Compressor fault diagnosis method and system based on multi-sensor fusion

Abstract

The invention discloses a compressor fault diagnosis method and system based on multi-sensor fusion, and the system comprises a data collection module which collects the operation data of a compressor in real time, a data processing module which carries out the data processing, and a fault diagnosis module which carries out the fault diagnosis. The data processing module comprises a preprocessing unit, a feature extraction unit and a data fusion unit, and the preprocessing unit carries out filtering, denoising and normalization processing on the collected data; the feature extraction unit extracts feature values from the data processed by the preprocessing unit; the data fusion unit generates a comprehensive feature vector from the feature values extracted by the feature extraction unit according to a time axis; the fault diagnosis module comprises a fault diagnosis calculation model and a feedback unit, and the fault diagnosis calculation model calculates and outputs the fault type and position according to the comprehensive feature vector generated by the data fusion unit; and the feedback unit feeds back the diagnosis result.

Description

Technical Field

[0001] The present invention relates to the technical field of compressor fault diagnosis, and in particular to a compressor fault diagnosis method and system based on multi-sensor fusion. Background Art

[0002] Compressors are key equipment in many industrial fields, and their normal operation is crucial to the production process. However, due to long-term operation and the influence of various external factors, compressors are prone to various faults, such as mechanical wear, gas blockage, electrical failure, cooling failure, lubrication failure, etc. Traditional fault diagnosis methods mainly rely on data from a single sensor, such as a vibration sensor or a temperature sensor. This method can often only detect the appearance of equipment vibration and temperature trend changes under complex working conditions, and it is difficult to accurately identify the fault type and location, and it is impossible to determine the root cause of the fault. Therefore, it is particularly important to develop a fault diagnosis method that can integrate data from multiple sensors.

[0003] The patent with the authorization announcement number "CN115596654B" discloses a reciprocating compressor fault diagnosis method and system based on state parameter learning. In the technical solution of this patent, it includes building a BP neural network; extracting the input parameters and output parameters in the historical operating data of the reciprocating compressor under normal conditions as training samples, and pre-training the BP neural network; predicting and calculating the input parameters in the training samples through the pre-trained BP neural network, obtaining the benchmark deviation between the predicted value and the actual value, and setting the safety threshold range using the benchmark deviation; extracting the input parameters in the current operating data of the reciprocating compressor as the input parameters of the BP neural network, and obtaining the predicted value of the output parameter; calculating the relative deviation between the predicted value and the measured value of the output parameter, and comparing it with the safety threshold range. If it exceeds the safety threshold range, it is considered that the reciprocating compressor is in a faulty state, and the fault is located according to the change of the relative deviation, and a fault diagnosis result is given. In the technical solution disclosed in this patent document, only the sample data of the normal state is considered when establishing the neural network model, and only the state parameters exceeding the safety threshold range are judged for faults. Its compressor fault detection method has the following shortcomings: ① Single state parameter data cannot fully reflect the operating status of the compressor, especially when individual state parameters change in the early stage of the fault, there is a lack of fault analysis capability; ② It only feeds back abnormal changes in individual state parameters (such as vibration, temperature, etc.) during the fault occurrence stage, and the corresponding identification of fault type, location and root cause is not accurate enough; ③ Fault analysis relies on preset conditions, lacks judgment of time series in actual situations, and has poor adaptability to complex working conditions.

[0004] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a compressor fault diagnosis method and system based on multi-sensor fusion, which can use multi-source heterogeneous data processing technology to fuse multiple sensor data of the compressor, form a comprehensive picture of various parameters of the equipment in the time series, analyze the correlation of parameter changes, and combine relevant machine learning algorithms to form the root cause of equipment failure, thereby guiding equipment fault troubleshooting and operation and maintenance work.

[0006] In order to achieve the above-mentioned purpose, in the technical solution of the present invention, on the one hand, a compressor fault diagnosis system based on multi-sensor fusion is provided, including a data acquisition module for real-time acquisition of compressor operation data, a data processing module for data processing, and a fault diagnosis module for fault diagnosis.

[0007] Furthermore, the data processing module includes a preprocessing unit, a feature extraction unit and a data fusion unit, wherein the preprocessing unit performs filtering, denoising and normalization processing on the collected data; the feature extraction unit extracts feature values ​​from the data processed by the preprocessing unit; the data fusion unit generates a comprehensive feature vector according to the time axis for each feature value extracted by the feature extraction unit, that is, each group of comprehensive feature vectors includes each feature value at the same time point; Furthermore, the fault diagnosis module includes a fault diagnosis calculation model and a feedback unit. The fault diagnosis calculation model calculates and outputs the fault type and location according to the comprehensive feature vector generated by the data fusion unit; and the feedback unit is used to feed back the diagnosis result.

[0008] Furthermore, the compressor operation data collected in real time by the data acquisition module includes vibration information monitored by a vibration sensor, temperature information monitored by a temperature sensor, pressure and flow information monitored by a pressure sensor and a flow sensor, current information monitored by a current sensor, and power information monitored by a power sensor.

[0009] Furthermore, in the technical solution of the present invention, the preprocessing unit performs filtering, denoising and normalization processing on the collected data, specifically including: Vibration information preprocessing: convert the time domain vibration signal into spectrum information through Fourier transform, and then extract 1 times, 2 times, 3 times, 4 times, 5 times, and 6 times frequency signals as feature values; Temperature information preprocessing: polynomial fitting of temperature signal, extracting temperature speed change signal and original temperature signal as characteristic values; Preprocessing of pressure and flow information: performing polynomial fitting on the pressure and flow information respectively, extracting pressure and flow rate change signals as well as original pressure and flow signals as characteristic values; Current information preprocessing, using current harmonics and current values ​​as characteristic values; Power information is preprocessed, and power is directly used as the eigenvalue.

[0010] Furthermore, in the technical solution of the present invention, the fault diagnosis calculation model refers to a calculation model for fault analysis established by training the collected physical quantity data using normal data and abnormal data of the equipment operation, and the specific steps include: S1. Analyze the correlation between equipment subsystems and components from previous failure and malfunction cases, and establish a fault database through historical data and fault records, including fault type, fault location, and fault time, to store fault cases and related data; S2. Based on the fault database, determine the key physical quantities that need to be monitored, including vibration, temperature, pressure and flow, current and power. Select the characteristic parameters that contribute most to fault diagnosis through correlation analysis and failure analysis methods. According to the change rules of these characteristic parameters, establish a temporal logic model to describe the relationship and change trend between the characteristic parameters: Based on the support vector machine machine learning algorithm, the equipment state vector, that is, the characteristic parameters, are classified into normal operation data and fault data. The specific objective function is defined as:

[0011] Where: is a predetermined constant. is the data representation of the linear hyperplane, is the variable to be optimized, and the constraints are: ; The decision tree machine learning algorithm is used as a classifier for compressor fault diagnosis to identify and locate faults. The decision tree constructed for equipment fault classification uses information gain to define the splitting criteria, where the empirical entropy of the data set is calculated as follows: ; S3. Collect a large amount of equipment operation data, including normal operation data and fault data, and label these data, distinguish normal data from fault data, label the fault type and location, and use the labeled data to train the established model.

[0012] Another aspect of the present invention further provides a compressor fault diagnosis method based on multi-sensor fusion, which adopts a compressor fault diagnosis system based on multi-sensor fusion as described above, and specifically includes the following contents: Data acquisition: The data acquisition module collects compressor operation data in real time, including vibration information monitored by vibration sensors, temperature information monitored by temperature sensors, pressure and flow information monitored by pressure sensors and flow sensors, current information monitored by current sensors, and power information monitored by power sensors; Data processing: The collected compressor operation data is preprocessed through the data processing module to obtain characteristic values, and a comprehensive characteristic vector is generated based on the obtained characteristic values; Fault diagnosis: Calculate and output the fault type and location based on the comprehensive feature vector, and feedback the diagnosis results.

[0013] Effective gain: In summary, the present invention proposes a compressor fault diagnosis method and system based on multi-sensor fusion. The present invention adopts a training method of multiple fault data sets and normal state data sets, and can directly perform fault classification model training through a decision tree; the present invention analyzes and models equipment faults, introduces time series to establish the correlation relationship between multiple sensor data states associated with the fault in the fault analysis model, and effectively solves the problem of locating the root cause of the fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of a compressor fault diagnosis system based on multi-sensor fusion according to the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0016] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0017] In the description of the present invention, it is necessary to understand that the orientations or positional relationships indicated by the terms "upper", "lower", "top", "bottom", "inside", "outside", etc. are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0018] In addition, in the description of the present invention, “several” means two or more than two, unless otherwise clearly and specifically defined.

[0019] In order to solve the problems raised in the above background technology, an embodiment of the present invention proposes a compressor fault diagnosis system based on multi-sensor fusion, which includes a data acquisition module for real-time collection of compressor operation data, a data processing module for data processing, and a fault diagnosis module for fault diagnosis.

[0020] Specifically, the data processing module includes a preprocessing unit, a feature extraction unit and a data fusion unit. The preprocessing unit filters, denoises and normalizes the collected data; the feature extraction unit extracts feature values ​​from the data processed by the preprocessing unit; the data fusion unit generates a comprehensive feature vector for each feature value extracted by the feature extraction unit according to the time axis, that is, each group of comprehensive feature vectors includes each feature value at the same time point.

[0021] Specifically, the fault diagnosis module includes a fault diagnosis calculation model and a feedback unit. The fault diagnosis calculation model calculates and outputs the fault type and location according to the comprehensive feature vector generated by the data fusion unit; the feedback unit feeds back the diagnosis result.

[0022] Specifically, the compressor operation data collected in real time by the data acquisition module includes vibration information monitored by vibration sensors, temperature information monitored by temperature sensors, pressure and flow information monitored by pressure sensors and flow sensors, current information monitored by current sensors, and power information monitored by power sensors.

[0023] Specifically, the preprocessing unit performs filtering, denoising and normalization processing on the collected data, which specifically includes: Vibration information preprocessing: convert the time domain vibration signal into spectrum information through Fourier transform, and then extract 1 times, 2 times, 3 times, 4 times, 5 times, and 6 times frequency signals as feature values; Temperature information preprocessing: polynomial fitting of temperature signal, extracting temperature speed change signal and original temperature signal as characteristic values; Preprocessing of pressure and flow information: performing polynomial fitting on the pressure and flow information respectively, extracting pressure and flow rate change signals as well as original pressure and flow signals as characteristic values; Current information preprocessing, using current harmonics and current values ​​as characteristic values; Power information is preprocessed, and power is directly used as the eigenvalue.

[0024] Specifically, the fault diagnosis calculation model refers to a fault analysis calculation model established by training the collected physical quantity data using the normal data and abnormal data of the equipment operation. The specific steps include: S1. Analyze the correlation between equipment subsystems and components from previous failure and malfunction cases, and establish a fault database through historical data and fault records, including fault type, fault location, and fault time, to store fault cases and related data; S2. Based on the fault database, determine the key physical quantities that need to be monitored, including vibration, temperature, pressure and flow, current and power. Select the characteristic parameters that contribute most to fault diagnosis through correlation analysis and failure analysis methods. According to the change rules of these characteristic parameters, establish a temporal logic model to describe the relationship and change trend between the characteristic parameters: Based on the support vector machine machine learning algorithm, the equipment state vector, that is, the characteristic parameters, are classified into normal operation data and fault data. The specific objective function is defined as:

[0025] Where: is a predetermined constant. is the data representation of the linear hyperplane, is the variable to be optimized, and the constraints are: ; The decision tree machine learning algorithm is used as a classifier for compressor fault diagnosis to identify and locate faults. The decision tree constructed for equipment fault classification uses information gain to define the splitting criteria, where the empirical entropy of the data set is calculated as follows: ; S3. Collect a large amount of equipment operation data, including normal operation data and fault data, and label these data, distinguish normal data from fault data, label the fault type and location, and use the labeled data to train the established model.

[0026] On the other hand, this embodiment further provides a compressor fault diagnosis method based on multi-sensor fusion, which adopts a compressor fault diagnosis system based on multi-sensor fusion as described above, and specifically includes the following contents: Data acquisition: The data acquisition module collects compressor operation data in real time, including vibration information monitored by vibration sensors, temperature information monitored by temperature sensors, pressure and flow information monitored by pressure sensors and flow sensors, current information monitored by current sensors, and power information monitored by power sensors; Data processing: The collected compressor operation data is preprocessed through the data processing module to obtain characteristic values, and a comprehensive characteristic vector is generated based on the obtained characteristic values; Fault diagnosis: Calculate and output the fault type and location based on the comprehensive feature vector, and feedback the diagnosis results.

[0027] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A compressor fault diagnosis system based on multi-sensor fusion, comprising a data acquisition module for real-time collection of compressor operation data, a data processing module for data processing, and a fault diagnosis module for fault diagnosis, characterized in that: The data processing module comprises: The pre-processing unit performs filtering, denoising and normalization on the collected data; A feature extraction unit, which extracts feature values ​​from each item of data processed by the preprocessing unit; A data fusion unit generates a comprehensive feature vector according to a time axis for each feature value extracted by the feature extraction unit, that is, each group of comprehensive feature vectors includes each feature value at the same time point; The fault diagnosis module comprises: A fault diagnosis calculation model calculates and outputs the fault type and location based on the comprehensive feature vector generated by the data fusion unit; Feedback unit, feeding back diagnosis results.

2. A compressor fault diagnosis system based on multi-sensor fusion according to claim 1, characterized in that: The compressor operation data collected in real time by the data acquisition module includes vibration information monitored by a vibration sensor, temperature information monitored by a temperature sensor, pressure and flow information monitored by a pressure sensor and a flow sensor, current information monitored by a current sensor, and power information monitored by a power sensor.

3. A compressor fault diagnosis system based on multi-sensor fusion according to claim 2, characterized in that: The preprocessing unit performs filtering, denoising and normalization processing on the collected data, specifically including: Vibration information preprocessing: convert the time domain vibration signal into spectrum information through Fourier transform, and then extract 1 times, 2 times, 3 times, 4 times, 5 times, and 6 times frequency signals as feature values; Temperature information preprocessing: polynomial fitting of temperature signal, extracting temperature speed change signal and original temperature signal as characteristic values; Preprocessing of pressure and flow information: performing polynomial fitting on the pressure and flow information respectively, extracting pressure and flow rate change signals as well as original pressure and flow signals as characteristic values; Current information preprocessing, using current harmonics and current values ​​as characteristic values; Power information is preprocessed, and power is directly used as the eigenvalue.

4. A compressor fault diagnosis system based on multi-sensor fusion according to claim 1, characterized in that: The fault diagnosis calculation model refers to a fault analysis calculation model established by training the collected physical quantity data using normal data and abnormal data of the equipment operation. The specific steps include: S1. Analyze the correlation between equipment subsystems and components from previous failure and malfunction cases, and establish a fault database through historical data and fault records, including fault type, fault location, and fault time, to store fault cases and related data; S2. Based on the fault database, determine the key physical quantities that need to be monitored, including vibration, temperature, pressure and flow, current and power. Select the characteristic parameters that contribute most to fault diagnosis through correlation analysis and failure analysis methods. According to the change rules of these characteristic parameters, establish a temporal logic model to describe the relationship and change trend between the characteristic parameters: Based on the support vector machine machine learning algorithm, the equipment state vector, that is, the characteristic parameters, are classified into normal operation data and fault data. The specific objective function is defined as: ； Where: is a predetermined constant. is the data representation of the linear hyperplane, is the variable to be optimized, and the constraints are: ； The decision tree machine learning algorithm is used as a classifier for compressor fault diagnosis to identify and locate faults. The decision tree constructed for equipment fault classification uses information gain to define the splitting criteria, where the empirical entropy of the data set is calculated as follows: ； S3. Collect a large amount of equipment operation data, including normal operation data and fault data, and label these data, distinguish normal data from fault data, label the fault type and location, and use the labeled data to train the established model.

5. A compressor fault diagnosis method based on multi-sensor fusion, using a compressor fault diagnosis system based on multi-sensor fusion as claimed in any one of claims 1 to 4, specifically comprising the following contents: Data acquisition: The data acquisition module collects compressor operation data in real time, including vibration information monitored by vibration sensors, temperature information monitored by temperature sensors, pressure and flow information monitored by pressure sensors and flow sensors, current information monitored by current sensors, and power information monitored by power sensors; Data processing: The collected compressor operation data is preprocessed through the data processing module to obtain characteristic values, and a comprehensive characteristic vector is generated based on the obtained characteristic values; Fault diagnosis: Calculate and output the fault type and location based on the comprehensive feature vector, and feedback the diagnosis results.

Images