Optimal Selection of Marine Turbocharger Condition Monitoring Signals and Fault Diagnosis Method

Abstract

The present invention discloses a method for optimizing state monitoring signals and fault diagnosis of a marine turbocharger. The method includes: arranging a plurality of measuring points on the marine turbocharger and collecting vibration signals and their corresponding fault states; performing validity identification on the vibration signals, including time-domain detection and frequency-domain detection; if the vibration signals are valid in both time-domain detection and frequency-domain detection, then the vibration signals are valid data; performing variational mode decomposition on them, and selecting some IMF components for signal reconstruction to obtain reconstructed signals; calculating the improved information entropy of each reconstructed signal, and determining the preferred measuring points according to the improved information entropy; selecting effective time-domain and frequency-domain characteristic parameters, and training a neural network model by using the effective characteristic parameters of the vibration signals of the preferred measuring points and their corresponding fault states to complete the fault diagnosis of the marine turbocharger. The present invention completes the screening of sensitive measuring points based on the improved information entropy, and selects effective characteristic parameters to complete the fault diagnosis of the marine turbocharger.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis of marine turbochargers for engines, and specifically relates to a method for optimizing state monitoring signals and fault diagnosis of marine turbochargers. Background Art

[0002] Vibration signal monitoring, diagnosis and health state assessment of marine turbochargers can effectively evaluate and warn the actual operating state of turbochargers, thereby preventing major accidents of turbochargers. It is an important research direction for the design, operation and maintenance, and management of marine turbochargers. Its main processes include: arrangement of sensors on the body of marine turbochargers, vibration signal acquisition and storage, data validity judgment, calculation and analysis of vibration signal characteristic parameters, diagnosis algorithm development, etc. At present, the research on fault diagnosis and health state assessment of marine turbochargers mainly has the following problems:

[0003] (1) Before analyzing and processing the vibration signal of the turbocharger, interference signals and noise signals are not removed, and effective vibration signals are not discriminated. During the actual operation of the turbocharger, there are often many noise and interference signals. The noise mainly includes environmental white noise, etc., and the interference signals mainly include the interference of the vibration signal of the marine engine on the turbocharger signal, etc. If these signals are not removed and isolated, it will lead to misdiagnosis of the faults of the turbocharger.

[0004] (2) A single sensor or a small number of signal categories are used for fault diagnosis. The vibration signals of the turbocharger contain fault information such as blade and rotor imbalance and bearing wear in the X, Y, and Z directions. There are abnormal phenomena such as longitudinal vibration caused by axial end play, transverse vibration and flutter caused by the interaction between the blade and the fluid, and excessive vibration caused by abnormal bearing wear. At the same time, the energy of the measured point signal is related to the distance of the turbocharger. Therefore, only arranging a single sensor and arranging it in an inappropriate position will result in insufficient fault information contained in the collected signal, resulting in poor fault diagnosis effect.

[0005] (3) When using the data of multiple sensors for fault diagnosis, sensitive measurement points are not screened. Fault diagnosis requires the fusion of multi-measurement point signals. If sensitive measurement points are not screened and measurement points are blindly arranged, it will lead to too many measurement points and increased costs. At the same time, if data layer fusion of multiple measurement points is used, the data calculation rate will decrease due to excessive data volume, and the real-time performance will deteriorate. In addition, when collecting signals, the effective information contained in sensors at different positions is different. Without screening sensitive measurement points, it may lead to too much interference information in the final fusion features, resulting in a greater impact on the accuracy of fault diagnosis.

[0006] (4) The characteristic laws under variable speed conditions in the operating environment of the supercharger are not analyzed. Due to the fluctuations and non-uniformity in engine combustion, and at the same time, the load of the diesel engine changes due to the influence of the propeller load on the ship propulsion system, the speed of the turbocharger will change due to the change in exhaust gas energy, which makes the change law of the fault information contained in the vibration signal difficult to distinguish. For example, there are significant similarities in the time-domain diagram, frequency-domain diagram, and various characteristic parameters between the fault signal at a lower speed and the fault-free vibration signal at a higher speed. If the characteristic parameters of the vibration signal at a single speed are directly applied to variable speed conditions, there will be a problem of confusion between fault characteristics and normal characteristic signals.

[0007] In view of the above problems existing in the existing fault diagnosis methods for marine turbochargers, it is necessary to study the optimal layout of measuring points, signal optimization and anti-interference analysis, and diagnosis methods under variable conditions for marine turbochargers. Summary of the Invention

[0008] The purpose of the present invention is to provide a method for optimizing the selection of state monitoring signals and fault diagnosis of marine turbochargers, so as to solve at least the problems faced by the fault diagnosis method of marine turbochargers, such as difficult screening of sensitive measuring points, lack of effective discrimination of vibration signals, difficult selection of effective characteristic parameters, and poor adaptability to variable speed conditions.

[0009] First, the present invention provides a method for optimizing the selection of state monitoring signals and fault diagnosis of marine turbochargers, and the method includes:

[0010] Arrange multiple measuring points on the marine turbocharger, and collect vibration signals and their corresponding fault states;

[0011] Conduct validity identification on the collected vibration signals, including time-domain detection and frequency-domain detection;

[0012] If the vibration signal is effective in both time-domain detection and frequency-domain detection, then the vibration signal is valid data;

[0013] Perform variational mode decomposition on the vibration signals that are valid data to obtain each order of IMF components, and select some IMF components for signal reconstruction to obtain a reconstructed signal;

[0014] Calculate the improved information entropy of each reconstructed signal, and determine the optimal measuring points according to the improved information entropy;

[0015] Calculate the time-domain and frequency-domain characteristic parameters of the vibration signals at the optimal measuring points. With the speed as the abscissa and the characteristic parameter value as the ordinate, each time-domain and frequency-domain characteristic parameter generates a scatter plot. Each scatter plot contains different fault states of the marine turbocharger, and different fault states use different markers;

[0016] If the markers corresponding to different fault states in the scatter plot each occupy a region, the regions are clearly distinguishable from each other, and the variation trends of the markers corresponding to different fault states with the rotational speed are gentle, then the time-domain and frequency-domain characteristic parameters corresponding to this scatter plot are effective characteristic parameters;

[0017] Use the effective characteristic parameters of the vibration signals of the optimized measuring points and their corresponding fault states to train a neural network model;

[0018] Perform single-fault cross-validation and multi-fault validation on the trained neural network model. The data used for single-fault cross-validation are all data containing only one fault state, and the data used for multi-fault validation are data containing multiple fault states. Containing multiple fault states means that a piece of data not only contains the fault state to be diagnosed but also contains fault states that do not need to be diagnosed, so as to verify whether the model can extract the fault state to be diagnosed from multiple fault states;

[0019] If the accuracies of both single-fault cross-validation and multi-fault validation meet the requirements, then the neural network model is qualified, and complete the fault diagnosis of the marine turbocharger based on the qualified neural network model.

[0020] In some embodiments, multiple measuring points are arranged in different directions, at different distances, and at different component positions of the marine turbocharger to collect vibration signals at different directions, at different distances, and at different component positions of the marine turbocharger.

[0021] In some embodiments, there are 8 measuring points, and the arrangement is as follows:

[0022] The first to fifth measuring points are in the Z direction along the vertical direction, the seventh and eighth measuring points are in the X direction along the axial direction of the marine turbocharger rotor, and the sixth measuring point is in the Y direction perpendicular to the X direction and the Z direction; among them, the X direction represents the longitudinal vibration caused by the axial movement of the turbocharger rotor, and the Y direction and the Z direction represent the transverse vibration and flutter generated by the interaction between the blades and the fluid and bearing wear;

[0023] The first to fourth measuring points are at the first distance from the turbocharger, and the fifth to eighth measuring points are at the second distance from the turbocharger, and the first distance is greater than the second distance;

[0024] The first measuring point and the fourth to eighth measuring points are arranged at the compressor end of the turbocharger, and the second and third measuring points are arranged at the turbine end of the turbocharger.

[0025] In some embodiments, the time-domain detection includes:

[0026] Determine whether the proportion of data points with voltage values less than or equal to the preset amplitude in the vibration signal is greater than or equal to the first proportion; if so, the vibration signal is invalid; if not, perform least-squares detrending on the vibration signal, and calculate the effective value of the vibration signal after detrending. If the effective value exceeds the preset threshold, the vibration signal is invalid; if the effective value does not exceed the preset threshold, the vibration signal is valid;

[0027] Frequency domain detection includes:

[0028] Obtain the spectrum of the vibration signal through Fourier transform, and combine with the turbine supercharger speed to calculate the ratio of the frequency domain energy below the fundamental frequency of the turbine supercharger spectrum to the overall frequency domain energy. If the ratio is greater than or greater than or equal to the preset ratio, the vibration signal is invalid; otherwise, the vibration signal is valid.

[0029] In some of the embodiments, performing least-squares detrending on the vibration signal includes:

[0030] Calculate b and a:

[0031]

[0032]

[0033] In the formula, represents the vibration signal, represents time, represents the number of data points of the vibration signal;

[0034] Calculate the vibration signal after detrending:

[0035]

[0036] In the formula, represents the vibration signal after detrending.

[0037] In some of the embodiments, selecting some IMF components for signal reconstruction to obtain a reconstructed signal includes:

[0038] If the maximum measurement frequency of the vibration sensor is , and the order number containing in each order of IMF components is the th order, then remove the IMF components from the 1st order to the th order;

[0039] If the fundamental frequency of the turbine supercharger speed is , and the order number containing in each order of IMF components is the th order, then select the highest order number of IMF components as the th order;

[0040] Sum the IMF components from the -th order to the -th order to obtain the reconstructed signal.

[0041] In some embodiments, calculate the improved information entropy of each reconstructed signal, and determine the preferred measurement points according to the improved information entropy, including:

[0042] Calculate the information parameters of the frequency-domain signal of the IMF components of the reconstructed signal:

[0043]

[0044] Wherein, is the energy parameter of the i -th IMF component, is the number of data points of the frequency-domain signal, is the i -th j spectral line in the spectrum of the -th IMF component, is the corresponding amplitude;

[0045] Calculate the total energy of the IMF components:

[0046]

[0047] Wherein, represents the number of IMF components of the reconstructed signal;

[0048] Calculate the probability of the event :

[0049]

[0050]

[0051] Calculate the improved information entropy:

[0052]

[0053]

[0054] Wherein, is the total number of rotational speed intervals, k is the k -th rotational speed interval, is the improved information entropy of the -th rotational speed interval calculated based on and k is the sum of the improved information entropy of all rotational speed intervals;

[0055] Take the average of the multi-state improved information entropy :

[0056]

[0057] In the formula, represents the number of fault states of the marine turbocharger, represents the th fault state of the marine turbocharger;

[0058] Select several of the largest measurement points or measurement points that exceed the preset information entropy threshold as the optimal measurement points.

[0059] In some of the embodiments, the method further includes:

[0060] Save the vibration signals that are valid data as follows:

[0061] Cut the signal into multiple sub-signals according to a specified length;

[0062] Add labels to each sub-signal, where the labels include rotational speed, signal type, measurement point location and direction, and timestamp; the signal type includes vibration acceleration signal and shaft center orbit signal;

[0063] Add the fault state corresponding to the sub-signal to the label.

[0064] In some of the embodiments, the fault states of the marine turbocharger include normal state, rotor slightly unbalanced state, and rotor severely unbalanced state;

[0065] The time-domain and frequency-domain characteristic parameters of the vibration signal include at least one of maximum value, mean value, absolute mean value, root mean square amplitude, mean square value, effective value, variance, standard deviation, skewness, kurtosis, waveform factor, peak factor, impulse factor, margin index, kurtosis index, skewness index, amplitude mean value, center frequency, mean square frequency, frequency variance, root mean square frequency, amplitude variance, and frequency standard deviation;

[0066] Perform a cubic polynomial curve fitting on the markers corresponding to each fault state in the scatter plot. If the amplitude variation of the fitting curve within the entire rotational speed range is less than the preset variation threshold, it is considered that the trend of the markers corresponding to different fault states in this scatter plot with respect to rotational speed is flat;

[0067] The neural network model is a one-dimensional convolutional neural network;

[0068] Before training the neural network model, perform dimensionality reduction processing on the high-dimensional feature vector composed of the effective characteristic parameters of the vibration signals of the selected measurement points through the t-distributed stochastic neighbor embedding dimensionality reduction algorithm.

[0069] According to the second aspect of the present invention, a marine turbocharger is provided, and the marine turbocharger uses the method for optimizing the state monitoring signals and diagnosing faults of the marine turbocharger described in any one of the first aspects to complete the fault diagnosis of the marine turbocharger.

[0070] Generally speaking, compared with the prior art by the above technical solutions conceived by the present invention, the following beneficial effects can be achieved:

[0071] The present invention proposes a method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger. Through the design of a preliminary test plan, the identification and preservation of the effectiveness of monitoring signals, the optimization and arrangement of measuring points, the multi-signal data fusion and diagnostic algorithm research based on the optimized measuring points, and the combination of a variable-condition fault diagnosis algorithm library function package, the effective eigenvalue selection, sensitive measuring point screening, fault diagnosis and health status evaluation of the marine turbocharger under variable rotational speed conditions are realized with convenience, high calculation efficiency and high accuracy.

[0072] (1) An overall strategy for the preliminary test of the turbocharger fault is proposed, including measuring point arrangement, data cleaning, and data preservation. The vibration signals in all directions of the turbocharger can be obtained more completely and representatively, and the effective signals are saved as directly usable data sets.

[0073] (2) The calculation method of information entropy is improved, and it is innovatively applied to the field of sensitive measuring point screening, realizing the screening of sensitive measuring points when a large number of measuring point positions are arranged. At the same time, the proposed measuring point screening method can provide general guidance for the measuring point arrangement method of the vibration sensors of the marine turbocharger, can reduce the number of measuring point arrangements, and effectively reduce the experimental cost.

[0074] (3) Through the analysis of the laws of various characteristic parameters of the vibration signals, a method for judging whether the characteristic parameters are effective by using a rotational speed-parameter scatter plot is proposed, and effective characteristic parameters that are sensitive to faults and have a small change trend with the rotational speed can be selected.

[0075] (4) A dimensionality reduction algorithm is used to perform dimensionality reduction processing on the high-dimensional feature vectors, removing redundant information and improving the operation efficiency; and a convolutional neural network is used to realize the recognition and classification of the feature vectors, with high accuracy.

[0076] Combined with the above effects, the method for optimizing the state monitoring signals and fault diagnosis of marine turbochargers proposed by the present invention analyzes and processes the vibration signals of marine turbochargers, realizing the fault diagnosis of marine turbochargers. Compared with other existing methods, the method for optimizing the state monitoring signals and fault diagnosis of marine turbochargers proposed by the present invention has significant advantages in terms of the screening effect of sensitive measuring points and the adaptability to variable rotational speed working conditions. The method proposed by the present invention has high practical value for monitoring the operating state of marine turbochargers and ensuring the safe and stable operation of the power system of ship engines. Description of the Drawings

[0077] Figure 1 It is the overall flowchart of a method for optimizing the state monitoring signals and fault diagnosis of a marine turbocharger provided by an embodiment of the present application;

[0078] Figure 2 It is a sensor layout plan provided by an embodiment of the present application;

[0079] Figure 3 It is the flowchart for identifying the effectiveness of monitoring signals provided by an embodiment of the present application;

[0080] Figure 4 It is the flowchart for data saving provided by an embodiment of the present application;

[0081] Figure 5 It is the flowchart for optimizing measuring points provided by an embodiment of the present application;

[0082] Figure 6 It is a schematic diagram of a data set provided by an embodiment of the present application;

[0083] Figure 7 It is the flowchart for multi-fault verification provided by an embodiment of the present application;

[0084] Figure 8 It is the flowchart for fault diagnosis under variable rotational speed provided by an embodiment of the present application;

[0085] Figure 9 It is the flowchart for the software development of fault diagnosis under variable working conditions provided by an embodiment of the present application. Detailed Embodiment

[0086] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. Based on the embodiments provided in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0087] Obviously, the accompanying drawings in the following description are only some examples or embodiments of the present application. For those of ordinary skill in the art, without creative efforts, the present application can also be applied to other similar scenarios based on these drawings. In addition, it can also be understood that although the efforts made in such a development process may be complex and lengthy, for those of ordinary skill in the art related to the content disclosed in the present application, some design, manufacturing, or production changes based on the technical content disclosed in the present application are only conventional technical means and should not be understood as insufficient disclosure of the content of the present application.

[0088] In the present application, the mention of "embodiment" means that the specific features, structures, or characteristics described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various positions in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those of ordinary skill in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments without conflict.

[0089] Unless otherwise defined, the technical terms or scientific terms involved in the present application should be of the ordinary meaning understood by those with ordinary skills in the technical field to which the present application belongs. The terms "a", "an", "one", "the" and similar words involved in the present application do not indicate a quantity limitation and can represent a singular or plural number. The terms "include", "comprise", "have" and any variations thereof involved in the present application are intended to cover non-exclusive inclusion; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include unlisted steps or units, or may also include other steps or units inherent to these processes, methods, products or devices. The terms "connect", "couple" and similar words involved in the present application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The term "plurality" involved in the present application refers to two or more. "And / or" describes the association relationship of associated objects and indicates that three relationships may exist. For example, "A and / or B" may represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after. The terms "first", "second", "third", etc. involved in the present application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0090] This application provides a method for arranging measuring points and optimizing signals of marine turbochargers, as well as a method for judging signal validity and fault diagnosis, which is applicable to the design and development requirements of related products for online fault diagnosis and health status assessment in aspects such as the arrangement and selection principles of sensors for marine turbocharger monitoring, signal acquisition and analysis, signal validity identification methods, intelligent diagnosis algorithms, and monitoring and diagnosis software.

[0091] Most of the existing fault diagnosis methods for marine turbochargers face problems such as difficult screening of sensitive measuring points, lack of validity discrimination for vibration signals, difficult selection of effective characteristic parameters, and poor adaptability to variable speed conditions. The present invention proposes a method for optimizing health monitoring signals and diagnosing faults under variable conditions of marine turbochargers, selects effective characteristic parameters, screens sensitive measuring points, and develops an intelligent diagnosis algorithm, and solves the following problems:

[0092] (1) Aiming at the problem of interference and noise in vibration signals, the online monitoring data of the turbocharger is cleaned. According to the statistical characteristics of the vibration signals, abnormal data points in the vibration signals are removed and replaced with reliable data points, improving the authenticity and reliability of the vibration signals;

[0093] (2) Aiming at the problems of insufficient measuring points and few signal types, vibration sensors are arranged at multiple positions and in multiple directions to collect vibration signals of multiple characteristic parts and representative directions of the turbocharger; and combined with the rotor center orbit for fault diagnosis of the turbocharger, solving the problem of incomplete acquisition of fault characteristics caused by single measuring points and data types;

[0094] (3) Aiming at the problem of blindly arranging and selecting measuring points, an improved information entropy calculation method is proposed, and a measuring point screening method is invented based on the improved information entropy, solving the adverse effects of information redundancy and interference signals on the diagnosis rate and accuracy when blindly selecting too many measuring point signals, and reducing the number of sensor arrangements and saving costs;

[0095] (4) Aiming at the problem of poor adaptability of existing algorithms to rotational speed, through the trend analysis of characteristic parameters of the vibration signals of the turbocharger at different rotational speeds, effective characteristic parameters that are sensitive to faults and have a relatively stable trend with rotational speed changes can be selected, avoiding the problem of poor adaptability of characteristic parameter selection to rotational speed.

[0096] Finally, a method for testing the single-fault - multi-fault data model is designed, providing a method for verifying the fault diagnosis model from multiple perspectives such as accuracy, stability, and generalization ability.

[0097] The method for optimizing monitoring signals and diagnosing faults of marine turbochargers provided by this application, such as Figure 1As shown in the figure, its process mainly includes five parts: the design of the preliminary test plan, the identification and preservation of the effectiveness of monitoring signals, the optimization and layout of measuring points, the multi-signal data fusion and diagnostic algorithm research based on the optimized measuring points, and the development of the function package of the variable working condition fault diagnosis algorithm. Through the combination of the above five parts, an efficient, convenient and highly diagnostic accuracy monitoring point optimization plan and diagnostic algorithm for marine turbochargers are formed, realizing the fault identification and health status assessment of the turbocharger under variable rotational speeds and anti-external interference during actual operation. The specific technical solutions are as follows:

[0098] 1. Design of the preliminary test plan

[0099] The layout method of measuring points for the preliminary test is as Figure 2 shown. In the preliminary test, multiple measuring points are arranged around the turbocharger and need to include the three directions of X, Y, and Z. The purpose is to obtain as many vibration signals as possible at different positions and in different directions of the turbocharger, providing sufficient effective original signal references for subsequent signal processing, measuring point screening, etc. The sensor layout method is as Figure 2 shown. Measuring points 1 to 5 are in the Z direction along the vertical direction, measuring points 7 and 8 are in the X direction along the axial direction of the marine turbocharger rotor, and measuring point 6 is in the Y direction perpendicular to the X and Z directions. Among them, the X direction represents the longitudinal vibration caused by the axial movement of the turbocharger rotor, and the Y and Z directions represent the transverse vibration and flutter generated by the interaction between the blades and the fluid, bearing wear, etc.; measuring points 1-4 are far from the turbocharger, and measuring points 5-8 are close to the turbocharger. Using the layout method of combining far and near measuring points can verify the useful information contained in the turbocharger vibration signals at different propagation distances; measuring points 1, 4, 5, 6, 7, and 8 are arranged at the compressor end of the turbocharger, and measuring points 2 and 3 are arranged at the turbine end of the turbocharger. Arranging measuring points at both the compressor end and the turbine end can verify the information content of the turbocharger vibration signals at different component locations. Therefore, the developed layout method of measuring points for the preliminary test enables the arranged 8 measuring points to obtain the vibration signals of the turbocharger more completely and representatively through the combination of different directions, different distances, and different component positions, laying a foundation for the subsequent measuring point screening work.

[0100] 2. Identification and preservation of the effectiveness of monitoring signals

[0101] After the measuring points are arranged, signals are collected, and at the same time, the effectiveness of the collected signals is identified and the data is saved. As Figure 3 shown, the identification of the effectiveness of monitoring signals is divided into two parts: time-domain detection and frequency-domain detection.

[0102] Time-domain detection is divided into three parts:

[0103] (1) Measure whether the time-domain signal is very weak or even zero. Here, the definition of a weak signal is that the voltage values of 80% or more of the data points are less than or equal to 10 mV. If this occurs, it indicates that there is a problem with the corresponding signal acquisition channel and the vibration acceleration signal cannot be acquired. The signal should be masked or deleted, and the acquisition channel should be checked.

[0104] (2) To avoid signal drift, perform detrending on the signal. Perform least-squares detrending on the signal according to the following formula:

[0105] ① Establish a model:

[0106]

[0107] ② Calculate the parameters b and a:

[0108]

[0109]

[0110] ③ Calculate the signal after detrending: :

[0111]

[0112] (3) Calculate the effective value characteristic parameters of the signal after detrending, and compare the calculated values with other adjacent measuring points in the same direction. If there are no adjacent measuring points in the same direction, compare with historical data or empirical values. If the effective value exceeds the normal value range by more than 15%, the signal of this channel is judged as an abnormal signal and is masked or deleted. The frequency-domain detection is as follows:

[0113] Frequency-domain detection:

[0114] The frequency-domain detection obtains the spectrum of the time-domain signal through Fourier transform. Combining with the speed of the turbocharger, calculate the ratio of the frequency-domain energy below the fundamental frequency of the turbocharger spectrum to the overall frequency-domain energy. If the ratio is greater than 10%, it is determined that the energy of the signal in the low-frequency region is unreasonable, and the signal of this channel is defined as an abnormal signal and is masked or deleted. Calculate the frequency-domain energy below the fundamental frequency according to the following formula:

[0115] ① Calculate the fundamental frequency of the turbocharger: :

[0116]

[0117] Where: is the speed of the turbocharger, with the unit of rpm, The unit of

[0118] ② Calculate the energy in the frequency domain below the fundamental frequency :

[0119]

[0120] In the formula, is the energy in the frequency domain below the fundamental frequency, is the frequency of the signal, The maximum value of is the fundamental frequency of the turbocharger , is the amplitude corresponding to the th spectral line.

[0121] If the signal passes the time-domain detection and frequency-domain detection, the signal is determined as a qualified signal to implement data cleaning. After cleaning the data, the valid data is saved. The data is directly saved in the form of an available data set for subsequent data calculation and the development of fault diagnosis algorithms.

[0122] As Figure 4 shown, the data saving is mainly divided into five parts:

[0123] (1) Signal segmentation: The collected signal is segmented into multiple sub-signals according to a length of 0.25 s for each segment.

[0124] (2) Adding labels: Measuring point and working condition information are added to the end of each segment of data. The label needs to contain the following information: rotational speed, signal type (vibration acceleration signal, shaft center orbit signal), measuring point position and direction, time stamp.

[0125] (3) Distinguishing the saving interval: The segmented and labeled signal is used for fault diagnosis. If the fault diagnosis result is normal, the data is saved every 60 minutes; if the fault diagnosis result is a fault, the data is saved every 1 minute; and the fault diagnosis result is added to the label of the data. The data here is saved to the temporary database.

[0126] (4) Data sorting: The program performs data cleaning every 24 hours, identifies the fault diagnosis labels of the saved data. If the fault diagnosis label is normal, the redundant normal data is deleted according to the time stamp, and only the most recent normal data is retained; if the fault diagnosis label is a fault, all the data is retained. The data here is saved to the long-term database, as Figure 6 shown.

[0127] (5) After the data sorting is completed, the temporary database is cleared to prepare for the next saving.

[0128] 3. Measuring point optimization and layout

[0129] As Figure 5As shown in the figure, the measuring point optimization and layout module includes four parts: vibration signal decomposition, signal reconstruction, improved information entropy calculation, and sensitive measuring point selection:

[0130] (1) Vibration signal decomposition: The original vibration signal is subjected to variational mode decomposition (VMD) calculation, and each segment of the vibration signal is decomposed into multiple intrinsic mode functions (IMFs). The first layer is the component with the highest frequency in the vibration signal, decreasing in turn, and the last layer is the required lowest frequency component.

[0131] (2) Signal reconstruction: Select some IMF components after the decomposition of the vibration signal. If the maximum measurement frequency of the vibration sensor is , and the order containing in each order of IMF components is the th order, then the IMF components from the 1st order to the th order are removed. The purpose is to remove the high-frequency noise in the signal and the signal outside the measurement range of the sensor. If the fundamental frequency of the rotational speed of the turbocharger is , and the order containing in each order of IMF components is the th order, then the highest order of the selected IMF components is the th order. The purpose is to remove the interference vibration signals with lower frequencies, such as the vibration signals of internal combustion engines, etc., and at the same time retain the vibration signals of the turbocharger at lower rotational speeds as much as possible. After selecting the appropriate order of IMF, the selected IMF components are added together to obtain the reconstructed signal.

[0132] The traditional method for calculating the energy entropy of vibration signals only uses the amplitudes of the time-domain or frequency-domain signals of each order of components, which all belong to the ordinate values of the signals, and does not use the time or frequency values of their abscissas. This calculation method ignores the information amount of one dimension of the signal, and using this method, it is impossible to calculate the entire information amount of the signal completely. Therefore, it is necessary to improve the traditional energy entropy calculation method by adding the frequency values of the abscissas of the frequency-domain signals to the information entropy calculation process, which is the improved information entropy.

[0133] (3) Improved information entropy calculation: Calculate the improved information entropy of the reconstructed signal of each measuring point according to the following five steps to evaluate the information amount contained in the signals collected by each measuring point.

[0134] ① Calculate the information parameters of the frequency-domain signal of the IMF component:

[0135]

[0136] In the formula: is the energy parameter of the i th IMF component, is the number of data points of the frequency-domain signal, is the iThe frequency value corresponding to the j th spectral line in the spectrum of each IMF component, is the corresponding amplitude.

[0137] ② Calculate the total energy of the m IMF components:

[0138]

[0139] ③ Calculate the probability of the event:

[0140]

[0141]

[0142] ④ Calculate the improved information entropy:

[0143]

[0144]

[0145] In the formula: k is the rotational speed. For example, from 35000 rpm to 60000 rpm, there are 11 rotational speeds in total (the step interval of the rotational speed during data acquisition is 2500 rpm), so the value of k ranges from 1 to 11. is based on and calculated improved information entropy of the k th rotational speed interval, is the total improved information entropy of all rotational speed intervals.

[0146] ⑤ Take the average of the multi-state information entropy:

[0147] To avoid the contingency and instability of the information entropy value of a single fault state, it is necessary to take the average of the information entropy of all fault states. For example, there are three states: normal state, slight rotor imbalance, and severe rotor imbalance. Then take the average of the information entropy of these three states:

[0148]

[0149] (4)Selection of sensitive measurement points Sort the information entropy of each measurement point obtained in the improved information entropy calculation, and select the 4 measurement points with the largest information entropy as sensitive measurement points.

[0150] The sensitive measurement points selected according to this method can be used as the measurement point layout positions for vibration signal acquisition of marine turbochargers.

[0151] 4. Research on multi-signal data fusion and diagnostic algorithm based on optimized measurement points

[0152] AsFigure 8 As shown in Figure 8 , the research module of multi-signal data fusion and diagnosis algorithm based on preferred measuring points includes five parts: characteristic parameter calculation, characteristic parameter trend analysis, effective characteristic parameter selection, feature vector dimensionality reduction, and fault diagnosis.

[0153] (1) Characteristic parameter calculation calculates the common time-domain and frequency-domain characteristic parameters of the vibration signal of marine turbochargers through formulas, and its calculation methods are shown in Table 1.

[0154] Table 1 Calculation method table of characteristic parameters

[0155]

[0156] (2) Characteristic parameter trend analysis will calculate the time-domain and frequency-domain characteristic parameters of each sub-signal according to the data saved in the data saving module using the characteristic parameter calculation method described in Table 1, and analyze the change trend of the characteristic parameters using a scatter plot. The method is as follows: the abscissa is the rotational speed, and the ordinate is the characteristic parameter value. A scatter plot is generated for each characteristic parameter, and each scatter plot should include the characteristic values of signals with different fault degrees (for example: normal state, rotor slightly unbalanced state, rotor severely unbalanced state). The scatter points of each fault degree are in different colors. The principles for judging the effectiveness of characteristic parameters based on the scatter plot are as follows:

[0157] ① In a scatter plot, the scatter points of each color occupy a separate area, and the scatter points of different colors are clearly distinguished.

[0158] ② On the premise of meeting condition ①, perform a cubic polynomial curve fitting on the scatter points of each color. If the amplitude variation of the fitting curve of the scatter points of each color within the entire rotational speed range is less than 15%, it is considered that the change trend of this characteristic parameter with the rotational speed is gentle, and the characteristic parameter represented by this scatter plot can be selected as an effective characteristic parameter. The amplitude variation of the fitting curve is calculated according to the following formula:

[0159]

[0160] In the formula, is the amplitude variation of the fitting curve, is the maximum value of the fitting curve, is the minimum value of the fitting curve.

[0161] Through this method, the selected effective characteristic parameters are the centroid frequency, mean square frequency, and root mean square frequency.

[0162] (3) Effective characteristic parameter selection summarizes the centroid frequency, mean square frequency, and root mean square frequency selected in the characteristic parameter trend analysis as effective characteristic parameters.

[0163] (4)Feature vector dimensionality reduction The high-dimensional feature vector composed of the effective feature parameters of the selected measuring points in the sensitive measuring point selection is processed by the t-distributed stochastic neighbor embedding dimensionality reduction algorithm to reduce the high-dimensional feature vector to three dimensions.

[0164] (5)Data classification The dimensionality-reduced feature parameters are classified and diagnosed through a one-dimensional convolutional neural network. The diagnosis is divided into two parts: single-fault cross-validation and multi-fault validation.

[0165] The data used for single-fault cross-validation only contains data with one type of fault, such as: normal state data, rotor slightly unbalanced fault data, rotor severely unbalanced fault data.

[0166] After the single-fault cross-validation, if the average accuracy rate ≥ 85%, it is determined to be qualified.

[0167] After the single-fault cross-validation is qualified, to further verify the robustness and generality of the fault diagnosis algorithm, multi-fault data can be further used for fault diagnosis verification.

[0168] The data used for multi-fault validation contains data with multiple types of faults, such as: normal state data, rotor slightly unbalanced fault data, rotor severely unbalanced + bearing wear double-fault data. And when performing multi-fault validation, the neural network model should not be retrained, but the neural network model trained with single-fault data should be directly used. The purpose of using multi-fault data for fault diagnosis is to test whether the algorithm used has generality, that is, whether the fault to be diagnosed can be extracted from the multi-fault data, such as: whether the rotor severely unbalanced fault can be diagnosed using the rotor severely unbalanced + bearing wear double-fault data, rather than being diagnosed as a normal signal or a rotor slightly unbalanced fault.

[0169] After the multi-fault validation, if the average accuracy rate ≥ 85%, it is determined to be qualified.

[0170] The process of multi-fault validation is as Figure 7 shown.

[0171] If the single-fault cross-validation is qualified, it proves that the algorithm has a high accuracy rate; if both the single-fault cross-validation and the multi-fault validation are qualified, it not only proves that the algorithm has a high accuracy rate, but also has high robustness and generality. This verification method can provide a reference for the effect evaluation of the fault diagnosis model.

[0172] 5. Software development for fault diagnosis under variable working conditions

[0173] As Figure 9As shown in the figure, the variable operating condition fault diagnosis software is developed using the MATLAB software platform for developing fault diagnosis algorithms, and the monitoring and diagnosis software is developed using the virtual instrument LabVIEW software. The developed monitoring and diagnosis software includes a signal acquisition parameter setting module, a signal acquisition module, a data display module, a fault diagnosis module, etc.

[0174] When using LabVIEW for fault diagnosis, it is necessary to use the algorithm script import function of LabVIEW software to integrate the fault diagnosis algorithm developed using MATLAB into LabVIEW, and realize the integrated process of signal acquisition and fault diagnosis based on virtual instruments through the method of mixed programming of M and G languages.

[0175] After developing software using LabVIEW, the project may contain multiple sub-files (VIs). At this time, it is necessary to package the relevant VIs in the form of a LabVIEW library function package (lvlib). By this method, the relevant VIs can be managed in an orderly manner. In addition, Ivlib allows setting access permissions, and the access permissions of VIs can be freely set to specify which VIs are public and which are private. This helps to protect sub-VIs that are not expected to be directly called and improve the security of the code.

[0176] In addition, it is necessary to integrate and package the monitoring and diagnosis software project developed using LabVIEW, and finally generate an executable file (.exe file) so that the monitoring and diagnosis software can run on a computer without installing LabVIEW and MATLAB software.

[0177] In summary, this application provides a method for optimizing the state monitoring signals and fault diagnosis of marine turbochargers. Through the design of the preliminary test plan, the identification and preservation of the effectiveness of monitoring signals, the optimization and layout of measuring points, the multi-signal data fusion and diagnosis algorithm research based on the optimized measuring points, and the combination of the variable operating condition fault diagnosis algorithm library function package, it realizes the effective eigenvalue selection, sensitive measuring point screening, fault diagnosis and health status assessment of marine turbochargers under variable speed operating conditions in a convenient, fast, high calculation efficiency and high accuracy manner.

[0178] (1) The overall strategy of the preliminary test for turbocharger faults is proposed, including measuring point layout, data cleaning, and data preservation. The vibration signals in all directions of the turbocharger can be obtained more completely and representatively, and the effective signals are saved as directly available data sets.

[0179] (2) The calculation method of information entropy is improved and innovatively applied to the field of sensitive measuring point screening, realizing the screening of sensitive measuring points when arranging a large number of measuring point positions. At the same time, the proposed measuring point screening method can provide general guidance for the vibration sensor measuring point arrangement method of marine turbochargers, reduce the number of measuring points arranged, and effectively reduce the experimental cost.

[0180] (3) Through the analysis of the laws of various characteristic parameters of vibration signals, a method for judging the effectiveness of characteristic parameters by means of a speed-parameter scatter plot is proposed, which can select effective characteristic parameters that are sensitive to faults and have a small change trend with speed.

[0181] (4) The dimensionality reduction algorithm is used to perform dimensionality reduction processing on high-dimensional feature vectors, removing redundant information and improving the operation efficiency; and a convolutional neural network is used to realize the recognition and classification of feature vectors, with high accuracy.

[0182] (5) Through the method of mixed programming of MATLAB and LabVIEW, a fault diagnosis algorithm and data monitoring and fault diagnosis software are developed. The relevant sub-VIs are integrated into the lvlib library file package, enhancing the manageability of project files, and the final data monitoring and fault diagnosis software is compiled into an executable file, realizing the integrated process of data acquisition and fault diagnosis based on virtual instruments.

[0183] Combining the above effects, the marine turbocharger fault diagnosis method proposed by the present invention analyzes and processes the vibration signals of marine turbochargers, realizing the fault diagnosis of marine turbochargers. Compared with other existing methods, the marine turbocharger fault diagnosis method proposed by the present invention has significant advantages in the sensitive measuring point screening effect and the adaptability to variable speed operating conditions. The method proposed by the present invention has high practical value for monitoring the operating state of marine turbochargers and ensuring the safe and stable operation of the ship engine power system.

[0184] It should be noted that the technical features of the above-described embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combinations of these technical features do not conflict, they should be considered as within the scope described in this specification. In addition, according to the needs of implementation, each step / component described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0185] Those skilled in the art can easily understand that the above-described embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A method for optimizing state monitoring signals and diagnosing faults of a marine turbocharger, characterized in that, The method includes: Arranging a plurality of measurement points on a marine turbocharger, and collecting vibration signals and their corresponding fault states; Performing validity identification on the collected vibration signals, including time-domain detection and frequency-domain detection; If the vibration signal is valid in both time-domain detection and frequency-domain detection, then the vibration signal is valid data; Performing variational mode decomposition on the vibration signal that is valid data to obtain each order of IMF components, and selecting some IMF components for signal reconstruction to obtain a reconstructed signal; Calculating the improved information entropy of each reconstructed signal, and determining the optimal measurement points according to the improved information entropy; Calculating the time-domain and frequency-domain characteristic parameters of the vibration signals of the optimal measurement points, using the rotational speed as the abscissa and the characteristic parameter values as the ordinate, generating a scatter plot for each time-domain and frequency-domain characteristic parameter, each scatter plot including different fault states of the marine turbocharger, and using different markers for different fault states; If the markers corresponding to different fault states in the scatter plot respectively occupy a region, the regions are clearly distinguishable from each other, and the variation trend of the markers corresponding to different fault states with the rotational speed is gentle, then the time-domain and frequency-domain characteristic parameters corresponding to this scatter plot are valid characteristic parameters; Training a neural network model using the valid characteristic parameters of the vibration signals of the optimal measurement points and their corresponding fault states; Performing single-fault cross-validation and multi-fault validation on the trained neural network model. The data used for single-fault cross-validation are all data containing only one fault state, and the data used for multi-fault validation are data containing multiple fault states. Containing multiple fault states means that a data not only contains the fault state to be diagnosed but also contains fault states that do not need to be diagnosed, to verify whether the model can extract the fault state to be diagnosed from multiple fault states; If the accuracies of both single-fault cross-validation and multi-fault validation meet the requirements, then the neural network model is qualified, and based on the qualified neural network model, the fault diagnosis of the marine turbocharger is completed; Among them, calculating the improved information entropy of each reconstructed signal and determining the optimal measurement points according to the improved information entropy includes: Calculating the information parameters of the frequency-domain signals of the IMF components of the reconstructed signal: wherein, is the energy parameter of the i th IMF component, is the number of data points of the frequency-domain signal, is the i th frequency value corresponding to the j th spectral line in the spectrum of the th IMF component, and is the corresponding amplitude; Calculation The total energy of the IMF components : In the formula, represents the number of IMF components of the reconstructed signal; Calculate the probability of an event : Calculating the improved information entropy: In the formula, is the total number of rotational speed intervals, k is the k th rotational speed interval, is the improved information entropy of the and calculated for the k th rotational speed interval, is the total improved information entropy of all rotational speed intervals; Take the average of the multi-state improved information entropy : In the formula, represents the number of fault states of the marine turbocharger, represents the th fault state of the marine turbocharger; Select a number of the largest measurement points or measurement points exceeding a preset information entropy threshold as the optimal measurement points.

2. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 1, wherein Arranging a plurality of measurement points in different directions, at different distances, and at different component positions of the marine turbocharger to collect vibration signals at different directions, at different distances, and at different component positions of the marine turbocharger.

3. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 2, wherein There are 8 measurement points, and the arrangement method is as follows: The first measurement point to the fifth measurement point are in the Z direction along the vertical direction, the seventh measurement point and the eighth measurement point are in the X direction along the axial direction of the marine turbocharger rotor, and the sixth measurement point is in the Y direction perpendicular to the X direction and the Z direction; among them, the X direction represents the longitudinal vibration caused by the axial displacement of the marine turbocharger rotor, and the Y direction and the Z direction represent the transverse vibration and flutter generated by the interaction between the blades and the fluid and bearing wear; The first measurement point to the fourth measurement point are at the first distance from the marine turbocharger, the fifth measurement point to the eighth measurement point are at the second distance from the marine turbocharger, and the first distance is greater than the second distance; The first measurement point and the fourth measurement point to the eighth measurement point are arranged at the compressor end of the marine turbocharger, and the second measurement point and the third measurement point are arranged at the turbine end of the marine turbocharger.

4. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 1, characterized in that, The time-domain detection includes: Judge whether the proportion of data points with voltage values less than or equal to the preset amplitude in the vibration signal is greater than or equal to the first proportion; if so, the vibration signal is invalid; if not, perform least-squares detrending on the vibration signal, and calculate the effective value of the vibration signal after detrending. If the effective value exceeds the preset threshold, the vibration signal is invalid; if the effective value does not exceed the preset threshold, the vibration signal is valid; Frequency domain detection includes: Obtain the spectrum of the vibration signal through Fourier transform, and combine with the turbocharger speed to calculate the ratio of the frequency domain energy below the fundamental frequency of the turbocharger spectrum to the overall frequency domain energy. If the ratio is greater than or greater than or equal to the preset ratio, the vibration signal is invalid; otherwise, the vibration signal is valid.

5. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 4, characterized in that, Performing least-squares detrending on the vibration signal includes: Calculate b and a: In the formula, represents the vibration signal, represents the time, represents the number of data points of the vibration signal; Calculate the vibration signal after detrending: In the formula, represents the vibration signal after removing the detrended term.

6. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 1, characterized in that, Selecting some IMF components for signal reconstruction to obtain a reconstructed signal, including: If the maximum measurement frequency of the vibration sensor is , and the order containing in each IMF component is the th order, then remove the IMF components from the 1st order to the th order; If the fundamental frequency of the rotational speed of the turbocharger is , and the order containing in each IMF component is the th order, then select the highest order of the IMF component as the th order; Sum up the IMF components from the -th order to the -th order to obtain the reconstructed signal.

7. The method for optimizing state monitoring signals and diagnosing faults of a marine turbocharger according to claim 1, characterized in that, The method further includes: Save the vibration signals that are valid data as follows: Divide the signal into multiple sub-signals according to the specified length; Add labels to each sub-signal. The labels include speed, signal type, measurement point position and direction, and timestamp; the signal type includes vibration acceleration signal and shaft center orbit signal; Add the fault status corresponding to the sub-signal to the label.

8. The method for optimizing the state monitoring signals and diagnosing faults of a marine turbocharger according to claim 1, characterized in that, The fault status of the marine turbocharger includes normal status, slight rotor imbalance status, and severe rotor imbalance status; The time-domain and frequency-domain characteristic parameters of the vibration signal include at least one of maximum value, mean value, absolute mean value, root mean square amplitude, mean square value, effective value, variance, standard deviation, skewness, kurtosis, waveform factor, peak factor, pulse factor, margin index, kurtosis index, skewness index, amplitude mean value, center frequency, mean square frequency, frequency variance, root mean square frequency, amplitude variance, and frequency standard deviation; Perform cubic polynomial curve fitting on the markers corresponding to each fault status in the scatter plot. If the amplitude variation of the fitting curve within the entire speed range is less than the preset variation threshold, it is considered that the variation trend of the markers corresponding to different fault statuses in the scatter plot with respect to speed is flat; The neural network model is a one-dimensional convolutional neural network; Before training the neural network model, perform dimensionality reduction processing on the high-dimensional feature vector composed of the effective characteristic parameters of the vibration signals at the preferred measurement points through the t-distributed stochastic neighbor embedding dimensionality reduction algorithm.

9. A marine turbocharger, characterized in that, The marine turbocharger uses the marine turbocharger status monitoring signal optimization and fault diagnosis method described in any one of claims 1 to 8 to complete the fault diagnosis of the marine turbocharger.

Images