Operating state monitoring method and system of connector production machine

Abstract

The invention relates to the field of data processing, in particular to an operation state monitoring method and system for a connector production machine, and the method comprises the steps: obtaining operation state data of the connector production machine, carrying out the preprocessing, taking preset local reference data as a benchmark, determining the relative change degree of data of each dimension, and calculating the abnormal fluctuation degree. And meanwhile, performing first-order difference symbolization processing on the local reference data, and counting symbol continuous lengths to form a sequence so as to analyze the abnormal tendency. And combining the abnormal fluctuation degree and the tendency to construct a fault risk index, and calculating a local abnormal weighting factor at each moment. And finally, based on the factor, improving an LOF anomaly detection algorithm and calculating and correcting an anomaly score, thereby effectively monitoring the operation state of the connector production machine. According to the method, the LOF algorithm is improved, the defects of a traditional algorithm in time sequence and multi-type data anomaly detection are overcome, the monitoring accuracy and the fault early warning capacity are improved, and the production efficiency and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing. In particular, it relates to a method and system for monitoring the operating state of a connector production machine. Background Art

[0002] A connector production machine is an automated device used for mass-producing electronic or electrical connectors. It produces precision connectors through processes such as stamping, injection molding, and assembly, and is applied in fields such as consumer electronics and automotive electronics. To ensure product quality, prevent equipment damage, improve production efficiency, and meet industry regulatory requirements, etc., it is necessary to monitor its operating state in order to detect abnormalities in a timely manner.

[0003] Monitoring the operating state of a connector production machine is important in many aspects. First, it ensures product quality. Real-time monitoring can promptly detect equipment abnormalities, avoid the production of defective products caused by failures, and ensure the stable and reliable quality of connectors. Second, it improves production efficiency. Early warning of potential equipment failures reduces sudden downtime and enables the production process to proceed continuously and stably. Third, it reduces maintenance costs. Implementing preventive maintenance extends the service life of the equipment. Fourth, it optimizes the production process. By analyzing the operating data, production bottlenecks can be identified, providing a basis for adjusting production plans and improving processes.

[0004] In the process of monitoring the operating state of a connector production machine, current traditional methods such as Although anomaly detection algorithms can, without assuming the data distribution, identify the anomaly degree between data and adjacent data based on the density difference of operating state data at different times, and are applicable to irregular data, they cannot identify anomalies in the time series of data because they ignore the time dimension of the data. In addition, there are various types of data anomalies, including statistical anomalies, context anomalies, collective anomalies, etc. Traditional monitoring methods are difficult to comprehensively and accurately identify and distinguish these different types of anomalies, thus affecting the accurate judgment of the equipment operating state and the effective early warning of faults. Summary of the Invention

[0005] To solve the problem that algorithms are difficult to identify time series anomalies, and in the face of multiple types of data anomalies, traditional monitoring methods are difficult to comprehensively and accurately distinguish and identify, thus affecting the accurate judgment of the equipment operating state and fault early warning, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a method for monitoring the operating state of a connector production machine includes: obtaining state data related to the operation of the connector production machine and preprocessing the state data; for the preprocessed state data, taking the preset local reference data as a benchmark, comparing the fluctuation differences between the data of each dimension and the data of adjacent dimensions, determining the relative change degree of the data of each dimension, and calculating the entropy of the relative change degree of the data of all dimensions to obtain the abnormal fluctuation degree of the data of each dimension; after performing a first-order difference and symbolization on the local reference data, counting the symbol continuous length sequence formed by the continuous lengths of the same symbols and analyzing the abnormal trend; constructing a fault risk index based on the abnormal fluctuation degree and the abnormal trend, constructing a fault risk index for quantifying the local abnormal degree of the data of each dimension, and calculating the local abnormal weighting factor of each moment according to the fault risk index of the data of each dimension and the maximum value of the fault risk index; improving the anomaly detection algorithm and calculating the corrected anomaly score to evaluate the abnormal degree of the operating state of the stamping machine at each moment, and completing the monitoring of the operating state of the connector production machine.

[0007] By preprocessing the state data of the connector production machine, combining the comparison and analysis with the local reference data, determining the relative change degree and abnormal fluctuation degree of the data of each dimension, and statistically analyzing the symbol continuous length sequence through the first-order difference and symbolization process to analyze the abnormal trend. Further, constructing a fault risk index to quantify the local abnormal degree, and calculating the local abnormal weighting factor of each moment based on this. Finally, using this factor to improve the traditional anomaly detection algorithm, calculating the corrected anomaly score, effectively solving the problem that the traditional algorithm is difficult to identify time series anomalies and multi-type data anomalies, and significantly improving the accuracy of monitoring the operating state of the stamping machine and the fault warning ability. By reasonably evaluating the abnormal fluctuations and trends of the data of each dimension, constructing a fault risk index, realizing the refined monitoring of the equipment operating state, timely discovering potential fault risks, reducing equipment fault losses, and improving production efficiency and equipment reliability.

[0008] Preferably, the preprocessing of the state data includes: Eliminating the incorrect data that exceeds the physical range from the state data, deleting duplicate records, smoothing the state data by using the moving average method, and filling the missing data by using the regression filling method to obtain complete state data, and performing standard deviation normalization processing on the complete state data.

[0009] Preferably, the relative change degree includes: Taking the dimension data in any state data as the marked dimension data, and taking the dimension data sequence corresponding to the preset number of moments before the marked dimension data as the local reference data of the marked dimension data; Calculate the absolute difference between the local reference data of the marked dimension data at two adjacent moments, add 1 to the absolute difference and take the reciprocal as the fluctuation smoothness; calculate the sum of the fluctuation smoothness between all adjacent moment data in the local reference data as the local smoothness sum; Take the ratio between the fluctuation smoothness and the local smoothness sum as the relative change degree of the marked dimension data.

[0010] By selecting any dimension data as the marked dimension data, determining its local reference data, calculating the reciprocal of the absolute difference between adjacent moment data plus 1 as the fluctuation smoothness, then summing to obtain the local smoothness sum, and finally taking the ratio to obtain the relative change degree, it can accurately reflect the fluctuation characteristics of each dimension data.

[0011] Preferably, the abnormal fluctuation degree includes: Take the dimension data in any state data as the marked dimension data, and use the dimension data sequence corresponding to the preset number of moments before the marked dimension data as the local reference data; Map the relative change degree of each data in the local reference data using a logarithmic function, sum the product of the relative change degree of all data in the local reference data and the result after logarithmic function mapping, and take the negative value of the summation result to obtain the abnormal fluctuation degree of the local reference data.

[0012] By performing logarithmic function mapping on the relative change degree of each data in the local reference data and summing and taking the negative value to obtain the abnormal fluctuation degree, it can effectively quantify the fluctuation complexity and uncertainty of the data. The logarithmic mapping can enhance the sensitivity to subtle fluctuations, improve the accuracy of anomaly detection, and assist in earlier and more accurate fault warning.

[0013] Preferably, the first-order difference and symbolization of the local reference data include: Perform first-order difference processing on the local reference data. According to each element after the first-order difference, mark the element less than 0 as -1, equal to 0 as 0, and greater than 0 as 1.

[0014] Preferably, the symbol continuous length sequence includes: Arrange the symbols of each element in order to form a local state difference sequence, count the continuous length of the same elements in the local state difference sequence, and arrange the continuous length of all elements in the order of appearance to form a symbol continuous length sequence.

[0015] By counting the consecutive lengths of the same elements and forming a symbol consecutive length sequence, it helps to highlight the continuously unchanged trend segments in the data. The lengths of these trend segments can quantify the duration of the data under a certain symbol, thereby providing a basis for identifying potential abnormal patterns; shorter consecutive lengths may imply frequent data fluctuations, belonging to normal random fluctuations; while longer consecutive lengths may indicate the existence of persistent abnormal trends, helping to improve the accuracy and timeliness of anomaly detection.

[0016] Preferably, the fault risk index includes: Traverse each element in the symbol consecutive length sequence, perform exponential function mapping on the ratio of each element in the symbol consecutive length sequence to the total time corresponding to the local reference data to obtain the relative consecutive length, sum up the relative consecutive lengths of all elements, and take the product of the summation result and the degree of abnormal fluctuation of the local reference data as the fault risk index of the local reference data.

[0017] Through exponential function mapping, the discrimination degree for different consecutive lengths can be enhanced, making the contribution of longer consecutive lengths to the fault risk index more significant, so as to more accurately reflect potential risks, be able to early and accurately warn of potential faults of equipment, and reduce false alarms and missed alarms.

[0018] Preferably, the local anomaly weighting factor includes: Taking the dimensional data of any state data in the stamping machine operation state data as the marked dimensional data, traverse the ratio between the fault risk index of the local reference data and the maximum value of the fault risk indexes at all moments of the marked dimensional data to obtain the normalized fault risk index, and take the average value of the sum of the normalized fault risk indexes of all dimensional data as the local anomaly weighting factor at the corresponding moment.

[0019] Preferably, the corrected anomaly score includes: Use the traditional algorithm to calculate the anomaly scores of the stamping machine operation state data at each moment, take the sum of the local anomaly weighting factor and the initial weight of the traditional anomaly detection algorithm as the comprehensive anomaly weight, and take the product of the comprehensive anomaly weight and the anomaly score of the traditional algorithm as the corrected anomaly score.

[0020] In a second aspect, a running state monitoring system for a connector production machine includes: a processor and a memory, and the memory stores computer program instructions, which implement the above-mentioned running state monitoring method for the connector production machine when the computer program instructions are executed by the processor.

[0021] The present invention has the following effects: 1. The present invention constructs a fault risk index by combining the degree of abnormal fluctuation and abnormal trend, and improves the anomaly detection algorithm by incorporating a local anomaly weighting factor, effectively solving the problem that traditional algorithms are difficult to identify anomalies in time series and multi-type data anomalies, significantly improving the accuracy of monitoring the operating state of the connector production machine, and enabling it to more accurately evaluate the degree of equipment anomalies. The traditional algorithm has difficulty in identifying time series anomalies and multi-type data anomalies, significantly improving the accuracy of monitoring the operating state of the connector production machine, and enabling it to more accurately evaluate the degree of equipment anomalies.

[0022] 2. The present invention comprehensively analyzes the local anomaly degree of data in each dimension, processes the data through multiple steps such as preprocessing, fluctuation analysis, and trend analysis, constructs a fault risk index, and then realizes a detailed evaluation of the equipment operating state. It helps to timely discover potential fault risks, enhance the fault warning ability, reduce the losses caused by equipment failures, and improve production efficiency and equipment reliability. Description of the Drawings

[0023] Figure 1 is a flowchart of the method from step S1 to step S5 in the method for monitoring the operating state of a connector production machine according to an embodiment of the present invention.

[0024] Figure 2 is a block diagram of the structure of a system for monitoring the operating state of a connector production machine according to an embodiment of the present invention. Detailed Embodiments

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments.

[0026] Referring to Figure 1 , a method for monitoring the operating state of a connector production machine includes steps S1 to S5, specifically as follows: S1: Obtain the state data related to the operation of the connector production machine, and preprocess the state data.

[0027] Preprocessing the state data includes: Eliminating the error data that exceeds the physical range from the state data, deleting duplicate records, smoothing the state data by using the moving average method, filling the missing data by using the regression filling method to obtain complete state data, and performing standard deviation normalization processing on the complete state data to eliminate the dimension problem of the data in each dimension of the state data.

[0028] It should be noted that the connector production machine includes various machine devices such as a stamping machine, an injection molding machine, and an electroplating device. For the convenience of description, this application takes the stamping machine as an example for description and analysis.

[0029] During the operation of the stamping machine, data in multiple dimensions is generated. In this application, the voltage, current, working temperature, stamping pressure, and stamping speed data of the stamping machine during operation are analyzed as the operation status data of the stamping machine.

[0030] Install a pressure sensor on the die of the stamping machine to collect stamping pressure data in real time; install a speed sensor on the traditional system of the stamping machine to collect stamping speed data in real time; install a temperature sensor at the core component of the stamping machine to collect the working temperature data of the stamping machine in real time; collect the voltage and current data of the stamping machine during operation through a voltage and current acquisition module in the industrial environment.

[0031] Since, during the process of collecting the operation status data of the production machine, the collected data may have missing points and other situations due to factors such as environmental interference. To avoid the impact of missing values on subsequent processing steps, this application uses the regression filling method to fill the missing values. The regression filling method is a well-known technology and will not be elaborated here.

[0032] S2: For the preprocessed status data, taking the preset local reference data as the benchmark, compare the fluctuation differences between the data in each dimension and the adjacent dimension data, determine the relative change degree of the data in each dimension, and calculate the entropy of the relative change degree of all dimension data to obtain the abnormal fluctuation degree of the data in each dimension.

[0033] The relative change degree includes: Taking the dimension data in any status data as the marked dimension data, and taking the dimension data sequence corresponding to the preset number of moments before the marked dimension data as the local reference data of the marked dimension data; Calculate the absolute difference between the data at two adjacent moments in the local reference data of the marked dimension data, add 1 to the absolute difference and take the reciprocal as the fluctuation smoothness; calculate the sum of the fluctuation smoothness between all adjacent moment data in the local reference data as the local smoothness sum; Take the ratio between the fluctuation smoothness and the local smoothness sum as the relative change degree of the marked dimension data.

[0034] It should be noted that each data point collected according to the moment corresponds to data in multiple dimensions. By dividing the local reference data according to the dimension data sequence corresponding to the preset number of moments, the local reference data corresponding to multiple dimension data can be obtained.

[0035] Specifically, the relative change degree satisfies the following relational expression: ; In the formula, represents the moment and the previous moments of the operation status data of the stamping machine, and the The relative change degree of the data of the th dimension in the local reference data composed of the data of dimensions, represents the data of the th dimension in the local reference data composed of the data of the th dimension in the stamping machine operation state data at the th moment and the previous moments, represents the data of the th dimension in the local reference data composed of the data of the th dimension in the stamping machine operation state data at the th moment and the previous moments, represents the number of moments before the

[0036] It should be noted that adding 1 to the denominator is to avoid a zero denominator, which may lead to inability to calculate. Implementers can select according to the actual situation. reflects the stability degree of the data of the th dimension in the sequence relative to the data of the previous dimension, reflects the smoothness degree of the fluctuation of the data of the th dimension in the sequence relative to the data of the previous dimension. The larger the value, the smaller the fluctuation degree of the data of the th dimension relative to the data of the previous dimension, and the smoother it is. The larger the value, the greater the fluctuation degree of the data of this dimension relative to the data of the previous dimension, and the less stable it is.

[0037] The abnormal fluctuation degree includes: Taking the data of a dimension in any state data as the marked dimension data, and using the dimension data sequence corresponding to the preset number of moments before the marked dimension data as the local reference data; Mapping the relative change degree of each data in the local reference data using the logarithmic function, summing the product of the relative change degrees of all data in the local reference data and the result after mapping with the logarithmic function, and taking the negative value of the summation result to obtain the abnormal fluctuation degree of the local reference data.

[0038] Specifically, the abnormal fluctuation degree satisfies the following relational expression: ; In the formula, represents the abnormal fluctuation degree of the local reference data, represents the number of moments before the th moment, represents the data of the th dimension in the stamping machine operation state data at the moments before the The relative change degree of the th dimensional data in the local reference data of dimensional data, represents the logarithmic function with the natural constant as the base.

[0039] It should be noted that the local stamping fluctuation entropy reflects the fluctuation regularity and complexity of the local reference data. The smaller the value, the more regular and stable the fluctuation in the local reference data; the larger the value, the more complex and irregular the fluctuation, corresponding to a higher instability of the stamping machine operation state and a greater possibility of abnormal fluctuation.

[0040] During the operation of the stamping machine, if the operation state is normal, among the dimensional data included in the operation state data, when there is a small fluctuation near a stable value, that is is larger, according to the actual meaning and formula of entropy, the calculated local stamping fluctuation entropy is smaller; if the operation state is abnormal, such as the unstable power supply system, etc., the dimensional data such as voltage and current may have large abnormal fluctuations, making the calculated smaller, according to the actual meaning and formula of entropy, resulting in a larger calculated local stamping fluctuation entropy.

[0041] Furthermore, in the working process of the stamping machine, some abnormal states may be long-term and trend-based, such as insufficient air pressure caused by air pressure system failures, oil circuit blockages caused by hydraulic system failures, etc. Such abnormalities may not be easily reflected only by the size of the data itself. Therefore, a fault risk index is constructed based on the degree of abnormal fluctuation, combining the degree of abnormal fluctuation and abnormal trend of each dimensional data in the operation state data, and comprehensively reflecting the degree of abnormality of each dimensional data. The specific steps are as follows: S3: After performing the first-order difference and symbolizing the local reference data, count the symbol continuous length sequence composed of the continuous lengths of the same symbols, and analyze the abnormal trend.

[0042] Perform the first-order difference processing on the local reference data. According to each element after the first-order difference, mark the element less than 0 as -1, equal to 0 as 0, and greater than 0 as 1.

[0043] Arrange the symbols of each element in order to form a local state difference sequence. Count the continuous lengths of the same elements in the local state difference sequence, and arrange the continuous lengths of all elements in the order of appearance to form a symbol continuous length sequence.

[0044] Exemplarily, the local state difference sequence: ; The 1st to 3rd elements are symbols 1 with a consecutive length of 3, the 4th element is a symbol 0 with a consecutive length of 1, the 5th to 6th elements are symbols 1 with a consecutive length of 2, and the 7th to 9th elements are symbols 0 with a consecutive length of 3. Therefore, the symbol consecutive length sequence obtained from the local state difference sequence is .

[0045] It should be noted that if each element value in the symbol consecutive length sequence is relatively small, it is normal random fluctuation and the possibility of having an abnormal trend is small; if each element value in the symbol consecutive length sequence is relatively large, the possibility of having an abnormal trend is large, and as the element value increases, the possibility of abnormal trend should have an exponential growth trend, rather than a linear trend.

[0046] That is to say, during the operation of the stamping machine, the stronger the trend of the operation state data, the more likely it indicates that the stamping machine has an abnormal operation trend state. At the same time, if the volatility of the operation state data is stronger, it indicates that the operation state of the stamping machine is more unstable, and the failure risk index is larger.

[0047] S4: Construct a failure risk index based on the degree of abnormal fluctuation and abnormal trend. Construct a failure risk index for quantifying the local abnormal degree of each dimension data. According to the failure risk index of each dimension data and the maximum value of the failure risk index, calculate the local abnormal weighting factor at each moment.

[0048] The failure risk index includes: Traverse each element in the symbol consecutive length sequence, perform an exponential function mapping on the ratio of each element in the symbol consecutive length sequence to the total time corresponding to the local reference data to obtain the relative consecutive length, sum up the relative consecutive lengths of all elements, and take the product of the summation result and the degree of abnormal fluctuation of the local reference data as the failure risk index of the local reference data.

[0049] Specifically, the failure risk index satisfies the following relational expression: ; In the formula, represents the failure risk index of the local reference data, represents the degree of abnormal fluctuation of the local reference data, represents the sequence length of the symbol consecutive length sequence, represents the exponential function with the natural constant as the base, represents the value of the th element in the symbol consecutive length sequence, represents the number of moments before the

[0050] That is to say, the value of the -th element in the symbol consecutive length sequence is the number of times a certain symbol (such as ) appears continuously in the local state difference sequence, reflecting the proportion of the length of the continuous symbol appearance relative to the total time window, that is, the relative duration of the symbol in the time series. By applying the de-exponential function, the influence of longer continuous lengths on the fault risk index can be amplified. Longer continuous lengths mean stronger trendiness, which may indicate a more significant abnormal trend in the equipment operating state.

[0051] The local anomaly weighting factor includes: Taking the dimensional data of any state data in the stamping machine operating state data as the marked dimensional data, traversing the ratio between the fault risk index of the local reference data and the maximum value of the fault risk indices at all times of the marked dimensional data, obtaining the normalized fault risk index, and taking the average value of the sum of the normalized fault risk indices of all dimensional data as the local anomaly weighting factor at the corresponding time.

[0052] Specifically, the local anomaly weighting factor satisfies the following relational expression: ; In the formula, represents the local anomaly weighting factor at the -th time, represents the number of dimensions of the stamping machine operating state, represents the fault risk index of the -th dimensional data in the local reference data corresponding to the -th time, represents the maximum value of the fault risk indices at all times in the -th dimensional data of the stamping machine operating state data.

[0053] That is to say, reflects the relative anomaly degree of the -th dimensional data in the stamping machine operating state data at the -th time. During the operation of the stamping machine, the greater the relative anomaly degree of each dimensional data at the -th time, the greater the degree of abnormal operation at this time. When using the

[0054] anomaly detection algorithm to calculate the anomaly degree at the When the anomaly detection algorithm monitors the operating state of a stamping machine, one data point represents one moment. At the same time, to further evaluate the comprehensive anomaly degree of all dimensional data during the operation of the stamping machine at the moment, therefore, this application constructs a local anomaly weighting factor based on the fault risk index to evaluate the anomaly state at each moment. The specific steps are as follows: S5: Improve based on the local anomaly weighting factor the anomaly detection algorithm and calculate the corrected anomaly score to evaluate the anomaly degree of the operating state of the stamping machine at each moment, and complete the monitoring of the operating state of the connector production machine.

[0055] The corrected anomaly score includes: Use the traditional algorithm to calculate the anomaly score of the operating state data of the stamping machine at each moment, and take the sum of the local anomaly weighting factor and the initial weight of the traditional anomaly detection algorithm as the comprehensive anomaly weight, and take the product between the comprehensive anomaly weight and the anomaly score of the traditional algorithm as the corrected anomaly score.

[0056] Specifically, the anomaly score satisfies the following relational expression: ; In the formula, represents the corrected anomaly score of the operating state data of the stamping machine at the moment after improving the algorithm, represents the local anomaly weighting factor at the moment, 1 represents the initial weight of the traditional anomaly detection algorithm, represents the anomaly score of the operating state data of the stamping machine calculated using the traditional algorithm at the moment.

[0057] That is to say, reflects the comprehensive anomaly weight of the operating state of the stamping machine at the moment. The larger is, the more obvious the anomaly degree of the operating state data of the stamping machine in the local time series. At the same time, the larger the anomaly score calculated by the traditional algorithm, the greater the difference degree between the operating state data of the stamping machine at the moment and that of the other moments, and the greater the comprehensive anomaly degree. Therefore, the larger the anomaly score of the improved algorithm is calculated.

[0058] ​So far, a sequence of abnormal scores at each moment during the operation of the connector production machine has been obtained. The sequence of abnormal scores is used as the input of the K-means clustering algorithm, with K taking the value of 3, and the output is the moments after clustering. That is to say, the improvement of each moment of the connector production machine After the abnormal scores are input into the K-means clustering algorithm, each moment will be assigned to a specific clustering cluster. The result of the clustering cluster is a set of moment sets, each set corresponding to a clustering cluster, and the moments within the same cluster have similar abnormal score characteristics. Calculate the mean of the abnormal scores at each moment after improvement within each clustering cluster respectively The mean of the abnormal scores, sort the clustering clusters in descending order of the mean, and record the sorted clustering clusters as the severe abnormal cluster, the controllable abnormal cluster, and the normal operation cluster respectively. Among them, the K-means clustering algorithm is a well-known technology and will not be elaborated here.

[0059] Take the clustering result of the clustering cluster to which the current moment belongs as the monitoring result of the operation state of the connector production machine, and realize the monitoring of the operation state of the connector production machine.

[0060] The present invention also provides an operation state monitoring system for a connector production machine. As Figure 2 shown, the system includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an operation state monitoring method for a connector production machine according to the first aspect of the present invention is realized. The system also includes other components well-known to those skilled in the art such as a communication bus and a communication interface, and their settings and functions are known in the art, so they will not be elaborated here.

[0061] It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several deformations and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A method for monitoring the operating status of a connector production machine, characterized in that: include: Obtaining status data related to the operation of the connector production machine and preprocessing the status data; For the preprocessed state data, the preset local reference data is used as a benchmark to compare the fluctuation difference between each dimension data and the adjacent dimension data, determine the relative change degree of each dimension data, and calculate the entropy of the relative change degree of all dimensional data to obtain the abnormal fluctuation degree of each dimensional data; After taking the first-order difference and symbolizing the local reference data, the symbol continuous length sequence composed of the continuous length of the same symbol is counted, and the abnormal trend is analyzed; A fault risk index is constructed based on the abnormal fluctuation degree and abnormal trend, and a fault risk index is constructed to quantify the local abnormal degree of each dimensional data. Based on the fault risk index of each dimensional data and the maximum value of the fault risk index, the local abnormal weighting factor at each moment is calculated; Improvement based on local anomaly weighting factor The anomaly detection algorithm calculates and corrects the anomaly score, evaluates the degree of abnormality of the stamping machine's operating status at each moment, and completes the monitoring of the connector production machine's operating status.

2. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The preprocessing of the state data includes: The erroneous data beyond the physical range will be eliminated from the status data, duplicate records will be deleted, the status data will be smoothed using the sliding flat method, the missing data will be filled using the regression filling method to obtain complete status data, and the complete status data will be normalized by standard deviation.

3. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The relative degree of change includes: Taking dimension data in any state data as marked dimension data, taking the dimension data sequence corresponding to a preset number of moments before the marked dimension data as local reference data for the marked dimension data; Calculate the absolute difference between the local reference data of the marked dimension data and the data at two adjacent moments, add 1 to the absolute difference and take the reciprocal as the fluctuation stability; calculate the sum of the fluctuation stability between all adjacent moment data in the local reference data as the local stability sum; The ratio of fluctuation stability to local stability is used as the relative change degree of the labeled dimension data.

4. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The abnormal fluctuation degree includes: Taking dimension data in any state data as marked dimension data, taking the dimension data sequence corresponding to a preset number of moments before the marked dimension data as local reference data; The relative change degree of each data in the local reference data is mapped using a logarithmic function, the product of the relative change degree of all data in the local reference data and the logarithmic function mapping result is summed, the sum result is negatively valued, and the abnormal fluctuation degree of the local reference data is obtained.

5. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The first-order difference and symbolization of the local reference data include: Perform first-order difference processing on the local reference data. According to each element after the first-order difference, mark the element less than 0 as -1, equal to 0 as 0, and greater than 0 as 1.

6. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The symbol continuous length sequence includes: Arrange the symbols of each element in order to form a local state difference sequence, count the continuous lengths of the same elements in the local state difference sequence, and arrange the continuous lengths of all elements in the order of appearance to form a symbol continuous length sequence.

7. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The failure risk index includes: Each element in the symbol continuous length sequence is traversed, and the ratio of each element in the symbol continuous length sequence to the total time corresponding to the local reference data is mapped by an exponential function to obtain the relative continuous length. The relative continuous lengths of all elements are summed up, and the product of the summation result and the abnormal fluctuation degree of the local reference data is used as the fault risk index of the local reference data.

8. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The local abnormality weighting factor includes: Taking the dimensional data of any state data in the stamping machine operation status data as the marked dimensional data, the ratio between the fault risk index of the local reference data and the maximum value of the fault risk index of the marked dimensional data at all times is traversed to obtain the normalized fault risk index, and the average value of the normalized fault risk index of all dimensional data is taken as the local abnormality weighting factor at the corresponding moment.

9. The method for monitoring the operating status of a connector production machine according to claim 1, characterized in that: The modified abnormality score includes: Use traditional The algorithm calculates the abnormal score of the stamping machine operation status data at each moment, and combines the local abnormal weighting factor with the traditional The sum of the initial weights of the anomaly detection algorithm is used as the comprehensive anomaly weight, and the comprehensive anomaly weight is combined with the traditional The product of the anomaly scores of the algorithms is taken as the modified anomaly score.

10. A system for monitoring the operation status of a connector production machine, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the operating status monitoring method of the connector production machine according to any one of claims 1-9 is implemented.

Images