Remaining service life prediction method of hydraulic component, computer equipment and storage medium

Abstract

The invention discloses a residual service life prediction method of a hydraulic element, computer equipment and a storage medium. The method comprises the following steps: receiving response data of the hydraulic element; inputting the response data into a pre-trained residual service life prediction model to predict a first residual service life; performing time-frequency analysis on the response data to obtain a current time-frequency feature, and determining a second remaining service life based on the current time-frequency feature and a reference time-frequency feature; and under the condition that the second remaining service life is within the interval range corresponding to the first remaining service life, determining that the second remaining service life is the target remaining service life of the hydraulic element. Therefore, the prediction accuracy and the prediction reliability of the remaining service life of the hydraulic element are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydraulic element detection, in particular to a hydraulic element residual service life prediction method, a computer device and a storage medium. BACKGROUND

[0002] As the core executive component of the working machine, the health status of the hydraulic element is directly related to the reliability and safety of the working machine. Accurate prediction of the residual service life of the hydraulic element to realize predictive maintenance has become the key to improving the competitiveness of the working machine market.

[0003] At present, the residual service life prediction methods of the hydraulic element mainly include three kinds: based on experience knowledge, based on mathematical model and data driven. The first two are difficult to achieve due to the complexity of the hydraulic system and the difficulty of obtaining accurate mathematical models. The latter, i.e. the data driven method, is difficult to accurately predict when the working condition of the working machine is complex and the service life of the component changes.

[0004] Therefore, how to improve the prediction accuracy of the residual service life of the hydraulic element under complex working conditions and improve the reliability of the working machine has become a technical problem to be solved. SUMMARY

[0005] The purpose of the present application is to provide a hydraulic element residual service life prediction method, a computer device and a storage medium, which can improve the accuracy and reliability of the residual service life prediction of the hydraulic element.

[0006] To achieve the above purpose: In a first aspect, the present application provides a hydraulic element residual service life prediction method, comprising: receiving response data of the hydraulic element; inputting the response data into a pre-trained residual service life prediction model to predict a first residual service life; performing time-frequency analysis on the response data to obtain a current time-frequency feature, and determining a second residual service life based on the current time-frequency feature and a reference time-frequency feature; in the case that the second residual service life is located in the interval range corresponding to the first residual service life, determining the second residual service life as the target residual service life of the hydraulic element.

[0007] In an embodiment, the method further comprises: obtaining working condition state information of the working machine corresponding to the hydraulic element; receiving the response data of the hydraulic element, comprising: receiving the response data of the hydraulic element when the working condition state information is determined as the target working condition.

[0008] In an embodiment, the method further comprises: obtaining the test bench test data and the real vehicle operation data of the sample hydraulic components in different remaining useful life stages, wherein the test bench test data and the real vehicle operation data each comprise sample response data of the sample hydraulic components with a remaining life label; training the neural network model using the test bench test data as a training set to obtain a base model; and performing transfer learning on the base model using the real vehicle operation data as a transfer set to obtain the remaining useful life prediction model.

[0009] In an embodiment, the method further comprises: obtaining sample working condition state information of the sample hydraulic components corresponding to sample working machines in different working conditions; and obtaining the test bench test data and the real vehicle operation data of the sample hydraulic components in different remaining useful life stages, comprising: determining a target working condition of the sample hydraulic components corresponding to the sample working machines in different working conditions according to the sample working condition state information of the sample hydraulic components corresponding to the sample working machines in different working conditions; obtaining test sample response data and real vehicle sample response data of the sample hydraulic components in different remaining useful life stages under the target working condition, respectively; and obtaining the test bench test data and the real vehicle operation data of the sample hydraulic components in different remaining useful life stages according to the test sample response data and the real vehicle sample response data of the sample hydraulic components in different remaining useful life stages under the target working condition.

[0010] In an embodiment, obtaining the test sample response data and the real vehicle sample response data of the sample hydraulic components in different remaining useful life stages under the target working condition comprises: obtaining a first quantity of test sample response data of the sample hydraulic components in different remaining useful life stages under the target working condition; and obtaining a second quantity of real vehicle sample response data of the sample hydraulic components in different remaining useful life stages under the target working condition, wherein the second quantity is less than the first quantity.

[0011] In an embodiment, the method further comprises: performing time-frequency analysis on the reference response data of the first reference component to obtain a first reference time-frequency feature; performing time-frequency analysis on the reference response data of the second reference component to obtain a second reference time-frequency feature, wherein the remaining useful life of the first reference component is greater than the remaining useful life of the second reference component; and determining the second remaining useful life based on the current time-frequency feature and the reference time-frequency features, comprising: determining the second remaining useful life according to the current time-frequency feature, the first reference time-frequency feature, and the second reference time-frequency feature.

[0012] In an embodiment, the time-frequency analysis on the response data to obtain the current time-frequency feature comprises: performing wavelet packet decomposition on the response data to extract energy features of subbands in a predetermined frequency band range as the current time-frequency feature.

[0013] In an embodiment, the method further comprises: in a case where the second remaining service life is located in the interval range corresponding to the first remaining service life, determining that the second remaining service life is a valid prediction result; and outputting an alarm information to prompt a user to perform an abnormal state treatment on the hydraulic element, if the second remaining service life in the continuous K valid prediction results is less than the life threshold, where K is a positive integer.

[0014] In a second aspect, an embodiment of the present application provides a computer device, comprising a processor and a memory storing a computer program, and when the processor executes the computer program, the above-mentioned method for predicting a remaining service life of a hydraulic element is implemented.

[0015] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, and the computer readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for predicting a remaining service life of a hydraulic element is implemented.

[0016] The method for predicting a remaining service life of a hydraulic element, the computer device and the storage medium provided by the embodiments of the present application receive response data of the hydraulic element, input the response data into a pre-trained remaining service life prediction model to obtain a first remaining service life, perform time-frequency analysis on the response data to obtain a current time-frequency feature, and determine a second remaining service life based on the current time-frequency feature and a reference time-frequency feature. In a case where the second remaining service life is located in an interval range corresponding to the first remaining service life, the second remaining service life is determined as a target remaining service life of the hydraulic element. In this way, the embodiments of the present application construct two kinds of predictions of the remaining service life of the hydraulic element, that is, a double-path prediction of the first remaining service life and the second remaining service life, and set a verification mechanism that the second remaining service life is determined as the target remaining service life of the hydraulic element in a case where the first remaining service life is located in an interval range corresponding to the second remaining service life, thereby improving the accuracy of the prediction of the target remaining service life, improving the usability and guiding value of the prediction result of the target remaining service life, and improving the reliability of the prediction result and reducing the risk of false positives. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of the method for predicting a remaining service life of a hydraulic element provided by the embodiments of the present application is shown.

[0018] Figure 2 A flowchart of the training method of the remaining service life prediction model provided by the embodiments of the present application is shown.

[0019] Figure 3 A training scene diagram of the remaining service life prediction model provided by the embodiments of the present application is shown.

[0020] Figure 4 An application scenario diagram of a remaining useful life prediction web platform provided for an embodiment of the present application is shown.

[0021] Figure 5 A structural diagram of a remaining useful life prediction device for a hydraulic element provided for an embodiment of the present application is shown.

[0022] Figure 6 A structural diagram of a computer device provided for an embodiment of the present application is shown. DETAILED DESCRIPTION

[0023] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The following description is made with reference to the accompanying drawings in which like reference numerals refer to like elements, and the term "exemplary" is used herein to mean "serving as an example, instance, or illustration." Unless otherwise noted, the description of a particular embodiment applies equally to like elements of other embodiments. It is to be understood that the specific embodiments described herein are merely exemplary and do not limit the scope of the present application.

[0024] The following first explains the terms that can be involved in the present embodiment: Neural network model: a computational model inspired by the structure of the human brain, composed of a large number of interconnected nodes or neurons arranged in layers, including input layer, hidden layer and output layer; through learning complex and nonlinear relationships in a large amount of data to perform pattern recognition, classification or prediction, each connection has a weight, the model learns by repeatedly adjusting these weights, so that for a given input, the desired output can be produced.

[0025] SE_ResNet convolutional neural network is an advanced and powerful deep learning model, which is a combination of ResNet and SENet technologies. ResNet solves the gradient vanishing problem in very deep neural networks by introducing "residual connection" or "jump connection", making it possible to train very deep networks, thereby enabling the learning of more complex features. The core of SENet is the "compression-activation" module, which can adaptively calibrate channel feature responses, allowing the model to learn which feature channels are important and enhance them while suppressing unimportant channels. SE_ResNet convolutional neural network is particularly suitable for processing complex stress signals, the depth of ResNet guarantees the feature extraction capability, and the SE mechanism can automatically focus on the frequency bands or features most relevant to life degradation in the signal, thereby improving the accuracy of the prediction.

[0026] Wavelet packet decomposition is a more refined signal frequency band analysis method than traditional wavelet decomposition. It can decompose a signal at multiple frequency band levels, not only decomposing the low-frequency part but also the high-frequency part. It is like using a set of precision sieves to filter the signal layer by layer, ultimately obtaining the sub-signals or sub-bands of the signal in different fine frequency bands. This improves the local time-frequency information more than traditional spectrum analysis. Wavelet packet decomposition can decompose the high-frequency pressure response signal of hydraulic components into multiple finer frequency bands to capture the subtle changes in energy distribution within a specific frequency band caused by the wear of hydraulic components.

[0027] Energy analysis refers to the calculation of the energy of each sub-band signal after wavelet packet decomposition. Specifically, it involves squaring or averaging the amplitude of all data points within a sub-band to obtain a numerical value representing the signal strength of that frequency band. In this context, when hydraulic components wear down, the frequency characteristics of their vibration or pressure response change, causing regular variations in the signal energy at these frequencies. For example, some frequency bands may experience energy attenuation while others may experience enhancement. By tracking these changes in sub-band energy, the health status of the component can be quantified.

[0028] See Figure 1 This application provides a method for predicting the remaining service life of a hydraulic component. This method can be executed by a device for predicting the remaining service life of a hydraulic component, which can be implemented in software and / or hardware, such as a computer or server. For example, the method can run on a hydraulic component remaining service life prediction network platform (hereinafter referred to as the prediction system). This network platform can be integrated into a server, which can be a local server or a cloud server. In this embodiment, the execution entity of the hydraulic component remaining service life prediction method is a server (such as a cloud server). The remaining service life prediction method provided in this embodiment includes: Step 101: Receive response data for the hydraulic components.

[0029] Hydraulic components are parts of a hydraulic system, such as hydraulic cylinders, hydraulic motors, hydraulic valves, and hydraulic pumps. Hydraulic cylinders (or oil cylinders) are the most important and typical hydraulic components, such as those used in machinery to perform lifting, pushing, pulling, and supporting actions. For simplicity, this application will primarily use hydraulic cylinders as an example to illustrate the remaining useful life prediction method.

[0030] Response data refers to transient physical signals that reflect the health status of hydraulic components. For example, response data may include, but is not limited to, at least one of the following: pressure response data, vibration response data, velocity response data, and acceleration response data. The following example uses pressure response data. Here, pressure response data is high-frequency analog data acquired by a high-frequency pressure sensor, and then a pressure signal reflecting the stable and healthy state of the hydraulic component is extracted from the high-frequency analog data. This pressure response data can be a high-frequency signal.

[0031] In some implementations, the method further includes: Obtain the operating status information of the working machinery corresponding to the hydraulic component; in step S101, receive the response data of the hydraulic component, including: receiving the response data of the hydraulic component when the operating status information is determined to be the target operating condition.

[0032] The hydraulic component-related work machinery refers to the mechanical equipment that installs and uses the hydraulic component. For example, the work machinery may include, but is not limited to, at least one of the following: construction machinery, fire-fighting machinery, agricultural machinery, or work robots. For instance, the work machinery may be at least one of excavators, cranes, aerial work platforms, loaders, and pump trucks.

[0033] Among them, operating condition information refers to the parameter signals of the current working status of the machinery and its hydraulic system, as well as the external load environment. Operating condition information is used to determine the current working condition of hydraulic components.

[0034] For example, determining the target operating condition based on the operating condition status information includes: identifying the operating condition with a usage frequency higher than a frequency threshold as the target operating condition. In other words, in this embodiment, the system filters out response data collected under operating conditions with low usage frequency, thereby reducing noise in the service life prediction. It is understood that the operating condition corresponding to the high usage frequency threshold typically refers to the operating condition of the machinery under stable operating conditions; conversely, the operating condition corresponding to the low usage frequency threshold could be, for example, an operating condition with a drastic change in load.

[0035] Here, the response data can be data that has been filtered for validity based on operating condition information. For example, validity filtering could involve removing unstable data collected during periods of drastic load changes. This improves the accuracy of subsequent predictions of remaining useful life based on valid response data, reduces the processing of invalid data, increases data processing efficiency, and enhances the performance of the prediction system.

[0036] In some embodiments, step 101, receiving response data for the hydraulic component, includes: receiving response data sent by a data acquisition terminal, wherein the response data is obtained by the data acquisition terminal based on preprocessed raw response data for the hydraulic component. The preprocessing of the raw response data may include, but is not limited to, at least one of the following: removing low-frequency response data from the raw response data; removing response data from the raw response data that is not acquired under the target operating condition.

[0037] For example, a data acquisition terminal is deployed on the operating machinery, and the data acquisition terminal may include: The analog signal acquisition unit includes: an ADC sampling chip for acquiring raw analog signals; The CAN transceiver module is used to obtain raw response data from the controller area network of the operating machinery; The microprocessor, as the control core, is used for signal processing, logic judgment, and data scheduling; The wireless transmission module, including a 4G / 5G communication module, is used to establish a connection with the cloud server and transmit data; The power supply module is used to supply power to the above modules.

[0038] For example, the raw response data can be an analog signal such as pressure or vibration from a hydraulic component undergoing reciprocating linear motion for a fixed duration. The raw response data can also be hydraulic system status information from the machine's controller, such as the main pump speed. The microprocessor determines whether the raw response data is usable; if usable, it is transmitted to the server via wireless transmission; otherwise, it is deleted and re-acquired. It should be noted that usability can be determined by whether the raw response data was collected under the target operating condition.

[0039] For example, the data acquisition terminal preprocesses the acquired raw response data. Specifically, the processing of the raw response data is as follows: The ADC chip of the data acquisition terminal performs high-speed, high-precision analog-to-digital conversion on the voltage signal output by the pressure sensor connected to the reciprocating linear motion hydraulic component, for example, at a sampling frequency of up to 20kHz, to obtain the raw, continuous high-frequency digital sequence, i.e., the raw response data; then, the microprocessor of the data acquisition terminal extracts the continuous high-frequency data sequence into a set of independent data frames according to a fixed duration, such as 50 milliseconds. Each frame of data represents the complete dynamic response of the hydraulic component in a specific action, such as a startup process. The processing of the raw response data also includes the following: The CAN transceiver module of the data acquisition terminal listens to and receives messages on the CAN bus of the operating machinery in real time, and parses out the working condition information related to the hydraulic system. This working condition information includes, but is not limited to, the main pump speed, the hydraulic cylinder speed command, the system pressure, and the oil temperature; then, the microprocessor of the data acquisition terminal timestamps each set of high-frequency analog data frames and associates them with the working condition information within the same time point or time window. Furthermore, the microprocessor determines whether the raw response data collected at the current moment is usable based on the associated operating condition information. It then marks the raw response data as operating condition signals and the corresponding raw response data. Specifically, when the raw response data indicates that the hydraulic cylinder is in a state of uniform and stable motion and the system load is stable, it is determined that the raw response data collected at this time can effectively reflect the health status of the hydraulic components and the data is usable. Conversely, when the raw response data indicates that the hydraulic cylinder is starting, stopping, or experiencing a sudden change in load, the pressure signal at this time contains too much operating condition interference and is determined to be unusable. Thus, only the raw response data marked as usable and its associated operating condition signals are packaged to obtain the response data of the hydraulic components corresponding to the target operating condition, while invalid data is directly deleted and not transmitted. The raw response data under the operating condition information is then collected again for judgment and processing. Here, the response data can effectively reflect the health status of hydraulic components. If the response data clearly conforms to or belongs to the corresponding target working condition, the original response data collected is considered usable. Conversely, if the response data clearly does not conform to or does not belong to the corresponding target working condition, the original response data collected is considered unusable, and the original response data is collected again for judgment and processing.

[0040] Thus, in the above implementation, the data acquisition terminal uses a microprocessor to simultaneously receive raw response data and operating status information, and preprocesses the raw response data and operating status information to filter out a large amount of invalid data, reducing the server load and improving the efficiency of the prediction system. At the same time, because a large amount of invalid data is filtered out, the transmission burden between the data acquisition terminal and the server is also reduced, and the transmission efficiency is improved. This provides a strong guarantee for real-time data transmission and real-time prediction results, and provides a favorable foundation for subsequent online viewing of results based on real-time prediction results and reducing equipment maintenance response time.

[0041] Step 102: Input the response data into the pre-trained remaining useful life prediction model to predict the first remaining useful life.

[0042] The first remaining useful life can be a range of remaining useful life percentages or a range of remaining useful life values. In short, the first remaining useful life is not a specific useful life value, but an estimated useful life range.

[0043] For example, in step 102, based on the high-frequency analog signal and the operating condition signal, a remaining service life prediction model trained by transfer learning is used to predict the first remaining service life prediction result for the hydraulic component. This includes: the server normalizes the response data to eliminate the influence of amplitude dimensions; and uses the normalized signal as the input feature of the remaining service life prediction model, and then outputs the first remaining service life for the hydraulic component. Specifically, the remaining service life prediction model processes the acquired response data and outputs, for example, a discrete remaining service life percentage range, such as 0% to 50%, which indicates that the current remaining service life of the hydraulic component or cylinder is 50% of its rated service life.

[0044] In some implementations, please refer to Figure 2 , Figure 2 A flowchart illustrating the training method for the remaining useful life prediction model provided in this application embodiment is shown below. Figure 2 As shown, the method includes: Step 201: Obtain bench test data and vehicle operation data of sample hydraulic components at different stages of remaining service life. The bench test data and vehicle operation data include sample response data of the sample hydraulic components with remaining service life labels. Step 202: Use bench test data as the training set to train the neural network model and obtain the basic model; Step 203: Using real vehicle operation data as a transfer set, perform transfer learning on the base model to obtain the remaining service life prediction model.

[0045] For example, bench test data is collected in a controlled laboratory environment, such as a pumping test bench, to obtain sample data for multiple defined life stages (e.g., 100%, 80%, 50%, 25%, 0%) from brand new (100%) to completely scrapped (0%), for example through accelerated durability testing. On real-world machinery such as excavators and cranes, life labels are typically estimated based on operating time or initial conditions and are considered stable over a period of time, collected via data acquisition terminals.

[0046] In some embodiments, the method further includes: acquiring sample operating condition status information of the sample hydraulic component corresponding to the sample working machinery under different working conditions; acquiring bench test data and vehicle operation data of the sample hydraulic components at different remaining service life stages, including: determining the target working condition of the sample working machinery corresponding to the sample hydraulic component based on the sample operating condition status information of the sample working machinery under different working conditions; acquiring test sample response data and vehicle sample response data of the sample hydraulic components at different remaining service life stages under the target working condition; and acquiring bench test data and vehicle operation data of the sample hydraulic components at different remaining service life stages based on the test sample response data and vehicle sample response data of the sample hydraulic components at different remaining service life stages under the target working condition.

[0047] For example, the system analyzes a large amount of historical data, such as the load spectrum recorded by the vehicle controller, and identifies one or more of the most frequently used operating conditions for the hydraulic components as target operating conditions. For instance, for the bucket cylinder of an excavator, target operating conditions may include "medium load digging," "heavy load digging," and "leveling operations," and these operating conditions with a usage frequency higher than a frequency threshold are defined as target operating conditions.

[0048] For example, please refer to Figure 3 , Figure 3 This is a schematic diagram of a training scenario for the remaining useful life prediction model provided in the embodiments of this application, such as... Figure 3 As shown in the left figure, accelerated durability tests were conducted on a batch of hydraulic components of the same model on a test bench according to the speed, pressure, temperature, and other parameters corresponding to the target operating conditions. During the test, the test was periodically interrupted, and the hydraulic components were removed for calibration to obtain test sample response data of hydraulic components at different remaining service life stages under the target operating conditions. Furthermore, hydraulic components at different remaining service life stages can also be obtained; for example, some have a remaining service life of 100%, some have a remaining service life of 80%, some have a remaining service life of 50%, some have a remaining service life of 25%, some have a remaining service life of 10%, and some have a remaining service life of 0%.

[0049] For example, a pressure sensor is used to acquire high-frequency pressure data of hydraulic components under various target operating conditions at each stage of their remaining service life, and this high-frequency pressure data is used as the response data for a real vehicle sample. For example, the high-frequency pressure data can also be rod-cavity high-frequency pressure data. For example, the high-frequency pressure data can also be pressure signals from other locations or other types of signals, such as vibration or flow signals.

[0050] Furthermore, in some embodiments, test sample response data and real vehicle sample response data of sample hydraulic components at different remaining service life stages under target operating conditions will be acquired respectively, including: acquiring a first number of test sample response data of sample hydraulic components at different remaining service life stages under target operating conditions; and acquiring a second number of real vehicle sample response data of sample hydraulic components at different service life stages under target operating conditions, wherein the second number is less than the first number.

[0051] For details, please refer to [link / reference]. Figure 3 ,like Figure 3 As shown in the right figure, bench test data is input into, for example, an SE_ResNet convolutional neural network for initial model training. Then, a small amount of real vehicle operation data is used for transfer learning to obtain a transfer model, namely the aforementioned remaining service life prediction model. For example, please refer to [further details omitted]. Figure 3 The transfer learning process involves using a small amount of real-vehicle operating data, followed by data transfer learning using this small amount of real-vehicle operating data, and then model transfer learning using a hydraulic component lifecycle database. This allows the remaining service life prediction model obtained from the training to adapt to all operating conditions. The hydraulic component lifecycle database here includes operating data corresponding to all operating conditions of the hydraulic components.

[0052] Thus, in the above implementation, by combining the two types of data, first using standard bench test data as the basis for building the model, and then using real vehicle operation data to train the basic model through transfer learning, the core problems of the scarcity of high-quality labeled data and the performance degradation of laboratory models in field applications faced by deep learning models in the industrial field are solved through the bench pre-training and real vehicle transfer learning mode. This enables the trained remaining service life prediction model to balance high accuracy and strong generalization ability.

[0053] In some implementations, test sample response data and real vehicle sample response data of sample hydraulic components at different stages of remaining service life under target operating conditions are acquired, including: High-frequency pressure data of sample hydraulic components under target working conditions and at different stages of remaining service life were obtained on the test bench and in the actual vehicle environment. Based on the test sample response data and real vehicle sample response data of sample hydraulic components at different remaining service life stages under target operating conditions, bench test data and real vehicle operation data of sample hydraulic components at different remaining service life stages were obtained, including: The high-frequency pressure data from the test and the high-frequency pressure data from the actual vehicle were segmented and homogenized according to a preset time length, and each segment of data was marked with a corresponding remaining service life label to form bench test data and actual vehicle operation data, respectively.

[0054] Here, bench test data includes, but is not limited to, high-frequency pressure data with a first label. This refers to the high-frequency pressure data of hydraulic components collected by pressure sensors mounted on the test bench under simulated target operating conditions at each preset remaining life stage. The aforementioned first label can be used to mark the first remaining life stage and the first target operating condition information corresponding to the high-frequency pressure data. This first target operating condition information can be used to characterize the corresponding operating parameters at the time the high-frequency pressure data was collected, such as the precise control load, speed, and oil temperature.

[0055] Here, real-vehicle operating data includes, but is not limited to, measured high-frequency pressure data with a second tag, i.e., the real-vehicle high-frequency pressure signal detected under actual target operating conditions at each preset remaining life stage. The aforementioned second tag can be used to mark the second remaining life stage and the second target operating condition information corresponding to the measured high-frequency pressure data. The second target operating condition information can be used to characterize the operating condition parameters corresponding to the acquisition of the real-vehicle high-frequency pressure data, such as the real-time speed and pressure of the main pump, the speed of the hydraulic cylinder, and the oil temperature.

[0056] It should be added that the actual vehicle operation data is obtained by collecting high-frequency pressure data of hydraulic components under multiple operating conditions at a small number of different remaining service life stages, such as high-frequency pressure data of the rod chamber.

[0057] Therefore, it is understandable that there is less real-vehicle operating data and more bench test data; real-vehicle operating data has higher noise levels and bench test data has lower noise levels; the lifespan labels of real-vehicle operating data are estimated and not accurate enough, while the lifespan labels of bench test data are accurate and known.

[0058] In the above embodiments, high-frequency pressure data from different sources and of different durations are uniformly processed into standard data segments of fixed length, and the influence of amplitude dimensions is eliminated. This provides a regular and consistent input for subsequent model training, laying a data foundation for efficient learning and training of the model. Furthermore, each data segment is precisely labeled with its corresponding remaining service life label, such as a percentage of remaining service life, providing the model with a clear learning objective. This allows the model to directly learn the mapping relationship between signal features and lifespan status from these labeled samples, making it possible to use neural network models such as SE_ResNet convolutional neural networks for learning and training. Moreover, since the bench test data and real-vehicle operation data have the same structure and format, compatibility between the base model and the target domain data is ensured, providing favorable conditions for efficient transfer.

[0059] In some implementations, the neural network model includes an SE_ResNet convolutional neural network. In step 202, the neural network model is trained using bench test data as the training set to obtain a base model, including: training the SE_ResNet convolutional neural network using bench test data as the training set to obtain a base model.

[0060] It should be added that, in some implementations, the neural network model may also include convolutional neural networks (CNN), recurrent neural networks (RNN), residual neural networks, and support vector machines (SVM); any neural network model that supports training on discretized data is acceptable.

[0061] In some implementations, the homogenized test high-frequency pressure data and real vehicle high-frequency pressure data are also converted into image data and input into the aforementioned neural network model, such as the SE_ResNet convolutional neural network, for training and transfer learning to output the first remaining service life prediction result.

[0062] Thus, in this embodiment, by introducing the SE_ResNet convolutional neural network, which is a combination of ResNet and SENet technologies, the residual structure of ResNet effectively solves the gradient vanishing problem in deep network training. Furthermore, through SENet's channel attention mechanism, the importance of each feature channel can be automatically evaluated, and the signal of key channels can be adaptively enhanced while suppressing secondary or noisy channels. In the prediction of the remaining service life of hydraulic components, it can automatically focus on specific frequency bands or resonance components in the pressure signal that are most relevant to wear, improving prediction accuracy. Therefore, by employing the SE_ResNet convolutional neural network, this embodiment can achieve convergence to a better performance level with the same training cycle and data volume, providing an advanced model architecture guarantee for achieving high-precision prediction.

[0063] Step 103: Perform time-frequency analysis on the response data to obtain the current time-frequency characteristics, and determine the second remaining service life based on the current time-frequency characteristics and the reference time-frequency characteristics.

[0064] The second remaining useful life can be a percentage of the remaining useful life or a value of the remaining useful life. In short, the second remaining useful life can be a specific life value.

[0065] Specifically, the first remaining useful life output in step S102 is a discrete type of data, providing a reliable range of useful life states rather than a specific value, such as 41%~55.3% or 41~55%. The second remaining useful life output in step S103 is a continuous value, a precise quantification point, such as 55.3% or 55.3%.

[0066] Here, the dual-path prediction, consisting of interval prediction in step 102 and numerical prediction in step 103, provides a basis for verifying accurate prediction results in the future.

[0067] In some implementations, step 103 involves performing time-frequency analysis on the response data to obtain the current time-frequency features, including: performing wavelet packet decomposition on the response data and extracting the energy features of sub-bands within a predetermined frequency band as the current time-frequency features.

[0068] Wavelet packet decomposition can not only analyze which frequency components a signal contains, but also accurately pinpoint the timing of these frequency components. Wavelet packet decomposition simultaneously unfolds a signal in both time and frequency dimensions, resulting in a time-frequency distribution map. For example, time-frequency analysis of the response data yields the current time-frequency characteristics, including: extracting the energy characteristics of sub-bands n to m within a preset frequency band by performing three-layer wavelet packet analysis on the response data. It should be noted that not all frequencies are sensitive to wear; some frequency bands may contain mechanical vibration noise, or the inherent frequencies of some hydraulic pumps. We need to focus on frequencies that truly reflect fault characteristics such as hydraulic cylinder seal wear or internal wall scratches. Here, sub-bands n to m refer to the sub-bands with the highest wear sensitivity. For example, by analyzing the frequency domain differences in the response data of brand-new hydraulic components and completely worn hydraulic components, we can determine the specific sub-bands within the predetermined frequency band that exhibit the changing energy absorption pattern as wear progresses.

[0069] Among them, energy characteristics may include, but are not limited to, one of the following: total energy, average energy, energy percentage, and energy entropy.

[0070] Thus, in this embodiment, wavelet packet decomposition can accurately extract the frequency domain energy features hidden deep within the response data that are extremely sensitive to wear, thereby improving the accuracy and reliability of identifying the degradation state of hydraulic components and laying the foundation for subsequent accurate prediction of the second remaining lifetime.

[0071] In some embodiments, the method further includes: Time-frequency analysis is performed on the reference response data of the first reference element to obtain the first reference time-frequency characteristics; time-frequency analysis is performed on the reference response data of the second reference element to obtain the second reference time-frequency characteristics, and the remaining service life of the first reference element is greater than the remaining service life of the second reference element; In step 103, the second remaining lifetime is determined based on the current time-frequency characteristics and the reference time-frequency characteristics, including: The second remaining service life is determined based on the current time-frequency characteristics, the first reference time-frequency characteristics, and the second reference time-frequency characteristics.

[0072] For example, the first reference element can be a brand new element, and the second reference element can be an element with zero remaining useful life, that is, a completely obsolete element. Of course, in other examples, the first reference element and the second reference element can also be reference elements with other combinations of remaining useful life, as long as the remaining useful life of the first reference element is greater than the remaining useful life of the second reference element.

[0073] Hereinafter, the first reference element is considered to be a brand new element, and the second reference element is considered to be an element with zero remaining service life.

[0074] Time-frequency characteristics of the first reference are obtained by performing time-frequency analysis on the reference response data of the first reference element, including: Wavelet packet decomposition is performed on the reference pressure response data of a brand-new component to extract the first energy information of the sub-band within a predetermined frequency band, which is used as the first reference time-frequency feature. Time-frequency characteristics of the second reference are obtained by performing time-frequency analysis on the reference response data of the second reference element, including: Wavelet packet decomposition is performed on the reference pressure response data of a component with zero remaining service life to extract the second energy information of the sub-band within the same predetermined frequency band, which is used as the second reference time-frequency feature. Based on the current time-frequency characteristics, the first reference time-frequency characteristics, and the second reference time-frequency characteristics, the second remaining service life is determined, including: The second remaining service life of the hydraulic component is determined based on energy characteristics, first energy information, and second energy information.

[0075] The brand-new components and components with zero remaining service life mentioned here are hydraulic components of the same type and / or model as the hydraulic components described above. It is understood that the brand-new components and components with zero remaining service life are hydraulic components that can be installed on the same working machinery as the aforementioned hydraulic components. The sub-band within the predetermined frequency band here is the same as the sub-band within the predetermined frequency band in the above embodiments.

[0076] Please refer to the following: Figure 3 On the test bench, under the main target working conditions, the reference pressure response data of brand-new hydraulic components when they are stable is collected; and the reference pressure response data of hydraulic components that have reached the end of their technical life due to wear and have zero remaining service life is collected under the same test bench and working conditions.

[0077] For example, the system performs three-level wavelet packet decomposition on the two types of reference pressure response data, converting the time-domain information of the signal to the frequency domain and finely separating it into multiple different sub-bands. It can be understood that the first energy information represents the energy level of a healthy component within that frequency band; the second energy information represents the energy level within that frequency band when the component completely fails.

[0078] Thus, in this embodiment, by quantifying the real-time signal energy from two baseline states, from brand new to scrap, and by utilizing the energy characteristics of the hydraulic component under test, a current continuous life value is estimated so that the continuous life value can be verified later.

[0079] It should be added that the first energy information mentioned above can be at least one of the total energy value, average energy value, energy percentage, and energy entropy of a brand-new component in a sub-band within a predetermined frequency band; the second energy information mentioned above can be at least one of the total energy value, average energy value, energy percentage, and energy entropy of a component with zero remaining service life in a sub-band within a predetermined frequency band; the energy characteristics mentioned above can refer to at least one of the total energy value, average energy value, energy percentage, and energy entropy of the component to be predicted, i.e., the hydraulic component mentioned above, in a sub-band within a predetermined frequency band.

[0080] The following example, taking the first energy information including a first average energy value, the second energy information including a second average energy value, and the energy characteristic including a third average energy value, determines the second remaining service life of the hydraulic component based on the energy characteristic, the first energy information, and the second energy information, including: Based on the third average energy value, the first average energy value, and the second average energy value, the second remaining service life prediction result of the hydraulic component is calculated using formula (1).

[0081] (1) Wherein, H represents the second remaining lifetime prediction result, Emax represents the first energy average value of subband n to subband m within the predetermined frequency band, Emin represents the second energy average value of subband n to subband m within the predetermined frequency band, and Ei represents the third energy average value of subband n to subband m within the predetermined frequency band.

[0082] Thus, in this embodiment, the real-time energy characteristics are compared with the first and second energy information determined through experiments in advance, achieving standardization and dimensionlessness of the measurement results. By establishing a clear, physically based benchmark, regardless of how the overall amplitude of the signal changes, the degree of component degradation can be consistently assessed through its relative position on the benchmark, eliminating deviations caused by sensor differences or system pressure fluctuations, and ensuring the consistency and comparability of prediction results from different devices and at different time points. Furthermore, in this embodiment, by clearly transforming the calculation into a deterministic calculation based on three key scalars—the current time-frequency characteristics, the first reference time-frequency characteristics, and the second reference time-frequency characteristics—complex, black-box reasoning processes are avoided. The calculation logic is clear, transparent, and efficient, reducing the computational burden on the server, making the system response faster, and also facilitating engineers' understanding and verification of the entire prediction process, thus improving the system's engineering practicality and maintainability. In summary, this implementation method transforms lifetime prediction into a stable, standardized, and efficient quantitative calculation process by comparing wavelet packet energy averaging with an absolute benchmark. This significantly enhances the robustness and feasibility of the method in real industrial environments while ensuring the accuracy of the prediction results.

[0083] In some implementations, step S103, determining the second remaining useful life based on the current time-frequency features and the baseline time-frequency features, may further include: inputting the current time-frequency features into a network model trained based on the baseline time-frequency features; obtaining the original output score of the fully connected layer (fc_out) when outputting the result; using the softmax function to convert the score into a probability distribution; and mapping the probability distribution to continuous values ​​from 0 to 100% (e.g., continuous_pred = p1*0 + p2*25% + p4*50% + p5*75% + p6*100%, where continuous_pred is the continuous remaining useful life prediction result, i.e., the output second remaining useful life, and p is the probability distribution of each original score). Of course, in other implementations, methods such as regression head replacement, ordered regression, fuzzy logic, and uncertainty quantization can also be used to calculate the attenuation of the hydraulic component, i.e., the second remaining useful life of the hydraulic component. In short, any method that can be used to calculate the continuous second remaining useful life of the hydraulic component is acceptable, and will not be elaborated further here.

[0084] In this way, by using the current time-frequency characteristics and the baseline time-frequency characteristics to calculate the second remaining service life, such as the percentage of remaining service life, a more intuitive and accurate quantitative assessment can be achieved. This improves the usability and guiding value of the prediction results, as well as their reliability, reducing the risk of false alarms. It also provides timely and reliable assurance for subsequent component maintenance based on the prediction results. Step 104: If the second remaining service life is within the range corresponding to the first remaining service life, determine the second remaining service life as the target remaining service life of the hydraulic component.

[0085] Specifically, the system compares the first remaining lifespan output in step S102 (e.g., x1%) with the second remaining lifespan output in step S103 (e.g., x2%). When x2% is within the range corresponding to x1%, the second remaining lifespan output in step S103 is directly used as the prediction result. Here, x1% can be a range, and x2% can be a specific value.

[0086] Thus, step 104 combines the interval judgment capability of deep learning with the numerical calculation capability of signal processing through a concise and rigorous logical threshold verification, ensuring that only high-confidence prediction results that have undergone double verification will be finally adopted and output, thereby greatly improving the reliability and practicality of the prediction results output by the entire system.

[0087] In summary, the remaining service life prediction method for hydraulic components provided by the above embodiments improves the accuracy of prediction results and reduces the risk of false alarms by constructing a dual path of interval prediction and numerical prediction, and setting up a verification mechanism based on interval prediction and numerical prediction. At the same time, by using response data to perform time-frequency analysis to obtain the current time-frequency characteristics, and calculating the second remaining service life prediction result based on the current time-frequency characteristics and the reference time-frequency characteristics, a more intuitive and certain quantitative assessment can be made, improving the usability and guiding value of the prediction results, as well as the reliability of the prediction results, and providing timeliness and reliability assurance for subsequent component maintenance based on the prediction results.

[0088] In some embodiments, the method further includes: If the second remaining useful life is within the range corresponding to the first remaining useful life, the second remaining useful life is determined to be a valid prediction result. If the second remaining service life in K consecutive valid prediction results is less than the service life threshold, an alarm message is output to prompt the user to handle the abnormal state of the hydraulic component, where K is a positive integer.

[0089] Specifically, when x2% falls within the range corresponding to x1%, the result is considered valid. The system issues an alarm when the remaining lifespan is less than a preset value y% for K consecutive outputs. y% is the percentage threshold, for example, 20%. This can also be understood as different lifespan thresholds corresponding to different hydraulic components, meaning different percentage values ​​of lifespan are required for their ultimate replacement.

[0090] In some implementations, for user convenience, the server displays the aforementioned alarm information, load spectrum of hydraulic components, historical fault information, predicted final service life of hydraulic components, and operating status information of the operating machinery via a network platform.

[0091] It should be added that, in order to facilitate customers and enterprises to perform maintenance or repairs in advance, the above alarm information can not only be displayed on the network platform provided by the prediction system, but also sent to the user's mobile terminal, such as a mobile phone, through the contact information reserved by the user in the system.

[0092] In some implementations, in step S103, after receiving the response data from the hydraulic component, the system processes the data using wavelet packet decomposition to extract the energy values ​​of sub-bands within a predetermined frequency band. Based on these sub-band energy values, a radar chart is generated to visualize the remaining service life of the component and displayed on the network platform. Simultaneously, the uniformized test high-frequency pressure data and the actual vehicle high-frequency pressure data are also converted into image data and displayed on the network platform. This visualizes the prediction process and improves its traceability.

[0093] To further understand the method for predicting the remaining service life of hydraulic components provided in the embodiments of this application, please refer to... Figure 4 , Figure 4 This is a schematic diagram illustrating the application scenario of the remaining useful life prediction web platform provided in the embodiments of this application, such as... Figure 4 As shown, the functions of the web platform provided by the prediction system (hereinafter referred to as the Remaining Useful Life Prediction Web Platform) include: input import and storage; visualization; and result output and early warning.

[0094] The input import and storage sections aim to provide online data import, offline data import, and data storage. The data here can be understood as the response data of the hydraulic components described in the above embodiments. Offline data import can be the remaining service life prediction model described in the above embodiments, and / or a hydraulic component's full lifecycle database, etc. In visualization, icon visualization can refer to displaying radar charts showing the target remaining service life and / or sub-band energy values ​​for hydraulic components, etc., while load spectrum visualization can refer to displaying the load spectrum for the corresponding operating machinery for the hydraulic component. In result output and early warning, prediction result display can show the target remaining service life of the hydraulic component; historical data query can display historical fault information of the hydraulic component; fault early warning refers to issuing a fault warning when K valid results are all below a percentage threshold.

[0095] Thus, the above embodiments provide an online result query interface that integrates prediction result query, abnormal status alarm, and component replacement early warning, so that users can better monitor and manage the remaining service life of hydraulic components.

[0096] It should be noted that the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0097] To achieve the above objectives, embodiments of the present invention also provide a device for predicting the remaining service life of hydraulic components. This device is applied to computer equipment. Please refer to [link to relevant documentation]. Figure 5 The device includes: The receiving module 51 is used to receive response data for the hydraulic components; Prediction module 52 is used to input response data into a pre-trained remaining useful life prediction model to predict the first remaining useful life; The determination module 53 is used to perform time-frequency analysis on the response data to obtain the current time-frequency characteristics, and to determine the second remaining service life based on the current time-frequency characteristics and the reference time-frequency characteristics; The output module 54 is used to determine the second remaining service life as the target remaining service life of the hydraulic component when the second remaining service life is within the range corresponding to the first remaining service life.

[0098] In some embodiments, the device further includes: a first acquisition module, used to acquire the working condition status information of the working machinery corresponding to the hydraulic component; the receiving module 51 is further used to receive the response data of the hydraulic component when the working condition status information is determined to be the target working condition.

[0099] In some embodiments, the apparatus further includes: The second acquisition module is used to acquire bench test data and vehicle operation data of sample hydraulic components at different remaining service life stages. Both bench test data and vehicle operation data include sample response data of sample hydraulic components with remaining service life labels. The first module is used to train the neural network model using bench test data as the training set to obtain the basic model. The second module is used to perform transfer learning on the base model using real vehicle operation data as a transfer set, and obtain the remaining service life prediction model.

[0100] In some embodiments, the device further includes: a third acquisition module, configured to acquire sample working condition status information of the sample hydraulic component corresponding to the sample working machinery under different working conditions; and a second acquisition module, further configured to: determine the target working condition of the sample hydraulic component corresponding to the sample working machinery under different working conditions based on the sample working condition status information of the sample hydraulic component corresponding to the working machinery under different working conditions; acquire test sample response data and actual vehicle sample response data of the sample hydraulic component at different remaining service life stages under the target working condition; and acquire bench test data and actual vehicle operation data of the sample hydraulic component at different remaining service life stages based on the test sample response data and actual vehicle sample response data of the sample hydraulic component at different remaining service life stages under the target working condition.

[0101] In some embodiments, the second acquisition module is further configured to acquire a first number of test sample response data of sample hydraulic components at different remaining service life stages under the target operating condition, and to acquire a second number of real vehicle sample response data of sample hydraulic components at different service life stages under the target operating condition, wherein the second number is less than the first number.

[0102] In some embodiments, the apparatus further includes: The first extraction module is used to perform time-frequency analysis on the reference response data of the first reference element to obtain the first reference time-frequency characteristics; The second extraction module is used to perform time-frequency analysis on the reference response data of the second reference element to obtain the time-frequency characteristics of the second reference element, wherein the remaining service life of the first reference element is greater than the remaining service life of the second reference element. The determining module 53 is also used to determine the second remaining service life based on the current time-frequency characteristics, the first reference time-frequency characteristics, and the second reference time-frequency characteristics.

[0103] In some implementations, the determining module 53 is further configured to perform wavelet packet decomposition on the response data and extract the energy features of sub-bands within a predetermined frequency band as the current time-frequency features.

[0104] In some embodiments, the device further includes an alarm information output device, used to determine the second remaining service life as a valid prediction result when the second remaining service life is within the range corresponding to the first remaining service life; if the second remaining service life in K consecutive valid prediction results is less than the service life threshold, then an alarm information is output to prompt the user to handle the abnormal state of the hydraulic component, where K is a positive integer.

[0105] It should be noted that the description of the hydraulic component remaining service life prediction device above is similar to the description of the hydraulic component remaining service life prediction method above, and the beneficial effects of the same method will not be repeated. For technical details not disclosed in the embodiments of the hydraulic component remaining service life prediction device of the present invention, please refer to the description of the embodiments of the hydraulic component remaining service life prediction method of the present invention.

[0106] Based on the same inventive concept as the foregoing embodiments, this invention provides a computer device, such as... Figure 6 As shown, the computer device includes: a processor 610 and a memory 611 storing computer programs; wherein, Figure 6 The processor 610 shown in the diagram does not indicate that there is only one processor 610, but only indicates the positional relationship of the processor 610 relative to other devices. In practical applications, there can be one or more processors 610; similarly, Figure 6 The memory 611 shown in the diagram has the same meaning, that is, it is only used to indicate the positional relationship of memory 611 relative to other devices. In practical applications, there can be one or more memories 611. When the processor 610 runs the computer program, it implements the method for predicting the remaining service life of hydraulic components described above.

[0107] The computer device may also include at least one network interface 612. Various components of the computer device are coupled together via a bus system 613. It is understood that the bus system 613 is used to implement communication between these components. In addition to a data bus, the bus system 613 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in… Figure 6 The general designated all buses as Bus System 613.

[0108] Based on the same inventive concept as the foregoing embodiments, this embodiment also provides a computer-readable storage medium storing a computer program. The computer-readable storage medium can be a magnetic random access memory (FRAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM), etc.; it can also be various devices including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc. When the computer program stored in the computer-readable storage medium is run by a processor, it implements the above-described process solution generation method. For the specific steps implemented when the computer program is executed by the processor, please refer to [link to relevant documentation]. Figures 1-4 The description of the illustrated embodiments will not be repeated here.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method of predicting the remaining useful life of a hydraulic component, characterized by, The method comprises: receiving response data of the hydraulic element; inputting the response data into a pre-trained remaining useful life prediction model to predict a first remaining useful life; performing time-frequency analysis on the response data to obtain a current time-frequency feature, and determining a second remaining useful life based on the current time-frequency feature and a reference time-frequency feature; in a case where the second remaining useful life is within a range corresponding to the first remaining useful life, determining the second remaining useful life as a target remaining useful life of the hydraulic element.

2. The method of claim 1, wherein, The method further comprises: obtaining working condition state information of a working machine corresponding to the hydraulic element; The method further comprises: receiving the response data of the hydraulic element when the working condition state information is determined as a target working condition.

3. The method of claim 1, wherein, The method further comprises: obtaining bench test data and real vehicle operation data of sample hydraulic elements in different remaining useful life stages, wherein the bench test data and the real vehicle operation data both comprise sample response data of the sample hydraulic elements with remaining life labels; training a neural network model using the bench test data as a training set to obtain a base model; performing transfer learning on the base model using the real vehicle operation data as a transfer set to obtain the remaining useful life prediction model.

4. The method of claim 3, wherein, The method further comprises: obtaining sample working condition state information of sample working machines corresponding to sample hydraulic elements under different working conditions; The method further comprises: determining a target working condition of the sample working machines corresponding to the sample hydraulic elements according to the sample working condition state information of the sample working machines corresponding to the sample hydraulic elements under different working conditions; obtaining test sample response data and real vehicle sample response data of the sample hydraulic elements in different remaining useful life stages under the target working condition respectively; obtaining the bench test data and the real vehicle operation data of the sample hydraulic elements in different remaining useful life stages according to the test sample response data and the real vehicle sample response data of the sample hydraulic elements in different remaining useful life stages under the target working condition.

5. The method of claim 4, wherein, The method further comprises: obtaining a first number of test sample response data of the sample hydraulic elements in different remaining useful life stages under the target working condition, and obtaining a second number of real vehicle sample response data of the sample hydraulic elements in different remaining useful life stages under the target working condition, wherein the second number is less than the first number.

6. The method of claim 1, wherein, The method further comprises: performing time-frequency analysis on reference response data of a first reference element to obtain a first reference time-frequency feature, and performing time-frequency analysis on reference response data of a second reference element to obtain a second reference time-frequency feature, wherein the remaining useful life of the first reference element is greater than the remaining useful life of the second reference element; The method further comprises: determining the second remaining useful life based on the current time-frequency feature and the reference time-frequency feature. According to the current time-frequency feature, the first reference time-frequency feature and the second reference time-frequency feature, the second remaining service life is determined.

7. The method of claim 1, wherein, the time-frequency analysis of the response data to obtain a current time-frequency feature comprises: wavelet packet decomposition is performed on the response data to extract energy features of sub-bands in a predetermined frequency band range as the current time-frequency feature.

8. The method of claim 1, wherein, The method further comprises: in a case where the second remaining service life is located in an interval range corresponding to the first remaining service life, determining that the second remaining service life is a valid prediction result; if the second remaining service life in the K consecutive valid prediction results is less than the life threshold, an alarm information is output to prompt the user to handle the abnormal state of the hydraulic element, wherein K is a positive integer.

9. A computer device, comprising: comprises: a processor and a memory storing a computer program, when the processor runs the computer program, the remaining service life prediction method of the hydraulic element in any one of claims 1 to 8 is realized.

10. A computer-readable storage medium, characterized in that, The computer program is stored in the computer readable storage medium, and the computer program is executed by the processor to realize the remaining service life prediction method of the hydraulic element in any one of claims 1 to 8.

Images