A method and system for predicting degradation of a heavy vehicle hydraulic system

Abstract

This invention relates to a method and system for predicting the degradation of a heavy-duty vehicle hydraulic system. The method involves transmitting real-time data on the vehicle's hydraulic system, including oil pressure, oil temperature, and flow rate, to the cloud. Degradation features are extracted sequentially in the cloud, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature, and flow rate data, yielding three degradation indices: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate. The extracted degradation feature data is then classified, and the extracted oil pressure, oil temperature, and flow rate data are labeled with different degradation stages. These labels, along with the degradation feature data, are used as input to a pre-built degradation prediction model for training. The three degradation indices, the degradation categories obtained from the degradation classification, and the degradation stage labels are then input into the trained degradation prediction model to output the degradation prediction result, thus completing the prediction. This invention effectively improves prediction accuracy, saves costs, and enables real-time prediction of vehicle performance during operation.

Description

Technical Field

[0001] This invention relates to the field of vehicle hydraulic system function prediction technology, and in particular to a method for predicting the degradation of heavy vehicle hydraulic systems based on data-driven cloud computing. Background Technology

[0002] With the continuous development and application of industrial technology, hydraulic transmission technology has significant advantages in the field of heavy machinery. Compared with mechanical transmission, gas transmission, and electric transmission, hydraulic transmission technology offers higher safety and reliability. Therefore, hydraulic systems are widely used in the automotive industry and are a crucial component of vehicle transmission systems. Furthermore, the working environment of hydraulic systems in heavy vehicle transmissions is generally harsh, leading to the gradual induction and accumulation of blockages and leaks. Given the extremely limited space for test sensor placement, the prediction of system degradation has become a bottleneck in the industry. Against this backdrop, exploring degradation prediction methods for heavy vehicle hydraulic systems based on real-vehicle data is of great significance and application value.

[0003] Existing technologies mostly employ the following methods for predicting and analyzing the degradation of hydraulic systems: 1) Physical model-based methods: Establishing a comprehensive simulation model to describe the physical characteristics and degradation modes of the hydraulic system, primarily using AMESim hydraulic simulation software. 2) Statistical theory-based methods: Establishing probabilistic models of parameter changes and failure loss based on historical data of the hydraulic system. Comparing the current multi-parameter probabilistic state space of the hydraulic system with the established probabilistic model to determine the current health status of the hydraulic system and analyze and evaluate its degradation trend. Commonly used methods include Bayesian networks, support vector machines, and support vector data description. 3) Information fusion-based methods: For complex systems like hydraulic transmission systems, relying solely on individual sensors often fails to obtain sufficient equipment operating information. The performance degradation assessment and prediction results obtained based on this information are also inaccurate. Utilizing multiple sensors to extract information from the hydraulic system and fusing it to obtain a more accurate equipment operating status, combined with artificial intelligence methods, can effectively improve the accuracy of the assessment. However, each of the aforementioned methods has its own advantages and disadvantages. The physical model-based method requires the establishment of a relatively comprehensive mathematical model, which is quite difficult for complex systems such as hydraulic transmission systems. Moreover, once the equipment is updated and iterated, the model will lose its original value. On the other hand, the method based on statistical theory, artificial intelligence and information fusion has high requirements for the hardware properties of the equipment. Otherwise, problems such as slow calculation speed or lag will occur. In addition, it requires a large amount of full life cycle data. More comprehensive data is difficult to obtain at the current stage.

[0004] Therefore, the evaluation results obtained by the above methods are often not accurate enough, the evaluation models are easily eliminated, and a lot of manpower and resources are required to inspect the vehicles, which cannot achieve real-time evaluation of the vehicles in operation. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this invention is to provide a method and system for predicting the degradation of hydraulic systems in heavy vehicles, which can effectively improve prediction accuracy, save costs, and achieve real-time prediction of vehicle operation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for predicting the degradation of a heavy-duty vehicle hydraulic system, comprising: transmitting real-time collected data on oil pressure, oil temperature, and flow rate of the vehicle hydraulic system to the cloud; sequentially extracting degradation features in the cloud; evaluating the hydraulic system performance based on the oil pressure, oil temperature, and flow rate data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate; classifying the extracted degradation feature data; labeling the extracted oil pressure, oil temperature, and flow rate data with different degradation stages; and using these labels, along with the degradation feature data, as input to a pre-constructed degradation prediction model to train the degradation prediction model; inputting the three degradation indicators, the degradation categories obtained from the degradation classification, and the degradation stage labels into the trained degradation prediction model; and outputting the degradation prediction result to complete the prediction.

[0007] Furthermore, hydraulic system performance evaluation is conducted based on oil pressure, oil temperature, and flow rate data. This includes: adopting a hierarchical composite fusion processing structure: a three-level progressive evaluation system of index—performance—system, with the three levels divided into system-level degradation evaluation, performance-level degradation evaluation, and index-level degradation evaluation. System-level degradation evaluation is derived from the fusion evaluation of various performance characteristics of the system, and the specific performance characteristics representing the service capability of the hydraulic system together constitute the performance level of the index system. Performance-level degradation evaluation is determined by the evaluation of a single index or by the fusion evaluation of multiple indices, and the indices representing each performance together constitute the index level. The index level covers the degradation characteristics of each system performance evaluation and comprehensively considers the index evaluation to obtain the corresponding performance layer degradation. Index-level degradation evaluation analyzes the extracted features to obtain the similarity of oil pressure curves, oil temperature curves, and lubrication flow rate.

[0008] Furthermore, the extracted degradation feature data is classified into degradation categories, including: dividing the extracted degradation feature data into a vector based on the dynamic similarity, amplitude, and time dimensions of pressure, flow rate, and temperature, to represent the comparison results between the normal state data of the hydraulic system and the data of a certain degradation stage; for each dimension, if the degradation feature extracted from the collected data is the same as the standard feature, the value is 1, otherwise it is 0.

[0009] Furthermore, K-means clustering is used for degenerate classification. Using k as the input parameter, the set of n objects is divided into k clusters, resulting in high similarity within clusters and low similarity between clusters. Cluster similarity is a measure of the mean of the objects within the cluster, considered as the centroid or centroid of the cluster; including:

[0010] Select k optimized initial objects, each representing the initial mean or center of a cluster;

[0011] For each remaining object, assign it to the most similar cluster based on its distance from the cluster mean;

[0012] Calculate the new mean for each cluster;

[0013] Repeat the above process until the criterion function converges.

[0014] Furthermore, the criterion function follows the rule of minimum sum of squared errors;

[0015] The sum of variance (SSE) value represents the proximity of feature data points to the centroid. The smaller the SSE value, the closer the data points are to the centroid, and the better the clustering effect.

[0016] When the SSE value is greater than the preset range, but not close to the centroid of other degradation stage feature data, and accumulates to the preset data amount, then some feature data characterize a new degradation stage or degradation type. At this time, combined with the actual vehicle mileage of the data source, the feature data is sorted in chronological order to form a new degradation stage or degradation type.

[0017] Furthermore, the pre-constructed degradation prediction model is a long short-term memory neural network model; the output of this model consists of two types of data: degradation stage and degradation category, and the number of hidden layers in the middle is calculated to be 11 according to Kolmogorov's theorem.

[0018] Furthermore, the Long Short-Term Memory Neural Network model incorporates the sigmoid function and the hyperbolic tangent function into the network, including forget gates, input gates, and output gates;

[0019] The forget gate has σ levels, with input C. t-1 ,X t, Then in C t-1 Output a specific value, within the range of 0 to 1; where 0 and 1 correspond to two special states: complete forgetting and complete retention, respectively; where C t-1 X represents the cell state at t-1. t This represents the input layer matrix at time t;

[0020] The input gate selects the data required by the model, and the σ layer of the gate determines the value to be updated. A new candidate value is generated by the hyperbolic tangent layer and added to the neuron state.

[0021] The process of performing neuron state updates starts from C. t-1 To C t , will f t Multiply by the old state to add a new candidate value; where Ct The cell state at time t;

[0022] The output gate determines the final output value.

[0023] A degradation prediction system for heavy-duty vehicle hydraulic systems includes: a first processing module that transmits real-time collected data on oil pressure, oil temperature, and flow rate of the vehicle's hydraulic system to the cloud, extracts degradation features sequentially in the cloud, and evaluates the hydraulic system performance based on the oil pressure, oil temperature, and flow rate data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate; a second processing module that classifies the extracted degradation feature data, labels the extracted oil pressure, oil temperature, and flow rate data with different degradation stages, and uses these labels, along with the degradation feature data, as input to a pre-built degradation prediction model to train the degradation prediction model; and a prediction output module that inputs the three degradation indicators, the degradation categories obtained from the degradation classification, and the degradation stage labels into the trained degradation prediction model, outputs the degradation prediction results, and completes the prediction.

[0024] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform any of the methods described above.

[0025] A computing device includes: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0026] The present invention has the following advantages due to the adoption of the above technical solutions:

[0027] 1. The performance evaluation system for hydraulic systems of heavy-duty vehicle transmission devices constructed in this invention possesses strong rationality, comprehensiveness, and universality. It can be applied to the performance evaluation of hydraulic systems of most heavy-duty vehicle transmission devices and has the basic conditions for technology transfer. This hydraulic system performance evaluation system is divided into three levels: system level, performance level, and index level. It can comprehensively characterize the performance of the hydraulic system of the vehicle transmission device during operation and uses a Long Short-Term Memory (LSTM) neural network to rationally allocate index weights.

[0028] 2. This invention establishes a condition monitoring system for the hydraulic transmission device of heavy vehicles based on cloud computing and user terminals. The hydraulic system's measuring points are arranged in a reasonable and comprehensive manner, enabling data sharing among operating vehicles. The status of vehicles in different regions can be aggregated and viewed uniformly. Moreover, even with a large number of vehicles, it can efficiently grasp the status information of each vehicle without excessive investment of manpower and resources, allowing for the timely prevention and resolution of degradation faults in the vehicle's hydraulic system. Data is categorized and stored using a cloud database, while the local machine is only responsible for data collection and processing. Because large amounts of data do not occupy local memory, local data processing is faster, and real-time data sharing is not limited by geographical location.

[0029] 3. This invention employs a Long Short-Term Memory (LSTM) neural network algorithm to construct a degradation prediction model and uses the K-means algorithm to classify degradation features. These two algorithms exhibit strong matching compatibility with the data features of hydraulic transmission systems. The model's computational sample spans a long period and involves a massive amount of data; the unique memory units of LSTM address the issues of high computational cost and prediction accuracy. If new degradation types emerge in the future due to advancements in vehicle hydraulic technology, these will be reflected in the data, consequently altering the weight allocation of the prediction model. Therefore, this prediction method demonstrates strong adaptability to related technological developments. Attached Figure Description

[0030] Figure 1 This is a schematic flowchart of a method for predicting the degradation of a heavy vehicle hydraulic system according to an embodiment of the present invention;

[0031] Figure 2 This is a schematic diagram of the signal acquisition subsystem structure in one embodiment of the present invention;

[0032] Figure 3 This is a schematic diagram of signal processing and feature extraction in one embodiment of the present invention;

[0033] Figure 4 This is a waveform diagram of oil pressure and oil temperature in one embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of the feature extraction method in one embodiment of the present invention;

[0035] Figure 6 This is a schematic diagram of hydraulic system performance evaluation in one embodiment of the present invention;

[0036] Figure 7 This is a schematic diagram of a long short-term memory neural network model structure in one embodiment of the present invention;

[0037] Figure 8 This is a schematic diagram of the input-output hierarchical structure of a long short-term memory neural network in one embodiment of the present invention;

[0038] Figure 9 This is a schematic diagram of a long short-term memory neural network model prediction in one embodiment of the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0040] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0041] This invention provides a method and system for predicting the degradation of a heavy-duty vehicle hydraulic system. The method transmits real-time data on oil pressure, oil temperature, and flow rate of the vehicle's hydraulic system to the cloud. Degradation features are extracted sequentially in the cloud, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature, and flow rate data, yielding three degradation indicators: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate. The extracted degradation feature data is then classified, and the extracted oil pressure, oil temperature, and flow rate data are labeled with different degradation stages. These labels, along with the degradation feature data, are used as input to a pre-built degradation prediction model for training. The three degradation indicators, the degradation categories obtained from the degradation classification, and the degradation stage labels are then input into the trained degradation prediction model to output the degradation prediction result, completing the prediction. This invention effectively improves prediction accuracy, saves costs, and enables real-time prediction of vehicle performance during operation.

[0042] In one embodiment of the present invention, a method for predicting the degradation of a heavy-duty vehicle hydraulic system is provided. In this embodiment, as... Figure 1 As shown, the method includes the following steps:

[0043] 1) The real-time collected data of oil pressure, oil temperature and flow rate of the vehicle hydraulic system are transmitted to the cloud. Degradation features are extracted sequentially in the cloud, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature and flow rate data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity and lubrication flow rate.

[0044] 2) Classify the extracted degradation feature data, label the extracted oil pressure, oil temperature and flow data with different degradation stages, and use them together with the degradation feature data as input to the pre-built degradation prediction model to train the degradation prediction model.

[0045] 3) Input the three degradation indicators, degradation categories obtained from degradation classification, and degradation stage labels into the trained degradation prediction model, output the degradation prediction results, and complete the prediction.

[0046] In step 1) above, the signal acquisition subsystem collects real-time data on the oil pressure, oil temperature, and flow rate of the vehicle's hydraulic system. The signal acquisition subsystem mainly includes a data acquisition card and sensors, including pressure sensors, temperature sensors, and flow meters. The installation locations of each sensor are as follows: Figure 2 As shown in the diagram. Temperature sensors are installed on the radiator pipes, the pipes between the pressure tank and the oil pump, and the return oil tank pipe; pressure sensors are installed on the pipes between the pressure tank and the constant pressure valve, the pipes between the check valve and the control valve assembly, the pump motor pipes, and the pipes between the constant pressure valve and the pump motor control valve; flow meters are installed on the hydraulic torque converter, the oil pump, the pump motor, the control valve assembly, and the viscous clutch.

[0047] In step 1) above, data is transmitted to the cloud via TCP / IP protocol to achieve data sharing. Furthermore, this embodiment uses MATLAB for degradation feature extraction, such as... Figure 3 As shown.

[0048] Specifically, data on oil pressure, oil temperature, and lubrication flow of the main components of the vehicle's hydraulic system, such as the hydraulic pump and pump motor, are collected. Then, feature extraction is performed on the collected data, and finally, the extracted features are converted into a new format and uploaded to a cloud database.

[0049] Based on the functional requirements of the signal acquisition subsystem, the DAQ assistant in MATLAB is first used to establish a connection with the NI acquisition card and configure its information. Then, the sensors are connected to the acquisition card to acquire pressure, temperature, and flow signals from various hydraulic components. The pressure signals are divided into filling oil pressure, drain oil pressure, lubricating oil pressure, and control oil pressure signals. The sampling frequency, sampling time, and operating condition can be set, and the oil pressure and temperature waveforms can be displayed. Figure 4 As shown.

[0050] like Figure 5As shown, the main purpose of feature extraction is to send the degradation indicators obtained from the feature extraction of disordered data to the classifier for classification. When the extracted features are abnormal, the degradation type assigned by the allocator will be provided and an alarm will be displayed to the user. After the signal passes through the feature extraction module, the data needs to be transmitted to the cloud. In the data transmission, the transmitted data is first converted into a unified JSON data format, and then the data is uploaded to the cloud database through the TCP / IP protocol.

[0051] In step 1) above, in this embodiment, the overall system degradation evaluation is directly affected by the degradation of its main internal components, such as the hydraulic pump, steering pump motor, and control valve assembly. These main components collectively determine the overall system degradation. The system evaluation is influenced by multiple performance evaluations. Performance evaluations are characterized by dispersed targeting and reliability, and require a large set of parameters for accumulating target features, such as system stability and transmission efficiency. The basis for performance evaluation comes from feature indicators extracted from data collected by sensors. The source of hydraulic system feature parameters includes data from oil temperature and oil pressure sensors. This constitutes a three-tiered progressive evaluation system of indicators—performance—system. The reliability, focus, and time required for target attribute evaluation differ among the three levels. Therefore, based on the application requirements of target attribute fusion evaluation, a layered composite fusion processing structure is adopted. The layered fusion framework verifies the target level in conjunction with the specific work of the target. Based on the different characteristics of the three levels, they are divided into system level, performance level, and indicator level. Figure 6 As shown, the hydraulic system performance is evaluated based on oil pressure, oil temperature, and flow rate data, specifically as follows:

[0052] A hierarchical composite integrated processing structure is adopted: a three-level progressive evaluation system of indicators, performance and system, and the three levels are divided into system-level degradation evaluation, performance-level degradation evaluation and indicator-level degradation evaluation;

[0053] System-level degradation evaluation is derived from the integrated evaluation of various performance characteristics of the system. Specific performance characteristics representing the service capability of the hydraulic system collectively constitute the performance levels of the indicator system. Each performance level comprehensively acquires evaluation aspects such as the durability, component working quality, transmission efficiency, and stability of typical components. Through analysis and reasoning of these performance characteristics, system-level degradation can be inferred. Examples include the hydraulic system's oil supply capacity, pump motor power transmission capacity, oil pressure stability, and oil temperature stability. Performance-level degradation evaluation is based on a holistic assessment of corresponding characteristic indicators.

[0054] Performance-level degradation evaluation is determined by the evaluation of a single indicator or by the fusion of multiple indicators. The indicators characterizing each performance level together constitute the indicator level, which covers the degradation characteristics of the performance evaluation of each system. The corresponding performance layer degradation is obtained by comprehensively considering the indicator evaluation. For example, the oil supply capacity of a hydraulic system is obtained by the fusion evaluation of the oil pump volumetric efficiency and lubrication flow rate; the power transmission capacity of the pump motor is obtained by the fusion evaluation of the pump motor speed, pump motor volumetric efficiency, and pump motor torque; the oil pressure stability is obtained by the fusion evaluation of the oil pressure similarity during gear shifting, the oil pressure similarity under specific working conditions, and the oil pressure characteristic parameter group; and the oil temperature stability is determined by the oil temperature similarity under specific working conditions.

[0055] The index-level degradation evaluation analyzes the extracted features to obtain the similarity of oil pressure curves, oil temperature curves, and lubrication flow. All the sensors with extracted features together constitute the data level.

[0056] In step 2) above, the extracted degradation feature data is classified as degradation, specifically as follows:

[0057] The extracted degradation feature data is divided into a vector based on nine dimensions, including dynamic similarity of pressure, flow rate, and temperature, amplitude, and time, to represent the comparison result between the normal state data of the hydraulic system and the data of a certain degradation stage.

[0058] For each dimension, if the degenerate feature extracted from the collected data is the same as the standard feature, the value is 1; otherwise, it is 0.

[0059] In this embodiment, K-means clustering is used for degenerate classification. Using k as the input parameter, the set of n objects is divided into k clusters, resulting in high similarity within clusters and low similarity between clusters. The cluster similarity is a measure of the mean of the objects within the cluster, considered as the centroid or centroid of the cluster. The method includes the following steps:

[0060] 2.1) Select k optimized initial objects, each representing the initial mean or center of a cluster;

[0061] 2.2) For each remaining object, assign it to the most similar cluster based on its distance from the mean of each cluster;

[0062] 2.3) Calculate the new mean for each cluster;

[0063] 2.4) Repeat the above process until the criterion function converges.

[0064] The criterion function is the minimum sum of squared errors rule J(c, μ), which is defined as follows:

[0065]

[0066] In the formula, x (i)μ represents the feature sample data point value. c (i) represents the centroid of the feature sample data in the i-th degradation stage. When the clustering model cost function is minimized, the feature sample data within each degradation stage are most similar, and the squared error between them and the centroid is minimized. By calculating the sum of the squared errors of the feature sample data in all degradation stages, we can verify whether the clustering into k degradation stages is the most reasonable. In other words, for each object in a cluster, we calculate the squared distance from the object to its cluster center and then sum them. This criterion attempts to make the generated K resulting clusters as compact and independent as possible.

[0067] The sum of squares (SSE) and variance (SSE) values ​​represent the proximity of feature data points to the centroid. A smaller SSE value indicates closer proximity to the centroid and better clustering. When the SSE value exceeds a preset range, is not close to the centroid of other degradation stage feature data, and accumulates to a preset data volume, some feature data represent a new degradation stage or type. In this case, combined with the actual vehicle mileage from the data source, the feature data is sorted chronologically to form a new degradation stage or type. The vector of comparison results between all standard state data and degradation data is entered programmatically. Partial data is shown in Table 1.

[0068] Table 1 Partial Hydraulic System Input Data

[0069]

[0070] In step 2) above, the pre-constructed degradation prediction model is a long short-term memory neural network model. The input data for this model consists of three degradation indicators obtained from the feature extraction module: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate, as well as degradation category and degradation stage labels obtained from the classifier. The output consists of two categories of data: degradation stage and degradation category. Based on Kolmogorov's theorem, the number of hidden layers is calculated to be 11. Figure 7 As shown.

[0071] Long Short-Term Memory (LSTM) neural networks are a type of recurrent neural network. They are characterized by self-recalling based on time series and can predict data. However, ordinary recurrent neural networks suffer from gradient explosion and vanishing gradients due to backpropagation across time steps, making them unsuitable for predicting data with large time spans. In contrast, neurons in LSM models are replaced by storage units, giving the model a memory unit that enables training and prediction of sequences with large time spans.

[0072] In this embodiment, the long short-term memory neural network model incorporates the sigmoid function and the hyperbolic tangent function into the network, including forget gate, input gate, and output gate; it achieves long-term data retention through summation operations, thus showing significant advantages in applications.

[0073] like Figure 8 As shown, the forget gate has σ layers, with input C. t-1 ,X t, Then in C t-1 Output a specific value, within the range of 0 to 1; where 0 and 1 correspond to two special states: complete forgetting and complete retention, respectively; where C t-1 X represents the cell state at t-1. t This represents the input layer matrix at time t;

[0074] The specific formula is as follows:

[0075] f t =σ(W f [h t-1 ,x t ]+b f )

[0076] In the formula: f t Here, σ is the forget gate at time t; σ is the sigmoid function; x t h t-1 W represents the input layer and hidden layer outputs at time t and t-1, respectively; f b is the corresponding weight matrix; f This is the corresponding bias term.

[0077] The input gate selects the data required by the model, and the σ layer of the gate determines the value to be updated. A new candidate value is generated by the hyperbolic tangent layer and added to the neuron state.

[0078] The specific calculation formula is as follows:

[0079] i t =σ(w i [h t-1 ,x t ]+b i )

[0080] C′ t =tanh(w c [h t-1 ,x t ]+b c )

[0081] In the formula: i t C′ is the input gate at time t; t The updated neuronal cell state; w i w c b is the corresponding weight matrix; i b c This is the corresponding bias term.

[0082] The process of performing neuron state updates starts from C. t-1 To C t , will f t Multiply by the old state to add a new candidate value; where C t The cell state at time t;

[0083] Specifically, the formula is:

[0084] C t =f t ×C t-1 +i t ×C′ t

[0085] In the formula: C t The cell state at time t.

[0086] The output gate determines the final output value:

[0087] O t =σ(w o [h t-1 ,x t ]+b o )

[0088] h t =O t ×tanh(C t )

[0089] In the formula: O t The output gate at time t; w o b is the corresponding weight matrix; o This is the corresponding bias term.

[0090] First, data preprocessing is performed. Missing values ​​are imputed using mean-based methods, and dimensionality reduction of the index parameters is applied to several types of feature data from the hydraulic system. The simplified standard data is then passed to the input layer. The input layer receives data that is divided into training and testing datasets. The hidden layer constructs a network model based on LSTM. Finally, the prediction results are output, such as... Figure 9 As shown.

[0091] In summary, this invention transmits the collected data to the cloud in real time for processing, enabling big data computation. By embedding degradation data classification and long short-term memory neural network prediction into the cloud, real-time computation of shared data across the platform is achieved. This invention enables data sharing among operating vehicles, allowing for the aggregation and unified viewing of vehicle status across different regions. Furthermore, even with a large number of vehicles, it efficiently monitors the status of each vehicle without requiring excessive manpower or resources, allowing for the timely prevention and resolution of degradation faults in the vehicle's hydraulic system.

[0092] In one embodiment of the present invention, a degradation prediction system for a heavy-duty vehicle hydraulic system is provided, comprising:

[0093] The first processing module transmits the real-time collected oil pressure, oil temperature and flow data of the vehicle hydraulic system to the cloud. In the cloud, degradation features are extracted sequentially, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature and flow data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity and lubrication flow.

[0094] The second processing module performs degradation classification on the extracted degradation feature data, labels the extracted oil pressure, oil temperature and flow data with different degradation stages, and uses them together with the degradation feature data as input to a pre-built degradation prediction model to train the degradation prediction model.

[0095] The prediction output module takes the three degradation indicators, the degradation category obtained from degradation classification, and the degradation stage label as input to the trained degradation prediction model, and outputs the degradation prediction result to complete the prediction.

[0096] The system provided in this embodiment is used to execute the above-described method embodiments. For specific processes and details, please refer to the above embodiments, which will not be repeated here.

[0097] A schematic diagram of a computing device structure is provided in one embodiment of the present invention. The computing device can be a terminal, which may include: a processor, a communication interface, memory, a display screen, and an input device. The processor, communication interface, and memory communicate with each other via a communication bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. When executed by the processor, the computer program implements a method for predicting the degradation of a heavy-duty vehicle hydraulic system. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, a management network, NFC (Near Field Communication), or other technologies. The display screen can be a liquid crystal display or an e-ink display. The input device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad mounted on the casing of the computing device, or an external keyboard, touchpad, or mouse. The processor can call logical instructions stored in the memory.

[0098] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] In one embodiment of the present invention, a computer program product is provided, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to perform the methods provided in the above-described method embodiments.

[0100] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, which stores server instructions that cause a computer to perform the methods provided in the above embodiments.

[0101] The computer-readable storage medium provided in the above embodiments has a similar implementation principle and technical effect to the above method embodiments, and will not be described again here.

[0102] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0103] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0104] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the degradation of a heavy-duty vehicle hydraulic system, characterized in that, include: The real-time collected data on oil pressure, oil temperature, and flow rate of the vehicle's hydraulic system is transmitted to the cloud. Degradation features are extracted sequentially in the cloud, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature, and flow rate data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity, and lubrication flow rate. The extracted degradation feature data is classified into degradation categories. The extracted oil pressure, oil temperature and flow data are labeled with different degradation stages and used together with the degradation feature data as input to a pre-built degradation prediction model to train the degradation prediction model. Input the three degradation indicators, degradation categories obtained from degradation classification, and degradation stage labels into the trained degradation prediction model, output the degradation prediction results, and complete the prediction; Hydraulic system performance is evaluated based on oil pressure, oil temperature, and flow rate data, including: A hierarchical composite integrated processing structure is adopted: a three-level progressive evaluation system of indicators, performance and system, and the three levels are divided into system-level degradation evaluation, performance-level degradation evaluation and indicator-level degradation evaluation; The system-level degradation evaluation is derived from the integrated evaluation of various performance characteristics of the system. The specific performance characteristics that characterize the service capability of the hydraulic system together constitute the performance level of the index system. Performance-level degradation evaluation is determined by the evaluation of a single indicator or by the fusion of multiple indicators. The indicators that characterize each performance level together constitute the indicator level. The indicator level covers the degradation characteristics of each system performance evaluation and comprehensively considers the indicator evaluation to obtain the corresponding performance layer degradation. The index-level degradation evaluation involves analyzing the extracted features to obtain the similarity of the oil pressure curve, the similarity of the oil temperature curve, and the lubrication flow rate.

2. The method for predicting the degradation of a heavy-duty vehicle hydraulic system as described in claim 1, characterized in that, The extracted degradation feature data is classified into degradation categories, including: The extracted degradation feature data is divided into a vector based on the dynamic similarity, amplitude, and time dimension of pressure, flow rate, and temperature, representing the comparison result between the normal state data of a hydraulic system and the data of a certain degradation stage. For each dimension, if the degenerate feature extracted from the collected data is the same as the standard feature, the value is 1; otherwise, it is 0.

3. The method for predicting the degradation of a heavy-duty vehicle hydraulic system as described in claim 2, characterized in that, Degenerate classification is performed using K-means clustering, with k as the input parameter. The set of n objects is divided into k clusters, resulting in high similarity within clusters and low similarity between clusters. Cluster similarity is a measure of the mean of the objects within the cluster, considered as the centroid or centroid of the cluster; including: Select k optimized initial objects, each representing the initial mean or center of a cluster; For each remaining object, assign it to the most similar cluster based on its distance from the cluster mean; Calculate the new mean for each cluster; Repeat the above process until the criterion function converges.

4. The method for predicting the degradation of a heavy-duty vehicle hydraulic system as described in claim 3, characterized in that, The criterion function is the rule of minimum sum of squared errors; The sum of variance (SSE) value represents the proximity of feature data points to the centroid. The smaller the SSE value, the closer the data points are to the centroid, and the better the clustering effect. When the SSE value is greater than the preset range, but not close to the centroid of other degradation stage feature data, and accumulates to the preset data amount, then some feature data characterize a new degradation stage or degradation type. At this time, combined with the actual vehicle mileage of the data source, the feature data is sorted in chronological order to form a new degradation stage or degradation type.

5. The method for predicting the degradation of a heavy-duty vehicle hydraulic system as described in claim 1, characterized in that, The pre-built degradation prediction model is a long short-term memory neural network model; the output of the model consists of two types of data: degradation stage and degradation category, and the number of hidden layers in the middle is calculated to be 11 according to Kolmogorov's theorem.

6. The method for predicting the degradation of a heavy-duty vehicle hydraulic system as described in claim 5, characterized in that, The Long Short-Term Memory (LSTM) neural network model incorporates the sigmoid function and the hyperbolic tangent function into the network, including forget gates, input gates, and output gates; The forget gate has σ levels, with input C. t-1 ,X t Then in C t-1 The output value is in the range of 0 to 1; where 0 and 1 correspond to two special states, namely complete forgetting and complete retention; where C t-1 X represents the cell state at t-1. t This represents the input layer matrix at time t; The input gate selects the data required by the model, and the σ layer of the gate determines the value to be updated. A new candidate value is generated by the hyperbolic tangent layer and added to the neuron state. The process of performing neuron state updates starts from C. t-1 To C t , will f t Multiply by the old state to add a new candidate value; where C t The cell state at time t; f t For the forget gate at time t; The output gate determines the final output value.

7. A degradation prediction system for heavy-duty vehicle hydraulic systems, used to implement the degradation prediction method for heavy-duty vehicle hydraulic systems as described in any one of claims 1 to 6, characterized in that, include: The first processing module transmits the real-time collected oil pressure, oil temperature and flow data of the vehicle hydraulic system to the cloud. In the cloud, degradation features are extracted sequentially, and the hydraulic system performance is evaluated based on the oil pressure, oil temperature and flow data to obtain three degradation indicators: oil pressure curve similarity, oil temperature curve similarity and lubrication flow. The second processing module performs degradation classification on the extracted degradation feature data, labels the extracted oil pressure, oil temperature and flow data with different degradation stages, and uses them together with the degradation feature data as input to a pre-built degradation prediction model to train the degradation prediction model. The prediction output module takes the three degradation indicators, the degradation category obtained from degradation classification, and the degradation stage label as input to the trained degradation prediction model, and outputs the degradation prediction result to complete the prediction.

8. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 6.

9. A computing device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 6.

Images