A method and system for intelligent analysis of electronic component production data

Abstract

The present invention relates to the field of data processing technology, and specifically to a method and system for intelligent analysis of electronic component production data, including: obtaining a parameter sequence, obtaining an initial fluctuation degree according to the slope of different parameter values ​​in the parameter segment sequence and the time interval between the parameter value and the extreme value, obtaining the final fluctuation degree of the parameter value in the parameter segment sequence according to the initial fluctuation degree, and then obtaining the distance influence weight of the parameter sequence, obtaining the correlation between different parameter sequences, and then obtaining the influence of the correlation of each parameter sequence on the other parameter sequences, obtaining the distance adjustment coefficient of the parameter sequence according to the distance influence weight and the correlation influence, and then obtaining the distance between sample data, and finally clustering the sample data and completing anomaly detection. The present invention analyzes the correlation between different parameter sequences, so that the final clustering result is more in line with the actual situation, and completes the detection of abnormal reference values.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for intelligent analysis of electronic component production data. Background Art

[0002] Electronic components are small machines and instruments. By analyzing and processing the data related to electronic component production equipment, we can timely understand the current status of the equipment and adjust and optimize the equipment parameters. Existing methods for optimizing equipment parameters usually cluster the different operating states of the equipment and perform outlier detection or further analysis and processing based on the clustering results. In actual situations, when electronic components are running, due to the influence of different parameter data on each other, the calculation of the distance between data cannot accurately represent the similarity of the data, making the clustering results inaccurate, which in turn leads to inaccurate detection results of abnormal data. Summary of the invention

[0003] In order to solve the above problems, the present invention provides a method and system for intelligent analysis of electronic component production data.

[0004] The electronic component production data intelligent analysis method and system of the present invention adopts the following technical solutions:

[0005] An embodiment of the present invention provides a method for intelligent analysis of electronic component production data, the method comprising the following steps:

[0006] Collecting timing monitoring data of electronic component production equipment, wherein the timing monitoring data includes a plurality of parameter sequences, and each parameter sequence includes a plurality of parameter values;

[0007] Divide the parameter sequence into several parameter segment sequences of equal length;

[0008] According to the slopes of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time intervals between the parameter values ​​and the extreme values, the initial fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained, and according to the initial fluctuation degree, the final fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained, and according to the final fluctuation degree, the distance influence weight of each parameter sequence is obtained;

[0009] According to the number of slope changes of parameter values ​​of different parameter sequences in the same parameter segment sequence and the time interval corresponding to the maximum slope value, the correlation between different parameter sequences is obtained, and the influence of the correlation of each parameter sequence on other parameter sequences is obtained according to the correlation, and the distance adjustment coefficient of each parameter sequence is obtained according to the distance influence weight and the correlation influence;

[0010] The distance between different sample data is obtained according to the difference between the distance adjustment coefficient and different parameter values ​​in the parameter sequence. The sample data is a sequence composed of the parameter values ​​at the same position of all parameter sequences according to the acquisition order of the parameter sequence. The sample data is clustered according to the distance between different sample data to obtain several clusters, and the sample data in each cluster is detected for anomalies to obtain several abnormal parameter values.

[0011] Furthermore, the initial fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence is obtained according to the slope of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time interval between the parameter value and the extreme value, and the specific steps include the following:

[0012] ;

[0013] In the formula, is the slope of the qth parameter value in the mth parameter segment sequence of the ith parameter sequence, is the slope mean of all parameter values ​​in the mth parameter segment sequence of the ith parameter sequence, is the time interval between the extreme value closest to the qth parameter value in the mth parameter segment sequence of the ith parameter sequence and the qth parameter value, where the extreme value refers to the extreme value of the parameter value, is the linear normalization function, is the number of parameter values ​​in the mth parameter segment sequence of the ith parameter sequence, is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

[0014] Furthermore, the step of obtaining the final fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence according to the initial fluctuation degree includes the following specific steps:

[0015] ;

[0016] In the formula, is the variance of the initial fluctuation of the parameter values ​​in all parameter segment sequences of the i-th parameter sequence, is an exponential function with a natural constant as base, is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

[0017] Furthermore, the step of obtaining the distance influence weight of each parameter sequence according to the final fluctuation degree includes the following specific steps:

[0018] ;

[0019] In the formula, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the number of parameter segment sequences of the i-th parameter sequence, is the distance influence weight of the i-th parameter sequence.

[0020] Furthermore, the correlation between different parameter sequences is obtained according to the number of slope changes of parameter values ​​in the same parameter segment sequence and the time interval corresponding to the maximum slope value of different parameter sequences, and the specific steps include the following:

[0021] ;

[0022] In the formula, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, represents the time interval corresponding to the maximum value of the slope, is the linear normalization function, is the number of times the slope of the parameter value changes in the mth parameter segment sequence of the i-th parameter sequence, is the number of changes in the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, To take the absolute value, is an exponential function with a natural constant as base, is the number of parameter segment sequences of the parameter sequence, is the correlation between the i-th parameter sequence and the j-th parameter sequence.

[0023] Furthermore, the step of obtaining, according to the correlation, that each parameter sequence is affected by the correlation of other parameter sequences comprises the following specific steps:

[0024] ;

[0025] In the formula, is the correlation between the i-th parameter sequence and the j-th parameter sequence, is the linear normalization function, is the total number of parameter sequences, The i-th parameter sequence is affected by the correlation of other parameter sequences, and the other parameter sequences refer to the remaining parameter sequences except the i-th parameter sequence among the several parameter sequences.

[0026] Furthermore, the distance adjustment coefficient of each parameter sequence is obtained according to the distance influence weight and the correlation influence, and the specific steps include the following:

[0027] ;

[0028] In the formula, is the distance influence weight of the i-th parameter sequence, The i-th parameter sequence is affected by the correlation of other parameter sequences. is the distance adjustment coefficient of the i-th parameter sequence.

[0029] Furthermore, the distance between different sample data is obtained according to the difference between the distance adjustment coefficient and different parameter values ​​in the parameter sequence, and the specific steps include the following:

[0030] ;

[0031] In the formula, is the distance adjustment coefficient of the i-th parameter sequence, is the ath parameter value in the i-th parameter sequence, is the bth parameter value in the i-th parameter sequence, , is the total number of parameter sequences, is the difference of different parameter values, is the distance between the a-th sample data and the b-th sample data.

[0032] Furthermore, clustering the sample data according to the distances between different sample data to obtain a number of clusters, and performing anomaly detection on the sample data in each cluster to obtain a number of abnormal parameter values ​​includes the following specific steps:

[0033] The sample data are clustered using the K-means algorithm according to the distance between different sample data to obtain several clusters. The sample data in each cluster are detected using the LOF anomaly detection algorithm to obtain several abnormal sample data in each cluster. The parameter values ​​in the abnormal sample data are abnormal parameter values.

[0034] The present invention also proposes an electronic component production data intelligent analysis system, comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the steps of the aforementioned method.

[0035] The beneficial effect of the technical solution of the present invention is that when each parameter sequence participates in the distance calculation between sample data, the distance adjustment coefficient of the parameter sequence is obtained according to the change trend of the parameter value in each parameter segment sequence of each parameter sequence combined with the time difference. Compared with the original K-means algorithm which only uses different parameter measurement values ​​of data at different times as the characterization of a certain operating state of the equipment, the present invention quantifies the correlation between different parameter sequences and introduces the correlation into the calculation of the sample data distance, so that the final clustering result is more in line with the actual situation, completes the detection of abnormal reference values, and achieves intelligent analysis of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 A flowchart of the steps of an intelligent analysis method for electronic component production data provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the electronic component production data intelligent analysis method and system proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0039] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0040] The specific scheme of the method for intelligent analysis of electronic component production data provided by the present invention is described in detail below with reference to the accompanying drawings.

[0041] See also Figure 1 , which shows a flowchart of a method for intelligent analysis of electronic component production data provided by an embodiment of the present invention, the method comprising the following steps:

[0042] Step S001, collecting timing monitoring data of electronic component production equipment.

[0043] It should be noted that the purpose of this embodiment is to obtain the adjustment coefficient of the distance between the data of the device at different times by combining the degree to which each parameter is affected by the changes in other parameters with the data change characteristics of each parameter itself, so as to obtain accurate abnormal data. Before starting the analysis, data needs to be collected first.

[0044] Specifically, the equipment management system collects the time series monitoring data of any electronic component production equipment on the current production line within the most recent day. The time series monitoring data includes several parameter sequences, and each parameter sequence includes several parameter values.

[0045] It should be noted that the parameter sequence includes four parameter sequences: voltage sequence, current sequence, temperature sequence and vibration sequence. Each parameter sequence is two-dimensional data. The horizontal axis of the two-dimensional data is time, and the vertical axis is the corresponding parameter value. In the parameter sequence of this embodiment, one parameter value corresponds to every one minute, and the length of each parameter sequence is the same.

[0046] At this point, time series monitoring data and several parameter sequences are obtained.

[0047] Step S002: Divide the parameter sequence into equal lengths to obtain a number of parameter segment sequences for each parameter sequence.

[0048] It should be noted that clustering is performed based on the different change patterns of various parameters of electronic component production equipment to obtain different operating states of the equipment. The data set consisting of all parameter values ​​at each moment is regarded as a sample data. When calculating the distance between different sample data, the difference in the time series change characteristics of different parameter sequences is taken into account. The adjustment coefficient of the parameter participating in the distance calculation is obtained according to the change of different parameter sequences, and then clustering and anomaly detection are completed according to the adjustment coefficient.

[0049] It should be noted that during the operation of electronic component production equipment, samples will be collected regularly to detect the current equipment parameter values, and some parameters that do not meet the standards will be adjusted to ensure that the final product meets the requirements. Different parameters of the equipment usually affect each other, which will cause corresponding parameter changes, such as temperature changes, vibration changes, etc., and such changes will cause the various parameters monitored by the equipment at this time to be not completely determined by the equipment itself, so it is necessary to analyze the local parameter sequence.

[0050] Specifically, a first value is preset, denoted as TH1. This embodiment is described as TH1=20. For any parameter sequence, starting from the first parameter value of the parameter sequence, every TH1 parameter values ​​are divided into a parameter segment sequence, and finally a plurality of parameter segment sequences of the parameter sequence are obtained. It should be noted that if the length of the last parameter segment sequence is less than TH1, the sequence composed of the remaining parameter values ​​is regarded as a parameter segment sequence.

[0051] At this point, several parameter segment sequences are obtained for each parameter sequence.

[0052] Step S003: according to the slopes of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time intervals between the parameter values ​​and the extreme values, the initial fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained; according to the initial fluctuation degree, the final fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained; according to the final fluctuation degree, the distance influence weight of each parameter sequence is obtained.

[0053] It should be noted that different parameter sequences have different change characteristics in time series. For example, for temperature sequences, since it is difficult for temperature changes to produce large numerical changes in a very short period of time, the proportion of this parameter is relatively high when calculating data distance. For voltage sequences or current sequences, the values ​​of these parameters can change greatly in a very short period of time, so the impact on these parameters is relatively weak. Therefore, it is necessary to quantify the changes between each parameter sequence and obtain the distance impact weight.

[0054] Specifically, according to the slope of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time interval between the parameter value and the extreme value, the initial fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence is obtained, as follows:

[0055] ;

[0056] In the formula, is the slope of the qth parameter value in the mth parameter segment sequence of the ith parameter sequence, is the slope mean of all parameter values ​​in the mth parameter segment sequence of the i-th parameter sequence, is the time interval between the extreme value closest to the qth parameter value in the mth parameter segment sequence of the ith parameter sequence and the qth parameter value, where the extreme value refers to the extreme value of the parameter value, is a linear normalization function, and the normalized object is all parameter segment sequences of the i-th parameter sequence. , is the number of parameter values ​​in the mth parameter segment sequence of the ith parameter sequence, is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

[0057] It should be noted that, considering that the parameter values ​​are changing more slowly in the process of changing, the parameter values ​​closer to the extreme point change more slowly than other parts, and the initial fluctuation degree of the parameter value has little effect on it. Therefore, in order to accurately characterize the degree of change of the parameter itself, it is necessary to reduce the weight of this type of data. In actual situations, the parameter value changes of different parameter sequences in time series have certain regularity. If the regularity of the parameter value changes in the parameter sequence in time series is weak, the credibility of the initial fluctuation degree of the parameter value in the parameter segment sequence obtained above is low. Therefore, on the basis of the above initial fluctuation degree, the regularity constraint of the parameter value change is added, and the regularity of the parameter sequence is characterized by the initial fluctuation degree of the parameter value in different parameter segment sequences of the parameter sequence.

[0058] Specifically, the final fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence is obtained according to the initial fluctuation degree, as follows:

[0059] ;

[0060] In the formula, is the variance of the initial fluctuation of the parameter values ​​in all parameter segment sequences of the i-th parameter sequence, is an exponential function with a natural constant as the base. The model is used to present the inverse proportional relationship and normalization processing, where U is the input value of the model. The implementer can set the inverse proportional function and normalization function according to the specific implementation situation. is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

[0061] It should be noted that It represents the regularity of the i-th parameter sequence. The higher the regularity of the parameter sequence, the higher the reliability of the initial fluctuation degree of the parameter value in the m-th parameter segment sequence, and the greater the weight.

[0062] Furthermore, the distance influence weight of each parameter sequence is obtained according to the final fluctuation degree, as follows:

[0063] ;

[0064] In the formula, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the number of parameter segment sequences of the i-th parameter sequence, is the distance influence weight of the i-th parameter sequence.

[0065] It should be noted that the distance influence weight of the parameter sequence is used to adjust the distance between different parameter values ​​in the parameter sequence. The larger the distance influence weight of the i-th parameter sequence, the smaller the change between adjacent parameter values ​​compared with other parameter sequences. Therefore, the proportion of this parameter sequence when calculating the data distance is relatively high.

[0066] At this point, the distance influence weight of each parameter sequence is obtained.

[0067] Step S004: according to the number of slope changes of parameter values ​​of different parameter sequences in the same parameter segment sequence and the time interval corresponding to the maximum slope value, the correlation between different parameter sequences is obtained, according to the correlation, the influence of the correlation of each parameter sequence on other parameter sequences is obtained, and according to the distance influence weight and the correlation influence, the distance adjustment coefficient of each parameter sequence is obtained.

[0068] It should be noted that the above steps obtain the distance influence weight of each parameter sequence. When calculating the distance between parameter values ​​at different times, the mutual influence between different parameter sequences also needs to be considered. That is, when calculating the weight of the influence of the parameter change on the distance for a certain parameter, the influence of the other parameters on the parameter is calculated at the same time, that is, the correlation between different parameter sequences is quantified.

[0069] Specifically, the correlation between different parameter sequences is obtained according to the number of slope changes of parameter values ​​in the same parameter segment sequence and the time interval corresponding to the maximum slope value, as follows:

[0070] ;

[0071] In the formula, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, represents the time interval corresponding to the maximum value of the slope, is a linear normalization function, and the normalized object is all parameter sequences , is the number of times the slope of the parameter value changes in the mth parameter segment sequence of the i-th parameter sequence, is the number of changes in the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, To take the absolute value, is an exponential function with a natural constant as the base. The model is used to present the inverse proportional relationship and normalization processing, where U is the input value of the model. The implementer can set the inverse proportional function and normalization function according to the specific implementation situation. is the number of parameter segment sequences of the parameter sequence, is the correlation between the i-th parameter sequence and the j-th parameter sequence.

[0072] It should be noted that, in this embodiment, when the slope change of adjacent parameter values ​​is greater than 0.1, it is considered that a slope change has occurred. Excluding the influence of noise on the slope, the slope reflects the rate of change of the parameter value. The correlation between the two parameter sequences is characterized by the slope change of the parameter values ​​of different parameter sequences in the same parameter segment sequence. If the correlation between the parameter sequences is high, then as one parameter sequence changes, the other parameter sequence will also change. Therefore, the correlation between different parameter sequences is characterized by the number of slope changes in the parameter segment sequence. At the same time, taking into account the mutual influence between the parameters, which may cause the parameter values ​​of the parameter sequence to have a delay, the delay is characterized by the time interval corresponding to the maximum slope value. The smaller the time interval, the higher the correlation between the two parameter sequences.

[0073] Furthermore, according to the correlation, each parameter sequence is affected by the correlation of other parameter sequences, as follows:

[0074] ;

[0075] In the formula, is the correlation between the i-th parameter sequence and the j-th parameter sequence, is a linear normalization function, and the normalized object is all parameter sequences , is the total number of parameter sequences, The i-th parameter sequence is affected by the correlation of other parameter sequences, and the other parameter sequences refer to the remaining parameter sequences except the i-th parameter sequence among the several parameter sequences.

[0076] It should be noted that the above-mentioned influence of the correlation of each parameter sequence with other parameter sequences and the distance influence weight of each parameter sequence are obtained, and the distance adjustment coefficient of the parameter sequence is finally obtained through the two.

[0077] Specifically, the distance adjustment coefficient of each parameter sequence is obtained according to the distance influence weight and the correlation influence, as follows:

[0078] ;

[0079] In the formula, is the distance influence weight of the i-th parameter sequence, The i-th parameter sequence is affected by the correlation of other parameter sequences. is the distance adjustment coefficient of the i-th parameter sequence.

[0080] At this point, the distance adjustment coefficient of each parameter sequence is obtained.

[0081] Step S005: obtain the distance between different sample data according to the distance adjustment coefficient and the difference between different parameter values ​​in the parameter sequence, cluster the sample data according to the distance between different sample data to obtain several clusters, and perform anomaly detection on the sample data in each cluster to obtain several abnormal parameter values.

[0082] It should be noted that the distance adjustment coefficient of each parameter sequence is obtained above, and the distance between parameter values ​​is obtained through the distance adjustment coefficient.

[0083] Specifically, the distance between different sample data is obtained according to the difference between the distance adjustment coefficient and the different parameter values ​​in the parameter sequence, as follows:

[0084] ;

[0085] In the formula, is the distance adjustment coefficient of the i-th parameter sequence, is the ath parameter value in the i-th parameter sequence, is the bth parameter value in the i-th parameter sequence, , is the total number of parameter sequences, is the difference of different parameter values, is the distance between the ath sample data and the bth sample data, and the sample data is a sequence composed of the parameter values ​​at the same position of all parameter sequences in the order in which the parameter values ​​are obtained.

[0086] It should be noted that the sample data can be regarded as a multidimensional data point in the time series monitoring data, which contains multiple parameter values. If anomaly detection is performed directly on all the sample data in the time series monitoring data, due to the large amount of sample data and the influence between different parameter sequences, in order to reduce the amount of calculation and obtain abnormal parameter values, it is necessary to cluster and detect anomalies on the sample data.

[0087] Specifically, the sample data are clustered according to the distances between different sample data to obtain several clusters, and the sample data in each cluster are subjected to anomaly detection to obtain several abnormal parameter values, as follows:

[0088] The sample data are clustered using the K-means algorithm according to the distance between different sample data to obtain several clusters. The sample data in each cluster are detected using the LOF anomaly detection algorithm to obtain several abnormal sample data in each cluster. The parameter values ​​in the abnormal sample data are abnormal parameter values, thereby achieving intelligent analysis of the time series monitoring data.

[0089] Through the above steps, an intelligent analysis method for electronic component production data is completed.

[0090] Another embodiment of the present invention provides an electronic component production data intelligent analysis system, the system comprising a memory and a processor, the processor executing the computer program stored in the memory, performing the following operations:

[0091] The timing monitoring data of electronic component production equipment is collected, wherein the timing monitoring data includes a plurality of parameter sequences, each parameter sequence includes a plurality of parameter values; the parameter sequence is divided into a plurality of parameter segment sequences of equal length to obtain each parameter sequence; the initial fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence is obtained according to the slope of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time interval between the parameter value and the extreme value; the final fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence is obtained according to the initial fluctuation degree; the distance influence weight of each parameter sequence is obtained according to the final fluctuation degree; the distance influence weight of each parameter sequence is obtained according to the slope of the parameter values ​​of different parameter sequences in the same parameter segment sequence The correlation between different parameter sequences is obtained by the number of rate changes and the time interval corresponding to the maximum value of the slope. The influence of the correlation of each parameter sequence on other parameter sequences is obtained according to the correlation. The distance adjustment coefficient of each parameter sequence is obtained according to the distance influence weight and the correlation influence. The distance between different sample data is obtained according to the distance adjustment coefficient and the difference between different parameter values ​​in the parameter sequence. The sample data is a sequence composed of parameter values ​​at the same position of all parameter sequences according to the acquisition order of the parameter sequence. The sample data is clustered according to the distance between different sample data to obtain several clusters. The sample data in each cluster is detected for abnormality to obtain several abnormal parameter values.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent analysis method for electronic component production data, characterized in that: The method comprises the following steps: Collecting timing monitoring data of electronic component production equipment, wherein the timing monitoring data includes a plurality of parameter sequences, and each parameter sequence includes a plurality of parameter values; Divide the parameter sequence into several parameter segment sequences of equal length; According to the slopes of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time intervals between the parameter values ​​and the extreme values, the initial fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained, and according to the initial fluctuation degree, the final fluctuation degree of the parameter values ​​in each parameter segment sequence of each parameter sequence is obtained, and according to the final fluctuation degree, the distance influence weight of each parameter sequence is obtained; According to the number of slope changes of parameter values ​​of different parameter sequences in the same parameter segment sequence and the time interval corresponding to the maximum slope value, the correlation between different parameter sequences is obtained, and the influence of the correlation of each parameter sequence on other parameter sequences is obtained according to the correlation, and the distance adjustment coefficient of each parameter sequence is obtained according to the distance influence weight and the correlation influence; The distance between different sample data is obtained according to the difference between the distance adjustment coefficient and different parameter values ​​in the parameter sequence, wherein the sample data is a sequence composed of parameter values ​​at the same position of all parameter sequences according to the acquisition order of the parameter sequence, the sample data is clustered according to the distance between different sample data to obtain a number of clusters, and the sample data in each cluster is detected for anomalies to obtain a number of abnormal parameter values; The specific steps of obtaining the distance influence weight of each parameter sequence according to the final fluctuation degree are as follows: ； In the formula, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the number of parameter segment sequences of the i-th parameter sequence, is the distance influence weight of the i-th parameter sequence; The specific steps of obtaining the distance adjustment coefficient of each parameter sequence according to the distance influence weight and the correlation influence are as follows: ； In the formula, The i-th parameter sequence is affected by the correlation of other parameter sequences. is the distance adjustment coefficient of the i-th parameter sequence.

2. According to claim 1, a method for intelligent analysis of electronic component production data is characterized in that: The step of obtaining the initial fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence according to the slope of different parameter values ​​in each parameter segment sequence of the parameter sequence and the time interval between the parameter value and the extreme value includes the following specific steps: ； In the formula, is the slope of the qth parameter value in the mth parameter segment sequence of the ith parameter sequence, is the slope mean of all parameter values ​​in the mth parameter segment sequence of the i-th parameter sequence, is the time interval between the extreme value closest to the qth parameter value in the mth parameter segment sequence of the ith parameter sequence and the qth parameter value, where the extreme value refers to the extreme value of the parameter value, is the linear normalization function, is the number of parameter values ​​in the mth parameter segment sequence of the ith parameter sequence, is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

3. The method for intelligent analysis of electronic component production data according to claim 1, characterized in that: The specific steps of obtaining the final fluctuation degree of the parameter value in each parameter segment sequence of each parameter sequence according to the initial fluctuation degree include the following: ； In the formula, is the variance of the initial fluctuation of the parameter values ​​in all parameter segment sequences of the i-th parameter sequence, is an exponential function with a natural constant as base, is the initial fluctuation degree of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the final fluctuation degree of the parameter value in the mth parameter segment sequence of the ith parameter sequence.

4. The method for intelligent analysis of electronic component production data according to claim 1, characterized in that: The step of obtaining the correlation between different parameter sequences according to the number of slope changes of parameter values ​​in the same parameter segment sequence and the time interval corresponding to the maximum slope value of different parameter sequences includes the following specific steps: ； In the formula, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the i-th parameter sequence, is the time corresponding to the maximum value of the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, represents the time interval corresponding to the maximum value of the slope, is the linear normalization function, is the number of times the slope of the parameter value changes in the mth parameter segment sequence of the i-th parameter sequence, is the number of changes in the slope of the parameter value in the mth parameter segment sequence of the jth parameter sequence, To take the absolute value, is an exponential function with a natural constant as base, is the number of parameter segment sequences of the parameter sequence, is the correlation between the i-th parameter sequence and the j-th parameter sequence.

5. The method for intelligent analysis of electronic component production data according to claim 1, characterized in that: The specific steps of obtaining the influence of the correlation of other parameter sequences on each parameter sequence according to the correlation are as follows: ； In the formula, is the correlation between the i-th parameter sequence and the j-th parameter sequence, is the linear normalization function, is the total number of parameter sequences, The i-th parameter sequence is affected by the correlation of other parameter sequences, and the other parameter sequences refer to the remaining parameter sequences except the i-th parameter sequence among the several parameter sequences.

6. The method for intelligent analysis of electronic component production data according to claim 1, characterized in that: The specific steps of obtaining the distance between different sample data according to the distance adjustment coefficient and the difference between different parameter values ​​in the parameter sequence are as follows: ； In the formula, is the distance adjustment coefficient of the i-th parameter sequence, is the ath parameter value in the i-th parameter sequence, is the bth parameter value in the i-th parameter sequence, , is the total number of parameter sequences, is the difference of different parameter values, is the distance between the a-th sample data and the b-th sample data.

7. The method for intelligent analysis of electronic component production data according to claim 1, characterized in that: The method of clustering the sample data according to the distances between different sample data to obtain a number of clusters, and performing anomaly detection on the sample data in each cluster to obtain a number of abnormal parameter values ​​includes the following specific steps: The sample data are clustered using the K-means algorithm according to the distance between different sample data to obtain several clusters. The sample data in each cluster are detected using the LOF anomaly detection algorithm to obtain several abnormal sample data in each cluster. The parameter values ​​in the abnormal sample data are abnormal parameter values.

8. An electronic component production data intelligent analysis system, the system comprising a memory and a processor, characterized in that: The processor executes the computer program stored in the memory to implement the steps of the method for intelligent analysis of electronic component production data as described in any one of claims 1 to 7.

Images