Dynamic equipment intelligent monitoring and maintenance system and method based on vibration signal fusion multi-source information

Abstract

The invention discloses a system and a method for intelligent monitoring and maintenance of moving equipment based on fusion of vibration signals and multi-source information, and relates to the technical field of multi-dimensional monitoring of the moving equipment, and the method comprises the following steps: S1, through a non-intrusive signal collection scheme, collecting parameters of normal moving equipment and fault moving equipment in the operation process, comprising vibration signals, external magnetic field signals, temperature signals and sound signals; key features are extracted from the collected vibration, external magnetic field, sound and temperature signals of the current fault moving equipment, feature indexes with the highest association degree with different fault states of the moving equipment are screened through an algorithm, and feature vectors are constructed; and based on the feature vectors of the moving equipment in different fault states, constructing a moving equipment fault judgment model, and based on the collected operating parameter features of the current moving equipment. By collecting and analyzing vibration, sound, temperature and external magnetic field signals of the dynamic equipment at the same time, the accuracy of equipment fault diagnosis is effectively improved, and the defect of a single signal source is overcome.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-dimensional monitoring of moving equipment, and in particular to a system and method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information. Background Art

[0002] Dynamic equipment is widely used in industry, but the complex working conditions and changing external environment of the equipment often lead to various forms of equipment failure, such as cavitation, impeller damage, mechanical seal failure, etc. These failures may affect the normal operation and efficiency of the equipment, and even cause safety hazards; Traditional monitoring methods are mostly based on a single vibration signal or pressure pulsation signal. However, due to the variety of fault types of moving equipment and often accompanied by different fault modes, rotating machinery faults are mainly reflected in abnormal vibration and electromagnetic field changes. Traditional single signal source monitoring methods are difficult to fully reflect the operating status and fault mode of the equipment, and have problems with low practicality and functionality. Therefore, there is an urgent need for a monitoring system that can integrate multi-source information to improve the comprehensiveness and accuracy of monitoring during the operation of moving equipment. Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0003] In view of the problems in the related art, the present invention proposes a system and method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information to overcome the above-mentioned technical problems existing in the existing related art.

[0004] To this end, the specific technical solution adopted by the present invention is as follows: A method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information, the method comprising the following steps: S1. Collect parameters of normal and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals and sound signals, through non-intrusive signal acquisition solutions; S2. Extract key features from the collected vibration, external magnetic field, sound and temperature signals of the currently faulty moving equipment, select the feature indicators with the highest correlation with different fault states of the moving equipment through algorithms, and construct feature vectors; S3. Based on the characteristic vectors of the dynamic equipment under different fault states, a dynamic equipment fault judgment model is constructed. Based on the collected operating parameter characteristics of the current dynamic equipment and combined with the dynamic equipment fault judgment model, multi-dimensional judgments are made on potential faults that have not occurred during the operation of the dynamic equipment, and faults that have occurred are diagnosed.

[0005] As a preferred embodiment, the S1 includes the following sub-steps: S11. Place high-sensitivity vibration sensors at key locations of the moving equipment, including bearings, rotors, mechanical seals, and impellers, to collect vibration characteristics of the moving equipment in different states, including vibration frequency, amplitude, vibration phase, and spectrum characteristics; S12. Based on the principle of electromagnetic induction, the induction coil installed outside the motor of the moving device is used to collect the external magnetic field signals of the moving device under different states, including characteristic frequency and amplitude; S13, monitoring the temperature distribution outside different moving devices through non-contact infrared temperature sensors, and combining the data collected by internal temperature sensors to obtain the temperature changes of the moving devices under different states; S14. A directional microphone is arranged outside the moving device in different states to collect the operating sound signal of the moving device in a contactless manner.

[0006] As a preferred embodiment, S2 comprises the following steps: S21, for the vibration signal, external magnetic field signal, temperature signal and sound signal of the moving device under the normal state, respectively extract the key features of the vibration signal, the external magnetic field signal, the temperature signal and the sound signal as the reference state parameter features of the moving device; S22. For the vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment under different fault states, the key features of the vibration signals, external magnetic field signals, temperature signals and sound signals are extracted respectively. Based on the Mahalanobis distance and the characteristics of the benchmark state parameters of the moving equipment, the correlation characteristic indicators under different fault states are screened to construct different fault feature vectors.

[0007] As a preferred embodiment, the S21 comprises the following steps: S211. Extracting time domain features and frequency domain features from the vibration signal of the moving equipment under the normal state, wherein the time domain features include mean, standard deviation, peak value, and kurtosis, and the frequency domain features include main frequency, spectrum, and frequency center; S212, extracting characteristic parameters of the external magnetic field signal of the moving device under the normal state, including peak value, maximum intensity change rate, main frequency change, harmonic characteristics, offset, and fluctuation amplitude; S213, extracting trend features and dynamic features from the collected temperature signal of the moving device in the normal state, wherein the trend features include the temperature mean, the temperature change rate, the maximum temperature, and the minimum temperature, and the dynamic features include the temperature fluctuation amplitude, the temperature rise rate, and the temperature gradient; S214, extracting characteristic parameters of the sound signal collected in a normal state, including short-time energy, signal zero-crossing rate, spectrum center frequency, bandwidth, and Mel frequency cepstrum coefficient; S215, extraction of normal moving equipment The vibration signal characteristics, external magnetic field signal characteristics, temperature signal characteristics and sound signal characteristic parameter values ​​are processed by averaging, and the characteristic values ​​after averaging are used as the characteristic vector of the reference state parameters of the dynamic equipment: ; in, Represents the characteristic vector of the reference state parameters of the dynamic equipment, They represent the key features after mean processing of vibration signal, external magnetic field signal, temperature signal and sound signal respectively.

[0008] As a preferred embodiment, the S22 comprises the following steps: S221, analyze the vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment under different fault conditions Group collection and grouping based on fault type, through Establish different fault type files respectively, and collect the data of the moving equipment under the same fault type. The group parameters are processed by averaging, and the characteristic parameter values ​​after averaging are recorded in the corresponding fault type file to construct the fault state feature vector: ; in, Representative The fault state feature vector under the fault type file, Respectively represent The key features of vibration signal, external magnetic field signal, temperature signal and sound signal after mean value processing under each fault type file; S222. Based on the Mahalanobis distance algorithm and the reference state parameter characteristics of the dynamic equipment, the Mahalanobis distances between the dynamic equipment acquisition parameters and the reference state parameter characteristics under different fault type files are calculated respectively. The algorithm formula is: ; in, represents the covariance matrix of the baseline state parameter eigenvector, Representative The Mahalanobis distance between each feature and the reference state parameter feature under each fault type file; S223, based on Mahalanobis distance threshold , select characteristic indicators with high correlation with each fault state, and the specific steps are as follows: Statistics all > These characteristics are the high-correlation characteristic indicators under different fault states. Based on the selected correlation characteristic indicators, the correlation characteristic vectors under different fault states are constructed: ; in, That is the The associated feature vector under each fault type file, Respectively represent The vibration, magnetic field, temperature and sound characteristics that are highly correlated with the fault status under each fault type file.

[0009] As a preferred embodiment, S3 includes the following sub-steps: S31. Based on the characteristic vectors of the dynamic equipment under different fault conditions, a dynamic equipment fault determination model is constructed. The specific steps are as follows: The feature vectors of normal state and different fault states are labeled, where the label of normal state is 0 and the label of fault state is 1 to N. Based on the associated feature vector, the feature parameters related to the fault are selected to construct the input feature matrix : ; The equipment fault judgment model is constructed based on the neural network. The output of the input layer is the input feature vector. The output of each hidden layer is transformed nonlinearly through the activation function. The number of neurons in the output layer is equal to the number of categories, which is N+1 categories, including normal state and N fault states. The activation function performs probability distribution conversion and converts the labeled feature matrix The neural model is trained with the label input under the fault state to obtain the dynamic equipment fault judgment model; S32. Collect vibration signals, external magnetic field signals, sound signals, and temperature signals during the current operation of the dynamic equipment, extract characteristic parameters, construct the current characteristic vector, and based on the obtained dynamic equipment fault judgment model, make multi-dimensional predictions on potential faults that have not occurred during the operation of the dynamic equipment, and diagnose faults that have occurred.

[0010] As a preferred embodiment, the S32 comprises the following steps: S321. Collect vibration signals, external magnetic field signals, sound signals, and temperature signals during the operation of the current moving equipment, extract characteristic parameters, and construct a real-time characteristic vector: ; in, They respectively represent the key features of the vibration signal, external magnetic field signal, temperature signal and sound signal collected during the operation of the current moving equipment; S322, input the real-time feature vector into the dynamic equipment fault determination model to detect whether there is a potential fault in the dynamic equipment: The output layer usually uses The activation function converts the real-time feature vector input and output into the probability distribution of each category, and selects the category with the highest probability as the final prediction result: When the output result is 0, the current device is in normal state; When the prediction result is 1 to N, it means that there is a potential failure in the dynamic equipment; S323. For the dynamic equipment with potential faults, the abnormal characteristic parameters are extracted to obtain the abnormal characteristic vector: ; in, Respectively represent the abnormal characteristics of the vibration signal, external magnetic field signal, temperature signal and sound signal of the current moving equipment, and calculate The cosine similarity of the associated feature vectors under each fault type file is sorted in descending order, and the fault type file with the first ranking is selected as the diagnostic output fault of the current abnormal fault of the dynamic equipment.

[0011] The intelligent monitoring and maintenance system for moving equipment based on vibration signal fusion of multi-source information includes data acquisition module, feature extraction module, model building module and output judgment module: The data acquisition module collects parameters of normal moving equipment and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals and sound signals, through a non-intrusive signal acquisition solution; The feature extraction module extracts key features from the vibration, external magnetic field, sound and temperature signals of the current faulty moving equipment collected by the data acquisition module, selects the feature indicators with the highest correlation with different fault states of the moving equipment through an algorithm, and constructs a feature vector; The model building module builds a dynamic equipment fault determination model through a neural network based on the feature vectors under different fault states of the dynamic equipment, and outputs the model to the output determination module; The output determination module, in combination with the operating parameter characteristics of the current dynamic equipment collected by the data acquisition module, uses the dynamic equipment fault determination model to perform multi-dimensional determination on potential faults that have not occurred during the operation of the dynamic equipment, and diagnose faults that have occurred.

[0012] The beneficial effects of the present invention are: 1. The present invention collects vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment in the fault state, and analyzes, compares and calculates the signal characteristic parameters in different fault states in combination with the signal characteristic parameters in the normal state, thereby obtaining the characteristic parameter vectors in different fault states. By collecting various types of signals, the operating state of the equipment can be more comprehensively understood, thereby improving the accuracy of the moving equipment fault diagnosis; 2. The present invention constructs a feature vector by comparing the signal characteristic parameters of the fault state and the normal state, which helps to highlight the fault characteristics to improve the diagnosis accuracy. At the same time, by using the characteristics of vibration, external magnetic field, temperature and sound signals, different fault types can be more accurately located to assist in the fault diagnosis of mobile equipment; 3. The present invention can analyze the various signal characteristics collected during the operation of the dynamic equipment in real time by building a dynamic equipment fault judgment model, timely judge abnormal signals to predict faults, which is helpful to carry out maintenance work in advance, and judge the most likely fault type of the current dynamic equipment by fault diagnosis of abnormal signals, thereby assisting maintenance personnel to repair the dynamic equipment, and enhancing practicality and functionality; 4. The present invention can comprehensively improve the accuracy of dynamic equipment fault diagnosis and the monitoring efficiency under complex working conditions through multi-source information fusion analysis of vibration, external magnetic field, sound and temperature signals, and realize automatic dynamic equipment fault identification and classification by using the constructed judgment model. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 is a flow chart of a method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to an embodiment of the present invention; Figure 2 It is a block diagram of a system for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to an embodiment of the present invention. DETAILED DESCRIPTION

[0015] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0016] According to an embodiment of the present invention, a system and method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information are provided.

[0017] The present invention is further described with reference to the accompanying drawings and specific embodiments: Example 1: Figure 1As shown, according to the method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to an embodiment of the present invention, the method comprises the following steps: S1. Collect parameters of normal and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals and sound signals, through non-intrusive signal acquisition solutions; S11. Place high-sensitivity vibration sensors at key locations of the moving equipment, including bearings, rotors, mechanical seals, and impellers, to collect vibration characteristics of the moving equipment in different states, including vibration frequency, amplitude, vibration phase, and spectrum characteristics; It should be noted that when different abnormalities occur in the moving equipment, the rotor imbalance failure manifests as radial vibration, the frequency of which is mainly concentrated on the fundamental frequency, the mechanical seal damage failure usually causes abnormal high-frequency vibration, and the impeller damage failure causes specific vibration spectrum characteristic changes.

[0018] S12. Based on the principle of electromagnetic induction, the induction coil installed outside the motor of the moving device is used to collect the external magnetic field signals of the moving device under different states, including characteristic frequency and amplitude; It should be noted that the induction coil reflects the load change by capturing the change of the stator magnetic field of the moving equipment motor. When the working conditions or load of the moving equipment change, the characteristic frequency and amplitude of the external magnetic field signal will also change; S13, monitoring the temperature distribution outside different moving devices through non-contact infrared temperature sensors, and combining the data collected by internal temperature sensors to obtain the temperature changes of the moving devices under different states; It should be noted that when the load of the moving equipment increases or a key component fails, the temperature signal will show an abnormal upward trend. By collecting the temperature change of the current moving equipment, the moving equipment failure can be determined; S14. A directional microphone is arranged outside the moving device in different states to collect the operating sound signal of the moving device in a contactless manner.

[0019] It should be noted that different fault types will cause changes in sound characteristics. For example, impeller damage will cause pulse-type sound characteristics, and mechanical friction faults will increase high-frequency noise. By collecting the sound signals of the moving equipment, the sound signal characteristics can be obtained through further feature extraction; S2. Extract key features from the collected vibration, external magnetic field, sound and temperature signals of the currently faulty moving equipment, select the feature indicators with the highest correlation with different fault states of the moving equipment through algorithms, and construct feature vectors; S21, for the vibration signal, external magnetic field signal, temperature signal and sound signal of the moving device under the normal state, respectively extract the key features of the vibration signal, the external magnetic field signal, the temperature signal and the sound signal as the reference state parameter features of the moving device; S211, extracting time domain features and frequency domain features from the vibration signal of the moving equipment under the normal state, wherein the time domain features include mean, standard deviation, peak value, and kurtosis, and the frequency domain features include main frequency, spectrum, and frequency center; It should be noted that in the vibration signal, the mean is the average value of the vibration signal to reflect the overall level of vibration amplitude; the peak is the maximum amplitude of the vibration signal to indicate the intensity of abnormal vibration; the kurtosis is a measure of the sharpness of the vibration signal, which will increase significantly when a fault occurs in the moving equipment; the dominant frequency is the main vibration frequency of the signal, which is used to identify specific fault types, such as rotor imbalance; the spectrum is the energy distribution in the frequency domain, which is used to analyze fault signals in a specific frequency band.

[0020] S212, extracting characteristic parameters of the external magnetic field signal of the moving device under the normal state, including peak value, maximum intensity change rate, main frequency change, harmonic characteristics, offset, and fluctuation amplitude; It should be noted that in the external magnetic field signal, the peak value reflects the degree of change in the motor load, the maximum intensity change rate reflects the speed of intensity change when the load changes rapidly, the main frequency change can indicate the characteristic frequency fluctuation of the magnetic field caused by the load change, the harmonic characteristics are the high-order harmonic components generated when the load fluctuates or the torque fluctuates, the offset is the average value offset of the magnetic field strength, indicating the overall load trend, and the fluctuation amplitude is the range of change of the magnetic field strength in a short period of time.

[0021] S213, extracting trend features and dynamic features from the collected temperature signals of the moving equipment in the normal state, wherein the trend features include the temperature mean, the temperature change rate, the maximum temperature, and the minimum temperature, and the dynamic features include the temperature fluctuation amplitude, the temperature rise rate, and the temperature gradient; It should be noted that in the temperature signal, the temperature mean and change rate are used to identify the thermal equilibrium state and heating or cooling trend. The maximum and minimum temperature values ​​reflect the thermal load and cooling capacity of the equipment. The temperature fluctuation amplitude indicates the stability of the temperature signal, which may cause increased fluctuations when the moving equipment fails. The temperature rise rate represents the speed of temperature rise and is used to diagnose equipment overload or insufficient heat dissipation. The temperature gradient is used to identify local overheating when the moving equipment fails.

[0022] S214, extracting characteristic parameters of the sound signal collected in a normal state, including short-time energy, signal zero-crossing rate, spectrum center frequency, bandwidth, and Mel frequency cepstrum coefficient; It should be noted that in sound signals, short-time energy is the local energy of the sound signal, which is used to reflect the intensity of noise or impact signals. The signal zero-crossing rate can reflect the high-frequency components and noise level of the sound signal. The spectrum center frequency is the main frequency of the sound signal, which is used to identify the source of the dynamic equipment failure. The bandwidth is the width of the sound signal energy distribution, indicating the complexity of the signal. The Mel frequency cepstrum coefficient is used to describe the frequency distribution of the sound signal. S215, extraction of normal moving equipment The vibration signal characteristics, external magnetic field signal characteristics, temperature signal characteristics and sound signal characteristic parameter values ​​are processed by averaging, and the characteristic values ​​after averaging are used as the characteristic vector of the reference state parameters of the dynamic equipment: ; in, Represents the characteristic vector of the reference state parameters of the dynamic equipment, They represent the key features after mean processing of vibration signal, external magnetic field signal, temperature signal and sound signal respectively.

[0023] It should be noted that by Collecting group parameters and calculating the mean can reduce the error of collected data. The value is generally set to 5, and can also be adjusted according to actual conditions; S22, extracting key features of vibration signals, external magnetic field signals, temperature signals and sound signals of moving equipment under different fault states, respectively, and screening correlation characteristic indicators under different fault states based on Mahalanobis distance combined with the reference state parameter characteristics of the moving equipment, and constructing different fault feature vectors; S221, analyze the vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment under different fault conditions Group collection and grouping based on fault type, through Establish different fault type files respectively, and collect the data of the moving equipment under the same fault type. The group parameters are processed by averaging, and the characteristic parameter values ​​after averaging are recorded in the corresponding fault type file to construct the fault state feature vector: ; in, Representative The fault state feature vector under the fault type file, Respectively represent The key features of vibration signal, external magnetic field signal, temperature signal and sound signal after mean value processing under each fault type file; It should be noted that Collecting group parameters and calculating the mean can reduce the error of collected data. The value of is consistent with that in S215.

[0024] S222. Based on the Mahalanobis distance algorithm and the reference state parameter characteristics of the dynamic equipment, the Mahalanobis distances between the dynamic equipment acquisition parameters and the reference state parameter characteristics under different fault type files are calculated respectively. The algorithm formula is: ; in, represents the covariance matrix of the baseline state parameter eigenvector, Representative The Mahalanobis distance between each feature and the reference state parameter feature under each fault type file; S223, based on Mahalanobis distance threshold , select characteristic indicators with high correlation with each fault state, and the specific steps are as follows: Statistics all > These characteristics are the high-correlation characteristic indicators under different fault states. Based on the selected correlation characteristic indicators, the correlation characteristic vectors under different fault states are constructed: ; in, That is the The associated feature vector under each fault type file, Respectively represent The vibration, magnetic field, temperature and sound characteristics that are highly correlated with the fault status under each fault type file.

[0025] It should be noted that based on the Mahalanobis distance combined with the characteristics of the benchmark state parameters of the dynamic equipment, the correlation characteristic indicators under different fault states are screened out, and different fault feature vectors are constructed, so as to realize the effective identification and diagnosis of the fault state of the dynamic equipment. According to the Mahalanobis distance, the distribution of samples in the feature space can be judged, and the characteristic parameters that contribute most to the classification can be selected. The Mahalanobis distance threshold It is necessary to calculate the covariance matrix of the baseline state feature vector to calculate the Mahalanobis distance under normal conditions and draw its distribution. According to the chi-square distribution table, select the appropriate confidence level to obtain the corresponding Mahalanobis distance threshold. Embodiment 2: S3, based on the feature vectors of the dynamic equipment under different fault states, a dynamic equipment fault determination model is constructed, based on the collected operating parameter characteristics of the current dynamic equipment, combined with the dynamic equipment fault determination model, a multi-dimensional determination is made on potential faults that have not occurred during the operation of the dynamic equipment, and a diagnosis is made on faults that have occurred; S31. Based on the characteristic vectors of the dynamic equipment under different fault conditions, a dynamic equipment fault determination model is constructed. The specific steps are as follows: The feature vectors of normal state and different fault states are labeled, where the label of normal state is 0 and the label of fault state is 1 to N. Based on the associated feature vector, the feature parameters related to the fault are selected to construct the input feature matrix : ; The equipment fault judgment model is constructed based on the neural network. The output of the input layer is the input feature vector. The output of each hidden layer is transformed nonlinearly through the activation function. The number of neurons in the output layer is equal to the number of categories, which is N+1 categories, including normal state and N fault states. The activation function performs probability distribution conversion and converts the labeled feature matrix The neural model is trained with the label input under the fault state to obtain the dynamic equipment fault judgment model; It should be noted that in the process of building a neural network model, forward propagation, loss function, and back propagation are required to ultimately train the model. The specific steps of forward propagation are as follows: ; ; in, is the weight matrix from the input layer to the hidden layer, is the bias vector, is the ReLU activation function.

[0026] S32, collecting vibration signals, external magnetic field signals, sound signals, and temperature signals during the operation of the current dynamic equipment, extracting characteristic parameters, constructing a current characteristic vector, and based on the obtained dynamic equipment fault judgment model, making multi-dimensional predictions on potential faults that have not occurred during the operation of the dynamic equipment, and diagnosing faults that have occurred; S321. Collect vibration signals, external magnetic field signals, sound signals, and temperature signals during the operation of the current moving equipment, extract characteristic parameters, and construct a real-time characteristic vector: ; in, They respectively represent the key features of the vibration signal, external magnetic field signal, temperature signal and sound signal collected during the operation of the current moving equipment; S322, input the real-time feature vector into the dynamic equipment fault determination model to detect whether there is a potential fault in the dynamic equipment: The output layer usually uses The activation function converts the real-time feature vector input and output into the probability distribution of each category, and selects the category with the highest probability as the final prediction result: When the output result is 0, the current device is in normal state; When the prediction result is 1 to N, it means that there is a potential failure in the dynamic equipment; S323. For the dynamic equipment with potential faults, the abnormal characteristic parameters are extracted to obtain the abnormal characteristic vector: ; in, Respectively represent the abnormal characteristics of the vibration signal, external magnetic field signal, temperature signal and sound signal of the current moving equipment, and calculate The cosine similarity of the associated feature vectors under each fault type file is sorted in descending order, and the fault type file with the first ranking is selected as the diagnostic output fault of the current abnormal fault of the dynamic equipment.

[0027] It should be noted that the larger the cosine similarity, the more similar the two feature vectors are, that is, the more similar the current abnormal characteristics of the dynamic equipment are to the associated feature vectors in the fault file, the higher the similarity between the current abnormality and the fault. Outputting the fault through output diagnosis can assist operators to perform timely maintenance and early repair of the abnormality of the dynamic equipment.

[0028] Example 3: Figure 2 As shown in the figure, the intelligent monitoring and maintenance system for dynamic equipment based on vibration signal fusion of multi-source information includes data acquisition module, feature extraction module, model building module and output judgment module: The data acquisition module collects parameters of normal and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals, and sound signals, through a non-intrusive signal acquisition solution; The feature extraction module extracts key features from the vibration, external magnetic field, sound and temperature signals of the current faulty moving equipment collected by the data acquisition module, selects the feature indicators with the highest correlation with different fault states of the moving equipment through algorithms, and constructs feature vectors; The model building module builds a dynamic equipment fault judgment model through a neural network based on the feature vectors under different fault states of the dynamic equipment, and outputs the model to the output judgment module; The output judgment module, combined with the operating parameter characteristics of the current dynamic equipment collected by the data acquisition module, uses the dynamic equipment fault judgment model to make multi-dimensional judgments on potential faults that have not occurred in the dynamic equipment operation process, and diagnose faults that have already occurred.

[0029] In summary, the present invention collects vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment in the fault state, and analyzes, compares and calculates the signal characteristic parameters in different fault states in combination with the signal characteristic parameters in the normal state, thereby obtaining the characteristic parameter vectors in different fault states. By collecting various types of signals, the operating state of the equipment can be more comprehensively understood, thereby improving the accuracy of the moving equipment fault diagnosis; By building a dynamic equipment fault judgment model, it is possible to analyze the various signal characteristics collected during the operation of the dynamic equipment in real time, and to promptly judge abnormal signals to predict faults, which helps to carry out maintenance work in advance. By diagnosing abnormal signals, the most likely fault type of the current dynamic equipment can be determined, thereby assisting maintenance personnel to repair the dynamic equipment, enhancing practicality and functionality.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information, characterized in that: The method comprises the following steps: S1. Collect parameters of normal and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals and sound signals, through non-intrusive signal acquisition solutions; S2. Extract key features from the collected vibration, external magnetic field, sound and temperature signals of the currently faulty moving equipment, select the feature indicators with the highest correlation with different fault states of the moving equipment through algorithms, and construct feature vectors; S3. Based on the characteristic vectors of the dynamic equipment under different fault states, a dynamic equipment fault judgment model is constructed. Based on the collected operating parameter characteristics of the current dynamic equipment and combined with the dynamic equipment fault judgment model, multi-dimensional judgments are made on potential faults that have not occurred during the operation of the dynamic equipment, and faults that have occurred are diagnosed.

2. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 1 is characterized in that: The S1 comprises the following sub-steps: S11. Place high-sensitivity vibration sensors at key locations of the moving equipment, including bearings, rotors, mechanical seals, and impellers, to collect vibration characteristics of the moving equipment in different states, including vibration frequency, amplitude, vibration phase, and spectrum characteristics; S12. Based on the principle of electromagnetic induction, the induction coil installed outside the motor of the moving device is used to collect the external magnetic field signals of the moving device under different states, including characteristic frequency and amplitude; S13, monitoring the temperature distribution outside different moving devices through non-contact infrared temperature sensors, and combining the data collected by internal temperature sensors to obtain the temperature changes of the moving devices under different states; S14. A directional microphone is arranged outside the moving device in different states to collect the operating sound signal of the moving device in a contactless manner.

3. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 1 is characterized in that: The S2 comprises the following steps: S21, for the vibration signal, external magnetic field signal, temperature signal and sound signal of the moving device under the normal state, respectively extract the key features of the vibration signal, the external magnetic field signal, the temperature signal and the sound signal as the reference state parameter features of the moving device; S22. For the vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment under different fault states, the key features of the vibration signals, external magnetic field signals, temperature signals and sound signals are extracted respectively. Based on the Mahalanobis distance and the characteristics of the benchmark state parameters of the moving equipment, the correlation characteristic indicators under different fault states are screened to construct different fault feature vectors.

4. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 3 is characterized in that: The S21 comprises the following steps: S211. Extracting time domain features and frequency domain features from the vibration signal of the moving equipment under the normal state, wherein the time domain features include mean, standard deviation, peak value, and kurtosis, and the frequency domain features include main frequency, spectrum, and frequency center; S212, extracting characteristic parameters of the external magnetic field signal of the moving device under the normal state, including peak value, maximum intensity change rate, main frequency change, harmonic characteristics, offset, and fluctuation amplitude; S213, extracting trend features and dynamic features from the collected temperature signal of the moving device in the normal state, wherein the trend features include the temperature mean, the temperature change rate, the maximum temperature, and the minimum temperature, and the dynamic features include the temperature fluctuation amplitude, the temperature rise rate, and the temperature gradient; S214, extracting characteristic parameters of the sound signal collected in a normal state, including short-time energy, signal zero-crossing rate, spectrum center frequency, bandwidth, and Mel frequency cepstrum coefficient; S215, extraction of normal moving equipment The vibration signal characteristics, external magnetic field signal characteristics, temperature signal characteristics and sound signal characteristic parameter values ​​are processed by averaging, and the characteristic values ​​after averaging are used as the characteristic vector of the reference state parameters of the dynamic equipment: ； in, Represents the characteristic vector of the reference state parameters of the dynamic equipment, They represent the key features after mean processing of vibration signal, external magnetic field signal, temperature signal and sound signal respectively.

5. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 4 is characterized in that: The S22 comprises the following steps: S221, analyze the vibration signals, external magnetic field signals, temperature signals and sound signals of the moving equipment under different fault conditions Group collection and grouping based on fault type, through Establish different fault type files respectively, and collect the data of the moving equipment under the same fault type. The group parameters are processed by averaging, and the characteristic parameter values ​​after averaging are recorded in the corresponding fault type file to construct the fault state feature vector: ； in, Representative The fault state feature vector under the fault type file, Respectively represent The key features of vibration signal, external magnetic field signal, temperature signal and sound signal after mean value processing under each fault type file; S222. Based on the Mahalanobis distance algorithm and the reference state parameter characteristics of the dynamic equipment, the Mahalanobis distances between the dynamic equipment acquisition parameters and the reference state parameter characteristics under different fault type files are calculated respectively. The algorithm formula is: ； in, represents the covariance matrix of the baseline state parameter eigenvector, Representative The Mahalanobis distance between each feature and the reference state parameter feature under each fault type file; S223, based on Mahalanobis distance threshold , select characteristic indicators with high correlation with each fault state, and the specific steps are: Statistics all > These characteristics are the high-correlation characteristic indicators under different fault states. Based on the selected correlation characteristic indicators, the correlation characteristic vectors under different fault states are constructed: ； in, That is the The associated feature vector under each fault type file, Respectively represent The vibration, magnetic field, temperature and sound characteristics that are highly correlated with the fault status under each fault type file.

6. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 5 is characterized in that: The S3 comprises the following sub-steps: S31. Based on the characteristic vectors of the dynamic equipment under different fault conditions, a dynamic equipment fault determination model is constructed. The specific steps are as follows: The feature vectors of normal state and different fault states are labeled, where the label of normal state is 0 and the label of fault state is 1 to N. Based on the associated feature vector, the feature parameters related to the fault are selected to construct the input feature matrix : ； The equipment fault judgment model is constructed based on the neural network. The output of the input layer is the input feature vector. The output of each hidden layer is transformed nonlinearly through the activation function. The number of neurons in the output layer is equal to the number of categories, which is N+1 categories, including normal state and N fault states. The activation function performs probability distribution conversion and converts the marked feature matrix The neural model is trained with the label input under the fault state to obtain the dynamic equipment fault judgment model; S32. Collect vibration signals, external magnetic field signals, sound signals, and temperature signals during the current operation of the dynamic equipment, extract characteristic parameters, construct the current characteristic vector, and based on the obtained dynamic equipment fault judgment model, make multi-dimensional predictions on potential faults that have not occurred during the operation of the dynamic equipment, and diagnose faults that have occurred.

7. The method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information according to claim 6 is characterized in that: The S32 comprises the following steps: S321. Collect vibration signals, external magnetic field signals, sound signals, and temperature signals during the operation of the current moving equipment, extract characteristic parameters, and construct a real-time characteristic vector: ； in, They respectively represent the key features of the vibration signal, external magnetic field signal, temperature signal and sound signal collected during the operation of the current moving equipment; S322, input the real-time feature vector into the dynamic equipment fault determination model to detect whether there is a potential fault in the dynamic equipment: The output layer usually uses The activation function converts the real-time feature vector input and output into the probability distribution of each category, and selects the category with the highest probability as the final prediction result: When the output result is 0, the current device is in normal state; When the prediction result is 1 to N, it means that there is a potential fault in the dynamic equipment; S323. For the dynamic equipment with potential faults, the abnormal characteristic parameters are extracted to obtain the abnormal characteristic vector: ； in, Respectively represent the abnormal characteristics of the vibration signal, external magnetic field signal, temperature signal and sound signal of the current moving equipment, and calculate The cosine similarity of the associated feature vectors under each fault type file is sorted in descending order, and the fault type file with the first ranking is selected as the diagnostic output fault of the current abnormal fault of the dynamic equipment.

8. Intelligent monitoring and maintenance system for moving equipment based on vibration signal fusion of multi-source information, characterized by: The system adopts the method for intelligent monitoring and maintenance of moving equipment based on vibration signal fusion of multi-source information as described in any one of claims 1 to 7, including a data acquisition module, a feature extraction module, a model building module, and an output determination module: The data acquisition module collects parameters of normal moving equipment and faulty moving equipment during operation, including vibration signals, external magnetic field signals, temperature signals and sound signals, through a non-intrusive signal acquisition solution; The feature extraction module extracts key features from the vibration, external magnetic field, sound and temperature signals of the current faulty moving equipment collected by the data acquisition module, selects the feature indicators with the highest correlation with different fault states of the moving equipment through an algorithm, and constructs a feature vector; The model building module builds a dynamic equipment fault determination model through a neural network based on the feature vectors under different fault states of the dynamic equipment, and outputs the model to the output determination module; The output determination module, in combination with the operating parameter characteristics of the current dynamic equipment collected by the data acquisition module, uses the dynamic equipment fault determination model to perform multi-dimensional determination on potential faults that have not occurred during the operation of the dynamic equipment, and diagnose faults that have occurred.

Images