Digital management system based on hardware production

Abstract

The invention provides a digital management system based on hardware production, and the system comprises the steps: obtaining the full life cycle data of equipment, distinguishing normal and abnormal data, and recognizing fault early warning data; a prediction model is trained through normal data, the minimum error of fault early warning data is calculated, and the minimum error range of the early warning interval is determined based on the minimum error; the early warning accuracy of the equipment is tested, and if the early warning accuracy exceeds the set accuracy threshold, fault prediction is carried out, so that the reliability and prediction accuracy of the production equipment are improved.

Description

Technical Field

[0001] The present invention relates to the fields of industrial automation and intelligent manufacturing. More specifically, the present invention relates to a digital management system based on hardware production. Background Art

[0002] In modern industrial production, the stability and reliability of hardware production equipment are crucial for ensuring production efficiency and product quality. With the advancement of intelligent manufacturing, more and more enterprises have started to use digital means to monitor and manage production equipment. These systems can usually collect the operating data of the equipment and predict potential faults and maintenance requirements through data analysis. However, when dealing with the full-life cycle data of equipment, the existing technologies often lack refined management of the data, especially in distinguishing normal operating data from abnormal data. In addition, the accuracy of fault warnings also needs to be improved because they usually do not take into account the specific operating conditions and historical maintenance records of the equipment.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the existing technologies: the existing prediction models often adopt a one-size-fits-all method and are not customized and trained according to the specific characteristics of each equipment, resulting in insufficient accuracy and practicability of the warning system. At the same time, the processing of abnormal data is not fine enough to effectively distinguish true fault warning signals from normal operating fluctuations. In addition, the setting of the warning interval is often too broad and lacks pertinence, resulting in limited effectiveness of the warning system in practical applications. Summary of the Invention

[0004] The present invention provides a digital management system based on hardware production, including: Obtaining the full-life cycle data of multiple hardware production equipment; the full-life cycle data is divided into normal data and abnormal data according to the usage duration; the abnormal data includes fault warning data within the warning interval in the full life cycle; Using the normal data in each hardware production equipment to train the prediction model of the corresponding equipment and obtaining the minimum error of the fault warning data in the equipment; Based on the average value of the minimum errors of the fault warning data in the multiple hardware production equipment, determining the minimum error range corresponding to the warning interval; Using multiple test hardware production equipment to test the warning accuracy rate of the minimum error range respectively, taking the proportion of the test hardware production equipment with the accuracy rate exceeding the accuracy rate threshold as the test passing rate. If the test passing rate exceeds the first test passing rate threshold, fault prediction of the hardware production equipment to be predicted is performed based on the minimum error range.

[0005] Further, each test hardware production device tests the accuracy rate of the minimum error range by performing the following steps: Train the prediction model of the test device using the normal data in the test hardware production device, and obtain the minimum error of the abnormal data in the test hardware production device; Obtain the abnormal data interval where the minimum error of the abnormal data is within the minimum error range; Obtain the accuracy rate based on the ratio of the intersection of the abnormal data interval and the warning interval to the duration of the warning interval.

[0006] Further, it also includes: if the test pass rate is between the second test pass rate threshold and the first test pass rate threshold, then adjust the prediction model parameters of the test hardware production device with an accuracy rate lower than the accuracy rate threshold or the proportion of normal data in the test hardware production device, and retrain the prediction model of the test hardware production device, re-obtain the accuracy rate and the test pass rate; Among them, the second test pass rate threshold is lower than the first test pass rate threshold.

[0007] Further, it also includes: if the test pass rate is lower than the second test pass rate threshold, then adjust one or more of the prediction model parameters of the training hardware production device, the proportion of normal data in the full life cycle data, and the range of the warning interval, and re-determine the minimum error range, re-obtain the accuracy rate and the test pass rate.

[0008] Further, adjusting the prediction model parameters of the training hardware production device and the test hardware production device includes adjusting one parameter or multiple parameters among the learning rate, weight initialization, and the size of the winning neighborhood in the model.

[0009] Further, the accuracy rate threshold is 75%.

[0010] Further, the first test pass rate is 95%, and the second test pass rate is 85%.

[0011] Further, the fault prediction of the hardware production device to be predicted based on the minimum error range includes: For each piece of normal data collected from the hardware production device to be predicted, train the prediction model of the hardware production device to be predicted using the collected normal data of the hardware production device to be predicted; For each piece of abnormal data collected from the hardware production device to be predicted, obtain the minimum error of the abnormal data in the hardware production device to be predicted; If the minimum error of the abnormal data in the hardware production device to be predicted is within the minimum error range, then perform the fault prediction of the hardware production device to be predicted.

[0012] Further, the lower limit of the minimum error range is 90% of the average value of the minimum error, and the upper limit is 110% of the average value of the minimum error.

[0013] Further, the method of using the normal data in each hardware production device to train the prediction model of the current hardware production device and obtaining the minimum error of the fault warning data in the current hardware production device includes: Using the normal data in each hardware production device to train the prediction model of the current hardware production device to obtain the best matching unit; Taking the minimum value of the distance between the fault warning data and the best matching unit as the minimum error of the fault warning data;

[0014] Wherein, represents the minimum error, represents the number of data points, represents the actual value, represents the value predicted by the prediction model.

[0015] According to the above embodiments of the present invention, there are at least the following beneficial effects: By implementing the digital management system based on hardware production of the present invention, the accuracy of fault prediction can be improved. The system obtains the full life cycle data of the hardware production device, divides it into normal data and abnormal data, and further identifies the fault warning data. By using the normal data to train the prediction model and minimizing the error of the fault warning data by optimizing the model parameters, the reliability of the prediction model can be improved. In addition, the system determines the minimum error range of the warning interval by calculating the average value of the minimum errors of the fault warning data of multiple devices, which can further enhance the accuracy of fault prediction.

[0016] In addition, the system can also dynamically adjust the prediction model and data usage strategy to meet the needs of different devices and production environments. When the test pass rate does not reach the preset threshold, the system will automatically adjust the model parameters or data ratio and retrain the model to improve the warning accuracy rate. This method can not only ensure the continuous optimization of the system, but also be flexibly adjusted according to different production conditions, so as to maintain high efficiency and high accuracy of fault prediction in various situations. By this method, the unexpected downtime can be effectively reduced, the maintenance cost can be reduced, and the overall production efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] By reading the following detailed description with reference to the drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the drawings, several embodiments of the present invention are shown by way of illustration and not limitation, wherein: Figure 1 The flowchart of the digital management system based on hardware production provided by an embodiment of the present invention. Specific embodiments

[0018] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and then implement the present invention, and do not limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to be able to fully convey the scope of the present invention to those skilled in the art.

[0019] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, equipment, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms: completely hardware, completely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0020] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0021] The following reference Figure 1 , Figure 1 is the flowchart of the digital management system based on hardware production provided by an embodiment of the present invention. As Figure 1 shown, a digital management system 100 based on hardware production includes: Step 101, obtaining the full life cycle data of multiple hardware production devices; the full life cycle data is divided into normal data and abnormal data according to the usage duration; the abnormal data includes fault warning data within the warning interval in the full life cycle; Step 102, training the prediction model of each corresponding hardware production device using the normal data in the device, and obtaining the minimum error of the fault warning data in the device; Step 103, determining the minimum error range corresponding to the warning interval based on the average value of the minimum errors of the fault warning data in the multiple hardware production devices; Step 104, using multiple test hardware production devices to test the warning accuracy rate of the minimum error range respectively, taking the proportion of the test hardware production devices with the accuracy rate exceeding the accuracy rate threshold as the test passing rate. If the test passing rate exceeds the first test passing rate threshold, fault prediction of the hardware production device to be predicted is performed based on the minimum error range.

[0022] It should be noted that this digital management system collects the full-life cycle data of the hardware production equipment, including the usage duration, performance indicators, maintenance records, etc. of the equipment, so as to achieve comprehensive monitoring of the equipment status. Among them, the full-life cycle data refers to the data accumulated during the entire cycle from the equipment being put into use to being retired. The system classifies these data into normal data and abnormal data, and further identifies fault warning data from the abnormal data, that is, the data indicating that the equipment may have a fault.

[0023] Specifically, the system analyzes the data generated during the normal operation of the equipment and uses machine learning algorithms to train prediction models. These models can learn the normal operation mode of the equipment and identify abnormal behaviors that deviate from these modes.

[0024] Furthermore, the system calculates the minimum error of the fault warning data, that is, the difference between the model prediction value and the actual value, and determines the minimum error range of the warning interval based on the average value of these errors. The warning interval refers to the time range during the equipment operation cycle that the system believes may have a fault.

[0025] Preferably, the system can set multiple test equipment to verify the warning accuracy rate of the minimum error range. The warning accuracy rate refers to the matching degree between the fault warning data predicted by the system and the data of the actually occurred faults. If the warning accuracy rate of the test equipment exceeds the set accuracy rate threshold, such as 75%, the system will consider that this minimum error range is valid and can be used for the fault prediction of the equipment to be predicted.

[0026] Even further, the system can also adjust the model parameters or data ratio according to the needs of the actual production environment to optimize the warning performance. For example, parameters such as the learning rate, weight initialization, or winner neighborhood size can be adjusted to improve the prediction accuracy of the model.

[0027] In some embodiments, each test hardware production equipment tests the accuracy rate of the minimum error range by performing the following steps: Use the normal data in the test hardware production equipment to train the prediction model of the test equipment, and obtain the minimum error of the abnormal data in the test hardware production equipment; Obtain the abnormal data interval where the minimum error of the abnormal data is within the minimum error range; Obtain the accuracy rate based on the ratio of the intersection of the abnormal data interval and the warning interval to the duration of the warning interval.

[0028] It should be noted that this embodiment describes in detail how to use the test hardware production equipment to verify the correct rate of the minimum error range. The test hardware production equipment refers to a set of equipment used to verify and test the accuracy of the prediction model. These equipment test the warning correct rate of the minimum error range by performing specific steps, that is, the accuracy of the system's prediction of faults.

[0029] Specifically, the test equipment first uses its normal data to train their respective prediction models. Normal data refers to the data collected when the equipment is not faulty, and these data reflect the standard operating conditions of the equipment.

[0030] Furthermore, the system will obtain the minimum error of the abnormal data in these test equipment. Abnormal data refers to data that is significantly different from the normal data and may indicate potential faults. Then, the system will determine the interval within the minimum error range where the minimum error of the abnormal data falls, that is, whether these error values are within the acceptable warning error range. Finally, the system calculates the ratio of the intersection of the abnormal data interval and the warning interval, and this ratio is the warning correct rate.

[0031] Preferably, in order to further improve the accuracy and robustness of the prediction model, the parameters of the prediction model in the test equipment can be adjusted. For example, the learning rate can be adjusted to optimize the learning speed of the model, or the weight initialization can be adjusted to improve the initial state of the model.

[0032] More specifically, the proportion of normal data in the training set can also be adjusted to better reflect the actual operation of the equipment. If the test pass rate is between two preset test pass rate thresholds, these parameters can be further fine-tuned to ensure that the warning performance of the system reaches the best.

[0033] Even further, more anomaly detection algorithms, such as Isolation Forest or Autoencoder, can be considered to enhance the system's ability to identify abnormal data. Through these refined operation steps and alternative solutions, the system can more accurately predict potential faults of the hardware production equipment, thereby improving production efficiency and equipment reliability.

[0034] In some embodiments, it further includes: if the test pass rate is between the second test pass rate threshold and the first test pass rate threshold, then adjust the parameters of the prediction model of the test hardware production equipment with a correct rate lower than the correct rate threshold or the proportion of normal data in the test hardware production equipment, and retrain the prediction model of the test hardware production equipment, reacquire the correct rate and the test pass rate; Wherein, the second test pass rate threshold is lower than the first test pass rate threshold.

[0035] It should be noted that this embodiment describes how to adjust the prediction model parameters or the proportion of normal data when the test pass rate is between two preset thresholds to optimize the system performance. Among them, the test pass rate refers to the proportion of the early warning correct rate in the test equipment exceeding the set threshold. The second test pass rate threshold is an index lower than the first test pass rate threshold, which is used to determine whether the system needs further adjustment.

[0036] Specifically, when the test pass rate fails to reach the first test pass rate threshold but is higher than the second test pass rate threshold, the system will identify those test equipment with an early warning correct rate lower than the set threshold. For these equipment, the system will adjust the parameters of its prediction model, such as the learning rate, weight initialization, the size of the winning neighborhood, etc., or adjust the proportion of normal data used for training the model. For example, the diversity of the equipment operation modes included in the normal data can be increased, or the learning rate of the model can be adjusted to speed up or slow down the learning process to better adapt to the actual operation conditions of the equipment.

[0037] Preferably, in addition to adjusting the prediction model parameters and the proportion of normal data, the range of the early warning interval can also be considered. The early warning interval refers to the time window when the system predicts that the equipment may have a failure. By narrowing or widening this interval, the sensitivity of the early warning can be more finely controlled. For example, if the early warning interval is set too wide, there may be too many false alarms; while if it is set too narrow, some real failure warnings may be missed.

[0038] Furthermore, the setting of the early warning interval can be optimized by analyzing historical failure data, or an adaptive mechanism can be introduced to enable the system to dynamically adjust the early warning interval according to real-time data. In addition, more data preprocessing steps, such as feature selection and data normalization, can be considered to further improve the performance and robustness of the model.

[0039] In some embodiments, it further includes: if the test pass rate is lower than the second test pass rate threshold, then adjust one or more of the prediction model parameters of the training hardware production equipment, the proportion of normal data in the full life cycle data, and the range of the early warning interval, and re-determine the minimum error range, re-obtain the correct rate and the test pass rate.

[0040] It should be noted that when the test pass rate fails to reach the second test pass rate threshold, the system will adjust the prediction model parameters of the training hardware production equipment, the proportion of normal data in the full life cycle data, or the range of the early warning interval. Here, the test pass rate refers to the proportion of the early warning correct rate in the test equipment exceeding the set threshold, and the range of the early warning interval refers to the time window when the system predicts that the equipment may have a failure.

[0041] Specifically, the system will first identify the reasons for the test pass rate being lower than the second test pass rate threshold. The parameters that may need to be adjusted include, but are not limited to, the learning rate of the prediction model, weight initialization, the size of the winning neighborhood, etc.

[0042] Furthermore, the system may also need to adjust the proportion of normal data used for training the model, or adjust the range of the warning interval to more accurately capture the timing of fault occurrence. For example, if the warning interval is set too wide, it may result in too many false alarms; if it is set too narrow, the real fault warning may be missed. Therefore, the system needs to adjust these parameters according to the actual test results to improve the accuracy of the warning.

[0043] Preferably, the system can adopt an automated parameter adjustment strategy, such as grid search or Bayesian optimization, to find the optimal combination of model parameters. In addition, the system can also introduce more data preprocessing steps, such as feature engineering and outlier handling, to improve the training effect of the model. When adjusting the range of the warning interval, the system can use statistical analysis methods, such as confidence intervals or confidence levels, to determine a reasonable warning time window.

[0044] More specifically, the system can use historical fault data to estimate the frequency and time distribution of fault occurrence, thereby providing data support for the setting of the warning interval. Through these refined operation steps and alternative solutions, the system can more flexibly adapt to different production environments and equipment characteristics, and achieve more accurate fault prediction.

[0045] In some embodiments, adjusting the prediction model parameters of the training hardware production equipment and the test hardware production equipment includes adjusting one parameter or multiple parameters among the learning rate, weight initialization, and the size of the winning neighborhood in the model.

[0046] It should be noted that this embodiment details how to adjust the prediction model parameters of the training and test hardware production equipment to improve the accuracy of fault prediction. Prediction model parameters refer to the settings that affect the model learning process and prediction results, such as the learning rate, weight initialization, and the size of the winning neighborhood. These parameters are crucial for the performance of the model because they directly affect how the model learns from the data and makes predictions.

[0047] Specifically, adjusting the parameters of the prediction model includes but is not limited to the following aspects: First, the learning rate is the step size that controls the update of the model's weights in each iteration, which can be set to a fixed value or adjusted dynamically as the training progresses; Second, weight initialization refers to assigning initial values to the model's weights before the start of training, which can be random values or values based on a certain strategy; Finally, the size of the winning neighborhood refers to considering the proximity of data points during the model training process, which can affect the model's sensitivity to abnormal data. The specific settings of these parameters depend on the complexity of the model, the characteristics of the data, and the expected prediction accuracy.

[0048] Preferably, in addition to directly adjusting the above parameters, some advanced adjustment strategies can also be adopted to optimize the model performance. For example, the method of adaptive learning rate can be used, such as the Adam optimizer, which can automatically adjust the learning rate according to the model's loss function. In addition, Xavier initialization or He initialization can be used for weight initialization. These methods consider the variance of the input data and help to avoid the problems of gradient disappearance or explosion during the training process.

[0049] Furthermore, for the adjustment of the size of the winning neighborhood, a density-based clustering algorithm such as DBSCAN can be introduced to dynamically determine the neighborhood size, so as to better capture the local structure in the data. Through these refined operation steps and alternative solutions, the model can be made more adaptable to specific data distributions and prediction tasks, thereby improving the accuracy and robustness of fault prediction.

[0050] In some embodiments, the correct rate threshold is 75%.

[0051] It should be noted that this embodiment clarifies that the threshold for the early warning correct rate is set to 75%. The early warning correct rate refers to the matching degree between the fault early warning data predicted by the system and the actually occurred fault data. This threshold is the standard for the system to judge whether the early warning is effective, ensuring that the early warning is considered reliable only when the prediction accuracy of the system reaches or exceeds this level.

[0052] Specifically, the 75% threshold of the early warning correct rate means that at least 75% of the predictions are accurate among all the predicted faults. This threshold can be adjusted according to the actual production requirements and the characteristics of the equipment to balance the sensitivity and specificity of the early warning system. For example, for critical production equipment, a higher early warning correct rate threshold may be required to reduce the possibility of false alarms and missed alarms. When setting this threshold, the impacts of false alarms and missed alarms on production and the prediction ability of the system need to be considered.

[0053] Preferably, the system can adopt a dynamic adjustment mechanism to optimize the warning accuracy rate threshold. For example, based on historical data and actual production conditions, statistical methods can be used to determine the optimal threshold. In addition, the system can also automatically adjust the warning accuracy rate threshold according to the operating status and maintenance records of the equipment to adapt to different production environments and equipment states.

[0054] More specifically, machine learning techniques, such as reinforcement learning, can be introduced to adjust the threshold in real time, enabling the system to continuously optimize the warning performance based on real-time feedback. Through these refined operation steps and alternative solutions, the system can more flexibly adapt to different production requirements and equipment characteristics, achieving more accurate fault prediction.

[0055] In some embodiments, the passing rate of the first test is 95%, and the passing rate of the second test is 85%.

[0056] It should be noted that in this embodiment, specific passing rate thresholds for the tests are defined. The passing rate threshold for the first test is set at 95%, and the passing rate threshold for the second test is set at 85%. The passing rate of the test refers to the proportion of equipment in the test of the hardware production equipment where the warning accuracy rate of the prediction model exceeds the set accuracy rate threshold. These thresholds are used to evaluate the effectiveness and reliability of the prediction model in practical applications.

[0057] Specifically, the passing rate threshold of 95% for the first test means that among all the tested equipment, at least 95% of the equipment has a warning accuracy rate of the prediction model exceeding the accuracy rate threshold of 75%. The passing rate threshold of 85% for the second test serves as a lower benchmark for determining whether the system needs further adjustment. The setting of these thresholds is based on the requirements for the performance of the production equipment and the accuracy of fault prediction, ensuring that the system can provide reliable warnings in most cases.

[0058] Preferably, the system can adopt an adaptive threshold adjustment strategy to optimize the passing rate of the test. For example, the system can dynamically adjust the passing rate thresholds of the first and second tests according to the historical performance data and maintenance records of the equipment to meet the needs of different equipment and production environments. In addition, the system can also automatically adjust these thresholds based on real-time monitoring data and prediction results to improve the adaptability and prediction accuracy of the system.

[0059] More specifically, intelligent algorithms, such as fuzzy logic or neural networks, can be introduced to analyze the equipment status and prediction results, thereby intelligently adjusting the passing rate threshold of the test. Through these refined operation steps and alternative solutions, the system can more flexibly adapt to different production conditions, achieving more accurate fault prediction and management.

[0060] In some embodiments, the fault prediction of the hardware production equipment to be predicted based on the minimum error range includes: For each piece of normal data of the hardware production equipment to be predicted collected, use the collected normal data of the hardware production equipment to be predicted to train the prediction model of the hardware production equipment to be predicted; For each piece of abnormal data of the hardware production equipment to be predicted collected, obtain the minimum error of the abnormal data in the hardware production equipment to be predicted; If the minimum error of the abnormal data in the hardware production equipment to be predicted is within the minimum error range, perform a fault prediction on the hardware production equipment to be predicted.

[0061] It should be noted that this embodiment describes the specific steps for fault prediction of the hardware production equipment to be predicted based on the minimum error range. The hardware production equipment to be predicted here refers to those equipment that are about to be used or are in use and require fault prediction to prevent potential faults. The minimum error range refers to an error interval determined by the system based on historical data and model prediction errors, and is used to judge whether the equipment may have a fault.

[0062] Specifically, the system first needs to collect the normal data of the hardware production equipment to be predicted and use these data to train the prediction model of the equipment. Normal data refers to the data collected when the equipment is not malfunctioning, and these data reflect the standard operating conditions of the equipment. Then, the system will collect the abnormal data of the equipment and calculate the minimum error of these data. The minimum error refers to the difference between the model prediction value and the actual value. If this minimum error falls within the previously determined minimum error range, the system will perform a fault prediction and prompt the possible faults.

[0063] Preferably, the system can adopt a variety of data collection technologies and model training methods to improve the accuracy of fault prediction. For example, sensors can be used to collect the real-time operation data of the equipment, or historical maintenance records can be used to enrich the data set. In terms of model training, deep learning or ensemble learning methods can be adopted to improve the model's ability to recognize complex data patterns.

[0064] More specifically, the system can set up an automatic update mechanism to regularly retrain and optimize the prediction model with the latest equipment data to ensure that the model can adapt to the changes in the equipment state. In addition, the system can also introduce an anomaly detection algorithm to identify and process the outliers in the data to improve the accuracy of the minimum error calculation. Through these refined operation steps and alternative solutions, the system can more effectively predict and manage the potential faults of the hardware production equipment.

[0065] In some embodiments, the lower limit of the minimum error range is 90% of the average value of the minimum error, and the upper limit is 110% of the average value of the minimum error.

[0066] It should be noted that this embodiment elaborates on the specific calculation method of the minimum error range. The lower limit of the minimum error range is set to 90% of the average value of the minimum error, and the upper limit is set to 110% of the average value of the minimum error. Here, the average value of the minimum error refers to the average value of the prediction errors of all data points during the model training process, and this value reflects the accuracy of the model prediction. By setting the upper and lower limits of the error range, the system can determine a reasonable error interval for evaluating and predicting possible faults of the device.

[0067] Specifically, the system first needs to calculate the average value of the minimum error of all test device fault warning data. This average value is obtained by adding up the prediction errors of all data points and then dividing by the total number of data points. Then, the system will set the lower and upper limits of the minimum error range based on this average value. For example, if the average value of the minimum error is 5, then the lower limit is 4.5 (90%), and the upper limit is 5.5 (110%). This range provides a benchmark for the system to determine whether the prediction error of the device is within an acceptable range.

[0068] Preferably, the system can adopt a dynamic adjustment mechanism to optimize the setting of the minimum error range. For example, the system can automatically adjust the upper and lower limits of the minimum error range based on the historical performance data and maintenance records of the device to meet the requirements of different devices and production environments. In addition, the system can also automatically adjust these ranges according to real-time monitoring data and prediction results to improve the adaptability and prediction accuracy of the system.

[0069] More specifically, machine learning techniques such as reinforcement learning or adaptive filters can be introduced to adjust the setting of the minimum error range in real time. Through these refined operation steps and alternative solutions, the system can more flexibly adapt to different production conditions and achieve more accurate fault prediction and management.

[0070] In some embodiments, the method of using the normal data in each hardware production device to train the prediction model of the current hardware production device and obtaining the minimum error of the fault warning data in the current hardware production device includes: Using the normal data in each hardware production device to train the prediction model of the current hardware production device to obtain the best matching unit; Taking the minimum value of the distance between the fault warning data and the best matching unit as the minimum error of the fault warning data; Wherein, represents the minimum error, represents the number of data points, represents the actual value, Represents the value predicted by the prediction model. This formula is used to calculate the prediction accuracy of the prediction model by minimizing the square root of the average of the squared differences between the actual value and the predicted value.

[0071] It should be noted that this embodiment describes how to use the normal data in the hardware production equipment to train the prediction model and calculate the minimum error of the fault warning data. Here, the normal data refers to the data collected when the equipment is not malfunctioning, and these data are used to build a model to predict the future performance and potential faults of the equipment. The prediction model refers to a model constructed by machine learning algorithms that can predict the behavior and state of the equipment based on the input data.

[0072] Specifically, the system first needs to collect the data of the hardware production equipment during normal operation. These data may include sensor readings such as temperature, pressure, vibration, as well as the operating parameters and maintenance records of the equipment. Then, use these data to train a prediction model that can learn the normal operating mode of the equipment and predict the possible faults that may occur in the future.

[0073] Furthermore, during the model training process, the system will calculate the minimum error of the fault warning data, that is, the difference between the model predicted value and the actual value. The formula for the minimum error is the square root of the average of the sum of the squared prediction errors, and the formula is as follows: Among them, Represents the actual value, Represents the predicted value, Represents the number of data points.

[0074] Preferably, the system can adopt a variety of machine learning algorithms to train the prediction model, such as support vector machines, random forests or neural networks, to adapt to different types of data and prediction tasks. In addition, the system can also introduce feature selection and data preprocessing steps to improve the training effect and prediction accuracy of the model.

[0075] More specifically, the system can set up an automatic feature selection mechanism to identify the data features that have the most impact on fault prediction. At the same time, the system can also adopt data normalization or standardization techniques to ensure the scale consistency of different features, thereby improving the performance of the model. Through these refined operation steps and alternative solutions, the system can more effectively predict and manage the potential faults of the hardware production equipment.

[0076] The above embodiments of the present invention have the following beneficial effects: The digital management system based on hardware production of the present invention can perform refined management on the full-life cycle data of production equipment, and improve the accuracy of fault warning by training a prediction model and determining the minimum error range. The system optimizes the fault prediction performance by testing the warning accuracy rate of the equipment and dynamically adjusting the model parameters or data ratio according to the test passing rate. This method can ensure that the system can maintain an efficient fault warning ability under various production conditions and reduce production interruptions caused by equipment failures.

[0077] The system of the present invention sets specific warning accuracy rate thresholds, test passing rate thresholds, and calculation methods for the minimum error range. The setting of these parameters provides clear operation guidance for the system, making fault prediction more accurate and reliable. In addition, when the system collects new equipment data, it can update the prediction model in real time to ensure the timeliness and accuracy of the prediction results. By this method, the operation efficiency and reliability of production equipment can be further improved, maintenance costs can be reduced, and the market competitiveness of the enterprise can be enhanced.

[0078] Furthermore, the storage medium of the implementation manner of the present application stores program instructions capable of implementing all the above methods. Among them, the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in various implementation manners of the present application. The foregoing storage medium includes: various media that can store program codes such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, or terminal devices such as computers, servers, mobile phones, and tablets.

[0079] The above description is only some preferred implementation manners of the present invention and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the implementation manner of the present invention is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, technical solutions formed by mutually replacing the above features with (but not limited to) technical features having similar functions disclosed in the implementation manner of the present invention.

Claims

1. A digital management system based on hardware production, characterized in that The system performs the following steps: Obtain the full life cycle data of multiple hardware production devices; the full life cycle data is divided into normal data and abnormal data according to the usage duration; the abnormal data contains fault warning data within the warning interval in the full life cycle; Use the normal data in each of the hardware production devices to train the prediction model of the corresponding device, and obtain the minimum error of the fault warning data in the device; Based on the average value of the minimum errors of the fault warning data in the multiple hardware production devices, determine the minimum error range corresponding to the warning interval; Use multiple test hardware production devices to test the warning accuracy rate of the minimum error range respectively, and take the proportion of the test hardware production devices with the accuracy rate exceeding the accuracy rate threshold as the test passing rate. If the test passing rate exceeds the first test passing rate threshold, perform fault prediction on the hardware production device to be predicted based on the minimum error range.

2. The digital management system based on hardware production according to claim 1, wherein Each test hardware production device tests the accuracy rate of the minimum error range by performing the following steps: Use the normal data in the test hardware production device to train the prediction model of the test device, and obtain the minimum error of the abnormal data in the test hardware production device; Obtain the abnormal data interval where the minimum error of the abnormal data is within the minimum error range; Obtain the accuracy rate based on the ratio of the intersection of the abnormal data interval and the warning interval to the duration of the warning interval.

3. The digital management system based on hardware production according to claim 2, wherein It further includes: If the test passing rate is between the second test passing rate threshold and the first test passing rate threshold, adjust the prediction model parameters of the test hardware production device with the accuracy rate lower than the accuracy rate threshold or the proportion of normal data in the test hardware production device, and retrain the prediction model of the test hardware production device, re-obtain the accuracy rate and the test passing rate; Wherein, the second test passing rate threshold is lower than the first test passing rate threshold.

4. The digital management system based on hardware production according to claim 2, wherein It further includes: If the test passing rate is lower than the second test passing rate threshold, adjust one or more of the prediction model parameters of the training hardware production device, the proportion of normal data in the full life cycle data, and the range of the warning interval, and re-determine the minimum error range, re-obtain the accuracy rate and the test passing rate.

5. The digital management system based on hardware production according to claim 4, wherein Adjusting the prediction model parameters of the training hardware production device and the test hardware production device includes adjusting one parameter or multiple parameters such as the learning rate, weight initialization, and the size of the winning neighborhood in the model.

6. The digital management system based on hardware production according to any one of claims 1-5, characterized in that, The accuracy rate threshold is 75%.

7. The digital management system based on hardware production according to claim 6, wherein The first test passing rate is 95%, and the second test passing rate is 85%.

8. The digital management system based on hardware production according to claim 1, characterized in that Performing fault prediction on the hardware production device to be predicted based on the minimum error range includes: Every time the normal data of a hardware production device to be predicted is collected, use the collected normal data of the hardware production device to be predicted to train the prediction model of the hardware production device to be predicted; Every time the abnormal data of a hardware production device to be predicted is collected, obtain the minimum error of the abnormal data in the hardware production device to be predicted; If the minimum error of the abnormal data in the hardware production device to be predicted is within the minimum error range, perform fault prediction on the hardware production device to be predicted.

9. The digital management system based on hardware production according to claim 1, wherein The lower limit of the minimum error range is 90% of the average value of the minimum error, and the upper limit is 110% of the average value of the minimum error.

10. The digital management system based on hardware production according to claim 1, wherein Training the prediction model of the current hardware production equipment by using the normal data in each hardware production equipment, and obtaining the minimum error of the fault warning data in the current hardware production equipment, includes: Training the prediction model of the current hardware production equipment by using the normal data in each hardware production equipment to obtain the best matching unit; Taking the minimum value of the distance between the fault warning data and the best matching unit as the minimum error of the fault warning data; Among them, represents the minimum error, represents the number of data points, represents the actual value, represents the value predicted by the prediction model.

Images