Squeeze riveter data management system and method based on artificial intelligence

Abstract

The invention discloses a squeeze riveter data management system and method based on artificial intelligence, and relates to the technical field of data analysis. The squeeze riveter is integrated with a plurality of sensors; collecting sensor data when the squeeze riveter breaks down in history, and analyzing and extracting key features; calculating a first judgment threshold according to the key features; obtaining credible parameters associated with the key features, and calculating a second credible threshold value; according to the first judgment threshold value, whether the squeeze riveter breaks down or not is judged, and fault early warning is carried out; calculating the credibility of the fault early warning, and judging whether the fault early warning is credible or not according to the second credible threshold value; and visual display is provided. According to the method, a fault early warning credibility evaluation mechanism is introduced, the reliability of fault early warning is judged according to the volatility of sensor data, and false alarm and missing alarm are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a data management system and method for a riveting machine based on artificial intelligence. Background Art

[0002] With the rapid development of intelligent manufacturing, equipment in modern production lines increasingly relies on sensor networks and data analysis technologies for real-time monitoring and fault warning. In industrial equipment such as riveting machines, the occurrence of faults can lead to decreased production efficiency, equipment damage, and even safety accidents. As a common manufacturing equipment, the operation of riveting machines involves multiple factors, such as pressure, temperature, vibration, and sound. Traditional fault diagnosis methods often rely on manual experience or rule-based preset methods, which are unable to capture subtle changes in equipment operation in real time and have poor adaptability under various working conditions. Existing technologies usually rely on a single type of sensor (such as vibration or temperature sensors), which makes it difficult to fully capture multi-dimensional state changes of equipment and is prone to missing certain potential fault signals. Many traditional fault diagnosis methods are based on fixed rules and experience and lack the ability to adapt to changes in equipment and the environment. When the operating state of the equipment changes, recalibration or manual intervention is often required, resulting in a waste of human resources. Summary of the Invention

[0003] The purpose of the present invention is to provide a data management system and method for a riveting machine based on artificial intelligence to solve the problems raised in the prior art.

[0004] To achieve the above object, the present invention provides the following technical solution: a data management method for a riveting machine based on artificial intelligence, the data management method for a riveting machine specifically comprising the following steps:

[0005] An integrated riveting machine; the riveting machine is integrated with several sensors;

[0006] Collecting sensor data from historical press riveting machine failures, analyzing the collected sensor data, and extracting key features of the press riveting machine failures;

[0007] Calculating a first judgment threshold for judging whether a malfunction occurs in the press riveting machine based on the key feature;

[0008] Obtaining historical data records of failures of the riveting machine, performing data item fitting analysis on the data records and the key features, obtaining credible parameters associated with the key features, and calculating a second credible threshold;

[0009] Monitor the operation of the riveting machine in real time, obtain sensor data, determine whether the riveting machine has failed based on the first judgment threshold, and issue a fault warning;

[0010] Calculating the credibility of the fault warning, and determining whether the fault warning is credible based on the second credibility threshold;

[0011] Provide visual display;

[0012] An integrated riveting machine, wherein the riveting machine integrates several sensors, specifically:

[0013] The sensors include visual detection sensors, pressure sensors, sound sensors, vibration sensors and temperature sensors;

[0014] The visual detection sensor is used to collect visual images of the oil, rivet head and mold of the riveting machine;

[0015] The pressure sensor is used to collect pressure change data of the riveting machine;

[0016] The sound sensor is used to collect sound data of the riveting machine bearing;

[0017] The vibration sensor is used to collect vibration data of the riveting machine motor and oil pump;

[0018] The temperature sensor is used to collect oil temperature data of the riveting machine oil pump;

[0019] Collect sensor data from historical riveting machine failures, analyze the collected sensor data, and extract key features of the riveting machine failures, specifically:

[0020] Step S3-1: collecting sensor data values when the riveting machine fails in history. The sensor data value collection format is characterized as follows: Among them, t1, t2, ..., t n represents n consecutive time points when the riveting machine fails, n represents the number of data collection times, and n is a positive integer; q 1 ,q 2 ,...,q n Indicates the corresponding time points t1, t2, ..., t n The sensor data value of the pressure riveting machine is collected at n consecutive time points when the pressure riveting machine is running normally in the history. 1 、w 2 、...、w n ]; among them, w 1 、w 2 、...、w n Representing sensor data values at n consecutive time points when the riveting machine operates normally; collecting M types of sensor data using the acquisition format;

[0021] Step S3-2, using the collected n historical sensor data values when the press riveting machine fails and the sensor data values when the press riveting machine operates normally as a basis, respectively calculating the fluctuation change rates of M types of sensor data when the press riveting machine fails and when the press riveting machine operates normally;

[0022] The characterization formula of the fluctuation change rate is: Where P represents the fluctuation rate of sensor data, i represents the data collection times label, i is a positive integer, i∈[1,n-1]; z represents the sensor data value of the riveting machine in the history, including the sensor data value when the riveting machine fails and the sensor data value when the riveting machine operates normally; n represents the data collection times, n is a positive integer;

[0023] Step S3-3, calculating the fluctuation change rates {Pf_1, Pf_2, ..., Pf_M} of the M types of sensor data when the press riveting machine fails; wherein Pf_1, Pf_2, ..., Pf_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the press riveting machine fails;

[0024] Step S3-4, calculating the fluctuation change rates {Pt_1, Pt_2, ..., Pt_M} of the M types of sensor data when the riveting machine operates normally; wherein Pt_1, Pt_2, ..., Pt_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the riveting machine operates normally;

[0025] Step S3-5: Using the normal distribution law, respectively compare the fluctuation change rate of the same sensor data when a fault occurs and the fluctuation change rate when the sensor data is operating normally, and determine the sensor data whose fluctuation change rate when a fault occurs is three standard deviations greater than the fluctuation change rate when the sensor data is operating normally as the key feature; the key feature is characterized by [Q1, Q2, ..., Q k ]; among them, Q1, Q2, ..., Q k represents the first, second, ..., kth key feature obtained, k represents the type label of the key feature, k is a positive integer, k∈[1,M);

[0026] Based on the key features, a first judgment threshold for determining whether the riveting machine has failed is calculated, specifically:

[0027] Step S4-1, obtain the key characteristic values {[Q 11 , Q 12 ,...,Q 1N ]、[Q 21 , Q 22 ,...,Q 2N ], ..., [Q k1 , Qk2 ,...,Q kN ]}; where Q kN Indicates the key characteristic value of the kth key characteristic when the pressure riveting machine fails in the history, collected for the Nth time; N represents the number of times the key characteristic value is collected, and N is a positive integer;

[0028] Step S4-2: Calculate a first judgment threshold value by using the key characteristic values obtained during the N historical times when the press riveting machine fails, wherein the first judgment threshold value is obtained based on a numerical difference between the key characteristic values when the press riveting machine fails and the key characteristic values when the press riveting machine operates normally;

[0029] Specifically:

[0030] Where r is the first judgment threshold, which represents the numerical difference between the key characteristic value when the riveting machine fails and the key characteristic value when the riveting machine operates normally; u and v represent the identifiers of the key characteristic values, u and v are positive integers, u∈[1, k], v∈[1, N]; Q uv represents the key feature value of the vth time of the uth key feature when the pressure riveting machine fails in the history; Q u ' represents the average value of the uth key feature when the riveting machine operates normally in the history;

[0031] Obtain historical data records of press riveting machine failures, perform data item fitting analysis on the data records and the key features, obtain credible parameters associated with the key features, and calculate a second credible threshold, specifically:

[0032] Step S5-1: Obtain historical data records of press riveting machine failures, obtain influencing parameters whose key characteristic values change and exceed the average change during the operation of the press riveting machine, and establish a data item fitting model of the influencing parameters and the key characteristics using the influencing parameters and the key characteristics as data items; wherein the collected influencing parameter values and the key characteristic values are kept consistent in time;

[0033] Establish different data item fitting models according to different influencing parameters;

[0034] The characterization formula for establishing the data item fitting model is:

[0035] Q′=α*X+β; where Q' represents the predicted value of the key feature; X represents the influencing parameter value; α represents the slope of the data item fitting model; β represents the intercept of the data item fitting model;

[0036] Step S5-2: fitting a model based on the data items of different influencing parameters and key features, calculating model parameters, drawing a scatter plot, and analyzing the scatter plot;

[0037] Among them, the slope and intercept of the data item fitting model are calculated, and the specific representation formula is: Wherein, α represents the slope of the data item fitting model; l represents the number of collected influencing parameter values and key characteristic values; X represents the collected influencing parameter value; represents the historical average value of the influencing parameter; Q represents the key characteristic value obtained by collection; Represents the historical average value of key characteristics;

[0038]

[0039] Calculate the parameters of the fitting model for the data items of different influencing parameters and key features in turn;

[0040] Step S5-3: fitting a model based on the data items of different influencing parameters to obtain several groups of key feature prediction values, calculating the differences between the several groups of key feature prediction values and the actual measured values, and determining the influencing parameters whose differences are greater than the average difference as credible parameters;

[0041] The model is fitted based on the data items of different influencing parameters to obtain several groups of key feature prediction values, and the difference between the several groups of key feature prediction values and subsequent actual measured values is calculated. The influencing parameters greater than the average value of the difference are determined as credible parameters. Specifically,

[0042] Where y represents the difference between the predicted value and the actual measured value of the key feature in the data item fitting model of different influencing parameters;

[0043] Among them, the credibility of the influencing parameters and key features that are less than or equal to the mean value of the differences is insufficient;

[0044] The influencing parameters greater than the average value of the difference are determined as credible parameters, and the credible parameter set is recorded as {x1, x2, ..., x g}; where x1, x2, ..., x g They represent the first, second, ..., g types of trusted parameters respectively; g represents the type label of the trusted parameter, and g is a positive integer;

[0045] The calculation of the second credible threshold is specifically as follows:

[0046] Based on the determined credible parameters, the change rate and standard deviation of historical data of different credible parameters are obtained to calculate the second credible threshold;

[0047] The calculation formula of the second credible threshold is represented as:

[0048] Where r' represents the second credible threshold; μ xrepresents the rate of change of the credible parameter; σ x represents the standard deviation of the credible parameter;

[0049] The operation process of the riveting machine is monitored in real time, sensor data is acquired, and whether the riveting machine has failed is determined based on the first judgment threshold, and a fault warning is issued, specifically:

[0050] Install fault warning device;

[0051] obtaining a key characteristic value through real-time monitoring, and issuing a fault warning through the fault warning device when a numerical fluctuation of the key characteristic value obtained through real-time monitoring is greater than the first judgment threshold;

[0052] Specifically, they are:

[0053] When|Q T -Q T -1|>r, an early warning is issued by the early warning device;

[0054] Among them, through |Q T -Q T-1 | represents the numerical fluctuation of the key eigenvalue, Q T-1 Represents Q T The key characteristic value collected at the last continuous time point, Q T Indicates the key characteristic value obtained from the current monitoring.

[0055] Calculating the credibility of the fault warning and determining whether the fault warning is credible based on the second credibility threshold is specifically as follows:

[0056] Step S8-1: After receiving the warning signal from the fault warning device, establish a data item fitting model of different influencing parameters and key features according to the method described in step S5-1, calculate the credible parameters under actual conditions based on the key features, and obtain the credible parameter values through real-time monitoring;

[0057] Step S8-2: Based on the credible parameter value obtained through real-time monitoring, when the fluctuation rate of the credible parameter value obtained through real-time monitoring is greater than the second credible threshold, the fault warning is determined to be credible, and staff are arranged to inspect and repair the riveting machine;

[0058] Specifically, they are:

[0059] when When the fault warning is judged to be credible, the staff is arranged to perform

[0060] Inspection and maintenance; where x δ-1 Represents x δ The credible parameter value collected at the last continuous time point; x δIndicates the credible parameter value obtained from current monitoring;

[0061] Set timestamp;

[0062] The timestamp is used to determine the continuity of the collected data.

[0063] The timestamp is to mark the acquired data each time the data is acquired, so as to prevent time misalignment from causing erroneous calculation of the numerical fluctuation of the key characteristic value and the fluctuation rate of the trustworthy parameter value;

[0064] An artificial intelligence-based riveting machine data management system, comprising a data acquisition module, a data transmission and analysis module, a judgment threshold calculation module, a trusted parameter analysis module, a trusted threshold calculation module, a real-time monitoring and early warning module, and a visual display module;

[0065] The data acquisition module is used to collect the operating data of the riveting machine in real time from the integrated sensor, providing raw data support for subsequent data analysis and processing;

[0066] The data transmission and analysis module is used to transmit the collected sensor data to the analysis end, perform preprocessing and feature extraction, calculate the fluctuation rate of the data, and extract key features from it;

[0067] The judgment threshold calculation module is used to calculate the first judgment threshold for judging the occurrence of a fault based on the analysis results of the key feature data in the history, and to set a reasonable threshold based on the difference between the fault data and the normal operation data to support real-time fault monitoring;

[0068] The trusted parameter analysis module is used to establish a data item fitting model based on historical data and key features, and to evaluate and identify trusted parameters related to the key features;

[0069] The credible threshold calculation module is used to calculate the second credible threshold according to the change rate and standard deviation of the credible parameter, providing a basis for the credibility evaluation of the fault warning;

[0070] The real-time monitoring and early warning module is used to monitor the operating status of the riveting machine in real time. When the real-time data exceeds the judgment threshold, a fault early warning is triggered. The reliability of the early warning is judged based on the credibility assessment result, and the staff is reminded to check in time.

[0071] The visual display module is used to display real-time data, fault warnings and credibility assessment results through a visual interface, helping operators to intuitively understand equipment status, fault risks and maintenance requirements.

[0072] Compared with the existing technology, the beneficial effects of the present invention are as follows: this method effectively solves the problems of single sensor dependence, lack of adaptive learning and dynamic adjustment capabilities, and low credibility of the early warning system in the existing technology through multi-sensor data fusion and artificial intelligence algorithms. First, by integrating multiple sensors such as vision, pressure, sound, vibration and temperature, it can comprehensively capture the operating status of the riveting machine, avoiding the limitation that a single sensor is difficult to reflect complex faults, and improving the accuracy of fault diagnosis; secondly, by utilizing machine learning and adaptive algorithms, this method can dynamically adjust the fault diagnosis model according to historical data and real-time monitoring results, overcoming the shortcomings of traditional methods that rely on fixed rules and manual intervention, and ensuring high efficiency and stability in long-term operation; finally, by introducing a credibility evaluation mechanism for fault warning, it can accurately judge the reliability of fault warning according to the volatility of sensor data and changes in historical data, reduce false alarms and missed alarms, and improve the trust and response speed of the system; these beneficial effects not only improve the fault diagnosis accuracy of the riveting machine, but also significantly enhance the operating safety and production efficiency of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a schematic diagram of the steps of a riveting machine data management method based on artificial intelligence of the present invention;

[0074] Figure 2 This is a schematic diagram of a data item fitting model of Example 2 of an artificial intelligence-based riveting machine data management system and method of the present invention;

[0075] Figure 3 The figure is a structural diagram of an artificial intelligence-based riveting machine data management system of the present invention. DETAILED DESCRIPTION

[0076] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0077] Example: Figure 1-Figure 3 As shown, the present invention provides a technical solution, a riveting machine data management method based on artificial intelligence, and the riveting machine data management method specifically includes the following steps:

[0078] An integrated riveting machine; the riveting machine is integrated with several sensors;

[0079] Collecting sensor data from historical press riveting machine failures, analyzing the collected sensor data, and extracting key features of the press riveting machine failures;

[0080] Calculating a first judgment threshold for judging whether a malfunction occurs in the press riveting machine based on the key feature;

[0081] Obtaining historical data records of failures of the riveting machine, performing data item fitting analysis on the data records and the key features, obtaining credible parameters associated with the key features, and calculating a second credible threshold;

[0082] Monitor the operation of the riveting machine in real time, obtain sensor data, determine whether the riveting machine has failed based on the first judgment threshold, and issue a fault warning;

[0083] Calculating the credibility of the fault warning, and determining whether the fault warning is credible based on the second credibility threshold;

[0084] Provide visual display;

[0085] An integrated riveting machine, wherein the riveting machine integrates several sensors, specifically:

[0086] The sensors include visual detection sensors, pressure sensors, sound sensors, vibration sensors and temperature sensors;

[0087] The visual detection sensor is used to collect visual images of the oil, rivet head and mold of the riveting machine;

[0088] The pressure sensor is used to collect pressure change data of the riveting machine;

[0089] The sound sensor is used to collect sound data of the riveting machine bearing;

[0090] The vibration sensor is used to collect vibration data of the riveting machine motor and oil pump;

[0091] The temperature sensor is used to collect oil temperature data of the riveting machine oil pump;

[0092] Collect sensor data from historical riveting machine failures, analyze the collected sensor data, and extract key features of the riveting machine failures, specifically:

[0093] Step S3-1: collecting sensor data values when the riveting machine fails in history. The sensor data value collection format is characterized as follows: Among them, t1, t2, ..., t n represents n consecutive time points when the riveting machine fails, n represents the number of data collection times, and n is a positive integer; q 1 ,q 2 ,...,q n Indicates the corresponding time points t1, t2, ..., t nThe sensor data value of the pressure riveting machine is collected at n consecutive time points when the pressure riveting machine is running normally in the history. 1 、w 2 、...、w n ]; among them, w 1 、w 2 、...、w n Representing sensor data values at n consecutive time points when the riveting machine operates normally; collecting M types of sensor data using the acquisition format;

[0094] Step S3-2, using the collected n historical sensor data values when the press riveting machine fails and the sensor data values when the press riveting machine operates normally as a basis, respectively calculating the fluctuation change rates of M types of sensor data when the press riveting machine fails and when the press riveting machine operates normally;

[0095] The characterization formula of the fluctuation change rate is: Where P represents the fluctuation rate of sensor data, i represents the data collection times label, i is a positive integer, i∈[1,n-1]; z represents the sensor data value of the riveting machine in the history, including the sensor data value when the riveting machine fails and the sensor data value when the riveting machine operates normally; n represents the data collection times, n is a positive integer;

[0096] Step S3-3, calculating the fluctuation change rates {Pf_1, Pf_2, ..., Pf_M} of the M types of sensor data when the press riveting machine fails; wherein Pf_1, Pf_2, ..., Pf_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the press riveting machine fails;

[0097] Step S3-4, calculating the fluctuation change rates {Pt_1, Pt_2, ..., Pt_M} of the M types of sensor data when the riveting machine operates normally; wherein Pt_1, Pt_2, ..., Pt_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the riveting machine operates normally;

[0098] Step S3-5: Using the normal distribution law, respectively compare the fluctuation change rate of the same sensor data when a fault occurs and the fluctuation change rate when the sensor data is operating normally, and determine the sensor data whose fluctuation change rate when a fault occurs is three standard deviations greater than the fluctuation change rate when the sensor data is operating normally as the key feature; the key feature is characterized by [Q1, Q2, ..., Q k ]; among them, Q1, Q2, ..., Q k represents the first, second, ..., kth key feature obtained, k represents the type label of the key feature, k is a positive integer, k∈[1,M);

[0099] In Example 1, a certain model A riveting machine is equipped with three sensors (M=3): a temperature sensor, a pressure sensor, and a vibration sensor (the values are rounded);

[0100] Fault data (n=5 consecutive time points):

[0101] Temperature (℃): q(1) 1 =100,q(1) 2 =105,q(1) 3 =110,q(1) 4 =115,q(1) 5 =120;

[0102] Pressure (MPa): q(2) 1 =50,q(2) 2 =55,q(2) 3 =60,q(2) 4 =65,q(2) 5 =70;

[0103] Vibration approx. (mm / s 2 ):q(3) 1 =2,q(3) 2 =4,q(3) 3 =6,q(3) 4 =8,q(3) 5 =10; Normal data (n=5 consecutive time points):

[0104] Temperature (°C): w(1) = 90, 91, 90, 92, 91;

[0105] Pressure (MPa): w(2) = 50, 51, 50, 52, 51;

[0106] Vibration approx. (mm / s 2 ): w(3)=1, 1, 2, 1, 2;

[0107] Calculate the volatility rate of change:

[0108] Pf_2=5; Pf_3=2;

[0109] Pt_1=1.25; Pt_2=1.25; Pt_3=0.75;

[0110] Temperature sensor:

[0111] Normal volatility mean μ 温度 =1.25;

[0112] Standard deviation σ 温度 =0.2;

[0113] The three-standard-deviation threshold is 1.25 + 3 × 0.2 = 1.85;

[0114] Temperature fluctuation rate Pf during fault 温度 =5>1.85, temperature is selected as the key feature.

[0115] Pressure sensor:

[0116] Normal volatility mean μ 压力 =1.25,

[0117] Standard deviation σ 压力 =0.1;

[0118] The three standard deviation threshold is 1.25 + 3 × 0.1 = 1.55;

[0119] Pressure fluctuation rate Pf during failure 压力 =5>1.55, pressure is selected as the key feature.

[0120] Vibration sensor:

[0121] Normal volatility mean μ 振动 =0.75,

[0122] Standard deviation σ vibration = 0.05;

[0123] The threshold of three times the standard deviation is 0.75 + 3 × 0.05 = 0.9;

[0124] Vibration fluctuation rate Pf during fault 振动 =2>0.9, vibration is selected as the key feature.

[0125] Based on the key features, a first judgment threshold for determining whether the riveting machine has failed is calculated, specifically:

[0126] Step S4-1, obtain the key characteristic values {[Q 11 , Q 12 ,...,Q 1N ]、[Q 21 , Q 22 ,...,Q 2N ], ..., [Q k1 , Q k2 ,...,Q kN ]}; where Q kN Indicates the key characteristic value of the kth key characteristic when the pressure riveting machine fails in the history, collected for the Nth time; N represents the number of times the key characteristic value is collected, and N is a positive integer;

[0127] Step S4-2: Calculate a first judgment threshold value by using the key characteristic values obtained during the N historical times when the press riveting machine fails, wherein the first judgment threshold value is obtained based on a numerical difference between the key characteristic values when the press riveting machine fails and the key characteristic values when the press riveting machine operates normally;

[0128] Specifically:

[0129] Where r is the first judgment threshold, which represents the numerical difference between the key characteristic value when the riveting machine fails and the key characteristic value when the riveting machine operates normally; u and v represent the identifiers of the key characteristic values, u and v are positive integers, u∈[1, k], v∈[1, N]; Q uv represents the key feature value of the vth time of the uth key feature when the pressure riveting machine fails in the history; Q u ' represents the average value of the uth key feature when the riveting machine operates normally in the history;

[0130] Obtain historical data records of press riveting machine failures, perform data item fitting analysis on the data records and the key features, obtain credible parameters associated with the key features, and calculate a second credible threshold, specifically:

[0131] Step S5-1: Obtain historical data records of press riveting machine failures, obtain influencing parameters whose key characteristic values change and exceed the average change during the operation of the press riveting machine, and establish a data item fitting model of the influencing parameters and the key characteristics using the influencing parameters and the key characteristics as data items; wherein the collected influencing parameter values and the key characteristic values are kept consistent in time;

[0132] Establish different data item fitting models according to different influencing parameters;

[0133] The characterization formula for establishing the data item fitting model is:

[0134] Q′=α*X+β; where Q' represents the predicted value of the key feature; X represents the influencing parameter value; α represents the slope of the data item fitting model; β represents the intercept of the data item fitting model;

[0135] Step S5-2: fitting a model based on the data items of different influencing parameters and key features, calculating model parameters, drawing a scatter plot, and analyzing the scatter plot;

[0136] Among them, the slope and intercept of the data item fitting model are calculated, and the specific representation formula is: Wherein, α represents the slope of the data item fitting model; l represents the number of collected influencing parameter values and key characteristic values; X represents the collected influencing parameter value; represents the historical average value of the influencing parameter; Q represents the key characteristic value obtained by collection; Represents the historical average value of key characteristics;

[0137]

[0138] Calculate the parameters of the fitting model for the data items of different influencing parameters and key features in turn;

[0139] Step S5-3: fitting a model based on the data items of different influencing parameters to obtain several groups of key feature prediction values, calculating the differences between the several groups of key feature prediction values and the actual measured values, and determining the influencing parameters whose differences are greater than the average difference as credible parameters;

[0140] The model is fitted based on the data items of different influencing parameters to obtain several groups of key feature prediction values, and the difference between the several groups of key feature prediction values and subsequent actual measured values is calculated. The influencing parameters greater than the average value of the difference are determined as credible parameters. Specifically,

[0141] Where y represents the difference between the predicted value and the actual measured value of the key feature in the data item fitting model of different influencing parameters;

[0142] Among them, the credibility of the influencing parameters and key features that are less than or equal to the mean value of the differences is insufficient;

[0143] The influencing parameters greater than the average value of the difference are determined as credible parameters, and the credible parameter set is recorded as {x1, x2, ..., x g}; where x1, x2, ..., x g They represent the first, second, ..., g types of trusted parameters respectively; g represents the type label of the trusted parameter, and g is a positive integer;

[0144] Example 2: Factory B recorded the key characteristic values of 5 failures (N=5):

[0145] Temperature sensor (℃): Q 11 =100, Q 12 =105,Q 13 =110,Q 14 =115,Q 15 =120;

[0146] Pressure sensor (MPa): Q 21 =50, Q 22 =55,Q 23 =60, Q 24 =65,Q 25 =70;

[0147] Vibration sensor (mm / s 2 ):Q 31 =2,Q 32 =4,Q 33 =3,Q 34 =3.5, Q 35 =2.5;

[0148] Average values of key characteristics during normal operation:

[0149] Temperature: Q1′=90;

[0150] Pressure: Q2′=50;

[0151] Vibration: Q3′=1;

[0152] Calculate the first judgment threshold:

[0153] Temperature difference:

[0154] Pressure difference:

[0155] Vibration Difference:

[0156]

[0157] Obtain the influencing parameters of key characteristic values that change and exceed the average change during the operation of the riveting machine. Factory B found that the following influencing parameters are related to the key characteristics:

[0158] The ambient temperature affects the temperature sensor of the riveting machine;

[0159] The hydraulic oil pressure affects the pressure sensor of the riveting machine;

[0160] The degree of mechanical wear affects the vibration sensor of the riveting machine;

[0161] like Figure 2 As shown, taking the degree of mechanical wear as an example:

[0162] Five sets of data were collected:

[0163] Mechanical wear degree (X): 0.1, 0.2, 0.3, 0.4, 0.5;

[0164] Vibration sensor (Q): 2.5, 4.2, 5.8, 8.3, 9.7;

[0165]

[0166]

[0167] X1=0.1; Q1=2.5; Q1′=0.0537× 0.1+6.0839≈6.089;

[0168] X2=0.2; Q2=4.2; Q2′≈6.094;

[0169] X3=0.3; Q3=5.8; Q3′≈6.100;

[0170] X4=0.4; Q4=8.3; Q4′≈6.105;

[0171] X5=0.5; Q5=9.7; Q5′≈6.111;

[0172] Calculate the difference:

[0173]

[0174] The average value of the difference is yavg = 2.0;

[0175] Then y=2.313>2.0; the degree of mechanical wear (X) is determined to be a credible parameter.

[0176] The calculation of the second credible threshold is specifically as follows:

[0177] Based on the determined credible parameters, the change rate and standard deviation of historical data of different credible parameters are obtained to calculate the second credible threshold;

[0178] The calculation formula of the second credible threshold is represented as:

[0179] Where r' represents the second credible threshold; μx represents the rate of change of the credible parameter; σ x represents the standard deviation of the credible parameter;

[0180] The operation process of the riveting machine is monitored in real time, sensor data is acquired, and whether the riveting machine has failed is determined based on the first judgment threshold, and a fault warning is issued, specifically:

[0181] Install fault warning device;

[0182] obtaining a key characteristic value through real-time monitoring, and issuing a fault warning through the fault warning device when a numerical fluctuation of the key characteristic value obtained through real-time monitoring is greater than the first judgment threshold;

[0183] Specifically, they are:

[0184] When|Q T -Q T-1 When |>r, an early warning is issued by the early warning device;

[0185] Among them, through |Q T -Q T-1 | represents the numerical fluctuation of the key eigenvalue, Q T-1 Represents Q T The key characteristic value collected at the last continuous time point, Q T Indicates the key characteristic value obtained from the current monitoring.

[0186] Calculating the credibility of the fault warning and determining whether the fault warning is credible based on the second credibility threshold is specifically as follows:

[0187] Step S8-1: After receiving the warning signal from the fault warning device, establish a data item fitting model of different influencing parameters and key features according to the method described in step S5-1, calculate the credible parameters under actual conditions based on the key features, and obtain the credible parameter values through real-time monitoring;

[0188] Step S8-2: Based on the credible parameter value obtained through real-time monitoring, when the fluctuation rate of the credible parameter value obtained through real-time monitoring is greater than the second credible threshold, the fault warning is determined to be credible, and staff are arranged to inspect and repair the riveting machine;

[0189] Specifically, they are:

[0190] when When the fault warning is judged to be credible, staff are arranged to inspect and repair the riveting machine; Among them, x δ-1 Represents x δ The credible parameter value collected at the last continuous time point; x δ Indicates the credible parameter value obtained from current monitoring;

[0191] Set timestamp;

[0192] The timestamp is used to determine the continuity of the collected data.

[0193] The timestamp is to mark the acquired data each time the data is acquired, so as to prevent time misalignment from causing erroneous calculation of the numerical fluctuation of the key characteristic value and the fluctuation rate of the trustworthy parameter value;

[0194] like Figure 3 As shown, a data management system for a riveting machine based on artificial intelligence includes a data acquisition module, a data transmission and analysis module, a judgment threshold calculation module, a trusted parameter analysis module, a trusted threshold calculation module, a real-time monitoring and early warning module, and a visual display module;

[0195] The data acquisition module is used to collect the operating data of the riveting machine in real time from the integrated sensor, providing raw data support for subsequent data analysis and processing;

[0196] The data transmission and analysis module is used to transmit the collected sensor data to the analysis end, perform preprocessing and feature extraction, calculate the fluctuation rate of the data, and extract key features from it;

[0197] The judgment threshold calculation module is used to calculate the first judgment threshold for judging the occurrence of a fault based on the analysis results of the key feature data in the history, and to set a reasonable threshold based on the difference between the fault data and the normal operation data to support real-time fault monitoring;

[0198] The trusted parameter analysis module is used to establish a data item fitting model based on historical data and key features, and to evaluate and identify trusted parameters related to the key features;

[0199] The credible threshold calculation module is used to calculate the second credible threshold according to the change rate and standard deviation of the credible parameter, providing a basis for the credibility evaluation of the fault warning;

[0200] The real-time monitoring and early warning module is used to monitor the operating status of the riveting machine in real time. When the real-time data exceeds the judgment threshold, a fault early warning is triggered. The reliability of the early warning is judged based on the credibility assessment result, and the staff is reminded to check in time.

[0201] The visual display module is used to display real-time data, fault warnings and credibility assessment results through a visual interface, helping operators to intuitively understand equipment status, fault risks and maintenance requirements.

[0202] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A data management method for a riveting machine based on artificial intelligence, characterized by: The data management method of the riveting machine specifically comprises the following steps: An integrated riveting machine; the riveting machine is integrated with several sensors; Collecting sensor data from historical press riveting machine failures, analyzing the collected sensor data, and extracting key features of the press riveting machine failures; Calculating a first judgment threshold for judging whether a malfunction occurs in the press riveting machine based on the key feature; Obtaining historical data records of failures of the riveting machine, performing data item fitting analysis on the data records and the key features, obtaining credible parameters associated with the key features, and calculating a second credible threshold; Monitor the operation of the riveting machine in real time, obtain sensor data, determine whether the riveting machine has failed based on the first judgment threshold, and issue a fault warning; Calculating the credibility of the fault warning, and determining whether the fault warning is credible based on the second credibility threshold; Provides visual display.

2. The artificial intelligence-based riveting machine data management method according to claim 1, characterized in that: An integrated riveting machine, wherein the riveting machine integrates several sensors, specifically: The sensors include visual detection sensors, pressure sensors, sound sensors, vibration sensors and temperature sensors; The visual detection sensor is used to collect visual images of the oil, rivet head and mold of the riveting machine; The pressure sensor is used to collect pressure change data of the riveting machine; The sound sensor is used to collect sound data of the riveting machine bearing; The vibration sensor is used to collect vibration data of the riveting machine motor and oil pump; The temperature sensor is used to collect oil temperature data of the riveting machine oil pump.

3. The artificial intelligence-based riveting machine data management method according to claim 2, characterized in that: Collect sensor data from historical riveting machine failures, analyze the collected sensor data, and extract key features of the riveting machine failures, specifically: Step S3-1: collecting sensor data values when the riveting machine fails in history. The sensor data value collection format is characterized as follows: Where t1, t2, ..., tn represent n consecutive time points when the riveting machine fails, n represents the number of data collection times, and n is a positive integer; q 1 ,q 2 ,...,q n Indicates the corresponding time points t1, t2, ..., t n The sensor data value of the pressure riveting machine is collected at n consecutive time points when the pressure riveting machine is running normally in the history. 1 、w 2 、...、w n ]; among them, w 1 、w 2 、...、w n Representing sensor data values at n consecutive time points when the riveting machine operates normally; collecting M types of sensor data using the acquisition format; Step S3-2, using the collected n historical sensor data values when the press riveting machine fails and the sensor data values when the press riveting machine operates normally as a basis, respectively calculating the fluctuation change rates of M types of sensor data when the press riveting machine fails and when the press riveting machine operates normally; Step S3-3, calculating the fluctuation change rates {Pf_1, Pf_2, ..., Pf_M} of the M types of sensor data when the press riveting machine fails; wherein Pf_1, Pf_2, ..., Pf_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the press riveting machine fails; Step S3-4, calculating the fluctuation change rates {Pt_1, Pt_2, ..., Pt_M} of the M types of sensor data when the riveting machine operates normally; wherein Pt_1, Pt_2, ..., Pt_M represent the fluctuation change rates of the 1st, 2nd, ..., Mth types of sensor data when the riveting machine operates normally; Step S3-5: Using the normal distribution law, respectively compare the fluctuation change rate of the same sensor data when a fault occurs and the fluctuation change rate when the sensor data is operating normally, and determine the sensor data whose fluctuation change rate when a fault occurs is three standard deviations greater than the fluctuation change rate when the sensor data is operating normally as the key feature; the key feature is characterized by [Q1, Q2, ..., Q k ]; among them, Q1, Q2, ..., Q k Represents the first, second, ..., kth key features obtained, k represents the type label of the key feature, k is a positive integer, k∈[1,M).

4. The artificial intelligence-based riveting machine data management method according to claim 3, characterized in that: Based on the key features, a first judgment threshold for determining whether the riveting machine has failed is calculated, specifically: Step S4-1, obtain the key characteristic values {[Q 11 , Q 12 ,...,Q 1N ]、[Q 21 , Q 22 ,...,Q 2N ], ..., [Q k1 , Q k2 ,...,Q kN ]}; where Q kN Indicates the key characteristic value of the kth key characteristic when the pressure riveting machine fails in the history, collected for the Nth time; N represents the number of times the key characteristic value is collected, and N is a positive integer; Step S4-2: Calculate a first judgment threshold by using the key characteristic values obtained during the N historical times when the press riveting machine fails, wherein the first judgment threshold is obtained based on the numerical difference between the key characteristic values when the press riveting machine fails and the key characteristic values when the press riveting machine operates normally.

5. The artificial intelligence-based riveting machine data management method according to claim 4, characterized in that: Obtain historical data records of press riveting machine failures, perform data item fitting analysis on the data records and the key features, obtain credible parameters associated with the key features, and calculate a second credible threshold, specifically: Step S5-1: Obtain historical data records of press riveting machine failures, obtain influencing parameters whose key characteristic values change and exceed the average change during the operation of the press riveting machine, and establish a data item fitting model of the influencing parameters and the key characteristics using the influencing parameters and the key characteristics as data items; wherein the collected influencing parameter values and the key characteristic values are kept consistent in time; Step S5-2: fitting a model based on the data items of different influencing parameters and key features, calculating model parameters, drawing a scatter plot, and analyzing the scatter plot; Step S5-3: Fit the model according to the data items of different influencing parameters to obtain several groups of key feature prediction values, calculate the differences between the several groups of key feature prediction values and the actual measured values, and determine the influencing parameters whose differences are greater than the average difference as credible parameters.

6. The artificial intelligence-based riveting machine data management method according to claim 5, characterized in that: The calculation of the second credible threshold is specifically as follows: Based on the determined credible parameters, the historical data change rates and standard deviations of different credible parameters are obtained to calculate the second credible threshold.

7. The artificial intelligence-based riveting machine data management method according to claim 6, characterized in that: The operation process of the riveting machine is monitored in real time, sensor data is acquired, and whether the riveting machine has failed is determined based on the first judgment threshold, and a fault warning is issued, specifically: Install fault warning device; A key characteristic value is obtained through real-time monitoring. When the value fluctuation of the key characteristic value obtained through real-time monitoring is greater than the first judgment threshold, a fault warning is issued through the fault warning device.

8. The artificial intelligence-based riveting machine data management method according to claim 7, characterized in that: Calculating the credibility of the fault warning and determining whether the fault warning is credible based on the second credibility threshold is specifically as follows: Step S8-1: After receiving the warning signal from the fault warning device, establish a data item fitting model of different influencing parameters and key features according to the method described in step S5-1, calculate the credible parameters under actual conditions based on the key features, and obtain the credible parameter values through real-time monitoring; Step S8-2: Based on the credible parameter value obtained through real-time monitoring, when the fluctuation rate of the credible parameter value obtained through real-time monitoring is greater than the second credible threshold, the fault warning is judged to be credible, and staff are arranged to inspect and repair the riveting machine.

9. The artificial intelligence-based riveting machine data management method according to claim 1, characterized in that: Specifically: Set timestamp; The timestamp is used to determine the continuity of the collected data.

10. An artificial intelligence-based riveting machine data management system, applying the artificial intelligence-based riveting machine data management method according to any one of claims 1 to 9, characterized in that: The data management system for the riveting machine includes a data acquisition module, a data transmission and analysis module, a judgment threshold calculation module, a trustworthy parameter analysis module, a trustworthy threshold calculation module, a real-time monitoring and early warning module, and a visual display module; The data acquisition module is used to collect the operating data of the riveting machine in real time from the integrated sensor, providing raw data support for subsequent data analysis and processing; The data transmission and analysis module is used to transmit the collected sensor data to the analysis end, perform preprocessing and feature extraction, calculate the fluctuation rate of the data, and extract key features from it; The judgment threshold calculation module is used to calculate the first judgment threshold for judging the occurrence of a fault based on the analysis results of the key feature data in the history, and to set a reasonable threshold based on the difference between the fault data and the normal operation data to support real-time fault monitoring; The trusted parameter analysis module is used to establish a data item fitting model based on historical data and key features, and to evaluate and identify trusted parameters related to the key features; The credible threshold calculation module is used to calculate the second credible threshold according to the change rate and standard deviation of the credible parameter, providing a basis for the credibility evaluation of the fault warning; The real-time monitoring and early warning module is used to monitor the operating status of the riveting machine in real time. When the real-time data exceeds the judgment threshold, a fault early warning is triggered. The reliability of the early warning is judged based on the credibility assessment result, and the staff is reminded to check in time. The visual display module is used to display real-time data, fault warnings and credibility assessment results through a visual interface, helping operators to intuitively understand equipment status, fault risks and maintenance requirements.

Images