A method and system for monitoring process faults in cable production equipment

Abstract

This application relates to the field of system monitoring and control technology, and in particular to a method and system for monitoring process faults in cable production equipment. The method includes: acquiring multi-dimensional operating parameters of the cable production equipment to obtain a dataset containing multiple sample points; for any sample point in the dataset, determining its neighborhood set; constructing the nearest neighbor dispersion of the sample point; constructing the neighborhood rheological difference of the sample point; determining an adaptive truncation threshold for the sample point based on the nearest neighbor dispersion and the neighborhood rheological difference; and using the adaptive truncation threshold to perform density peak clustering analysis on the dataset to identify process faults in the cable production equipment. This application improves monitoring accuracy.

Description

Technical Field

[0001] This application relates to the field of system monitoring and control technology, and in particular to a method and system for monitoring faults in the processing of cable production equipment. Background Technology

[0002] Extruders and cabling machines are core equipment in the cable manufacturing process. Their main function is to strand conductor cores and coat them with insulation or sheathing materials to give the cable excellent electrical properties and mechanical protection. Because cable processing typically involves complex conditions of high temperature, high pressure, and continuous high-speed traction, the coupling between process parameters such as traction speed, screw speed, and heating temperature is extremely strong. Any slight abnormal fluctuation in these parameters can lead to irreversible quality defects such as insulation layer eccentricity, surface roughness, bubbles, or out-of-tolerance outer diameter, and in severe cases, even equipment damage or production interruption. Therefore, real-time and accurate fault monitoring of the cable production equipment is crucial for timely detection of potential hazards, reducing scrap rates, and ensuring the safe and stable operation of the production line.

[0003] In related technologies, density peak clustering (DPC) is often used for unsupervised fault detection of industrial process data. This algorithm is based on the assumption that cluster centers are surrounded by low local density neighbors, and has advantages such as not needing to pre-specify the number of cluster centers and being able to automatically discover clusters of arbitrary shapes.

[0004] However, during the production process of cable processing equipment, the monitoring data exhibits multi-scale variable density distribution characteristics in the feature space due to the influence of multiple stages such as start-up acceleration, steady-state operation, and process switching. The original DPC algorithm typically relies on a globally fixed cutoff distance hyperparameter when calculating local density. This parameter setting cannot simultaneously adapt to the distribution characteristics of both dense and sparse data areas. It is highly susceptible to mismatches between the cutoff distance and the local data distribution, leading to unclear boundaries of normal states under different operating conditions. This results in low accuracy in fault monitoring, making it difficult to meet the high-precision monitoring requirements of precision cable processing. Summary of the Invention

[0005] To address the issue of low detection accuracy of traditional DPC algorithms in cable production equipment, this application provides a method and system for monitoring faults in the processing of cable production equipment.

[0006] Firstly, this application provides a method for monitoring faults in the processing of cable production equipment, employing the following technical solution: A method for monitoring faults in the processing of cable production equipment includes: acquiring multi-dimensional operating parameters of the cable production equipment to obtain a dataset containing multiple sample points; For any sample point in the dataset, determine its neighborhood set; construct the nearest neighbor dispersion of the sample point, which is used to characterize the degree of dispersion between the sample point and other sample points in its neighborhood set; Construct the neighborhood rheological dissimilarity of sample points. The neighborhood rheological dissimilarity is used to characterize the difference between the nearest neighbor dispersion of a sample point and the nearest neighbor dispersion of other sample points in its neighborhood set. Based on nearest neighbor dispersion and neighborhood rheological difference, an adaptive truncation threshold for sample points is determined; and density peak clustering analysis is performed on the dataset using the adaptive truncation threshold to identify processing faults in cable production equipment.

[0007] This method introduces nearest-neighbor dispersion, which characterizes the looseness of local topology, and neighborhood rheological dissimilarity, which characterizes environmental consistency, and constructs an adaptive truncation threshold based on these two indicators. This method dynamically adjusts the truncation distance for each sample point based on the looseness of its local topology and its consistency with the surrounding environment. In data-dense areas, the threshold automatically decreases to more accurately distinguish small fault clusters; in data-sparse areas, the threshold automatically increases to maintain the connectivity of normal transitional data. Compared to traditional methods that rely on a globally fixed truncation distance to determine data anomalies, the method in this application effectively solves the problem of existing technologies' difficulty in distinguishing between normal sparse operating conditions and anomalous outliers, improving the accuracy and robustness of cable processing fault monitoring under complex and variable operating conditions.

[0008] Optionally, the step of constructing the nearest neighbor dispersion of a sample point includes: calculating the root mean square value of the Euclidean distance between the sample point and each sample point in its neighborhood set as the base value of the nearest neighbor dispersion; and multiplying the base value by a preset correction factor as the nearest neighbor dispersion.

[0009] The root mean square (RMS) value is more sensitive to large values ​​than the arithmetic mean, and can more accurately reflect the average radius of dispersion between a sample point and its neighboring sample points, i.e., the average sparsity of the data. Furthermore, a correction factor is introduced to adjust the RMS value, thereby improving the accuracy and robustness of the nearest neighbor dispersion calculation.

[0010] Optionally, the steps for obtaining the correction factor include: taking the difference between the maximum and minimum distances between a sample point and each sample point in its neighborhood set as the distance range; calculating the average distance between a sample point and each sample point in its neighborhood set; taking the ratio of the distance range to the average distance as the adjustment amount; and taking the sum of 1 and the adjustment amount as the correction factor.

[0011] In the initial stages of cable processing faults, parameter fluctuations are often nonlinear, leading to anisotropic or irregular spatial distribution of neighboring points. This increases the difference between the maximum and minimum distances, causing the nearest neighbor dispersion to take into account the uneven distribution of neighbors. When the distribution around a data point is extremely uneven, the correction factor increases, thus weighting and amplifying the nearest neighbor dispersion.

[0012] Optionally, the steps for constructing the neighborhood rheological dissimilarity of a sample point include: taking the mean of the nearest neighbor dispersion of the sample point and other sample points in its neighborhood set as the average dispersion, calculating the absolute value of the difference between the sample point and the average dispersion; determining a nonlinear weighting factor based on the nearest neighbor dispersion of the sample point; and taking the product of the normalized result of the absolute value of the difference and the nonlinear weighting factor as the neighborhood rheological dissimilarity.

[0013] The average dispersion reflects the dispersion level of other data points in the neighborhood of a sample point. The absolute value of the difference between the sample point and the average dispersion is calculated. The larger this absolute value, the more abrupt the current sample point appears compared to other points in its neighborhood, and thus the greater the likelihood of an anomaly. The nonlinear weighting factor can moderately amplify the sensitivity to differences when the sample point is in a highly discrete region, helping to accurately identify abrupt anomalies that disrupt the continuity of the rheological process, thereby reducing the false alarm rate caused by fluctuations in normal operating conditions.

[0014] Optionally, the step of determining the adaptive cutoff threshold of a sample point based on the nearest neighbor dispersion and the neighborhood rheological difference includes: for any sample point, the product of the nearest neighbor dispersion and the gain coefficient of the sample point is used as the adaptive cutoff threshold of the sample point, wherein the gain coefficient is a monotonically increasing function of the neighborhood rheological difference.

[0015] For normal sparse transition points with small differences, the gain coefficient prompts a moderate increase in the cutoff distance, expanding the search radius to ensure the connectivity of the sparse cluster and prevent normal fluctuations from being misjudged as breakpoints. For anomalous outliers with large differences, this mechanism limits their search range, making it difficult for them to find high-density neighbors in subsequent local density calculations, thus highlighting their isolation.

[0016] Optionally, for any sample point in the dataset, the step of determining its neighborhood set includes: calculating the Euclidean distance between any data point and all other data points; For any data point, select a preset number of data points that are closest to it in Euclidean distance as its neighbors, and multiple neighbors form a neighborhood set.

[0017] The K-nearest neighbor algorithm based on Euclidean distance is used to determine the neighborhood set, ensuring the objectivity and consistency of local topology analysis.

[0018] Optionally, the multi-dimensional operating parameters of the cable production equipment include at least one of the following: cable outer diameter, cable ellipticity, production line traction speed, extruder screw speed, barrel temperature, and die head melt pressure.

[0019] Optionally, in the step of determining the neighborhood set of any sample point in the dataset, the preset number range is 1% to 2% of the total number of sample points.

[0020] Optionally, after obtaining the multi-dimensional operating parameters of the cable production equipment, the process may also include the step of performing Z-score standardization on the obtained multi-dimensional operating data.

[0021] The data was standardized to eliminate the impact of differences in the dimensions of data such as temperature, rotation speed, and size.

[0022] Secondly, this application provides a process fault monitoring system for cable production equipment, which adopts the following technical solution: A process fault monitoring system for cable production equipment includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the process fault monitoring method for cable production equipment described above is implemented.

[0023] The above-mentioned method for monitoring the faults in the processing of cable production equipment is generated into a computer program and stored in a memory so that it can be loaded and executed by a processor. Thus, a system can be built based on the memory and the processor for convenient use.

[0024] This application has the following technical advantages: By constructing nearest-neighbor dispersion and neighborhood rheological difference, the local sparsity of the data and environmental consistency are reflected respectively. Based on this, an adaptive cutoff distance is dynamically generated for each sample point, which enables the effective differentiation between normal sparse transition states and abnormal outliers in the subsequent detection process, thereby improving the accuracy of fault monitoring under complex working conditions. Attached Figure Description

[0025] Figure 1 This is a flowchart of a method for monitoring faults in the processing of cable production equipment according to an embodiment of this application.

[0026] Figure 2 This is an adaptive cutoff distance map corresponding to different sample points in a method for monitoring faults in the processing of cable production equipment according to an embodiment of this application.

[0027] Figure 3 This is a comparison chart of the detection results of the traditional density peak clustering algorithm and the density peak clustering algorithm in this application. Detailed Implementation

[0028] This application discloses a method for monitoring faults in the cable manufacturing process. By collecting multi-dimensional data from the cable manufacturing process, it constructs a nearest-neighbor dispersion measure reflecting the looseness of the local topological structure of the data, and a neighborhood rheological difference measure reflecting the consistency between a data point and its surrounding environment. Based on this, an adaptive truncation threshold index is constructed, dynamically assigning a unique truncation distance to each data point. This improves the local density calculation method of the DPC algorithm, solving the problem that traditional algorithms cannot adapt to multi-scale variable density distributions, and achieving accurate monitoring of faults in the cable manufacturing process.

[0029] Reference Figure 1 A method for monitoring faults in the processing of cable production equipment includes steps S1-S5.

[0030] S1: Obtain multi-dimensional operating parameters of the cable production equipment to obtain a dataset containing multiple sample points.

[0031] Taking a typical cable extrusion production line as an example, real-time operating data is collected by deploying a variety of high-precision sensors.

[0032] Specifically, a non-contact laser diameter gauge is installed behind the cooling water tank. The non-contact laser diameter gauge can be a laser scanning diameter gauge with bidirectional detection on the X and Y axes, which can collect the outer diameter data and ellipticity data of the cable in real time to monitor the stability of the extrusion size.

[0033] A precision tension sensor is installed at the traction wheel assembly to collect the real-time tension value of the cable, in order to prevent structural damage caused by the cable being too loose or too tight.

[0034] A rotary encoder or Hall effect sensor is installed on the motor shaft to collect the extruder screw speed and production line traction speed in order to monitor the matching of extrusion rate and traction rate.

[0035] Thermocouple sensors are distributed throughout the extruder barrel, including the feeding, compression, and metering sections, as well as the die head, to collect real-time temperature data for each temperature zone, ensuring the rheological properties of the plastic melt. A melt pressure sensor is installed at the die head to collect melt pressure data, reflecting the screw's conveying stability. In this embodiment, the data acquisition system synchronously collects data from all the aforementioned sensors at 1-second intervals.

[0036] For ease of description, the above outer diameter data, ellipticity data, real-time tension values, extruder screw speed, production line traction speed, and real-time temperature data for each temperature zone are collectively referred to as raw data. Since the collected raw data often contains environmental noise and dimensional differences, preprocessing is required. First, Gaussian filtering is used to smooth the raw data sequence. Gaussian filtering uses a Gaussian function as a weighting function to perform a weighted average of the data, effectively suppressing high-frequency random noise caused by electromagnetic interference while preserving the main trend characteristics of the data. Second, considering the significant differences in the physical dimensions and orders of magnitude of data such as temperature (°C), speed (r / min), and diameter (mm), directly calculating the distance would lead to larger numerical features dominating the results and masking the changes in smaller numerical features. Therefore, Z-score normalization (Standard Score) is used to process the denoised data, mapping all dimensional feature data to the same distribution space with a mean of 0 and a standard deviation of 1, thus obtaining a standardized dataset containing multiple sample points, where each sample point includes multi-dimensional data.

[0037] S2: For any sample point in the dataset, determine its neighborhood set; construct the nearest neighbor dispersion of the sample point, which is used to characterize the degree of dispersion between the sample point and other sample points in its neighborhood set.

[0038] During cable extrusion, the melt flow is stable under normal steady-state production, and the data points are densely distributed in the feature space. However, when minor faults occur (such as feed pulsation) or during transitional phases, the density between data points decreases due to the extrusion expansion effect of the material and melt fracture, resulting in uneven local density. To measure the degree of local topological loosening caused by changes in operating conditions, nearest-neighbor dispersion is constructed in this step.

[0039] First, a KNN (K-Nearest Neighbors) search is performed on the dataset containing multiple sample points. For each sample point in the dataset, its Euclidean distance to all other points is calculated, and a predetermined number of the closest sample points are selected as its neighborhood set in ascending order of distance. In this embodiment, The value can be set according to the data size, for example, it can be 1% to 2% of the total data.

[0040] Based on the KNN search results, a nearest neighbor dispersion is constructed. The root mean square value of the Euclidean distance between a sample point and each sample point in its neighborhood set is calculated as the base value of the nearest neighbor dispersion. The product of the base value and a preset correction factor is used as the nearest neighbor dispersion.

[0041] Specifically, the formula for calculating the nearest neighbor dispersion can be expressed as: In the formula, Indicates the first The nearest neighbor dispersion of each sample point; The number of nearest neighbors can also be understood as the number of the first nearest neighbors. For ease of description, in this embodiment, the sample points in the neighborhood set corresponding to each sample point are referred to as the nearest neighbor points, and the number of sample points in the neighborhood set is referred to as the number of nearest neighbors. For sample points The set of K nearest neighbors; Represents sample points Its first Euclidean distance between nearest neighbors; Represents sample points The maximum distance between the nearest neighbor; Represents sample points The minimum distance to its nearest neighbor; and Represents sample points The average distance between the nearest neighbor.

[0042] The formula consists of two parts, the first half of the formula... The root mean square value of the distance within the neighborhood was calculated, which reflects the sample point. The average radius of dispersion between the data cluster and its neighboring points. The larger this value, the sparser the data cluster.

[0043] The second half of the formula It is a dimensionless correction factor. In the early stages of cable processing faults, parameter fluctuations often exhibit nonlinearity, leading to spatial anisotropy or irregularity in the distribution of neighboring points, thus amplifying the difference between the maximum and minimum distances. Introducing this correction factor allows for a weighted amplification of the average discrete radius by utilizing the non-uniformity of the neighboring distribution. Therefore, The indicator not only considers the average distance but also incorporates the morphological characteristics of the distribution. The larger the nearest neighbor dispersion value, the more unstable the operating conditions at that point and the looser the local topology of the data. Compared to calculating only the average distance, this indicator can more sensitively capture early fault characteristics.

[0044] S3: Construct the neighborhood rheological dissimilarity of the sample point. The neighborhood rheological dissimilarity is used to characterize the difference between the nearest neighbor dispersion of the sample point and the nearest neighbor dispersion of other sample points in its neighborhood set.

[0045] The cable manufacturing process exhibits rheological characteristics of thermal inertia hysteresis and shear thinning. During rapid speed changes or gradual faults such as filter clogging, the data distribution shows a gradual gradient change. This means that normal sparse points are usually surrounded by other sparse points, while the density structure around abnormal points differs significantly from their own. Based solely on… It is impossible to distinguish between normal sparse operating conditions and anomalous outliers. Therefore, this step constructs a neighborhood rheological dissimilarity measure to analyze the continuity of the discreteness between the sample points and their surrounding environment.

[0046] In this step, the mean of the nearest neighbor dispersion of a sample point and other sample points in its neighborhood set is used as the average dispersion, and the absolute value of the difference between the sample point and the average dispersion is calculated; the nonlinear weighting factor is determined based on the nearest neighbor dispersion of the sample point; the product of the normalized result of the difference and the nonlinear weighting factor is used as the neighborhood rheological difference degree.

[0047] Specifically, the formula for calculating the neighborhood rheological difference of any sample point can be expressed as: In the formula, Indicates the first The neighborhood rheological difference of each sample point; For the first The nearest neighbor dispersion of each sample point; Represents sample points The arithmetic mean of the nearest neighbor dispersion of multiple neighbor points; It is a natural constant; It is the natural logarithm.

[0048] molecular It directly measures the difference between the dispersion of a sample point itself and the average dispersion of its surrounding neighbors. If the difference is small, it indicates that the point is consistent with the environment (e.g., both are in a sparse transition zone); if the difference is large, it indicates that the point is abrupt (it may be an outlier). It is mainly used to normalize the molecular part. By leveraging the compression properties of the logarithmic function, a nonlinear adjustment for its own dispersion is introduced. When the nearest neighbor dispersion of a sample point is large, this term will moderately amplify the dissimilarity, because the structure of a high-dispersion region is inherently more unstable and requires higher sensitivity. Through this index, A larger value indicates that the continuity of the rheological process has been disrupted at that point, representing an abrupt change; conversely, a smaller value indicates a normal rheological process. This effectively eliminates the interference of pseudo-sparseness, improving the algorithm's adaptability to complex working conditions.

[0049] Step S4: Determine the adaptive cutoff threshold for sample points based on nearest neighbor dispersion and neighborhood rheological difference.

[0050] For any sample point, an adaptive truncation threshold is constructed based on its corresponding nearest-neighbor dispersion and neighborhood rheological dissimilarity. In this embodiment, the product of the nearest-neighbor dispersion and the gain coefficient of the sample point is used as the adaptive truncation threshold of the sample point, where the gain coefficient is a monotonically increasing function of the neighborhood rheological dissimilarity.

[0051] The formula for constructing the adaptive truncation threshold index can be expressed as: In the formula, Indicates the first Adaptive cutoff distance for each sample point; For the first The nearest neighbor dispersion of each sample point; No. The neighborhood rheological difference of each sample point; It is an exponential function with the natural constant as its base.

[0052] For sparse regions, the nearest neighbor dispersion is large. Therefore, the search radius for a point when looking for density contributions increases to ensure connectivity within the sparse region and prevent the disruption of normal sparse clusters. (Exponential term) It is a gain coefficient. When When the value is small, the exponent term approaches 1. Mainly composed of Decision. When When the value is large, the exponent term increases, thus appropriately amplifying the cutoff distance. Using the above method, the search range can be expanded for points in boundary or anomalous states; if it is a normal transition point, it is easier to establish connections with surrounding potential clusters; if it is an outlier, even expanding the radius makes it difficult to find high-density neighbors, thus highlighting its isolation in subsequent density calculations.

[0053] S5: Perform density peak clustering analysis on the dataset using an adaptive truncation threshold to identify processing faults in cable production equipment.

[0054] Utilizing adaptive truncation threshold The traditional DPC algorithm is improved and fault monitoring is implemented.

[0055] Specifically, during the detection process, the formula for calculating the local density of each sample point can be expressed as: ; Indicates the first Local density of each sample point; It is an exponential function with the natural constant as its base; Indicates the first Adaptive cutoff distance for each sample point; combined with Figure 2 Each point Each corresponds to an adaptive cutoff distance. In dense data areas, the adaptive cutoff distance automatically decreases, enabling the algorithm to accurately distinguish closely adjacent clusters of small faults. In sparse data areas, the adaptive cutoff distance automatically increases, ensuring the continuity of transitional data and avoiding misjudging normal fluctuations as breakpoints. Represents sample points With sample points The Euclidean distance between them.

[0056] Next, the relative distance to each point is calculated, which is the minimum distance to points with a higher density than itself. For the point with the highest density, its... Take the maximum value of the distances between this point and all other sample points.

[0057] Finally, a decision map is generated based on local density and relative distance, and fault determination is performed. The specific determination process follows conventional techniques in this field and will not be elaborated upon here. Figure 3 In the same dataset (where all data points are labeled, meaning both normal and abnormal data are known, and the decision graph obtained using the traditional DPC algorithm), some normal data points are incorrectly identified as abnormal, leading to a large number of anomalous data points and false positives. However, the DPC algorithm in this application reduces false positives in the resulting decision graph, improving detection accuracy.

[0058] This application improves the DPC algorithm by introducing an adaptive truncation threshold, effectively solving the monitoring problem caused by the multi-scale density distribution of cable production data, and improving the accuracy and robustness of fault monitoring.

[0059] This application also discloses a process fault monitoring system for cable production equipment, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a process fault monitoring method for cable production equipment according to this application is implemented.

[0060] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0061] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method for monitoring faults in the processing of cable manufacturing equipment, characterized in that, Obtain multi-dimensional operating parameters of cable production equipment to obtain a dataset containing multiple sample points; For any sample point in the dataset, determine its neighborhood set; construct the nearest neighbor dispersion of the sample point, which is used to characterize the degree of dispersion between the sample point and other sample points in its neighborhood set; Construct the neighborhood rheological dissimilarity of sample points. The neighborhood rheological dissimilarity is used to characterize the difference between the nearest neighbor dispersion of a sample point and the nearest neighbor dispersion of other sample points in its neighborhood set. Based on nearest neighbor dispersion and neighborhood rheological difference, an adaptive truncation threshold for sample points is determined. And density peak clustering analysis is performed on the dataset using an adaptive truncation threshold to identify processing faults in cable production equipment; The steps for constructing the neighborhood rheological dissimilarity of a sample point include: taking the mean of the nearest neighbor dispersion of the sample point and other sample points in its neighborhood set as the average dispersion, and calculating the absolute value of the difference between the sample point and the average dispersion. The nonlinear weighting factor is determined based on the nearest neighbor dispersion of the sample points; the product of the normalized result of the absolute value of the difference and the nonlinear weighting factor is used as the neighborhood rheological difference.

2. The method for monitoring process faults in cable production equipment according to claim 1, characterized in that, The steps for constructing the nearest neighbor dispersion of a sample point include: calculating the root mean square value of the Euclidean distance between the sample point and each sample point in its neighborhood set as the base value of the nearest neighbor dispersion; and multiplying the base value by a preset correction factor as the nearest neighbor dispersion.

3. The method for monitoring process faults in cable production equipment according to claim 2, characterized in that, The steps for obtaining the correction factor include: taking the difference between the maximum and minimum distances between a sample point and each sample point in its neighborhood as the distance range; calculating the average distance between a sample point and each sample point in its neighborhood; taking the ratio of the distance range to the average distance as the adjustment amount; and taking the sum of 1 and the adjustment amount as the correction factor.

4. The method for monitoring process faults in cable production equipment according to claim 1, characterized in that, The steps for determining the adaptive cutoff threshold of a sample point based on the nearest neighbor dispersion and the neighborhood rheological difference include: for any sample point, the product of the nearest neighbor dispersion and the gain coefficient of the sample point is used as the adaptive cutoff threshold of the sample point, wherein the gain coefficient is a monotonically increasing function of the neighborhood rheological difference.

5. The method for monitoring process faults in cable production equipment according to claim 1, characterized in that, For any sample point in the dataset, the steps to determine its neighborhood set include: calculating the Euclidean distance between any data point and all other data points; For any data point, select a preset number of data points that are closest to it in Euclidean distance as its neighbors, and multiple neighbors form a neighborhood set.

6. The method for monitoring process faults in cable production equipment according to claim 1, characterized in that, The multi-dimensional operating parameters of cable production equipment include at least one of the following: cable outer diameter, cable ellipticity, production line traction speed, extruder screw speed, barrel temperature, and die head melt pressure.

7. The method for monitoring process faults in cable production equipment according to claim 5, characterized in that, For any sample point in the dataset, in the step of determining its neighborhood set, the preset quantity range is 1% to 2% of the total number of sample points.

8. The method for monitoring process faults in cable production equipment according to claim 1, characterized in that, After obtaining the multi-dimensional operating parameters of the cable production equipment, the following steps are also included: performing Z-score standardization processing on the obtained multi-dimensional operating data.

9. A process fault monitoring system for cable production equipment, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a process fault monitoring method for cable production equipment according to any one of claims 1-8.

Images