Spinning machine fault monitoring method and system

Abstract

The invention relates to the technical field of data processing, in particular to a spinning machine fault monitoring method and system. The method comprises the following steps: acquiring original data in a monitoring data sequence, performing iterative expansion on the original data according to the first final expansion inevitability of the original data to acquire a first data segment, and dividing the monitoring data sequence into a plurality of data segments by parity of reasoning; constructing a box plot for each dimension of data of the monitoring data in each data segment to obtain a box plot to which each dimension of data of each monitoring data belongs; obtaining the abnormal degree of each piece of monitoring data, obtaining the fluctuation condition of neighborhood data of each piece of monitoring data, correcting the abnormal degree of each piece of monitoring data, and obtaining the corrected abnormal degree of each piece of monitoring data; and judging the fault of the spinning machine based on the corrected abnormal degree, so that the box plot can monitor the abnormal data more accurately, and the abnormal detection result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a spinning machine fault monitoring method and system. Background Art

[0002] Once a spinning machine fails, if it cannot be discovered and repaired in time, it will lead to a long period of downtime. The fault monitoring system can monitor the operating status of the spinning machine in real time, discover potential fault hazards in time, and enable maintenance personnel to monitor the status of the spinning machine in real time, perform maintenance in advance or repair it in the early stage of the failure, thereby avoiding the reduction of the quality of the produced yarn. For example, it can timely detect intermittent faults in the operation of the spinning machine (poor circuit contact, loose mechanical parts) and perform shutdown monitoring to avoid further damage to the machine. Therefore, it is necessary to collect monitoring data during the operation of the spinning machine, including vibration data and temperature data, and monitor the abnormal data in the collected monitoring data, so as to timely detect whether intermittent abnormalities occur during the operation of the spinning machine.

[0003] The patent application document with the publication number CN113626497A currently discloses a semiconductor problem machine positioning system and positioning method based on a box plot, which is divided into the following modules: a data acquisition module, which collects the original yield data of each machine in each site; a box plot drawing module, which makes a box plot corresponding to each machine based on the original yield data of each machine collected by the data acquisition module; a box plot analysis module, which counts the yield abnormal values ​​of each machine and finds out the machines with concentrated yield abnormal values.

[0004] When all the collected monitoring data are normal data, the distribution of the monitoring data may be skewed (for example, the spinning machine processes long-fiber raw materials in a shorter period of time and short-fiber raw materials in a longer period of time). Therefore, when constructing a box plot for all the collected monitoring data to detect abnormal data, the monitoring data generated when processing long-fiber raw materials in a shorter period of time may be identified as abnormal values. In addition, there is a small amount of abnormal data caused by intermittent machine failures in the monitoring data collected when processing short-fiber raw materials for a long time, which may be judged as normal data, affecting the accuracy of abnormal data monitoring. Summary of the invention

[0005] In order to solve the technical problem that when the distribution of monitoring data may belong to skewed distribution, a box plot is inaccurate when abnormal data is analyzed, the present invention provides a spinning machine fault monitoring method and system.

[0006] In a first aspect, the present invention provides a spinning machine fault monitoring method, which adopts the following technical solution: A method for monitoring faults of a spinning machine comprises the following steps: Collect monitoring data; sequentially obtain the data of the first N sampling moments in the monitoring data sequence as the original data, and obtain the first final expansion necessity of the original data ; Represents the fluctuation of the first iteration data; represents the necessity of the first expansion of the original data; norm() represents the normalization function; based on the final expansion necessity, the original data is iteratively expanded to obtain the first data segment; the data of N sampling moments after the first data segment are iteratively expanded as new original data to obtain the second data segment, and so on, until there is no data in the monitoring data sequence, and then multiple data segments are obtained; a box plot is constructed for each dimension of the monitoring data in each data segment to obtain multiple box plots; Get the abnormality level of each monitoring data , Represents the abnormality degree of the bth monitoring data; as well as The distance between the vibration data and the temperature data of the b-th monitoring data and the nearest boundary of the box plot to which they belong; represents the anomaly weight of the temperature data; Representing the abnormal weight of the vibration data; obtaining the corrected abnormality degree of each monitoring data, wherein the corrected abnormality degree is proportional to the fluctuation of the neighborhood data of the monitoring data; judging the fault of the spinning machine according to the corrected abnormality degree of each monitoring data.

[0007] The innovation of the present invention lies in obtaining the original data, mapping the data before and after the expansion of the original data into a two-dimensional rectangular coordinate system, observing the change of the centroid of the original data before and after the expansion and whether the original data before and after the expansion has a new extension direction, so as to obtain the first final expansion necessity of the original data, iteratively expand the original data based on the final expansion necessity, obtain the first data segment, and divide the monitoring data into multiple data segments by analogy, so that the data in each data segment does not have a skewed distribution, thereby improving the accuracy of subsequent anomaly detection; further, constructing a box plot for each dimension of the monitoring data in each data segment, obtaining the degree of abnormality of each monitoring data, and then correcting the degree of abnormality according to the fluctuation of the neighborhood data of the monitoring data to perform abnormal data identification, which can distinguish the noise data from the abnormal data in the monitoring data and improve the anomaly detection result.

[0008] If the centroid of the original data before and after expansion changes and has a new extension direction, it means that the expansion of the original data is necessary to be small.

[0009] Preferably, the necessity of obtaining the first expansion of the original data includes: ; In the formula, Represents the necessity of the first expansion of the original data; Represents the distance between the centroid of the original data and the centroid of the first updated data; The value of the i-th dimension data representing the centroid of the original data; Represents the value of the i-th dimension data of the centroid of the first updated data; Represents the number of dimensions of the data; || represents the absolute value symbol; The sum of the distances from the centroid of the convex hull region representing the original data to each vertex of the convex hull region; Represents the sum of the distances from the centroid of the convex hull area of ​​the first updated data to each vertex of its convex hull area; Represents the mean of the first updated data; Represents the median of the first updated data; exp() represents an exponential function with a natural constant as the base.

[0010] According to the deviation between the centroids before and after the expansion of the original data and whether there is a new extension direction, the necessity of the first expansion of the original data is obtained, so as to facilitate the subsequent determination of whether the original data needs to be expanded.

[0011] Preferably, the step of obtaining the centroid of the original data and the centroid of the first updated data includes: The number of sampling moments N is preset, and a two-dimensional rectangular coordinate system is constructed with amplitude as the horizontal coordinate and temperature as the vertical coordinate. The original data is mapped into the two-dimensional rectangular coordinate system and a convex hull detection is performed on it to obtain the convex hull area of ​​the original data, and the centroid of the convex hull area of ​​the original data is obtained and recorded as the centroid of the original data; M data after the Nth sampling moment in the monitoring data sequence are recorded as the first iteration data, which are added to the original data to obtain the first update data, and the first update data is mapped into the two-dimensional rectangular coordinate system, and a convex hull detection is performed on the first update data in the two-dimensional rectangular coordinate system to obtain the convex hull area of ​​the first update data, and the centroid of the convex hull area of ​​the first update data is recorded as the centroid of the first update data in the two-dimensional rectangular coordinate system.

[0012] This makes it easier to determine whether the original data needs to be updated based on the distance between the centroids.

[0013] Preferably, the acquisition of the fluctuation of the first iteration data includes: Use the least square method to fit a curve to the i-th dimension data of the first iteration data, and obtain the fitting value of each i-th dimension data of the first iteration data; ; In the formula, Represents the fluctuation of the first iteration data; Represents the number of dimensions of the data; Represents the sum of the fitted values ​​of all i-th dimension data of the first iteration data; Represents the sum of the actual values ​​of all i-th dimension data of the first iteration data; Represents the standard deviation of all i-th dimension data of the first iteration data; norm() represents the normalization function.

[0014] It reflects the fluctuation of the data to be added to the original data, and the subsequent first expansion of the original data will inevitably be corrected to make it more accurate.

[0015] Preferably, the iterative expansion of the original data based on the final expansion necessity to obtain the first data segment includes: A threshold value of expansion necessity is preset. If the first final expansion necessity of the original data is greater than the expansion necessity threshold, the first iteration data is added to the original data, and M data after the first iteration data are used as the second iteration data. The second iteration data is added to the original data to obtain the second updated data. According to the second updated data and the original data, the second final expansion necessity of the original data is obtained to expand the original data. The process stops when the mth final expansion necessity of the original data is less than the expansion necessity threshold, and the final expanded original data is used as the first data segment.

[0016] Preferably, obtaining the corrected abnormality degree of each monitoring data includes: Preset the number of neighborhood data U, obtain U data before each monitoring data and U data after each monitoring data as the neighborhood data of each monitoring data, and obtain the fluctuation of the neighborhood data of each monitoring data according to the calculation formula of the fluctuation of the first iteration data; ; In the formula, Represents the corrected abnormality degree of the bth monitoring data; Represents the abnormality degree of the bth monitoring data; Represents the fluctuation of the neighborhood data of the bth monitoring data; norm() represents the normalization function.

[0017] It can eliminate the influence of noise data on abnormal data identification.

[0018] Preferably, judging the fault of the spinning machine according to the degree of correction abnormality of each monitoring data comprises: A correction abnormality threshold T2 is preset. If the correction abnormality of any monitoring data is greater than the correction abnormality threshold T2, the monitoring data will be regarded as abnormal data. At this time, the system will issue an early warning and the staff will inspect the spinning machine.

[0019] In a second aspect, the present invention provides a spinning machine fault monitoring system, which adopts the following technical solution: A spinning machine fault monitoring system comprises: 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 spinning machine fault monitoring method is implemented.

[0020] By adopting the above technical solution, the above-mentioned spinning machine fault monitoring method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0021] The present invention has the following technical effects: the purpose of the present invention is to obtain original data, and by mapping the data before and after the expansion of the original data into a two-dimensional rectangular coordinate system, the first final expansion necessity of the original data is obtained according to the change of the centroid of the original data before and after the expansion and whether the original data has a new extension direction, the original data is iteratively expanded based on the final expansion necessity to obtain the first data segment, and so on, the monitoring data is divided into multiple data segments, so that the data in each data segment does not have a skewed distribution, thereby improving the accuracy of subsequent anomaly detection; further, a box plot is constructed for each dimension of the monitoring data in each data segment, the degree of abnormality of each monitoring data is obtained, and then the degree of abnormality is corrected according to the fluctuation of the neighborhood data of the monitoring data to perform abnormal data identification, which can distinguish the noise data from the abnormal data in the monitoring data, thereby improving the anomaly detection result. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-restrictive manner, and the same or corresponding reference numerals represent the same or corresponding parts.

[0023] Figure 1 The present invention is a flowchart of a method for monitoring faults of a spinning machine according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0025] It should be understood that when the terms "first", "second", etc. are used in the claims, descriptions, and drawings of the present invention, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their collections.

[0026] The embodiment of the present invention discloses a spinning machine fault monitoring method, referring to Figure 1 , comprising steps S1 to S3: S1: Collect monitoring data.

[0027] It should be noted that it is known that when a spinning machine is operating stably, the energy it consumes is relatively fixed, but when a fault occurs, the components need to overcome greater resistance during operation, which will result in more energy consumption. This energy consumption generally causes changes in the amplitude, and the generation of heat is inevitable during the operation of the spinning machine. For example, the friction between mechanical components, the rolling friction inside the bearings, the gear meshing friction, etc. will also generate heat. Therefore, each component has its normal operating temperature range, but when a fault occurs, the temperature will exceed the normal temperature range. Therefore, the vibration data and temperature data during the operation of the spinning machine can both reflect whether the spinning machine has a fault. It is necessary to consider the characteristics of the two types of data at the same time and conduct joint abnormal monitoring.

[0028] In an embodiment of the present invention, the preset sampling time is 2 seconds / time, and vibration sensors and temperature sensors are installed at key positions of the spinning machine to collect vibration data and temperature data at the same time. The vibration data and temperature data at each sampling time are used as one monitoring data, and the one monitoring data contains two dimensional data. The monitoring data at each sampling time are sorted in turn to obtain a monitoring data sequence.

[0029] S2: Obtain the original data in the monitoring data sequence, iteratively expand the original data according to the necessity of the first final expansion of the original data, obtain the first data segment, and so on, divide the monitoring data sequence into multiple data segments; construct a box plot for each dimension of the monitoring data in each data segment, and obtain the box plot of each dimension of each monitoring data; obtain the abnormality degree of each monitoring data according to the box plot of each dimension of each monitoring data.

[0030] It should be noted that it is known that during the operation of the spinning machine, the characteristics of the fiber raw materials processed by the spinning machine may change at any time, such as long fiber raw materials, short fiber raw materials, uneven fiber thickness or the humidity of the fiber raw materials. Therefore, when the characteristics of the fiber raw materials processed by the spinning machine change, the interaction between the fiber raw materials and the spinning machine will change, and the collected monitoring data will also change, such as the increase of vibration data and temperature data for a period of time. Therefore, if all the collected monitoring data are normal data, the distribution of the monitoring data may be skewed (for example, the spinning machine processes long fiber raw materials in a shorter period of time and processes short fiber raw materials in a longer period of time). Therefore, when constructing a box plot for all the collected monitoring data to detect abnormal data, the monitoring data generated when processing long fiber raw materials in a shorter period of time may be identified as abnormal values, and there is a small amount of abnormal data (such as local overheating or vibration abnormality) caused by intermittent machine failures in the monitoring data collected when processing short fiber raw materials for a long time. It is very likely to be judged as normal data, affecting the accuracy of abnormal data monitoring. Therefore, in an embodiment of the present invention, the monitoring data sequence is divided into several segments according to the distribution of the collected monitoring data, and anomaly detection is performed on the monitoring data in each segment respectively, so as to improve the accuracy of anomaly detection.

[0031] Step S2 includes steps S20 and S21, which are specifically as follows: S20: obtaining original data in the monitoring data sequence, and expanding the original data according to the necessity of the first final expansion of the original data to obtain the first data segment, and so on, dividing the monitoring data sequence into multiple data segments.

[0032] It should be noted that when segmenting the data in the monitoring data sequence, it is necessary to ensure that the data in each segment does not have a skewed distribution, so as to facilitate the subsequent improvement of the accuracy of abnormal data detection. It is known that the monitoring data collected initially is more likely to be generated by processing the same type of raw materials, and the processed raw materials may be changed in the middle. Therefore, in the present invention, the original data is first obtained according to the monitoring data sequence, and then the remaining data in the monitoring data sequence is added to the original data in turn. According to the data distribution before and after the original data, the necessity of expansion of the original data is obtained, and the original data is iteratively expanded according to the necessity of expansion to obtain the first data segment, and so on, the monitoring data sequence is divided into multiple data segments.

[0033] It should be further explained that the necessity of expanding the original data can be obtained by mapping the data before and after the expansion of the original data into a two-dimensional rectangular coordinate system, observing the changes in the center of mass of the original data before and after the expansion and whether the original data before and after the expansion has a new extension direction. If the center of mass of the original data before and after the expansion has changed and has a new extension direction, it means that the expansion necessity of the original data is relatively small.

[0034] In the embodiment of the present invention, the number of sampling moments N and the number of iterative data M are preset, and the data of the first N sampling moments in the monitoring data sequence are sequentially obtained as the original data; A two-dimensional rectangular coordinate system is constructed with amplitude as the horizontal coordinate and temperature as the vertical coordinate, the original data is mapped into the two-dimensional rectangular coordinate system and a convex hull detection is performed on the system to obtain the convex hull area of ​​the original data, and the centroid of the convex hull area of ​​the original data is obtained and recorded as the centroid of the original data; Record the M data after the Nth sampling moment in the monitoring data sequence as the first iteration data, add them to the original data to obtain the first update data, map the first update data to a two-dimensional rectangular coordinate system, perform convex hull detection on the first update data in the two-dimensional rectangular coordinate system to obtain the convex hull area of ​​the first update data, and record the centroid of the convex hull area of ​​the first update data as the centroid of the first update data in the two-dimensional rectangular coordinate system; Obtain the first expansion necessity of the original data: ; In the formula, Represents the necessity of the first expansion of the original data; Represents the distance between the centroid of the original data and the centroid of the first updated data; The value of the i-th dimension data representing the centroid of the original data; Represents the value of the i-th dimension data of the centroid of the first updated data; Represents the number of dimensions of the data; || represents the absolute value symbol; The sum of the distances from the centroid of the convex hull region representing the original data to each vertex of the convex hull region; Represents the sum of the distances from the centroid of the convex hull area of ​​the first updated data to each vertex of its convex hull area; Represents the mean of the first updated data; Represents the median of the first updated data; exp() represents an exponential function with a natural constant as the base; The larger the value is, the more likely it is that adding some data to the original data may cause the data distribution to be skewed, affecting the subsequent box plot's monitoring of abnormal data in the original data. At this time, the first expansion of the original data is less inevitable, and there is no need to expand the original data. The larger the value is, the more the original data is extended after adding some data. This may cause the original data to present a skewed distribution, affecting the subsequent box plot to monitor abnormal data in the original data. At this time, the first expansion of the original data is less inevitable. The smaller the value of , the more there is no skewed distribution in the data distribution of the first updated data. At this time, the first expansion of the original data is more inevitable and the more it needs to be expanded.

[0035] It should be noted that the necessity of the first expansion of the original data represents whether the first iteration data can be included in the original data. If the first iteration data and the original data are collected under different fiber raw materials, the necessity of expansion is small at this time, that is, the necessity of the first expansion of the original data is relatively accurate, and the adjustment degree is small. At this time, the first iteration data is in a state of gradual increase, decrease or stability at the same time. If the first iteration data and the original data are data collected under the same fiber raw material, the expansion necessity at this time is large, that is, the first expansion necessity of the original data obtained is relatively accurate, and the adjustment degree is small, and the first iteration data is in a relatively stable state at this time; However, if the first iteration data and the original data are collected under the same fiber raw material and the first iteration data contains some abnormal data, it means that this part of the abnormal data is generated when processing the same batch of raw materials. The first iteration data should be included in the original data for abnormal monitoring to be more accurate. However, at this time, the necessity of expansion may be relatively small due to the influence of abnormal data, so it needs to be adjusted larger, and the first iteration data fluctuates at this time; therefore, it is necessary to analyze the fluctuation of the first iteration data and adjust the necessity of the first expansion of the original data.

[0036] In the embodiment of the present invention, a least square method is used to fit a curve to the i-th dimension data of the first iteration data to obtain a fitting value of each i-th dimension data of the first iteration data; Get the fluctuation of the first iteration data: ; In the formula, Represents the fluctuation of the first iteration data; Represents the number of dimensions of the data; Represents the sum of the fitted values ​​of all i-th dimension data of the first iteration data; Represents the sum of the actual values ​​of all i-th dimension data of the first iteration data; Represents the standard deviation of all i-th dimension data of the first iteration data; norm() represents the normalization function; The larger the value of , the greater the fluctuation of the first iteration data; Obtain the first final augmentation necessity of the original data: ; In the formula, Represents the necessity of the first final expansion of the original data; Represents the fluctuation of the first iteration data; Represents the necessity of the first expansion of the original data; norm() represents the normalization function; Plus 1 means plus The value of is mapped to the interval (1, 2) and can be corrected accurately.

[0037] A threshold value T of expansion necessity is preset. If the first final expansion necessity of the original data is greater than the expansion necessity threshold, the first iteration data is added to the original data, and M data after the first iteration data is used as the second iteration data. The second iteration data is added to the original data to obtain the second update data. According to the second update data and the original data, the second final expansion necessity of the original data is obtained to expand the original data, and the original data is expanded until the mth final expansion necessity of the original data is less than the expansion necessity threshold. The original data finally expanded is used as the first data segment. Then, the data at N sampling moments after the first data segment are iteratively expanded as new original data to obtain the second data segment, and so on, until there is no data in the monitoring data sequence, and the monitoring data sequence is divided into multiple data segments.

[0038] S21: construct a box plot for each dimension of the monitoring data in each data segment to obtain the box plot of each dimension of each monitoring data; obtain the abnormality degree of each monitoring data according to the box plot of each dimension of each monitoring data.

[0039] It should be noted that the above steps divide the monitoring data sequence into multiple data segments according to the distribution of the monitoring data, ensuring that the data in each data segment does not have a skewed distribution. Next, it is necessary to construct a box plot for the data in each data segment and perform anomaly detection on the data in each data segment. However, since the monitoring data is two-dimensional data, it is necessary to construct a box plot for each dimension of all the monitoring data in each data segment for anomaly detection, and obtain the degree of abnormality of each monitoring data based on the abnormal performance of each dimension of each monitoring data in its box plot.

[0040] In an embodiment of the present invention, a box plot is constructed for each dimension of data in each data segment to obtain multiple box plots, that is, a box plot belonging to each dimension of data of each monitoring data is obtained. If any dimension of data of any monitoring data is located between two boundaries of the box plot to which it belongs, the distance between each dimension of the monitoring data and the nearest boundary of the box plot to which it belongs is obtained. If any dimension of data of any monitoring data is located between two boundaries of the box plot to which it belongs, the distance between the dimension of the monitoring data and the nearest boundary of the box plot to which it belongs is 0; It should be noted that the two boundaries of the box plot are existing knowledge, which are the positions of the first quartile and the third quartile of the box plot.

[0041] Get the abnormality level of each monitoring data: ; In the formula, Represents the abnormality degree of the bth monitoring data; represents the distance between the vibration data of the bth monitoring data and the nearest boundary of the box plot to which it belongs; norm() represents the normalization function; The distance between the temperature data representing the bth monitoring data and the nearest boundary of the box plot to which it belongs; represents the anomaly weight of the temperature data; Represents the abnormal weight of vibration data; in the embodiment of the present invention, preset , In other embodiments, the implementer may preset as well as value; It should be noted that during the operation of the spinning machine, the friction between mechanical parts is an important factor in generating heat and vibration. When the friction between parts increases, on the one hand, the vibration of the parts will be intensified, and on the other hand, it will cause the temperature to rise. Therefore, if a defect occurs in a part and causes abnormal vibration data, the temperature data will also be abnormal. Therefore, when obtaining the abnormality degree of each monitoring data, more attention should be paid to the abnormality of the vibration data.

[0042] S3: Obtain the fluctuation of the neighborhood data of each monitoring data, correct the abnormality of each monitoring data, and obtain the corrected abnormality of each monitoring data; obtain abnormal data according to the corrected abnormality of each monitoring data, and judge the fault of the spinning machine.

[0043] Step S3 includes steps S30 and S31, which are specifically as follows: S30: Obtain fluctuations of neighborhood data of each monitoring data, correct the abnormality of each monitoring data, and obtain the corrected abnormality of each monitoring data.

[0044] It should be noted that due to the reasons of the vibration sensor and temperature sensor themselves, as well as the influence of the vibration of other machines in the spinning workshop, the collected monitoring data contains a certain amount of noise, and the noise data and the abnormal data are similar in the box plot, which will affect the identification of abnormal data, resulting in inaccurate abnormality of each monitoring data obtained. Therefore, it is necessary to adjust the abnormality of each monitoring data. It is known that noise is random and isolated, so the fluctuation of the neighborhood data of the noise data is small; however, the abnormal data generated by the intermittent abnormality of the spinning machine will last for a period of time and the volatility of the surrounding data is large. Therefore, the fluctuation of the neighborhood data of each monitoring data is obtained to correct the abnormality of each monitoring data.

[0045] In the embodiment of the present invention, the number of neighborhood data U is preset, and U data before and after each monitoring data are obtained as the neighborhood data of each monitoring data. The fluctuation of the neighborhood data of each monitoring data is obtained according to the calculation formula of the fluctuation of the first iteration data; Get the corrected abnormality level of each monitoring data: ; In the formula, Represents the corrected abnormality degree of the bth monitoring data; Represents the abnormality degree of the bth monitoring data; Represents the fluctuation of the neighborhood data of the b-th monitoring data; norm() represents the normalization function; if the fluctuation of the neighborhood data of the b-th monitoring data is large, it means that the b-th monitoring data is more likely to be abnormal data rather than noise data, so the degree of correction of the b-th monitoring data is greater.

[0046] S31: Acquire abnormal data according to the corrected abnormal degree of each monitoring data, and judge the fault of the spinning machine.

[0047] In an embodiment of the present invention, a corrected abnormality degree threshold T2 is preset. If the corrected abnormality degree of any monitoring data is greater than the corrected abnormality degree threshold T2, the monitoring data is taken as abnormal data, and all abnormal data are obtained. At this time, the system issues an early warning, and the staff conducts corresponding inspections on the spinning machine, so as to find out the causes of these intermittent faults in time and prevent further damage to the spinning machine.

[0048] An embodiment of the present invention further discloses a spinning machine fault monitoring 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, a spinning machine fault monitoring method according to the present invention is implemented.

[0049] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0050] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high bandwidth memory, a hybrid storage cube, etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of the device or accessible or connectable to the device.

[0051] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

[0052] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A spinning machine fault monitoring method, characterized in that: Includes steps: Collect monitoring data; sequentially obtain the data of the first N sampling moments in the monitoring data sequence as the original data, and obtain the first final expansion necessity of the original data ; Represents the fluctuation of the first iteration data; represents the first expansion necessity of the original data; norm() represents the normalization function; based on the final expansion necessity, the original data is iteratively expanded to obtain the first data segment; The data at N sampling moments after the first data segment are iteratively expanded as new original data to obtain the second data segment, and so on, until there is no data in the monitoring data sequence, and then the process stops, thus obtaining multiple data segments; Constructing a box plot for each dimension of the monitoring data in each data segment to obtain multiple box plots; Get the abnormality level of each monitoring data , Represents the abnormality degree of the bth monitoring data; as well as The distance between the vibration data and the temperature data of the b-th monitoring data and the nearest boundary of the box plot to which they belong; represents the anomaly weight of the temperature data; represents the abnormal weight of vibration data; Obtaining a corrected abnormality degree of each monitoring data, wherein the corrected abnormality degree is proportional to the fluctuation of the neighboring data of the monitoring data; According to the degree of correction abnormality of each monitoring data, the fault of the spinning machine is judged.

2. A spinning machine fault monitoring method according to claim 1, characterized in that: The necessity of obtaining the first expansion of the original data includes: ； In the formula, Represents the necessity of the first expansion of the original data; Represents the distance between the centroid of the original data and the centroid of the first updated data; The value of the i-th dimension data representing the centroid of the original data; Represents the value of the i-th dimension data of the centroid of the first updated data; Represents the number of dimensions of the data; || represents the absolute value symbol; The sum of the distances from the centroid of the convex hull region representing the original data to each vertex of the convex hull region; Represents the sum of the distances from the centroid of the convex hull area of ​​the first updated data to each vertex of its convex hull area; Represents the mean of the first updated data; Represents the median of the first updated data; exp() represents an exponential function with a natural constant as the base.

3. A spinning machine fault monitoring method according to claim 2, characterized in that: The obtaining of the centroid of the original data and the centroid of the first updated data includes: The number of sampling moments N is preset, and a two-dimensional rectangular coordinate system is constructed with amplitude as the horizontal coordinate and temperature as the vertical coordinate. The original data is mapped into the two-dimensional rectangular coordinate system and a convex hull detection is performed on it to obtain the convex hull area of ​​the original data, and the centroid of the convex hull area of ​​the original data is obtained and recorded as the centroid of the original data; M data after the Nth sampling moment in the monitoring data sequence are recorded as the first iteration data, which are added to the original data to obtain the first update data, and the first update data is mapped into the two-dimensional rectangular coordinate system, and a convex hull detection is performed on the first update data in the two-dimensional rectangular coordinate system to obtain the convex hull area of ​​the first update data, and the centroid of the convex hull area of ​​the first update data is recorded as the centroid of the first update data in the two-dimensional rectangular coordinate system.

4. A spinning machine fault monitoring method according to claim 1, characterized in that: The acquisition of the fluctuation of the first iteration data includes: Use the least square method to fit a curve to the i-th dimension data of the first iteration data, and obtain the fitting value of each i-th dimension data of the first iteration data; ； In the formula, Represents the fluctuation of the first iteration data; Represents the number of dimensions of the data; Represents the sum of the fitted values ​​of all i-th dimension data of the first iteration data; Represents the sum of the actual values ​​of all i-th dimension data of the first iteration data; Represents the standard deviation of all i-th dimension data of the first iteration data; norm() represents the normalization function.

5. A spinning machine fault monitoring method according to claim 1, characterized in that: The iterative expansion of the original data based on the final expansion necessity to obtain the first data segment includes: A threshold value of expansion necessity is preset. If the first final expansion necessity of the original data is greater than the expansion necessity threshold, the first iteration data is added to the original data, and M data after the first iteration data are used as the second iteration data. The second iteration data is added to the original data to obtain the second updated data. According to the second updated data and the original data, the second final expansion necessity of the original data is obtained to expand the original data. The process stops when the mth final expansion necessity of the original data is less than the expansion necessity threshold, and the final expanded original data is used as the first data segment.

6. A spinning machine fault monitoring method according to claim 1, characterized in that: The obtaining of the corrected abnormality degree of each monitoring data includes: Preset the number of neighborhood data U, obtain U data before each monitoring data and U data after each monitoring data as the neighborhood data of each monitoring data, and obtain the fluctuation of the neighborhood data of each monitoring data according to the calculation formula of the fluctuation of the first iteration data; ； In the formula, Represents the corrected abnormality degree of the bth monitoring data; Represents the abnormality degree of the bth monitoring data; Represents the fluctuation of the neighborhood data of the bth monitoring data; norm() represents the normalization function.

7. A spinning machine fault monitoring method according to claim 1, characterized in that: The method of judging the fault of the spinning machine according to the degree of abnormal correction of each monitoring data comprises: A correction abnormality threshold T2 is preset. If the correction abnormality of any monitoring data is greater than the correction abnormality threshold T2, the monitoring data will be regarded as abnormal data. At this time, the system will issue an early warning and the staff will inspect the spinning machine.

8. A spinning machine fault monitoring system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a spinning machine fault monitoring method according to any one of claims 1-7 is implemented.

Images