Complex industrial system anomaly detection method and system based on physical knowledge embedding

Abstract

This application proposes a complex industrial system anomaly detection method and system embedded with physical knowledge, the method includes: collecting sensor data and preprocessing it; extracting the spectral characteristics of vibration data and noise data according to historical data and system structure parameters, setting the thresholds of temperature data, pressure data, and flow data in combination with equipment safety requirements, and constructing a physical knowledge feature set; using different wavelet analysis processes to extract key frequency features and all characteristic frequencies for vibration data and noise data, respectively, using sparse transformation to capture feature information for temperature data, pressure data, and flow data to obtain a trend threshold set, and constructing a measurement data feature set; comparing the physical knowledge feature set with the measurement data feature set, and performing anomaly detection on industrial production equipment based on the set distance threshold and amplitude threshold. Through the scheme proposed in this application, anomaly detection and early warning of industrial production equipment can be achieved.

Description

Technical Field

[0001] The present application relates to the field of industrial production equipment monitoring and maintenance, and in particular to a complex industrial system anomaly detection method and system with embedded physical knowledge. Background Art

[0002] In the field of industrial production, the normal operation of equipment is crucial to production efficiency and product quality. However, due to factors such as long-term operation of equipment and changes in the external environment, various abnormalities may occur in the equipment, such as abnormal vibration, abnormal temperature, abnormal pressure, etc. Traditional equipment monitoring methods mainly rely on a single indicator or a few indicators to judge the status of the equipment, which has problems such as insufficient sensitivity and low accuracy.

[0003] In addition, existing anomaly detection methods often lack comprehensive consideration of the physical structure characteristics and operating status information of the equipment, resulting in the inability to fully evaluate the operating status of the equipment. Summary of the invention

[0004] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0005] To this end, the first purpose of this application is to propose a complex industrial system anomaly detection method embedded with physical knowledge, which realizes comprehensive monitoring and intelligent diagnosis of the status of industrial production equipment by integrating physical principles, data analysis and machine learning technology.

[0006] The second objective of this application is to propose a complex industrial system anomaly detection system embedded with physical knowledge.

[0007] The third objective of the present application is to provide an electronic device.

[0008] To achieve the above objectives, the first embodiment of the present application proposes a complex industrial system anomaly detection method embedded with physical knowledge, including:

[0009] Collecting sensor data and preprocessing the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization;

[0010] According to historical data and system structure parameters, the spectrum characteristics of pre-processed vibration data and noise data are extracted, and the thresholds of pre-processed temperature data, pressure data, and flow data are set in combination with equipment safety requirements to construct a physical knowledge feature set;

[0011] Different wavelet analysis processes are used to extract key frequency features and all characteristic frequencies from the preprocessed vibration data and noise data. Sparse transformation is used to capture characteristic information of the preprocessed temperature data, pressure data, and flow data to obtain a trend threshold set and construct a measurement data feature set.

[0012] The physical knowledge feature set is compared with the measurement data feature set, and abnormality detection is performed on the industrial production equipment based on the set distance threshold and amplitude threshold.

[0013] Optionally, collecting sensor data and preprocessing the sensor includes:

[0014] Get the collected data from different sensors, where the collected data of each sensor is a time series, expressed as , Indicates the time points of data collection, Indicates different sensors;

[0015] Use filters or signal processing techniques to remove noise from the data collected by different sensors;

[0016] The collected data after noise removal is normalized, and the expression is:

[0017]

[0018] in, is the normalized collected data. is the collected data after noise removal, For sensor The mean of It is a sensor The standard deviation of .

[0019] Optionally, the spectral features of the pre-processed vibration data and noise data are extracted according to the historical data and system structure parameters, and the thresholds of the pre-processed temperature data, pressure data, and flow data are set in combination with the equipment safety requirements to construct a physical knowledge feature set, including:

[0020] For the vibration data measured by the acceleration sensor, collect its historical data, and calculate the spectrum feature set that may appear in the vibration data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ;

[0021] For the noise data measured by the acoustic sensor, collect its historical data and calculate the spectrum feature set that may appear in the noise data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ;

[0022] For the temperature data measured by the temperature sensor, set the upper and lower temperature thresholds according to the safety requirements of complex industrial system equipment and ;

[0023] For the pressure data measured by the pressure sensor, set the upper and lower pressure thresholds according to the safety requirements of complex industrial system equipment and ;

[0024] For the flow data measured by the flow sensor, set the upper and lower flow thresholds according to the safety requirements of complex industrial system equipment. and ;

[0025] According to the physical knowledge collection of vibration fault characteristics , Physical knowledge collection of vibration fault characteristics , upper and lower temperature thresholds and , upper and lower pressure thresholds and , Upper and lower flow thresholds and , construct a physical knowledge feature set, the expression is:

[0026]

[0027] in, Represents the physical knowledge feature set.

[0028] Optionally, the process of extracting key frequency features from the pre-processed vibration data using a wavelet analysis process includes:

[0029] The high-frequency noise in the vibration data is removed by an adaptive wavelet denoising method, and the expression is:

[0030]

[0031] in, is the wavelet coefficient after threshold denoising, is the wavelet basis function, is the vibration signal after denoising, Indicates time;

[0032] The adaptive wavelet odd function matching method is used to automatically match the vibration signal characteristics within the window. The expression of the matching process is:

[0033]

[0034] in, To reconstruct the signal, is the wavelet coefficient after matching;

[0035] Through the matched wavelet coefficients, the denoised vibration signal is subjected to wavelet transform analysis to obtain a time-frequency spectrum, where the wavelet transform formula is:

[0036]

[0037] in, The scale parameter is , the translation parameters are The complex conjugate of the wavelet function, For each scale The wavelet coefficients under ;

[0038] Extract key frequency features from the time-frequency spectrum to form a key frequency feature set, which is expressed as:

[0039]

[0040] in, represents the extracted key frequency features, is a set of key frequency features.

[0041] Optionally, the process of extracting all characteristic frequencies from the preprocessed noise data using a wavelet analysis process includes:

[0042] The interference noise in the noise data is removed by an adaptive MEL filter, and the expression is:

[0043]

[0044] in, is the noise signal after denoising, is a MEL filter, represents the convolution operation, is the original noise signal;

[0045] The denoised noise signal is transformed from the time domain to the time-frequency domain using wavelet transform to obtain spectrum fragments, where the wavelet transform formula is:

[0046]

[0047] In the time-frequency domain, the spectral clustering algorithm is used to perform cluster analysis on the spectrum fragments. The analysis process of the spectral clustering algorithm is as follows:

[0048]

[0049] in, is the spectral similarity matrix, is the spectrum cluster center;

[0050] The fault characteristic frequency band boundary under strong noise interference is determined according to the spectral clustering result after clustering, and a zero-phase filter is designed according to the fault characteristic frequency band boundary. The expression form of the zero-phase filter is:

[0051]

[0052] The denoised noise signal is filtered by the zero phase filter to reconstruct the components, and all characteristic frequency sets are obtained by Fourier transforming the reconstructed components. The expression of all characteristic frequency sets is:

[0053]

[0054] in, is the set of all characteristic frequencies, Represents all the extracted feature frequencies.

[0055] Optionally, sparse transformation is used to capture feature information of the preprocessed temperature data, pressure data, and flow data to obtain a trend threshold set, including:

[0056] The signal It is expressed as the product of an overcomplete basis and a sparse coefficient, and the expression is:

[0057]

[0058] in, Indicates the pre-processed temperature signal, pressure signal or flow signal. For an over-complete base, is the sparse coefficient;

[0059] The sparse coefficients are optimized by sparse optimization algorithm. Optimize and get the signal according to the optimized sparse coefficient The sparse representation of is expressed as:

[0060]

[0061] in, For signal The sparse representation of is the sparse coefficient obtained through sparse optimization;

[0062] The key features extracted from sparse representation are used to predict trends. The expression is:

[0063]

[0064] in, is the sparse coefficient, is the basis function, Indicates a point in the future Predicted trend signals;

[0065] Set trend thresholds based on predicted signal trends ,in, and Indicates the upper and lower temperature trend thresholds, and Indicates the upper and lower pressure trend thresholds, and Indicates the upper and lower traffic trend thresholds.

[0066] Optionally, also include:

[0067] The constructed measurement data feature set is expressed as:

[0068]

[0069] in, is the measurement data feature set.

[0070] Optionally, the comparing the physical knowledge feature set with the measurement data feature set and performing abnormality detection on the industrial production equipment based on a set distance threshold and amplitude threshold includes:

[0071] Define the distance function:

[0072]

[0073] in, and They represent the corresponding feature vectors in the physical knowledge feature set and the measurement data feature set, respectively. Indicates the defined distance value;

[0074] like When the set distance threshold is exceeded, a preliminary warning is issued for the status of industrial production equipment;

[0075] After the preliminary warning is made, the amplitude of each feature vector in the measurement data feature set is calculated. If there is a situation where the amplitude exceeds the set amplitude threshold, it is determined that the industrial production equipment is abnormal.

[0076] To achieve the above-mentioned purpose, the second embodiment of the present application proposes a complex industrial system anomaly detection system embedded with physical knowledge, including:

[0077] A preprocessing module, used to collect sensor data and preprocess the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization;

[0078] The physical knowledge feature set construction module is used to extract the spectrum characteristics of the pre-processed vibration data and noise data based on historical data and system structure parameters, set the thresholds of the pre-processed temperature data, pressure data, and flow data in combination with equipment safety requirements, and construct the physical knowledge feature set;

[0079] The measurement data feature set construction module is used to extract key frequency features and all characteristic frequencies from the pre-processed vibration data and noise data using different wavelet analysis processes, and to capture feature information from the pre-processed temperature data, pressure data, and flow data using sparse transformation to obtain a trend threshold set, thereby constructing a measurement data feature set;

[0080] The anomaly detection module is used to compare the physical knowledge feature set with the measurement data feature set, and perform anomaly detection on industrial production equipment based on a set distance threshold and amplitude threshold.

[0081] To achieve the above-mentioned purpose, the third aspect of the present application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0082] The memory stores computer-executable instructions;

[0083] The processor executes the computer-executable instructions stored in the memory to implement the method as described in any one of the first aspects above.

[0084] The technical solution provided by the embodiments of the present application brings at least the following beneficial effects:

[0085] (1) By comprehensively considering the physical structure characteristics and operating status information of industrial production equipment, a comprehensive set of characteristic indicators is constructed, which can more comprehensively describe the equipment status;

[0086] (2) The spectrum analysis method based on adaptive wavelet window can accurately analyze the signal at different frequency resolutions and more effectively capture the changes in device status;

[0087] (3) Comprehensive consideration of multiple indicators and analysis methods can effectively identify abnormal conditions of equipment, improve the accuracy and reliability of detection, and reduce the false alarm rate;

[0088] (4) This method combines multiple technical means, but is simple to implement, easy to operate and apply to industrial production sites, providing effective support for equipment management and maintenance.

[0089] In summary, the entire framework realizes comprehensive monitoring and intelligent diagnosis of the status of industrial production equipment by integrating physical principles, data analysis and machine learning technology, which helps preventive maintenance and early detection of faults, reduces unplanned downtime, and improves overall production efficiency and safety.

[0090] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0092] Figure 1 It is a flow chart of a complex industrial system anomaly detection method embedded with physical knowledge according to an embodiment of the present application;

[0093] Figure 2 It is a block diagram of a complex industrial system anomaly detection device embedded with physical knowledge according to an embodiment of the present application;

[0094] Figure 3 It is a block diagram of an electronic device. DETAILED DESCRIPTION

[0095] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0096] The following describes a complex industrial system anomaly detection method and device with embedded physical knowledge according to an embodiment of the present application with reference to the accompanying drawings.

[0097] Figure 1 is a flow chart of a complex industrial system anomaly detection method embedded with physical knowledge according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0098] Step 101 , collecting sensor data and preprocessing the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization.

[0099] In the embodiment of the present application, the collected data of different sensors are obtained, wherein the collected data of each sensor is a time series, which is expressed as , Indicates the time points of data collection, Indicates different sensors.

[0100] It can be understood that vibration data can be represented as a time series of vibration acceleration, noise data can be represented as a time series of sound, temperature data can be represented as a time series of temperature, pressure data can be represented as a time series of pressure, and flow data can be represented as a time series of flow.

[0101] In one embodiment of the present application, an industrial production device is equipped with an acceleration sensor, a temperature sensor, and a pressure sensor to collect sensor data over a period of time, including vibration data, temperature data, and pressure data. These data form a time series, which is expressed as:

[0102] Vibration sensor data (vibration amplitude, unit: mm):

[0103]

[0104] Temperature sensor data (temperature, unit: Celsius):

[0105]

[0106] Pressure sensor data (pressure, unit: MPa):

[0107]

[0108] In order to ensure the accuracy, completeness, consistency and reliability of the data and to improve the quality and accuracy of data analysis, the collected data must first be subjected to noise removal.

[0109] In one embodiment of the present application, a filter or signal processing technology is used to remove noise from the collected data of different sensors.

[0110] Then, in order to eliminate the dimensional differences in the output values ​​of different sensors, the collected data after noise removal needs to be normalized. The expression is:

[0111]

[0112] in, is the normalized collected data. is the collected data after noise removal, For sensor The mean of It is a sensor The standard deviation of .

[0113] In one embodiment of the present application, it is assumed that the data has been subjected to noise removal and data normalization processing, and the following normalized data is obtained:

[0114] Vibration sensor data (normalized):

[0115]

[0116] Temperature sensor data (normalized):

[0117]

[0118] Pressure sensor data (normalized):

[0119]

[0120] Step 102, extracting the spectral features of the pre-processed vibration data and noise data based on historical data and system structure parameters, setting thresholds of the pre-processed temperature data, pressure data, and flow data in combination with equipment safety requirements, and constructing a physical knowledge feature set.

[0121] In the embodiment of the present application, for the vibration data measured by the acceleration sensor, its historical data is collected, and based on the structural parameters of the complex industrial system, a set of spectral features that may appear in the vibration data when a failure occurs is calculated. , further, based on historical data and fault records, the spectrum feature set generated when the fault occurs is collected , the final comprehensive and , and built a physical knowledge set of vibration fault characteristics , which reflects the typical vibration characteristics of the system under fault conditions.

[0122] For the noise data measured by the acoustic sensor, historical data analysis and complex industrial system structure parameters are also used to calculate the possible spectral feature set of the noise data when a fault occurs. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics , which helps to identify fault signals in noise.

[0123] In addition, temperature, pressure, and flow are key slowly changing parameters, and thresholds are set according to equipment safety requirements.

[0124] For the temperature data measured by the temperature sensor, set the upper and lower temperature thresholds according to the safety requirements of complex industrial system equipment and ;

[0125] For the pressure data measured by the pressure sensor, set the upper and lower pressure thresholds according to the safety requirements of complex industrial system equipment and ;

[0126] For the flow data measured by the flow sensor, set the upper and lower flow thresholds according to the safety requirements of complex industrial system equipment. and .

[0127] Finally, according to the physical knowledge of vibration fault characteristics , Physical knowledge collection of vibration fault characteristics , upper and lower temperature thresholds and , upper and lower pressure thresholds and , Upper and lower flow thresholds and , construct a physical knowledge feature set, the expression is:

[0128]

[0129] in, Represents a set of physical knowledge features.

[0130] By constructing a set of physical knowledge features, a solid foundation is provided for subsequent anomaly detection, condition monitoring and fault diagnosis.

[0131] Moreover, by systematically integrating multi-source data and expertise in this step, not only the accuracy and timeliness of industrial equipment status assessment are improved, but also a scientific basis is provided for preventive maintenance and immediate intervention, effectively reducing the risk of production interruption caused by equipment failure.

[0132] Step 103, using different wavelet analysis processes to extract key frequency features and all characteristic frequencies from the preprocessed vibration data and noise data, and using sparse transformation to capture characteristic information from the preprocessed temperature data, pressure data, and flow data to obtain a trend threshold set, and construct a measurement data feature set.

[0133] In one embodiment of the present application, the process of extracting key frequency features from the preprocessed vibration data using a wavelet analysis process includes:

[0134] The high-frequency noise in the vibration data is removed by the adaptive wavelet denoising method, and the expression is:

[0135]

[0136] in, is the wavelet coefficient after threshold denoising, is the wavelet basis function, is the vibration signal after denoising, Indicates time.

[0137] Wavelet denoising utilizes the multi-scale characteristics of wavelet transform to decompose the signal at different frequencies, thereby separating the signal from the noise. Adaptive wavelet denoising removes noise more accurately by dynamically adjusting the threshold of the wavelet coefficients while retaining the useful information of the signal as much as possible.

[0138] Then, the adaptive wavelet odd function matching method is used to automatically match the vibration signal features in the window. The expression of the matching process is:

[0139]

[0140] in, To reconstruct the signal, are the matched wavelet coefficients.

[0141] It can be understood that the above process seems to be a restatement of the wavelet reconstruction process, but adaptive wavelet odd function matching should be understood as an optimization process, which aims to make the reconstructed signal better fit or match the characteristics of the original signal by adjusting the parameters of the wavelet basis (such as scale and position), especially those with physical significance, such as vibration modes of specific frequencies.

[0142] Through the matched wavelet coefficients, the denoised vibration signal is subjected to wavelet transform analysis to obtain a time-frequency spectrum, where the wavelet transform formula is:

[0143]

[0144] in, The scale parameter is , the translation parameters are The complex conjugate of the wavelet function is used to ensure the completeness of the transformation. For each scale The wavelet coefficients below.

[0145] Finally, the key frequency features are extracted from the time-frequency spectrum to form a key frequency feature set, which is expressed as:

[0146]

[0147] in, represents the extracted key frequency features, is a set of key frequency features.

[0148] It can be understood that the time-frequency spectrum obtained by wavelet transform can clearly show the energy distribution of the signal at different time points and different frequencies. From this time-frequency spectrum, the present application can identify and extract key frequency characteristic signals related to the failure mode of the vibration signal, that is, those specific frequency components that are closely related to potential failures or abnormal events.

[0149] In another embodiment of the present application, the process of extracting all characteristic frequencies from the preprocessed noise data using a wavelet analysis process includes:

[0150] The interference noise in the noise data is removed by adaptive MEL filter, and the expression is:

[0151]

[0152] in, is the noise signal after denoising, is a MEL filter, represents the convolution operation, is the original noise signal.

[0153] It should be noted that the MEL filter is designed to simulate the perception characteristics of the human auditory system to sounds of different frequencies, and effectively remove noise by adaptively adjusting the filter parameters while maintaining important information in the signal. The adaptive MEL filter can be dynamically adjusted according to the signal characteristics to more accurately match and remove noise.

[0154] Then, the denoised noise signal is transformed from the time domain to the time-frequency domain using wavelet transform to obtain spectrum fragments, where the wavelet transform formula is:

[0155]

[0156] In this way, local features of the signal are revealed at different scales and locations.

[0157] Then, the spectrum fragments are clustered and analyzed using the spectral clustering algorithm in the time-frequency domain. The analysis process of the spectral clustering algorithm is as follows:

[0158]

[0159] in, is the spectral similarity matrix, is the spectral cluster center.

[0160] Identifying patterns in the signal spectrum through clustering algorithms, especially in the strong noise background, helps to extract the fault characteristic frequency bands, and by minimizing the sum of squared differences of the spectrum similarity matrix, the cluster centers are determined to divide the fault characteristic frequency bands.

[0161] Furthermore, the fault characteristic frequency band boundary under strong noise interference is determined according to the spectral clustering results after clustering, and a zero-phase filter is designed according to the fault characteristic frequency band boundary. The expression of the zero-phase filter is:

[0162]

[0163] It can be understood that the purpose of designing a zero-phase filter is to keep the phase of the signal unchanged during the filtering process, which is crucial to restoring the original form of the signal.

[0164] Finally, the denoised noise signal is filtered by a zero-phase filter to reconstruct the components, and all characteristic frequency sets are obtained by Fourier transforming the reconstructed components. The expression of all characteristic frequency sets is:

[0165]

[0166] in, is the set of all characteristic frequencies, Represents all the extracted feature frequencies.

[0167] In addition, for slow-changing signals such as temperature, pressure, and flow, the sparse transformation method is used to capture their characteristic information and predict their trends to obtain a trend threshold set. The specific process is: It is expressed as the product of an overcomplete basis and a sparse coefficient, and the expression is:

[0168]

[0169] in, Indicates the pre-processed temperature signal, pressure signal or flow signal. For an over-complete base, are sparse coefficients. An overcomplete basis means that the number of bases is greater than the dimension of the signal, which allows the signal to be represented in a sparse manner, that is, most of the coefficients are close to zero, and only a small number of non-zero coefficients carry the main information of the signal.

[0170] Then, the sparse coefficients are optimized by the sparse optimization algorithm. Optimize and get the signal according to the optimized sparse coefficient The sparse representation of is expressed as:

[0171]

[0172] in, For signal The sparse representation of is the sparse coefficient obtained through sparse optimization;

[0173] Furthermore, the key features extracted from the sparse representation are used to predict the trend, and the expression is:

[0174]

[0175] in, is the sparse coefficient, is the basis function, Indicates a point in the future Predicted trend signals;

[0176] Finally, based on the predicted signal trend, a trend threshold set is set ,in, and Indicates the upper and lower temperature trend thresholds, and Indicates the upper and lower pressure trend thresholds, and Indicates the upper and lower traffic trend thresholds.

[0177] Finally, the constructed measurement data feature set is expressed as:

[0178]

[0179] in, is a set of measurement data features.

[0180] Step 104 , comparing the physical knowledge feature set with the measurement data feature set, and performing anomaly detection on the industrial production equipment based on the set distance threshold and amplitude threshold.

[0181] In one embodiment of the present application, a distance function is defined:

[0182]

[0183] in, and They represent the corresponding feature vectors in the physical knowledge feature set and the measurement data feature set, respectively. Indicates the defined distance value;

[0184] like When the set distance threshold is exceeded, a preliminary warning is issued for the status of industrial production equipment;

[0185] After a preliminary warning is issued, the amplitude of each feature vector in the measurement data feature set is calculated. If there is an amplitude exceeding the set amplitude threshold, the industrial production equipment is judged to be abnormal.

[0186] In one embodiment of the present application, once the distance value is detected to exceed the threshold, the system will trigger an abnormal alarm mechanism, which may include but is not limited to:

[0187] Immediate notification: Send alarms to operators or maintenance teams via email, SMS, APP push, etc.

[0188] Log records: record the abnormality occurrence time, severity, duration, related equipment status and other information in detail for subsequent analysis;

[0189] Automatic intervention: Based on preset rules, safety measures may be implemented, such as reducing speed, switching to backup equipment, adjusting process parameters, etc., to reduce damage.

[0190] In order to implement the above embodiments, the present application also proposes a complex industrial system anomaly detection system embedded with physical knowledge.

[0191] Figure 2 1 is a block diagram of a complex industrial system anomaly detection system 10 with embedded physical knowledge according to an embodiment of the present application, including:

[0192] A preprocessing module 100 is used to collect sensor data and preprocess the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization;

[0193] The physical knowledge feature set construction module 200 is used to extract the spectrum characteristics of the pre-processed vibration data and noise data according to the historical data and system structure parameters, set the thresholds of the pre-processed temperature data, pressure data, and flow data in combination with the equipment safety requirements, and construct the physical knowledge feature set;

[0194] The measurement data feature set construction module 300 is used to extract key frequency features and all characteristic frequencies from the pre-processed vibration data and noise data using different wavelet analysis processes, and to capture feature information from the pre-processed temperature data, pressure data, and flow data using sparse transformation to obtain a trend threshold set, thereby constructing a measurement data feature set;

[0195] The anomaly detection module 400 is used to compare the physical knowledge feature set with the measurement data feature set, and perform anomaly detection on the industrial production equipment based on the set distance threshold and amplitude threshold.

[0196] Regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0197] Figure 3 A schematic block diagram of an example electronic device 700 that can be used to implement an embodiment of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0198] like Figure 3 As shown, the device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0199] A number of components in the device 700 are connected to the I / O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a disk, an optical disk, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows the device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0200] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above. For example, in some embodiments, the scheme proposed in the present application may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the method shown in the present application in any other appropriate manner (e.g., by means of firmware).

[0201] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0202] The program code for implementing the method of the present application can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, implements the functions / operations specified in the flow chart and / or block diagram. The program code can be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0203] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0204] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0205] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0206] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0207] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution of this application can be achieved, and this document is not limited here.

[0208] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A complex industrial system anomaly detection method embedded with physical knowledge, characterized in that: include: Collecting sensor data and preprocessing the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization; According to historical data and system structure parameters, the spectrum characteristics of pre-processed vibration data and noise data are extracted, and the thresholds of pre-processed temperature data, pressure data, and flow data are set in combination with equipment safety requirements to construct a physical knowledge feature set; Different wavelet analysis processes are used to extract key frequency features and all characteristic frequencies from the preprocessed vibration data and noise data. Sparse transformation is used to capture characteristic information of the preprocessed temperature data, pressure data, and flow data to obtain a trend threshold set and construct a measurement data feature set. Comparing the physical knowledge feature set with the measurement data feature set, and performing abnormality detection on the industrial production equipment based on a set distance threshold and amplitude threshold; The method extracts the spectral features of the pre-processed vibration data and noise data based on the historical data and system structure parameters, sets the thresholds of the pre-processed temperature data, pressure data, and flow data in combination with the equipment safety requirements, and constructs a physical knowledge feature set, including: For the vibration data measured by the acceleration sensor, collect its historical data, and calculate the spectrum feature set that may appear in the vibration data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ; For the noise data measured by the acoustic sensor, collect its historical data and calculate the spectrum feature set that may appear in the noise data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ; For the temperature data measured by the temperature sensor, set the upper and lower temperature thresholds according to the safety requirements of complex industrial system equipment and ; For the pressure data measured by the pressure sensor, set the upper and lower pressure thresholds according to the safety requirements of complex industrial system equipment and ; For the flow data measured by the flow sensor, set the upper and lower flow thresholds according to the safety requirements of complex industrial system equipment. and ; According to the physical knowledge collection of vibration fault characteristics , Physical knowledge collection of vibration fault characteristics , upper and lower temperature thresholds and , upper and lower pressure thresholds and , Upper and lower flow thresholds and , construct a physical knowledge feature set, the expression is: in, represents the physical knowledge feature set; The process of extracting key frequency features from the pre-processed vibration data using wavelet analysis includes: The high-frequency noise in the vibration data is removed by an adaptive wavelet denoising method, and the expression is: in, is the wavelet coefficient after threshold denoising, is the wavelet basis function, is the vibration signal after denoising, Indicates time; The adaptive wavelet odd function matching method is used to automatically match the vibration signal characteristics within the window. The expression of the matching process is: in, To reconstruct the signal, is the wavelet coefficient after matching, is the number of layers of wavelet decomposition; Through the matched wavelet coefficients, the denoised vibration signal is subjected to wavelet transform analysis to obtain a time-frequency spectrum, where the wavelet transform formula is: in, The scale parameter is , the translation parameters are The complex conjugate of the wavelet function, For each scale The wavelet coefficients under ; Extract key frequency features from the time-frequency spectrum to form a key frequency feature set, which is expressed as: in, represents the extracted key frequency features, is a set of key frequency features; The process of extracting all characteristic frequencies from the preprocessed noise data using wavelet analysis includes: The interference noise in the noise data is removed by an adaptive MEL filter, and the expression is: in, is the noise signal after denoising, is a MEL filter, represents the convolution operation, is the original noise signal, Indicates the total number of MEL filters; The denoised noise signal is transformed from the time domain to the time-frequency domain using wavelet transform to obtain spectrum fragments, where the wavelet transform formula is: In the time-frequency domain, the spectral clustering algorithm is used to perform cluster analysis on the spectrum fragments. The analysis process of the spectral clustering algorithm is as follows: in, is the spectral similarity matrix, is the spectral cluster center, For the The cluster center to which the spectrum fragments belong; The fault characteristic frequency band boundary under strong noise interference is determined according to the spectral clustering result after clustering, and a zero-phase filter is designed according to the fault characteristic frequency band boundary. The expression form of the zero-phase filter is: in, is the frequency response of the original filter; The denoised noise signal is filtered by the zero phase filter to reconstruct the components, and all characteristic frequency sets are obtained by Fourier transforming the reconstructed components. The expression of all characteristic frequency sets is: in, is the set of all characteristic frequencies, represents all the feature frequencies extracted; Sparse transformation is used to capture feature information of preprocessed temperature data, pressure data, and flow data to obtain a set of trend thresholds, including: The signal It is expressed as the product of an overcomplete basis and a sparse coefficient, and the expression is: in, Indicates the pre-processed temperature signal, pressure signal or flow signal. For an over-complete base, is the sparse coefficient; The sparse coefficients are optimized by sparse optimization algorithm. Optimize and get the signal according to the optimized sparse coefficient The sparse representation of is expressed as: in, For signal The sparse representation of is the sparse coefficient obtained through sparse optimization; The key features extracted from sparse representation are used to predict trends. The expression is: in, is the sparse coefficient, is the basis function, Indicates a point in the future Predicted trend signals; Set trend thresholds based on predicted signal trends ,in, and Indicates the upper and lower temperature trend thresholds, and Indicates the upper and lower pressure trend thresholds, and Indicates the upper and lower traffic trend thresholds.

2. The method according to claim 1, characterized in that The collecting of sensor data and preprocessing of the sensor include: Get the collected data from different sensors, where the collected data of each sensor is a time series, expressed as , Indicates the time points of data collection, Indicates different sensors; Use filters or signal processing techniques to remove noise from the data collected by different sensors; The collected data after noise removal is normalized, and the expression is: in, is the normalized collected data. is the collected data after noise removal, For sensor The mean of It is a sensor The standard deviation of .

3. The method according to claim 2, characterized in that Also includes: The constructed measurement data feature set is expressed as: in, is the measurement data feature set.

4. The method according to claim 3, characterized in that The comparing the physical knowledge feature set with the measurement data feature set, and performing abnormality detection on the industrial production equipment based on a set distance threshold and amplitude threshold, includes: Define the distance function: in, and They represent the corresponding feature vectors in the physical knowledge feature set and the measurement data feature set, respectively. Indicates the defined distance value; like When the set distance threshold is exceeded, a preliminary warning is issued for the status of industrial production equipment; After the preliminary warning is made, the amplitude of each feature vector in the measurement data feature set is calculated. If there is a situation where the amplitude exceeds the set amplitude threshold, it is determined that the industrial production equipment is abnormal.

5. A complex industrial system anomaly detection system embedded with physical knowledge, characterized in that: include: A preprocessing module, used to collect sensor data and preprocess the sensor data, wherein the sensor data includes vibration data, noise data, temperature data, pressure data and flow data, and the preprocessing process includes noise removal and data normalization; The physical knowledge feature set construction module is used to extract the spectrum characteristics of the pre-processed vibration data and noise data based on historical data and system structure parameters, set the thresholds of the pre-processed temperature data, pressure data, and flow data in combination with equipment safety requirements, and construct the physical knowledge feature set; The measurement data feature set construction module is used to extract key frequency features and all characteristic frequencies from the pre-processed vibration data and noise data using different wavelet analysis processes, and to capture feature information from the pre-processed temperature data, pressure data, and flow data using sparse transformation to obtain a trend threshold set, thereby constructing a measurement data feature set; An anomaly detection module, used to compare the physical knowledge feature set with the measurement data feature set, and perform anomaly detection on industrial production equipment based on a set distance threshold and amplitude threshold; The method extracts the spectral features of the pre-processed vibration data and noise data based on the historical data and system structure parameters, sets the thresholds of the pre-processed temperature data, pressure data, and flow data in combination with the equipment safety requirements, and constructs a physical knowledge feature set, including: For the vibration data measured by the acceleration sensor, collect its historical data, and calculate the spectrum feature set that may appear in the vibration data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ; For the noise data measured by the acoustic sensor, collect its historical data and calculate the spectrum feature set that may appear in the noise data when a failure occurs based on the structural parameters of the complex industrial system. , and collect the spectrum feature set generated when the fault occurs based on historical data and fault records , according to the spectrum feature set and , build a physical knowledge set of vibration fault characteristics ; For the temperature data measured by the temperature sensor, set the upper and lower temperature thresholds according to the safety requirements of complex industrial system equipment and ; For the pressure data measured by the pressure sensor, set the upper and lower pressure thresholds according to the safety requirements of complex industrial system equipment and ; For the flow data measured by the flow sensor, set the upper and lower flow thresholds according to the safety requirements of complex industrial system equipment. and ; According to the physical knowledge collection of vibration fault characteristics , Physical knowledge collection of vibration fault characteristics , upper and lower temperature thresholds and , upper and lower pressure thresholds and , Upper and lower flow thresholds and , construct a physical knowledge feature set, the expression is: in, represents the physical knowledge feature set; The process of extracting key frequency features from the pre-processed vibration data using wavelet analysis includes: The high-frequency noise in the vibration data is removed by an adaptive wavelet denoising method, and the expression is: in, is the wavelet coefficient after threshold denoising, is the wavelet basis function, is the vibration signal after denoising, Indicates time; The adaptive wavelet odd function matching method is used to automatically match the vibration signal characteristics within the window. The expression of the matching process is: in, To reconstruct the signal, is the wavelet coefficient after matching, is the number of layers of wavelet decomposition; Through the matched wavelet coefficients, the denoised vibration signal is subjected to wavelet transform analysis to obtain a time-frequency spectrum, where the wavelet transform formula is: in, The scale parameter is , the translation parameters are The complex conjugate of the wavelet function, For each scale The wavelet coefficients under ; Extract key frequency features from the time-frequency spectrum to form a key frequency feature set, which is expressed as: in, represents the extracted key frequency features, is a set of key frequency features; The process of extracting all characteristic frequencies from the preprocessed noise data using wavelet analysis includes: The interference noise in the noise data is removed by an adaptive MEL filter, and the expression is: in, is the noise signal after denoising, is a MEL filter, represents the convolution operation, is the original noise signal, Indicates the total number of MEL filters; The denoised noise signal is transformed from the time domain to the time-frequency domain using wavelet transform to obtain spectrum fragments, where the wavelet transform formula is: In the time-frequency domain, the spectral clustering algorithm is used to perform cluster analysis on the spectrum fragments. The analysis process of the spectral clustering algorithm is as follows: in, is the spectral similarity matrix, is the spectral cluster center, For the The cluster center to which the spectrum fragments belong; The fault characteristic frequency band boundary under strong noise interference is determined according to the spectral clustering result after clustering, and a zero-phase filter is designed according to the fault characteristic frequency band boundary. The expression form of the zero-phase filter is: in, is the frequency response of the original filter; The denoised noise signal is filtered by the zero phase filter to reconstruct the components, and all characteristic frequency sets are obtained by Fourier transforming the reconstructed components. The expression of all characteristic frequency sets is: in, is the set of all characteristic frequencies, represents all the feature frequencies extracted; Sparse transformation is used to capture feature information of preprocessed temperature data, pressure data, and flow data to obtain a set of trend thresholds, including: The signal It is expressed as the product of an overcomplete basis and a sparse coefficient, and the expression is: in, Indicates the pre-processed temperature signal, pressure signal or flow signal. For an over-complete base, is the sparse coefficient; The sparse coefficients are optimized by sparse optimization algorithm. Optimize and get the signal according to the optimized sparse coefficient The sparse representation of is expressed as: in, For signal The sparse representation of is the sparse coefficient obtained through sparse optimization; The key features extracted from sparse representation are used to predict trends. The expression is: in, is the sparse coefficient, is the basis function, Indicates a point in the future Predicted trend signals; Set trend thresholds based on predicted signal trends ,in, and Indicates the upper and lower temperature trend thresholds, and Indicates the upper and lower pressure trend thresholds, and Indicates the upper and lower traffic trend thresholds.

6. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 4.

Images