Inspection test method and device for maintenance of printing equipment

Abstract

The invention relates to the technical field of printing equipment maintenance, and discloses an inspection test method and device for printing equipment maintenance, and the method comprises the steps: carrying out the vibration signal collection and preprocessing of target positions of a supporting structure, a fixing structure, a paper conveying assembly and a transmission assembly of printing equipment, and obtaining a vibration feature data set; performing time domain analysis and frequency domain analysis to obtain a multi-dimensional vibration feature vector; inputting the multi-dimensional vibration feature vector into a maximum posterior probability model for fault feature calculation to obtain a preliminary fault judgment result; according to the preliminary fault judgment result, a printing equipment transmission system dynamic model is constructed, the inherent frequency and vibration mode characteristics of the system are calculated through modal analysis, and a vibration transfer function is obtained; based on the vibration transfer function, state evaluation is conducted on the printing equipment, the vibration intensity and the dynamic balance index are calculated through multi-parameter collaborative analysis, an equipment state evaluation report is obtained, and accurate state evaluation and early warning of the printing equipment are achieved through the method.

Description

Technical Field

[0001] The present invention relates to the technical field of printing equipment maintenance, and in particular to an inspection and testing method and device for printing equipment maintenance. Background Art

[0002] The stability and reliability requirements of double-sided printing equipment in the production process are constantly increasing. Traditional printing equipment maintenance methods mainly rely on manual experience for regular inspections and fault diagnosis, lacking scientific detection methods and quantitative evaluation standards, resulting in low maintenance efficiency and prone to missed detection and misjudgment.

[0003] During the operation of printing equipment, abnormal vibration often occurs in key parts such as supporting structures, fixed structures, paper transport components and transmission components. These vibrations not only affect the printing quality, but may also cause equipment failure. Existing vibration detection methods often only focus on a single vibration parameter, ignoring the comprehensive characteristics of the vibration signal in the time domain and frequency domain, and are unable to accurately identify the fault type and locate the fault location. Summary of the invention

[0004] The invention provides an inspection and testing method and device for printing equipment maintenance, and the invention realizes accurate evaluation and early warning of the state of the printing equipment.

[0005] In a first aspect, the present invention provides a printing equipment maintenance inspection and testing method, the printing equipment maintenance inspection and testing method comprising: Collect and preprocess vibration signals at target positions of the support structure, fixed structure, paper transport component, and transmission component of the printing equipment to obtain a vibration feature data set; Performing time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; Inputting the multi-dimensional vibration feature vector into a maximum a posteriori probability model to calculate fault features and obtain a preliminary fault determination result; According to the preliminary fault determination result, a dynamic model of the transmission system of the printing equipment is constructed, and the natural frequency and vibration mode characteristics of the system are calculated by modal analysis to obtain the vibration transfer function; Based on the vibration transfer function, the state of the printing equipment is evaluated, and the vibration intensity and dynamic balance index are calculated through multi-parameter collaborative analysis to obtain an equipment state evaluation report.

[0006] In a second aspect, the present invention provides an inspection and testing device for maintaining a printing device, the inspection and testing device for maintaining a printing device comprising: The acquisition module is used to collect and pre-process the vibration signals of the target positions of the supporting structure, the fixed structure, the paper transport component and the transmission component of the printing equipment to obtain a vibration characteristic data set; An analysis module, used for performing time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; A calculation module, used for inputting the multi-dimensional vibration feature vector into a maximum a posteriori probability model to calculate the fault feature and obtain a preliminary fault determination result; A construction module is used to construct a dynamic model of the transmission system of the printing equipment according to the preliminary fault determination result, calculate the natural frequency and vibration mode characteristics of the system through modal analysis, and obtain a vibration transfer function; The evaluation module is used to evaluate the status of the printing equipment based on the vibration transfer function, calculate the vibration intensity and dynamic balance index through multi-parameter collaborative analysis, and obtain an equipment status evaluation report.

[0007] The third aspect of the present invention provides an inspection and testing device for printing equipment maintenance, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the inspection and testing device for printing equipment maintenance executes the above-mentioned inspection and testing method for printing equipment maintenance.

[0008] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned inspection and testing method for printing device maintenance.

[0009] In the technical solution provided by the present invention, a multi-dimensional vibration feature data acquisition and preprocessing system is established, and time domain analysis and frequency domain analysis methods are combined to accurately extract vibration feature information, and the maximum a posteriori probability model is used to calculate fault characteristics. The accurate identification and positioning of fault types are achieved through Bayesian reasoning, which significantly reduces the misjudgment rate of fault judgment. Based on dynamic modeling and modal analysis methods, a complete vibration transfer function system is established to accurately describe the vibration transfer characteristics between various components of the equipment. Through a multi-parameter collaborative analysis method, a comprehensive evaluation of the vibration intensity and dynamic balance state of the equipment is achieved, a standardized state evaluation system is established, and through a quantitative evaluation of the equipment health status index, accurate evaluation and early warning of the equipment status are achieved.

[0010] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0011] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A schematic diagram of an embodiment of a method for inspecting and testing printing equipment maintenance in an embodiment of the present invention; Figure 2 A schematic diagram of an embodiment of a testing device for printing equipment maintenance in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of an inspection and testing device for maintaining printing equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0015] To facilitate understanding of this embodiment, a printing equipment maintenance inspection and testing method disclosed in an embodiment of the present invention is first described in detail. Figure 1 As shown, the method comprises the following steps: 101. Collect and preprocess vibration signals at target positions of a supporting structure, a fixed structure, a paper transport component, and a transmission component of a printing device to obtain a vibration feature data set; It is understandable that the execution subject of the present invention may be a testing device for printing equipment maintenance, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject.

[0016] Specifically, acceleration sensors are installed on the bearing seats at both ends of the printing plate cylinder and the printing base gun to monitor the vibration information of various parts of the equipment in real time during operation. In order to capture the displacement changes of the printing equipment, displacement sensors are installed at the support bearings of the pulley roller and the non-stick roller. The vibration acceleration signal is obtained through the acceleration sensor, and the displacement signal of each part of the equipment is obtained through the displacement sensor. The installation position of the sensor needs to be accurately selected according to the working characteristics of the printing equipment and the vibration characteristics of the key components to ensure that the collected signals are representative and can truly reflect the operating status of the equipment. In the data collection process, the original vibration signals are collected by the acceleration sensor and the displacement sensor, and these signals are appropriately divided according to the operating status of the equipment. The working process of the printing equipment includes three stages: startup, stable operation and shutdown. The original vibration signals are divided into data segments at these time nodes. In this way, the vibration characteristics of different stages are obtained under different working conditions, and the health status of the equipment at each stage and the possible hidden faults are more accurately identified. The signal data of each stage forms an independent data segment. Each data segment is associated with the operating parameters of the equipment. The real-time collected working condition parameters such as printing speed, paper tension and motor speed are synchronized with the vibration signal to form a multi-dimensional correlation data matrix. In this process, the parameters such as printing speed, paper tension and motor speed are used as external working condition factors to model the influence of vibration signals, so that the change of vibration signals is more closely related to the operation status of the equipment. The vibration signals in the correlation data matrix are normalized to eliminate the influence caused by the difference in signal amplitude under different working conditions, so that the data can be compared under a unified scale. Based on the signal of each measuring point, the mean and standard deviation of the vibration signal of each data segment are calculated, and a standardized transformation is performed to eliminate the numerical fluctuations caused by different measuring points or different working conditions, and the standardized vibration data is obtained. According to the position of the measuring point and the acquisition sequence, the standardized vibration data is constructed into a three-dimensional data structure, and a vibration feature data set containing spatial dimension, time dimension and feature dimension is obtained. The spatial dimension reflects the distribution of measuring points in different parts of the printing equipment, the time dimension reflects the time sequence of data acquisition, and the feature dimension involves the changes in the amplitude, frequency and other characteristics of the vibration signal.

[0017] 102. Perform time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; Specifically, the root mean square value and kurtosis value are calculated for each measuring point data in the vibration feature data set. The root mean square value is used to reflect the vibration intensity of the signal, while the kurtosis is used to measure the pulse characteristics of the signal, which helps to identify the existing impact faults. The peak value is divided by the root mean square value to obtain the peak factor, which reflects the degree of extreme values ​​in the signal, and the root mean square value is divided by the arithmetic mean to obtain the waveform factor, which reveals the regularity of the signal morphology. Through this step, the time domain feature parameter group is obtained. The time domain feature parameter group is fitted with data, and the fluctuation trend of the data is calculated by the least squares method to obtain the time domain trend characteristics and identify the long-term change law of the data. Through this process, the long-term change trend of the vibration signal in the time dimension is extracted. The vibration feature data set is converted to the frequency domain for analysis. The time domain signal is converted into the frequency domain signal by the fast Fourier transform to obtain the basic features in the frequency domain. The fast Fourier transform can decompose the periodic components in the vibration signal into different frequency components, so that the system can more accurately analyze the response characteristics of the equipment at different frequencies. The basic features in the frequency domain are divided into frequency bands. The frequency range of 0-60Hz is divided into low frequency band, 60-120Hz is divided into medium frequency band, and 120-200Hz is divided into high frequency band. The energy proportion in each frequency band provides the system with the distribution of frequency components. By calculating the energy proportion of each frequency band, the energy distribution of the vibration signal in different frequency ranges can be identified, which helps the system determine the source of the fault. For example, abnormal energy in the high frequency band is related to friction or wear of the equipment, while abnormal energy in the low frequency band is related to structural faults of the equipment such as imbalance or looseness. Based on the energy distribution of the frequency band, the characteristic frequency is extracted, and the amplitude and phase information of the main frequency component of each frequency band are calculated. For example, the amplitude of the main frequency component can reflect the vibration intensity at a specific frequency, while the phase information helps the system determine the propagation characteristics of the vibration and whether there are problems such as phase shift. The time domain trend features and frequency domain feature parameters are reduced in dimension. Through dimensionality reduction processing, the feature dimension is reduced, the analysis efficiency is improved, and important feature information is retained. Through methods such as principal component analysis, the time domain and frequency domain features are fused into more compact dimensionality reduction features. The reduced-dimensional features are reconstructed according to the spatial positions of the measuring points, and a spatial feature distribution model is established. Through this model, the correlation coefficients between the measuring points are calculated to reveal the mutual relationship between the vibration signals of different measuring points. Spatial correlation features can help the system identify which components or areas have vibration features that affect each other and reveal potential fault transmission paths. The spatial correlation features are standardized and the feature values ​​of different measuring points are adjusted to a unified scale for comparison and analysis. A multidimensional vibration feature vector is obtained.

[0018] 103. Input the multi-dimensional vibration feature vector into the maximum a posteriori probability model to calculate the fault feature and obtain a preliminary fault determination result; Specifically, the multidimensional vibration feature vector is decoupled and split to better analyze and process the spatial and frequency features in the vibration signal. The vibration feature vector is decoupled and split according to the four spatial dimensions of the support structure, fixed structure, paper transport component and transmission component, and the three frequency dimensions of the low frequency band, medium frequency band and high frequency band, to obtain 12 sub-feature vectors. The vibration signal is refined in the two dimensions of space and frequency to capture the operating status of the equipment from different angles. The 12 sub-feature vectors are input into the feature mapping layer of the maximum a posteriori probability model for feature conversion. The feature mapping layer includes three layers of fully connected networks. Each layer of the network processes the input sub-feature vector through a nonlinear activation function, and gradually extracts deeper and more discriminative features. After these layers of mapping, the 12 sub-feature vectors will be converted into 12 64-dimensional mapping feature vectors. The mapping feature vector is input into the probability density calculation layer of the maximum a posteriori probability model. The model uses radial basis kernel function for calculation. The radial basis kernel function can calculate the conditional probability density of different fault types according to the similarity of input features. The kernel bandwidth parameter is set to 0.5, so that a conditional probability is assigned to each fault type in the high-dimensional feature space, and a 96×8-dimensional conditional probability matrix is ​​obtained, in which each row represents a feature vector, each column corresponds to a fault type, and each element in the matrix represents the matching degree of the corresponding feature vector and the fault type. In order to optimize the fault diagnosis results, Bayesian reasoning is performed on the conditional probability matrix. Combined with the preset 8 typical fault prior probability vectors, the posterior probability value of each component under each fault type is calculated by the maximum a posteriori probability criterion. Through this step, the system can calculate the posterior probability of each fault type using the current data under the condition of known prior probability, so as to make a more accurate assessment of the possibility of fault, and obtain a 12×8-dimensional fault probability distribution matrix, in which each column represents the fault probability distribution of a component. Based on the fault probability distribution, each element is compared with the preset fault feature threshold, and all fault types exceeding the threshold and their corresponding component positions are marked to obtain a fault candidate marking matrix. The fault type and probability of each component are displayed. All fault types in the candidate marking matrix are sorted from high to low according to the posterior probability value, and the corresponding component location and frequency band information are recorded to obtain a weighted fault type sequence. In order to verify the accuracy of fault diagnosis, the cosine similarity between the weighted fault type sequence and the feature template in the historical fault database is calculated to measure the similarity between the current fault characteristics and the historical fault mode, and to screen out the specific fault mode with the highest matching degree. For the fault mode location results, the confidence of the diagnosis result is evaluated by calculating the standard deviation (σ value) of the posterior probability. A smaller standard deviation means that the posterior probability distribution is more concentrated and the fault judgment result has a higher credibility; a larger standard deviation means that there is greater uncertainty in the fault judgment.According to the different σ values, the confidence is divided into three categories: when σ < 0.1, it is judged as high confidence, when 0.1 ≤ σ < 0.2, it is medium confidence, and when σ ≥ 0.2, it is low confidence. Combining the fault type, fault location, frequency characteristics and confidence score, a comprehensive preliminary fault judgment result is obtained.

[0019] The 96×8-dimensional conditional probability matrix is ​​normalized to ensure that the conditional probability values ​​of each column are within a uniform scale range, so that these values ​​can be used for subsequent mathematical operations and ensure the stability and accuracy of the calculation results. After normalization, the standardized conditional probability matrix obtained provides standardized data input for subsequent Bayesian reasoning calculations. The data of the 8 typical fault prior probability vectors is expanded, and each prior probability value is replicated 96 times to generate a 96×8-dimensional extended prior probability matrix. The standardized conditional probability matrix and the extended prior probability matrix are matrix-dot multiplied. Through matrix dot multiplication, the prior probability and the corresponding conditional probability of each fault type are considered at the same time to obtain the joint probability of each component under each fault type. The matrix is ​​summed by column to obtain a 96×8-dimensional joint probability matrix, which contains the joint probability information of all components under different fault types. The 96×8-dimensional joint probability matrix is ​​marginalized. By accumulating by row, the joint probability of each fault type is merged to obtain an 8-dimensional marginal probability vector, which reflects the overall probability distribution of each fault type in all components and all measurement points. Based on the joint probability matrix and the marginal probability vector, the Bayesian posterior probability calculation method is used to infer the posterior probability of each component under each fault type. The Bayesian inference formula can calculate the posterior probability of each fault type based on the known conditional probability and prior probability. The posterior probability reflects the possibility of each fault type under given conditions. Through this process, a 96×8 dimensional posterior probability matrix is ​​obtained. In order to combine the posterior probability with the spatial and frequency characteristics of the equipment, the 96×8 dimensional posterior probability matrix is ​​reorganized, and the 96 measurement points are remapped to the combined structure of 4 components and 3 frequency bands to obtain a 12×8 dimensional initial fault probability distribution. The 12×8 dimensional initial fault probability distribution is smoothed to reduce the probability value fluctuation caused by data fluctuation or noise, making the final diagnosis result more stable and reliable. The Gaussian kernel function is used to perform weighted averaging on the probability values ​​of adjacent measurement points. The weighted process can smooth the data and retain important trend information. Through the weighted average of the Gaussian kernel function, the final 12×8 dimensional fault probability distribution is obtained, which accurately reflects the probability of failure of each component in each frequency band.

[0020] 104. Based on the preliminary fault judgment results, a dynamic model of the printing equipment transmission system is constructed, and the system's natural frequency and vibration mode characteristics are calculated through modal analysis to obtain the vibration transfer function; Specifically, the fault location information is combined with the structural parameters of the printing equipment to establish the dynamic model of each component of the equipment. Based on the principle of the coil spring-mass system, a separate dynamic model is established for each component, and the stiffness coefficient and mass of each component are set. Through this step, the dynamic parameter matrix of each component is obtained, which contains important information such as the mass, stiffness and damping of each component. These parameters reflect the dynamic characteristics of each component of the equipment. The component dynamic parameter matrix is ​​combined and modeled. The motion equation of the multi-degree-of-freedom system is derived through the Lagrange equation. The Lagrange equation can describe the motion behavior of the system under the action of external force, and the dynamic equation of the system is derived through it. In this process, it is assumed that the damping coefficient of the system is 0.05 times the mass matrix and 0.02 times the stiffness matrix, so as to simplify the calculation and ensure the actual operability of the model, and obtain the system dynamic equation. The system dynamic equation is converted into a standard form to obtain the n natural frequencies and corresponding vibration mode vectors of the system. The natural frequency represents the frequency of the natural vibration of the system without external excitation, and the vibration mode describes the vibration mode of each component of the system at the natural frequency. In order to ensure the reliability of the calculation results, the orthogonality of the natural frequency and vibration mode vector is verified to ensure that different vibration modes do not interfere with each other, and the standardized modal matrix is ​​obtained. The frequency response function is calculated based on the standardized modal matrix to obtain the frequency response function curve. The frequency response function can describe the vibration response of the system at different frequencies. By calculating the frequency response function based on the standardized modal matrix, the frequency response function curve of the system is obtained, which reflects the amplitude and phase characteristics of the system at different frequencies. The frequency response function curve is decomposed into amplitude and phase components. The least squares method is used to fit these curves, and the numerator and denominator polynomial coefficients of the transfer function are solved to obtain the continuous transfer function of the system, which reflects the dynamic characteristics of the equipment at various frequencies. The continuous system transfer function is bilinearly transformed to obtain the discrete transfer function. The continuous transfer function is converted into a transfer function in the discrete time domain, which is convenient for calculation and analysis in the digital control system. The discrete transfer function can describe the dynamic response of the system in discrete time. The discrete transfer function is organized according to the transfer path of the support structure, fixed structure, paper transport component and transmission component to form a 4×4 transfer function matrix, which reflects the vibration transfer relationship between the components. By combining these transfer functions in parallel, the vibration transfer function of the entire printing device is finally obtained.

[0021] 105. Based on the vibration transfer function, the printing equipment status is evaluated, and the vibration intensity and dynamic balance index are calculated through multi-parameter collaborative analysis to obtain the equipment status evaluation report.

[0022] Specifically, the vibration transfer function is analyzed in the time domain. The vibration response of the support structure, fixed structure, paper transport component and transmission component at each frequency point is calculated by the convolution integral method. The vibration transfer function is combined with the external excitation signal, and the response of the device in the time domain is obtained by the convolution integral formula. The vibration response analysis can provide the dynamic behavior of the device at each frequency and calculate a series of key vibration indicators, including the root mean square value of displacement, the root mean square value of velocity and the root mean square value of acceleration. Through these indicators, a 16-dimensional vibration response characteristic matrix is ​​obtained, in which each dimension corresponds to the vibration intensity of a measuring point at different frequencies. These response characteristics provide the vibration characteristics of each component of the device under different working conditions. The vibration level evaluation value of each measuring point is calculated based on the 16-dimensional vibration response characteristic matrix. The vibration response characteristics are combined with the standard vibration level evaluation system to evaluate whether the vibration intensity of each measuring point exceeds the normal range according to the working requirements and relevant standards of the equipment. The vibration level evaluation value provides a preliminary judgment basis for subsequent fault detection and performance optimization. The resonant frequency characteristics in the vibration transfer function are analyzed to obtain the resonance characteristic curve. The resonance frequency is the key frequency point of the equipment vibration. When the equipment operating frequency approaches or reaches the resonance frequency, the vibration response will be significantly amplified, causing equipment damage. By analyzing the resonance characteristic curve, these key frequencies can be identified, and corresponding measures can be taken to prevent the equipment from operating at these frequencies. The dynamic amplification factor is calculated based on the resonance characteristic curve. The dynamic amplification factor is a coefficient that describes the increase in the system response at a specific frequency, which can help the system understand the degree of amplification of the vibration response of the equipment at these frequencies. By calculating the dynamic amplification factor, the dynamic response amplification factor of the support structure, fixed structure, paper transport component and transmission component is obtained, which reflects the vibration amplification effect of different components near the resonance frequency and helps identify the vulnerable parts and potential failure points of the equipment at a specific frequency. Dynamic balancing calculation is performed on the dynamic response amplification factor and the vibration data of the rotor system. Dynamic balancing calculation is used to evaluate and correct the vibration imbalance problem of the rotor system. The imbalance amount and phase angle are solved by the influence coefficient method to obtain the dynamic balancing correction parameters. The dynamic balancing correction parameters are used to adjust the operating status of the equipment to reduce the vibration and energy loss caused by imbalance. The equipment health status is classified based on the vibration level evaluation value, dynamic response amplification factor and dynamic balancing correction parameters. Combined with the specific working environment and historical fault data of the equipment, the health status of the equipment is divided into different levels, such as "good", "maintenance required" or "faulty", according to the set thresholds and evaluation indicators, to obtain the equipment health status index. Through this index, the operating status of the equipment is clarified and a basis is provided for equipment repair and maintenance decisions. Based on the equipment health status index, an equipment status assessment report is generated to describe the current status of the equipment, potential faults and maintenance recommendations.

[0023] In the embodiment of the present invention, by establishing a multi-dimensional vibration feature data acquisition and preprocessing system, combining time domain analysis and frequency domain analysis methods, vibration feature information is accurately extracted, the maximum a posteriori probability model is used to calculate fault characteristics, and accurate identification and positioning of fault types are achieved through Bayesian reasoning, which significantly reduces the misjudgment rate of fault judgment; based on dynamic modeling and modal analysis methods, a complete vibration transfer function system is established to accurately describe the vibration transfer characteristics between various components of the equipment; through a multi-parameter collaborative analysis method, a comprehensive evaluation of the vibration intensity and dynamic balance state of the equipment is achieved, a standardized state evaluation system is established, and through a quantitative evaluation of the equipment health status index, accurate evaluation and early warning of the equipment status are achieved.

[0024] In a specific embodiment, the process of executing step 101 may specifically include the following steps: Install acceleration sensors on the bearing seats at both ends of the printing plate cylinder and the printing base gun of the printing equipment, and install displacement sensors on the pulley roller and non-stick roller support bearings; Data is collected based on acceleration sensors and displacement sensors to obtain original vibration signals, and the original vibration signals are segmented based on the time nodes of the printing equipment startup, stable operation and shutdown states to obtain segmented vibration data; Establish a corresponding relationship between the segmented vibration data and the real-time collected working parameters of printing speed, paper tension, and motor speed to obtain a correlation data matrix, and perform amplitude normalization processing on the vibration signal in the correlation data matrix, calculate the mean and standard deviation of each measuring point, perform standardization transformation, and obtain standardized vibration data; A three-dimensional data structure is established for the standardized vibration data according to the measurement point locations and the acquisition sequence, and a vibration feature data set including space dimension, time dimension and feature dimension is obtained.

[0025] Specifically, appropriate sensors are installed at key parts of the printing equipment to monitor the vibration of the equipment in real time. Acceleration sensors are installed on the bearing seats at both ends of the printing plate cylinder and the printing base gun to capture the acceleration signal of the equipment during operation, reflecting the vibration intensity and frequency characteristics of each component. At the same time, displacement sensors are installed at the pulley roller and non-stick roller support bearings to monitor the displacement changes of these parts and provide displacement data on the vibration of the equipment. The data of the acceleration sensor and the displacement sensor can effectively complement each other and comprehensively describe the dynamic behavior of the equipment. Data acquisition is performed based on the acceleration sensor and the displacement sensor to obtain the original vibration signal. The original vibration signal is preprocessed. The original vibration signal is divided into multiple data segments according to the time nodes of the startup, stable operation and shutdown status of the printing equipment. By detecting the equipment status change point, the signal is divided into multiple relatively independent time periods with the help of information such as the equipment's working mode switching, speed change or shutdown mark. Assume that the original vibration signal is ,in It is a time variable. When dividing the data segment, it is divided into sub-segments, each of which represents a vibration signal in a specific state: ; in Indicates The vibration signal of a time period, and Respectively represent the start and end time of the signal segment. Associate each data segment with other operating parameters collected in real time. Operating parameters include printing speed, paper tension, and motor speed. These operating parameters directly affect the vibration characteristics during the operation of the equipment. Establishing a corresponding relationship between them and the vibration signal helps to reveal the interaction between the equipment state and the operating condition. Assume that in the In a time period, the real-time collected working condition parameters are (printing speed), (paper tension) and (motor speed), the correlation data matrix between vibration signal and working condition parameters is established by the following formula: ; Among them, the matrix Each row in represents the vibration signal of a time period and its corresponding operating parameters. The associated data matrix is ​​processed, the amplitude of the vibration signal therein is normalized, the mean and standard deviation of each measuring point are calculated, and a standardized transformation is performed to obtain standardized vibration data. The amplitude inconsistency caused by different measuring equipment or differences in the equipment's own state is eliminated, so that the vibration signals of different data segments are comparable. Based on the measuring point location and data acquisition timing, the standardized vibration data is converted into a three-dimensional data structure. The dimensions of the three-dimensional data structure correspond to the spatial dimension, the temporal dimension, and the feature dimension, respectively. Assume that there is on the device measurement points, data collection involves time periods, each with Features (such as RMS value, waveform factor, etc.), vibration feature data set Represented as a A three-dimensional matrix.

[0026] In a specific embodiment, the process of executing step 102 may specifically include the following steps: The root mean square value and kurtosis value of each measuring point data in the vibration characteristic data set are calculated, and the peak value is divided by the root mean square value to obtain the peak factor, and the root mean square value is divided by the arithmetic mean to obtain the waveform factor, and the time domain characteristic parameter group is obtained; The time domain characteristic parameter group is fitted, and the data fluctuation trend is calculated by the least square method to obtain the time domain trend characteristics. The vibration characteristic data set is converted to the frequency domain by fast Fourier transform to obtain the frequency domain basic characteristics. The basic features in the frequency domain are divided into frequency bands, 0-60 Hz is divided into a low frequency band, 60-120 Hz is divided into a medium frequency band, and 120-200 Hz is divided into a high frequency band. The energy proportion of each frequency band is calculated to obtain the frequency band energy distribution; Extract characteristic frequencies according to the frequency band energy distribution, calculate the amplitude and phase information of the main frequency components of each frequency band, obtain frequency domain characteristic parameters, and combine the time domain trend characteristics and frequency domain characteristic parameters to obtain dimension reduction features; The reduced-dimensional features are reconstructed according to the spatial positions of the measuring points, a spatial feature distribution model is established, the correlation coefficients between the measuring points are calculated, the spatial correlation features are obtained, and the spatial correlation features are standardized to obtain a multidimensional vibration feature vector.

[0027] Specifically, representative characteristic parameters are extracted from the time domain, including root mean square value, kurtosis value, peak factor and waveform factor, etc., to describe the basic characteristics of the vibration signal of the equipment during operation. , the time domain feature extraction is performed through the following steps: Calculate the root mean square value, which reflects the overall strength of the vibration signal. The formula for the root mean square value is: ; in, is the number of sampling points of the signal, It is The vibration signal value of each sampling point. The root mean square value is an important indicator to measure the vibration intensity. A high value indicates that the equipment has a large vibration and may have a fault. The kurtosis is calculated to reflect the sharpness of the signal distribution. For the kurtosis value of the time domain signal, the following formula is used to calculate: ; in, is the mean of the signal, is the standard deviation, is the sampling value of the signal. The kurtosis value can reveal whether there are spikes or pulses in the signal, and whether there are drastic vibration changes in the equipment. Calculate the peak factor and waveform factor. The calculation method of the peak factor is: ; in, It is the maximum peak value of the signal. The crest factor can reflect the height of the peak value in the signal. A higher crest factor indicates that there are sudden and large fluctuations in the vibration signal, indicating that the equipment is abnormal. The calculation of the waveform factor is: ; in, It is the arithmetic mean of the signal. The waveform factor reflects the overall shape of the signal. A lower waveform factor indicates that the signal is relatively stable, while a higher waveform factor implies that the device has certain dynamic fluctuations. By calculating the time domain characteristics, the time domain characteristic parameter group of each measuring point is obtained to form a time domain characteristic vector reflecting the device status. The time domain characteristic parameter group is fitted to extract the fluctuation trend of the signal. The least squares method is used to help the system fit the trend line of the signal and then calculate the time domain trend characteristics. The goal of the least squares method is to find a fitting line , making it consistent with the original signal The error between is minimized. The expression of the sum of squared errors is: ; By solving this equation, we get is the best fitting curve, which describes the overall fluctuation trend of the signal. The time domain trend feature is the coefficient of the trend line, which can represent the long-term change trend of the equipment vibration. The vibration feature data set is converted to the frequency domain through fast Fourier transform. Fast Fourier transform is an efficient algorithm for converting time domain signals into frequency domain representation. In the frequency domain, the vibration signal is decomposed into different frequency components, each component represents the energy of the signal at the corresponding frequency. The formula for fast Fourier transform is: ; in, is the frequency domain signal, is the time domain signal, is the frequency, is an imaginary unit, is the number of sampling points of the signal. Through the fast Fourier transform, the spectrum of the signal is obtained, revealing the energy distribution of the vibration signal at different frequencies. The basic characteristics in the frequency domain are divided into frequency bands. The frequency of the vibration signal is divided into low frequency band, medium frequency band and high frequency band. The frequency range from 0 to 200Hz is divided into three frequency bands: 0-60Hz is the low frequency band, 60-120Hz is the medium frequency band, and 120-200Hz is the high frequency band. In each frequency band, the energy proportion is calculated to evaluate the vibration intensity of the frequency band. The calculation formula for the frequency band energy proportion is: ; in, Indicates the energy proportion in a certain frequency band. and is the start and end frequency of the frequency band, is the energy of the frequency domain signal. In this process, the frequency band energy distribution provides the vibration characteristics of the equipment in each frequency band. According to the frequency band energy distribution, the characteristic frequency is extracted, and the amplitude and phase information of the main frequency component of the frequency band are calculated. The main frequency component is the most significant frequency component in the spectrum, and its amplitude and phase information provide further fault characteristics. For example, the larger the amplitude of the main frequency component, the stronger the vibration at this frequency; and the phase information can reflect the periodic characteristics of the vibration. The time domain trend characteristics and frequency domain characteristic parameters are combined to obtain the dimension reduction features. The high-dimensional feature space is compressed into a low-dimensional space through the dimension reduction algorithm (such as principal component analysis, feature selection, etc.), the most representative features are retained, the calculation burden is reduced, and the accuracy of fault diagnosis is improved. The dimension reduction features contain time domain and frequency domain information, which can comprehensively reflect the operating status of the equipment. The dimension reduction features will be reconstructed according to the spatial position of the measuring point to form a spatial feature distribution model. Through spatial reconstruction, the vibration characteristics of each measuring point are relocated in the spatial domain of the equipment to form a comprehensive spatial feature map. In order to reveal the relationship between the various components of the equipment, the correlation coefficient between each measuring point is calculated. The correlation coefficient is a measure of the strength of the linear relationship between two variables and is calculated using the Pearson correlation coefficient: ; in, and are the vibration characteristics of the two measuring points, and is their mean, is the number of measuring points. By calculating the correlation coefficients between the measuring points, the spatial correlation features are obtained, which can reveal the vibration synchronization or differences of different parts of the equipment. The spatial correlation features are standardized to eliminate the differences in vibration amplitudes between different measuring points, making the comparison of spatial features more fair and accurate. The standardized spatial correlation features constitute the final multi-dimensional vibration feature vector.

[0028] In a specific embodiment, the process of executing step 103 may specifically include the following steps: The multi-dimensional vibration eigenvector is decoupled and split according to the four spatial dimensions of the supporting structure, the fixed structure, the paper transport component and the transmission component, and the three frequency dimensions of the low frequency band, the medium frequency band and the high frequency band, and 12 sub-eigenvectors are obtained; The 12 sub-feature vectors are input into the feature mapping layer of the maximum a posteriori probability model for feature conversion to obtain 12 64-dimensional mapping feature vectors. The feature mapping layer includes a three-layer fully connected network. The 12 64-dimensional mapping feature vectors are input into the probability density calculation layer of the maximum a posteriori probability model, and the radial basis kernel function is used to calculate the conditional probability density of 8 typical fault types in the fault feature space. The kernel bandwidth parameter is set to 0.5, and a 96×8-dimensional conditional probability matrix is ​​obtained. The Bayesian inference calculation is performed on the 96×8 dimensional conditional probability matrix and the eight preset typical fault prior probability vectors. The posterior probability value of each component under each fault type is calculated using the maximum a posteriori probability criterion to obtain a 12×8 dimensional fault probability distribution. Each element in the 12×8 dimensional fault probability distribution is compared with the fault feature threshold, and all fault types exceeding the threshold and their corresponding component locations are marked to obtain a fault candidate marking matrix; All fault types in the fault candidate labeling matrix are sorted from high to low according to the posterior probability values, and the corresponding component positions and frequency band information are recorded to obtain a weighted fault type sequence; The cosine similarity calculation is performed between the weighted fault type sequence and the feature template in the historical fault database to screen out the specific fault mode with the highest matching degree and obtain the fault mode location result; The confidence score is calculated for the fault mode location result, and the reliability is quantified by the standard deviation σ of the posterior probability. When σ<0.1, it is judged as high confidence, when 0.1≤σ<0.2, it is judged as medium confidence, and when σ≥0.2, it is judged as low confidence. The preliminary fault judgment result including fault type, fault location, frequency characteristics and confidence is obtained.

[0029] Specifically, the multi-dimensional vibration characteristic vector is decoupled and split according to the structural characteristics of the equipment, and is divided into multiple sub-characteristic vectors in combination with the frequency domain information. In the vibration characteristic analysis of printing equipment, the structure of the equipment includes supporting structure, fixed structure, paper transport component and transmission component, and the vibration signal also has a frequency dimension. The vibration characteristic vector is divided into 12 sub-characteristic vectors, which correspond to four spatial dimensions (support structure, fixed structure, paper transport component, transmission component) and three frequency dimensions (low frequency band, medium frequency band, high frequency band). The vibration characteristics are split according to the structural components of the equipment where each measuring point is located and the measured frequency band information. For each measuring point , defining its spatial dimension index (one of the supporting structure, fixed structure, paper transport component or transmission component), and the frequency dimension index (belonging to the low frequency band, medium frequency band or high frequency band). Based on this information, the vibration eigenvector Split and obtain the corresponding 12 sub-feature vectors : ; in, It belongs to the spatial dimension and frequency dimension No. The vibration characteristics of each measuring point. Through decoupling, the corresponding features are extracted for each equipment component and frequency band. Each sub-feature vector is input to the feature mapping layer of the maximum a posteriori probability model for feature conversion. The maximum a posteriori probability model infers and classifies the fault type by maximizing the posterior probability criterion. In the feature mapping layer of this model, each input sub-feature vector will be nonlinearly mapped through a three-layer fully connected network to output a higher-dimensional feature vector. At this stage, each sub-feature vector Through a series of weighted and biased operations, it is mapped into a 64-dimensional feature vector , and its calculation process is expressed as: ; in, is the weight matrix, is the bias term, is an activation function (such as ReLU or sigmoid). Through this process, each sub-feature vector is converted into a 64-dimensional mapping feature vector. The 64-dimensional mapping feature vector is input into the probability density calculation layer of the maximum a posteriori probability model. This layer calculates the conditional probability density of 8 typical fault types in the fault feature space based on each feature vector. In order to calculate the conditional probability density, the radial basis kernel function is used, which measures the similarity between feature vectors and infers the probability density of each fault type based on the similarity. The calculation formula of the radial basis kernel function is: ; in, is the Euclidean distance between eigenvectors, is the kernel bandwidth parameter. In this calculation, the kernel bandwidth Set to 0.5 to control the smoothness of the kernel function and the sensitivity of the calculation. By calculating the output of the radial basis kernel function, a 96×8 dimensional conditional probability matrix is ​​obtained , where each element Indicates The eigenvector in The conditional probability density under the fault type. The Bayesian reasoning calculation is performed on the 96×8 dimensional conditional probability matrix and the preset 8 typical fault prior probability vectors. The core of Bayesian reasoning is the maximum a posteriori probability criterion, that is, the most likely fault type is determined by maximizing the a posteriori probability. The Bayesian formula is: ; in, is the posterior probability, indicating that given the vibration characteristic data In case of fault type The probability of is the conditional probability density, indicating the characteristic data in the fault type the possibility of the following; Is the fault type Through this Bayesian reasoning process, we get a 12×8 dimensional fault probability distribution matrix , where each element Indicates The feature vectors in the fault type Compare the fault probability distribution with the fault feature threshold, mark all fault types exceeding the threshold and their corresponding component locations, and obtain the fault candidate marking matrix. Each element in is compared to the threshold value. threshold, the element is marked as a possible fault type and its location. Through this step, a fault candidate marking matrix is ​​obtained, which contains the possible fault types for each component and each frequency band. The fault types in the fault candidate marking matrix are sorted according to the posterior probability values, and the probability values ​​are sorted from high to low. The corresponding component positions and frequency band information are recorded to obtain a weighted fault type sequence. Through sorting, priority is given to those fault types with higher posterior probability values ​​to improve the efficiency of fault diagnosis. The cosine similarity of the weighted fault type sequence and the feature template in the historical fault database is calculated to screen out the specific fault mode with the highest matching degree. The calculation formula for cosine similarity is: ; in, and are two eigenvectors, · represents the vector dot product, and is their norm. By calculating the cosine similarity between feature vectors, the fault mode most similar to the current fault feature is identified. According to the standard deviation of the posterior probability The confidence level of the failure mode is scored. The smaller the standard deviation, the more stable the judgment of the failure mode and the higher the confidence level. When , it is judged as high confidence; when When , it is judged as medium confidence; when Through this process, a preliminary fault judgment result including fault type, fault location, frequency characteristics and confidence is obtained.

[0030] In a specific embodiment, the execution step performs Bayesian reasoning calculation on the 96×8 dimensional conditional probability matrix and the preset 8 typical fault prior probability vectors, and calculates the posterior probability value of each component under each fault type by the maximum a posteriori probability criterion, and the process of obtaining the 12×8 dimensional fault probability distribution can specifically include the following steps: Each column of the 96×8 dimensional conditional probability matrix is ​​normalized to obtain a standardized conditional probability matrix, and data expansion is performed on 8 typical fault prior probability vectors. Each prior probability value is replicated 96 times to generate a 96×8 dimensional prior probability matrix to obtain an expanded prior probability matrix. Perform matrix dot multiplication on the standardized conditional probability matrix and the extended prior probability matrix, and sum the results by column to obtain a 96×8 dimensional joint probability matrix; The 96×8-dimensional joint probability matrix is ​​marginalized, and the marginal probability distribution is calculated by row accumulation to obtain an 8-dimensional marginal probability vector. The Bayesian posterior probability is calculated based on the 96×8-dimensional joint probability matrix and the 8-dimensional marginal probability vector. The posterior probability of each fault type is calculated using the conditional probability formula to obtain a 96×8-dimensional posterior probability matrix. The 96×8 dimensional posterior probability matrix is ​​reorganized according to the spatial dimension and frequency dimension, and the 96 measurement points are remapped to the combination structure of 4 components and 3 frequency bands to obtain a 12×8 dimensional initial fault probability distribution; The 12×8 dimensional initial fault probability distribution is smoothed, and the Gaussian kernel function is used to perform weighted averaging on the probability values ​​of adjacent measuring points to obtain a 12×8 dimensional fault probability distribution.

[0031] Specifically, each column of the 96×8 dimensional conditional probability matrix is ​​normalized. The conditional probability matrix , for each element in the matrix Normalize to ensure that the sum of each column is 1. This is achieved using the following formula: ; in, is the conditional probability matrix Middle For the first The probability of a failure type, is the normalized conditional probability value. After normalization, each column Represents the probability distribution of a certain fault type at different measurement points, and the sum of each column is 1. Expand the prior probability of 8 typical faults. Suppose there is a prior probability vector containing 8 elements ,in Representative In order to match the 96×8 conditional probability matrix, each prior probability value is replicated 96 times to obtain a 96×8 expanded prior probability matrix , the value of each column is equal to the prior probability of the corresponding fault type ,Right now: ; By extension, we get a conditional probability matrix The matching prior probability matrix . The standardized conditional probability matrix and the expanded prior probability matrix Perform a dot product operation and sum the results by column to obtain the joint probability matrix The specific operation of matrix dot multiplication is: ; The joint probability matrix Perform marginalization processing. Obtain the marginal probability vector of each fault type by row accumulation , which means that in the case of all measurement points, The total probability of failure. The marginalization formula is as follows: ; What you get is an 8-dimensional vector, representing the total probability of each fault type. The purpose of the marginalization operation is to remove the measurement point dimension and retain only the fault type dimension, providing the required prior probability for subsequent Bayesian reasoning. By obtaining the joint probability matrix and the marginal probability vector, the posterior probability of each fault type is calculated based on the Bayesian formula. The basic form of the Bayesian formula is: ; in, is the posterior probability, indicating that at a given measurement point In the case of The probability of a failure; is the conditional probability, indicating that the fault type Next, measuring point The probability of observing the value of ; is the prior probability, indicating The prior probability of the occurrence of a fault; It is the measuring point The marginal probability of all possible faults is In order to calculate the posterior probability, the joint probability matrix With the marginal probability vector Combined, Bayesian inference is applied to obtain a 96×8 dimensional posterior probability matrix , and its calculation formula is: ; Through this process, the posterior probability of each measuring point for each fault type is calculated. These posterior probabilities reflect the possibility of a certain fault type under the condition of known measuring point data and prior information. The 96×8 dimensional posterior probability matrix is ​​reorganized and mapped to a combination structure of 4 components and 3 frequency bands to obtain a 12×8 dimensional initial fault probability distribution matrix. By combining the spatial dimension and the frequency dimension, the data of each measuring point is reorganized into the corresponding components and frequency bands. The number of measuring points of the equipment is 96, which are divided into 4 components (support structure, fixed structure, paper transport component, transmission component) according to the spatial dimension and 3 frequency bands (low frequency band, medium frequency band, high frequency band) according to the frequency dimension. The posterior probability matrix is ​​reorganized into a 12×8 dimensional matrix according to the spatial position and frequency band information of the measuring point. The 12×8 dimensional initial fault probability distribution matrix is ​​smoothed to reduce the influence of noise, so that the probability values ​​of adjacent measuring points are smoother and continuous, and the stability of fault judgment is improved. The Gaussian kernel function is used to perform weighted average on the probability values ​​of adjacent measuring points. The form of the Gaussian kernel function is: ; in, and is the probability value of the two measurement points, is the standard deviation of the Gaussian kernel, which determines the degree of smoothing. By applying the Gaussian kernel function to the 12×8 dimensional initial fault probability distribution matrix, a 12×8 dimensional fault probability distribution matrix is ​​obtained.

[0032] In a specific embodiment, the process of executing step 104 may specifically include the following steps: The fault location information in the preliminary fault judgment results is combined with the structural parameters of the printing equipment, and the dynamic model of a single component is established through a spiral spring-mass system. The stiffness coefficients of the support structure, fixed structure, paper transport component and transmission component and the mass of each component are set to obtain the component dynamic parameter matrix. The dynamic parameter matrix of the components is combined and modeled, and the motion equation of the multi-degree-of-freedom system is established based on the Lagrange equation. The damping coefficient is set to 0.05 times the mass matrix and 0.02 times the stiffness matrix to obtain the system dynamic equation. The system dynamics equation is converted into a standard form to obtain n natural frequencies and corresponding vibration mode vectors, and the orthogonality of the natural frequencies and vibration mode vectors is verified to obtain a standardized modal matrix; The frequency response function is calculated based on the standardized modal matrix to obtain the frequency response function curve, and the frequency response function curve is decomposed into amplitude and phase characteristics. The transfer function of the continuous system is obtained by fitting the numerator and denominator polynomial coefficients of the transfer function through the least squares method; The continuous system transfer function is bilinearly transformed to obtain the discrete transfer function, which is then organized into a 4×4 transfer function matrix according to the transfer paths of the supporting structure, fixed structure, paper transport component, and transmission component. The vibration transfer function is obtained through parallel combination.

[0033] Specifically, based on the fault location information in the preliminary fault determination results and combined with the structural parameters of the printing equipment, a dynamic model of a single component is established. For each component, it is assumed that its dynamic behavior is described by a coil spring-mass system. The coil spring-mass system is a simplified dynamic model that can simulate the relationship between the elastic properties and mass properties of a component. In this model, the mass of the component and stiffness The basic parameters of the system are formed. The dynamic model of each component is expressed in the following form: ; in, is the quality of the component, is the stiffness, is the displacement of the component, is the external force applied to the component, is the acceleration. By substituting the stiffness coefficient and mass of the support structure, fixed structure, paper transport component and transmission component into the model, the dynamic parameter matrix of each component is obtained. The component dynamic parameter matrix is ​​combined and modeled to form a multi-degree-of-freedom system. In order to establish the motion equation of the system, it is derived based on the Lagrange equation. The Lagrange equation is a common method to describe the dynamic behavior of a multi-degree-of-freedom system. Assume that the generalized coordinates of the system are , the Lagrange equation is in the form of: ; in, is the kinetic energy of the system, is potential energy, is a generalized force, are generalized coordinates, is the generalized velocity. For a multi-degree-of-freedom system, its total kinetic energy and potential energy It is constructed by the mass, velocity and stiffness matrices of each component. Assume that the mass matrix is , the stiffness matrix is , then the system's equation of motion is written as: ; in, is the mass matrix, is the damping matrix, is the stiffness matrix, For external force, and are acceleration and velocity respectively. Assuming the damping coefficient is 0.05 times the mass matrix, and the stiffness matrix 0.02 times, that is: ; Through this assumption, the dynamic equations of the system are obtained. These equations describe the dynamic behavior of the multi-degree-of-freedom system, including the interaction between the components, external forces and damping effects. After establishing the system's motion equations, modal analysis is performed to obtain the system's natural frequency and corresponding vibration vector. By solving the system's eigenvalue equations: ; in, is the natural frequency, is the corresponding vibration mode. By solving this equation, we can get the system The orthogonality of these natural frequencies and vibration modes is verified, that is, whether their orthogonality conditions are met: ; If the verification is passed, the standardized modal matrix is ​​obtained. On this basis, the frequency response function is calculated. Frequency response function It describes the response characteristics of the system at different frequencies. It is obtained by applying a sinusoidal excitation of unit amplitude and calculating the response of the system. The amplitude and phase characteristics of the frequency response function are expressed as: ; in, and are the frequency domain representations of the input and output signals of the system, respectively. The transfer function of the continuous system is obtained by fitting the numerator and denominator polynomial coefficients of the transfer function by the least squares method. The standard form of the transfer function is: ; in, is a complex frequency variable, and is the coefficient of the transfer function. The transfer function of the continuous system is bilinearly transformed to obtain the discrete transfer function. Bilinear transformation is a commonly used method for converting from a continuous system to a discrete system. The formula is: ; in, is the sampling period, is a discretization operation. Through this transformation, the continuous transfer function is converted into a discrete transfer function, and the discrete transfer function is organized into a 4×4 transfer function matrix according to the transfer path of the support structure, fixed structure, paper transport component and transmission component. This matrix can represent the coupling relationship and vibration transmission process between various components. The vibration transfer function of the system is obtained by parallel combination. The process of parallel combination is similar to transferring vibration between multiple parallel systems. By synthesizing the transfer functions of each subsystem, the vibration transfer function of the entire system is obtained.

[0034] In a specific embodiment, the process of executing step 105 may specifically include the following steps: The vibration transfer function is analyzed in the time domain. The vibration response of the support structure, fixed structure, paper transport component and transmission component at each frequency point is calculated by convolution integral. The displacement root mean square value, velocity root mean square value and acceleration root mean square value are calculated to obtain a 16-dimensional vibration response characteristic matrix. The vibration level evaluation value of each measuring point is calculated based on the 16-dimensional vibration response characteristic matrix, and the resonance frequency characteristics in the vibration transfer function are analyzed to obtain the resonance characteristic curve; The dynamic amplification factor is calculated based on the resonance characteristic curve, and the dynamic response amplification factors of the support structure, the fixed structure, the paper transport component and the transmission component are obtained respectively; Perform dynamic balancing calculation on the dynamic response amplification factor and the vibration data of the rotor system, solve the imbalance amount and phase angle through the influence coefficient method, and obtain the dynamic balancing correction parameters; The health status is classified based on the vibration level assessment value, dynamic response amplification factor and dynamic balance correction parameter to obtain the equipment health status index, and an equipment status assessment report is generated based on the equipment health status index.

[0035] Specifically, the vibration transfer function is analyzed in the time domain. The vibration transfer function is expressed as , is the relationship between the input signal and the output signal, and is a complex multi-degree-of-freedom system function. In order to solve the time domain response, a convolution integral is performed. Assume that the input signal and the vibration transfer function It is known that the output signal It is calculated by convolution integral as follows: ; in, is the time domain response of the vibration transfer function, is the input signal. Through convolution calculation, the time domain response of the support structure, fixed structure, paper transport component and transmission component at each frequency point is obtained. Assuming that the vibration transfer function of each component is known, the vibration response at different frequency points is obtained by analyzing the response of each component. Based on the time domain response, the root mean square value (RMS) of displacement, the root mean square value of velocity (VRS) and the root mean square value of acceleration (ARS) are calculated. These values ​​are important parameters of vibration characteristics and are used to quantify the vibration intensity of the equipment. The calculation formula of the root mean square value is: ; in, is the vibration signal of a certain measuring point, is the total time length of the signal. Similarly, the calculation method of the velocity root mean square value and the acceleration root mean square value is similar to the displacement root mean square value, which are the root mean square of the velocity signal and the acceleration signal, respectively. By performing the above calculations on the vibration signals of 16 measuring points, a 16-dimensional vibration response characteristic matrix is ​​obtained, which provides vibration information of the equipment at different component positions and different frequency points. Based on the 16-dimensional vibration response characteristic matrix, the vibration level evaluation value of each measuring point is calculated. By comprehensively analyzing the vibration amplitude of each measuring point, the vibration level of each measuring point is evaluated using empirical formulas or vibration standards. For example, according to the ISO 10816 standard, different levels of vibration amplitudes correspond to different equipment health levels. The vibration level evaluation value is calculated using the following formula: ; Among them, Threshold is the standard threshold value of the vibration level. Based on these evaluation values, the resonant frequency characteristics in the vibration transfer function are analyzed to obtain the resonance characteristic curve. The resonant frequency characteristic is the frequency point where the system responds particularly strongly. When the external excitation frequency is close to the natural frequency, the system will resonate, resulting in significant amplification of the vibration. The resonant frequency is determined by the natural frequency of the system. These natural frequencies are obtained by solving the characteristic equations of the system and are usually expressed as the vibration mode of the system. According to the resonance characteristic curve, the dynamic amplification factor is calculated. The dynamic amplification factor refers to the degree of amplification of the vibration response near the resonant frequency relative to other frequency responses. Dynamic response amplification factor It is expressed as: ; in, is the frequency The amplitude of the system frequency response function under It is the maximum response value at the resonant frequency. The dynamic response magnification factor reflects the response degree of each structure (support structure, fixed structure, paper transport component, transmission component) at the resonant frequency. By calculating the dynamic response magnification factor of each frequency band, the resonance characteristics of the system at different frequencies are analyzed. For the dynamic balance problem of the system, the influence coefficient method is used to solve the imbalance amount and phase angle. Dynamic balancing analysis is used to determine the unbalanced part of the system and make corrections. The influence coefficient method calculates the imbalance amount by measuring the vibration response of the system at different points and phase angle , these values ​​are solved by the following formula: ; in, is the vibration response of each measuring point, is a system parameter related to the unbalance, is a constant. By solving these unbalanced values ​​and phase angles, dynamic balancing correction parameters are obtained, which are helpful to adjust the balance state of the system in subsequent operations and reduce unnecessary vibration. Based on the vibration level evaluation value, dynamic response amplification factor and dynamic balancing correction parameters, the health status of the equipment is classified and the equipment health status index is obtained. Health status index It is a comprehensive indicator to measure the status of the equipment, calculated by weighted average method: ; in, is the weight of each indicator, Vibration Level is the vibration level assessment value, is the dynamic response amplification factor, It is the dynamic balance correction parameter. Based on the obtained health status index, an equipment status assessment report is generated. The report includes information such as the vibration condition of the equipment, the risk of failure, and necessary maintenance measures, which helps to timely discover potential failure problems and take appropriate maintenance measures.

[0036] The above describes the inspection and testing method for maintaining the printing device in the embodiment of the present invention. The following describes the inspection and testing device for maintaining the printing device in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, an inspection and testing device for maintaining printing equipment includes: The acquisition module 201 is used to collect and pre-process vibration signals at target positions of the support structure, fixed structure, paper transport component and transmission component of the printing device to obtain a vibration feature data set; An analysis module 202 is used to perform time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; A calculation module 203 is used to input the multi-dimensional vibration feature vector into the maximum a posteriori probability model to calculate the fault feature and obtain a preliminary fault determination result; A construction module 204 is used to construct a dynamic model of the transmission system of the printing equipment according to the preliminary fault determination result, calculate the natural frequency and vibration shape characteristics of the system through modal analysis, and obtain a vibration transfer function; The evaluation module 205 is used to evaluate the status of the printing equipment based on the vibration transfer function, calculate the vibration intensity and dynamic balance index through multi-parameter collaborative analysis, and obtain an equipment status evaluation report.

[0037] Through the coordinated cooperation of the above-mentioned components, by establishing a multi-dimensional vibration feature data acquisition and preprocessing system, combining time domain analysis and frequency domain analysis methods, accurately extracting vibration feature information, using the maximum a posteriori probability model to calculate fault characteristics, and using Bayesian reasoning to accurately identify and locate fault types, the misjudgment rate of fault judgment is significantly reduced; based on dynamic modeling and modal analysis methods, a complete vibration transfer function system is established to accurately describe the vibration transfer characteristics between various components of the equipment; through multi-parameter collaborative analysis methods, a comprehensive evaluation of the vibration intensity and dynamic balance status of the equipment is achieved, a standardized status assessment system is established, and through the quantitative evaluation of the equipment health status index, accurate evaluation and early warning of the equipment status are achieved.

[0038] above Figure 2 The inspection and testing device for maintaining the printing equipment in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The inspection and testing device for maintaining the printing equipment in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0039] Figure 31 is a schematic diagram of the structure of a printing equipment maintenance inspection and testing device provided by an embodiment of the present invention. The printing equipment maintenance inspection and testing device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 may be temporary storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the printing equipment maintenance inspection and testing device 300. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the printing equipment maintenance inspection and testing device 300 to implement the steps of the above-mentioned printing equipment maintenance inspection and testing method.

[0040] The inspection and testing device 300 for printing device maintenance may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The structure of the inspection and testing equipment for printing equipment maintenance shown does not constitute a limitation on the inspection and testing equipment for printing equipment maintenance provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0041] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the inspection and testing method for maintaining the printing equipment.

[0042] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0043] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0044] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A printing equipment maintenance inspection and testing method, characterized in that: The method comprises: Collect and preprocess vibration signals at target positions of the support structure, fixed structure, paper transport component, and transmission component of the printing equipment to obtain a vibration feature data set; Performing time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; Inputting the multi-dimensional vibration feature vector into a maximum a posteriori probability model to calculate fault features and obtain a preliminary fault determination result; According to the preliminary fault determination result, a dynamic model of the transmission system of the printing equipment is constructed, and the natural frequency and vibration mode characteristics of the system are calculated by modal analysis to obtain the vibration transfer function; Based on the vibration transfer function, the state of the printing equipment is evaluated, and the vibration intensity and dynamic balance index are calculated through multi-parameter collaborative analysis to obtain an equipment state evaluation report.

2. The inspection and testing method for printing equipment maintenance according to claim 1, characterized in that: The vibration signal collection and preprocessing of the target positions of the support structure, the fixed structure, the paper transport component and the transmission component of the printing device to obtain a vibration feature data set includes: Install acceleration sensors on the bearing seats at both ends of the printing plate cylinder and the printing base gun of the printing equipment, and install displacement sensors on the pulley roller and non-stick roller support bearings; Data is collected based on the acceleration sensor and the displacement sensor to obtain an original vibration signal, and data segments are divided into segments based on the time nodes of the printing device startup, stable operation and shutdown to obtain segmented vibration data; Establishing a corresponding relationship between the segmented vibration data and the working condition parameters of the printing speed, paper tension, and motor speed collected in real time to obtain a correlation data matrix, and performing amplitude normalization processing on the vibration signal in the correlation data matrix, calculating the mean and standard deviation of each measuring point, performing a standardized transformation, and obtaining standardized vibration data; A three-dimensional data structure is established for the standardized vibration data according to the measurement point positions and the acquisition timing, so as to obtain a vibration feature data set including a spatial dimension, a temporal dimension and a feature dimension.

3. The inspection and testing method for printing equipment maintenance according to claim 2, characterized in that: The time domain analysis and frequency domain analysis of the vibration feature data set are performed to obtain a multi-dimensional vibration feature vector, including: Calculating the root mean square value and kurtosis value of each measuring point data in the vibration characteristic data set, dividing the peak value by the root mean square value to obtain the peak factor, dividing the root mean square value by the arithmetic mean to obtain the waveform factor, and obtaining a time domain characteristic parameter group; Performing data fitting on the time domain characteristic parameter group, calculating the data fluctuation trend by the least square method to obtain the time domain trend feature, and converting the vibration characteristic data set into the frequency domain by fast Fourier transform to obtain the frequency domain basic feature; The frequency domain basic features are divided into frequency bands, 0-60 Hz is divided into a low frequency band, 60-120 Hz is divided into a medium frequency band, and 120-200 Hz is divided into a high frequency band, and the energy proportion of each frequency band is calculated respectively to obtain the frequency band energy distribution; Extracting characteristic frequencies according to the frequency band energy distribution, calculating the amplitude and phase information of the main frequency component of each frequency band to obtain frequency domain characteristic parameters, and combining the time domain trend characteristics and the frequency domain characteristic parameters to obtain dimension reduction features; The dimension reduction features are reconstructed according to the spatial positions of the measuring points, a spatial feature distribution model is established, the correlation coefficients between the measuring points are calculated to obtain spatial correlation features, and the spatial correlation features are standardized to obtain a multi-dimensional vibration feature vector.

4. The inspection and testing method for printing equipment maintenance according to claim 3, characterized in that: The step of inputting the multi-dimensional vibration feature vector into the maximum a posteriori probability model to calculate the fault feature and obtain a preliminary fault determination result includes: Decouple and split the multidimensional vibration eigenvector according to four spatial dimensions of the support structure, the fixed structure, the paper transport component and the transmission component and three frequency dimensions of the low frequency band, the medium frequency band and the high frequency band to obtain 12 sub-eigenvectors; Inputting the 12 sub-feature vectors into a feature mapping layer of a maximum a posteriori probability model for feature conversion to obtain 12 64-dimensional mapping feature vectors, wherein the feature mapping layer includes a three-layer fully connected network; The 12 64-dimensional mapping feature vectors are input into the probability density calculation layer of the maximum a posteriori probability model, and the conditional probability density of 8 typical fault types in the fault feature space is calculated using a radial basis kernel function, and the kernel bandwidth parameter is set to 0.5 to obtain a 96×8-dimensional conditional probability matrix; The 96×8 dimensional conditional probability matrix and the preset 8 typical fault prior probability vectors are subjected to Bayesian reasoning calculation, and the posterior probability value of each component under each fault type is calculated by the maximum a posteriori probability criterion to obtain a 12×8 dimensional fault probability distribution; Compare each element in the 12×8 dimensional fault probability distribution with the fault feature threshold, mark all fault types exceeding the threshold and their corresponding component positions, and obtain a fault candidate marking matrix; All fault types in the fault candidate labeling matrix are sorted from high to low according to the posterior probability values, and the corresponding component positions and frequency segment information are recorded to obtain a weighted fault type sequence; Performing cosine similarity calculation on the weighted fault type sequence and the feature template in the historical fault database, screening out the specific fault mode with the highest matching degree, and obtaining the fault mode location result; The confidence score is calculated for the fault mode location result, and the reliability is quantified by the standard deviation σ of the posterior probability. When σ<0.1, it is judged as high confidence, when 0.1≤σ<0.2, it is judged as medium confidence, and when σ≥0.2, it is judged as low confidence, and a preliminary fault judgment result including fault type, fault location, frequency characteristics and confidence is obtained.

5. The inspection and testing method for printing equipment maintenance according to claim 4, characterized in that: The 96×8 dimensional conditional probability matrix and the preset 8 typical fault prior probability vectors are subjected to Bayesian reasoning calculation, and the posterior probability value of each component under each fault type is calculated by the maximum a posteriori probability criterion to obtain a 12×8 dimensional fault probability distribution, including: Each column of the 96×8 dimensional conditional probability matrix is ​​normalized to obtain a standardized conditional probability matrix, and data expansion is performed on the eight typical fault prior probability vectors, each prior probability value is replicated 96 times to generate a 96×8 dimensional prior probability matrix, and an expanded prior probability matrix is ​​obtained; Performing a matrix dot multiplication operation on the standardized conditional probability matrix and the extended prior probability matrix, and summing the results by column to obtain a 96×8 dimensional joint probability matrix; The 96×8-dimensional joint probability matrix is ​​marginalized, and the marginal probability distribution is calculated by row accumulation to obtain an 8-dimensional marginal probability vector, and Bayesian posterior probability calculation is performed based on the 96×8-dimensional joint probability matrix and the 8-dimensional marginal probability vector, and the posterior probability of each fault type is calculated by a conditional probability formula to obtain a 96×8-dimensional posterior probability matrix; The 96×8 dimensional posterior probability matrix is ​​reorganized according to the spatial dimension and the frequency dimension, and the 96 measurement points are remapped to a combination structure of 4 components and 3 frequency bands to obtain a 12×8 dimensional initial fault probability distribution; The 12×8 dimensional initial fault probability distribution is smoothed, and a Gaussian kernel function is used to perform weighted averaging on probability values ​​of adjacent measuring points to obtain a 12×8 dimensional fault probability distribution.

6. The inspection and testing method for printing equipment maintenance according to claim 5, characterized in that: According to the preliminary fault determination result, a dynamic model of the transmission system of the printing equipment is constructed, and the natural frequency and vibration mode characteristics of the system are calculated by modal analysis to obtain the vibration transfer function, including: The fault location information in the preliminary fault determination result is combined with the structural parameters of the printing device, a dynamic model of a single component is established through a coil spring-mass system, the stiffness coefficients of the support structure, the fixed structure, the paper transport component and the transmission component and the mass of each component are set, and a component dynamic parameter matrix is ​​obtained; The dynamic parameter matrix of the component is combined and modeled, and the motion equation of the multi-degree-of-freedom system is established based on the Lagrange equation, and the damping coefficient is set to 0.05 times of the mass matrix and 0.02 times of the stiffness matrix to obtain the system dynamic equation; The system dynamics equation is converted into a standard form to obtain n natural frequencies and corresponding vibration mode vectors, and the orthogonality of the natural frequencies and vibration mode vectors is verified to obtain a standardized modal matrix; A frequency response function is calculated based on the standardized modal matrix to obtain a frequency response function curve, and the frequency response function curve is decomposed into amplitude and phase characteristics, and the numerator and denominator polynomial coefficients of the transfer function are fitted by the least squares method to obtain a continuous system transfer function; The continuous system transfer function is bilinearly transformed to obtain a discrete transfer function, and the discrete transfer function is organized into a 4×4 transfer function matrix according to the transfer paths of the support structure, the fixed structure, the paper transport component and the transmission component, and the vibration transfer function is obtained through parallel combination.

7. The inspection and testing method for printing equipment maintenance according to claim 6, characterized in that: Based on the vibration transfer function, the printing equipment is evaluated, and the vibration intensity and dynamic balance index are calculated through multi-parameter collaborative analysis to obtain an equipment status evaluation report, including: Performing time domain response analysis on the vibration transfer function, calculating the vibration response of the support structure, the fixed structure, the paper transport component and the transmission component at each frequency point by convolution integral, and calculating the root mean square value of displacement, the root mean square value of velocity and the root mean square value of acceleration to obtain a 16-dimensional vibration response characteristic matrix; Calculating the vibration level evaluation value of each measuring point based on the 16-dimensional vibration response characteristic matrix, and analyzing the resonance frequency characteristics in the vibration transfer function to obtain a resonance characteristic curve; Calculate the dynamic amplification factor based on the resonance characteristic curve to obtain the dynamic response amplification coefficients of the support structure, the fixed structure, the paper transport component and the transmission component respectively; Performing dynamic balancing calculation on the dynamic response amplification factor and the vibration data of the rotor system, solving the unbalance amount and the phase angle by the influence coefficient method, and obtaining dynamic balancing correction parameters; A health status classification is performed based on the vibration level evaluation value, the dynamic response amplification factor and the dynamic balance correction parameter to obtain an equipment health status index, and an equipment status evaluation report is generated according to the equipment health status index.

8. An inspection and testing device for printing equipment maintenance, characterized in that: The device is used to perform the inspection and testing method for printing equipment maintenance according to any one of claims 1 to 7, the device comprising: The acquisition module is used to collect and pre-process the vibration signals of the target positions of the supporting structure, the fixed structure, the paper transport component and the transmission component of the printing equipment to obtain a vibration characteristic data set; An analysis module, used for performing time domain analysis and frequency domain analysis on the vibration feature data set to obtain a multi-dimensional vibration feature vector; A calculation module, used for inputting the multi-dimensional vibration feature vector into a maximum a posteriori probability model to calculate the fault feature and obtain a preliminary fault determination result; A construction module is used to construct a dynamic model of the transmission system of the printing equipment according to the preliminary fault determination result, calculate the natural frequency and vibration mode characteristics of the system through modal analysis, and obtain a vibration transfer function; The evaluation module is used to evaluate the status of the printing equipment based on the vibration transfer function, calculate the vibration intensity and dynamic balance index through multi-parameter collaborative analysis, and obtain an equipment status evaluation report.

9. An inspection and testing device for printing equipment maintenance, characterized in that: The inspection and testing device for printing device maintenance comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the inspection and testing device for printing device maintenance to execute the inspection and testing method for printing device maintenance according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the inspection and testing method for printing equipment maintenance according to any one of claims 1 to 7 is implemented.

Images