, the wear diagnosis results including worn parts and wear values ​​of the worn parts.

[0015] According to embodiments of this disclosure, by collecting particulate matter morphology and composition data of exhaust gas directly related to the operating state of the fuel system, as well as collecting fuel injection parameter data and pressure signal data, and extracting injection characteristics, pressure characteristics, and particulate matter morphology and composition characteristics as inputs to a wear diagnosis model, wear diagnosis of the fuel system can be achieved. Since particulate matter morphology and composition data can enhance the direct correlation with fuel system wear, combined with basic collected fuel injection parameter data and pressure signal data, the comprehensiveness and accuracy of wear diagnosis can be improved. Attached Figure Description

[0016] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0017] Figure 1 A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to an embodiment of the present disclosure is shown schematically.

[0018] Figure 2 A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to another embodiment of the present disclosure is shown schematically.

[0019] Figure 3 A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to yet another embodiment of the present disclosure is shown.

[0020] Figure 4 A block diagram of a fuel system wear diagnosis system for a diesel engine according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0021] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0022] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0023] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0024] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0025] In related technologies, typical diesel engine fault diagnosis methods include vibration analysis, acoustic emission analysis, pressure analysis, and thermodynamic parameter analysis. However, these methods require the installation of multiple intrusive sensors to detect faults in the diesel engine based on detected thermodynamic parameters, fuel pressure, or flow rate. In practical applications, they exhibit significant limitations: First, some fault diagnosis technologies employ intrusive detection methods, requiring the disassembly of critical components such as fuel injectors and high-pressure fuel pumps for offline measurements, affecting continuous equipment operation and introducing significant installation defects. Second, most mainstream fault diagnosis technologies suffer from a significant reliance on a single parameter, depending solely on data such as fuel pressure and flow rate. However, fault diagnosis methods relying on a single parameter struggle to accurately identify specific fault causes and cannot comprehensively reflect wear conditions. Third, existing fault diagnosis methods exhibit significant lag, failing to detect early-stage minor faults that do not yet have a noticeable impact on the overall performance of the power system. Therefore, by the time a fault is discovered, it often indicates severe damage, high repair costs, and unnecessary scrapping. However, reports of advanced diesel engine fuel system fault diagnosis technologies that can simultaneously circumvent these three limitations are currently very limited.

[0026] To address the aforementioned issues, indirect monitoring technologies have gradually become a key focus for both academia and industry in recent years. Fault diagnosis techniques based on the coupling of vibration signals, lubricating oil condition, and thermodynamic parameters have gradually become mainstream. However, due to the complexity of diesel engine systems, vibration signal analysis is easily interfered with by other mechanical noises from the engine, making it difficult to strictly correspond to the specific relationship between vibration signal characteristics and fuel system faults, resulting in low specificity. Furthermore, for the critical issue of monitoring internal wear conditions in fuel system fault detection, lubricating oil detection cannot directly reflect this, requiring the use of other detection methods. In summary, fault detection methods based on vibration signals and lubricating oil condition are insufficient to meet the current needs of diesel engine fuel system wear characteristic detection and fault detection. Further exploration is needed of other non-invasive fault detection methods that can be coupled with vibration / lubricating oil detection and have a certain theoretical and technical foundation.

[0027] Exhaust particulate matter is directly related to the operating condition of the fuel system and has a certain degree of specificity. For example, poor fuel atomization or fuel injector wear can lead to increased particle size and higher carbon soot content. The content of metal elements in exhaust particulate matter is also highly correlated with the wear of high-pressure fuel pump bearings or fuel injector coatings. Therefore, exhaust particulate matter can serve as one of the important diagnostic indicators of fuel system wear.

[0028] In view of this, this application provides a method for diagnosing wear in the fuel system of a diesel engine in order to at least solve at least one of the above-mentioned problems.

[0029] Figure 1A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to an embodiment of the present disclosure is shown.

[0030] During operation of S110, fuel injection parameter data and pressure signal data in the fuel line are acquired during the fuel injection and combustion process.

[0031] In operation of S120, an electrostatic deposition sampler is used to collect particulate matter from the exhaust gas during fuel injection combustion, and a microscopic camera is used to capture images of the particulate matter in order to extract particulate morphology data from the images.

[0032] In operation S130, a laser-induced breakdown spectrometer is used to emit pulsed laser light onto the particulate matter, exciting the plasma in the particulate matter to emit characteristic spectra, so as to obtain the composition data of the particulate matter based on the characteristic spectra.

[0033] In operation S140, feature extraction is performed on the fuel injection parameter data, particulate matter morphology data, and composition data to obtain fuel injection features, particulate matter morphology features, and composition features.

[0034] By operating S150, time-frequency analysis is performed on the pressure signal data to obtain pressure characteristics.

[0035] In operation S160, the injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics are input into the wear diagnosis model to obtain the wear diagnosis results, which include the worn parts and the wear values ​​of the worn parts.

[0036] For example, various injection parameter data can be obtained from the electronic control unit of the diesel engine. The injection parameter data may include data such as injection timing, injection pulse width, and injection quantity.

[0037] Pressure signal data can be collected using fuel pressure sensors configured in high-pressure fuel lines, common rails, and low-pressure fuel lines. The fuel pressure sensor can adaptively adjust detection parameters such as sampling step size based on diesel engine load and speed. The detection range of the fuel pressure sensor should be no less than 0~300 MPa, and the accuracy should reach ±0.1%.

[0038] The electrostatic deposition sampler uses an electric field strength of at least 5 kV / cm and minimizes particle overlap. The microscopic imaging device should have a resolution of 0.1 micrometers or higher to ensure accurate capture of particle morphology, and the camera frame rate should be able to reach 5000 fps to ensure accurate capture of particle images and avoid motion blur.

[0039] For example, image recognition technology can be used to calculate and identify data such as the equivalent volume diameter, equivalent surface area diameter, equivalent area diameter, inner diameter, minimum circumscribed diameter, average diameter, perimeter, and projected area of ​​particulate matter, thereby obtaining the morphological data of particulate matter.

[0040] Laser-induced breakdown spectroscopy (LAS) generates plasma by focusing an ultrashort pulse laser to ablate the surface of particulate matter, thereby emitting a characteristic spectrum. By analyzing the peak positions and intensities of the characteristic spectrum, the types of elements in the particulate matter can be identified and their contents calculated, thus obtaining the composition data of the particulate matter.

[0041] For example, after collecting multi-source data such as fuel injection parameter data, pressure signal data, particulate matter morphology data and composition data, the multi-source data can be processed synchronously. Kalman filtering is used to eliminate sensor noise, the timestamps of the multi-source data are aligned, and the error is ensured to be less than 0.1 ms.

[0042] For example, the injection parameter data, pressure signal data, particulate matter morphology data, and composition data can be normalized before feature extraction or analysis operations are performed.

[0043] For example, a wear diagnostic model is used to identify worn components in a fuel system and determine the wear value of those components.

[0044] For example, sample injection parameter data, sample pressure signal data, sample particulate matter morphology data, and sample composition data of multiple diesel engines of the same or similar models under normal operation and common fault conditions can be collected. Based on the sample injection parameter data, sample pressure signal data, sample particulate matter morphology data, and sample composition data, the initial wear diagnosis model can be trained to obtain a trained wear diagnosis model.

[0045] For example, the wear diagnosis model can employ a deep neural network (DNN) model to achieve deep fusion of multi-source data such as injection characteristics, pressure characteristics, particulate matter morphology characteristics, and composition characteristics. This disclosure is not limited to this; a hybrid model of GA (Genetic Algorithm)-DNN can also be used to improve model convergence speed and generalization ability while ensuring model accuracy.

[0046] For example, GA-DNN includes an algorithmic flow consisting of initializing the population, constructing the fitness function, and outputting the optimal DNN structure. The initial population size is 100, the crossover rate is 0.8, and the mutation rate is 0.01. The fitness function includes a weighted average of the F1 score and the model's convergence speed. The DNN structure employs a 4-layer hidden layer design with 256-128-64-32 nodes.

[0047] According to embodiments of this disclosure, by collecting particulate matter morphology and composition data of exhaust gas directly related to the operating state of the fuel system, as well as collecting fuel injection parameter data and pressure signal data, and extracting injection characteristics, pressure characteristics, and particulate matter morphology and composition characteristics as inputs to a wear diagnosis model, wear diagnosis of the fuel system can be achieved. Since particulate matter morphology and composition data can enhance the direct correlation with fuel system wear, combined with basic collected fuel injection parameter data and pressure signal data, the comprehensiveness and accuracy of wear diagnosis can be improved.

[0048] According to embodiments of this disclosure, the morphological data of particulate matter includes geometric morphology data and particle size distribution data. Feature extraction from the morphological data of particulate matter to obtain morphological features may include: extracting features from the geometric morphology data of the particulate matter to obtain the roundness and aspect ratio of the particulate matter; extracting features from the particle size distribution data of the particulate matter to obtain the fractal dimension of the particulate matter; and obtaining the morphological features of the particulate matter based on the roundness, aspect ratio, and fractal dimension.

[0049] Circularity measures how close a particle is to a perfect circle, with a value ranging from 0 to 1. The projected area and perimeter of the particle can be extracted from its geometric morphology data, and the circularity can be determined based on these parameters.

[0050] The aspect ratio is the ratio of the longest dimension to the shortest dimension of a particle, used to describe the degree of elongation of the particle. The larger the value, the more slender the particle; a value close to 1 indicates that the particle is close to spherical or square.

[0051] Fractal dimension is used to quantify the complexity and compactness of particle aggregate structures. A higher fractal dimension value indicates a denser, more compact structure; a lower value indicates a looser structure. The fractal dimension can be obtained by dividing an image into grids of different sizes using box counting and statistically analyzing the relationship between the number of grid cells containing particle pixels and the grid size.

[0052] According to embodiments of this disclosure, features such as roundness, aspect ratio, and fractal dimension can provide a unique perspective for wear diagnosis from the level of combustion chemistry, improve the model's ability to distinguish fault types, and provide richer evidence for assessing the degree of wear.

[0053] For example, poor atomization caused by fuel injector wear can produce larger droplets, which may form more irregular, low-sphericity carbon agglomerates after combustion. Cavitation caused by high-pressure fuel pump wear can directly strip away metal particles, producing highly spherical, small-sized spherical metal particles. Changes in fractal dimension can reflect whether the overall distribution is a single component failure or a complex failure of multiple components.

[0054] According to embodiments of this disclosure, feature extraction of particulate matter composition data to obtain the compositional features of particulate matter may include: extracting the features of metal elements from the particulate matter composition data, determining the content of each metal element and the target content ratio, wherein the target content ratio represents the ratio of the contents of two target metal elements; and obtaining the compositional features of particulate matter based on the content of each metal element and the target content ratio.

[0055] For example, the metallic elements may include copper, zinc, nickel, and other metallic elements. An abnormal increase in the content of metallic elements is directly related to component wear. For example, copper levels exceeding a preset threshold indicate wear on the fuel injector; the preset threshold for copper could be, for example, 12 ppb. Zinc levels exceeding a preset threshold indicate wear on the high-pressure fuel pump; the preset threshold for zinc could be, for example, 8 ppb.

[0056] The target content ratio can be, for example, the ratio of copper to zinc, but this disclosure is not limited to this. The target content ratio can also be selected based on the main constituent elements of each component of the fuel injection system. Different target content ratios can reflect the degree of wear of different components.

[0057] According to embodiments of this disclosure, by extracting the content of metal elements and the target content ratio, the component where wear occurs can be indicated. The element ratio can effectively distinguish coexisting wear components and quantify the wear ratio of different components, significantly improving the accuracy and reliability of the wear diagnosis model in locating specific wear components in complex fault scenarios.

[0058] According to embodiments of this disclosure, performing time-frequency analysis on pressure signal data to obtain pressure characteristics may include: using wavelet packets to perform multiple iterative decompositions on the pressure signal data to obtain multiple sub-band signals in different frequency ranges; determining the energy proportion of each sub-band signal based on the energy of each sub-band signal; determining the energy entropy and the target sub-band signal representing the main frequency based on the energy proportion of each sub-band signal; and obtaining the pressure characteristics based on the energy entropy and the target sub-band signal.

[0059] For example, after eliminating pressure sensor noise using Kalman filtering, wavelet transform can be used to extract frequency domain features and perform frequency domain decomposition of the pressure signal.

[0060] Wavelet packet decomposition is a method for multi-scale frequency domain analysis of signals. With each additional iteration of the wavelet packet, the number of frequency bands doubles. When decomposing to the Nth layer, a total of 2n ... N Sub-bands. For example, using wavelet packet decomposition to the 5th level, 16 sub-band signals can be extracted. Each sub-band signal corresponds to a potential physical process or event in the diesel engine fuel system. For example, the low-frequency band may correspond to the reciprocating frequency of the high-pressure fuel pump plunger, and the mid-frequency band may correspond to the pressure oscillation mode of the common rail, etc.

[0061] The capability of a sub-band signal can be determined based on its amplitude, and then the percentage of energy in each sub-band can be calculated. When the fuel system is in a healthy state, energy is highly concentrated in a few main frequency bands associated with normal cyclical operation. However, wear or malfunctions can cause new abnormal vibrations or alter existing oscillation patterns.

[0062] Energy entropy is a measure of the degree of disorder in a fuel system. Low energy entropy indicates relatively regular pressure fluctuations and a stable fuel system. High energy entropy indicates chaotic pressure fluctuations, a disordered fuel system, and potential wear or malfunction. Dominant frequency represents the dominant frequency component in pressure fluctuations; in cases of wear in the fuel system, this dominant frequency may shift. It can serve as one of the main indicators for locating faults.

[0063] According to embodiments of this disclosure, the energy proportion of sub-bands is obtained by multi-level decomposition of pressure signal data, thereby obtaining energy entropy and dominant frequency. Energy entropy can quantify the complexity and disorder of pressure fluctuations as a whole, while dominant frequency characteristics can locate the frequency components dominated by specific mechanical action anomalies and their energy attenuation or shift. This enables more precise differentiation of wear types and earlier, more sensitive detection of wear degree.

[0064] According to embodiments of this disclosure, the fuel system includes multiple cylinder blocks. Extracting features from the injection parameter data to obtain injection features may include: for each cylinder block, performing statistical feature extraction on the injection parameter data to obtain injection statistical features of the cylinder block; extracting features from the differences in injection parameter data between multiple injections to obtain injection stability features of the cylinder block; extracting features from the differences in injection parameter data between the multiple cylinder blocks to obtain injection difference features between the multiple cylinder blocks; and obtaining injection features based on the injection statistical features of any cylinder block, the injection difference features between the multiple cylinder blocks, and the injection stability features of any cylinder block.

[0065] For example, the fuel injection parameter data may include parameters such as fuel injection timing, fuel injection pulse width, and fuel injection quantity.

[0066] The injection statistics of the cylinder block can include the average value and variance of various parameters such as injection timing, injection pulse width and injection quantity. The injection statistics can reflect the basic working state of the injection system. For example, if the average value of a cylinder deviates significantly from the set value, it may indicate that there is a flow characteristic deviation of its injector or a leak in the high-pressure oil circuit.

[0067] The injection stability characteristics of a cylinder block can include features such as the standard deviation and coefficient of variation of the injection quantity between cycles, which can reflect the consistency of the working state of a single cylinder block. For example, wear or poor sealing inside the injector can lead to inconsistent injections each time.

[0068] Injection variation characteristics between multiple cylinder blocks can include, for example, the standard deviation and range of the average injection quantity per cylinder, which can be used to diagnose multi-cylinder imbalance. Fuel system wear usually does not occur synchronously and uniformly; injection variation characteristics can be used to locate the cylinder block where the fault has occurred.

[0069] According to embodiments of this disclosure, by separating and extracting in-cylinder statistical features, in-cylinder stability features, and inter-cylinder difference features, a three-dimensional diagnostic system for capturing wear within the cylinder and inter-cylinder coordination imbalance can be established. This not only helps to detect system-level fuel supply anomalies but also locates specific faulty cylinders, improving diagnostic accuracy.

[0070] According to embodiments of this disclosure, inputting injection features, pressure features, particulate morphology features, and composition features into a wear diagnosis model to obtain wear diagnosis results may include: inputting injection features, pressure features, particulate morphology features, and composition features into a feature fusion module of the wear diagnosis model to obtain fused features; inputting the fused features into a wear classification module of the wear diagnosis model to obtain worn components; and inputting the fused features into a wear value determination module of the wear diagnosis model to obtain the wear value of the worn components.

[0071] For example, the feature fusion module can perform nonlinear transformation and deep fusion on the fuel injection features, pressure features, particulate matter morphology features and composition features to obtain a high-dimensional, dense comprehensive feature vector.

[0072] The wear diagnosis model can include two task modules: a wear classification module and a wear value determination module. The wear classification module is used to learn the mapping function between the morphological, compositional, pressure, and injection characteristics of particulate matter and the faults of specific components. The wear value determination module is used to learn the mapping function between the morphological, compositional, pressure, and injection characteristics of particulate matter and the specific wear values.

[0073] According to embodiments of this disclosure, by using shared fusion features and multiple task modules, the diagnostic capabilities of the model are improved while achieving efficient model learning.

[0074] Figure 2 A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to another embodiment of the present disclosure is shown.

[0075] like Figure 2As shown, for each cylinder block, statistical features are extracted from the injection parameter data 201 of the cylinder block to obtain the injection statistical features 204 of the cylinder block; features are extracted from the differences in the injection parameter data 201 of the cylinder block between multiple injections to obtain the injection stability features 206 of the cylinder block; features are extracted from the differences in the injection parameter data 201 of multiple cylinder blocks to obtain the injection difference features 205 between multiple cylinder blocks; based on the injection statistical features 204 of any cylinder block, the injection difference features 205 between multiple cylinder blocks and the injection stability features 206 of any cylinder block, the injection feature 217 is obtained.

[0076] Feature extraction is performed on the geometric morphology data in the particulate matter morphology data 202 to obtain the roundness 207 and aspect ratio 208 of the particulate matter; feature extraction is performed on the particle size distribution data in the particulate matter morphology data 202 to obtain the fractal dimension 209 of the particulate matter; based on the roundness 207, aspect ratio 208 and fractal dimension 209, the morphological features 214 of the particulate matter are obtained.

[0077] The composition data 203 of particulate matter is used to extract the characteristics of metal elements, and the content 210 and target content ratio 211 of each metal element are determined. The target content ratio 211 represents the ratio of the contents of two target metal elements. Based on the content 210 and target content ratio 211 of each metal element, the composition characteristics 215 of the particulate matter are obtained.

[0078] The pressure signal data 204 is iteratively decomposed multiple times using wavelet packets to obtain multiple sub-band signals in different frequency ranges; the energy ratio of each sub-band signal is determined based on the energy of each sub-band signal; the energy entropy 212 and the target sub-band signal 213 representing the main frequency are determined based on the energy ratio of each sub-band signal; the pressure feature 216 is obtained based on the energy entropy 212 and the target sub-band signal 213.

[0079] The oil injection feature 217, the particulate morphology feature 214, the particulate composition feature 215, and the pressure feature 216 are input into the wear diagnosis model M218 to obtain the wear diagnosis result 219.

[0080] According to embodiments of this disclosure, the wear value determination module includes a wear value determination submodule, and the worn components include fuel injectors and high-pressure fuel pumps. The module that inputs fused features into the wear diagnosis model to obtain wear values ​​may include: when the worn component is a fuel injector or a high-pressure fuel pump, inputting the fused features into the wear value determination submodule to obtain the wear value of the fuel injector or the wear value of the high-pressure fuel pump.

[0081] For example, the wear classification module can output the probability of wear for each component. If the wear probability of a certain component is significantly higher than the wear probability of other components, it can be determined that only that component is worn. For example, if the probability of fuel injector wear is 0.85 and the probability of high-pressure pump wear is 0.10, then it can be determined that the fuel injector is worn. In the case of wear of a single component, the wear value of the sub-wear component can be directly determined by the wear value.

[0082] If the probability distribution output by the wear classification module shows that both components have high wear probabilities, it indicates that the wear is a combination of wear on both components. A probability threshold for wear on each component can be pre-configured. When the wear probabilities of two components are both greater than the threshold, it indicates that combination wear has occurred. For example, if the probability of injector wear is 0.55 and the probability of high-pressure pump wear is 0.45, it can be determined as combination wear of the injector and high-pressure pump.

[0083] According to embodiments of this disclosure, the wear value determination module includes a wear ratio determination submodule. The module that inputs the fused features into the wear diagnosis model to obtain the wear value may include: in the case of combined wear of the fuel injector and the high-pressure pump, inputting the content of each metal element in the particulate matter and the target content ratio into the wear ratio determination submodule to obtain the wear ratio of the fuel injector and the high-pressure pump; and inputting the wear ratio and the fused features into the wear value determination submodule to obtain the wear value of the fuel injector and the high-pressure pump respectively.

[0084] The wear ratio determination submodule is used to learn the functional mapping relationship between the metal element content, the target metal element ratio and the wear ratio between components.

[0085] For example, after inputting the wear ratio and fusion feature into the wear value determination submodule, the total wear value can be obtained first based on the fusion feature, and then the wear values ​​of the fuel injector and high-pressure fuel pump can be determined according to the total wear value and the wear ratio.

[0086] According to embodiments of this disclosure, under combined wear, by first determining the wear ratio and then further determining the wear values ​​of the fuel injector and high-pressure fuel pump based on the wear ratio and the comprehensive vector, the accuracy of wear diagnosis of each component under multi-component combined wear can be improved.

[0087] According to embodiments of this disclosure, after obtaining the wear values ​​of the fuel injector and the high-pressure fuel pump, the method may further include: inputting the fusion features, the worn component, and the wear value of the worn component into the remaining time prediction module of the wear diagnosis model to obtain the remaining usage time of the high-pressure fuel pump and the fuel injector.

[0088] For example, the remaining time prediction module can use a Bayesian network to predict the remaining service time of the high-pressure oil pump and the injector based on fused features, worn parts, and the wear value of the worn parts.

[0089] For example, a remaining time threshold can be used to trigger a shutdown command when the remaining usage events of the high-pressure oil pump and injector are below the threshold.

[0090] Figure 3 A flowchart illustrating a method for diagnosing fuel system wear in a diesel engine according to yet another embodiment of the present disclosure is shown.

[0091] like Figure 3 As shown, the oil injection feature 310, pressure feature 340, particulate morphology feature 320 and particulate composition feature 330 are input into the feature fusion module M351 of the wear diagnosis model M350 to obtain the fused features; the fused features are input into the wear classification module M352 of the wear diagnosis model to obtain the worn component 360.

[0092] When the worn components are fuel injectors or high-pressure fuel pumps, the fusion feature is input into the wear value determination submodule M354 to obtain the wear value of the fuel injector or the wear value 370 of the high-pressure fuel pump. When the worn components are a combination of fuel injector and high-pressure pump wear, the content of each metal element in the particulate matter and the target content ratio are input into the wear ratio determination submodule M353 to obtain the wear ratio of the fuel injector and the high-pressure fuel pump; the wear ratio and the fusion feature are input into the wear value determination submodule M354 to obtain the wear value 370 of the fuel injector and the high-pressure fuel pump respectively.

[0093] The training process of the wear diagnosis model is described in further detail below.

[0094] During the model pre-learning phase, over 15 days of normal sample data and over 15 days of abnormal sample data for common fault conditions (including injector wear faults and high-pressure fuel pump faults) were collected for diesel engine models and similar models. Both normal and abnormal sample data included sample injection parameter data, sample pressure signal data, particulate matter morphology data, and sample composition data. Sample composition data included metallic elements such as copper, zinc, and nickel. Under fault conditions, the content of a single metallic element should be above the detection limit of 0.5 ppb; the morphology of particulate matter should be at least 0.1 μm higher than the resolution; and the maximum internal pressure of the high-pressure fuel line should be kept below 300 MPa as much as possible. Particulate matter morphology data included the equivalent volume diameter, equivalent surface area diameter, equivalent area diameter, inner diameter, minimum circumscribed diameter, average diameter, perimeter, projected area, original particulate matter images, and original data such as the pressure-time distribution inside the high-pressure fuel line, as well as particulate matter roundness, fractal dimension, and aspect ratio. Feature extraction was performed on sample injection parameter data, sample pressure signal data, particulate matter morphology data, and sample composition data to obtain injection features, pressure features, and particulate matter morphology and composition features. These features were then used to train an initial wear diagnosis model. The wear classification module of the wear diagnosis model learned the specific correspondence between particulate matter morphology features, metal element ratios, pressure signals, and specific faults. The wear value determination module of the wear diagnosis model learned the relationship between the copper / zinc ratio of the particulate matter, the wear ratio of the injector coating and the high-pressure pump bearing, and the wear value, thus obtaining the wear diagnosis model. Wear alarm thresholds can be set, and sample data under different operating conditions can be collected to train the initial wear diagnosis model, resulting in wear alarm thresholds that are dynamically adjusted according to operating conditions.

[0095] During the model validation phase, diesel engine exhaust particulate matter was continuously monitored for 15 days, and sample test data were collected under different engine conditions. Thresholds of 12 ppb for copper and 8 ppb for zinc were used. If the concentrations of copper or zinc in the particulate matter reached or exceeded these thresholds, a fault condition was immediately reported, and testing was stopped. Specifically, excessive copper concentrations reported fuel injector wear faults, and excessive zinc concentrations reported high-pressure fuel pump faults. During monitoring, the electrostatic deposition sampler, laser-induced breakdown spectrometer, and microscope imaging device should be preheated 1 hour in advance, and the electric field strength of the electrostatic deposition sampler should be adjusted to above 5 kV / cm, and the frame rate of the high-speed camera should be adjusted to above 5000 fps. Particulate matter sampling was carried out using the electrostatic deposition sampler.

[0096] By using a laser-induced breakdown spectrometer and its built-in spectrometer to automatically analyze the emission spectrum of carbon soot plasma, the metal elements contained in the exhaust particulate matter are qualitatively detected and quantitatively analyzed. The qualitative and quantitative data are stored in the data storage and analysis equipment. At the same time, data such as the copper / zinc ratio are calculated. The wear ratio of the fuel injector coating and the high-pressure pump bearing is clearly identified at the qualitative level, and a preliminary conclusion is given on the wear status. The results are stored in the data storage and analysis equipment, and warnings and prompts are given on computer and other output devices.

[0097] The morphology of particulate matter is scanned using a microscopic imaging device, and the original images are stored in a data storage and analysis device. Simultaneously, based on image recognition software, the equivalent volume diameter, equivalent surface area diameter, equivalent area diameter, inner diameter, minimum circumscribed diameter, average diameter, perimeter, and projected area of ​​the particulate matter are calculated and identified. Feature extraction / dimensionality reduction parameters such as particulate roundness, fractal dimension, and aspect ratio are calculated and stored in the data storage and analysis device, and linked to the basic data and feature extraction / dimensionality reduction data collected by the data acquisition module in the database. The system automatically identifies the degree of motion blur in the particulate matter images and automatically deletes images with high blurriness. A limit of 1.5 for the fractal dimension of the particulate matter is set; if the real-time value exceeds the threshold, a warning and alert are issued on the computer or other output devices.

[0098] Pressure fluctuations in the high-pressure fuel line are measured using a fuel pressure sensor. The sampling step size and other detection parameters are adaptively adjusted based on changes in load and engine speed. Combined with manual adjustments to operating conditions and engine speed, the pressure range is ensured to remain within 300 MPa. Data is stored in a data storage and analysis device and correlated with other basic data. After eliminating pressure sensor noise using Kalman filtering, wavelet transform is used to extract frequency domain features, and the pressure signal is decomposed in the frequency domain. Wavelet packet decomposition is performed to the 5th level, extracting the energy proportion of 16 sub-frequency bands. The pressure fluctuation entropy value is then calculated, identifying the dominant frequency and energy entropy, which are stored in the data storage and analysis device. This data is then linked to the basic data collected by the data acquisition module and the feature extraction / dimensionality reduction data in the database. An entropy value > 2.5 is used as a threshold; if the real-time value exceeds the threshold, a warning and alert are issued on the computer or other output devices.

[0099] The data acquisition module automatically synchronizes multimodal signals, aligns timestamps according to database data association, and normalizes sample test data using a combination of Z-score standardization and Min-Max scaling. The data is then stored in a data storage and analysis device and associated with the basic data acquired by the data acquisition module.

[0100] Feature extraction processing is performed on the sample test data to obtain test injection characteristics, test pressure characteristics, particulate matter test morphology characteristics, and test composition characteristics. These characteristics are then input into the wear diagnosis model. The model outputs the wear layout and predicted wear conditions for each component (e.g., nozzle orifice diameter changes). Wear values ​​are graded into three levels: normal, slight, and severe. Based on practical experience, shutdown conditions are determined; for example, if the high-pressure oil pump plunger wear reaches 0.02 mm, a shutdown command is triggered.

[0101] A Bayesian network lifetime prediction system is employed, updating the remaining lifetime probability based on historical data and real-time features. A failure probability threshold of 0.95 is set, triggering a shutdown command when the failure probability threshold exceeds 0.95. During operation, the adaptive threshold alarm model is continuously adjusted and optimized by combining pre-learning data, historical data, and manual fault observations. Five-fold cross-validation is performed to ensure an accuracy rate of 96% and an AUC value of over 0.98 in ROC curve analysis. If the accuracy rate consistently fails to meet these standards, a shutdown command is triggered. After 15 days of operation, disassembly inspection and continuous testing are conducted to verify the threshold alarm and optimal maintenance time prediction results, ensuring a confidence level higher than 90%. The errors between the actual and predicted values ​​of the disassembly inspection of key components in the fuel system are compared, ensuring a relative error of less than 5%. The test and verification results are used as pre-learning data for subsequent tests.

[0102] According to embodiments of this disclosure, by integrating the data hierarchy of diesel engine exhaust particulate matter physicochemical properties, morphological characteristics, and fuel system operating parameters, multi-dimensional data utilization and in-depth data mining are achieved. Real-time in-situ analysis of metallic elements is realized through spectral detection. This results in highly sensitive diagnostics, ensuring that the fuel injector orifice diameter change detection limit reaches 2 μm and the accuracy exceeds 97%.

[0103] Figure 4 A block diagram of a system for diagnosing fuel system wear in a diesel engine according to an embodiment of the present disclosure is shown schematically.

[0104] The system includes an acquisition module 410, a first data extraction module 420, a second data extraction module 430, a feature extraction module 440, an analysis module 450, and a wear diagnosis module 460.

[0105] The acquisition module 410 is used to acquire fuel injection parameter data and pressure signal data in the fuel line during the fuel injection and combustion process.

[0106] The first data extraction module 420 is used to collect particulate matter from exhaust gas during fuel injection combustion using an electrostatic deposition sampler, and to capture images of the collected particulate matter using a microscopic camera to extract particulate matter morphology data from the images.

[0107] The second data extraction module 430 is used to emit pulsed lasers to the collected particulate matter using a laser-induced breakdown spectrometer to excite the plasma in the particulate matter to emit characteristic spectra, so as to obtain the composition data of the particulate matter based on the characteristic spectra.

[0108] The feature extraction module 440 extracts features from the fuel injection parameter data, particulate matter morphology data, and composition data to obtain fuel injection features, particulate matter morphology features, and composition features.

[0109] Analysis module 450 is used to perform time-frequency analysis on pressure signal data to obtain pressure characteristics.

[0110] The wear diagnosis module 460 is used to input the injection characteristics, pressure characteristics, particulate morphology characteristics and composition characteristics into the wear diagnosis model to obtain wear diagnosis results, which include the worn parts and the wear values ​​of the worn parts.

[0111] According to embodiments of this disclosure, the particulate matter morphology data includes geometric morphology data and particle size distribution data. The feature extraction module includes a first feature extraction submodule, a second feature extraction submodule, and a first determination submodule.

[0112] The first feature extraction submodule is used to extract features from the geometric morphology data of particles to obtain the roundness and aspect ratio of the particles.

[0113] The second feature extraction submodule is used to extract features from the particle size distribution data of particulate matter to obtain the fractal dimension of the particulate matter.

[0114] The first determination submodule is used to obtain the morphological characteristics of particles based on roundness, aspect ratio, and fractal dimension.

[0115] According to embodiments of this disclosure, the feature extraction module includes a third feature extraction submodule and a second determination submodule.

[0116] The third feature extraction submodule is used to extract the features of metal elements from the composition data of particulate matter, determine the content of each metal element and the target content ratio, where the target content ratio represents the ratio of the contents of the two target metal elements.

[0117] The second determining submodule is used to obtain the compositional characteristics of particulate matter based on the content of each metal element and the target content ratio.

[0118] According to an implementation of this disclosure, the feature extraction module includes a decomposition submodule, a third determination submodule, and a fourth determination submodule.

[0119] The decomposition submodule is used to perform multiple iterative decompositions on the pressure signal data using wavelet packets to obtain multiple sub-band signals in different frequency ranges.

[0120] The third determining submodule is used to determine the energy ratio of each sub-band signal based on the energy of each sub-band signal.

[0121] The fourth determination submodule is used to determine the energy entropy and the target sub-band signal characterizing the main frequency based on the energy proportion of each sub-band signal.

[0122] The fifth determination submodule is used to obtain pressure characteristics based on energy entropy and target sub-band signals.

[0123] According to embodiments of this disclosure, the fuel system includes multiple cylinder blocks; the feature extraction module includes:

[0124] The fourth feature extraction submodule is used to extract statistical features from the injection parameter data of each cylinder block to obtain the injection statistical features of the cylinder block.

[0125] The fifth feature extraction submodule is used to extract features from the differences in the injection parameters of the cylinder block between multiple injections, so as to obtain the injection stability features of the cylinder block.

[0126] The sixth feature extraction submodule is used to extract features from the differences in the fuel injection parameter data of multiple cylinder blocks to obtain the fuel injection difference features between multiple cylinder blocks.

[0127] The sixth determination submodule is used to obtain the injection characteristics based on the injection statistics of any cylinder block, the injection difference characteristics between multiple cylinder blocks, and the injection stability characteristics of any cylinder block.

[0128] According to embodiments of this disclosure, the wear diagnosis module includes an input module, a first diagnosis submodule, and a second diagnosis submodule.

[0129] The input module is used to input the injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics into the feature fusion module of the wear diagnosis model to obtain fused features.

[0130] The first diagnostic submodule is used to input the fused features into the wear classification module of the wear diagnostic model to obtain the worn parts.

[0131] The second diagnostic submodule is used to input the fused features into the wear value determination module of the wear diagnostic model to obtain the wear value of the worn component.

[0132] According to embodiments of this disclosure, the wear value determination module includes a wear value determination submodule, and the worn components include a fuel injector and a high-pressure fuel pump; the second diagnostic submodule includes a first diagnostic unit.

[0133] The first diagnostic unit is used to input the fusion features into the wear value determination submodule when the worn component is a fuel injector or a high-pressure fuel pump, so as to obtain the wear value of the fuel injector or the wear value of the high-pressure fuel pump.

[0134] According to embodiments of this disclosure, the wear value determination module includes a wear ratio determination submodule, and the worn components include combined wear of the injector and the high-pressure pump; the second diagnostic submodule includes:

[0135] The second diagnostic unit is used to input the content and target content ratio of each metal element in the particulate matter into the wear ratio determination submodule when the worn parts are a combination of fuel injector and high-pressure pump wear, so as to obtain the wear ratio of fuel injector and high-pressure pump.

[0136] The third diagnostic unit is used to input the wear ratio and fusion characteristics into the wear value determination submodule to obtain the wear values ​​of the fuel injector and the high-pressure fuel pump.

[0137] According to embodiments of this disclosure, the system further includes a third diagnostic module.

[0138] The third diagnostic module is used to input the fusion features, worn parts, and wear values ​​of the worn parts into the remaining time prediction module of the wear diagnostic model to obtain the remaining service time of the high-pressure oil pump and the injector.

[0139] Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure, or at least part of the functions of any one or more of them, can be implemented in one module. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be implemented by dividing them into multiple modules. Any one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as hardware circuitry, such as a Field-Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-a-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, one or more of the modules, submodules, units, and subunits according to embodiments of the present disclosure can be at least partially implemented as computer program modules, which, when run, can perform corresponding functions.

[0140] For example, any and more of the acquisition module 410, the first data extraction module 420, the second data extraction module 430, the feature extraction module 440, the analysis module 450, and the wear diagnosis module 460 can be combined into one module / unit / subunit, or any one of these modules / units / subunits can be split into multiple modules / units / subunits. Alternatively, at least some of the functions of one or more of these modules / units / subunits can be combined with at least some of the functions of other modules / units / subunits and implemented in one module / unit / subunit.

[0141] According to embodiments of this disclosure, at least one of the acquisition module 410, the first data extraction module 420, the second data extraction module 430, the feature extraction module 440, the analysis module 450, and the wear diagnosis module 460 can be at least partially implemented as hardware circuits, such as field-programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits (ASICs), or any other reasonable means of integrating or packaging circuits, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the acquisition module 410, the first data extraction module 420, the second data extraction module 430, the feature extraction module 440, the analysis module 450, and the wear diagnosis module 460 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0142] It should be noted that the stress analysis device part in the embodiments of this disclosure corresponds to the stress analysis method part in the embodiments of this disclosure. The specific description of the stress analysis device part is referred to in the stress analysis method part, and will not be repeated here.

[0143] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this disclosure is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. A method for diagnosing wear in the fuel system of a diesel engine, characterized in that, The method includes: Acquire the fuel injection parameter data and pressure signal data in the fuel line during the fuel injection and combustion process of the fuel system; The particulate matter in the exhaust gas during the fuel injection combustion process is collected using an electrostatic deposition sampler, and images of the particulate matter are captured using a microscopic imaging device to extract the morphological data of the particulate matter from the images; A laser-induced breakdown spectrometer is used to emit pulsed laser light onto the particulate matter to excite the plasma emission characteristic spectrum in the particulate matter, so as to obtain the composition data of the particulate matter based on the characteristic spectrum; Feature extraction is performed on the fuel injection parameter data, the morphology data of the particulate matter, and the composition data of the particulate matter to obtain fuel injection features, morphology features, and composition features of the particulate matter. Time-frequency analysis was performed on the pressure signal data to obtain pressure characteristics; The oil injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics are input into the wear diagnosis model to obtain wear diagnosis results, which include worn parts and wear values ​​of worn parts.

2. The method according to claim 1, characterized in that, The morphological data of the particles includes geometric morphology data and particle size distribution data; Feature extraction is performed on the morphological data of the particulate matter to obtain the morphological features of the particulate matter, including: Feature extraction is performed on the geometric morphology data of the particles to obtain the roundness and aspect ratio of the particles; Feature extraction is performed on the particle size distribution data of the particulate matter to obtain the fractal dimension of the particulate matter; Based on the circularity, aspect ratio, and fractal dimension, the morphological characteristics of the particles are obtained.

3. The method according to claim 1, characterized in that, Feature extraction is performed on the composition data of the particulate matter to obtain the compositional features of the particulate matter, including: The composition data of the particulate matter is subjected to feature extraction of metal elements to determine the content of each metal element and the target content ratio, wherein the target content ratio represents the ratio of the contents of two target metal elements. The compositional characteristics of the particulate matter are obtained based on the content of each metal element and the target content ratio.

4. The method according to claim 1, characterized in that, Time-frequency analysis is performed on the pressure signal data to obtain pressure characteristics, including: The pressure signal data is iteratively decomposed multiple times using wavelet packets to obtain multiple sub-band signals in different frequency ranges; The energy percentage of each sub-band signal is determined based on the energy of each sub-band signal. Based on the energy proportion of each sub-band signal, determine the energy entropy and the target sub-band signal characterizing the main frequency; The pressure characteristics are obtained based on the energy entropy and the target sub-band signal.

5. The method according to claim 1, characterized in that, The fuel system includes multiple cylinders; feature extraction is performed on the injection parameter data to obtain injection features, including: For each cylinder block, statistical features are extracted from the injection parameter data of the cylinder block to obtain the injection statistical features of the cylinder block; The differences in the fuel injection parameter data of the cylinder block between multiple injections are extracted to obtain the fuel injection stability characteristics of the cylinder block. Feature extraction is performed on the differences in fuel injection parameter data between the multiple cylinder blocks to obtain fuel injection difference features between the multiple cylinder blocks; The injection characteristics are obtained based on the injection statistics of any one of the cylinder blocks, the injection difference characteristics among multiple cylinder blocks, and the injection stability characteristics of any one of the cylinder blocks.

6. The method according to any one of claims 1 to 5, characterized in that, The injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics are input into the wear diagnosis model to obtain wear diagnosis results, including: The injection characteristics, pressure characteristics, particulate morphology characteristics, and composition characteristics are input into the feature fusion module of the wear diagnosis model to obtain fused features. The fused features are input into the wear classification module of the wear diagnosis model to obtain the worn component; The fused features are input into the wear value determination module of the wear diagnosis model to obtain the wear value of the worn component.

7. The method according to claim 6, characterized in that, The wear value determination module includes a wear value determination submodule, and the wear components include fuel injectors and high-pressure fuel pumps; The fused features are input into the module of the wear diagnosis model to obtain the wear value, including: When the worn component is a fuel injector or a high-pressure fuel pump, the fusion feature is input into the wear value determination submodule to obtain the wear value of the fuel injector or the wear value of the high-pressure fuel pump.

8. The method according to claim 7, characterized in that, The wear value determination module includes a wear ratio determination submodule, and the wear components include the combined wear of the fuel injector and the high-pressure pump; The fused features are input into the module of the wear diagnosis model to obtain the wear value, including: When the wear components are a combination of fuel injector and high-pressure pump wear, the content of each metal element in the particulate matter and the target content ratio are input into the wear ratio determination submodule to obtain the wear ratio of fuel injector and high-pressure pump. The wear ratio and the fusion feature are input into the wear value determination submodule to obtain the wear values ​​of the fuel injector and the high-pressure fuel pump respectively.

9. The method according to claim 6, characterized in that, The method further includes: The fusion features, the worn components, and the wear values ​​of the worn components are input into the remaining time prediction module of the wear diagnosis model to obtain the remaining service time of the high-pressure oil pump and the injector.

10. A system for diagnosing wear in the fuel system of a diesel engine, characterized in that, The system includes: The acquisition module is used to acquire the fuel injection parameter data and the pressure signal data in the fuel line during the fuel injection and combustion process of the fuel system. The first data extraction module is used to collect particulate matter in the exhaust gas during the fuel injection combustion process using an electrostatic deposition sampler, and to capture images of the collected particulate matter using a microscopic camera to extract the morphological data of the particulate matter from the images. The second data extraction module is used to emit pulsed lasers to the collected particulate matter using a laser-induced breakdown spectrometer to excite the plasma emission characteristic spectrum in the particulate matter, so as to obtain the composition data of the particulate matter based on the characteristic spectrum. The feature extraction module performs feature extraction on the fuel injection parameter data, the morphological data of the particulate matter, and the composition data of the particulate matter, respectively, to obtain fuel injection features, morphological features of the particulate matter, and composition features. The analysis module is used to perform time-frequency analysis on the pressure signal data to obtain pressure characteristics; The wear diagnosis module is used to input the oil injection characteristics, pressure characteristics, morphological characteristics and composition characteristics of the particles into the wear diagnosis model to obtain wear diagnosis results, which include the worn parts and the wear values ​​of the worn parts.

Images