Hydraulic cylinder micro-leakage monitoring method and monitoring device based on information fusion

Abstract

The present application discloses a method and device for monitoring micro-leakage of hydraulic cylinders based on information fusion, and relates to the technical field of monitoring micro-leakage of hydraulic cylinders. The method includes: collecting circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data to obtain a time series feature vector set, and then obtaining fault data; locating the fault of the hydraulic cylinder through the fault data to obtain the result of locating the fault of the hydraulic cylinder; obtaining the fault position of the hydraulic cylinder according to the result of locating the fault of the hydraulic cylinder; extracting features of the fault position to obtain the features of the fault position, and obtaining a spectrum diagram of the fault position of the hydraulic cylinder based on the features of the fault position; analyzing the spectrum diagram to determine whether there is electromagnetic wave interference, and if there is interference, calibrating and compensating the analysis results to obtain the fault position of the hydraulic cylinder. The present application can diagnose and locate the fault of micro-leakage of the hydraulic cylinder, and effectively improve the accuracy of monitoring micro-leakage of the hydraulic cylinder.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of hydraulic cylinder micro-leakage monitoring, and in particular to a hydraulic cylinder micro-leakage monitoring method and monitoring device based on information fusion. Background Art

[0002] Hydraulic cylinders are widely used in the field of mechanical equipment. In daily work, the most common failure of hydraulic cylinders is micro-leakage. If a hydraulic cylinder micro-leakage failure occurs, it will not only affect the quality of the equipment, but also cause equipment scrapping, casualties, etc. Therefore, it is very necessary to monitor the micro-leakage of hydraulic cylinders.

[0003] At present, the monitoring of micro-leakage of hydraulic cylinders mainly adopts observation method, auditory method and experimental method. Manually judge whether micro-leakage occurs by observing the slow movement of the hydraulic cylinder, feeling the change of oil temperature of the hydraulic cylinder, the vibration of the hydraulic cylinder or by listening to the sound of the hydraulic cylinder when working. Although the above methods are intuitive and simple, it is difficult to monitor the micro-leakage phenomenon inside the hydraulic cylinder. Therefore, the accuracy of manual monitoring of micro-leakage of hydraulic cylinders is low.

[0004] In summary, there is an urgent need for a method to improve the accuracy of micro-leakage monitoring of hydraulic cylinders. Summary of the invention

[0005] The embodiments of the present application provide a hydraulic cylinder micro-leakage monitoring method and a monitoring device based on information fusion, which are used to effectively improve the accuracy of hydraulic cylinder micro-leakage monitoring.

[0006] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0007] In a first aspect, a method for monitoring micro-leakage of a hydraulic cylinder based on information fusion is provided, which is applied to a monitoring device, and the method comprises:

[0008] Collect circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data;

[0009] Performing feature extraction on the circumferential acoustic emission sensor data, the temperature sensor data, and the pressure sensor data to obtain a time series feature vector set, and obtaining fault data according to the time series feature vector set;

[0010] Locating the fault of the hydraulic cylinder by using the fault data to obtain a hydraulic cylinder fault locating result;

[0011] According to the hydraulic cylinder fault location result, the fault position of the hydraulic cylinder is obtained;

[0012] Extracting features of the fault location to obtain features of the fault location, and obtaining a frequency spectrum of the fault location based on the features of the fault location;

[0013] According to the frequency spectrum of the hydraulic cylinder fault position, determining whether the frequency spectrum has electromagnetic wave interference;

[0014] In the case where electromagnetic wave interference exists in the spectrum diagram of the fault position of the hydraulic cylinder, spectrum analysis is performed on the spectrum diagram to obtain an initial spectrum analysis result;

[0015] Performing calibration compensation on the initial spectrum analysis result to obtain a final spectrum analysis result;

[0016] The matching degree between the final spectrum analysis result and the spectrum diagram is calculated, and when the matching degree between the final spectrum analysis result and the spectrum diagram of the hydraulic cylinder fault position is greater than a first preset threshold, the fault position of the hydraulic cylinder is determined.

[0017] In a possible implementation manner of the first aspect, extracting features from the circumferential acoustic emission sensor data, the temperature sensor data, and the pressure sensor data to obtain a time series feature vector set, and obtaining fault data according to the time series feature vector set includes:

[0018] Inputting the circumferential acoustic emission sensor data, the temperature sensor data and the pressure sensor data into a pre-trained neural network model to obtain a time series feature vector set of the hydraulic cylinder fault;

[0019] Determining the working state of the hydraulic cylinder according to the time series feature vector set of the hydraulic cylinder fault;

[0020] The fault data is obtained according to the working state of the hydraulic cylinder.

[0021] In a possible implementation manner of the first aspect, the working state of the hydraulic cylinder includes a normal state and a fault state, and determining the working state of the hydraulic cylinder according to a time series feature vector set of the hydraulic cylinder fault includes:

[0022] Acquire circumferential acoustic emission signals, temperature signals and pressure signals;

[0023] Building a fault feature model based on the time series feature vector set of the hydraulic cylinder fault;

[0024] Dividing the circumferential acoustic emission signal, the temperature signal and the pressure signal within a preset time unit to obtain a first time series characteristic signal and a second time series characteristic signal;

[0025] Inputting the first time series characteristic signal and the second time series characteristic signal into the fault characteristic model to obtain a first prediction result and a second prediction result;

[0026] Calculate the matching degree between the first prediction result, the second prediction result and the working state to obtain a first matching degree and a second matching degree;

[0027] If the first matching degree and the second matching degree are both greater than a preset matching degree, the first prediction result, the second prediction result and the working state of the hydraulic cylinder are compared to determine the working state of the hydraulic cylinder.

[0028] In a possible implementation manner of the first aspect, locating the fault of the hydraulic cylinder by using the fault data to obtain a hydraulic cylinder fault locating result includes:

[0029] Extracting features from the fault data to obtain features of the fault data;

[0030] Inputting the fault data features into a pre-trained genetic algorithm model to obtain an optimal fault location result;

[0031] According to the optimal fault location result, a hydraulic cylinder fault location result is obtained.

[0032] In a possible implementation of the first aspect, inputting the fault data feature into a pre-trained genetic algorithm model to obtain an optimal fault location result includes:

[0033] S1. Generate a binary coding strategy according to the fault data characteristics, determine the maximum and minimum ranges of the initial value through the binary coding strategy and generate an initial group, wherein the binary coding strategy is the initial value of the fault data;

[0034] S2. Calculate the fitness of individuals in the initial population based on the objective function, and record the optimal fitness of each generation. When the optimal fitness no longer changes, exit the iteration to obtain the first adaptive population.

[0035] S3, performing crossover and mutation on the first adaptive population;

[0036] S4, repeating steps S1 to S3, determining whether the fitness values ​​obtained in this and previous iterations reach a preset target fitness value, and if so, stopping iteration and outputting the hydraulic cylinder fault optimization positioning result.

[0037] In a possible implementation manner of the first aspect, after locating the fault of the hydraulic cylinder by using the fault data to obtain the hydraulic cylinder fault locating result, the method further includes:

[0038] The fault data is stored in a preset diagnosis module, and a fault model tree is constructed;

[0039] Determining the fault information of the hydraulic cylinder according to the fault location result of the hydraulic cylinder;

[0040] The hydraulic cylinder fault location result is transmitted to the diagnosis module, and the hydraulic cylinder fault location result is compared with the fault model tree to obtain the hydraulic cylinder fault location information.

[0041] In a possible implementation manner of the first aspect, judging, according to the frequency spectrum of the hydraulic cylinder fault position, whether there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position includes:

[0042] Performing time-frequency analysis on the frequency spectrum of the hydraulic cylinder fault position by short-time Fourier transform to obtain a time-frequency diagram;

[0043] Identify the electromagnetic interference feature to be measured on the time-frequency diagram, where the electromagnetic interference feature to be measured is a vertical stripe whose duration is less than a preset time period and whose frequency bandwidth is greater than a preset bandwidth;

[0044] For each of the electromagnetic interference features, the interference intensity is calculated using a preset interference intensity calculation formula;

[0045] Calculating the interference density of the electromagnetic interference feature on the time-frequency diagram;

[0046] It is determined whether electromagnetic wave interference exists in the frequency spectrum of the hydraulic cylinder fault position according to the interference density and the interference intensity.

[0047] In a possible implementation manner of the first aspect, judging, according to the interference density and the interference intensity, whether there is electromagnetic wave interference in the frequency spectrum diagram of the hydraulic cylinder fault position includes:

[0048] Calculating the mean, standard deviation, skewness and kurtosis of the interference intensity, and calculating the intensity dispersion degree according to the mean and the standard deviation, wherein the intensity dispersion degree is obtained by dividing the standard deviation by the mean;

[0049] Calculating an intensity distribution shape index according to the skewness and the kurtosis, wherein the intensity distribution shape index is the sum of the absolute value of the skewness and the absolute value of the kurtosis;

[0050] Calculating scores of the interference density, the intensity dispersion degree and the intensity distribution shape index respectively by corresponding scoring functions;

[0051] If the scores of any two of the interference density, the intensity dispersion degree and the intensity distribution shape index are greater than the corresponding preset thresholds, it is determined that there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position; wherein, if the interference density is less than the preset score trough and when the intensity dispersion degree and the intensity distribution shape index are both greater than the corresponding preset thresholds, it is determined that there is no electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position.

[0052] In a possible implementation manner of the first aspect, the calibrating and compensating the initial spectrum analysis result to obtain a final spectrum analysis result includes:

[0053] In the initial spectrum analysis result, determining discrete peaks exceeding a preset expected amplitude, and performing median filtering on the discrete peaks to remove the discrete peaks;

[0054] Interpolating the frequency points of the removed discrete peaks to obtain interpolated spectrum analysis results;

[0055] The baseline of the spectrum is determined in the interpolated spectrum analysis result, and the baseline is subtracted from the interpolated spectrum analysis result to perform baseline correction to obtain a final spectrum analysis result.

[0056] In a second aspect, the present application provides a monitoring device, comprising:

[0057] Circumferential acoustic emission sensor;

[0058] Temperature sensor;

[0059] Pressure sensor;

[0060] a memory configured to store instructions; and

[0061] The processor is configured to call the instructions from the memory and implement the above-mentioned hydraulic cylinder micro-leakage monitoring method based on information fusion when executing the instructions.

[0062] Through the above technical solution, circumferential acoustic emission data, temperature data and pressure data are obtained, and feature extraction is performed on the data to obtain a time series feature vector set and then the fault data. Fault location is performed according to the fault data of the hydraulic cylinder, and the hydraulic cylinder fault is located through the fault data to determine the fault position of the hydraulic cylinder. Secondly, through the spectrum diagram of the hydraulic cylinder fault position and spectrum analysis, it is determined whether there is electromagnetic wave interference in the spectrum diagram. If there is interference, the spectrum analysis results are calibrated and compensated to determine the fault position of the hydraulic cylinder. The present application can diagnose and locate micro-leakage faults in hydraulic cylinders, which improves the accuracy of micro-leakage monitoring of hydraulic cylinders compared to the method of manually determining whether a hydraulic cylinder is leaking.

[0063] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of a flow chart of a method for monitoring micro-leakage of a hydraulic cylinder based on information fusion provided in an embodiment of the present application;

[0065] Figure 2 A schematic diagram of a time-frequency diagram provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0067] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0068] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0069] Figure 1 The following schematically shows a flow chart of a method for monitoring micro-leakage of a hydraulic cylinder based on information fusion according to an embodiment of the present application. Figure 1 As shown, an embodiment of the present application provides a method for monitoring micro-leakage of a hydraulic cylinder based on information fusion, and the method may include the following steps.

[0070] S110, collecting circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data;

[0071] S120, extracting features from the circumferential acoustic emission sensor data, the temperature sensor data, and the pressure sensor data to obtain a time series feature vector set, and obtaining fault data according to the time series feature vector set;

[0072] S130, locating the fault of the hydraulic cylinder through the fault data to obtain a hydraulic cylinder fault locating result;

[0073] S140, obtaining a fault position of the hydraulic cylinder according to the hydraulic cylinder fault location result;

[0074] S150, extracting features of the fault location to obtain fault location features, and obtaining a frequency spectrum of the fault location based on the fault location features;

[0075] S160, judging whether there is electromagnetic wave interference in the spectrum diagram according to the spectrum diagram of the hydraulic cylinder fault position;

[0076] S170, when there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position, performing spectrum analysis on the spectrum diagram to obtain an initial spectrum analysis result;

[0077] S180, calibrating and compensating the initial spectrum analysis result to obtain a final spectrum analysis result;

[0078] S190, calculating the matching degree between the final spectrum analysis result and the spectrum diagram, and determining the fault location of the hydraulic cylinder when the matching degree between the final spectrum analysis result and the spectrum diagram of the fault location of the hydraulic cylinder is greater than a first preset threshold.

[0079] First, circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data are collected by placing pressure sensors, circumferential acoustic emission sensors, and temperature sensors around the piston of the hydraulic cylinder, wherein the collected data includes data of the hydraulic cylinder in both normal and faulty modes. In the event of a micro-leakage in the hydraulic cylinder, changes in pressure and temperature will occur in the hydraulic cylinder, and such changes can be captured by the pressure sensor and the temperature sensor. Secondly, when micro-leakage occurs inside the hydraulic cylinder, whether due to aging, damage, or other reasons of the seal, these micro-leakages are usually accompanied by deformation or fracture of the material, thereby generating acoustic emission signals, and the circumferential acoustic emission sensor can capture these signals. Therefore, this embodiment uses pressure sensors, circumferential acoustic emission sensors, and temperature sensors to obtain data to monitor micro-leakage of the hydraulic cylinder.

[0080] After that, the circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are feature extracted to obtain a time series feature vector set, which can be obtained through a neural network model. The neural network model can achieve autonomous learning and decision-making by simulating the working mode of neurons in the human brain. The neural network model includes an input layer, a hidden layer and an output layer. These layers together constitute the basic structure of the neural network, enabling the neural network to process and transform information and ultimately produce output results. Among them, the input layer is responsible for receiving external input data, the hidden layer can transform and process the input data, and the output layer is to output the processed information in a specific form.

[0081] Specifically, the circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are input into a pre-trained neural network model, the data are divided according to a preset time period, and the circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are identified and extracted through the pre-trained neural network model to obtain a time series feature vector set, and the fault data is obtained according to the time series feature vector. Fault feature identification refers to identifying the characteristics of possible faults through a neural network model. In this embodiment, the preset time period can be the six-day working time of the hydraulic cylinder. Dividing the preset time period refers to dividing the time into different time periods within a certain time range, and each time period has similar performance information. For example, the seven-day working time is divided into one, two, and three-day working time, and the data within one, two, and three-day working time are respectively input into the pre-trained neural network model for fault feature identification and extraction.

[0082] After the fault data is obtained, the fault of the hydraulic cylinder may be located by the fault data. In this embodiment, the fault data may be input into a pre-built genetic algorithm model to locate the fault of the hydraulic cylinder.

[0083] Specifically, first determine the initial population. Calculate the individual fitness in the initial population according to the objective function, and record the optimal fitness each time until the optimal fitness each time does not change, and determine the first adaptive population. Perform crossover and mutation operations on the first adaptive population to increase the diversity of the population and obtain a more adaptive population. Finally, repeat the steps of determining the initial population, obtaining the first adaptive population, and performing crossover and mutation on the first adaptive population. When the target fitness value is reached, stop the iteration and output the hydraulic cylinder fault optimization positioning result.

[0084] According to the hydraulic cylinder fault location result, the fault location of the hydraulic cylinder is obtained. Specifically, the historical fault information is stored in a preset diagnosis module, and a fault model tree is constructed, wherein the fault model tree is used to intuitively represent the complex fault information structure. The fault location result is mapped to the fault information of the hydraulic cylinder, the hydraulic cylinder fault location result is transmitted to the diagnosis module, and the hydraulic cylinder fault location result is compared with the fault model tree to obtain the hydraulic cylinder fault location information.

[0085] After the fault location is obtained, the fault location is firstly feature extracted to obtain the fault location feature, and the frequency spectrum of the hydraulic cylinder fault location is obtained based on the fault location feature. The feature extraction of the fault location can be performed by filtering method or wavelet transform method, which is not limited in this application. Since the wavelet transform method has the advantages of data compression and denoising, the wavelet transform method is selected in this embodiment to extract the fault location feature. After obtaining the fault location feature, the original circumferential acoustic emission signal, temperature signal and pressure signal are converted from the time domain to the frequency domain using short-time Fourier transform to obtain the frequency domain signal, and then the frequency spectrum can be generated according to the frequency domain signal to obtain the frequency spectrum of the hydraulic cylinder fault location.

[0086] The spectrum diagram of the hydraulic cylinder fault position refers to obtaining the original signal of the hydraulic cylinder fault position, and generating a spectrum diagram from the original signal of the fault position for analysis. For example, when the hydraulic cylinder is working, a micro-leakage fault may occur due to piston wear. When a micro-leakage fault occurs due to piston wear, the temperature at the location of the piston will also rise. By obtaining the temperature signal at the location of the piston, the time domain signal is converted into a frequency domain signal to obtain the spectrum diagram of the hydraulic cylinder fault position. Electromagnetic wave interference refers to the interference of electromagnetic wave signals to the spectrum diagram, which is manifested as the phenomenon of aliasing of the peak spectrum lines of the spectrum diagram.

[0087] According to the spectrum diagram of the hydraulic cylinder fault position, determine whether there is electromagnetic wave interference in the spectrum diagram. First, determine whether there is interference through the change of the frequency peak of the spectrum diagram. When there is interference, the frequency peak of the spectrum diagram will suddenly increase. If there is interference, then determine whether the interference is electromagnetic wave interference. If it is electromagnetic wave interference, there will be aliasing of the peak spectrum lines of the spectrum diagram. According to the spectrum diagram of the hydraulic cylinder fault position, it can be obtained that the peak spectrum lines of the spectrum diagram are aliased, and it is determined that there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position.

[0088] Spectral analysis refers to the process of converting a time domain signal into a frequency domain signal, obtaining a spectrum diagram with amplitude as the vertical axis and frequency as the horizontal axis, and analyzing the spectrum diagram. The conversion of a time domain signal into a frequency domain signal can be obtained by short-time Fourier transform. In this embodiment, the initial spectrum analysis result refers to the analysis result obtained after analyzing the spectrum diagram of the hydraulic cylinder fault position when electromagnetic wave interference exists. After determining that the spectrum diagram of the hydraulic cylinder fault position has electromagnetic wave interference, the spectrum diagram of the hydraulic cylinder fault position is subjected to spectrum analysis to obtain the initial spectrum analysis result.

[0089] Calibration compensation is an operation to eliminate the deviation between the measurement result and the reference value, and to optimize and improve the accuracy and reliability of the analysis result. In this embodiment, the final spectrum analysis result refers to the spectrum analysis result obtained after the initial spectrum analysis result is calibrated and compensated. Since there is electromagnetic wave signal interference in the initial spectrum analysis result, there is a deviation between the analysis result and the actual result. In order to eliminate the deviation between the analysis result and the actual result, the initial spectrum analysis result is calibrated and compensated. When there is electromagnetic wave interference, the intensity of the peak spectrum line will also increase. By calculating the intensity of the peak spectrum line, the first measurement intensity is obtained, and the overlap coefficient of the peak spectrum line is set. The first measurement intensity is calculated according to the preset overlap coefficient. The second measurement intensity is obtained, and it is determined whether the first measurement intensity and the second measurement intensity meet the preset accuracy requirements. If the preset accuracy requirements are met, the calculation is stopped, and the calibration compensation result is output to obtain the final spectrum analysis result.

[0090] The first preset threshold refers to a critical value set in a certain system. When a certain parameter or indicator reaches or exceeds this value, a specific action or behavior will be triggered. In this embodiment, the first preset threshold means that when the final spectrum analysis result and the spectrum diagram of the hydraulic cylinder fault position match 80%, the fault position of the hydraulic cylinder can be determined through the final spectrum analysis result. After obtaining the final spectrum analysis result, the final spectrum analysis result and the spectrum diagram are matched. When the matching degree between the final spectrum analysis result and the spectrum diagram of the hydraulic cylinder fault position is greater than the first preset threshold, the fault position of the hydraulic cylinder is determined. In this embodiment, the first preset threshold is set to 80%, and the final spectrum analysis result and the spectrum diagram are matched. If the matching degree is greater than 80%, it is determined that the final spectrum analysis result meets the matching requirement, and the fault position of the hydraulic cylinder is determined through the final spectrum analysis result.

[0091] In summary, by acquiring circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data, the hydraulic cylinder fault diagnosis and location are performed, and based on the spectrum diagram of the hydraulic cylinder fault position, it is determined whether there is signal interference, and the signal interference is calibrated and compensated to determine the fault position of the hydraulic cylinder. It can not only effectively identify and monitor the micro-leakage problem of the hydraulic cylinder, and timely eliminate the micro-leakage fault of the hydraulic cylinder, but also improve the safety of the hydraulic cylinder system and the working efficiency of the hydraulic cylinder system, and the monitoring of hydraulic cylinder micro-leakage plays a vital role in the normal operation of the equipment.

[0092] Through the above technical scheme, firstly, circumferential acoustic emission sensors, temperature sensors, and pressure sensors are used to obtain circumferential acoustic emission data, temperature data, and pressure data, and the data is input into a pre-trained neural network model to obtain a time series feature vector set and then obtain fault data, and the hydraulic cylinder is diagnosed through the fault data. Fault location is performed according to the fault data of the hydraulic cylinder, and the fault data is input into a pre-trained genetic algorithm model to locate the hydraulic cylinder fault and determine the fault position of the hydraulic cylinder. Secondly, the signal at the fault position is obtained and a spectrum analysis is performed to determine whether there is interference from other signals at the fault position of the hydraulic cylinder, and whether it is electromagnetic wave signal interference. If there is interference, the spectrum analysis result is calibrated and compensated to determine the fault position of the hydraulic cylinder. The present application can timely and accurately perform micro-leakage fault diagnosis and positioning on the hydraulic system, ensure the normal operation of hydraulic equipment, and play an important role in improving production efficiency, reducing maintenance costs, and reducing unnecessary economic losses.

[0093] In one implementation of this embodiment, feature extraction is performed on circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data to obtain a time series feature vector set, and fault data is obtained according to the time series feature vector set, including:

[0094] S210, inputting the circumferential acoustic emission sensor data, the temperature sensor data, and the pressure sensor data into a pre-trained neural network model to obtain a time series feature vector set of the hydraulic cylinder fault;

[0095] S220, determining the working state of the hydraulic cylinder according to the time series feature vector set of the hydraulic cylinder fault;

[0096] S230. Obtain fault data according to the working state of the hydraulic cylinder.

[0097] A pre-trained neural network model refers to a neural network model that has been pre-trained with a large number of data sets. The time series feature vector involves the analysis and processing of time series data. In the scenario of hydraulic cylinder micro-leakage monitoring in this embodiment, the time series feature vector of the hydraulic cylinder in the fault state is obtained through the pre-trained neural network model, thereby capturing the cause of the hydraulic cylinder micro-leakage, including capturing the changes in temperature, pressure, and signal of the hydraulic cylinder during micro-leakage; the fault data refers to the circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data of the hydraulic cylinder in the case of micro-leakage.

[0098] The circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are input into a pre-trained neural network model to obtain a time series feature vector set of the hydraulic cylinder fault. Specifically, the circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are input into a pre-trained neural network model, the circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data are divided according to preset time periods, and fault features are identified and extracted to obtain a time series feature vector set.

[0099] Fault feature identification refers to identifying the features of possible faults through a neural network model. The time series feature vector after fault feature identification is used to express the fault features that change over time. In this embodiment, the preset time period can be the six-day working time of the hydraulic cylinder. Dividing the preset time period means dividing the time into different time periods within a certain time range, and each time period has similar performance information. For example, the six-day working time is divided into one, two, and three days of working time, and the circumferential acoustic emission sensor data, temperature sensor data, and pressure sensor data are input into the pre-trained neural network model within one, two, and three days of working time, respectively, for fault feature identification and extraction. After obtaining the time series feature vector set, the working state of the hydraulic cylinder is determined according to the time series feature vector.

[0100] Specifically, a fault feature model can be constructed to determine the working state of the hydraulic cylinder according to the time series feature vector set of the hydraulic cylinder fault, that is, whether the hydraulic cylinder fails can be predicted according to the fault feature model, thereby determining the working state of the hydraulic cylinder. The fault feature model can be a deep learning model.

[0101] This embodiment obtains the time series feature vector set of hydraulic cylinder faults through the pre-trained neural network model, and then determines the working state of the hydraulic cylinder to obtain fault data. It can improve the accuracy and robustness of hydraulic cylinder micro-leakage monitoring, and timely and accurately perform fault diagnosis on hydraulic cylinder micro-leakage to ensure the normal operation of the hydraulic cylinder.

[0102] In one implementation of this embodiment, the working state of the hydraulic cylinder includes a normal state and a fault state. The working state of the hydraulic cylinder is determined according to a time series feature vector set of the hydraulic cylinder fault, including:

[0103] S310, acquiring circumferential acoustic emission signals, temperature signals, and pressure signals;

[0104] S320, constructing a fault feature model based on a time series feature vector set of the hydraulic cylinder fault;

[0105] S330, dividing the circumferential acoustic emission signal, the temperature signal, and the pressure signal within a preset time unit to obtain a first time series characteristic signal and a second time series characteristic signal;

[0106] S340, inputting the first time series characteristic signal and the second time series characteristic signal into a fault characteristic model to obtain a first prediction result and a second prediction result;

[0107] S350, calculating the matching degree between the first prediction result, the second prediction result and the working state to obtain a first matching degree and a second matching degree;

[0108] S360: If both the first matching degree and the second matching degree are greater than the preset matching degree, the first prediction result and the second prediction result are compared with the working state of the hydraulic cylinder to determine the working state of the hydraulic cylinder.

[0109] The working state of the hydraulic cylinder includes normal state and fault state. The normal state is that the hydraulic cylinder can perform linear reciprocating motion or swinging motion, and the motion is smooth and there is no transmission gap. The working state of the hydraulic cylinder under fault conditions will show insufficient hydraulic cylinder pressure, speed reduction, creeping phenomenon, and the piston rod cannot move. The micro-leakage failure of the hydraulic cylinder is mainly caused by damage or aging of the seal, contamination of the hydraulic oil of the hydraulic cylinder, etc. The working state of the hydraulic cylinder is determined by the time series feature vector of the hydraulic cylinder fault, and it is judged whether the working state of the hydraulic cylinder is a fault state or a normal state.

[0110] Based on the time series feature vector set of hydraulic cylinder fault, a fault feature model can be constructed. Through the deep learning algorithm, the time series feature vector of hydraulic cylinder fault can be used to construct a fault model to predict the working state of the hydraulic cylinder. Deep learning is a machine learning method that simulates the neural network of the human brain to process and predict data by constructing a deep neural network.

[0111] The circumferential acoustic emission signal, temperature signal and pressure signal are divided within the preset time unit to obtain the first time series characteristic signal and the second time series characteristic signal. The working time of the hydraulic cylinder is set. Within the preset hydraulic cylinder working time, the preset time is divided into two time periods of different nodes. Within the time periods of two different nodes, the characteristic vector of the hydraulic cylinder fault is divided into the first time series characteristic signal and the second time series characteristic signal. The first time series characteristic signal and the second time series characteristic signal obtained after the characteristic vector is divided are input into the fault characteristic model to obtain the first prediction result and the second prediction result respectively. For example,

[0112] In this embodiment, the preset time unit can be three days, that is, the circumferential acoustic emission signal, temperature signal and pressure signal are divided within three days to obtain the first timing characteristic signal and the second timing characteristic signal. The working time of the hydraulic cylinder is set to three days, and the signal is divided on the first day, the second day and the third day respectively to obtain the first timing characteristic signal and the second timing characteristic signal.

[0113] The first prediction result, the second prediction result and the working state are matched to calculate the first matching degree and the second matching degree, and the specific steps are as follows:

[0114] The first prediction result and the second prediction result of the hydraulic cylinder currently obtained are matched and calculated. The matching method is as follows: the first prediction result is split, and the change curves of the circumferential acoustic emission signal, the temperature sensor signal and the pressure sensor signal in the first prediction result are obtained, and the change curves of the circumferential acoustic emission signal, the temperature sensor signal and the pressure sensor signal are standardized, and the matching degree between the current curve change and the curve change in the normal state is calculated to obtain the first matching degree. Similarly, the change curves of the circumferential acoustic emission signal, the temperature sensor signal and the pressure sensor signal in the second prediction result are split, and the change curves of the circumferential acoustic emission signal, the temperature sensor signal and the pressure sensor signal of the second prediction result are standardized, and the matching degree between the curve change of the second prediction result and the curve change in the normal state is calculated to obtain the second matching degree. Standardization refers to converting data into a unified standard format for comparison and analysis between different data sources. In this embodiment, the minimum and maximum value standardization is adopted to convert the original data into a range of 0 to 1 for further analysis and processing.

[0115] If both the first matching degree and the second matching degree are greater than the preset matching degree, the first prediction result and the second prediction result are compared with the working state of the hydraulic cylinder. When both the first prediction result and the second prediction result are in normal state, the current hydraulic cylinder is in normal state. When both the first prediction result and the second prediction result are in fault state, the current hydraulic cylinder is in fault state. If the first prediction result and the second prediction result are in normal state and fault state respectively, the weight values ​​of the first prediction result and the second prediction result need to be calculated, and the prediction result with the larger weight value is the working state of the current hydraulic cylinder. For example, the weight values ​​of the first prediction result and the second prediction result can be calculated by the weighted summation method. The weighted summation method is to assign different weights to different indicators, and then multiply the scores of each indicator by the corresponding weights, and add them to obtain the final score. Different weights are assigned to the first prediction result and the second prediction result, and then the scores of the first prediction result and the second prediction result are multiplied by the corresponding assigned weights to determine the weight value of the prediction result.

[0116] Through the time series characteristic vector of hydraulic cylinder fault, the working state of the hydraulic cylinder is determined, potential risks are discovered in time, and the faults of the hydraulic cylinder are predicted, which improves the safety of hydraulic cylinder monitoring and effectively improves the timeliness of hydraulic cylinder monitoring.

[0117] In one implementation of this embodiment, locating the fault of the hydraulic cylinder by using the fault data to obtain the hydraulic cylinder fault locating result includes:

[0118] S410, extracting features from the fault data to obtain fault data features;

[0119] S420, inputting the fault data features into a pre-trained genetic algorithm model to obtain an optimal fault location result;

[0120] S430. Obtain a hydraulic cylinder fault location result according to the optimal fault location result.

[0121] The fault data features are the features of the circumferential acoustic emission signal, temperature signal and pressure signal when the hydraulic cylinder is in a faulty state. The data in the faulty state is acquired and the features of the data in the faulty state are extracted. The fault data features are mainly manifested in that the similarity with the data in the normal state is lower than the threshold. The fault data features are obtained by extracting the features of the fault data. The fault of the hydraulic cylinder is located through the pre-trained genetic algorithm model. The fault of the hydraulic cylinder is located based on the genetic algorithm. The optimal fault location result is obtained by step-by-step iteration and optimization. The hydraulic cylinder fault location result is determined through the optimal fault location result.

[0122] Input the fault data features into the pre-built genetic algorithm model. The specific steps include:

[0123] S1. Generate a binary coding strategy according to the characteristics of the fault data, determine the maximum and minimum ranges of the initial value through the binary coding strategy and generate an initial group, wherein the binary coding strategy is the initial value of the fault data;

[0124] S2. Calculate the fitness of individuals in the initial population based on the objective function, and record the optimal fitness of each generation. When the optimal fitness no longer changes, exit the iteration and obtain the first adaptive population.

[0125] S3, crossover and mutation of the first adaptation population;

[0126] S4, repeat steps S1 to S3 to determine whether the fitness values ​​obtained in this and previous iterations reach the preset target fitness values. If the target fitness values ​​are reached, the iteration is stopped and the hydraulic cylinder fault optimization positioning result is output.

[0127] According to the fault data characteristics, a code using only two different characters, usually "0" and "1", is generated. By generating a binary coding strategy, the maximum and minimum ranges of the hydraulic cylinder fault data characteristics are determined and an initial group is generated.

[0128] The objective function is a function for evaluating individual fitness. The objective function directly affects the convergence speed of the genetic algorithm and whether the optimal fitness can be found. The initial population is brought into the objective function, and the obtained optimal fitness is marked until the optimal fitness no longer changes, then the iteration is exited and the first adaptive population is determined.

[0129] Based on the first adaptive population, crossover and mutation operations are performed. Crossover and mutation operations can increase the diversity of the algorithm, help global search, and thus improve the efficiency and performance of the algorithm. Crossover and mutation operations are performed on the first adaptive population under the hydraulic cylinder fault state to increase the accuracy of the hydraulic cylinder fault location results. By judging whether the fitness value obtained in this and the previous iterations reaches the preset target fitness value, if the target fitness value is reached, the iteration is stopped and the hydraulic cylinder fault optimization location result is output. Through the hydraulic cylinder fault optimization location result, the hydraulic cylinder fault location is gradually optimized and accurately located, and finally the hydraulic cylinder fault location result is determined.

[0130] Based on the pre-trained genetic algorithm model, the fault of the hydraulic cylinder is located through the fault data to obtain the hydraulic cylinder fault location result, which can narrow the search range of the hydraulic cylinder fault position and accurately obtain the location of the hydraulic cylinder fault.

[0131] In one implementation manner of this embodiment, after locating the fault of the hydraulic cylinder through the fault data and obtaining the hydraulic cylinder fault locating result, the method further includes:

[0132] S510, storing the fault data in a preset diagnosis module and constructing a fault model tree;

[0133] S520, determining the fault information of the hydraulic cylinder according to the hydraulic cylinder fault location result;

[0134] S530, transmitting the hydraulic cylinder fault location result to the diagnosis module, and comparing the hydraulic cylinder fault location result with the fault model tree to obtain the hydraulic cylinder fault location information.

[0135] The diagnostic module is a standardized module for handling hydraulic cylinder equipment fault diagnosis and management. The diagnostic module can store and clear fault information, and provide functions such as recording, clearing, and archiving fault information. First, a fault model tree is constructed based on historical fault information, and the historical fault information is stored in the diagnostic module. The fault information is determined by a one-to-one mapping of the hydraulic cylinder fault location results. Secondly, the fault tree is used to analyze complex fault information and determine the cause of the fault.

[0136] The steps to construct a fault model tree are:

[0137] 1. Determine the cause of the failure and the possible consequences.

[0138] 2. Analyze the hydraulic cylinder failure system to determine the various modes that may cause the failure.

[0139] 3. Based on the relationship between the causes and types of hydraulic cylinder failures, establish a logical relationship model of the fault tree.

[0140] 4. Detailed analysis and definition of each basic event of the cause and type of hydraulic cylinder failure, including description of hydraulic cylinder failure phenomenon, occurrence conditions and probability.

[0141] 5. By quantitatively calculating and analyzing the fault tree, the availability and reliability of the fault tree are evaluated, and the probability and impact of hydraulic cylinder failure are identified.

[0142] The hydraulic cylinder fault location result is mapped one-to-one with the fault information, and the hydraulic cylinder fault location result is transmitted to the diagnosis module, and the fault model tree is used to query in the diagnosis module to determine the fault information of the hydraulic cylinder. Subsequently, the fault information is transmitted to the mobile terminal using the SIM card.

[0143] By storing fault data, using the diagnostic module to obtain the hydraulic cylinder fault location information, and sending the fault information location to the mobile terminal device through the SIM card, the hydraulic cylinder fault location information can be obtained in a timely manner, which helps to identify the fault location and take corresponding actions to ensure that the hydraulic cylinder fault can be resolved in a timely manner.

[0144] In one implementation of this embodiment, judging whether there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position according to the spectrum diagram of the hydraulic cylinder fault position includes:

[0145] S610, performing time-frequency analysis on the frequency spectrum of the hydraulic cylinder fault position by using short-time Fourier transform to obtain a time-frequency diagram;

[0146] S620, identifying and obtaining a feature of the electromagnetic interference to be measured on the time-frequency diagram, where the feature of the electromagnetic interference to be measured is a vertical stripe whose duration is less than a preset time period and whose frequency bandwidth is greater than a preset bandwidth;

[0147] S630, for each electromagnetic interference feature, using a preset interference intensity calculation formula to calculate the interference intensity;

[0148] S640, calculating the interference density of the electromagnetic interference feature on the time-frequency diagram;

[0149] S650: Determine whether there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position according to the interference density and the interference intensity.

[0150] In this embodiment, first, a short-time Fourier transform is performed on the spectrum of the hydraulic cylinder fault position to obtain a time-frequency diagram. Specifically, the time domain signal is divided into multiple short-time windows, and then Fourier transform is applied to each window. In specific implementation, a suitable window function (such as Hanning window or Hamming window) and window length can be selected, and the window length is usually 1 / 10 to 1 / 100 of the total length of the signal. There can be 50% overlap between windows to improve the time resolution. Fast Fourier transform (FFT) is performed on each window to obtain the spectrum within the time period. The spectra of all windows are arranged in chronological order to obtain a time-frequency diagram. The horizontal axis of the time-frequency diagram represents time, the vertical axis represents frequency, and the color depth represents signal strength. The time-frequency diagram can simultaneously reflect the characteristics of the signal in the time domain and frequency domain, which is conducive to observing the changes of the signal over time. For example, for a signal lasting 10 seconds, a window length of 0.1 seconds can be selected, and the window is slid once every 0.05 seconds, and finally the spectrum of 200 time points is obtained to form a time-frequency diagram.

[0151] Next, the electromagnetic interference feature to be tested is identified on the time-frequency diagram. The electromagnetic interference feature is manifested as vertical stripes with a duration less than a preset time period and a bandwidth greater than a preset bandwidth. The implementation principle is based on the typical performance of electromagnetic interference on the time-frequency diagram: Since electromagnetic interference is usually short-lived and broadband, it will appear as narrow and high vertical stripes on the time-frequency diagram, such as Figure 2 As shown, Figure 2 The horizontal axis is time. Figure 2The vertical axis is the frequency V. In specific implementation, a time threshold (such as 0.01 seconds) and a bandwidth threshold (such as 10kHz) can be set first. Then, image processing techniques, such as edge detection algorithms (such as the Canny algorithm), are used to identify vertical edges in the time-frequency diagram. For each detected edge, its duration and bandwidth are calculated. If the duration is less than the preset time threshold and the bandwidth is greater than the preset bandwidth threshold, it is marked as a potential electromagnetic interference feature. For example, if a vertical stripe with a duration of 0.005 seconds and a bandwidth of 15kHz is detected, it can be identified as an electromagnetic interference feature.

[0152] Then, for each identified electromagnetic interference feature, the interference strength is calculated using a preset interference strength calculation formula. The interference strength calculation formula is:

[0153]

[0154] Where I is the interference intensity, N is the number of video points covered by vertical stripes, S(t i ,f i ) is the signal strength of the ith signal in the time-frequency diagram.

[0155] Next, calculate the interference density of the electromagnetic interference features on the time-frequency graph. The interference density refers to the number of interference features within a unit time and frequency range. In specific implementation, the time-frequency graph can be divided into several time-frequency blocks, each of which has a time span of Δt (such as 0.1 seconds) and a frequency span of Δf (such as 1kHz). Then count the number of interference features contained in each time-frequency block to obtain an interference density matrix. Finally, calculate the average interference density of the entire time-frequency graph. The calculation formula is: D = N / (T*F), where D is the average interference density, N is the total number of interference features, T is the total signal duration, and F is the total signal bandwidth. For example, if 500 interference features are identified in a signal that lasts 10 seconds and has a bandwidth of 100kHz, the average interference density D = 500 / (10*100000) = 0.0005 / (seconds*Hz).

[0156] Finally, according to the interference density and interference intensity, it is judged whether there is electromagnetic interference in the spectrum diagram of the hydraulic cylinder fault position. The statistical characteristics and distribution characteristics of the interference intensity and the interference density can be used to determine whether there is electromagnetic interference.

[0157] The above method realizes the analysis of electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position through steps such as short-time Fourier transform, feature recognition, intensity calculation, density statistics and comprehensive judgment. Not only can a single interference feature be located and quantified, but the overall interference condition can also be evaluated. By combining time domain and frequency domain analysis, short-term, broadband interference signals can be captured and judged through objective numerical indicators. It can effectively improve the accuracy of hydraulic cylinder fault diagnosis and effectively distinguish real fault signals from external interference, thereby providing a reliable basis for subsequent fault location and maintenance.

[0158] In another embodiment, judging whether there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position according to the spectrum diagram of the hydraulic cylinder fault position further includes:

[0159] S10, obtaining the frequency spectrum of each preset time period, and determining whether a frequency peak in the frequency spectrum of each preset time period has a sudden change;

[0160] S20, if the frequency peak value suddenly changes, it is determined that there is interference in the frequency spectrum of the hydraulic cylinder fault position;

[0161] S30, performing peak spectrum line aliasing identification on the peak of the spectrum graph;

[0162] S40. When there is aliasing of the peak spectrum lines, it is determined that there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position.

[0163] Short-time Fourier transform can obtain the spectrum information of the signal at different time points by dividing the signal into multiple short time periods and performing Fourier transform on each time period.

[0164] This embodiment obtains the circumferential acoustic emission signal, temperature signal and pressure signal of the hydraulic cylinder fault position, and converts the original circumferential acoustic emission signal, temperature signal and pressure signal from the time domain to the frequency domain using short-time Fourier transform, so as to obtain a frequency domain signal. Then, a spectrum graph can be generated according to the frequency domain signal, and the peak value in the spectrum graph can be used to identify whether there is interference in the spectrum graph of the hydraulic cylinder fault position.

[0165] By acquiring the spectrum graph of each preset time period, it is determined whether the frequency peak in the spectrum graph of each preset time period changes suddenly. In actual situations, when the frequency peak in the spectrum graph changes irregularly, the spectrum graph is interfered.

[0166] When the hydraulic cylinder is in a normal working state, the spectrum of the circumferential acoustic emission signal, temperature signal and pressure signal of the hydraulic cylinder is obtained, and the amplitude and peak value of the spectrum of the circumferential acoustic emission signal, temperature signal and pressure signal change regularly. When there is interference in the spectrum of the hydraulic cylinder fault position, the frequency peak of the spectrum of the hydraulic cylinder fault position will suddenly change. Therefore, it can be determined that there is interference in the spectrum of the hydraulic cylinder fault position through the sudden change of the frequency peak in the spectrum.

[0167] Electromagnetic interference can cause inaccurate positioning of the hydraulic cylinder fault location. This interference may come from strong electromagnetic fields in the surrounding area, such as wireless communication modules, switching power supplies, etc. Under strong electromagnetic interference, the peak spectrum line of the spectrum diagram of the hydraulic cylinder fault location will produce irregular displacement or fluctuation, which is usually manifested as peak spectrum line aliasing.

[0168] Therefore, this embodiment first determines whether there are frequency peaks of different shapes through the spectrum diagram, which can be obtained by analyzing the spectrum diagram of the hydraulic cylinder fault position, wherein spectrum analysis is the process of converting the signal from the time domain to the frequency domain, and the amplitude distribution of the signal at different frequencies can be obtained through Fourier transform. When the signal is sampled, if the sampling frequency is not enough to support the highest frequency component of the signal, the sampled signal will not be able to reconstruct the original signal, resulting in the loss of spectrum characteristics and aliasing. For example, a signal has only one spectrum line somewhere in the frequency domain, but when it is sampled with different sampling frequencies, signals of different shapes may be obtained, such as DC curves, triangular wave signals, etc. These signals cannot maintain the spectrum characteristics of the original signal, resulting in aliasing. By identifying the frequency peak, it can be preliminarily determined whether there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position.

[0169] The peak of the spectrum graph is identified for peak spectral line aliasing. Since the working environment of the hydraulic cylinder may have a variety of electromagnetic signal interference, such as electromagnetic noise caused by natural phenomena, such as lightning, solar activity, etc., and electromagnetic interference generated by electronic equipment and artificial devices. The interference caused by these environmental factors will lead to inaccurate positioning of the hydraulic cylinder fault position. Therefore, the peak of the spectrum graph at the hydraulic cylinder fault position is identified for peak spectral line aliasing.

[0170] The spectrum diagram of the hydraulic cylinder fault position is analyzed to determine whether there is electromagnetic wave interference, and the hydraulic cylinder fault position is further accurately located. Through the analysis of the spectrum diagram, not only the fault diagnosis capability of the hydraulic cylinder micro-leakage monitoring equipment can be effectively improved, but also the effect of micro-leakage monitoring of the hydraulic cylinder can be effectively improved.

[0171] In one implementation of this embodiment, judging whether there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position according to the interference density and the interference intensity includes:

[0172] S710, calculating the mean, standard deviation, skewness and kurtosis of the interference intensity, and calculating the intensity dispersion degree according to the mean and standard deviation, wherein the intensity dispersion degree is obtained by dividing the standard deviation by the mean;

[0173] S720, calculating an intensity distribution shape index according to the skewness and the kurtosis, wherein the intensity distribution shape index is the sum of the absolute value of the skewness and the absolute value of the kurtosis;

[0174] S730, respectively calculating scores of interference density, intensity dispersion and intensity distribution shape index by corresponding scoring functions;

[0175] S740. If the scores of any two of the interference density, intensity dispersion and intensity distribution shape index are greater than the corresponding preset thresholds, it is determined that there is electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position. If the interference density is less than the preset score trough and the intensity dispersion and intensity distribution shape index are both greater than the corresponding preset thresholds, it is determined that there is no electromagnetic wave interference in the spectrum diagram of the hydraulic cylinder fault position.

[0176] In this embodiment, the mean of the interference intensity is first calculated, that is, the arithmetic mean of all interference intensity values, which reflects the central tendency of the overall interference intensity. The standard deviation is used to measure the degree of dispersion of the interference intensity, which is the square root of the arithmetic mean of the sum of the squares of the differences between each interference intensity value and the mean. Skewness describes the asymmetry of the distribution of interference intensity. Positive skewness indicates that the tail on the right side of the distribution is longer, while negative skewness is the opposite. Kurtosis reflects the sharpness of the distribution. A high kurtosis indicates that the distribution is sharper, while a low kurtosis indicates that the distribution is flatter. The degree of dispersion of intensity is calculated by dividing the standard deviation by the mean, also known as the coefficient of variation, which eliminates the influence of the dimension so that data of different scales can be compared. For example, suppose there are 10 interference intensity values: 100, 120, 80, 90, 110, 105, 95, 115, 85, 100. The calculated mean is 100 and the standard deviation is approximately 12.47. The degree of dispersion of intensity is 12.47 / 100≈0.1247. The intensity dispersion reflects the relative dispersion of the interference intensity. The larger the value, the greater the fluctuation of the interference intensity.

[0177] The intensity distribution shape index is defined as the sum of the absolute value of the skewness and the absolute value of the kurtosis, which comprehensively reflects the asymmetry and sharpness of the distribution. The larger the absolute value of the skewness, the more asymmetric the distribution is; the larger the absolute value of the kurtosis, the more the distribution deviates from the normal distribution. By adding these two indices, a single value is obtained to describe the overall shape characteristics of the distribution. For example, assuming that the calculated skewness is 0.5 and the kurtosis is 3.2, the intensity distribution shape index is |0.5|+|3.2|=3.7. This value is large, indicating that the interference intensity distribution is both asymmetric and sharp. In practical applications, a threshold can be set based on experience. If the intensity distribution shape index exceeds this threshold, it can be considered that the interference intensity distribution has significant non-normal characteristics, which may mean that there is an abnormal interference pattern. The shape characteristics of the interference intensity distribution can be quantified, which helps to identify possible abnormal interference patterns.

[0178] The scoring function is obtained based on the experimental data analysis of the interference characteristics. For the interference density, the scoring function can be: S D =100*D0, where D0 is the interference density, S D is the interference density score. For the intensity dispersion, the scoring function can be: S CV =100×(1-e (-CV / 0.5) );

[0179] In the formula, S CV is the score of intensity dispersion, and CV is the intensity dispersion.

[0180] For the intensity distribution shape indicator, the scoring function can be:

[0181] S s =100×(1-e (-S / 2) );

[0182] In the formula, S S is the score of the intensity distribution shape index, and S is the intensity distribution shape index.

[0183] Through the above scoring function, statistical features can be converted into intuitive scores, providing comparable quantitative indicators for the final judgment.

[0184] Finally, a comprehensive judgment is made based on the previously calculated scores to determine whether there is electromagnetic interference. Specifically, if multiple indicators show abnormalities at the same time, it is more likely that there is interference. Specifically, if the scores of any two of the interference density, intensity dispersion, and intensity distribution shape indicators are greater than the corresponding preset thresholds, it is determined that there is electromagnetic interference. The preset threshold here is determined based on a large amount of experimental data and expert experience, for example, it can be set to 0.7 or 0.8. However, a special case needs to be considered: when the interference density is low, but the intensity dispersion and intensity distribution shape indicators are both high, it may indicate that there is a small amount of but very significant interference. In order to deal with this situation, an additional judgment condition is introduced: if the interference density is less than the preset score trough (for example, 0.3), even if the intensity dispersion and intensity distribution shape indicators are both greater than the corresponding preset thresholds, it is determined that there is no electromagnetic interference. Too low interference density means that interference events are very rare and may be accidental phenomena caused by other factors. For example, assuming that the interference density score is 0.2, the intensity dispersion score is 0.9, and the intensity distribution shape indicator score is 0.85, the preset thresholds are all 0.7, and the preset score trough is 0.3. Although the last two indicators exceeded the threshold, the interference density was lower than the preset score trough, and it was finally determined that there was no electromagnetic wave interference. This not only avoided the misjudgment that might be caused by a single indicator, but also took into account special circumstances, thus improving the accuracy and reliability of the judgment.

[0185] The above method comprehensively describes the distribution characteristics of interference by calculating multiple statistical characteristics of interference intensity (mean, standard deviation, skewness, kurtosis) and interference density. By introducing the intensity dispersion degree and intensity distribution shape indicators, the volatility and non-normal characteristics of interference are further quantified. By considering the synergy and special circumstances of multiple indicators, the accuracy and robustness of judgment are greatly improved. It can not only effectively identify common electromagnetic wave interference, but also deal with some complex interference patterns, providing a reliable interference identification basis for hydraulic cylinder fault diagnosis.

[0186] According to the above S10-S40, in another embodiment, the peak spectral line aliasing identification is performed on the peak of the spectrum diagram, including:

[0187] S100, acquiring a spectrum line corresponding to a peak value of the spectrum diagram within the same period to obtain a peak spectrum line;

[0188] S200, calculating the intensity threshold of the peak spectrum line to obtain a first intensity threshold;

[0189] S300: Compare the first intensity threshold with the preset peak spectral line intensity. If the first intensity threshold is greater than the preset peak spectral line intensity, it is determined that peak spectral line aliasing exists.

[0190] Aliasing refers to the phenomenon that in the process of signal processing, due to insufficient sampling frequency or improper signal processing, the originally separated frequency components overlap in the spectrum, thus affecting the accurate identification and analysis of signal characteristics. The spectral line corresponding to the peak of the spectrum graph usually represents the intensity or amplitude of a specific frequency component in the signal. In spectrum analysis, the signal is decomposed into components of different frequencies, and the amplitudes of these components are represented as peaks on the spectrum graph.

[0191] The spectral line corresponding to the peak value of the spectrum graph is obtained within the same period to obtain the peak spectrum line. First, the spectral line or frequency range to be extracted is determined and the determined frequency range is marked. For example, on the spectrum graph of the hydraulic cylinder fault position, the spectral line with a fixed frequency corresponding to the signal whose amplitude repeatedly changes with time within the same period is selected, and the frequencies between the spectrum line with an amplitude of 1 and the next spectrum line with an amplitude reaching 1 are collected and marked.

[0192] The influence of spectral line aliasing on intensity is significant. When spectral line aliasing affects the accuracy of intensity calculation, it will also cause distortion of signal spectrum information, making it impossible to accurately identify the spectral components of the signal, thus affecting subsequent signal processing and analysis. Therefore, it is necessary to calculate the intensity of the peak spectral line to determine whether the peak spectral line has aliasing.

[0193] The intensity threshold of the peak spectrum line is calculated to obtain a first intensity threshold. Specifically, a signal within a preset frequency is collected, and the intensity of the collected signal is calculated. The intensity threshold of the peak spectrum line is calculated by a spectrum line presentation method. The spectrum line presentation method is a method of judging the intensity of the spectrum line by directly observing the brightness or darkness of the spectrum line to obtain the first intensity threshold. The spectrum line presentation method can directly observe the spectrum line, and the presentation of the spectrum line is observed by an instrument to calculate the intensity of the spectrum line.

[0194] The first intensity threshold is compared with the preset peak spectral line intensity, and if the first intensity threshold is greater than the preset peak spectral line intensity, it is determined that there is peak spectral line aliasing. For example, if the preset peak spectral line intensity is 0.6 and the first intensity threshold is 0.8, it is determined that there is peak spectral line aliasing.

[0195] Identifying peak spectral line aliasing of the peaks of the spectrum graph can effectively improve signal processing efficiency and simplify the data processing process, reduce or avoid unnecessary signal interference, and thus improve the quality and efficiency of signal processing.

[0196] In one implementation manner of this embodiment, calibration compensation is performed on the initial spectrum analysis result to obtain the final spectrum analysis result, including:

[0197] S810, in the initial spectrum analysis result, determining discrete peaks exceeding a preset expected amplitude, and performing median filtering on the discrete peaks to remove the discrete peaks;

[0198] S820, interpolating the frequency points of the removed discrete peaks to obtain interpolated spectrum analysis results;

[0199] S830: Determine the baseline of the spectrum in the interpolated spectrum analysis result, and subtract the baseline from the interpolated spectrum analysis result to perform baseline correction to obtain a final spectrum analysis result.

[0200] In this embodiment, firstly, the discrete peaks exceeding the preset expected amplitude are determined in the initial spectrum analysis results, and the discrete peaks are removed to remove the sharp peaks that may be caused by noise, interference or abnormal signals, so as to obtain a smoother and more reliable spectrum analysis result. First, the 95% quantile of the spectrum amplitude can be selected as the threshold. Then, the entire spectrum is traversed to find all the peak points exceeding this threshold. For each peak point exceeding the preset expected amplitude, a median filter is applied for smoothing. The principle of median filtering is to select a window (for example, 5 points or 7 points) around the peak point, sort all the amplitude values ​​in this window, and then replace the amplitude of the peak point with the middle value. For example, if the amplitude value at the frequency point f is 100, and the values ​​around it are 20, 25, 30, 35, and 40 respectively, then after median filtering, the amplitude value at the frequency point f will be replaced by 30. This method can effectively remove isolated spikes while maintaining the shape of the overall spectrum. The window size of the median filter is selected according to the resolution of the spectrum and the expected peak width, and an odd number of points is usually selected to ensure a clear median value. By adopting the above method, the abnormal sharp peaks in the spectrum are significantly reduced, making the spectrum smoother and better reflecting the overall frequency characteristics of the signal without being disturbed by local outliers.

[0201] Afterwards, fill in the "gaps" caused by removing discrete peaks to ensure the continuity and smoothness of the spectrum. Specifically, the interpolation process is actually to estimate the reasonable value of the removed point based on the surrounding data points. This embodiment uses linear interpolation, which assumes that the value between two known points changes along a straight line. For example, if the amplitude values ​​of frequencies f1 and f3 are known, to interpolate the amplitude value of point f2, the formula can be used: A(f2)=A(f1)+(A(f3)-A(f1))*(f2-f1) / (f3-f1), where A represents the amplitude value.

[0202] The baseline of the spectrum is determined in the interpolated spectrum analysis result, and the baseline is subtracted from the interpolated spectrum analysis result to perform baseline correction to obtain the final spectrum analysis result, so as to eliminate the background trend or systematic deviation in the spectrum and highlight the real signal characteristics. The baseline is usually caused by the instrument background, environmental noise or low-frequency components of the signal, which may mask some weak but important spectrum features. The moving minimum method can be used to determine the baseline. Specifically, the moving minimum method is to find the minimum value in a sliding window and connect these minimum points to form a baseline.

[0203] After the baseline is determined, it is subtracted from the interpolated spectrum to obtain the corrected spectrum. This method can significantly improve the signal-to-noise ratio of the spectrum, making the weak features that may have been masked by the baseline clearly visible, while also eliminating the long-term drift caused by the system or environment, making the spectrum results obtained at different times or under different conditions more comparable.

[0204] The spectrum analysis calibration and compensation method described above effectively improves the accuracy and reliability of spectrum analysis. First, abnormal peaks are removed by median filtering, eliminating false signals that may be caused by noise or interference. Then, the continuity of the spectrum is restored by interpolation technology, ensuring the smoothness and integrity of the spectrum curve. Finally, background trends and system deviations are eliminated by baseline correction, highlighting the true signal characteristics. It can not only effectively remove various interferences, but also retain the essential characteristics of the signal, greatly improving the quality and interpretability of the spectrum analysis results, making the spectrum analysis more stable and reliable, providing a solid foundation for subsequent signal feature extraction and pattern recognition, and effectively improving the accuracy of hydraulic cylinder fault diagnosis.

[0205] In another embodiment, for the above S100-S300, calibration compensation is performed on the initial spectrum analysis result to obtain the final spectrum analysis result, further comprising:

[0206] S1000, find the peak spectrum line in the presence of aliasing;

[0207] S2000, calculating the intensity of the peak spectrum line in the presence of aliasing to obtain a first measurement intensity;

[0208] S3000, performing a secondary intensity calculation on the first measurement intensity according to a preset overlap coefficient to obtain a second measurement intensity;

[0209] S4000, determining whether the first measurement intensity and the second measurement intensity meet a preset accuracy requirement, and if so, stopping the calculation, outputting a calibration compensation result, and obtaining a final spectrum analysis result.

[0210] Calibration compensation is the process of optimizing and improving the accuracy and reliability of analytical results to ensure the accuracy and reliability of analytical results, thereby better meeting prediction needs.

[0211] Find the peak spectrum line in the presence of aliasing. Specifically, use the spectrum line presentation method to directly observe the darkness of the spectrum line to judge the spectrum line intensity to find the peak spectrum line in the presence of aliasing, obtain the peak spectrum line with large black pixels, and obtain the peak spectrum line in the presence of aliasing.

[0212] The intensity calculation of the peak spectral line in the presence of aliasing is mainly based on the spectral line intensity formula:

[0213] Among them, B is the electromagnetic induction intensity and μ is the electromagnetic conductivity.

[0214] The first measurement intensity is obtained by the spectral line intensity formula. The first measurement intensity is subjected to a secondary intensity calculation according to the preset overlap coefficient to obtain the second measurement intensity. For example, if the first measurement intensity is 0.7 and the set overlap coefficient is ±0.1, the first measurement intensity is calculated according to the preset overlap coefficient to obtain 0.6 and 0.8. The first measurement intensity is obtained according to the preset overlap coefficient and the value is averaged, and the obtained average value is calculated using the spectral line intensity formula to obtain the second measurement intensity. It is determined whether the first measurement intensity and the second measurement intensity meet the preset accuracy requirements. If the preset accuracy requirements are met, the calculation is stopped, and the calibration compensation result is output to obtain the final spectrum analysis result.

[0215] The final spectrum analysis result is obtained by calibrating and compensating the initial spectrum analysis result, and the initial spectrum analysis result is further calibrated and compensated to improve the accuracy of the analysis result and effectively improve the locking accuracy of the fault position of the hydraulic cylinder.

[0216] The present application also provides a monitoring device, which includes:

[0217] Circumferential acoustic emission sensor;

[0218] Temperature sensor;

[0219] Pressure sensor;

[0220] a memory configured to store instructions; and

[0221] The processor is configured to call instructions from the memory and implement the above-mentioned hydraulic cylinder micro-leakage monitoring method based on information fusion when executing the instructions.

[0222] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0223] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0224] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0225] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0226] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0227] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0228] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0229] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0230] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for monitoring micro-leakage of hydraulic cylinders based on information fusion, characterized in that: Applied to a monitoring device, the method comprises: Collect circumferential acoustic emission sensor data, temperature sensor data and pressure sensor data; Inputting the circumferential acoustic emission sensor data, the temperature sensor data and the pressure sensor data into a pre-trained neural network model to obtain a time series feature vector set of the hydraulic cylinder fault; Determining the working state of the hydraulic cylinder according to the time series feature vector set of the hydraulic cylinder fault; Obtaining fault data according to the working state of the hydraulic cylinder; Locating the fault of the hydraulic cylinder by using the fault data to obtain a hydraulic cylinder fault locating result; According to the hydraulic cylinder fault location result, the fault position of the hydraulic cylinder is obtained; Extracting features of the fault location to obtain features of the fault location, and obtaining a frequency spectrum of the fault location based on the features of the fault location; According to the frequency spectrum of the hydraulic cylinder fault position, determining whether the frequency spectrum has electromagnetic wave interference; In the case where electromagnetic wave interference exists in the spectrum diagram of the fault position of the hydraulic cylinder, spectrum analysis is performed on the spectrum diagram to obtain an initial spectrum analysis result; Performing calibration compensation on the initial spectrum analysis result to obtain a final spectrum analysis result; The matching degree between the final spectrum analysis result and the spectrum diagram is calculated, and when the matching degree between the final spectrum analysis result and the spectrum diagram of the hydraulic cylinder fault position is greater than a first preset threshold, the fault position of the hydraulic cylinder is determined.

2. The method according to claim 1, characterized in that The working state of the hydraulic cylinder includes a normal state and a fault state, and determining the working state of the hydraulic cylinder according to the time series feature vector set of the hydraulic cylinder fault includes: Acquire circumferential acoustic emission signals, temperature signals and pressure signals; Building a fault feature model based on the time series feature vector set of the hydraulic cylinder fault; Dividing the circumferential acoustic emission signal, the temperature signal and the pressure signal within a preset time unit to obtain a first time series characteristic signal and a second time series characteristic signal; Inputting the first time series characteristic signal and the second time series characteristic signal into the fault characteristic model to obtain a first prediction result and a second prediction result; Calculate the matching degree between the first prediction result, the second prediction result and the working state to obtain a first matching degree and a second matching degree; If both the first matching degree and the second matching degree are greater than a preset matching degree, the first prediction result, the second prediction result and the working state of the hydraulic cylinder are compared to determine the working state of the hydraulic cylinder.

3. The method according to claim 1, characterized in that The method of locating the fault of the hydraulic cylinder by using the fault data to obtain a hydraulic cylinder fault locating result includes: Extracting features from the fault data to obtain fault data features; Inputting the fault data features into a pre-built genetic algorithm model to obtain an optimal fault location result; According to the optimal fault location result, a hydraulic cylinder fault location result is obtained.

4. The method according to claim 3, characterized in that: The step of inputting the fault data features into a pre-trained genetic algorithm model to obtain an optimal fault location result includes: S1. Generate a binary coding strategy according to the fault data characteristics, determine the maximum and minimum ranges of the initial value through the binary coding strategy and generate an initial group, wherein the binary coding strategy is the initial value of the fault data; S2. Calculate the fitness of individuals in the initial population based on the objective function, and record the optimal fitness of each generation. When the optimal fitness no longer changes, exit the iteration to obtain the first adaptive population. S3, performing crossover and mutation on the first adaptive population; S4, repeating steps S1 to S3, determining whether the fitness values ​​obtained in this and previous iterations reach a preset target fitness value, and if so, stopping iteration and outputting the hydraulic cylinder fault optimization positioning result.

5. The method according to claim 1, characterized in that After locating the fault of the hydraulic cylinder by using the fault data to obtain the hydraulic cylinder fault locating result, the method further includes: The fault data is stored in a preset diagnosis module, and a fault model tree is constructed; Determining the fault information of the hydraulic cylinder according to the hydraulic cylinder fault location result; The hydraulic cylinder fault location result is transmitted to the diagnosis module, and the hydraulic cylinder fault location result is compared with the fault model tree to obtain the hydraulic cylinder fault location information.

6. The method according to claim 1, characterized in that The determining, based on the frequency spectrum of the hydraulic cylinder fault position, whether there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position includes: Performing time-frequency analysis on the frequency spectrum of the hydraulic cylinder fault position by short-time Fourier transform to obtain a time-frequency diagram; Identify the electromagnetic interference feature to be measured on the time-frequency diagram, where the electromagnetic interference feature to be measured is a vertical stripe whose duration is less than a preset time period and whose frequency bandwidth is greater than a preset bandwidth; For each of the electromagnetic interference features, the interference intensity is calculated using a preset interference intensity calculation formula; Calculating the interference density of the electromagnetic interference feature on the time-frequency diagram; It is determined whether electromagnetic wave interference exists in the frequency spectrum of the hydraulic cylinder fault position according to the interference density and the interference intensity.

7. The method according to claim 6, characterized in that The determining, according to the interference density and the interference intensity, whether there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position includes: Calculating the mean, standard deviation, skewness and kurtosis of the interference intensity, and calculating the intensity dispersion degree according to the mean and the standard deviation, wherein the intensity dispersion degree is obtained by dividing the standard deviation by the mean; Calculating an intensity distribution shape index according to the skewness and the kurtosis, wherein the intensity distribution shape index is the sum of the absolute value of the skewness and the absolute value of the kurtosis; Calculating scores of the interference density, the intensity dispersion degree and the intensity distribution shape index respectively by corresponding scoring functions; If the scores of any two of the interference density, the intensity dispersion degree and the intensity distribution shape index are greater than the corresponding preset thresholds, it is determined that there is electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position; wherein, if the interference density is less than the preset score trough and when the intensity dispersion degree and the intensity distribution shape index are both greater than the corresponding preset thresholds, it is determined that there is no electromagnetic wave interference in the frequency spectrum of the hydraulic cylinder fault position.

8. The method according to claim 7, characterized in that The calibrating and compensating the initial spectrum analysis result to obtain the final spectrum analysis result includes: In the initial spectrum analysis result, discrete peaks exceeding a preset expected amplitude are determined, and median filtering is performed on the discrete peaks to remove the discrete peaks; Interpolating the frequency points of the removed discrete peaks to obtain interpolated spectrum analysis results; The baseline of the spectrum is determined in the interpolated spectrum analysis result, and the baseline is subtracted from the interpolated spectrum analysis result to perform baseline correction to obtain a final spectrum analysis result.

9. A monitoring device, characterized in that: include: Circumferential acoustic emission sensor; Temperature sensor; Pressure sensor; a memory configured to store instructions; as well as A processor is configured to call the instruction from the memory and implement the hydraulic cylinder micro-leakage monitoring method based on information fusion according to any one of claims 1 to 8 when executing the instruction.

Images