Packaging machinery health state evaluation method and system based on multi-source data fusion

Abstract

The invention discloses a packaging machinery health state evaluation method and system based on multi-source data fusion, and relates to the technical field of multi-source data fusion, and the method comprises the steps: carrying out the fault root cause positioning according to a feature vector obtained through the decomposition of a main feature value, and obtaining a main evaluation dimension which causes the reduction of a comprehensive health degree index; according to the method, the evaluation dimension mapped by the component with the maximum absolute value in the feature vector corresponding to the main feature value is analyzed, accurate positioning of the fault root cause is achieved, the process effectively strips coupling correlation among dimension data on the basis of the multivariate statistical principle, the problems of misjudgment and missed judgment caused by judgment depending on experience of operation and maintenance personnel in the traditional technology are solved, and the accuracy of the fault root cause is improved. When the comprehensive health degree index of the equipment is reduced, the problem can be quickly locked from electrical, mechanical, process or environment dimensions, the troubleshooting time is greatly shortened, the maintenance response efficiency is improved, and a clear direction is provided for targeted equipment maintenance.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-source data fusion, and particularly relates to a packaging machine health state evaluation method and system based on multi-source data fusion. BACKGROUND

[0002] With the transformation of the packaging industry to high speed and automation, packaging machines as the core equipment of production lines in the fields of food, medicine, daily chemical products, etc., their continuous and stable operation directly determines the production efficiency, product quality and cost. The equipment is operated under high load for a long time, and is affected by environmental dust, temperature fluctuation and process switching, which is prone to problems such as electrical aging, mechanical wear and tear, process parameter drift, etc. Sudden failure not only causes production line downtime loss, but also may cause product recall risk. Therefore, real-time and accurate evaluation and early warning of the health state of the packaging machine has become a key requirement for enterprises to ensure production continuity and reduce operation and maintenance costs.

[0003] At present, the health evaluation of packaging machines still mainly relies on traditional monitoring and experience judgment. Although intelligent means have been introduced in some fields, there are obvious limitations: traditional methods mostly rely on single-dimensional data, such as monitoring only mechanical vibration or motor current, and a few combine process parameters for simple analysis, without forming a multi-dimensional collaborative system; in recent years, the development of industrial Internet of Things has promoted the attempt of multi-source data fusion, but most of them only stop at data splicing and independent analysis, without mining the coupling correlation of each dimension, such as mechanical abnormalities leading to process fluctuations, and the evaluation mostly uses fixed thresholds, lacking comprehensive quantification of overall health state, making it difficult to shift from “post-fault maintenance” to “pre-fault warning”, and the fault root cause positioning also mostly relies on manual experience, which is low in efficiency.

[0004] There are still many key problems in current technology applications: first, the depth of data fusion is insufficient, or the evaluation is one-sided due to the neglect of data correlation, or it is simply spliced by weighting, which cannot extract deep health information; second, there is a lack of comprehensive health quantification ability, mostly with binary output of normal or abnormal, which is difficult to predict the degradation trend and has poor timeliness of early warning; third, the accuracy of fault root cause positioning is low, manual experience is easily affected by subjectivity, and multi-parameter fluctuation is prone to misjudgment, prolonging the maintenance time; fourth, the model adaptability is poor, the dimension weight and warning threshold are mostly fixed values, which cannot be dynamically optimized according to the running period or working condition of the equipment, the precision decreases after long-term use, and it is difficult to meet the fine operation and maintenance requirements.

[0005] Therefore, it is necessary to invent a packaging machine health state evaluation method and system based on multi-source data fusion to solve the above problems. SUMMARY

[0006] The purpose of the present application is to provide a packaging machine health state evaluation method and system based on multi-source data fusion to solve the problems raised in the background.

[0007] In order to achieve the above object, the present application provides the following technical scheme: a packaging machine health state evaluation method based on multi-source data fusion, comprising the following steps: S1, collecting the running data of the packaging machine in multiple preset dimensions to obtain at least two monitoring index data in each dimension, and constructing the corresponding feature vector of each dimension; S2, combining the feature vector corresponding to each dimension into a health state matrix, and performing normalization processing to obtain a normalized matrix; S3, assigning weights to each dimension, constructing a weight diagonal matrix, and multiplying the normalized matrix by the weight diagonal matrix to obtain a weighted state matrix; S4, calculating the covariance matrix of the weighted state matrix, performing eigenvalue decomposition to obtain the principal eigenvalue, and then calculating the comprehensive health degree index according to the principal eigenvalue; S5, determining the equipment health state according to the comprehensive health degree index, and issuing a warning when the comprehensive health degree index is lower than the set threshold; S6, performing fault root cause positioning according to the feature vector obtained by the principal eigenvalue decomposition to obtain the main evaluation dimension that causes the comprehensive health degree index to decrease; S7, according to the comparison result of the main evaluation dimension and the maintenance feedback data, the weight diagonal matrix and the threshold are adaptively optimized.

[0008] Preferably, the plurality of preset dimensions are four evaluation dimensions, including an electrical dimension, a mechanical dimension, a process dimension and an environmental dimension.

[0009] Preferably, the monitoring index data of the electrical dimension includes motor current harmonic distortion rate THD and drive power factor PF, and the corresponding feature vector V elec =[THD, PF]; The monitoring index data of the mechanical dimension includes the effective value V rms of the driving end bearing vibration speed and the kurtosis value K urt of the gearbox vibration signal, and the corresponding feature vector V mech =[V rms , K urt ]; The monitoring index data of the process dimension includes the unit time qualified product rate P rat and the sealing temperature fluctuation standard deviation T std , and the corresponding feature vector V proc =[P rat , T std ]; The monitoring index data of the environmental dimension includes the spindle bearing temperature T bea and the environmental dust concentration D ust , and the corresponding feature vector V env= [T bea , D ust ].

[0010] Preferably, the normalization processing is implemented by using a membership function to map the elements of the health state matrix to the interval [0, 1]; wherein the lower the numerical value represents the better the monitoring index data of the equipment state, and the descending half trapezoidal membership function is used, and the higher the numerical value represents the better the monitoring index data of the equipment state, and the ascending half trapezoidal membership function is used.

[0011] Preferably, the calculation formula of the comprehensive health index is: HI = 100 x (1 - λ1 / λ ideal ), wherein HI is the comprehensive health index, λ1 is the current principal eigenvalue, λ ideal is the maximum principal eigenvalue of the equipment in the brand-new ideal state.

[0012] Preferably, the determination of the equipment health state is specifically: when the comprehensive health index HI is greater than or equal to a, it is determined as a healthy state; when the comprehensive health index b is less than or equal to HI and less than a, it is determined as a sub-healthy state; when the comprehensive health index c is less than or equal to HI and less than b, it is determined as a pre-warning state; when the comprehensive health index HI is less than c, it is determined as an alarm state.

[0013] Preferably, the fault root cause positioning is realized by analyzing the dimension corresponding to the component with the largest absolute value in the characteristic vector corresponding to the principal eigenvalue.

[0014] Preferably, the specific process of the adaptive optimization is: S71, record the fault reason R of the industrial equipment after maintenance, the monitoring index data before and after maintenance, and the change of the comprehensive health index HI; S72, perform consistency comparison on the main evaluation dimension and the evaluation dimension corresponding to the fault reason R, and count the number of positioning deviations; S73, when the number of positioning deviations exceeds a preset threshold, adjust the diagonal line elements of the weight diagonal matrix by using a gradient descent algorithm; S74, dynamically update the threshold in step S5 based on the normal fluctuation range of HI of the industrial equipment under different operation cycles or different environments.

[0015] The packaging machinery health state evaluation system based on multi-source data fusion comprises the following modules: A data acquisition and feature construction module is used to acquire the operation data of the packaging machinery in multiple preset dimensions, obtain at least two monitoring index data in each dimension, and construct the characteristic vector corresponding to each dimension. a data normalization processing module configured to combine the feature vectors of each dimension into a health state matrix and perform normalization processing to obtain a normalized matrix; a weight allocation and weighted calculation module configured to allocate weights to each dimension, construct a weight diagonal matrix, and multiply the normalized matrix by the weight diagonal matrix to obtain a weighted state matrix; a comprehensive health index calculation module configured to calculate a covariance matrix of the weighted state matrix, perform eigenvalue decomposition to obtain a principal eigenvalue, and calculate a comprehensive health index according to the principal eigenvalue; a health state assessment and early warning module configured to determine the health state of the equipment according to the comprehensive health index and issue a warning when the comprehensive health index is lower than a set threshold; a fault root cause positioning module configured to perform fault root cause positioning according to the eigenvectors obtained by the principal eigenvalue decomposition to obtain the main evaluation dimensions that cause the comprehensive health index to decrease; an adaptive optimization module configured to perform adaptive optimization on the weight diagonal matrix and the threshold according to the comparison result of the main evaluation dimensions and the maintenance feedback data.

[0016] Technical effects and advantages of the present application: 1. The present application collects running data of four preset dimensions of packaging machinery, i.e., electrical, mechanical, process, and environment, constructs a feature vector containing at least two monitoring indexes for each dimension, and combines the feature vectors of each dimension into a health state matrix for subsequent analysis, breaking the limitation of traditional evaluation methods that rely only on single dimension data, comprehensively covering key influencing factors of equipment operation, making the health state assessment more in line with the actual operation scene of the equipment, effectively avoiding the one-sidedness of the evaluation caused by the lack of data dimensions, and laying a data foundation for accurate evaluation of the overall health state of the equipment. 2. The present application normalizes the health state matrix by using a membership function, adopts a descending half trapezoidal membership function for monitoring indexes with lower values representing better equipment states, and adopts an ascending half trapezoidal membership function for monitoring indexes with higher values representing better equipment states, thereby uniformly mapping monitoring data of different dimensions and different evaluation standards to the [0, 1] interval, completely eliminating the interference of data scale differences on the evaluation results, providing standardized data support for subsequent weight allocation and weighted calculation of each dimension, and significantly improving the accuracy and reliability of the evaluation results. 3、The application realizes accurate positioning of fault root cause by analyzing the evaluation dimension mapped by the component with the largest absolute value in the feature vector corresponding to the principal eigenvalue, which effectively strips the coupling relationship between the dimensional data based on the principle of multivariate statistics, avoids the misjudgment and missed judgment caused by the experience of operation and maintenance personnel, and quickly locuses the problem source in the electrical, mechanical, process or environmental dimension when the comprehensive health index of the equipment decreases, greatly shortens the fault troubleshooting time, improves the maintenance response efficiency, and provides a clear direction for targeted equipment maintenance; 4、The application compares the main evaluation dimension obtained by positioning the fault root cause with the real evaluation dimension corresponding to the fault cause through recording the fault cause after equipment maintenance, the monitoring index data before and after maintenance and the change of the comprehensive health index, adjusts the weight diagonal matrix by using the gradient descent algorithm when the positioning deviation frequency exceeds the preset threshold, and dynamically updates the early warning threshold based on the normal fluctuation range of HI under different operation cycles or different environments, so that the evaluation model can be continuously optimized with the change of the equipment operation state and the adjustment of the working condition, the long-term evaluation precision decline problem caused by fixed parameters is effectively avoided, and the adaptability of the model in complex production scenes is significantly enhanced; 5、The application has the whole process design of data acquisition and feature construction, data normalization processing, weight distribution and weighted calculation, comprehensive health index calculation, health state evaluation and early warning, fault root cause positioning and self-adaptive optimization, and constructs the equipment health management closed loop of "monitoring-evaluation-early warning-positioning-optimization", which can not only early warn the potential fault of the equipment and avoid the production line downtime loss and product quality risk caused by sudden fault, but also reduce the problem of excessive maintenance or insufficient maintenance through accurate root cause positioning, further reduce the operation and maintenance cost of enterprises combined with model self-adaptive optimization, and finally guarantee the continuous and stable operation of the packaging production line and the consistency of product quality. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a method step flowchart of the application.

[0018] Figure 2 It is a system architecture schematic diagram of the application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0020] The application provides a method for evaluating the health state of a packaging production line, comprising the following steps: Figure 1The packaging machine health state evaluation method based on multi-source data fusion shown comprises the following steps: S1, collecting running data of the packaging machine in multiple preset dimensions to obtain at least two monitoring index data in each dimension and constructing a corresponding feature vector in each dimension; It should be known that the running data includes electrical data, mechanical data, process data and environmental data; The collection of electrical data is realized by real-time collection of current and voltage waveform signals of the motor running through the power quality analyzer or the smart meter installed in the power supply circuit of the main drive motor of the equipment; The collection of mechanical data is realized by collecting mechanical vibration signals through the vibration acceleration sensor installed on the drive end bearing seat and the gearbox housing, and obtaining the vibration time domain signal after analog-digital conversion by the data acquisition card; The collection of process data is realized by collecting temperature signals through the temperature sensor installed on the sealing device; the qualified product rate data is obtained by the visual detection system or the product counter at the end of the production line, and is read through the interface of the equipment monitoring system (such as the SCADA system); The collection of environmental data is realized by collecting temperature signals through the temperature sensor installed near the measuring point of the main shaft bearing; and collecting environmental dust data through the dust concentration sensor installed in the key area (such as near the electrical control cabinet) of the equipment.

[0021] Further, in the above technical solution, the multiple preset dimensions are four evaluation dimensions, including electrical dimension, mechanical dimension, process dimension and environmental dimension.

[0022] Further, in the above technical solution, the monitoring index of the electrical dimension includes motor current harmonic distortion rate THD and drive power factor PF, and the corresponding feature vector V elec =[THD, PF]; The monitoring index data of the mechanical dimension includes drive end bearing vibration speed effective value V rms and gearbox vibration signal kurtosis value K urt , and the corresponding feature vector V mech =[V rms , K urt ]; The monitoring index data of the process dimension includes unit time qualified product rate P rat and sealing temperature fluctuation standard deviation T std , and the corresponding feature vector V proc =[P rat , T std ]; The monitoring index data of the environmental dimension includes main shaft bearing temperature T bea and environmental dust concentration D ust , and the corresponding feature vector Venv = THD bea , D ust ].

[0023] It is to be known that the motor current harmonic distortion rate THD is directly output by the power quality analyzer after Fourier transform analysis of the collected current waveform, or calculated by the industrial computer according to the collected waveform data, for quantifying the distortion degree of the motor input current waveform relative to the standard sine wave, and an abnormal increase in the value often indicates that the motor or the driver has a potential fault or abnormal load; The driving power factor PF is directly measured and output by the power quality analyzer or the smart meter, reflecting the utilization efficiency of electric energy, and a too low power factor indicates that there is reactive power loss, which may be related to motor aging, load mismatch or power supply quality problems; The driving end bearing vibration speed effective value V rms , which is obtained after integration and root mean square calculation of the time domain signal collected by the vibration acceleration sensor, is used to evaluate the overall vibration intensity of the bearing, and is a commonly used parameter for judging bearing wear, looseness or imbalance and the like.

[0024] The gearbox vibration signal kurtosis value K urt , which is obtained after high-order statistics (fourth moment) calculation of the time domain signal collected by the vibration acceleration sensor, is very sensitive to the impact component in the signal, and an increase in the value usually indicates that the gearbox gears, bearings and other components have early pitting, cracking and other local damage; The unit time qualified product rate P rat , which is calculated by the total output and the qualified output in a production line monitoring system statistical period, has a calculation formula of P rat = qualified output / total output, and directly reflects the production efficiency and output quality of the equipment, and is a comprehensive index for measuring the process performance of the equipment; The sealing temperature fluctuation standard deviation T std , which is obtained by calculating the standard deviation of the continuous temperature values collected by the temperature sensor in a monitoring period, is used to quantify the stability of temperature control in the sealing process, and a too large fluctuation directly affects the sealing quality, which may be caused by heating element aging or temperature control system failure; The main shaft bearing temperature T bea , which is directly measured by a PT100 thermal resistance or thermocouple temperature sensor installed on the bearing seat, monitors the bearing working temperature, and an abnormal temperature rise is an important sign of poor lubrication, aggravated wear or overload; The environmental dust concentration D ust , which is measured and output by a laser dust sensor or a particle counter, quantifies the dust level of the environment around the equipment, and a too high dust value will aggravate mechanical wear, affect heat dissipation and even cause electrical failure.

[0025] S2, combine the feature vectors of each dimension into a health state matrix and perform normalization processing to obtain a normalized matrix; It should be known that the feature vectors of each evaluation dimension are taken as a row of a matrix, combined into a multi-row and two-column matrix in a preset order, referred to as a health state matrix X. For example, after the data collected at time t is constructed, the feature vectors of four dimensions are respectively: The electrical dimension feature vector: V elec (t) = [THD(t), PF(t)]; The mechanical dimension feature vector: V mech (t) = [V rms (t), K urt (t)]; The process dimension feature vector: V proc (t) = [P rat (t), T std (t)]; The environmental dimension feature vector: V env (t) = [T bea (t), D ust (t)]; The health state matrix X(t) at this moment is constructed as follows: , The matrix X(t) is a 4x2 matrix, each row of which represents an evaluation dimension, and each column represents the data sequence of a specific monitoring indicator at different times.

[0026] Further, in the above technical solution, the normalization processing is realized by using a membership function to map the elements of the health state matrix to the interval [0, 1]; wherein, the monitoring indicator data of the lower numerical value representing the better equipment state adopts a descending half trapezoidal membership function, and the monitoring indicator data of the higher numerical value representing the better equipment state adopts an ascending half trapezoidal membership function.

[0027] It should be known that the type of membership function of each monitoring indicator is determined as follows: For the electrical dimension: the motor current harmonic distortion rate THD adopts a descending half trapezoidal membership function; the drive power factor PF adopts an ascending half trapezoidal membership function.

[0028] For the mechanical dimension: the drive end bearing vibration speed effective value V rms adopts a descending half trapezoidal membership function; the gearbox vibration signal kurtosis value K urt adopts a descending half trapezoidal membership function.

[0029] For the process dimension: the unit time qualified product rate P ratThe ascending half trapezoidal membership function is adopted; the standard deviation of the sealing temperature fluctuation is T std The descending half trapezoidal membership function is adopted.

[0030] For the environmental dimension: the main shaft bearing temperature T bea The descending half trapezoidal membership function is adopted; the environmental dust concentration D ust The descending half trapezoidal membership function is adopted.

[0031] The descending half trapezoidal membership function is defined as follows: assuming that the actual value of the monitoring index data is x, and the optimal value range of the health state is [m, n], the normalized value μ(x) is calculated as follows: , The ascending half trapezoidal membership function is defined as follows: assuming that the actual value of the monitoring index data is y, and the optimal value range of the health state is [g, h], the normalized value μ(y) is calculated as follows: , Wherein, m, n, g and h are set according to the historical data of the equipment or the experience of experts.

[0032] S3, assigning weights to each dimension, constructing a weight diagonal matrix, and multiplying the normalized matrix and the weight diagonal matrix to obtain a weighted state matrix; It should be known that the value of the weight is determined based on the historical failure data of the equipment, the experience of experts in the field or the analytic hierarchy process, and the sum of all weights is 1.

[0033] S4, calculating the covariance matrix of the weighted state matrix, and performing eigenvalue decomposition to obtain the principal eigenvalue, and then calculating the comprehensive health degree index according to the principal eigenvalue; It should be known that the weighted state matrix is denoted as X weighted ; the covariance matrix is denoted as C, and the calculation formula of the covariance matrix of the weighted state matrix is as follows: , Wherein, N is the sample number, i.e. the number of rows of the matrix, which is 4 in this example, T represents the transpose operation of the matrix, and the calculated covariance matrix C is a 2x2 matrix.

[0034] The eigenvalue decomposition satisfies the following relationship: , wherein λ i is the eigenvalue of the covariance matrix C, v i is the corresponding eigenvector, the solved eigenvalues are arranged in descending order and denoted as λ1, λ2, λ3, λ4, wherein λ1 is the largest eigenvalue, called the principal eigenvalue, which represents the variance of the data in the main change direction and best reflects the overall change trend of the health state of the equipment.

[0035] Further, in the above technical solution, the calculation formula of the comprehensive health degree index is: HI=100×(1-λ1 / λ ideal ), Wherein, HI is the comprehensive health degree index, λ1 is the current main eigenvalue, λ ideal is the maximum main eigenvalue of the equipment in the brand-new ideal state.

[0036] It should be known that the maximum main eigenvalue λ ideal of the equipment in the brand-new ideal state is determined in the following way: in the initial stage of equipment installation and commissioning, continuously collect multiple sets of normal operation data, calculate and record multiple sets of main eigenvalues according to steps S1 to S4, and take the maximum value as λ ideal ; or directly use the ideal value calibrated when the equipment is factory acceptance tested.

[0037] The meaning of the calculation formula of the comprehensive health degree index is that: in the brand-new state, the main eigenvalue λ1 of the equipment should be close to λ ideal , at this time HI≈100, representing excellent health degree. With the degradation of the performance of the equipment, the main eigenvalue λ1 will gradually increase, resulting in the decrease of the value of HI; when λ1=λ ideal , HI=0, representing extremely poor health state.

[0038] S5, determining the health state of the equipment according to the comprehensive health degree index, and issuing a warning when the comprehensive health degree index is lower than the set threshold value; Further, in the above technical solution, the determination of the health state of the equipment is specifically: When the comprehensive health degree index HI≥a, it is determined that the health state is good; When the comprehensive health degree index b≤HI When the comprehensive health degree index c≤HI When the comprehensive health degree index HI

[0039] It should be known that the thresholds a, b and c are determined according to the type of the equipment, the working condition and the historical operation data, and exemplarily a=85, b=75 and c=60 can be taken; The health state indicates that the running indicators of the equipment in each dimension are good, and the overall performance is at an ideal level, so the equipment can be safely put into production and only needs to be regularly maintained and monitored; The sub-health state indicates that the equipment has a slight performance degradation or individual indicators deviate slightly from the optimal interval, which does not affect the current normal operation, but the monitoring frequency needs to be strengthened, the change trend needs to be concerned, and preventive maintenance can be prepared; The pre-warning state indicates that the performance of the equipment has obviously degraded, there is a high risk of failure, and the system should immediately issue a pre-warning signal, such as highlighting a yellow alarm on the monitoring interface, sending an in-station message or an email notification, prompting maintenance personnel to intervene, and arranging for inspection and maintenance as soon as possible in combination with the root cause positioning result of step S6 to prevent the state from further deteriorating; The alarm state indicates that the health condition of the equipment has seriously deteriorated, and the failure may occur at any moment or has already occurred. The system must issue the highest level of alarm signal, such as triggering an audible and light alarm, sending an emergency message, etc., to require maintenance personnel to immediately stop and check, and to perform emergency repair or failure handling to ensure the safety of the equipment and production.

[0040] S6, performing root cause positioning according to the eigenvectors obtained from the principal eigenvalue decomposition to obtain the main evaluation dimension causing the decline of the comprehensive health degree index; It should be noted that the core of this step S6 is to utilize the mathematical properties of principal component analysis to accurately locate the main failure root dimension causing the decline of the comprehensive health degree index from the data-driven perspective, thereby realizing the tracing from the overall state evaluation to the specific problem dimension and providing decision support for subsequent targeted maintenance.

[0041] Further, in the above technical solution, the root cause positioning is achieved by analyzing the dimension corresponding to the component with the largest absolute value in the eigenvector corresponding to the principal eigenvalue. The specific operation is as follows: S61, extracting the eigenvector v1 corresponding to the principal eigenvalue λ1; S62, calculating the absolute value of each component in the eigenvector v1; S63, finding the component with the largest absolute value; S64, determining the serial number of the component in the eigenvector, which is mapped back to the original evaluation dimension order, for example, serial number 1 corresponds to the electrical dimension, serial number 2 corresponds to the mechanical dimension, and so on, thereby identifying the main evaluation dimension causing the deterioration of the health state.

[0042] It should be noted that this step is based on the principle of multivariate statistics and can effectively decouple the coupling relationship between multiple dimension parameters, avoiding misjudgment due to parameter correlation, thereby more accurately pointing to the core problem dimension. For example, if the component corresponding to the mechanical dimension in the eigenvector v1 has the largest absolute value, it indicates that the degradation of the current equipment health state is mainly due to the abnormality of the mechanical components, such as bearing wear or gear box failure, thereby guiding the maintenance personnel to preferentially check the mechanical system.

[0043] S7, adaptively optimizing the weight diagonal matrix and the threshold value according to the comparison result of the main evaluation dimension and the maintenance feedback data.

[0044] Further, in the above technical solution, the specific process of the adaptive optimization is: S71, record the failure cause R after the maintenance of the industrial equipment, the monitoring index data before and after the maintenance, and the change of the comprehensive health index HI; It should be noted that the present step aims to establish a maintenance case library for model optimization. The recording operation is automatically completed by the system, and when the monitoring system detects that the equipment has experienced a maintenance event, such as the closure of a maintenance work order or the change of the equipment state from "stop" to "run", the data archiving program is automatically triggered; the failure cause R is selected from predefined options or manually input by the maintenance personnel in the maintenance work order, and the failure cause is mapped to the corresponding evaluation dimension, such as "bearing wear" mapped to the mechanical dimension and "driver failure" mapped to the electrical dimension; the monitoring index data before and after the maintenance refers to the health state matrix X before collected at the time of failure or the last time before maintenance after ; the change of the comprehensive health index HI is the record of HI before and HI after . All these data will form a complete maintenance case and be stored in the historical database.

[0045] S72, compare the main evaluation dimension with the evaluation dimension corresponding to the failure cause R for consistency, and count the number of positioning deviations; It should be noted that the consistency comparison is specifically: comparing the main evaluation dimension located in step S6 with the real evaluation dimension actually corresponding to the failure cause R in the maintenance record, if they are consistent, it is counted as one correct positioning; if they are not consistent, it is counted as one positioning deviation, and the system maintains a positioning deviation counter for each dimension, for example, Count elec , Count mech , Count proc , Count env Each time a positioning deviation occurs, the counter of the corresponding dimension is incremented by 1. The number of positioning deviations refers to the current value of reading these counters.

[0046] S73, when the number of positioning deviations exceeds a preset threshold, adjust the diagonal elements of the weight diagonal matrix using the gradient descent algorithm; It is to be understood that the preset threshold is an empirical value, for example, it can be set to 5 times, when the positioning deviation times of a certain dimension exceeds the threshold, it indicates that the weight currently assigned to the dimension may be unreasonable, leading to frequent misjudgment of the system, thereby triggering the optimization program. The core idea of adjusting the diagonal elements of the weight diagonal matrix by using the gradient descent algorithm is to construct a loss function to punish the cases where the "fault root cause positioning dimension" is inconsistent with the "true fault dimension" in the historical fault cases, and to optimize the weight diagonal matrix by minimizing the loss function. The loss function is constructed as follows: for each case k in the historical database where positioning deviation occurs, the true fault dimension of the case is obtained, denoted as jtrue(k), for example, when k=1, jtrue(1) is the electrical dimension, and the main evaluation dimension positioned in step S6 is obtained, denoted as jpred(k), when jtrue(k) is equal to jpred(k), the loss L k is 0, indicating correct positioning, no loss; when jtrue(k) is not equal to jpred(k), the loss This loss term means that when the weight w jtrue(k) of the true fault dimension is less than the weight w jpred(k) of the main evaluation dimension positioned in step S6, a large loss value will be generated, and the optimization process will minimize this loss by increasing w jtrue(k) and / or decreasing w jpred(k) , so as to encourage the next time to encounter similar situations to prefer to identify j true(k) dimension as the principal component. Finally, the losses of all historical deviation cases are summed to obtain the total loss L=ΣL k .

[0047] After adjusting the diagonal elements of the weight diagonal matrix by the gradient descent algorithm, the corresponding positioning deviation counter is cleared and starts counting again.

[0048] S74, based on the normal fluctuation range of HI of the industrial equipment under different operating periods or different environments, dynamically updating the threshold in step S5.

[0049] It is to be understood that the different operating cycles refer to the overhaul cycle, wear cycle, etc. of the equipment; the different environments refer to different seasons, production of different products, etc. The normal HI baseline value of the equipment is different in the early, middle and late stages of its life. For example, the HI value of a new equipment can be stable at above 95 for a long time, while the HI value of an old equipment fluctuates between 80-85 when it is in good condition. This step realizes the dynamic updating of the threshold value through statistical analysis of historical data. The system retrieves the HI values in all health states in the past period (e.g. half a year) and calculates the distribution (e.g. mean μ and standard deviation σ) of the HI values. Then the threshold values in step S5 are reset according to the distribution, for example: The threshold value a can be updated as μ-k a ×σ; The threshold value b can be updated as a-Δ1; The threshold value c can be updated as b-Δ2; wherein k a is a multiple coefficient set according to experience, the value range is between 1-2, Δ1 and Δ2 are preset fixed interval values; The packaging machinery health state evaluation system based on multi-source data fusion, as shown in Figure 2 includes the following modules: A data acquisition and feature construction module is used to acquire the running data of the packaging machinery in multiple preset dimensions, obtain at least two monitoring index data in each dimension, and construct a feature vector corresponding to each dimension; A data normalization processing module is used to combine the feature vector corresponding to each dimension into a health state matrix and perform normalization processing to obtain a normalized matrix; A weight allocation and weighted calculation module is used to allocate weights to each dimension, construct a weight diagonal matrix, and multiply the normalized matrix by the weight diagonal matrix to obtain a weighted state matrix; A comprehensive health degree index calculation module is used to calculate the covariance matrix of the weighted state matrix, perform eigenvalue decomposition to obtain the principal eigenvalues, and then calculate the comprehensive health degree index according to the principal eigenvalues; A health state evaluation and early warning module is used to determine the health state of the equipment according to the comprehensive health degree index, and issue a warning when the comprehensive health degree index is below a set threshold value; A fault root cause positioning module is used to perform fault root cause positioning according to the feature vectors obtained by the principal eigenvalue decomposition to obtain the main evaluation dimensions that cause the decrease of the comprehensive health degree index; An adaptive optimization module is used to compare the main evaluation dimensions with the maintenance feedback data to adaptively optimize the weight diagonal matrix and the threshold value.

[0050] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art will appreciate that the technical solutions described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalent features, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for assessing the health status of packaging machinery based on multi-source data fusion, characterized in that, It includes the following steps: S1. Collect the operation data of the packaging machine in multiple preset dimensions, obtain at least two monitoring index data for each dimension, and construct the feature vector corresponding to each dimension; S2. Combine the feature vectors corresponding to each dimension into a health status matrix, and perform normalization processing to obtain a normalized matrix; S3. Assign weights to each dimension, construct a weight diagonal matrix, and multiply the normalized matrix by the weight diagonal matrix to obtain a weighted status matrix; S4. Calculate the covariance matrix of the weighted status matrix, perform eigenvalue decomposition to obtain the main eigenvalues, and then calculate the comprehensive health index according to the main eigenvalues; S5. Determine the health status of the equipment according to the comprehensive health index, and issue a warning when the comprehensive health index is lower than the set threshold; S6. Locate the root cause of the failure according to the eigenvector obtained by decomposing the main eigenvalue, and obtain the main evaluation dimension that causes the decline of the comprehensive health index; S7. Adaptively optimize the weight diagonal matrix and the threshold according to the comparison result between the main evaluation dimension and the maintenance feedback data.

2. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The multiple preset dimensions are four evaluation dimensions, including the electrical dimension, the mechanical dimension, the process dimension and the environmental dimension.

3. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 2, characterized in that, The electrical dimension monitoring data includes motor current harmonic distortion rate (THD) and drive power factor (PF), corresponding to the feature vector V. elec =[THD,PF]; The monitoring data for the mechanical dimension includes the effective value of the vibration velocity V of the drive-end bearing. rms and gearbox vibration signal kurtosis value K urt The corresponding feature vector V mech =[V rms K urt ]; The monitoring indicators for the process dimension include the yield rate P per unit time. rat and the standard deviation of sealing temperature fluctuation T std The corresponding feature vector V proc =[P rat T std ]; The environmental monitoring data includes the spindle bearing temperature T. bea and environmental dust concentration D ust The corresponding feature vector V env =[T bea D ust ].

4. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The normalization processing is realized by using a membership function, and the elements of the health status matrix are mapped to the interval [0, 1]; among them, for the monitoring index data with lower values representing better equipment status, a descending half-trapezoidal membership function is used, and for the monitoring index data with higher values representing better equipment status, an ascending half-trapezoidal membership function is used.

5. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The calculation formula of the comprehensive health index is: HI=100×(1-λ1 / λ ideal ) Where HI is the comprehensive health index, λ1 is the current main feature value, and λ ideal This represents the maximum principal characteristic value of the device under ideal conditions.

6. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The determination of the health status of the equipment is specifically: When the comprehensive health index HI≥a, it is determined to be in a healthy state; When the comprehensive health index b≤HI<a, it is determined to be in a sub-healthy state; When the comprehensive health index c≤HI<b, it is determined to be in a warning state; When the comprehensive health index HI<c, it is determined to be in an alarm state.

7. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The root cause of the failure location is realized by analyzing the dimension corresponding to the component with the largest absolute value in the eigenvector corresponding to the main eigenvalue.

8. The method for assessing the health status of packaging machinery based on multi-source data fusion according to claim 1, characterized in that, The specific process of the adaptive optimization is: S71. Record the failure cause R after the industrial equipment is repaired, the monitoring index data before and after the repair, and the change of the comprehensive health index HI; S72. Compare the main evaluation dimension with the evaluation dimension corresponding to the failure cause R for consistency, and count the number of positioning deviation times; S73. When the number of positioning deviation times exceeds the preset threshold, use the gradient descent algorithm to adjust the diagonal elements of the weight diagonal matrix; S74. Dynamically update the threshold in step S5 based on the normal fluctuation range of HI in different operation cycles or different environments of the industrial equipment.

9. A packaging machinery health status assessment system based on multi-source data fusion is applied to the packaging machinery health status assessment method based on multi-source data fusion as described in any one of claims 1-8, characterized in that, It includes the following modules: A data acquisition and feature construction module, which is used to collect the operation data of the packaging machine in multiple preset dimensions, obtain at least two monitoring index data for each dimension, and construct the feature vector corresponding to each dimension; A data normalization processing module, which is used to combine the feature vectors corresponding to each dimension into a health status matrix and perform normalization processing to obtain a normalized matrix; The weight allocation and weighted calculation module is used to allocate weights to each dimension, construct a weight diagonal matrix, and multiply the normalized matrix with the weight diagonal matrix to obtain a weighted state matrix; The comprehensive health index calculation module is used to calculate the covariance matrix of the weighted state matrix, perform eigenvalue decomposition to obtain the principal eigenvalues, and then calculate the comprehensive health index based on the principal eigenvalues. The health status assessment and early warning module is used to determine the health status of the equipment based on the comprehensive health index and to issue an early warning when the comprehensive health index is lower than a set threshold. The fault root cause localization module is used to locate the fault root cause based on the feature vector obtained by the decomposition of the main feature value, and to obtain the main evaluation dimensions that lead to the decline of the comprehensive health index. The adaptive optimization module is used to adaptively optimize the weight diagonal matrix and threshold based on the comparison results between the main evaluation dimensions and maintenance feedback data.

Images