Abnormal identification method and system for slaughtering line based on digital twin

Abstract

The present invention relates to the field of industrial equipment anomaly detection, and in particular to a slaughter line anomaly identification method and system based on digital twins. The method comprises: collecting operating data and health reference data of a splitting saw in a slaughter line through a data acquisition unit and a digital twin interface unit; obtaining a preliminary abnormality judgment result of the operating state of the splitting saw by performing digital twin reference correction and cumulative control chart anomaly detection processing on the instantaneous current data of the driving unit; obtaining an abnormality source tendency discriminant factor by fusing and evaluating the physical feature data and health reference data of the splitting saw; obtaining an abnormality identification result of the current processing cycle of the splitting saw by performing a comprehensive analysis of the preliminary abnormality judgment result and the abnormality source tendency discriminant factor, thereby improving the accuracy of abnormality identification of equipment in the slaughter line.

Description

Technical Field

[0001] The present invention relates to the field of industrial equipment anomaly detection, and in particular to a slaughter line anomaly recognition method and system based on digital twins. Background Art

[0002] Slaughtering and processing are crucial components of the meat supply chain, with production efficiency, product quality, and equipment reliability significantly impacting the entire supply chain. In modern slaughtering lines, the splitting process is a crucial step in precisely dividing the slaughtered animal along the spine. This process is typically performed by an automatic or semi-automatic splitting saw. As a core piece of equipment, the sharpness of the saw blade and the operating condition of the drive motor are directly related to splitting quality and production continuity. To ensure stable operation of the splitting saw and prevent production interruptions and product quality degradation caused by sudden failures, equipment condition monitoring and anomaly detection methods are commonly employed. Early methods primarily relied on setting fixed thresholds for single parameters such as motor current, voltage, or overall equipment vibration amplitude to generate alarms. These methods are insensitive to dynamic process changes and minor anomalies. To address this, some monitoring systems have incorporated the concepts of statistical process control, employing sequential analysis methods such as cumulative sum control charts and exponentially weighted moving average control charts. These methods accumulate small deviations from target values ​​of process parameters to more sensitively detect persistent small excursions from the process mean, thereby improving the early detection of anomalies. However, in actual slaughter operations, the objects being processed exhibit significant individual differences in size, breed, leanness, and bone density. This variability in the objects being processed causes significant natural fluctuations in key operating parameters such as the motor's load current and vibration response during normal operation of the splitting saw. For example, when splitting objects with large bones or tough meat, even if the equipment is in perfect condition, the motor current and vibration will transiently increase. Conversely, when splitting smaller objects or those with softer meat, these parameters will be relatively low.

[0003] Existing anomaly detection methods for butchering line split saws, particularly those employing statistical process control models such as cumulative sum charts, often rely on historical data or fixed parameter settings. These models' control benchmarks are often inadequate to accurately match the normal operating parameters of the split saw, which vary dynamically due to individual differences in the processing object. This mismatch between the control benchmark and actual operating conditions leads to unreliable anomaly detection results. When the processing object's characteristics cause the operating parameters to be higher than normal, the model can easily misinterpret this normal condition as an anomaly, generating a false alarm. Conversely, when the equipment exhibits true early degradation but the processing object's characteristics cause the operating parameters to be lower, the model fails to recognize this degradation, leading to missed alarms. Therefore, developing a mechanism that dynamically adapts the sensitive benchmark of the statistical process control model to changes in processing object characteristics, thereby significantly improving the accuracy and reliability of real-state anomaly detection in the split saw, is an urgent issue. Summary of the Invention

[0004] In view of this, an embodiment of the present invention provides a slaughtering line abnormality identification method and system based on digital twins to solve the problems of false alarms and missed alarms of splitting saws in the slaughtering line.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] In a first aspect, the present application provides a method for identifying abnormalities in a slaughtering line based on digital twins, the method comprising the following steps:

[0007] Step S1: collecting operating data and health reference data of the splitting saw in the slaughtering line through the data acquisition unit and the digital twin interface unit;

[0008] Step S2: Obtain a preliminary abnormality judgment result of the split saw operation state by performing digital twin reference correction and cumulative sum control chart abnormality detection processing on the instantaneous current data of the drive unit;

[0009] Step S3: Obtaining an abnormality source tendency discriminant factor by fusing and evaluating the physical feature data of the split saw with the health reference data;

[0010] Step S4: Obtain an abnormality recognition result of the current processing cycle of the split saw by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor;

[0011] Step S5: intelligent application of abnormal state of the splitting saw in the slaughter line based on the abnormality recognition result in the splitting saw processing cycle;

[0012] The data acquisition unit and the digital twin interface unit are used to collect operation data and health reference data of the splitting saw in the slaughtering line, including: in the data acquisition unit of the slaughtering line, various sensors installed at key positions of the splitting saw are used to periodically collect data, and the data collected during the data acquisition process include: processing cycle data of the drive unit, instantaneous current data of the drive unit, maximum temperature data of the shell of the drive unit at the end of each processing cycle, root mean square data of vibration energy in a frequency band corresponding to the fault characteristic frequency of the outer ring of the output end bearing of the drive unit, root mean square data of vibration energy in a high-frequency vibration characteristic frequency band of the drive unit, and a volatility index of the instantaneous current data of the drive unit;

[0013] In the digital twin interface unit of the slaughtering line, through the digital twin interface module deployed in the control system, the digital twin model matching the characteristics of the current processed object is called in each processing cycle to obtain reference data of the health state corresponding to the processing stage of the current processing cycle of the current processing object, specifically including the reference standard deviation of the instantaneous current data of the drive unit, the instantaneous current health data of the drive unit, the maximum shell temperature health data of the drive unit at the end of each processing cycle, the root mean square health data of the vibration energy in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit, the root mean square health data of the vibration energy in the high-frequency vibration characteristic frequency band of the drive unit, and the volatility health index of the instantaneous current data of the drive unit;

[0014] The method of obtaining an abnormal source tendency discrimination factor by fusing and evaluating the physical feature data of the split saw with the health reference data includes: obtaining a normalized deviation index reflecting the degree of abnormality of each physical dimension by normalizing the microscopic physical feature data and the health reference data; obtaining an abnormal contribution value of each feature dimension by applying a feature sensitivity adjustment function to the normalized deviation index; and obtaining an abnormal source tendency discrimination factor for fault source discrimination by performing a structured fusion evaluation on the abnormal contribution value.

[0015] Preferably, the method of obtaining a preliminary abnormality judgment result of the operation state of the split saw by performing digital twin reference correction and cumulative control diagram abnormality detection processing on the instantaneous current data of the drive unit includes:

[0016] By performing a deviation assessment on the instantaneous current data of the drive unit and the instantaneous current health data of the drive unit, the instantaneous current deviation data of the drive unit is obtained; a bilateral cumulative sum control chart is constructed by using the deviation sequence formed by the instantaneous current deviation data of the drive unit as input, and the upper cumulative sum and the lower cumulative sum are calculated respectively; by performing a decision limit assessment on the upper cumulative sum and the lower cumulative sum, a preliminary abnormality judgment result of the operating status of the split saw is obtained.

[0017] Preferably, the step of performing deviation evaluation on the instantaneous current data of the drive unit and the health data of the instantaneous current matching of the drive unit to obtain the instantaneous current deviation data of the drive unit includes:

[0018] The calculation result of subtracting the instantaneous current data of the drive unit from the instantaneous current health data of the drive unit is used as the instantaneous current first deviation of the drive unit, and the calculation result of dividing the instantaneous current first deviation of the drive unit by the reference standard deviation of the instantaneous current data of the drive unit is used as the instantaneous current deviation data of the drive unit.

[0019] Preferably, the normalization processing of the microscopic physical characteristic data and the health reference data to obtain a normalized deviation index reflecting the abnormality degree of each physical dimension includes:

[0020] The characteristic tolerance coefficient of the drive unit is set, including: the temperature characteristic tolerance coefficient of the drive unit, the root mean square tolerance coefficient of the vibration energy in the frequency band corresponding to the characteristic frequency of the fault of the outer ring of the output end bearing of the drive unit, the root mean square tolerance coefficient of the vibration energy in the high-frequency vibration characteristic frequency band of the drive unit, and the current fluctuation characteristic tolerance coefficient of the drive unit; the difference between the physical characteristics of the drive unit and the healthy reference data is used as the numerator, the calculation result of multiplying the healthy reference data of the drive unit by the corresponding characteristic tolerance coefficient of the drive unit is used as the denominator, and the formed fraction is used as the first characteristic abnormality degree of the drive unit; the larger value between the first characteristic abnormality degree of the drive unit and the constant 0 is used as the normalized deviation index reflecting the abnormality degree of each physical dimension.

[0021] Preferably, the abnormal contribution value of each feature dimension is obtained by applying a feature sensitivity adjustment function to the normalized deviation index, including:

[0022] A feature sensitivity adjustment function is set, and the normalized deviation index of the abnormality degree of each physical dimension is mapped through the feature sensitivity adjustment function to obtain the abnormal contribution value of each feature dimension.

[0023] Preferably, the abnormal source tendency discrimination factor for fault source discrimination is obtained by performing structured fusion evaluation on the abnormal contribution value, including:

[0024] Obtain the temperature anomaly contribution value of the drive unit in all feature dimensions, the vibration energy anomaly contribution value in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit, the vibration energy anomaly contribution value in the high-frequency vibration characteristic frequency band of the drive unit, and the current data fluctuation anomaly contribution value of the drive unit; add the temperature anomaly contribution value of the drive unit and the specific vibration characteristic anomaly contribution value of the drive unit as the numerator, add the high-frequency vibration anomaly contribution value of the drive unit and the current data fluctuation anomaly contribution value of the drive unit as the denominator, and use the formed fraction as an abnormal source tendency discrimination factor for fault source discrimination.

[0025] Preferably, the method of obtaining the abnormality identification result of the current processing cycle of the split saw by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor includes:

[0026] An upper discrimination threshold and a lower discrimination threshold are set for identifying abnormalities of the split saw; if the abnormal source tendency discrimination factor is greater than the upper discrimination threshold, it is determined that the split saw has a fault caused by the drive unit itself in the current processing cycle; if the abnormal source tendency discrimination factor is less than the lower discrimination threshold, it is determined that the split saw has a fault caused by tool performance degradation or interaction between the tool and the processing object in the current processing cycle; if the abnormal source tendency discrimination factor is between the upper discrimination threshold and the lower discrimination threshold, it is determined that the split saw has a mixed fault in the current processing cycle and manual inspection is required.

[0027] In the second aspect, the present application provides a slaughter line abnormality identification system based on digital twins, which includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the slaughter line abnormality identification method based on digital twins in the present application is implemented.

[0028] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0029] The present invention proposes a method for identifying abnormalities in the splitting saw of a slaughtering line based on digital twins. This method is aimed at practical problems in slaughtering production scenarios, such as the complex and changeable operating status of the splitting saw equipment, the large individual differences in the processing objects, and the difficulty in tracing the source of faults. By constructing an abnormality detection and discrimination mechanism that integrates collected data and digital twin reference data, it achieves high-precision identification of the operating status of key equipment and intelligent judgment of the source of abnormalities. Specifically, the present invention improves the accuracy and adaptability of abnormality detection, and achieves explainable discrimination and quantitative judgment of the source of faults. The present invention proposes an abnormality source tendency discrimination factor to further deepen the attribution discrimination of the cause of the abnormality. The abnormality source tendency discrimination factor fully considers the behavioral patterns of four types of key characteristic data, including: the maximum temperature data of the housing, the root mean square data of the vibration energy in the fault characteristic frequency band of the outer ring of the motor output bearing, the root mean square data of the vibration energy in the high-frequency vibration characteristic frequency band, and the volatility index of the instantaneous current data of the drive unit. The above-mentioned characteristic data are normalized and compared with their respective theoretical healthy reference data, and are uniformly mapped in combination with the characteristic sensitivity adjustment function, so that the quantitative expression of the anomaly has physical interpretability and discriminant value. It can effectively distinguish anomalies caused by drive unit failure from those caused by saw blade loss or workpiece interference, providing a strong basis for subsequent precise maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 This is a flow chart of the method and system for identifying abnormalities in a slaughtering line based on digital twins provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0032] The embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0033] It should be noted that the terms "first," "second," and the like in the specification of the present disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0034] In order to illustrate the technical solution of the present invention, specific embodiments are provided below.

[0035] See also Figure 1 , is a flowchart of a method and system for identifying abnormalities in a slaughtering line based on digital twins provided in the first embodiment of the present invention, such as Figure 1 As shown, the method may include:

[0036] Step S1: collect operating data and health reference data of the splitting saw in the slaughtering line through the data acquisition unit and the digital twin interface unit.

[0037] In this step, the data acquisition unit and digital twin interface unit installed in the splitting saw equipment and its associated system are used to collect the operating data of the splitting saw in the slaughter line. The data acquisition unit periodically collects data through various sensors installed in key parts of the splitting saw. The collected data includes:

[0038] 1. Processing cycle data of the drive unit;

[0039] 2. Instantaneous current data of the drive unit;

[0040] 3. The maximum shell temperature data of the drive unit in each processing cycle;

[0041] 4. The instantaneous vibration acceleration signal collected by the vibration sensor installed at the bearing position of the output end of the drive unit is processed by fast Fourier transform in each processing cycle to calculate the root mean square data of the vibration energy within the preset frequency band corresponding to the characteristic frequency of the fault of the outer race of the motor output end bearing. In this embodiment, the frequency band corresponding to the characteristic frequency of the fault of the outer race of the motor output end bearing is set to the range of plus or minus 5 Hz of the center frequency;

[0042] 5. The instantaneous vibration acceleration signal collected from a vibration sensor installed near the tool is processed by fast Fourier transform during each machining cycle to calculate the root mean square vibration energy data within a preset high-frequency vibration characteristic frequency band. In this embodiment of the present invention, the high-frequency vibration characteristic frequency band is set to range from 1 kHz to 5 kHz.

[0043] 6. Fluctuation index of the instantaneous current data of the drive unit. In this embodiment, the current data within each processing cycle is first centered, that is, the current mean value within the window is subtracted to remove the DC component, and then the standard deviation of the processed data is calculated. The standard deviation value is the fluctuation index of the instantaneous current data of the drive unit.

[0044] Through the digital twin interface module deployed in the control system, the present invention calls the digital twin model that matches the characteristics of the current processed object in each processing cycle to obtain reference data of the current processed object in the health state corresponding to the processing stage of the current processing cycle, specifically including:

[0045] 1. The instantaneous healthy current data corresponding to the actual observed current value of the drive unit;

[0046] 2. A reference standard deviation used for normalizing the correction error signal in subsequent steps. In this embodiment, the reference standard deviation is obtained based on a large amount of historical health data statistics. It represents a fixed value of the average fluctuation level of the residual data under various operating conditions and is set to 1.5% of the rated current.

[0047] 3. The healthy temperature reference data corresponding to the maximum shell temperature data of the drive unit in each processing cycle collected above;

[0048] 4. Healthy reference data corresponding to the root mean square vibration energy data within the frequency band corresponding to the fault characteristic frequency of the outer ring of the output end bearing of the drive unit collected above;

[0049] 5. Health reference data corresponding to the root mean square vibration energy data within the high-frequency vibration characteristic frequency band of the drive unit collected above;

[0050] 6. The health reference data corresponding to the volatility index of the instantaneous current data of the drive unit collected above.

[0051] Step S2: A preliminary abnormality judgment result of the split saw operation state is obtained by performing digital twin reference correction and cumulative control chart abnormality detection processing on the instantaneous current data of the drive unit.

[0052] After completing the operational data collection and corresponding health reference data acquisition of key components of the split saw equipment in step S1, the present invention further proceeds to the intelligent anomaly identification stage for the equipment's operational status. Given the significant individual variability of objects processed in slaughter lines (e.g., weight, shape, and type), as well as the significant dynamic operating conditions during processing, directly determining anomalies based on fixed thresholds or static reference baselines often fails to accurately identify the equipment's true operational status, potentially leading to false alarms or missed detections. The present invention first obtains instantaneous current deviation data for the drive unit by performing a deviation assessment between the drive unit's instantaneous current data and its instantaneous current health data. A bilateral cumulative sum control chart is then constructed using the deviation sequence formed by the drive unit's instantaneous current deviation data as input, and the upper and lower cumulative sums are calculated. Finally, a preliminary anomaly determination result for the split saw's operational status is obtained by performing a decision limit assessment on the upper and lower cumulative sums.

[0053] Specifically, first, for each sampling point in the instantaneous current data of the drive unit collected in step S1, the instantaneous current deviation data of the drive unit is evaluated in combination with the mean value of the instantaneous current theoretical health data of the drive unit obtained from the digital twin model that matches the current processing object and the moment corresponding to the sampling point.

[0054] In one embodiment, it is assumed that the first The actual observation data of the instantaneous current at each sampling point is The theoretical mean value of the instantaneous current health data of the drive unit that matches the moment corresponding to the sampling point is , then the drive unit's The calculation formula for the instantaneous current deviation data of each sampling point is:

[0055]

[0056] in, Indicates the drive unit The instantaneous current deviation data of each sampling point is used to eliminate the The individual characteristics of the changes and the processing process at all times Fluctuations in the instantaneous current of the drive unit caused by normal dynamic changes; Indicates the drive unit Actual observation data of instantaneous current at each sampling point; Indicates the The theoretical health data mean of the instantaneous current of the drive unit that matches the moment corresponding to each sampling point; represents the reference standard deviation used to normalize the signal for correction bias.

[0057] Afterwards, the calculated instantaneous current deviation data of the drive unit is used as input, and the standard bilateral CUSUM control chart is applied for real-time monitoring to calculate the upper side accumulation and the lower side accumulation. The calculation formula for the upper side accumulation of the bilateral CUSUM control chart is:

[0058]

[0059] The calculation formula for the lower cumulative value of the two-sided CUSUM control chart is:

[0060]

[0061] in, Indicates that the CUSUM control chart is The upper side accumulation of sampling points, Indicates the drive unit Instantaneous current deviation data of each sampling point; Indicates the relaxation parameter of the CUSUM control chart. In a healthy state, the standard deviation is approximately 1, so in this embodiment, is 0.5; Represents the maximum value calculation function; Indicates that the CUSUM control chart is The lower side accumulation of sampling points.

[0062] Set the decision limits of the CUSUM control chart. In this example, due to In a healthy state, the standard deviation is approximately 1, so the decision limit for the CUSUM control chart is set to 4.0. When either the upper or lower cumulative sum exceeds the set decision limit, a preliminary abnormal equipment operation event is determined to have occurred. At this time, the time window in which the preliminary abnormal event occurred and the cumulative CUSUM value that triggered the alarm are recorded. Specifically, the value of the upper or lower cumulative sum that exceeds the limit is used as the cumulative CUSUM value that triggers the alarm and the number of sample points from the start of accumulation to the alarm.

[0063] It should be noted that in order to demonstrate the execution process and technical effect of the bilateral CUSUM control chart in this step, the following calculation process is provided:

[0064] Assumed reference standard deviation ; Relaxation parameter of CUSUM control chart ; Decision limits of CUSUM control chart ; Initial cumulative sum In actual scenarios, due to early failure of the equipment, the actual current data The current starts to be slightly higher than the theoretical health current data of the current processing object predicted by the digital twin model. This deviation (e.g., the mean deviation , i.e. 1 ) For conventional The threshold alarm method based on the principle is too weak to be detected immediately.

[0065] Sample data and calculation process:

[0066] Table 1 below shows the Start with the calculation process for 10 consecutive sampling points:

[0067]

[0068] Table 1

[0069] The result shows that at the 10th sampling point, the upper cumulative sum . Corresponding technical effect: Due to Exceeded the decision limit The CUSUM control chart issued an alarm signal at the 10th sampling point in this embodiment. It can be seen from the example data that the correction deviation of each sampling point is not large, and it is difficult to detect with the conventional threshold method. Since the present invention applies the CUSUM control chart to the deviation data after the digital twin reference correction, it utilizes the cumulative effect of CUSUM on continuous small offsets and successfully detects this early small anomaly. This proves that the design of this step can effectively solve the technical problem of difficulty in detecting early faults due to changes in the processing object, and significantly improves the accuracy and sensitivity of preliminary anomaly detection.

[0070] Through the above process, this step utilizes the dynamic reference for each data point provided by the digital twin model to optimize the input signal of the CUSUM control chart, enabling it to more effectively identify the actual state anomalies of the equipment itself from complex data that includes individual differences in the processing objects and the dynamic changes of the processing process itself, thereby significantly improving the accuracy of preliminary anomaly detection and serving as the basis for subsequent precise fault diagnosis.

[0071] Step S3, obtaining an abnormality source tendency discrimination factor by fusing and evaluating the physical feature data of the split saw with the health reference data.

[0072] After confirming in step S2 that the split saw equipment has experienced an abnormal operating state using a cumulative sum control chart optimized based on the digital twin reference, the root cause of the abnormal event is analyzed and determined. While preliminary anomaly detection improves alarm accuracy, it only indicates that the equipment has an abnormality and cannot precisely determine the specific cause of the abnormality. For example, it is impossible to determine whether a fault has occurred in the drive unit itself or in the interaction between the tool and the workpiece. Because different sources of problems are ultimately addressed in very different ways, it is necessary to clearly distinguish the source of the problem. To effectively distinguish the cause of the fault, this step continues by analyzing various physical characteristics of the equipment during the time window when the anomaly occurred. Specifically, the health of the drive unit is often directly reflected by its temperature changes and the vibration characteristics of specific mechanical components. If the motor has internal problems such as overheating, poor lubrication, bearing wear, or rotor imbalance, its temperature data will exceed the normal range, or the frequency components related to these faults in its vibration signal will be significantly enhanced. On the other hand, the performance of the tool and its interaction with the workpiece are more reflected by the high-frequency vibration signals generated during the cutting process and the dynamic response of the drive unit to load changes. When a saw blade becomes worn, chipped, or loose, its contact with the workpiece produces more intense friction, impact, or unstable vibrations. These phenomena manifest as a significant increase in the energy or amplitude of the high-frequency vibration signal. This unstable cutting process also causes irregular fluctuations in the motor load, resulting in greater fluctuations in the input current of the drive motor than normal.

[0073] Therefore, the core of this step is to synchronously collect and analyze microscopic physical characteristic data from the above multiple sources within the time window where the anomaly has been confirmed. By comparing these actually observed microscopic characteristic performances with their theoretical healthy states under the current specific processing object and operating conditions, the respective deviation levels are quantified, and these deviation levels pointing to different potential fault sources are integrated and compared. First, by normalizing the microscopic physical characteristic data and the healthy reference data, a normalized deviation index reflecting the degree of anomaly in each physical dimension is obtained. Then, by applying a feature sensitivity adjustment function to the normalized deviation index, the anomaly contribution value of each feature dimension is obtained. Finally, by performing a structured fusion evaluation on the anomaly contribution values, an anomaly source tendency discriminant factor for fault source identification is obtained.

[0074] Specifically, the temperature characteristic tolerance coefficient of the drive unit, the vibration characteristic tolerance coefficient of the drive unit, the high-frequency vibration characteristic tolerance coefficient of the drive unit, and the current fluctuation characteristic tolerance coefficient of the drive unit are set;

[0075] The result of subtracting the maximum temperature data of the shell of the drive unit at the end of each processing cycle from the maximum temperature health data of the shell of the drive unit at the end of each processing cycle is used as the numerator, and the result of multiplying the temperature characteristic tolerance coefficient of the drive unit by the maximum temperature health data of the shell of the drive unit at the end of each processing cycle is used as the denominator, and the resulting fraction is used as the first temperature abnormality degree of the drive unit; the larger value between the first temperature abnormality degree of the drive unit and the constant 0 is used as the normalized temperature rise abnormality degree of the drive unit;

[0076] The result of subtracting the root mean square data of the vibration energy in the frequency band corresponding to the characteristic frequency of the fault of the outer race of the bearing at the output end of the drive unit from the root mean square health data of the vibration energy in the frequency band corresponding to the characteristic frequency of the fault of the outer race of the bearing at the output end of the drive unit is used as the numerator, and the result of multiplying the root mean square health data of the vibration energy in the frequency band corresponding to the characteristic frequency of the fault of the outer race of the bearing at the output end of the drive unit is used as the denominator, and the resulting fraction is used as the abnormality degree of the first specific vibration characteristic of the drive unit; the larger value between the abnormality degree of the first specific vibration characteristic of the drive unit and the constant 0 is used as the normalized abnormal amplitude of the specific vibration characteristic of the drive unit;

[0077] Similarly, the normalized abnormal amplitude of the high-frequency vibration characteristics of the driving unit and the normalized abnormal amplitude of the instantaneous current data fluctuation characteristics of the driving unit are obtained. Then, a characteristic sensitivity adjustment function is set, and the normalized deviation index of each physical temperature abnormality degree is mapped through the characteristic sensitivity adjustment function to obtain the abnormal contribution value of each characteristic dimension. In the embodiment of the present invention, the characteristic sensitivity adjustment function is set as , by fixing the index accordingly to highlight significant deviations.

[0078] Finally, the temperature anomaly contribution value of the drive unit, the specific vibration characteristic anomaly contribution value of the drive unit, the high-frequency vibration anomaly contribution value of the drive unit, and the current data fluctuation anomaly contribution value of the drive unit in all feature dimensions are obtained; the temperature anomaly contribution value of the drive unit and the specific vibration characteristic anomaly contribution value of the drive unit are added as the numerator, and the high-frequency vibration anomaly contribution value of the drive unit and the current data fluctuation anomaly contribution value of the drive unit are added as the denominator, and the formed fraction is used as the abnormal source tendency discrimination factor for fault source discrimination.

[0079] In one embodiment, it is assumed that the sensitivity adjustment function is ; The temperature anomaly contribution of the drive unit is ; The specific vibration characteristic abnormal contribution of the drive unit is ; The abnormal contribution of high frequency vibration of the drive unit is ; The abnormal contribution value of the current data fluctuation of the drive unit is , then the calculation formula of the abnormal source tendency discriminant factor is:

[0080]

[0081] in, It represents the discriminant factor of abnormal source tendency; Indicates the temperature anomaly contribution value of the drive unit; Indicates the abnormal contribution value of the specific vibration characteristics of the drive unit; Indicates the abnormal contribution value of high-frequency vibration of the drive unit; Indicates the abnormal contribution value of the current data fluctuation of the drive unit.

[0082] It should be noted that the core function of the abnormal source tendency discrimination factor proposed in this step is to achieve a more accurate discrimination of the root cause of the abnormality by systematically integrating and comparing multiple micro features that directly reflect the status of specific physical components or process characteristics, under the premise that the operation abnormality of the equipment has been accurately detected by the CUSUM based on digital twin reference optimization in step S2. When calculating the normalized abnormality of each micro feature, the abnormal source tendency discrimination factor uses the theoretical health reference value predicted by the digital twin model and multiplies it by a preset tolerance percentage coefficient to construct a dynamic tolerance deviation space, rather than relying on a fixed absolute alarm limit. In the numerator of the abnormal source tendency discrimination factor, that is, A drive unit fault evidence strength index is constructed, in which Indicates the temperature anomaly contribution value of the drive unit. It quantifies the degree of thermal anomaly by comparing the actual temperature performance of the drive unit with its theoretical healthy temperature reference value under the current specific working conditions and the preset alarm limit. Similarly, It represents the abnormal contribution value of the specific vibration characteristics of the drive unit. It quantifies the abnormal degree of the drive unit in specific mechanical vibration in the same way. Both of these normalized abnormality metrics use the dynamic health reference data provided by the digital twin model to compensate for the impact of changes in the processing object and working conditions on the baseline readings of these microscopic characteristics. Through a unified feature sensitivity adjustment function, the two features from different physical dimensions are transformed and accumulated. The numerator can effectively indicate the comprehensive signal strength of the internal faults of the drive unit itself, such as overheating, wear of key mechanical parts, imbalance, etc. Correspondingly, the denominator of the abnormal source tendency discrimination factor, that is, , a tool and processing object interaction abnormality intensity index was constructed, where Indicates the abnormal contribution of high-frequency vibration of the drive unit. This feature quantifies the abnormal severity of friction, impact, or chatter caused by factors such as saw blade wear, poor edge condition, or unstable installation during cutting by comparing the actual high-frequency vibration index collected from the sensor close to the tool position with its theoretical healthy reference and alarm limit under the current working conditions. The feature represents the contribution of abnormal current data fluctuations in the drive unit. This feature quantifies the smoothness of the drive unit's response to changes in cutting load by analyzing the residual fluctuations in the current signal after removing the primary load trend. When the interaction between the tool and the workpiece is unstable (e.g., when the saw blade encounters an obstruction or the cutting force is uneven), the current typically exhibits fluctuations beyond the normal range. These two normalized anomaly metrics are also transformed and accumulated using a unified feature sensitivity adjustment function, summing up the comprehensive signal strength indicating an anomaly in the tool-workpiece interaction process. By calculating the ratio of the numerator to the denominator, the anomaly source tendency discriminant factor provides a direct quantitative comparison of the strength of microscopic physical evidence for two different potential fault sources. The numerical value of the anomaly source tendency discriminant factor and its relationship to the preset discrimination threshold reveal which fault-related microscopic features are more prominent among the currently confirmed anomaly events, providing clear guidance for determining the fault cause.

[0083] In order to verify the effectiveness and operability of the abnormal source tendency discriminant factor proposed in the present invention, the following calculation example is provided:

[0084] First, the basic parameters are set, including the setting of the tolerance coefficient, in which the temperature characteristic tolerance coefficient of the drive unit is set. ; Set the vibration characteristic tolerance coefficient of the drive unit ; Set the high frequency vibration tolerance coefficient of the drive unit ; Set the instantaneous current data fluctuation tolerance coefficient of the drive unit ; Set the discrimination threshold interval to Afterwards, the abnormal contribution value of each feature is obtained through the sensitivity adjustment function, and the abnormal source tendency discrimination factor is obtained.

[0085] In the scenario of early bearing failure of the drive unit, the early wear of the motor bearing will cause its operating temperature and the vibration energy of the specific fault frequency to deviate significantly from its theoretical healthy reference value under this working condition, while the impact on the tool's high-frequency vibration and current fluctuation is relatively small. Assume a set of example data, after the alarm occurs in step S2, the maximum shell temperature data of the drive unit at the end of each processing cycle is collected and calculated within the abnormal window ; RMS data of vibration energy in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit ; RMS data of vibration energy within the high-frequency vibration characteristic frequency band of the drive unit ; Fluctuation index of instantaneous current data of drive unit At the same time, the digital twin model provides health reference data for the current working conditions ; ; ; The abnormal tendency discriminant factor is calculated from the data:

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] The abnormal tendency discrimination factor obtained by calculation is much greater than the upper threshold ,According to the judgment criteria, this result deepens the judgment that the anomaly is more likely to be caused by a fault in the drive unit itself, which is consistent with the simulation scenario.

[0092] In the case of moderate saw blade wear, the main cause of saw blade wear is instability in the cutting process, which is reflected in the tool's high-frequency vibration and current fluctuation significantly deviating from its theoretical healthy reference value, while the impact on the motor's own temperature and specific vibration is relatively small. Assume a set of example data, after the alarm occurs in step S2, the maximum housing temperature data of the drive unit at the end of each processing cycle is collected and calculated within the abnormal window. ; RMS data of vibration energy in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit ; RMS data of vibration energy within the high-frequency vibration characteristic frequency band of the drive unit ; Fluctuation index of instantaneous current data of drive unit At the same time, the digital twin model provides health reference data for the current working conditions ; ; ; The abnormal tendency discriminant factor is calculated from the data:

[0093]

[0094]

[0095]

[0096]

[0097]

[0098] The calculated abnormal tendency discriminant factor is far less than the lower threshold of 0.8. Based on the judgment criteria, this result further confirms that the abnormality is more likely due to degraded tool performance or abnormal interaction between the tool and the workpiece, which is consistent with the simulation scenario.

[0099] Step S4, obtaining the abnormality recognition result of the current processing cycle of the split saw by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor.

[0100] After obtaining the preliminary judgment results and the abnormal source tendency discrimination factor, the upper discrimination threshold and the lower discrimination threshold for the split saw abnormality recognition can be set.

[0101] An upper discrimination threshold and a lower discrimination threshold for identifying abnormalities of the split saw are set. In an embodiment of the present invention, the upper discrimination threshold is set to 1.2 and the lower discrimination threshold is set to 0.8. The threshold settings can be adjusted according to actual needs and are not required. If the abnormal source tendency discrimination factor is greater than the upper discrimination threshold, it is determined that the split saw has a fault caused by the drive unit itself in the current processing cycle; if the abnormal source tendency discrimination factor is less than the lower discrimination threshold, it is determined that the split saw has a fault caused by tool performance degradation or interaction between the tool and the processing object in the current processing cycle; if the abnormal source tendency discrimination factor is between the upper discrimination threshold and the lower discrimination threshold, it is determined that the split saw has a mixed fault in the current processing cycle and manual inspection is required.

[0102] Specifically, a factor significantly greater than the upper threshold indicates that the combined strength of the microphysical evidence supporting an internal fault in the drive unit is far stronger than the microphysical evidence supporting tool performance degradation or process instability. In this case, the diagnostic conclusion is more likely to identify the drive unit as the root cause of the anomaly. A factor significantly less than the lower threshold indicates that the combined strength of the microphysical evidence supporting tool performance degradation or process instability predominates. The diagnostic conclusion is more likely to identify the tool or its interaction with the workpiece as the root cause of the anomaly.

[0103] When the abnormal source tendency discriminant factor is between the preset upper and lower thresholds, it indicates that the comprehensive strength of the two types of micro-evidence is equivalent, or neither shows an overwhelming advantage. In this case, the failure mode is more complex, and there are mixed failures caused by multiple factors.

[0104] Step S5, obtaining the abnormality recognition result of the current processing cycle of the split saw by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor.

[0105] The digital twin-based slaughtering line abnormality recognition method consisting of the aforementioned steps S1 to S4, and the abnormality recognition system constructed based on this method, are integrated and applied to the operation management and maintenance decision-making process of the splitting saw equipment in the actual slaughtering production line. The application of this method can solve the problems of unreliable abnormality detection and insufficient judgment ability of the existing technology when facing the complex dynamic working conditions of the slaughtering line, and realize the intelligent management of the health status of key equipment.

[0106] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. The method for identifying abnormalities in slaughtering lines based on digital twins is characterized by: The method for identifying abnormalities in a slaughtering line based on digital twins includes: Step S1: collecting operating data and health reference data of the splitting saw in the slaughtering line through the data acquisition unit and the digital twin interface unit; Step S2: Obtain a preliminary abnormality judgment result of the split saw operation state by performing digital twin reference correction and cumulative sum control chart abnormality detection processing on the instantaneous current data of the drive unit; Step S3: Obtaining an abnormality source tendency discriminant factor by fusing and evaluating the physical feature data of the split saw with the health reference data; Step S4: Obtain an abnormality recognition result of the current processing cycle of the split saw by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor; Step S5: intelligent application of abnormal state of the splitting saw in the slaughter line based on the abnormality recognition result in the splitting saw processing cycle; The data acquisition unit and the digital twin interface unit are used to collect operation data and health reference data of the splitting saw in the slaughtering line, including: in the data acquisition unit of the slaughtering line, various sensors installed at key positions of the splitting saw are used to periodically collect data, and the data collected during the data acquisition process include: processing cycle data of the drive unit, instantaneous current data of the drive unit, maximum temperature data of the shell of the drive unit at the end of each processing cycle, root mean square data of vibration energy in a frequency band corresponding to the fault characteristic frequency of the outer ring of the output end bearing of the drive unit, root mean square data of vibration energy in a high-frequency vibration characteristic frequency band of the drive unit, and a volatility index of the instantaneous current data of the drive unit; In the digital twin interface unit of the slaughtering line, through the digital twin interface module deployed in the control system, the digital twin model matching the characteristics of the current processed object is called in each processing cycle to obtain reference data of the health state corresponding to the processing stage of the current processing cycle of the current processing object, specifically including the reference standard deviation of the instantaneous current data of the drive unit, the instantaneous current health data of the drive unit, the maximum shell temperature health data of the drive unit at the end of each processing cycle, the root mean square health data of the vibration energy in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit, the root mean square health data of the vibration energy in the high-frequency vibration characteristic frequency band of the drive unit, and the volatility health index of the instantaneous current data of the drive unit; The method of obtaining an abnormal source tendency discrimination factor by fusing and evaluating the physical feature data of the split saw with the healthy reference data includes: obtaining a normalized deviation index reflecting the degree of abnormality of each physical dimension by normalizing the microscopic physical feature data and the healthy reference data; obtaining an abnormal contribution value of each feature dimension by applying a feature sensitivity adjustment function to the normalized deviation index; and obtaining an abnormal source tendency discrimination factor for fault source discrimination by performing a structured fusion evaluation on the abnormal contribution values. The abnormal source tendency discrimination factor for fault source discrimination is obtained by performing structured fusion evaluation on the abnormal contribution value, including: obtaining the temperature abnormality contribution value of the drive unit in all feature dimensions, the vibration energy abnormality contribution value in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit, the vibration energy abnormality contribution value in the high-frequency vibration characteristic frequency band of the drive unit, and the current data fluctuation abnormality contribution value of the drive unit; adding the temperature abnormality contribution value of the drive unit and the vibration energy abnormality contribution value in the frequency band corresponding to the fault characteristic frequency of the outer ring of the bearing at the output end of the drive unit as the numerator, adding the vibration energy abnormality contribution value in the high-frequency vibration characteristic frequency band of the drive unit and the current data fluctuation abnormality contribution value of the drive unit as the denominator, and using the formed fraction as the abnormal source tendency discrimination factor for fault source discrimination.

2. The method for identifying abnormalities in a slaughtering line based on digital twins according to claim 1 is characterized in that: The method of obtaining a preliminary abnormality judgment result of the split saw operation state by performing digital twin reference correction and cumulative control chart abnormality detection processing on the instantaneous current data of the drive unit includes: By performing a deviation assessment on the instantaneous current data of the drive unit and the instantaneous current health data of the drive unit, the instantaneous current deviation data of the drive unit is obtained; a bilateral cumulative sum control chart is constructed by using the deviation sequence formed by the instantaneous current deviation data of the drive unit as input, and the upper cumulative sum and the lower cumulative sum are calculated respectively; by performing a decision limit assessment on the upper cumulative sum and the lower cumulative sum, a preliminary abnormality judgment result of the operating status of the split saw is obtained.

3. The method for identifying abnormalities in a slaughtering line based on digital twins according to claim 2 is characterized in that: The step of performing deviation evaluation on the instantaneous current data of the drive unit and the healthy data of the instantaneous current matching of the drive unit to obtain the instantaneous current deviation data of the drive unit includes: The calculation result of subtracting the instantaneous current data of the drive unit from the instantaneous current health data of the drive unit is used as the instantaneous current first deviation of the drive unit, and the calculation result of dividing the instantaneous current first deviation of the drive unit by the reference standard deviation of the instantaneous current data of the drive unit is used as the instantaneous current deviation data of the drive unit.

4. The method for identifying abnormalities in a slaughtering line based on digital twins according to claim 1, characterized in that: By normalizing the microscopic physical characteristic data and the health reference data, a normalized deviation index reflecting the abnormality degree of each physical dimension is obtained, including: The characteristic tolerance coefficient of the drive unit is set, including: the temperature characteristic tolerance coefficient of the drive unit, the root mean square tolerance coefficient of the vibration energy in the frequency band corresponding to the characteristic frequency of the fault of the outer ring of the output end bearing of the drive unit, the root mean square tolerance coefficient of the vibration energy in the high-frequency vibration characteristic frequency band of the drive unit, and the current fluctuation characteristic tolerance coefficient of the drive unit; the difference between the physical characteristics of the drive unit and the healthy reference data is used as the numerator, the calculation result of multiplying the healthy reference data of the drive unit by the corresponding characteristic tolerance coefficient of the drive unit is used as the denominator, and the formed fraction is used as the first characteristic abnormality degree of the drive unit; the larger value between the first characteristic abnormality degree of the drive unit and the constant 0 is used as the normalized deviation index reflecting the abnormality degree of each physical dimension.

5. The method for identifying abnormalities in a slaughtering line based on digital twins according to claim 1, characterized in that: The abnormal contribution value of each feature dimension is obtained by applying the feature sensitivity adjustment function to the normalized deviation index, including: A feature sensitivity adjustment function is set, and the normalized deviation index of the abnormality degree of each physical dimension is mapped through the feature sensitivity adjustment function to obtain the abnormal contribution value of each feature dimension.

6. The method for identifying abnormalities in a slaughtering line based on digital twins according to claim 1, characterized in that: The above-mentioned abnormality identification result of the current processing cycle of the split saw is obtained by comprehensively analyzing the preliminary abnormality judgment result and the abnormality source tendency discrimination factor, including: An upper discrimination threshold and a lower discrimination threshold are set for identifying abnormalities of the split saw; if the abnormal source tendency discrimination factor is greater than the upper discrimination threshold, it is determined that the split saw has a fault caused by the drive unit itself in the current processing cycle; if the abnormal source tendency discrimination factor is less than the lower discrimination threshold, it is determined that the split saw has a fault caused by tool performance degradation or interaction between the tool and the processing object in the current processing cycle; if the abnormal source tendency discrimination factor is between the upper discrimination threshold and the lower discrimination threshold, it is determined that the split saw has a mixed fault in the current processing cycle and manual inspection is required.

7. The slaughter line abnormality recognition system based on digital twin is characterized by: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the digital twin-based slaughter line abnormality identification method according to any one of claims 1 to 6 is implemented.

Images