Hydraulic pump motor degradation evaluation method and system based on LSTM neural network and similarity calculation

Abstract

The invention relates to a hydraulic pump motor degradation evaluation method and system based on an LSTM neural network and similarity calculation. The hydraulic pump motor degradation evaluation method comprises the steps that key indexes representing hydraulic pump motor degradation and input parameters influencing hydraulic pump motor working conditions are extracted from test data in different working conditions; selecting and dividing the obtained original test data into a training data set and a test data set; performing normalization and data screening on samples in the training data set to obtain data enhancement samples for model training; samples in the test data set are only subjected to normalization processing and are used for model testing; setting model parameters based on data features, obtaining a volumetric efficiency calculation result, and calculating a root-mean-square error; determining operation conditions of the hydraulic pump motor in the test data, and analyzing change trends of degradation index parameters of the hydraulic pump motor in different time periods; the similarity of the degradation index parameters of the hydraulic pump motor in different time periods is calculated, and the degradation degree is obtained; and respectively multiplying a prediction result of the neural network and a similarity calculation result by a weight coefficient of 0.5 to realize degradation evaluation of the hydraulic pump motor.

Description

Technical Field

[0001] The present invention relates to the technical field of hydraulic degradation assessment, and in particular to a hydraulic pump motor degradation assessment method and system based on LSTM neural network and similarity calculation. Background Art

[0002] With the development of computer technology, a large number of degradation indicators can be collected by analyzing test data under different operating conditions of hydraulic pumps and motors. Data analysis methods can map test data to the degradation status of hydraulic pumps and motors, playing an increasingly important role in hydraulic system reliability assessment.

[0003] In the research of data analysis methods, as product reliability improves, the collection of failure data becomes increasingly difficult. Consequently, methods based on performance degradation (RUL) assessment are gaining increasing attention. Deep learning methods, with their powerful data learning capabilities, have been widely used in the field of degradation assessment. By fully mining the information contained in the data based on data-driven models, they can produce highly accurate and stable degradation assessment models.

[0004] Various methods have been proposed to address the degradation assessment of hydraulic pumps and motors, primarily including physical modeling and data modeling. The former investigates the degradation mechanisms of hydraulic pumps and motors by modeling and analyzing the internal flow and pressure, plunger dynamics, and swashplate control systems of various pump / motor components. However, determining the parameters of the physical model in this approach is difficult. Data-driven modeling approaches primarily include statistical theory and neural networks. Commonly used statistical methods include Bayesian networks, hidden Markov models, support vector machines, and support vector data descriptions.

[0005] A data analysis method combining neural networks and similarity calculations provides a new approach for hydraulic pump and motor degradation assessment. It first uses an LSTM neural network to calculate the volumetric efficiency parameters of the hydraulic pump and motor, reflecting the impact of operating conditions on degradation. Data analysis is then used to determine the variation patterns of degradation indicators, reflecting the impact of time on degradation. Finally, these indicators are multiplied by weight coefficients to achieve degradation assessment.

[0006] In the degradation assessment of a hydraulic system, some model parameters of the physical model method will change non-linearly with the vehicle operating conditions and operation time. Therefore, it is difficult to give accurate values, which affects the model accuracy. In the data analysis method, the artificial neural network method, which is widely used, has a weak applicability to sample information with a small amount of data. It is difficult to determine the network structure and has excessive redundancy. The training time of the model is difficult to control and the real-time performance is poor. Methods such as support vector machines are prone to not outputting the optimal solution due to parameter optimization problems, and the model still needs to be improved. The data similarity calculation method has high requirements for test data. In the degradation test, the similarity calculation of the degradation index of the hydraulic pump motor in the same group of test data does not consider the change of working conditions, and there are certain errors in the results. Summary of the Invention

[0007] In view of the above problems, the purpose of the present invention is to provide a hydraulic pump motor degradation assessment method and system based on LSTM neural network and similarity calculation, which overcomes the difficult situation that the existing degradation assessment technology is difficult to process multi-dimensional and multi-condition test data. On the premise of ensuring degradation assessment, considering the change of working conditions, the accuracy of hydraulic pump motor degradation assessment under various working conditions is improved.

[0008] To achieve the above purpose, in the first aspect, the technical solution adopted by the present invention is: a hydraulic pump motor degradation assessment method based on LSTM neural network and similarity calculation, which includes: determining degradation indicators and input parameters: analyzing the mechanical structure and degradation principle of the hydraulic pump motor, and extracting key indicators representing the degradation of the hydraulic pump motor and input parameters affecting the working conditions of the hydraulic pump motor from the test data within different working conditions in complex working conditions and different time histories; constructing a data set: selecting the obtained original test data and dividing it into a training data set and a test data set; data preprocessing: normalizing and data screening the samples in the training data set to obtain data augmentation samples for model training; the samples in the test data set are only normalized for model testing; building an LSTM neural network: setting model parameters based on data characteristics, including the number of neural network layers, the number of neurons, the learning rate, and the training termination condition; training the LSTM neural network: obtaining the calculation result of the volumetric efficiency and calculating the root mean square error RMSE; test data analysis: determining the operating conditions of the hydraulic pump motor in the test data and analyzing the change trend of the degradation index parameters of the hydraulic pump motor in different time periods; realizing degradation assessment: analyzing the degradation state according to the change trend, calculating the curve similarity, and multiplying the prediction results and similarity calculation results of the neural network by a weight coefficient of 0.5 respectively to realize the degradation assessment of the hydraulic pump motor.

[0009] Further, in the step of determining degradation indicators and input parameters, the calculation method of the volumetric efficiency is: considering the external characteristics of the pump, directly calculating the volumetric efficiency of the pump motor according to the input speed of the pump and the output speed of the motor.

[0010] Furthermore, the volumetric efficiency of the pump motor is:

[0011]

[0012] In the formula, ε is the swash plate angle; a 马达 is the output speed of the motor; n 泵 is the input speed of the pump; η 泵马达 is the volumetric efficiency of the pump motor.

[0013] Furthermore, in the step of determining the degradation index and input parameters, the calculation method of the volumetric efficiency is as follows: According to the degradation principle inside the steering pump / motor system, the volumetric efficiency of the pump and the volumetric efficiency of the motor are calculated based on the theoretical flow rate and the actual flow rate, and the product of the volumetric efficiency of the pump and the volumetric efficiency of the motor is the volumetric efficiency of the hydraulic pump motor system.

[0014] Furthermore, the volumetric efficiency is:

[0015] η 泵马达 = η 泵 ·η 马达

[0016] In the formula, η 泵马达 is the volumetric efficiency of the pump motor; η 泵 is the volumetric efficiency of the pump; η 马达 is the volumetric efficiency of the motor.

[0017] Furthermore, in the step of data preprocessing, when the LSTM neural network predicts the volumetric efficiency, it will receive the current hydraulic pump motor input parameter information from the input layer and the hidden state of the previous moment through the input gate, and calculate the information from different sources through the Sigmoid function to determine whether it is activated.

[0018] Furthermore, calculating the curve similarity includes: analyzing the degradation state according to the change trend, calculating the Euclidean distance of the hydraulic pump motor degradation index parameters in different time periods, obtaining the similarity of the hydraulic pump motor degradation index parameters in different time periods, and further the degradation degree of the hydraulic pump motor degradation index parameters over time.

[0019] Second aspect, the technical solution adopted by the present invention is as follows: A hydraulic pump and motor degradation evaluation system based on an LSTM neural network and similarity calculation, which includes: a module for determining degradation indicators and input parameters: analyzing the mechanical structure and degradation principle of the hydraulic pump and motor, and extracting key indicators characterizing the degradation of the hydraulic pump and motor and input parameters affecting the working conditions of the hydraulic pump and motor from the test data in different working conditions in complex working conditions and different time histories; a module for constructing a data set: selecting the obtained original test data and dividing it into a training data set and a test data set; a data preprocessing module: normalizing and screening the samples in the training data set to obtain data augmentation samples for model training; the samples in the test data set are only normalized for model testing; a module for building an LSTM neural network: setting model parameters based on data features, mainly including the number of neural network layers, the number of neurons, the learning rate, and the training termination condition; a module for training the LSTM neural network: obtaining the calculation result of the volumetric efficiency and calculating the root mean square error RMSE; a module for analyzing test data: determining the operating conditions of the hydraulic pump and motor in the test data and analyzing the change trend of the degradation index parameters of the hydraulic pump and motor in different time periods; a module for realizing degradation evaluation: analyzing the degradation state according to the change trend, calculating the curve similarity module, and multiplying the prediction result and the similarity calculation result of the neural network by a weight coefficient of 0.5 respectively to realize the degradation evaluation of the hydraulic pump and motor.

[0020] Third aspect, the technical solution adopted by the present invention is as follows: 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 execute any of the above methods.

[0021] Fourth aspect, the technical solution adopted by the present invention is as follows: A computing device, which 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 executing any of the above methods.

[0022] Due to the above technical solutions adopted by the present invention, it has the following advantages:

[0023] 1. The present invention adopts the LSTM neural network method to reflect the influence of working conditions on the degradation state of the hydraulic pump and motor and predict the key parameter volumetric efficiency; adopts the test data analysis method to study the change trend of the degradation index parameters of the hydraulic pump and motor in different time periods, calculates its similarity according to the Euclidean distance, and finally multiplies by the weight coefficient respectively to realize the degradation evaluation. It can improve the accuracy of degradation evaluation under variable working conditions.

[0024] 2. The data feature extraction ability of the present invention is strong, and it can effectively utilize a large amount of normal data; at the same time, it considers the working condition changes and degradation states, and can avoid the influence of working conditions on the degradation states of hydraulic pump motors under complex working conditions; the accuracy of degradation assessment is improved. Description of the Drawings

[0025] Figure 1 is a flowchart of a method for evaluating the degradation of a hydraulic pump motor based on an LSTM neural network and similarity calculation in an embodiment of the present invention;

[0026] Figure 2 is a schematic diagram of an LSTM neural network in an embodiment of the present invention;

[0027] Figure 3 is the pump motor bench test data from 28h to 60h in an embodiment of the present invention;

[0028] Figure 4 is a graph of the change in RMSE during the training process in an embodiment of the present invention;

[0029] Figure 5 is a graph of the change trend of performance parameters when the hydraulic pump motor runs for 30h in an embodiment of the present invention;

[0030] Figure 6 is a graph of the change trend of the working pressure difference of the hydraulic pump motor at different running times in an embodiment of the present invention;

[0031] Figure 7 is a graph of the change trend of the output speed of the hydraulic pump motor at different running times in an embodiment of the present invention;

[0032] Figure 8 is a graph of the change trend of the output torque of the hydraulic pump motor at different running times in an embodiment of the present invention. Detailed Embodiments

[0033] 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 of the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention fall within the scope of protection of the present invention.

[0034] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, 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.

[0035] The present invention adopts two-stage data analysis. The first stage reflects the influence of working conditions on the degradation state of hydraulic pump motors. Based on the bench test data, the working conditions of the hydraulic pump motors are determined, and the performance changes of the hydraulic pump motors under different working conditions are analyzed. By using research methods of deep learning such as LSTM neural network, a data-driven model is constructed to predict the key parameter of volumetric efficiency. The second stage reflects the influence of the growth of time history on the degradation state, etc. The operating conditions in the test data are determined, the change trends of the degradation index parameters of the hydraulic pump motors in different time periods are analyzed, and their similarity is calculated according to the Euclidean distance. Finally, the weight coefficients are multiplied respectively to achieve degradation assessment.

[0036] In an embodiment of the present invention, a method for evaluating the degradation of a hydraulic pump motor based on an LSTM neural network and similarity calculation is provided. In this embodiment, as Figure 1 shown, the method includes the following steps:

[0037] Step 1: Determine the degradation index and input parameters.

[0038] During the long-term steering operation of the hydraulic pump motor of a special vehicle, a series of functional degradations will occur with the increase of the vehicle's driving mileage, the switching of the vehicle's operating conditions, and the change of the vehicle's driving environment. During the degradation process, the test data of the hydraulic pump motor contains rich degradation characteristic information. In complex working conditions and different time histories, the key characteristics of the degradation information also have different manifestations. Extracting the key indicators characterizing the degradation of the hydraulic pump motor from the test data within different working conditions is the key to realizing the degradation assessment of the hydraulic pump motor.

[0039] The volumetric efficiency of the hydraulic pump motor is the key indicator for judging its degradation performance. According to the provisions of JBT7043-2006, the volumetric efficiency of a variable axial piston pump with a displacement greater than 25 ml / r should be greater than or equal to 91%. The size of the volumetric efficiency can characterize the degradation state of the variable axial piston pump. The volumetric efficiency is also widely used by domestic and foreign scholars in the research on the degradation state of hydraulic pumps and hydraulic motors. Therefore, the volumetric efficiency is the key indicator for evaluating the degradation of the hydraulic pump motor.

[0040] There are mainly two main calculation methods for the volumetric efficiency. One is to mainly consider the external characteristics of the pump and directly calculate the volumetric efficiency η 泵 of the pump motor according to the input speed n 马达 of the pump and the output speed n 泵马达 of the motor, as shown in Equation (1). During the operation of the hydraulic pump motor, the input speed of the pump comes from the engine, and its output speed is closely related to the volumetric efficiency. Therefore, the output speed can also be used as a key indicator for evaluating the degradation state of the system.

[0041]

[0042] In the formula, ε is the swash plate angle.

[0043] The second method mainly focuses on the degradation principle inside the steering pump / motor system, such as in Equations (2) to (4). It mainly calculates the volumetric efficiency η 理论 of the pump and the volumetric efficiency η 实际 of the motor according to the theoretical flow rate q 泵 and the actual flow rate q 马达 . The product of the volumetric efficiencies of the pump / motor is the volumetric efficiency η 泵马达 of the hydraulic pump-motor, which can reflect that the reduction of the system volumetric efficiency is mainly caused by the reduction of the volumetric efficiencies of the variable displacement piston pump and the fixed displacement piston motor. This method mainly focuses on the internal structure of the hydraulic pump-motor, and it is necessary to accurately calculate the internal and external leakage flow rates of the pump-motor, which is difficult to achieve in actual engineering, but it is of great significance for determining the degradation index of the hydraulic pump-motor.

[0044]

[0045] η 泵马达 = η 泵 ·η 马达 (4)

[0046] In the formula: n p is the rotational speed of the pump; n m is the rotational speed of the motor; C p ·ΔP is the maximum displacement of the pump; V max is the maximum displacement of the motor; V is the displacement of the motor; ε is the relative variable rate of the pump.

[0047] The volumetric efficiency mainly characterizes the overall degradation characteristics of the hydraulic pump-motor system, and the hydraulic pump-motor system is a multi-component system affected by multiple factors. It is necessary to further analyze its internal degradation characteristics with reference to Equation (2). To determine other important indicators for system degradation assessment, analysis should be carried out in combination with the actual operating characteristics and test data of the hydraulic pump-motor. According to the literature..., the leakage coefficient of the pump is related to the pressure difference between the high and low pressure circuits of the hydraulic pump-motor. As the pressure difference between the high and low pressure sides increases, the leakage coefficient of the hydraulic pump-motor will increase accordingly. Therefore, the pressure of the high and low pressure side circuits is also an important indicator for studying system degradation, and the pressure difference between the high and low pressure side circuits is mainly affected by the load during vehicle driving, which is manifested as the torque of the hydraulic pump-motor affecting the leakage coefficient. Among them, the motor torque is determined by the working conditions and actual load of the special vehicle, and the torque of the pump can reflect the degradation of the hydraulic pump-motor.

[0048] In summary, the key indicators characterizing the degradation of the hydraulic pump-motor system mainly include volumetric efficiency, high-pressure (MPa), motor speed (r / min), and pump torque (Nm). After determining the degradation indicators, when studying the degradation of the hydraulic pump-motor system based on test data with different time histories, there should also be a basic premise, that is, to ensure the unity of the working conditions of the special vehicle. The working conditions of the vehicle are mainly determined by the input speed (r / min), output torque (Nm), and swashplate angle.

[0049] Step 2: Construct a data set. The neural network methods for constructing a data model mainly include feedforward neural networks (such as BP, CNN, FCNN, etc.) and feedback neural networks (such as RNN, HNN, BAM, etc.). Among them, the feedforward neural network has a simple structure and can effectively extract the hidden features of high-dimensional data, making it suitable for processing static data and pattern recognition problems. The feedback neural network can learn historical information and is more suitable for processing time series. Among them, HNN is mainly applied to image storage and finding the optimal solution, and BAM is mainly used for parallel processing of large-scale data.

[0050] During the prediction process of volumetric efficiency, the training data has strong time-series requirements. The LSTM network, as a new type of RNN, matches this feature very well. The added memory module in the LSTM network compared with the traditional RNN enables the network to have the function of preserving the feedback error during the gradient propagation process, making it not easily fall into local optimal solutions, and can establish connections for information at relatively long distances in long sequences. Therefore, by virtue of the non-linear feature adaptability and self-learning ability of the deep neural network, this paper proposes to use the LSTM neural network to learn the variation law of the volumetric efficiency of the hydraulic pump-motor with working conditions. The typical feature of the LSTM neural network lies in its long-term memory function, which can calculate how much of the cell state containing data of different time histories of the hydraulic pump-motor can be retained until the next moment. This can associate the volumetric efficiency data at relatively long distances in the long time series and learn the dynamic change characteristics of the volumetric efficiency of the pump-motor under continuously varying working conditions of the vehicle.

[0051] The schematic diagram of the pump-motor volumetric efficiency prediction model based on the LSTM neural network is as Figure 2 shown. This model can train the test data of the pump-motor under long-term complex working conditions. The input of the model is the volumetric efficiency parameter at the previous moment and parameters such as the input speed of the hydraulic pump-motor and the swashplate angle. The forget gate, input gate, and output gate inside the model continuously correct the weights according to the input data, improving the accuracy of the model and reducing the prediction error of the volumetric efficiency. Finally, the volumetric efficiency value of the hydraulic pump-motor at the current moment is obtained.

[0052] Step 3: Data preprocessing. To improve the accuracy of the volumetric efficiency prediction model, the hydraulic pump and motor test data should be processed, mainly including data screening and normalization. First, screen the data of the swash plate angle and volumetric efficiency within a reasonable range to avoid the influence of the sensor accuracy limit on the volumetric efficiency prediction, and then perform normalization processing on the test data. Among them, the data screening conditions and normalization methods for various sensors are shown in Table 1, and the original test data is as Figure 3 shown. Select 5000 data points and divide the test data into a training data set and a test data set according to a ratio of 8:2.

[0053] Table 1 Sensor data screening conditions and normalization methods

[0054]

[0055]

[0056] Step 4: Build an LSTM neural network. In the process of constructing the volumetric efficiency prediction model, use MATLAB software to build an LSTM deep learning network and set the model parameters, and train the model based on the processed test data. To obtain the optimal network structure and prediction effect, set different model parameters (learning rate, number of neurons, number of layers, termination time, etc.) for training and compare the training results. In the neural network model, the learning rate is a key parameter affecting the model stability. Too high a learning rate will make the model difficult to converge, resulting in higher errors, while too low a learning rate will prolong the training time and easily fall into local minima. To avoid such phenomena, set the initial value of the learning rate to 0.005, and reduce the learning rate by multiplying by a factor of 0.2 after 100 rounds of training to ensure the stable convergence of the model.

[0057] During the training process, use RMSE to describe the error. First, set the termination condition for model training to an error lower than 0.05. After multiple rounds of data-driven model training, it can be obtained that the error between the model prediction value and the test value will be lower than 0.05 within 200 rounds. To further reduce the model error and find the optimal model parameters of the LSTM deep learning network, set the termination condition for data-driven model training to a maximum of 200 training rounds. On the basis of ensuring the accuracy during the training process, use the minimum number of neurons and number of layers to reduce the training time. The selection of the number of neurons and number of layers in the model refers to the training results of different model parameters in Table 3.2. Finally, set the number of neurons in the LSTM network to 16 and the number of neural network layers to 4.

[0058] Table 2 RMSE corresponding to different model parameters

[0059]

[0060] When using an LSTM neural network to predict volumetric efficiency, it receives information on the current hydraulic pump motor input parameters from the input layer and the hidden state of the previous moment through the input gate, calculates the information from different sources using the sigmoid function, and determines whether to activate it. The construction formula for the output of the input gate is:

[0061]

[0062] The main function of the forget gate is to process the data in the memory cell, select whether to retain the state of the previous memory cell, and then superimpose the processing result on the state information processed by the input gate to form a new memory cell state. Its construction formula is:

[0063] f t = σ(W f · [h t-1 , x t + b f ) (6)

[0064] The output gate mainly determines the output of the current LSTM memory module based on the output of the previous memory cell and in combination with the dynamic control of the current forget gate. Its construction formula:

[0065]

[0066] In the formula, σ(Z) and tan(z) are the sigmoid and tanh activation functions respectively; h t-1 is the short-term memory output of the neuron at the previous moment; x t is the input of the neuron at this moment; C t-1 is the long-term memory output Cell state of the neuron at the previous moment.

[0067] Step 5, Train the LSTM neural network. Set the model parameters and train the volumetric efficiency prediction model. Input the test set data into the model, and the results can be obtained by inverse normalization to get the calculation results of the volumetric efficiency of the hydraulic pump motor under different working conditions, and compare them with the test set data to calculate the model error value. The loss function used for the error value is the RMSE function of the LSTM network. The change in the performance index during the model training process is as Figure 4 shown. The initial learning rate is 0.005. After iterating 0 - 100 times, the value of RMSE decreases from 0.98 to 0.068. When iterating 100 - 200 times (near 170 times), the learning rate automatically decreases by 20%, and the value of RMSE remains stable and decreases to 0.022.

[0068] The calculation method of RMSE is shown in Equation (8).

[0069]

[0070] Where N is the number of samples; Y i is the predicted value; f(x i ) is the true value.

[0071] Step 6, test data analysis. From Figure 3 the test data, it can be seen that the test data of the third stage (28 h to 60 h) should be selected for studying the working conditions of the swash plate change. Select the working conditions with higher frequencies and similar values in different time periods. Finally, the input speed of the selected working condition is about 2500 r / min, and the input torque is about 1000 Nm.

[0072] According to the working condition information, the bench test data is divided into 5 stages, namely the typical data of the hydraulic pump motor running for 30 - 31 hours, 35 - 36 hours, 43 - 44 hours, 47 - 48 hours, and 51 - 52 hours, and analyze the change of the volumetric efficiency during the degradation process of the hydraulic pump motor.

[0073] Select 6000 points of data running for about two minutes under Condition 1, as Figure 5 shown, and analyze the change trend of its performance parameters; to further screen the test data of the swash plate fully deflected to the left from the middle position during this period, take the average value of multiple segments of data to reduce the interference of test errors and the contingency of single - group data.

[0074] The data of the hydraulic pump motor running for 35 - 36 hours, 43 - 44 hours, 47 - 48 hours, and 51 - 52 hours can be referred to Figure 5 given.

[0075] Step 7, calculate the curve similarity. Based on the statistical data of different time histories, summarize the data of key performance - influencing parameters such as pressure, speed, and torque of the hydraulic pump motor under Condition 1 as the swash plate is fully deflected from the middle position, as Figures 6 to 8 shown, study its change trend, analyze the degradation state, and calculate its Euclidean distance.

[0076] Table 3 Euclidean distances of key performance parameters of the hydraulic pump motor at different running times

[0077]

[0078] According to Table 3, normalize the calculation results of the Euclidean distance to obtain the degradation trend of the performance parameters with time.

[0079] Step 8, achieve degradation assessment. Finally, multiply the prediction results of the neural network and the calculation results of the normalized similarity by a weight coefficient of 0.5 respectively to achieve the degradation assessment of the hydraulic pump motor.

[0080] In one embodiment of the present invention, a hydraulic pump and motor degradation evaluation system based on an LSTM neural network and similarity calculation is provided, which includes:

[0081] Determine degradation index and input parameter module: Analyze the mechanical structure and degradation principle of the hydraulic pump and motor. In complex working conditions and different time histories, extract the key indicators characterizing the degradation of the hydraulic pump and motor and the input parameters affecting the working conditions of the hydraulic pump and motor from the test data within different working conditions;

[0082] Construct data set module: Select the acquired original test data and divide it into a training data set and a test data set;

[0083] Data preprocessing module: Normalize and screen the samples in the training data set to obtain data augmentation samples for model training; The samples in the test data set are only normalized for model testing;

[0084] Build LSTM neural network module: Set model parameters based on data features, mainly including the number of neural network layers, the number of neurons, the learning rate, and the training termination condition;

[0085] Train LSTM neural network module: Obtain the calculation result of the volumetric efficiency and calculate the root mean square error RMSE;

[0086] Test data analysis module: Determine the operating conditions of the hydraulic pump and motor in the test data, and analyze the change trend of the degradation index parameters of the hydraulic pump and motor in different time periods;

[0087] Implement degradation evaluation module: Analyze the degradation state according to the change trend, calculate the curve similarity module, and multiply the prediction result and similarity calculation result of the neural network by a weight coefficient of 0.5 respectively to realize the degradation evaluation of the hydraulic pump and motor.

[0088] In the above embodiment, in the step of determining the degradation index and input parameters, the calculation method of the volumetric efficiency is: Considering the external characteristics of the pump, directly calculate the volumetric efficiency of the pump and motor according to the input speed of the pump and the output speed of the motor.

[0089] In this embodiment, the volumetric efficiency of the pump and motor is:

[0090]

[0091] In the formula, ε is the swash plate angle; n 马达 is the output speed of the motor; n 泵 is the input speed of the pump; η 泵马达 is the volumetric efficiency of the pump and motor.

[0092] In the above embodiments, in the step of determining the degradation index and the input parameters, the calculation method of the volumetric efficiency is as follows: According to the degradation principle inside the steering pump / motor system, the volumetric efficiency of the pump and the volumetric efficiency of the motor are calculated based on the theoretical flow rate and the actual flow rate, and the product of the volumetric efficiency of the pump and the volumetric efficiency of the motor is the volumetric efficiency of the hydraulic pump-motor system.

[0093] In this embodiment, the volumetric efficiency is:

[0094] η 泵马达 = η 泵 ·η 马达

[0095] In the formula, η 泵马达 is the volumetric efficiency of the pump-motor; η 泵 is the volumetric efficiency of the pump; η 马达 is the volumetric efficiency of the motor.

[0096] In the above embodiments, in the step of data preprocessing, when the LSTM neural network predicts the volumetric efficiency, it receives the current hydraulic pump-motor input parameter information from the input layer and the hidden state of the previous moment through the input gate, and calculates the information from different sources through the sigmoid function to determine whether it is activated.

[0097] In the above embodiments, calculating the curve similarity includes: analyzing the degradation state according to the change trend, calculating the Euclidean distance of the hydraulic pump-motor degradation index parameters in different time periods, obtaining the similarity of the hydraulic pump-motor degradation index parameters in different time periods, and further obtaining the degradation degree of the hydraulic pump-motor degradation index parameters over time.

[0098] The system provided in this embodiment is used to execute the above method embodiments. For the specific process and detailed content, please refer to the above embodiments and will not be repeated here.

[0099] In an embodiment of the present invention, a computing device is provided. The computing device may be a terminal, which may include: a processor, a communications interface, a memory, a display screen, and an input device. Among them, the processor, the communications interface, and the memory complete communication with each other through a communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, the methods in the above embodiments are implemented; the internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communications interface is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen may be a liquid crystal display screen or an electronic ink display screen. The input device may be a touch layer covering the display screen, or may be a button, a trackball, or a touchpad provided on the housing of the computing device, or may also be an external keyboard, touchpad, or mouse, etc. The processor can call the logical instructions in the memory.

[0100] In addition, when the logical instructions in the above-mentioned memory are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this 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 for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

[0101] In an embodiment of the present invention, a computer program product is provided. The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided in the above method embodiments.

[0102] In an embodiment of the present invention, a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium stores server instructions. The computer instructions cause the computer to execute the methods provided in the above embodiments.

[0103] A computer-readable storage medium provided by the above embodiments has the same implementation principle and technical effects as the above method embodiments, and will not be described in detail here.

[0104] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or a device for implementing the functions specified in a plurality of blocks.

[0105] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or a device for implementing the functions specified in a plurality of blocks.

[0106] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flows Figure 1 one or more flows and / or blocks Figure 1 or a device for implementing the functions specified in a plurality of blocks.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A hydraulic pump and motor degradation assessment method based on an LSTM neural network and similarity calculation, characterized in that Including: Determine degradation indicators and input parameters: Analyze the mechanical structure and degradation principle of the hydraulic pump motor. In complex working conditions and different time histories, extract the key indicators characterizing the degradation of the hydraulic pump motor and the input parameters affecting the working conditions of the hydraulic pump motor from the test data in different working conditions; Construct a data set: Select the obtained original test data and divide it into a training data set and a test data set; Data preprocessing: Normalize and screen the samples in the training data set to obtain data augmentation samples for model training; The samples in the test data set are only normalized for model testing; Build an LSTM neural network: Set model parameters based on data characteristics, including the number of neural network layers, the number of neurons, the learning rate, and the training termination condition; Train the LSTM neural network: Obtain the calculation result of the volumetric efficiency and calculate the root mean square error RMSE; Test data analysis: Determine the operating conditions of the hydraulic pump motor in the test data and analyze the change trend of the degradation index parameters of the hydraulic pump motor in different time periods; Implement degradation assessment: Analyze the degradation state according to the change trend, calculate the curve similarity, and multiply the prediction result of the neural network and the calculation result of the similarity by a weight coefficient of 0.5 to implement the degradation assessment of the hydraulic pump motor.

2. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 1, wherein, In the step of determining the degradation indicators and input parameters, the calculation method of the volumetric efficiency is: Considering the external characteristics of the pump, directly calculate the volumetric efficiency of the pump motor according to the input speed of the pump and the output speed of the motor.

3. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 2, wherein The volumetric efficiency of the pump motor is: Where ε is the swashplate angle; n 马达 is the output speed of the motor; n 泵 is the input speed of the pump; η 泵马达 is the volumetric efficiency of the pump - motor.

4. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 1, characterized in that In the step of determining the degradation indicators and input parameters, the calculation method of the volumetric efficiency is: Based on the degradation principle inside the steering pump / motor system, calculate the volumetric efficiency of the pump and the volumetric efficiency of the motor according to the theoretical flow rate and the actual flow rate. The product of the volumetric efficiency of the pump and the volumetric efficiency of the motor is the volumetric efficiency of the hydraulic pump motor system.

5. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 4, characterized in that, The volumetric efficiency is: η 泵马达 = η 泵 · η 马达 where η 泵马达 is the volumetric efficiency of the pump motor; η 泵 is the volumetric efficiency of the pump; η 马达 is the volumetric efficiency of the motor.

6. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 1, characterized in that In the step of data preprocessing, when the LSTM neural network predicts the volumetric efficiency, it will receive the current hydraulic pump motor input parameter information from the input layer and the hidden state of the previous moment through the input gate, calculate the information from different sources through the sigmoid function, and determine whether to activate it.

7. The hydraulic pump and motor degradation evaluation method based on the LSTM neural network and similarity calculation according to claim 1, characterized in that, Calculate the curve similarity, including: Analyze the degradation state according to the change trend, calculate the Euclidean distance of the degradation index parameters of the hydraulic pump motor in different time periods, obtain the similarity of the degradation index parameters of the hydraulic pump motor in different time periods, and further the degradation degree of the degradation index parameters of the hydraulic pump motor over time.

8. A hydraulic pump and motor degradation evaluation system based on an LSTM neural network and similarity calculation, characterized in that, Including: Determine degradation indicators and input parameters module: Analyze the mechanical structure and degradation principle of the hydraulic pump motor. In complex working conditions and different time histories, extract the key indicators characterizing the degradation of the hydraulic pump motor and the input parameters affecting the working conditions of the hydraulic pump motor from the test data in different working conditions; Construct data set module: Select the obtained original test data and divide it into a training data set and a test data set; Data preprocessing module: Normalize and screen the samples in the training data set to obtain data augmentation samples for model training; The samples in the test data set are only normalized for model testing; Build the LSTM neural network module: Set the model parameters based on the data features, mainly including the number of neural network layers, the number of neurons, the learning rate, and the training termination condition; Train the LSTM neural network module: Obtain the calculation result of the volumetric efficiency and calculate the root mean square error RMSE; Test data analysis module: Determine the operating conditions of the hydraulic pump motor in the test data and analyze the change trend of the degradation index parameters of the hydraulic pump motor in different time periods; Implement the degradation evaluation module: Analyze the degradation state according to the change trend, calculate the curve similarity module, and multiply the prediction result of the neural network and the similarity calculation result by a weight coefficient of 0.5 respectively to achieve the degradation evaluation of the hydraulic pump motor.

9. A computer-readable storage medium 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 7.

10. A computing device, characterized in that, Comprising: 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 in claims 1 to 7.

Images