A cable production data acquisition method and system

Abstract

The application belongs to the technical field of data processing, and particularly relates to a cable production data acquisition method and system. The method comprises the following steps: acquiring a reference data sequence and a data sequence to be aligned in a cable production process; calculating a local fluctuation degree based on data point amplitude information of the data sequence to be aligned, and calculating a window stability index; calculating a local shape complexity based on data point shape information of the data sequence to be aligned, and generating a dynamic scale factor according to the local shape complexity; calculating a composite reliability weight of each data point in the data sequence to be aligned, and performing dynamic time warping based on the composite reliability weight to fuse the reference data sequence and the data sequence to be aligned, so as to realize data acquisition of the cable production process. The application can effectively avoid noise interference, and greatly improve the robustness of data alignment and the accuracy of acquired data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing. More particularly, the present application relates to a cable production data acquisition method and system. BACKGROUND

[0002] In the cable production process, in order to realize fine monitoring and tracing of product quality, it is necessary to collect a large amount of process data from each process such as wire drawing, twisting, and insulation extrusion of the production line. These data are diverse in source and structure, forming a typical multi-source heterogeneous data stream, for example, high-frequency vibration sensor data for monitoring mechanical stability, and low-frequency temperature and pressure sensor data reflecting process state, which differ greatly in data type and sampling frequency.

[0003] In order to build a complete digital archive of the cable in the whole production cycle, it is necessary to accurately align and fuse these data streams that are dispersed in time and different in rate. The existing technology usually adopts a dynamic time warping (DTW) algorithm to solve such time series alignment problems. DTW can effectively align sequences with time axis scaling or translation by finding the optimal nonlinear matching path between two time series, and is very suitable for sensor data alignment with different sampling rates.

[0004] However, in the complex electromagnetic and mechanical environment of cable production, sensor data is easily disturbed by instantaneous interference, producing noise spikes far from the normal fluctuation range. When calculating the warping distance, the standard DTW algorithm treats each data point in the sequence equally. When processing data sequences containing such singular points, in order to minimize the overall warping distance, the algorithm will force normal data points in the other sequence to be matched with these singular points, resulting in serious distortion of the warping path that does not conform to the physical law. This distorted alignment path will generate false interpolation and resampling data, ultimately causing the fused data to be severely contaminated and unable to truly reflect the comprehensive state of the cable at that moment, thereby seriously affecting the accuracy of subsequent quality tracing and process optimization analysis. SUMMARY

[0005] To solve the technical problem of path distortion in dynamic time warping algorithm for multi-source data alignment caused by noise interference, the present application provides solutions in the following aspects.

[0006] In a first aspect, the present application provides a cable production data acquisition method, comprising: acquiring a reference data sequence and a data sequence to be aligned in a cable production process, and synchronously performing segmentation processing on the reference data sequence and the data sequence to be aligned; based on data point amplitude information of the data sequence to be aligned, calculating a local fluctuation degree for representing a deviation degree of a single data point from its adjacent data, and calculating a window stability index for measuring an overall fluctuation state of the data sequence to be aligned after segmentation; based on data point shape information of the data sequence to be aligned, calculating a local shape complexity for representing a local shape sharpness of a signal, and generating a dynamic scale factor according to the local shape complexity; based on the local fluctuation degree, the window stability index and the dynamic scale factor, calculating a composite reliability weight of each data point in the data sequence to be aligned, and performing dynamic time warping based on the composite reliability weight to fuse the reference data sequence and the data sequence to be aligned, so as to realize data acquisition of the cable production process.

[0007] The present application evaluates a composite reliability weight for each data point before performing DTW, and comprehensively considers the amplitude deviation degree and the shape abnormality degree of the data point. For a low-reliability data point which may be noise, the weight is significantly reduced, so that the influence of the noise point is weakened when calculating the warping path by DTW, and the alignment path is prevented from being distorted due to noise, so that the accuracy and robustness of data fusion are improved, and the quality of the finally acquired data is ensured.

[0008] Preferably, the local fluctuation degree satisfies the relationship: ; wherein, is the local fluctuation degree of the i th data point, is the i th data point in the data sequence to be aligned, w is a window size for defining a local neighborhood, and o is an index of a data point in the local neighborhood.

[0009] By calculating the deviation of the data point from its local mean value, the abnormality degree of the point can be preliminarily evaluated from the value size, and the greater the local fluctuation degree of a data point, the higher the possibility of the data point being noise, thereby providing a judgment basis at the amplitude level for subsequent reliability evaluation.

[0010] Preferably, the window stability index satisfies the relationship: ; wherein, is the window stability index of the k th data window, is the window variance of the k th data window, is a very small positive number to prevent the denominator from being 0.

[0011] In a very stable window, even a small local fluctuation can be abnormal; while in a window that is fluctuating wildly itself, a local fluctuation of the same size can be a reasonable process variation. Therefore, the window stability index provides a dynamic benchmark for the severity of local fluctuation.

[0012] Preferably, the local morphological complexity satisfies the relationship: ; wherein, is the local morphological complexity of the i-th data point, , and is the adjacent three data points in the data sequence to be aligned, is a very small positive number to prevent the denominator from being zero.

[0013] Preferably, the dynamic scale factor satisfies the relationship: ; wherein, is the dynamic scale factor of the i-th data point, is a preset basic scale coefficient.

[0014] Preferably, the composite reliability weight satisfies the relationship: ; wherein, is the composite reliability weight of the i-th data point.

[0015] By fusing the information of the three dimensions into a unified weight, the evaluation of the reliability of the data points is more comprehensive and intelligent, which can accurately identify and suppress noise points with high amplitude and abnormal morphology, while retaining process mutation points with high amplitude but normal morphology.

[0016] Preferably, when the dynamic time warping is performed based on the composite reliability weight, the weighted cost used when constructing the cost matrix is: ; wherein, is the data point in the reference data sequence and the original warping distance cost between the data point in the data sequence to be aligned, is the composite reliability weight of the data point .

[0017] Preferably, the reference data sequence is a reading sequence of a laser length measuring instrument, and the data sequence to be aligned is a signal sequence of a high-frequency vibration sensor.

[0018] In a second aspect, the present application provides a cable production data acquisition system, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned cable production data acquisition method is realized.

[0019] By adopting the technical scheme, the cable production data acquisition method is generated into a computer program and stored in a memory to be loaded and executed by a processor, so that a terminal device is manufactured according to the memory and the processor, and use is facilitated.

[0020] The application not only considers the deviation degree of the data points in the amplitude, but also introduces analysis of the local form of the data, and by evaluating the sharpness of the signal, the application can intelligently identify the instantaneous noise peak and the real process mutation.

[0021] Further, by applying the composite reliability weight to the dynamic time warping algorithm, the DTW can actively avoid the interference of the noise points when planning the alignment path, effectively solving the problem that the standard DTW algorithm is sensitive to the singular points and is prone to path distortion. Finally, the application can generate a smoother and more reasonable alignment path that conforms to the physical law, greatly improving the robustness and accuracy of data alignment in a complex industrial noise environment. BRIEF DESCRIPTION OF DRAWINGS

[0022] The above and other objects, features and advantages of the exemplary embodiments of the present application will be more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which several embodiments of the present application are shown by way of example, and wherein like reference numerals refer to like elements throughout. In the drawings:

[0023] Figure 1 is a flowchart schematically showing a cable production data acquisition method in the present application;

[0024] Figure 2 is a sequence alignment schematic diagram schematically showing application of a standard DTW algorithm;

[0025] Figure 3 is a sequence alignment and reliability weight adjustment schematic diagram schematically showing application of an embodiment of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of, rather than all of, the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0027] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0028] The embodiments of the present application disclose a cable production data acquisition method, referring to Figure 1 , comprising steps S1-S4:

[0029] S1, obtaining a reference data sequence and a data sequence to be aligned in a cable production process, and synchronously segmenting the reference data sequence and the data sequence to be aligned.

[0030] In an optional embodiment, a reading sequence of a laser length measuring instrument capable of stably representing the physical position of the cable can be selected as the reference data sequence, which is stable and continuous, and suitable as a reference for alignment. Meanwhile, a signal sequence of a high-frequency vibration sensor that needs to be aligned can be selected as the data sequence to be aligned, which has a high sampling frequency and is easily disturbed by electromagnetic and mechanical noise in the production site, resulting in singular points.

[0031] Further, in order to perform real-time processing, a fixed-length time window, for example, 1 second, can be set, and the reference data sequence and the data sequence to be aligned are synchronously and slidingly segmented along the time axis. In the kth time window, a reference data sub-sequence and a data sub-sequence to be aligned are obtained, that is, there is a corresponding reference data sub-sequence and a data sub-sequence to be aligned in each time window.

[0032] In this way, by clearly obtaining the data sequences and synchronously segmenting the two data sequences, standardized data inputs are provided for subsequent local feature analysis and real-time data processing.

[0033] S2, based on the amplitude information of the data points of the data sequence to be aligned, calculating a local fluctuation degree for representing the deviation degree of a single data point from its adjacent data, and calculating a window stability index for measuring the overall fluctuation state of the segmented data sequence to be aligned.

[0034] In an optional embodiment, for each data point in each data sub-sequence to be aligned, the local fluctuation degree of the data point in the time window and the window stability index of the window can be calculated. Wherein, the local fluctuation degree satisfies the relationship:

[0035]

[0036] Wherein, is the local fluctuation degree of the ith data point, is the ith data point in the data sequence to be aligned, w is the window size for defining the local neighborhood, which is used to determine how many neighbor data points are taken on the left and right of a data point when calculating the local mean, and o is the index of the data points in the local neighborhood.

[0037] For example, a vibration signal data sequence intercepted is {..., 10.1, 10.3, 15.8, 10.2, 10.4,...}, for the data point is 15.8, the window size of the local neighborhood is set to 2, then its neighborhood is {10.1, 10.3, 15.8, 10.2, 10.4}, and the mean of the neighborhood is 11.36, and the corresponding local fluctuation degree is , which is larger, indicating that the data point can be an amplitude abnormal point.

[0038] To measure the overall fluctuation state of the current entire data sub-sequence to be aligned, so as to judge the severity of the local fluctuation, a window stability index can be calculated, and the window stability index satisfies the relationship:

[0039]

[0040] , wherein is the window stability index of the kth data window, is the window variance of the kth data window, To prevent the denominator from being a very small positive number, for example, in this scheme is taken as .

[0041] For example, the variance of the sequence {10.1, 10.3, 15.8, 10.2, 10.4} is 4.94, and the corresponding window stability index is 0.2.

[0042] In this way, by performing amplitude analysis on the data sequence to be aligned, the numerical abnormality of the data points is described from the local and global levels, laying a foundation for deeper reliability evaluation.

[0043] S3, based on the data point shape information of the data sequence to be aligned, calculating the local shape complexity for representing the local shape sharpness of the signal, and generating a dynamic scale factor according to the local shape complexity.

[0044] In an optional embodiment, to distinguish the morphologically sharp noise peaks in the data sequence to be aligned from the relatively smooth true process variations, the local shape complexity for representing the local shape sharpness of the signal can be calculated based on the data point shape information of the data sequence to be aligned, and the local shape complexity satisfies the relationship:

[0045]

[0046] , wherein is the local shape complexity of the ith data point, and the larger the value, the more sharp the local shape of the signal, and the more likely it is noise; , and are adjacent three data points in the data sequence to be aligned. To prevent the extreme small positive number with denominator 0, the value is taken as .

[0047] Exemplarily, for the data point is 15.8, the previous data point is 10.3, and the next data point is 10.2, thus , the denominator is , and the corresponding local morphological complexity is 1.98, which is large, indicating the signal morphology at 15.8 is relatively sharp, similar to a noise spike.

[0048] Further, the dynamic scale factor can be generated by the local morphological complexity, and the generation logic is that the more complex the morphology is, the more like noise it is, and the punishment intensity of the amplitude fluctuation of the point should be larger, and the dynamic scale factor satisfies the relationship:

[0049]

[0050] wherein, is the dynamic scale factor of the i-th data point, is a preset basic scale coefficient, which is used to control the overall sensitivity of the subsequent composite reliability weight to the anomaly, and exemplarily, the value can be 0.5.

[0051] Exemplarily, the obtained local morphological complexity of 1.98 data can be used to calculate is 3.62.

[0052] In this way, by performing morphological analysis on the data sequence to be aligned, the difference in shape between noise and real signal can be intelligently identified, and a dynamic penalty factor is generated, which provides a morphological basis for the final composite reliability weight calculation.

[0053] S4, based on the local fluctuation degree, the window stability index and the dynamic scale factor, the composite reliability weight of each data point in the data sequence to be aligned is calculated, and the dynamic time warping based on the composite reliability weight is performed to fuse the reference data sequence and the data sequence to be aligned, so as to realize data acquisition of the cable production process.

[0054] In an optional embodiment, the composite reliability weight of each data point in the data sequence to be aligned can be calculated in combination with the obtained local fluctuation degree, window stability index and dynamic scale factor, which can comprehensively integrate the dual information of amplitude deviation and morphological anomaly, and the composite reliability weight satisfies the relationship:

[0055]

[0056] wherein, a composite reliability weight for the i-th data point, the value of the composite reliability weight is between 0 and 1, when a point is a high reliability point, tends to 1; when a point is a low reliability noise point, tends to 0.

[0057] It is worth noting that since the window stability index is inversely proportional to the variance, the smaller the fluctuation, the larger the window stability index , thus leading to a smaller composite reliability weight . The key to the construction of the logic here is that the role of the window stability index is not to independently determine the weight, but to adjust the composite reliability weight together as a local fluctuation degree punishment amplifier, so as to judge how abnormal a certain size of local fluctuation is in the current overall environment.

[0058] For example, in an extremely stable data window, the corresponding data point variance is extremely small, and the window stability index will be extremely large. In such a stable case, even if a very small local fluctuation occurs, the local fluctuation degree is not large, but this fluctuation will be out of place and is likely to be noise. Because the window stability index is large, it will cause the local fluctuation degree to be amplified by a large margin in the exponential term , thus leading to the composite reliability weight tending to 0, achieving the purpose of heavily punishing the small change and identifying it as a low-reliability noise point in a stable background.

[0059] In an extremely volatile data window, such as the device startup or process switching stage, the data itself fluctuates violently, and the variance is extremely large at this time, leading to a small window stability index . In such a volatile environment, when a local fluctuation degree of the same value as in the extremely stable data window occurs, the fluctuation may only be part of the overall fluctuation and is normal, and should not be considered as noise. Therefore, the small will weaken the influence of in the exponential term , even if is not small, the value of the entire exponential term will be small, leading to a high level of the calculated composite reliability weight.

[0060] In this optional embodiment, when constructing the cost matrix of the DTW algorithm, the original warping distance cost can be weighted by a composite reliability weight, thereby performing dynamic time warping based on the composite reliability weight. Specifically, the weighted cost used when constructing the cost matrix is ​​as follows:

[0061]

[0062] in, Data points in the baseline data sequence Data points in the data sequence to be aligned The cost of the original normalized distance between them For data points The composite reliability weight. When If the weight is extremely low, it indicates that the point is a noise point. In subsequent path planning, the system will identify such low-weight matching points and prioritize the selection of a smoother path that conforms to physical laws, thus avoiding drastic distortion of the path due to noise points.

[0063] Furthermore, based on the found path, the data subsequence to be aligned can be resampled or interpolated to ensure its time points are perfectly aligned with the reference data subsequence. The values ​​of the reference data subsequence and all aligned non-reference data subsequences at each time point are concatenated to form a multi-dimensional feature vector, which is a digital slice of the cable at that moment, thus completing the final data acquisition task.

[0064] like Figure 2 The diagram illustrates the sequence alignment using the standard DTW algorithm. As you can see, when a noise spike appears, the red alignment line is pulled up unreasonably, directly connecting to the highest point of the noise spike in the blue sequence. This means that in order to complete the matching task, the standard DTW algorithm is forced to match a normal data segment in the gray sequence with a completely abnormal noise point in the blue sequence. If the data is fused based on this distorted path, it will cause serious distortion.

[0065] like Figure 3 The diagram shows a sequence alignment and reliability weight adjustment using an embodiment of the present invention. Sub-figure (a) shows the signal and reliability weights, and sub-figure (b) shows the sequence alignment result. It can be seen that when a noise spike occurs, the green alignment line is not pulled up due to the penalty of the composite reliability weight. This means that the matching cost connecting to the noise peak is reduced to zero through the composite reliability weight. Simultaneously, during process switching, although the fluctuations are large, it can be seen that the composite reliability weight only approaches 0 at the moment of switching in the latter half, and the subsequent weight remains high, thus ensuring normal sequence matching.

[0066] In this way, by calculating and applying the composite reliability weight, noise can be intelligently identified and avoided, so as to generate a distortion-free alignment path, and finally realize high-fidelity data acquisition and fusion.

[0067] The embodiment of the present application further discloses a cable production data acquisition system, comprising a processor and a memory, and the memory stores computer program instructions.

[0068] The above system further comprises a communication bus and a communication interface and other components well known to those skilled in the art, the settings and functions of which are known in the art, and thus will not be described here.

[0069] In the description of the present specification, the meaning of "a plurality of", "several" is at least two, for example, two, three or more, etc., unless otherwise explicitly specifically limited.

[0070] Although the present specification has shown and described several embodiments of the present application, it will be apparent to those skilled in the art that such embodiments are provided only by way of example. Many modifications, changes and alternative ways will occur to those skilled in the art without departing from the spirit and principles of the present application. It should be understood that various alternatives to the embodiments of the present application described herein can be employed in practicing the present application.

Claims

1. A method of collecting data for cable production, characterized by, The method comprises the following steps: obtaining a reference data sequence and a data sequence to be aligned in a cable production process, and synchronously performing segmentation processing on the reference data sequence and the data sequence to be aligned; calculating a local fluctuation degree for representing a deviation degree of a single data point from its adjacent data based on data point amplitude information of the data sequence to be aligned, and calculating a window stability index for measuring an overall fluctuation state of the data sequence to be aligned after segmentation; the local fluctuation degree satisfies the relationship: wherein, is the local fluctuation degree of the i-th data point, is the i-th data point in the data sequence to be aligned, w is the window size for defining the local neighborhood, and o is the index of the data point within the local neighborhood. calculating a local morphological complexity for representing a local morphological sharpness degree of a signal based on data point morphological information of the data sequence to be aligned, and generating a dynamic scale factor according to the local morphological complexity; the local morphological complexity satisfies the relationship: wherein, is the local morphological complexity of the ith data point, , and are the three adjacent data points in the data sequence to be aligned, is a very small positive number to prevent the denominator from being zero. calculating a composite reliability weight of each data point in the data sequence to be aligned based on the local fluctuation degree, the window stability index and the dynamic scale factor, and performing dynamic time warping based on the composite reliability weight to fuse the reference data sequence and the data sequence to be aligned, so as to realize data acquisition of the cable production process.

2. The method of claim 1, wherein, The window stability index satisfies the relationship: wherein, is a window stability indicator for the kth data window, is a window variance for the kth data window, is a small positive number to prevent the denominator from being zero.

3. The method of claim 1, wherein, The dynamic scale factor satisfies the relationship: wherein, is a dynamic scale factor for the i-th data point, is a preset base scale coefficient.

4. The method of claim 1, wherein, The composite reliability weight satisfies the relationship: wherein, is the composite reliability weight for the i-th data point.

5. The method of claim 1 or 4, wherein, When the dynamic time warping based on the composite reliability weight is performed, a weighted cost used in constructing a cost matrix is: wherein, is the original distance between data points in the data sequence to be aligned, is the composite reliability weight of data point .​ 6. The method of claim 1, wherein, The reference data sequence is a reading sequence of a laser length measuring instrument, and the data sequence to be aligned is a signal sequence of a high-frequency vibration sensor.

7. A cable production data acquisition system characterized by, The method comprises the following steps: a processor and a memory, the memory storing computer program instructions, when the computer program instructions are executed by the processor, a cable production data acquisition method according to any one of claims 1-6 is realized.

Images