A fault diagnosis method and system for a hydraulic component universal device

Abstract

This invention discloses a fault diagnosis method and system for general-purpose hydraulic components, relating to the field of hydraulic component fault diagnosis technology. The method includes the following steps: based on the hydraulic components, identifying the control channels of the general-purpose equipment; acquiring transient signals in real time within each control channel; introducing preset interference constraints; generating interference factors through a high-entropy random disturbance source; analyzing the amplitude, frequency, and waveform state disturbances of the transient signals to generate interference identifiers; forming a health state vector by combining the operating status; comparing the deviation of the health state vector from the health baseline feature map; if the deviation is less than a standard deviation threshold, executing a status assessment strategy in the first diagnostic channel to identify a stable health trend; if the deviation is greater than or equal to the standard deviation threshold, executing a fault diagnosis strategy in the second diagnostic channel to identify fault signs. This invention achieves efficient division of labor between health assessment and fault diagnosis, enabling accurate identification of fault signs.

Description

Technical Field

[0001] This invention relates to the field of hydraulic component fault diagnosis technology, specifically to a fault diagnosis method and system for general hydraulic component equipment. Background Technology

[0002] Hydraulic systems, including general-purpose equipment such as mechanical components and hydraulic oil, operate in closed loops. They lack the intuitive visibility of traditional machinery and cannot obtain real-time parameters through conventional testing methods like electrical equipment. In addition, the variability of the working environment and the complexity of the operating conditions further increase the difficulty of system monitoring and maintenance. Against this background, research on fault diagnosis technology for hydraulic systems has significant engineering value. Traditional hydraulic component diagnostic systems have very limited ability to diagnose early or nascent faults. Firstly, these systems focus on a single characteristic of transient signals, such as peak pressure or average speed, neglecting the synergistic correlation of multiple dimensions, including amplitude, frequency, and waveform. For example, judging the state solely by pressure amplitude may misinterpret normal amplitude fluctuations under high load as faults, or fail to detect faults due to overlooked resonance risks caused by frequency shifts. Secondly, in the early stages of a fault, its impact on the system's dynamic characteristics is extremely weak, and the resulting abnormal signals are often drowned out by strong background noise and normal operating condition fluctuations, resulting in a very low signal-to-noise ratio. Furthermore, for transient... Signal monitoring is usually passive. When interference is detected, simple filtering or empirical elimination methods are often used, which can easily lead to interference components from non-target components being mixed into the acquired transient signals. This reduces the effectiveness of subsequent signal analysis and makes it difficult to focus on the true operating state of hydraulic components. Finally, when the system faces multiple concurrent or coupled faults, the signal characteristics generated by different faults are superimposed. Traditional methods directly use a single channel for fault diagnosis, which can easily lead to excessive execution of fault diagnosis algorithms in healthy states, increasing the system's computing power consumption. In fault states, using state assessment strategies will result in delayed fault identification, making it difficult to balance the accuracy and efficiency of monitoring. Summary of the Invention

[0003] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a fault diagnosis method and system for general-purpose hydraulic components. It collects transient signals from each control channel in real time and integrates the operating status, accurately reflecting the dynamic response characteristics of hydraulic components under different operating conditions. By employing a design with different execution strategies under dual diagnostic channels, it achieves efficient division of labor between health assessment and fault diagnosis, avoiding the limitations of a single diagnostic mode. This provides targeted and highly efficient technical support for subsequent condition maintenance and fault handling of hydraulic components, effectively reducing maintenance costs and improving equipment reliability, thus solving the problems mentioned in the background technology.

[0004] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, this application provides a fault diagnosis method for general-purpose hydraulic component equipment, the method comprising: Based on hydraulic components, the control channels of general equipment are identified; transient signals are collected in real time in each control channel, preset interference constraints are introduced, interference factors are generated through high-entropy random disturbance sources, and the amplitude, frequency and waveform state disturbance of transient signals are analyzed to generate interference labels. Based on the operating status of hydraulic components and combined with interference indicators to form a health status vector, the degree of deviation between the health status vector and the health baseline feature map is compared, and a diagnostic channel is selected, including a first diagnostic channel and a second diagnostic channel: if the deviation is less than the standard deviation threshold, a status assessment strategy is executed in the first diagnostic channel to identify a stable health trend; if the deviation is greater than or equal to the standard deviation threshold, a fault diagnosis strategy is executed in the second diagnostic channel to identify fault signs.

[0005] Furthermore, the hydraulic components include at least one of a hydraulic pump, a hydraulic valve, and a hydraulic cylinder; the control channels include at least an electro-hydraulic proportional valve control channel and a hydraulic pump displacement control channel; and the transient signals include at least two signals selected from pressure, vibration, acoustic, and temperature sensors.

[0006] Furthermore, by introducing preset interference constraints and generating interference factors through high-entropy random disturbance sources, the amplitude, frequency, and waveform state disturbances of transient signals are analyzed to generate interference identifiers, including: Load preset interference constraints, including the first interference constraint, the second interference constraint, and the third interference constraint, and obtain the corresponding constraint range. Generate several types of interference factors, including the first interference factor, the second interference factor, and the third interference factor, through a high-entropy random disturbance source. Combine the corresponding transient signals collected in real time from each control channel to perform signal disturbance operations, including amplitude disturbance, frequency disturbance, and waveform disturbance. Based on the transient signal after disturbance, the amplitude, frequency, and waveform combination are extracted and fused into a first monitoring vector. The first monitoring vector is compared with the corresponding preset benchmark monitoring vector to obtain different deviations. Different weights are assigned to different deviations, and the first monitoring vector is modally decomposed to obtain the basic mode and the fluctuation mode. Mode pairs are constructed based on the basic mode and the fluctuation mode, and fused into a second monitoring vector through deviation, weight, and mode pairs. This second monitoring vector is then imported into a preset spatiotemporal change model to obtain the interference identifier.

[0007] Furthermore, the first monitoring vector mode is decomposed to obtain the fundamental mode and the wave mode, including: The first monitoring vector is initially decomposed using a variational mode decomposition algorithm to obtain several eigenmode functions; Based on the intrinsic mode function, the density, frequency, and waveform similarity of extreme points are extracted, and a feature space composed of the density, frequency, and waveform similarity of extreme points at the same time is constructed. Based on the feature space, a certain frequency is fixed, and the feature space is cut to obtain the first cross section. Based on the first cross section, the extreme points and waveform similarity within the cross section are extracted to form an initial sphere center set. The monitoring radius is preset, and the first cross section is spherically aggregated. The first mode candidate set falling within the sphere is screened, and the feature points falling outside the sphere are marked as the first undetermined mode. Based on the feature space, a certain waveform similarity is fixed, and the feature space is cut to obtain a second cross section. Based on the second cross section, the x-axis is preset as the frequency range and the y-axis as the extreme point density range. The rectangles of the second cross section are aggregated, and the candidate set of the second mode falling within the rectangle is filtered. The feature points falling outside the rectangle are marked as the second undetermined modes. The intersection of the first candidate mode set and the second undetermined mode set is taken as the basic mode; the intersection of the first undetermined mode set and the second candidate mode set is taken as the wave mode.

[0008] Furthermore, the health baseline feature map was obtained through an unsupervised clustering algorithm, which was based on the K-means seed clustering algorithm.

[0009] Furthermore, the state assessment strategy is implemented, including: Within the first diagnostic channel, the deviation levels continuously collected under the same operating conditions are divided into several time segments according to a fixed-length operating cycle window. Each time segment contains the deviation levels of multiple sampling points, and all deviation levels are less than the standard deviation threshold. Key trend features are extracted from time series segments to construct a high-dimensional feature matrix. The similarity between the high-dimensional feature matrix and the corresponding feature matrices of each time series segment is calculated and processed to form a similarity matrix. The high-dimensional feature matrix is ​​then compressed into a low-dimensional space to obtain a low-dimensional feature matrix. The low-dimensional feature matrix is ​​positively correlated with the similarity matrix. A loss function based on Gaussian distribution is constructed, and the low-dimensional distribution is optimized through gradient descent to identify the clustering state of healthy and stable trends.

[0010] Furthermore, implement fault diagnosis strategies, including: Within the second diagnostic channel, all health status vectors with deviations greater than or equal to the standard deviation threshold are acquired and aggregated into set H. Quantitative element groups for each health status vector in set H are extracted, and the maximum and minimum subsets for each quantitative element are constructed. The difference between the maximum and minimum subsets for each quantitative element is marked as the maximum span value of that element. Simultaneously, the standard deviation coefficients for each quantitative element in set H are acquired, and the product of the maximum span value and the standard deviation coefficient for each quantitative element is marked as the single-element diagnostic value. Using each single-element diagnostic value as an axis, a multidimensional curve is constructed, and the trajectory of the multidimensional curve moving with the time axis is obtained. Combining the shape of the trajectory in a three-dimensional cross-section, fault signs are identified.

[0011] Furthermore, by combining the trajectory's morphology in a three-dimensional cross-section, fault signs are identified, including: If the trajectory is spherical, it is determined to be without fault signs; if the trajectory is ellipsoidal, it is determined to be a minor fault sign; if the trajectory is conical, it is determined to be a moderate fault sign; if the trajectory is hyperboloidal, it is determined to be a severe fault sign; if the trajectory does not have any of the above shapes, a second identification is performed.

[0012] Secondly, this application provides a fault diagnosis system for general-purpose hydraulic component equipment, the system comprising a first module, a second module, and a third module: The first module is used to identify the control channels of general equipment based on hydraulic components; it collects transient signals in real time in each control channel, introduces preset interference constraints, generates interference factors through high-entropy random disturbance sources, analyzes the amplitude, frequency and waveform state disturbance of transient signals, and generates interference labels. The second module is used to determine the operating status of hydraulic components and combine them with interference indicators to form a health status vector. It compares the degree of deviation between the health status vector and the health baseline feature map and selects a diagnostic channel, including a first diagnostic channel and a second diagnostic channel. The third module: If the deviation is less than the standard deviation threshold, the status assessment strategy is executed in the first diagnostic channel to identify a healthy and stable trend; if the deviation is greater than or equal to the standard deviation threshold, the fault diagnosis strategy is executed in the second diagnostic channel to identify fault signs.

[0013] Thirdly, this application provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the fault diagnosis method for a general hydraulic component device provided in the first aspect above.

[0014] (III) Beneficial Effects This invention provides a fault diagnosis method and system for general-purpose hydraulic component equipment, which has the following beneficial effects: 1. This invention identifies the general equipment control channel corresponding to hydraulic components, generates interference factors by combining high-entropy random disturbance sources, and conducts transient signal disturbance analysis to generate interference identifiers. This effectively uncovers the potential correlation between transient signals and interference factors, providing data-driven decision-making basis for accurate assessment of the health status and fault diagnosis of hydraulic components. 2. This invention employs a dual-section hybrid clustering approach, using spherical clustering in the first section and rectangular clustering in the second section. This combination of spherical and rectangular clustering takes into account both the clustering of fundamental modes and the intervalicity of fluctuation modes. Finally, through intersection analysis of the clustering results, the identification of fundamental and fluctuation modes is obtained, thus solving the problem that it is difficult to distinguish modes with a single feature dimension. 3. This invention constructs a health status vector based on the operating status and interference indicators of hydraulic components. By comparing the degree of deviation from the health baseline feature map, it achieves adaptive selection of diagnostic channels. For different degrees of deviation, it executes status assessment or fault diagnosis strategies respectively. This not only ensures the accuracy of stable trend identification in the healthy state, but also improves the timeliness of fault sign capture in the fault state, thus greatly improving the accuracy of health diagnosis of hydraulic components. Attached Figure Description

[0015] Figure 1 This is a schematic diagram illustrating the steps of a fault diagnosis method according to an exemplary embodiment; Figure 2 This is a schematic diagram of a fault diagnosis system according to an exemplary embodiment. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of 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.

[0017] Example 1: This invention provides a fault diagnosis method for general hydraulic component equipment; Figure 1 This is a schematic diagram illustrating the steps of a fault diagnosis method according to an exemplary embodiment; please refer to... Figure 1 The method includes the following steps: S1: Based on hydraulic components, identify the control channels of general equipment; collect transient signals in real time within each control channel, introduce preset interference constraints, generate interference factors through high-entropy random disturbance sources, analyze the amplitude, frequency, and waveform state disturbance of transient signals, and generate interference labels; wherein, the hydraulic components include at least one of hydraulic pumps, hydraulic valves, and hydraulic cylinders, the control channels include, but are not limited to, electro-hydraulic proportional valve control channels and hydraulic pump displacement control channels, and the transient signals include two or more signals selected from pressure, vibration, acoustic, and temperature sensors; Preset interference constraints are introduced, and interference factors are generated through high-entropy random disturbance sources. The amplitude, frequency, and waveform state disturbance of transient signals are analyzed to generate interference labels. This includes: loading preset interference constraints, obtaining the constraint range, generating several types of interference factors through high-entropy random disturbance sources, and combining the corresponding transient signals collected in real time in each control channel to perform signal disturbance operations. Specifically, it includes: The preset interference constraints include a first interference constraint, a second interference constraint, and a third interference constraint; The first interference constraint is a safety amplitude constraint, used to limit the maximum amplitude and energy of the interference factor, ensuring that the disturbance operation will not damage the physical integrity of the hydraulic components or the stability of the system. The second interference constraint is an effective bandwidth constraint, used to limit the frequency range of the interference factor, ensuring that it is within the effective dynamic response bandwidth of the hydraulic component, so as to ensure that the excitation signal can be correctly responded to by the component. The third interference constraint is a waveform interference constraint, used to limit the waveform distortion type and proportion of the transient signal. Its constraint range includes the basic waveform, such as the combined weight of sine wave, square wave and triangular wave. Interference factors corresponding to each interference constraint are generated within the constraint range through a high-entropy random disturbance source, including the first interference factor, the second interference factor and the third interference factor. Among them, the first interference factor is the amplitude adjustment coefficient, generated within the first interference constraint range by a Gaussian distribution random number generator; the second interference factor is the frequency offset, generated within the second interference constraint range by a Logistic chaotic mapping algorithm; the third interference factor is the waveform combination weight vector, randomly generated by a Dirichlet distribution, the sum of the vector elements is 1, and each element corresponds to the proportion of a basic waveform, satisfying the proportion limit of the third interference constraint. Based on the first, second, and third interference factors, multi-dimensional perturbations are applied to the acquired real-time transient signals to generate perturbed transient signals. These multi-dimensional perturbations include amplitude perturbation, frequency perturbation, and waveform perturbation. Amplitude perturbation: The original amplitude of the real-time transient signal is multiplied by the first interference factor to generate an amplitude perturbation signal. Frequency perturbation: A Fourier transform is performed on the real-time transient signal, and the frequency offset corresponding to the second interference factor is superimposed at characteristic frequencies, followed by an inverse Fourier transform to generate a frequency perturbation signal. Waveform perturbation: Based on the waveform combination weights of the third interference factor, at characteristic moments of the real-time transient signal, such as the rising edge of a pressure wave or the peak point of a vibration signal, the distortion component of the corresponding basic waveform is superimposed to generate a waveform perturbation signal. The disturbance signal; the final transient signal after disturbance is a weighted sum of amplitude disturbance signal, frequency disturbance signal, and waveform disturbance signal, with the weights determined by the importance coefficients of each control channel; the above process identifies the general equipment control channel corresponding to the hydraulic component, generates interference factors by combining high-entropy random disturbance sources, and conducts transient signal disturbance analysis to generate interference identifiers, effectively explores the potential correlation between transient signals and interference factors, and provides data-driven decision-making basis for accurate assessment of the health status and fault diagnosis of hydraulic components; the amplitude, frequency, and waveform combination of the transient signal after disturbance are extracted and fused into the first monitoring vector V1: V1=[y1,y2,y3]; where y1 represents amplitude, y2 represents frequency, and y3 represents waveform combination; Based on the first monitoring vector, the first monitoring vector is compared with the corresponding preset benchmark monitoring vector to obtain different deviations, including the first deviation, the second deviation, and the third deviation. Different weights are assigned to different deviations, including the first weight, the second weight, and the third weight. The acquisition of the first weight is related to the amplitude deviation, the acquisition of the second weight is related to the frequency deviation, and the acquisition of the third weight is related to the waveform combination deviation. Under the constraint that the sum of the three is 1, the weights are dynamically adjusted according to the real-time operating conditions of the hydraulic system, thereby realizing accurate monitoring and diagnosis of the health status of hydraulic components. The first weight is determined by the degree of influence of amplitude deviation on the mechanical impact damage of hydraulic components. Its significance lies in quantifying the contribution of amplitude deviation from the baseline amplitude to the instantaneous impact damage of hydraulic components, such as valve cores, seals, and cylinders. Amplitude directly determines the instantaneous load intensity of the hydraulic system. A larger first deviation value indicates an abnormal amplitude, such as excessively high or low amplitude, which can lead to a sudden increase in the impact load on components. For example, excessively high pressure amplitude can cause seal compression damage or insufficient power output, while excessively low speed amplitude can cause actuator jamming. The first weight automatically increases. Conversely, a smaller first deviation value results in a correspondingly lower first weight. Its function is to dynamically adjust the weight of amplitude deviation in comprehensive decision-making based on its risk level. During calculation, the corresponding weight is obtained by looking up a preset nonlinear mapping table (e.g., a piecewise function), achieving adaptive weighting for amplitude anomalies and ensuring that severe amplitude impacts are identified first. The second weight is derived from the ability of frequency deviation to affect the resonance risk of the hydraulic system. Its significance lies in quantifying the risk of system vibration amplification and structural fatigue caused by frequency deviation from the reference range. Frequency determines the response characteristics of the hydraulic system to disturbances in different frequency bands. Abnormal bandwidth can trigger system resonance. For example, if the frequency shifts to the inherent frequency range of a component, it will lead to an increase in the amplitude of pipeline vibration, accelerating the fatigue damage of components such as hydraulic pump bearings and pipeline joints. In the calculation process, an exponential function is used to adjust the weight. For example, when the second deviation is low, the second weight automatically decreases; when the second deviation is high, the second weight increases accordingly, and the weight increases exponentially after the deviation exceeds the resonance risk threshold. Its function is to highlight the early warning role of frequency deviation through a nonlinear weight allocation strategy, enhance the sensitivity to system resonance scenarios, and avoid cascading failures caused by high-frequency vibration in advance. The acquisition of the third weight stems from the indicative role of waveform combination deviation in the early identification of hydraulic faults. Its significance lies in distinguishing the contribution of normal waveform fluctuations to fault diagnosis from abnormal waveform distortion. Waveform combinations directly reflect the morphological integrity of transient signals; the larger the waveform combination deviation, the more signal distortion exists. For example, peaks in pressure waveforms and plateaus in speed waveforms often indicate component wear, such as valve core wear and oil contamination—typical characteristics of early faults. A preset third deviation threshold is used. When the third deviation is greater than or equal to the threshold, the third weight is set to a higher value, indicating that waveform distortion has a significant indicative role in fault diagnosis. When the third deviation is less than the threshold, the weight coefficient is set to a lower value, indicating that slight waveform fluctuations are mostly normal operating condition interference and do not require excessive weighting. Its function is to filter out invalid operating condition interference in transient signals through tiered weight settings. During calculation, conditional statements are used to assign weights after threshold comparison, ensuring that the waveform dimension only plays an important role in valid fault characteristic scenarios, thus improving the accuracy of fault diagnosis. Simultaneously, the first monitoring vector is modally decomposed to obtain the fundamental mode and the fluctuation mode. The first monitoring vector is initially decomposed using a variational mode decomposition algorithm to obtain several intrinsic mode functions (IMFs), denoted as IMF_1, IMF_2, ..., IMF_n, where each IMF corresponds to a transient feature component and n represents the number of IMFs. Based on the IMFs, the extreme point density, frequency, and waveform similarity are extracted, and a feature space is plotted at the same time, with the extreme point density as the X-axis, the frequency as the Y-axis, and the waveform similarity as the Z-axis. It should be noted that: extreme point density: the number of extreme points of the IMF per unit time, including the number of maxima and minima; frequency: the center frequency of the IMF output by variational mode decomposition; waveform similarity: the cross-correlation coefficient between the IMF and the preset standard waveform, with a value range of 0 to 1. The standard waveform is generated based on the transient signal of the hydraulic system under healthy conditions; based on the feature space, a certain frequency is fixed, and the feature space is cut to obtain the first cross section. Based on the first cross-section, IMF feature points within the first cross-section are mapped back to the first cross-section. The average extreme points and average waveform similarity within this cross-section are extracted to form an initial sphere center. A preset monitoring radius is used to spherically aggregate the first cross-section, and a first mode candidate set falling within the sphere is selected, which meets the basic stable features of low frequency, high similarity, and low fluctuation. Feature points falling outside the sphere are marked as the first undetermined modes. The preset monitoring radius is the maximum Euclidean distance between the feature points of healthy samples and the initial sphere center. Based on the feature space, a certain waveform similarity is fixed, and the feature space is cut to obtain the second cross-section. Based on the second cross section, the IMF feature points within the second cross section are mapped to the second cross section. The x-axis is preset as the frequency range and the y-axis as the extreme point density range. The rectangles of the second cross section are aggregated, and the candidate set of the second mode falling within the rectangle is selected. The second mode meets the fluctuation characteristics of high frequency, low similarity, and high fluctuation. The feature points falling outside the rectangle are marked as the second undetermined mode. The intersection of the first candidate mode set and the second undetermined mode set is taken as the basic mode, which is an IMF that simultaneously satisfies the condition of being within the sphere of the first cross section and not within the rectangle of the second cross section. These IMFs simultaneously possess the characteristics of low frequency, high similarity, low volatility, and are neither high frequency nor high volatility. The intersection of the first undetermined mode set and the second candidate mode set is taken as the fluctuation mode, which is an IMF that simultaneously satisfies the condition of being within the rectangle of the second cross section and not within the sphere of the first cross section. These IMFs simultaneously possess the characteristics of high frequency, low similarity, high volatility, and are neither low frequency nor low volatility. IMFs that are neither in the basic mode nor the fluctuation mode from the first and second undetermined modes are taken, and their energy proportion is calculated. If the energy proportion is greater than or equal to 50%, that is, close to the dominance of the basic mode, they are included in the basic mode. If the energy proportion is less than or equal to 20%, that is, meets the condition of the secondary importance of the fluctuation mode, they are included in the fluctuation mode. The rest are marked as transition modes, not classified for the time being, and stored in the system. The above-described process employs a two-section hybrid clustering approach, strongly binding the modal classification of hydraulic transient signals to the physical characteristics of low-frequency stability or high-frequency interference, and regular or distorted waveforms. The first section is a plane with a fixed low frequency, such as 50Hz, representing the density of extreme points at the upper frequency limit of the hydraulic basic stable mode—a plane of waveform similarity. This focuses on identifying which IMFs within the low-frequency band possess the basic characteristics of regular waveforms and smooth fluctuations. The second section is a plane with a fixed low waveform similarity, such as 0.5, representing the density of extreme points at the lower similarity limit of the hydraulic fluctuation mode. This focuses on identifying which IMFs at low similarity possess the interference characteristics of high-frequency and violent fluctuations. This approach solves the problem of difficulty in distinguishing modes using a single feature dimension. The combination of spherical and rectangular clustering takes into account both the clustering of basic modes and the intervalicity of interference modes. Based on the fundamental mode and the fluctuation mode, mode pairs are constructed and fused into a second monitoring vector through deviation, weight, and mode pair: V2=[y4,y5,y6], where y4 represents deviation, y5 represents weight, and y6 represents mode pair. This vector is then imported into a preset spatiotemporal variation model to obtain an interference identifier, which is used to consider the spatiotemporal variation characteristics of transient signals and dynamically adjusts with changes in time and space, representing the unique identity of the general-purpose hydraulic component equipment under different time and space conditions. The interference identifier is represented in the form of: factor(t)=g(V2(t),t); where t represents time, factor(t) represents the interference identifier at time t, V2(t) represents the second monitoring vector of the general-purpose hydraulic component equipment at time t, g(·) represents the time variation model, including time series analysis, and g(V2(t),t) is specifically expressed as: In the formula, e1 is the number of feature dimensions in y4, e2 is the number of feature dimensions in y5, e3 is the number of feature dimensions in y6, p represents the p-th feature, and f p (t) represents the value of the p-th feature in the second monitoring vector V2 at time t, ω p The weight of the p-th feature is represented by β(t), which is a dynamic time bias term used to adjust the influence of spatiotemporal factors. As time t progresses, the second monitoring vector V2(t) of the general-purpose hydraulic component equipment will change, resulting in dynamic changes in the interference identifier. The above process generates reasonable interference constraints and disturbance factors, and through the interference quantification and identification capabilities of scientific disturbance analysis, accurately characterizes the type, intensity and scope of interference, breaks the steady-state operation mode of hydraulic components under a single working condition, generates characteristics with interference properties, effectively distinguishes between disturbance fluctuations and fault distortions, and enhances the accurate identification of misjudgments or fault signs in health status assessment.

[0018] S2: Based on the operating status of hydraulic components and combined with interference indicators, a health status vector is formed. The deviation of the health status vector from the health baseline feature map is compared, and a diagnostic channel is selected, including a first diagnostic channel and a second diagnostic channel. If the deviation is less than the standard deviation threshold, a status assessment strategy is executed in the first diagnostic channel to identify a stable health trend. If the deviation is greater than or equal to the standard deviation threshold, a fault diagnosis strategy is executed in the second diagnostic channel to identify fault signs. Based on the operational status, operational elements are extracted, including pressure gradient, expansion speed, and high temperature duration. Combined with interference indicators, a health status vector is formed through quantitative processing. The operational status includes cold start, low load, and high load. It should be noted that the working elements extracted under different operating conditions are as follows: pressure gradient: the rate of change of pressure in the hydraulic working chamber per unit time. The larger the absolute value of the pressure gradient, the more drastic the load change of the hydraulic component; extension and retraction speed: the speed of extension and retraction of the hydraulic cylinder piston rod, i.e., the rate of change of displacement with respect to time; high temperature duration: the continuous duration for which the oil temperature in the oil circuit where the hydraulic component is located exceeds the normal operating range. The longer the high temperature duration, the more frequently the oil temperature exceeds the standard, and the faster the hydraulic component ages. Among them, the steps for obtaining the health baseline feature map include: collecting all historical equipment operating parameters and historical transient signals under historical health conditions, and identifying the health baseline feature map through an unsupervised clustering algorithm; where the equipment operating parameters include, but are not limited to, load pressure, oil temperature, and hydraulic pump speed, and the unsupervised clustering algorithm is based on the K-means seed clustering algorithm; The specific steps include: Preprocessing: Outlier removal, noise suppression, and dimensional normalization are performed on historically collected equipment operating parameters and transient signals to generate a standardized health dataset; Unsupervised operating condition clustering: Cluster Analysis: The K-means seed clustering algorithm is used to perform cluster analysis on the standardized health dataset, automatically dividing it into several health condition clusters. This includes: determining the optimal number of clusters K for each health condition through weighted voting; performing initial mean clustering based on the optimal number of clusters K to obtain initial health condition clusters; selecting seed sample sets based on the initial health condition clusters; and correcting the clustering results using Mahalanobis distance to obtain optimized health condition clusters. Feature Extraction: For each health condition cluster, core health baseline features are extracted from transient signals, including peak values, time-domain features, and frequency-domain features. The equipment operating parameters of each health condition cluster are correlated with the corresponding core health baseline features, with load pressure, oil temperature, and hydraulic pump speed as the horizontal axis parameters and peak values, time-domain feature values, and frequency-domain feature values ​​as the vertical axis parameters. A multi-dimensional mapping table is established; simultaneously, the feature standard deviation of samples within the healthy operating condition cluster is calculated and marked as a feature reliability index, with a value ranging from 0 to 1. The closer the value is to 1, the more stable the feature under that operating condition is; finally, a healthy baseline feature map containing the mapping relationship between operating condition parameters and feature parameters and the feature reliability index is obtained; it should be noted that time-domain feature values ​​can be represented as step response time, and frequency-domain feature values ​​can be identified as frequency response functions. In the above process, a healthy baseline library containing all operating condition-specific models is constructed; by establishing an analysis model of hydraulic components under a single operating condition, all different stable operating conditions of the component within the entire operating envelope can be automatically identified, and a dedicated, high-fidelity dynamic model and data statistical profile can be established for each operating condition; If the deviation is less than the standard deviation threshold, a status assessment strategy will be implemented within the first diagnostic channel, including: Time window segmentation: Within the first diagnostic channel, the deviation levels continuously collected under the same operating condition are divided into several time segments according to a fixed-length operating condition cycle window. Each time segment contains the deviation levels of multiple sampling points, and all deviation levels are less than the standard deviation threshold. Trend feature extraction: Key trend features of each time segment are extracted. Key trend features include the average deviation of the window, the maximum deviation of the window, the slope of the deviation change, and the variance of the deviation fluctuation. The key trend features of several time segments are integrated into a high-dimensional feature matrix, i.e., an m×4 high-dimensional feature matrix is ​​constructed, where m is the number of operating condition cycle windows. It should be noted that the average deviation within a window is the average of all health deviations within a time window, reflecting the overall health level of the hydraulic components during that period. For example, in a 100-second window with 10 sampling points, the average deviation is obtained by adding these 10 values ​​and dividing by 10. The lower the value, the closer the components are to the health baseline during that period; the higher the value, the worse the overall health level. The maximum deviation within a window is the maximum value among all health deviations within a time window, reflecting the upper limit of fluctuation in the health status of the hydraulic components during that period. For example, the largest value among the 10 deviations in a 100-second window is the maximum deviation. If the average deviation is not high, but the maximum deviation is close to the standard deviation threshold, it indicates a momentary health fluctuation, possibly due to a load shock or instantaneous leakage, which requires attention. The slope of the deviation change is the change in the health deviation from the beginning to the end of a time window. Rate reflects the trend and speed of change of the health status during this period. For example, the slope is obtained by subtracting the deviation of the first sampling point from the deviation of the last sampling point in the window, and then dividing by the length of this period. A positive slope indicates that the health status is slowly deteriorating during this period, manifested as the deviation increasing and approaching the standard deviation threshold. A negative slope indicates that the health status is gradually improving. A slope close to 0 indicates that the health status is basically stable. Deviation fluctuation variance is the average of the squared deviations of all deviations from the average deviation within a time window, i.e., a time segment. It reflects the stability of the hydraulic component's health status during this period. For example, if the deviations within the window fluctuate slightly around the average, the variance is small; if they fluctuate wildly, the variance is large. The smaller the variance, the more stable the health status during this period; the larger the variance, the more drastic the fluctuations in the health status, which may lead to frequent load changes, hydraulic shocks, and poor stability. These four indicators combined can fully characterize the health trend within a time window. Similarity calculation: The similarity between the high-dimensional feature matrix and the corresponding feature matrices of each time segment is calculated. For each sample point, it is assumed that the points around it are closer to it, while the points further away are sparser. For example, the Euclidean distance between any two time segment feature matrices in the high-dimensional feature matrix is ​​calculated. Based on the Gaussian distribution probability density function, the Euclidean distance is converted into a similarity value. The closer the distance, the higher the similarity; the farther the distance, the lower the similarity. In order to balance the overall distribution, symmetry processing is completed by calculating the bidirectional similarity mean, forming an m×m symmetric similarity matrix. Low-dimensional space compression: A Gaussian distribution is used to compress the high-dimensional feature matrix into a low-dimensional space, such as 2D or 3D, to obtain a low-dimensional feature matrix. Initial coordinates of each window feature point are randomly generated in the low-dimensional space. A Gaussian distribution model consistent with that in the high-dimensional space is used to calculate the similarity between low-dimensional feature points, ensuring that the distance between feature vectors in the low-dimensional space is positively correlated with the similarity in the high-dimensional space. Trend change determination: A loss function based on a Gaussian distribution is constructed, using the Kullback-Leibler divergence as an error metric to quantify the difference between the high-dimensional and low-dimensional similarity distributions. The coordinates of the low-dimensional feature points are iteratively adjusted using a gradient descent algorithm to approximate the high-dimensional distribution. The near-high-dimensional structure involves performing a "near-close" operation on feature points with high similarity in high-dimensional space but far apart in low-dimensional space, and a "push-away" operation on feature points with low similarity in high-dimensional space but close apart in low-dimensional space. This is done while dynamically adapting the high-dimensional bandwidth parameters to a Gaussian distribution until the error difference between two adjacent iterations is less than 1e-4, achieving convergence. Finally, the optimized low-dimensional feature matrix is ​​output, preserving the stable clustering structure in high-dimensional space and capturing the clustering state of stable operating conditions. A health trend trajectory visualization model is constructed using scatter plots combined with time-series connections to identify stable health trends, output stable trend features, and complete the health status trend assessment. If the deviation is greater than or equal to the standard deviation threshold, a fault diagnosis strategy is executed, including: Within the second diagnostic channel, all health status vectors with deviations greater than or equal to the standard deviation threshold are acquired and aggregated into a set H. Quantitative element groups for each health status vector in set H are extracted, including pressure gradient quantization value, expansion / contraction speed quantization value, and high temperature duration quantization value. The maximum and minimum subsets of each quantification element are then constructed. The difference between the maximum and minimum subsets corresponding to each quantification element is marked as the maximum span value of that element, i.e., the maximum span value of pressure gradient, expansion / contraction speed, and high temperature duration. The maximum span value reflects the dynamic fluctuation range of the corresponding quantification element; the larger the value, the higher the probability of significant abnormal fluctuations in that element. At the same time, the standard deviation coefficients corresponding to each quantitative element in set H are obtained, namely the standard deviation coefficients of pressure gradient, expansion and contraction speed, and high temperature duration. The larger the value of the standard deviation coefficient, the greater the degree of variation of the corresponding quantitative element in set H, and the more obvious the discreteness of the fault characteristics corresponding to the element. The product of the maximum span value and the standard deviation coefficient corresponding to each quantitative element is marked as the single-element diagnostic value, namely, pressure diagnostic value, velocity diagnostic value, and temperature diagnostic value. The larger the diagnostic value corresponding to a single quantitative element, the more prominent the fault indication under that element dimension. Using the single-element diagnostic values ​​as the axis, and pressure diagnostic value as the X-axis, velocity diagnostic value as the Y-axis, temperature diagnostic value as the Z-axis, and time as the fourth axis, a four-dimensional curve of pressure diagnostic value, velocity diagnostic value, temperature diagnostic value, and time is constructed to obtain a three-dimensional cross-section composed of pressure diagnostic value, velocity diagnostic value, and temperature diagnostic value. Based on the three-dimensional cross-section, the movement trajectory of the three-dimensional cross-section on the time axis is identified. Based on the DTW algorithm, the three-dimensional cross-section is compared and analyzed with the system's built-in trajectory: If the three-dimensional cross-sectional trajectory is a target sphere, and the target sphere is a compact sphere, similar to a ping-pong ball suspended in the air rotating at a constant speed around its central axis, and the center of the sphere does not shift when extended along the time axis, that is, the distance from each point on its surface to the center of the sphere remains consistent, without obvious shift or expansion, this trajectory indicates that the hydraulic component is in the optimal operating state, with stable pressure, uniform speed, controllable temperature, no interference or abnormality, and is judged as a balanced mode, showing no signs of failure; If the three-dimensional cross-sectional trajectory is in the shape of a target ellipsoid, the target ellipsoid is similar to the three-dimensional trajectory of a uniformly twisted spring. When extended along the time axis, the ellipsoid makes a periodic spiral motion around the ideal center. That is, with a regular ellipsoid as the base, a slow spiral extension along the time axis is superimposed. It is like the spiral progression of a spring being stretched at a uniform speed, and also like the trajectory of a planet slowly approaching a star while revolving around it. If a single or double parameter of pressure, speed, or temperature shows periodic fluctuations, the ratio of the major axis to the minor axis of the trajectory ellipse is ≤1.5, and the spiral amplitude fluctuation range is only 5% to 15%, which is a slight deviation but does not exceed the safe range. This trajectory indicates that there is a slight abnormal trend in the hydraulic component. It is necessary to continuously monitor the changes in the working condition and judge it as a warning mode, which is a sign of a slight fault. If the three-dimensional cross-sectional trajectory is a target cone shape, similar to the three-dimensional trajectory formed by sand grains falling in a funnel in a single direction, and as it extends along the time axis, the target cone continuously contracts along a certain dimension, that is, the apex of the target cone precisely points to a certain deteriorating parameter dimension, such as the temperature dimension, and the cross-sectional radius continuously contracts along this dimension over time, just like the directional movement of sand grains gradually converging from the opening of the funnel to the bottom; the core manifestation is a continuous increase in the single parameter temperature, which is a continuous deterioration. Although other parameters do not show obvious abnormalities, the overall safety is reduced due to the influence of the core parameter. This trajectory indicates that there are signs of moderate failure in the hydraulic components, such as a decrease in the efficiency of the heat dissipation system or a slight seal leak. The root cause needs to be investigated in time and the risk mode should be determined. If the three-dimensional cross-sectional trajectory resembles a divergent hyperboloid, similar to the three-dimensional trajectory formed by high-speed, multi-directional fragmentation after an explosion, and extends along the time axis, the hyperboloid diffuses synchronously along at least two dimensions, i.e., the trajectory diffuses synchronously along two or more dimensions of pressure, velocity, and temperature, much like the shape of an explosion shock wave radiating in all directions, and the diffusion amplitude continues to increase over time, forming an irregular divergent trajectory; this trajectory indicates a systemic failure of the hydraulic components, such as: valve port jamming combined with oil cavitation, seal leakage accompanied by pressure shock, requiring immediate shutdown and handling, classified as a critical mode, and a sign of severe failure; If the 3D cross-sectional trajectory does not have the aforementioned trajectory shape, secondary recognition is triggered. The sampling frequency of the original transient signal corresponding to the health state vector is increased, features are re-extracted to generate a 3D cross-section, and the trajectory shape is compared again. All the data mentioned above are reference values, and specific settings are based on actual conditions. The above process constructs a health state vector based on the operating status of hydraulic components and interference indicators. By comparing the degree of deviation from the health baseline feature map, the diagnostic channel is adaptively selected. State assessment or fault diagnosis strategies are executed for different degrees of deviation. This ensures the accuracy of stable trend identification in the healthy state and improves the timeliness of fault sign capture in the fault state, significantly improving the accuracy of hydraulic component health diagnosis. At the same time, the introduction of preset interference constraints and the generation of interference indicators enable the system to fully consider the impact of interference factors on the health state, avoid misjudging interference as faults, and significantly improve the system's adaptability and robustness.

[0019] Example 2: This invention provides a fault diagnosis system for general-purpose hydraulic component equipment; Figure 2 This is a schematic diagram of a fault diagnosis system module according to an exemplary embodiment; please refer to... Figure 2 The system includes a first module, a second module, and a third module, and the first module, the second module, and the third module are interconnected. The first module identifies the control channels of general-purpose equipment based on hydraulic components. It collects transient signals in real time within each control channel, introduces preset interference constraints, generates interference factors through high-entropy random disturbance sources, and analyzes the amplitude, frequency, and waveform state disturbances of the transient signals to generate interference identifiers. The second module forms a health state vector based on the operating status of hydraulic components and the interference identifiers. It compares the deviation of the health state vector from the health baseline feature map and selects a diagnostic channel, including a first diagnostic channel and a second diagnostic channel. The third module executes a status assessment strategy in the first diagnostic channel if the deviation is less than a standard deviation threshold to identify a stable health trend; if the deviation is greater than or equal to the standard deviation threshold, it executes a fault diagnosis strategy in the second diagnostic channel to identify fault signs.

[0020] Example 3: Embodiment 3 of the present invention provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, it implements a fault diagnosis method for a general hydraulic component device provided in Embodiment 1.

[0021] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0022] In a possible implementation, the present invention can also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, is used to cause the terminal device to perform the steps of implementing the fault diagnosis method for a general hydraulic component device in Embodiment 1.

[0023] In the application, the various formulas mentioned are all calculated by removing dimensions and taking their numerical values. The formulas are derived from the most recent real-world situation by collecting a large amount of data and conducting software simulations. The formulas are set by those skilled in the art according to the actual situation.

[0024] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0025] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0026] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A fault diagnosis method for general-purpose hydraulic component equipment, characterized in that, The method includes: Based on hydraulic components, the control channels of general equipment are identified; transient signals are collected in real time in each control channel, preset interference constraints are introduced, interference factors are generated through high-entropy random disturbance sources, and the amplitude, frequency and waveform state disturbance of transient signals are analyzed to generate interference labels. The process of generating an interference identifier includes: extracting amplitude, frequency, and waveform combinations from the transient signal after disturbance and fusing them into a first monitoring vector; comparing the first monitoring vector with the corresponding preset benchmark monitoring vector to obtain different deviations; assigning different weights to different deviations; performing mode decomposition on the first monitoring vector to obtain a fundamental mode and a fluctuating mode; constructing mode pairs based on the fundamental mode and the fluctuating mode; fusing them into a second monitoring vector through deviations, weights, and mode pairs; and importing the second monitoring vector into a preset spatiotemporal variation model to obtain an interference identifier. The first monitoring vector mode decomposition yields the fundamental mode and wave mode, including: The first monitoring vector is initially decomposed using a variational mode decomposition algorithm to obtain several eigenmode functions; Based on the intrinsic mode function, the density, frequency, and waveform similarity of extreme points are extracted, and a feature space composed of the density, frequency, and waveform similarity of extreme points at the same time is constructed. Based on the feature space, a certain frequency is fixed, and the feature space is cut to obtain the first cross section. Based on the first cross section, the extreme points and waveform similarity within the cross section are extracted to form an initial sphere center set. The monitoring radius is preset, and the first cross section is spherically aggregated. The first mode candidate set falling within the sphere is screened, and the feature points falling outside the sphere are marked as the first undetermined mode. Based on the feature space, a certain waveform similarity is fixed, and the feature space is cut to obtain a second cross section. Based on the second cross section, the x-axis is preset as the frequency range and the y-axis as the extreme point density range. The rectangles of the second cross section are aggregated, and the candidate set of the second mode falling within the rectangle is filtered. The feature points falling outside the rectangle are marked as the second undetermined modes. The intersection of the first candidate mode and the second undetermined mode is taken as the basic mode; The intersection of the first undetermined mode and the candidate set of the second mode is taken as the wave mode; Based on the operating status of hydraulic components and combined with interference indicators to form a health status vector, the degree of deviation between the health status vector and the health baseline feature map is compared, and a diagnostic channel is selected, including a first diagnostic channel and a second diagnostic channel: if the deviation is less than the standard deviation threshold, a status assessment strategy is executed in the first diagnostic channel to identify a stable health trend; if the deviation is greater than or equal to the standard deviation threshold, a fault diagnosis strategy is executed in the second diagnostic channel to identify fault signs.

2. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 1, characterized in that, The hydraulic components include at least one of a hydraulic pump, a hydraulic valve, and a hydraulic cylinder. The control channels include at least an electro-hydraulic proportional valve control channel and a hydraulic pump displacement control channel. The transient signals include at least two signals selected from pressure, vibration, acoustic, and temperature sensors.

3. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 1, characterized in that, The process involves introducing preset interference constraints, generating interference factors through high-entropy random disturbance sources, and analyzing the amplitude, frequency, and waveform state disturbances of transient signals, including: The system loads preset interference constraints, including a first interference constraint, a second interference constraint, and a third interference constraint, and obtains the corresponding constraint range. It generates several types of interference factors, including a first interference factor, a second interference factor, and a third interference factor, through a high-entropy random disturbance source. It combines the corresponding transient signals collected in real time from each control channel to perform signal disturbance operations, including amplitude disturbance, frequency disturbance, and waveform disturbance.

4. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 1, characterized in that, The health baseline feature map was obtained by identification through an unsupervised clustering algorithm, which was established based on the K-means seed clustering algorithm.

5. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 1, characterized in that, The execution status assessment strategy includes: Within the first diagnostic channel, the deviation levels continuously collected under the same operating conditions are divided into several time segments according to a fixed-length operating cycle window. Each time segment contains the deviation levels of multiple sampling points, and all deviation levels are less than the standard deviation threshold. Key trend features are extracted from time series segments to construct a high-dimensional feature matrix. The similarity between the high-dimensional feature matrix and the corresponding feature matrices of each time series segment is calculated and processed to form a similarity matrix. The high-dimensional feature matrix is ​​then compressed into a low-dimensional space to obtain a low-dimensional feature matrix. The low-dimensional feature matrix is ​​positively correlated with the similarity matrix. A loss function based on Gaussian distribution is constructed, and the low-dimensional distribution is optimized through gradient descent to identify the clustering state of healthy and stable trends.

6. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 1, characterized in that, The execution of the fault diagnosis strategy includes: Within the second diagnostic channel, all health status vectors with deviations greater than or equal to the standard deviation threshold are acquired and aggregated into set H. Quantitative element groups for each health status vector in set H are extracted, and the maximum and minimum subsets for each quantitative element are constructed. The difference between the maximum and minimum subsets for each quantitative element is marked as the maximum span value of that element. Simultaneously, the standard deviation coefficients for each quantitative element in set H are acquired, and the product of the maximum span value and the standard deviation coefficient for each quantitative element is marked as the single-element diagnostic value. Using each single-element diagnostic value as an axis, a multidimensional curve is constructed, and the trajectory of the multidimensional curve moving with the time axis is obtained. Combining the shape of the trajectory in a three-dimensional cross-section, fault signs are identified.

7. The fault diagnosis method for a general-purpose hydraulic component equipment according to claim 6, characterized in that, The shape of the combined trajectory in a three-dimensional cross-section is used to identify signs of failure, including: If the trajectory is spherical, it is determined to be without fault signs; if the trajectory is ellipsoidal, it is determined to be a minor fault sign; if the trajectory is conical, it is determined to be a moderate fault sign; if the trajectory is hyperboloidal, it is determined to be a severe fault sign; if the trajectory does not have any of the above shapes, a second identification is performed.

8. A fault diagnosis system for a general-purpose hydraulic component equipment, characterized in that, The system includes: The first module is used to identify the control channels of general equipment based on hydraulic components; it collects transient signals in real time in each control channel, introduces preset interference constraints, generates interference factors through high-entropy random disturbance sources, analyzes the amplitude, frequency and waveform state disturbance of transient signals, and generates interference labels. The process of generating an interference identifier includes: extracting amplitude, frequency, and waveform combinations from the transient signal after disturbance and fusing them into a first monitoring vector; comparing the first monitoring vector with the corresponding preset benchmark monitoring vector to obtain different deviations; assigning different weights to different deviations; performing mode decomposition on the first monitoring vector to obtain a fundamental mode and a fluctuating mode; constructing mode pairs based on the fundamental mode and the fluctuating mode; fusing them into a second monitoring vector through deviations, weights, and mode pairs; and importing the second monitoring vector into a preset spatiotemporal variation model to obtain an interference identifier. The first monitoring vector mode decomposition yields the fundamental mode and wave mode, including: The first monitoring vector is initially decomposed using a variational mode decomposition algorithm to obtain several eigenmode functions; Based on the intrinsic mode function, the density, frequency, and waveform similarity of extreme points are extracted, and a feature space composed of the density, frequency, and waveform similarity of extreme points at the same time is constructed. Based on the feature space, a certain frequency is fixed, and the feature space is cut to obtain the first cross section. Based on the first cross section, the extreme points and waveform similarity within the cross section are extracted to form an initial sphere center set. The monitoring radius is preset, and the first cross section is spherically aggregated. The first mode candidate set falling within the sphere is screened, and the feature points falling outside the sphere are marked as the first undetermined mode. Based on the feature space, a certain waveform similarity is fixed, and the feature space is cut to obtain a second cross section. Based on the second cross section, the x-axis is preset as the frequency range and the y-axis as the extreme point density range. The rectangles of the second cross section are aggregated, and the candidate set of the second mode falling within the rectangle is filtered. The feature points falling outside the rectangle are marked as the second undetermined modes. The intersection of the first candidate mode and the second undetermined mode is taken as the basic mode; The intersection of the first undetermined mode and the candidate set of the second mode is taken as the wave mode; The second module is used to form a health status vector based on the operating status of hydraulic components and interference indicators, compare the degree of deviation between the health status vector and the health baseline feature map, and select a diagnostic channel, including a first diagnostic channel and a second diagnostic channel. The third module executes a status assessment strategy in the first diagnostic channel to identify a healthy and stable trend if the deviation is less than the standard deviation threshold; if the deviation is greater than or equal to the standard deviation threshold, it executes a fault diagnosis strategy in the second diagnostic channel to identify fault signs.

9. A computer-readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements a fault diagnosis method for a general hydraulic component device according to any one of claims 1-7.

Images