A method and system for processing abnormal data of a production line based on digital twin

Abstract

The present invention relates to the field of intelligent management technology, and discloses a method and system for processing abnormal data of a production line based on digital twins. The method includes: collecting digital twin model data of the production line to obtain the number of devices on the production line and determining the data analysis period; collecting the operation data of each device within the data analysis period, arranging them in chronological order as a data matrix, and determining whether there is an abnormality in the device; when it is determined that there is an abnormality in the device, extracting the abnormal characteristics, comparing them with the historical operation data, and determining the fault type of the device; determining the device risk value according to the fault type and the abnormal characteristics, and sorting the devices with abnormalities; correspondingly traversing the historical operation data of each abnormal device, obtaining the historical influence value, and determining whether to adjust the sorting to obtain the final sorting of abnormal devices. This application improves the accuracy of abnormal detection of the production line and reduces the maintenance cost of the production line.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management. Specifically, it relates to a method and system for processing abnormal data of a production line based on digital twin. Background Art

[0002] In modern manufacturing, the complexity and automation level of production lines are increasing day by day, and the operating status of equipment directly affects production efficiency and product quality. To ensure the efficient and stable operation of production lines, real-time monitoring and abnormal detection of equipment status become particularly important. However, due to the variety of production line equipment and complex operating environments, traditional detection methods based on fixed time intervals are difficult to adapt to changing working conditions and equipment status.

[0003] Traditional equipment abnormal detection methods usually rely on fixed detection cycles or simple threshold judgments. When facing complex working conditions and multi-source data, false alarms or missed alarms are likely to occur. Currently, most methods lack flexibility in setting the data analysis cycle and do not fully consider the impact of the number of equipment, resulting in too long or too short analysis cycles, affecting the timeliness and accuracy of abnormal detection. In addition, after equipment abnormal detection, if there are multiple abnormalities at the same time, it is impossible to sort and give early warnings according to the severity of the abnormalities, which easily causes delays in maintaining serious areas, thus increasing maintenance costs.

[0004] Therefore, it is necessary to design a method and system for processing abnormal data of a production line based on digital twin to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a method and system for processing abnormal data of a production line based on digital twin, aiming to solve the problems of poor flexibility in current production line detection, low detection accuracy, and inability to perform intelligent sorting and early warning when there are multiple abnormalities.

[0006] On the one hand, the present invention proposes a method for processing abnormal data of a production line based on digital twin, including:

[0007] Collecting digital twin model data of the production line to obtain the number of equipment on the production line, and determining the data analysis cycle according to the number of equipment;

[0008] Collecting the operation data of each equipment within the data analysis cycle, arranging the operation data in chronological order as a data matrix, and analyzing the data matrix to determine whether the equipment has an abnormality; when there is at least one abnormal data point in the data matrix, it is determined that the equipment has an abnormality;

[0009] Extract abnormal features when it is determined that the device has an abnormality. The abnormal features include abnormal data and the type of abnormal data. Compare the abnormal features with the historical operation data to determine the fault type of the device.

[0010] Determine the device risk value according to the fault type and abnormal features, and sort the devices with abnormalities according to the device risk value.

[0011] Correspondingly traverse the historical operation data of each abnormal device to obtain the historical impact value, and judge whether to adjust the sorting according to the historical impact value to obtain the final sorting of abnormal devices.

[0012] Further, when determining the data analysis period according to the number of devices, the data analysis period is calculated by the following formula:

[0013]

[0014] where T represents the data analysis period, N represents the number of devices, M represents the number of data acquisition channels, σi represents the data volatility of the i-th channel, f i represents the sampling frequency of the i-th channel, α represents the influence coefficient of data transmission rate on processing capacity, R represents the data transmission rate, and p represents the processor processing capacity.

[0015] Further, when arranging the operation data in chronological order as a data matrix and analyzing the data matrix to determine whether the device has an abnormality, it includes:

[0016] Determine the sampling frequency within the data analysis period, divide the operation data into several data groups according to the sampling frequency, and arrange the several data groups in sampling time as a 2 data matrix.

[0017] Randomly select a point in the data matrix as the point to be judged.

[0018] Taking the point to be judged as the center and L as the diameter to determine the data judgment range.

[0019] Calculate the average data value of the data judgment range according to the values of each data group within the data judgment range.

[0020] When the difference between the data value at the point to be judged and the average data value of the data judgment range is greater than the data threshold, determine that the point to be judged is an abnormal data point.

[0021] When there is at least one abnormal data point in the data matrix, determine that the device has an abnormality.

[0022] Further, the average data value of the data judgment range is calculated by the following formula:

[0023]

[0024] Among them, Q avg represents the average data value of the data judgment range, L represents the diameter of the data judgment range, (x0, y0) represents the coordinates of the point to be judged, and Q(x, y) represents the temperature value of the point (x, y) on the data matrix.

[0025] Furthermore, when comparing the abnormal feature with the historical operation data to determine the fault type of the device, it includes:

[0026] Calculating the similarity between the abnormal feature and the historical operation data, and taking the historical fault type corresponding to the highest similarity value as the fault type of the device; among them,

[0027] Calculating the similarity between the abnormal data in the abnormal feature and the abnormal data in the historical operation data:

[0028]

[0029] Among them, Stj represents the similarity of the abnormal data of the jth type, J represents the number of abnormal data types in the abnormal feature, V represents the abnormal data in the abnormal feature, Vi represents the abnormal data of the ith fault in the historical operation data, Vmax represents the maximum data value in the historical operation data, Vmin represents the minimum data value in the historical operation data, and n represents the number of historical faults;

[0030] Calculating the similarity between the abnormal data type in the abnormal feature and the abnormal data type in the historical operation data:

[0031]

[0032] Among them, Sy represents the similarity of the abnormal data type, n represents the number of historical faults, e represents the abnormal data type in the abnormal feature, ei represents the abnormal data type of the ith fault in the historical operation data, and E(e, ei) represents the type consistency function. When e = ei, E = 1;

[0033] Calculating the similarity between the abnormal feature and the historical operation data:

[0034] S = a * Stj + b * Sy;

[0035] Among them, S represents the similarity between the abnormal feature and the historical operation data, and a, b represent the weight coefficients, and a + b = 1.

[0036] Furthermore, when determining the device risk value according to the fault type and the abnormal feature, the device risk value is obtained by calculating the following formula:

[0037]

[0038] Among them, F represents the device risk value, J represents the number of abnormal data types in the abnormal features, Vj represents the abnormal data of the j-th type, μj represents the historical average value of the j-th type of data, σk represents the standard deviation of the j-th type of data, and r1, r2 represent adjustment coefficients.

[0039] Furthermore, when obtaining the historical influence value by traversing the historical operation data of each abnormal device, it includes:

[0040] Traverse the historical operation data of each abnormal device to obtain historical fault events of the same fault type as the device, collect the occurrence frequency and maintenance duration of the historical fault events, and calculate the historical influence value according to the occurrence frequency and maintenance duration. The historical influence value is obtained by the following formula:

[0041]

[0042] Among them, I represents the historical influence value, ρ represents the occurrence frequency, tb represents the maintenance duration of the b-th time, μt represents the historical average maintenance duration, σt represents the standard deviation of the historical maintenance duration, and B represents the total number of historical fault events.

[0043] Furthermore, when judging whether to adjust the sorting according to the historical influence value, it includes:

[0044] Compare the historical influence value with the influence value threshold, and judge whether to adjust the sorting according to the comparison result;

[0045] When the historical influence value is greater than the influence value threshold, it is determined to adjust the sorting;

[0046] When the historical influence value is less than or equal to the influence value threshold, it is determined not to adjust the sorting, and the current sorting is used as the final sorting of the abnormal devices.

[0047] Furthermore, when it is determined to adjust the sorting, it includes:

[0048] Obtain the influence value ratio according to the historical influence value and the influence value threshold. The influence value ratio is the quotient of the historical influence value and the influence value threshold;

[0049] When the influence value ratio is less than or equal to 1.2, move up the sorting of the corresponding abnormal device by one position;

[0050] When the influence value ratio is greater than 1.2 and less than or equal to 1.4, move up the sorting of the corresponding abnormal device by two positions;

[0051] When the influence value ratio is greater than 1.4, move up the sorting of the corresponding abnormal device by three positions.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows: By introducing digital twin model data and dynamically adjusting the data analysis period, the limitations of the traditional fixed-time interval detection method are overcome, ensuring the efficiency and flexibility of data analysis. By arranging the device operation data in chronological order as a data matrix and analyzing it, device anomalies can be detected promptly and accurately. After detecting an anomaly, by extracting the anomaly features and comparing them with the historical operation data, the fault type of the device can be accurately determined. Calculate the device risk value based on the fault type and anomaly features, and sort the devices with anomalies to ensure that high-risk devices are processed first. By traversing the historical data, calculate the historical impact value and dynamically adjust the device sorting, effectively avoiding maintenance delays in serious areas. The anomaly detection accuracy of the production line is improved, and the maintenance cost of the production line is reduced.

[0053] On the other hand, the present application also provides a digital twin-based production line anomaly data processing system for applying the above-mentioned digital twin-based production line anomaly data processing method, including:

[0054] A collection unit, configured to collect digital twin model data of the production line to obtain the number of devices on the production line, and determine the data analysis period according to the number of devices;

[0055] A judgment unit, configured to collect the operation data of each device within the data analysis period, arrange the operation data in chronological order as a data matrix, and analyze the data matrix to determine whether the device has an anomaly; when there is at least one anomaly data point in the data matrix, it is determined that the device has an anomaly;

[0056] A processing unit, configured to extract anomaly features when it is determined that the device has an anomaly, where the anomaly features include anomaly data and anomaly data types, and compare the anomaly features with the historical operation data to determine the fault type of the device;

[0057] An early warning unit, configured to determine the device risk value according to the fault type and anomaly features, and sort the devices with anomalies according to the device risk value; the early warning unit is also configured to traverse the historical operation data of each abnormal device correspondingly, obtain the historical impact value, and determine whether to adjust the sorting according to the historical impact value to obtain the final sorting of abnormal devices.

[0058] It can be understood that the above-mentioned digital twin-based production line anomaly data processing method and system have the same beneficial effects, which will not be elaborated here. Description of the Drawings

[0059] Various other advantages and benefits will become clear to those of ordinary skill in the art by reading the following detailed description of the preferred embodiments. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Also, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0060] Figure 1 It is a flowchart of the method for processing abnormal data of a production line based on digital twin provided by an embodiment of the present invention;

[0061] Figure 2 It is a functional block diagram of the system for processing abnormal data of a production line based on digital twin provided by an embodiment of the present invention. Detailed Embodiments

[0062] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. Hereinafter, the present invention will be described in detail with reference to the drawings and in conjunction with the embodiments.

[0063] In modern manufacturing, as the complexity and automation level of production lines continue to increase, production efficiency and product quality increasingly depend on the stable operation of equipment. Any abnormality of the equipment may lead to production line shutdown, product quality decline, and even safety hazards. Therefore, real-time monitoring of equipment status and rapid detection and handling of abnormalities have become the key to ensuring the stable and efficient operation of production lines. However, due to the diverse types of production line equipment and complex operating environments, traditional equipment abnormality detection methods face many challenges.

[0064] Traditional equipment abnormality detection usually relies on monitoring at fixed time intervals and simple threshold judgments. This method is inadequate in the face of a dynamically changing production environment. On the one hand, the fixed detection period cannot be flexibly adjusted according to the actual situation, resulting in a long detection period delaying the discovery of abnormalities, or a short period causing frequent false alarms and missed detections. On the other hand, simple threshold judgments cannot comprehensively consider the complexity of equipment operation, resulting in a significant reduction in the accuracy of abnormality detection when dealing with various types of equipment and complex working conditions.

[0065] In addition, the existing technology lacks effective anomaly handling and prioritization mechanisms after detecting equipment anomalies. When multiple anomalies occur simultaneously in a production line, it is usually impossible to accurately distinguish the severity and urgency of the anomalies, resulting in delays in the maintenance of important equipment, which in turn affects the normal operation of the entire production line, increasing maintenance costs and downtime. Therefore, it is necessary to design a production line anomaly data processing method and system based on digital twins to solve the problems existing in the current technology.

[0066] In some embodiments of the present application, referring to Figure 1 as shown, a production line anomaly data processing method based on digital twins includes:

[0067] S100: Collect digital twin model data of the production line to obtain the number of equipment in the production line, and determine the data analysis period according to the number of equipment.

[0068] S200: Collect the operation data of each equipment within the data analysis period, arrange the operation data in chronological order as a data matrix, and analyze the data matrix to determine whether the equipment has anomalies. When there is at least one anomaly data point in the data matrix, it is determined that the equipment has an anomaly.

[0069] S300: When it is determined that the equipment has an anomaly, extract the anomaly features, where the anomaly features include the anomaly data and the anomaly data type, and compare the anomaly features with the historical operation data to determine the fault type of the equipment.

[0070] S400: Determine the equipment risk value according to the fault type and the anomaly features, and sort the equipment with anomalies according to the equipment risk value.

[0071] S500: Correspondingly traverse the historical operation data of each abnormal equipment to obtain the historical impact value, and determine whether to adjust the sorting according to the historical impact value to obtain the final sorting of the abnormal equipment.

[0072] Specifically, in S100, the number of devices in the production line is obtained by collecting data from the digital twin model of the production line. The digital twin model is a virtual and dynamically updated digital mapping of the production line, which can reflect the operating conditions of the actual production line in real time. Based on the number of devices, the data analysis cycle is dynamically adjusted to ensure that the analysis cycle is neither too long to delay anomaly detection nor too short to cause false alarms or system overload due to frequent detections. In S200, after determining the data analysis cycle, the operating data of each device is collected within this cycle. These operating data are arranged in chronological order to form a data matrix. Matrix processing can quickly screen the operating data and accurately determine whether there is an anomaly in the device. If one or more abnormal data points appear in the data matrix, it is determined that the device is abnormal. In S300, when a device anomaly is detected, the anomaly features are further extracted. The anomaly features include the specific values and types of the abnormal data (such as temperature anomaly, vibration anomaly, etc.). These anomaly features are compared with the historical operating data of the device. The historical operating data is the operating data when an anomaly occurred in the past, including all the fault types that have occurred in the past and the operating data and abnormal data types corresponding to each fault type. By matching the anomaly patterns, the fault type of the device is determined, improving the accuracy of fault determination. In S400, after determining the fault type, the risk value of the device is calculated based on the fault type and its corresponding anomaly features. The risk value reflects the potential impact degree of the device fault on the production line. Then, the devices with anomalies are sorted according to the calculated risk value, and the high-risk devices are given priority for processing to ensure that the core devices of the production line can be maintained in a timely manner and major shutdown accidents are avoided. In S500, after initially completing the device risk ranking, the historical operating data of each abnormal device is traversed, and the historical impact value is calculated. The historical impact value is based on the frequency of device failures and the time required for maintenance, reflecting the long-term impact of device failures on the production line. According to the historical impact value, it is judged whether it is necessary to adjust the device risk ranking to ensure that the final ranking of abnormal devices is more reasonable and further optimize the maintenance decision-making.

[0073] It can be understood that by using digital twin technology and dynamically adjusting the data analysis cycle, the timeliness and accuracy of anomaly detection are ensured. The matrix-based data analysis method can quickly capture the operating state of the device, thereby improving the speed of anomaly detection. After detecting an anomaly, by extracting and comparing the anomaly features, the fault type is accurately determined, and the devices are sorted according to the risk value, and the high-risk devices are given priority for processing. Through the calculation of the historical impact value and the adjustment of the ranking, the reasonable allocation of maintenance resources is ensured, the downtime and maintenance costs of the production line are reduced, and the monitoring efficiency of the production line is improved.

[0074] In some embodiments of the present application, when determining the data analysis cycle according to the number of devices, the data analysis cycle is obtained by calculating the following formula:

[0075]

[0076] Where T represents the data analysis cycle, N represents the number of devices, M represents the number of data acquisition channels, σi represents the data volatility of the i-th channel, fi represents the sampling frequency of the i-th channel, α represents the influence coefficient of data transmission rate on processing power, R represents the data transmission rate, and p represents the processor processing power.

[0077] Specifically, data volatility σi represents the amplitude or instability of the data collected by a data channel or sensor over a period of time. It can be obtained by calculating the standard deviation and variance of the channel data. For example, a temperature sensor may have a small volatility in a stable operating environment, but the volatility increases in drastic environmental changes. The processing power of the processor reflects the amount of data that the central processor can process or the number of instructions executed per unit time. The influence coefficient α is used to quantify the impact of the data transmission rate on the system processing power. In an environment with a stable network and sufficient bandwidth, α is preferably 0.1-0.3, and in an environment with an unstable network and more interference, α is preferably 0.5-0.7.

[0078] It is understandable that by dynamically adjusting the data analysis cycle, the flexibility and adaptability of the system when facing different numbers of devices and different types of data are ensured. Compared with the traditional fixed analysis cycle method, it can more accurately reflect the actual operation status of the production line, reduce the occurrence of false alarms and missed alarms, and improve the accuracy of anomaly detection. In addition, the calculation of the data analysis cycle comprehensively considers factors such as data volatility, sampling frequency, data transmission rate and processing capacity, realizes efficient data processing and resource allocation, and effectively improves the operation efficiency of the production line.

[0079] In some embodiments of the present application, the operation data is arranged in chronological order into a data matrix, and the data matrix is analyzed to determine whether the device is abnormal, including:

[0080] Determine the sampling frequency within the data analysis cycle, divide the operating data into several data groups according to the sampling frequency, and arrange the several data groups into a 2 The data matrix.

[0081] A point is randomly selected in the data matrix as the point to be judged.

[0082] The data judgment range is determined with the point to be judged as the center and L as the diameter.

[0083] The average data value of the data judgment range is calculated based on the values of each data group within the data judgment range.

[0084] When the difference between the data value at the point to be determined and the average data value of the data determination range is greater than the data threshold, the point to be determined is determined to be an abnormal data point.

[0085] When there is at least one abnormal data point in the data matrix, it is determined that the device is abnormal.

[0086] In some embodiments of the present application, the average data value of the data judgment range is obtained by the following formula:

[0087]

[0088] where Q avg represents the average data value of the data judgment range, L represents the diameter of the data judgment range, (x0, y0) represents the coordinates of the point to be judged, and Q(x, y) represents the temperature value at the point (x, y) on the data matrix.

[0089] Specifically, according to the sampling frequency within the data analysis period, the operation data of the device is divided into several data groups. The sampling frequency determines the number of data groups. The sampling frequency can be determined according to the lowest sampling frequency of the sensors in the data group. These data groups are arranged in a two-dimensional data matrix a×a in the order of sampling time from left to right and from top to bottom. When the number in the data group is insufficient to form a square matrix, the cyclic supplementation method is used for data supplementation.

[0090] It can be understood that by constructing a data matrix and combining the method of judging the neighborhood average value. Compared with the traditional single-point threshold judgment method, it can more effectively filter out random noise, improve the accuracy of anomaly detection, and reduce the false alarm rate. In addition, by setting the data judgment range, the sensitivity of the algorithm can be flexibly adjusted to adapt to different types of devices and working conditions.

[0091] In some embodiments of the present application, when comparing the abnormal features with the historical operation data to determine the fault type of the device, it includes:

[0092] Calculate the similarity between the abnormal features and the historical operation data, and take the historical fault type corresponding to the highest similarity value as the fault type of the device. Among them,

[0093] Calculate the similarity between the abnormal data in the abnormal features and the abnormal data in the historical operation data:

[0094]

[0095] where Stj represents the similarity of the abnormal data of the jth type, J represents the number of abnormal data types in the abnormal features, V represents the abnormal data in the abnormal features, Vi represents the abnormal data of the ith fault in the historical operation data, Vmax represents the maximum data value in the historical operation data, Vmin represents the minimum data value in the historical operation data, and n represents the number of historical faults.

[0096] Calculate the similarity between the abnormal data type in the abnormal feature and the abnormal data type in the historical operation data:

[0097]

[0098] Among them, Sy represents the similarity of the abnormal data type, n represents the number of historical faults, e represents the abnormal data type in the abnormal feature, ei represents the abnormal data type of the i-th fault in the historical operation data, and E(e, ei) represents the type consistency function. When e = ei, E = 1.

[0099] Calculate the similarity between the abnormal feature and the historical operation data:

[0100] S = a * Stj + b * Sy;

[0101] Among them, S represents the similarity between the abnormal feature and the historical operation data, a and b represent the weight coefficients, and a + b = 1.

[0102] It can be understood that the abnormal data type includes temperature abnormality, pressure abnormality, rotational speed abnormality, etc. According to the type of detection sensor, the abnormal data type can include multiple types. Suppose a device has an abnormality during current operation, the abnormal data is the temperature value V = 80°C, and the abnormal data type is temperature abnormality. Then the number of abnormal data types in the above formula is 1. Vmax represents the highest temperature reached by the device in the historical operation data, Vmin represents the lowest temperature reached by the device in the historical operation data. Vi represents the temperature data of the device during each fault, and n represents the total number of faults of the device in history.

[0103] It can be understood that by introducing the calculation of the similarity between the abnormal feature and the historical data, the determination of the fault type is made more accurate. Compared with the traditional simple matching method, by weighing the importance of the abnormal data value and the data type, the accuracy and robustness of the abnormal determination are improved. Especially in the scenario where the device fault types are diverse and the historical data is rich, it can more comprehensively and accurately identify the fault types of the device, thereby improving the stability of the production line and the efficiency of fault handling.

[0104] In some embodiments of the present application, when determining the device risk value according to the fault type and the abnormal feature, the device risk value is obtained by the following formula:

[0105]

[0106] Among them, F represents the device risk value, J represents the number of abnormal data types in the abnormal feature, Vj represents the abnormal data of the j-th type, μj represents the historical average value of the j-th type of data, σk represents the standard deviation of the j-th type of data, and r1, r2 represent the adjustment coefficients.

[0107] It is understandable that by comprehensively considering the standard deviation of abnormal data and the importance of different types of abnormalities, the risk value of the device can be accurately quantified, effectively improving the rationality of the sorting of abnormal devices. Especially in a production line with a large variety of devices and a complex operating environment, it can reduce the probability of false alarms and missed reports, ensuring the stable operation of the production line and the reasonable allocation of maintenance resources.

[0108] In some embodiments of the present application, when traversing the historical operation data of each abnormal device to obtain the historical impact value, it includes:

[0109] Traverse the historical operation data of each abnormal device to obtain historical fault events of the same fault type as the device, collect the occurrence frequency and maintenance duration of the historical fault events, and calculate the historical impact value based on the occurrence frequency and maintenance duration. The historical impact value is obtained by the following formula:

[0110]

[0111] Where, I represents the historical impact value, ρ represents the occurrence frequency, tb represents the maintenance duration of the bth time, μt represents the historical average maintenance duration, σt represents the standard deviation of the historical maintenance duration, and B represents the total number of historical fault events.

[0112] It is understandable that by integrating the frequency and maintenance duration of fault events, the influence of historical faults can be comprehensively evaluated. Compared with simple fault count statistics, it is beneficial to accurately reflect the actual impact of each fault on the device operation. By standardizing the maintenance duration and considering the relative differences between different events, the historical impact value becomes more representative.

[0113] In some embodiments of the present application, when judging whether to adjust the sorting according to the historical impact value, it includes: comparing the historical impact value with the impact value threshold, and judging whether to adjust the sorting according to the comparison result.

[0114] Specifically, when the historical impact value is greater than the impact value threshold, it is determined to adjust the sorting. When the historical impact value is less than or equal to the impact value threshold, it is determined not to adjust the sorting, and the current sorting is used as the final sorting of abnormal devices.

[0115] In some embodiments of the present application, when it is determined to adjust the sorting, it includes: obtaining an impact value ratio based on the historical impact value and the impact value threshold, and the impact value ratio is the quotient of the historical impact value and the impact value threshold.

[0116] Specifically, when the impact value ratio is less than or equal to 1.2, the sorting of the corresponding abnormal device is raised by one position. When the impact value ratio is greater than 1.2 and less than or equal to 1.4, the sorting of the corresponding abnormal device is raised by two positions. When the impact value ratio is greater than 1.4, the sorting of the corresponding abnormal device is raised by three positions.

[0117] Specifically, assume that the historical impact value I of a certain device is 15, and the set impact value threshold Imax is 12. According to the formula, the impact value ratio r = 15 / 12 = 1.25 is calculated. Since 1.2 < r = 1.25 ≤ 1.4, according to the adjustment strategy, the device is moved up two positions in the ranking. Assume that the device was originally ranked 5th. After adjustment, the priority of the device will rise to 3rd. If the device was originally ranked 1st, it remains unchanged. If the device was originally ranked 2nd, it is promoted to 1st.

[0118] It can be understood that by comparing the historical impact value with the set threshold, the ranking of the device is dynamically adjusted. Considering the severity of the device's historical faults, and by setting a multi-level adjustment mechanism, the device priority is flexibly adjusted according to the impact value ratio, so that devices with high risks can be processed in a timely manner. It improves the rationality of the abnormal device ranking and the efficiency of maintenance work, helps to optimize resource scheduling, and reduces downtime.

[0119] In the above embodiment, by introducing digital twin model data, the data analysis period is dynamically adjusted, overcoming the limitations of the traditional fixed-time interval detection method, and ensuring the efficiency and flexibility of data analysis. By arranging the device operation data in chronological order as a data matrix and analyzing it, device abnormalities can be detected in a timely and accurate manner. After detecting an abnormality, by extracting the abnormal features and comparing them with the historical operation data, the fault type of the device is accurately determined. According to the fault type and abnormal features, the device risk value is calculated, and the devices with abnormalities are ranked to ensure that high-risk devices are given priority for processing. By traversing the historical data, the historical impact value is calculated and the device ranking is dynamically adjusted, effectively avoiding the maintenance delay in serious areas. It improves the accuracy of abnormal detection on the production line and reduces the maintenance cost of the production line.

[0120] In another preferred manner based on the above embodiment, refer to Figure 2 As shown, this embodiment provides a digital twin-based production line abnormal data processing system for applying the above digital twin-based production line abnormal data processing method, including:

[0121] An acquisition unit, configured to acquire the digital twin model data of the production line to obtain the number of devices on the production line, and determine the data analysis period according to the number of devices;

[0122] A judgment unit, configured to acquire the operation data of each device within the data analysis period, arrange the operation data in chronological order as a data matrix, and analyze the data matrix to determine whether the device has an abnormality; when there is at least one abnormal data point in the data matrix, it is determined that the device has an abnormality;

[0123] A processing unit, configured to extract abnormal features when it is determined that the device has an abnormality. The abnormal features include abnormal data and the type of abnormal data, compare the abnormal features with historical operation data, and determine the fault type of the device;

[0124] An early warning unit, configured to determine the device risk value according to the fault type and abnormal features, and sort the devices with abnormalities according to the device risk value; the early warning unit is also configured to traverse the historical operation data of each abnormal device correspondingly, obtain the historical influence value, and judge whether to adjust the sorting according to the historical influence value to obtain the final sorting of abnormal devices.

[0125] It can be understood that by introducing digital twin model data and dynamically adjusting the data analysis period, the limitations of the traditional fixed-time-interval detection method are overcome, ensuring the efficiency and flexibility of data analysis. By arranging the device operation data in chronological order into a data matrix and analyzing it, device abnormalities can be detected in a timely and accurate manner. After detecting an abnormality, by extracting the abnormal features and comparing them with the historical operation data, the fault type of the device can be accurately determined. Calculate the device risk value according to the fault type and abnormal features, and sort the devices with abnormalities to ensure that high-risk devices are processed first. By traversing the historical data, calculating the historical influence value and dynamically adjusting the device sorting, the maintenance delay in serious areas is effectively avoided. The accuracy of abnormality detection on the production line is improved, and the maintenance cost of the production line is reduced.

[0126] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0127] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0128] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the function specified in one process Figure 1 one process or a plurality of processes and / or blocks Figure 1 specified in one block or a plurality of blocks.

[0129] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the function specified in one process Figure 1 one process or a plurality of processes and / or blocks Figure 1 specified in one block or a plurality of blocks.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention. Any modification or equivalent replacement without departing from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for processing abnormal data of a production line based on digital twin, characterized in that, Including: Collect the digital twin model data of the production line to obtain the number of devices on the production line, and determine the data analysis period according to the number of devices; Collect the operation data of each device within the data analysis period, arrange the operation data in chronological order as a data matrix, and analyze the data matrix to determine whether the device has an abnormality; When there is at least one abnormal data point in the data matrix, it is determined that the device has an abnormality; When it is determined that the device has an abnormality, extract the abnormal features, where the abnormal features include abnormal data and the type of abnormal data, and compare the abnormal features with the historical operation data to determine the fault type of the device; Determine the device risk value according to the fault type and the abnormal features, and sort the devices with abnormalities according to the device risk value; Correspondingly traverse the historical operation data of each abnormal device to obtain the historical influence value, and determine whether to adjust the sorting according to the historical influence value to obtain the final sorting of the abnormal devices; When determining the device risk value according to the fault type and the abnormal features, the device risk value is obtained by calculation using the following formula: ； Wherein, F represents the device risk value, J represents the number of types of abnormal data in the abnormal features, Vj represents the abnormal data of the jth type, μj represents the historical average value of the jth type of data, σj represents the standard deviation of the jth type of data, and r1, r2 represent adjustment coefficients.

2. The method for processing abnormal data of a production line based on digital twin according to claim 1, wherein, When arranging the operation data in chronological order as a data matrix and analyzing the data matrix to determine whether the device has an abnormality, it includes: Determine the sampling frequency within the data analysis period, divide the operation data into several data groups according to the sampling frequency, and arrange the several data groups in the order of sampling time as a 2 data matrix; Randomly select a point in the data matrix as the point to be judged; Taking the point to be judged as the center and L as the diameter to determine the data judgment range; Calculate the average data value of the data judgment range according to the values of each data group within the data judgment range; When the difference between the data value at the point to be judged and the average data value of the data judgment range is greater than the data threshold, it is determined that the point to be judged is an abnormal data point; When there is at least one abnormal data point in the data matrix, it is determined that the device has an abnormality.

3. The method for processing abnormal data of a production line based on digital twin according to claim 2, wherein The average data value of the data judgment range is obtained by calculation using the following formula: ； Among them, Q avg represents the average data value of the data judgment range, (x0, y0) represents the coordinates of the point to be judged, and Q(x, y) represents the temperature value of the point (x, y) on the data matrix.

4. The method for processing abnormal data of a production line based on digital twin according to claim 1, wherein When correspondingly traversing the historical operation data of each abnormal device to obtain the historical influence value, it includes: Traverse the historical operation data of each abnormal device to obtain historical fault events with the same fault type as the device, collect the occurrence frequency and maintenance duration of the historical fault events, and calculate the historical influence value according to the occurrence frequency and maintenance duration. The historical influence value is obtained by calculation using the following formula: ； Wherein, I represents the historical influence value, ρ represents the occurrence frequency, tb represents the maintenance duration of the bth time, μt represents the historical average maintenance duration, σt represents the standard deviation of the historical maintenance duration, and B represents the total number of historical fault events.

5. The method for processing abnormal data of a production line based on digital twin according to claim 4, characterized in that When determining whether to adjust the sorting according to the historical influence value, it includes: Compare the historical influence value with the influence value threshold, and determine whether to adjust the sorting according to the comparison result; When the historical influence value is greater than the influence value threshold, it is determined to adjust the sorting; When the historical impact value is less than or equal to the impact value threshold, it is determined that the sorting is not adjusted, and the current sorting is used as the final sorting of the abnormal devices.

6. The method for processing abnormal data of a production line based on digital twin according to claim 5, wherein, When it is determined to adjust the sorting, it includes: Obtaining an impact value ratio based on the historical impact value and the impact value threshold, where the impact value ratio is the quotient of the historical impact value and the impact value threshold; When the impact value ratio is less than or equal to 1.2, the sorting of the corresponding abnormal device is raised by one position; When the impact value ratio is greater than 1.2 and less than or equal to 1.4, the sorting of the corresponding abnormal device is raised by two positions; When the impact value ratio is greater than 1.4, the sorting of the corresponding abnormal device is raised by three positions.

7. A production line abnormal data processing system based on digital twin, which is used to apply the production line abnormal data processing method based on digital twin according to any one of claims 1-6, and is characterized in that It includes: A collection unit, configured to collect digital twin model data of a production line to obtain the number of devices on the production line, and determine a data analysis period according to the number of devices; A judgment unit, configured to collect the operation data of each device within the data analysis period, arrange the operation data in chronological order as a data matrix, and analyze the data matrix to determine whether the device is abnormal; When there is at least one abnormal data point in the data matrix, it is determined that the device is abnormal; A processing unit, configured to extract abnormal features when it is determined that the device is abnormal, where the abnormal features include abnormal data and the type of abnormal data, and compare the abnormal features with historical operation data to determine the fault type of the device; An early warning unit, configured to determine a device risk value according to the fault type and the abnormal features, and sort the devices with abnormalities according to the device risk value; the early warning unit is also configured to traverse the historical operation data of each abnormal device correspondingly, obtain a historical impact value, and determine whether to adjust the sorting according to the historical impact value to obtain the final sorting of the abnormal devices.

Images