Data driving and QMU fused equipment quality state evaluation method

Abstract

The invention provides an equipment quality state evaluation method fusing data driving and QMU, and belongs to the technical field of equipment evaluation and detection. The method comprises the steps of collecting test data and task environment data of equipment to construct a data set, and dividing the data set into a training set and a test set; based on the data set, importance calculation is carried out on the multi-dimensional features through a limit gradient boosting tree, the features meeting the equipment quality state contribution degree requirement are screened out, and an observation list of the data set is generated; based on the observation list of the training set, a Gaussian process regression model is established, and hyper-parameters of the Gaussian process regression model are optimized; predicting the equipment state by using a Gaussian process regression model, performing uncertainty quantization on a prediction result, and outputting a mean value and a variance of an equipment state prediction value; on the basis of the performance threshold PTL or PTU of the equipment, a predicted value of the margin is calculated in combination with the average value of equipment state prediction; and calculating a confidence coefficient CF based on the variance of the predicted value and the predicted value of the margin, and affirming the quality state of the equipment.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment evaluation and detection, and more particularly relates to an equipment quality status evaluation method integrating data drive and QMU. Background Art

[0002] Equipment condition assessment is the process of inspecting, testing, and diagnosing the equipment's operating status. The goal is to understand the equipment's performance, potential failures, and wear, providing a basis for subsequent maintenance and repairs. Equipment condition assessment involves multiple inspections, including appearance, vibration, sound, temperature, oil sampling, and more in-depth testing using specialized instruments and software.

[0003] Regular equipment status assessments can promptly identify potential failures and performance degradation trends, enabling appropriate maintenance measures to be implemented to avoid production interruptions and safety incidents caused by sudden equipment failures. Timely maintenance and repairs can mitigate equipment wear and aging, extend equipment life, and reduce replacement costs. Equipment status assessments can help understand equipment performance and enable targeted adjustments and optimizations to improve operational efficiency and production quality.

[0004] Using a Gaussian Process Regression (GPR) model to predict device performance parameters, it is possible to predict the performance parameters of each device component and, consequently, identify weaknesses within the device and its components. The GPR model is a nonparametric probabilistic model that leverages the data's prior distribution to estimate the posterior distribution. In practical applications, it can better incorporate external environmental factors, test values, and model noise, resulting in relatively accurate predictions. The model's greatest advantage is that it not only provides a point estimate of the predicted value but also expresses and quantifies the uncertainty of the predicted value. The basic concept is to assume that the predicted performance parameter follows a Gaussian distribution. By inputting normalized measured parameter values, the input data are fitted using the Gaussian process prior distribution. A Bayesian framework is then used to calculate the posterior distribution, and the mean and variance of the predicted results are output to predict parameter trends. A disadvantage of this method is that it assumes that the predicted values at each moment follow a Gaussian distribution, which can lead to significant cumulative errors over long periods of time. Summary of the Invention

[0005] In view of the above problems, the purpose of the present invention is to provide a device quality status assessment method that integrates data-driven and QMU. The device quality status is evaluated by integrating eXtreme Gradient Boosting (XGBoost), GPR and Quantification of Margin and Uncertainty (QMU), thereby improving the efficiency and accuracy of prediction.

[0006] To achieve the above-mentioned purpose, the present invention is implemented through the following technical solutions: In a first aspect, an embodiment of the present application provides a device quality status assessment method integrating data-driven and QMU, including: Collect equipment test data and mission environment data to build a dataset, and divide the dataset into a training set and a test set; Based on the data set, the importance of multi-dimensional features is calculated through the extreme gradient boosting tree. According to the importance threshold, the features that meet the requirements of the contribution of equipment quality status are screened out to generate the observation list of the data set; Based on the observation list of the training set, a Gaussian process regression model is established and its hyperparameters are optimized; The optimized Gaussian process regression model is used to predict the equipment status, and the uncertainty of the prediction results is quantified to output the mean and variance of the equipment status prediction value; The predicted value of margin is calculated based on the performance threshold PTL or PTU of the equipment and the mean value of the equipment status prediction; The confidence coefficient CF is calculated based on the variance of the equipment status prediction value and the predicted value of the margin, and the confidence coefficient CF is used to identify the quality status of the equipment.

[0007] In an optional embodiment, the collecting device test data and task environment data to construct a data set, and dividing the data set into a training set and a test set, includes: Collect test data and task environment data of the same model equipment in multiple time periods to form a new data table, and build a data set based on the data table ,in and represent the independent variable and the dependent variable respectively, Indicates the samples, and Represent the number of samples and feature dimensions respectively; The dataset D is divided into a training set and a test set according to a preset ratio.

[0008] In an optional embodiment, the importance of multidimensional features is calculated based on the data set by using an extreme gradient boosting tree, and features that meet the equipment quality status contribution requirements are screened out according to the importance threshold, and an observation list of the data set is generated, including: Based on the dataset D, several regression trees are formed to build the initial XGBoost model, as follows:

[0009]

[0010]

[0011] Where L is the objective function, is the loss function, represents the predicted value of the sample, is the regularization term, and represents the regularization coefficient, represents the feature importance, For the A regression tree model, Indicates the Related functions in the tree, represents the regression tree space, Indicates that the sample A function in the tree structure that maps to the corresponding leaf index, Represents the number of leaves; Solve the initial XGBoost model through an iterative algorithm, optimize the objective function and calculate the feature importance score; Based on the feature importance score, the features with scores below the preset threshold are filtered out according to the preset threshold; Generate an observation list of dataset D based on the filtered features.

[0012] In an optional embodiment, the establishing of a Gaussian process regression model based on the observation list of the training set and optimizing its hyperparameters includes: The feature x of the observation list of the training set is used as the input of the model, and the device status y is used as the output. The established Gaussian process model is:

[0013] in, is the selected mean function, and is a constant, is the selected covariance function, is the variance of the noise, is the Dirac function, that is hour ; The hyperparameters that need to be optimized are ; Hyperparameters are obtained by maximizing the log-likelihood function The optimal value of The log-likelihood function is:

[0014] in, is the matrix of covariance function, is the unit matrix, is the mean vector.

[0015] In an optional embodiment, the device state is predicted using the optimized Gaussian process regression model, and uncertainty of the prediction result is quantified to output the mean and variance of the device state prediction value, including: Output value of real-time test data of the equipment Output value of the training set Based on , it is brought into the optimized Gaussian regression process model, which obeys the following joint Gaussian distribution:

[0016] in, and are the input vectors of the training set and the test set respectively, is the mean vector outputted by the test data, ; The device's input data is obtained through the log-likelihood function and the joint Gaussian distribution. The mean of the state prediction values under and variance ,Right now:

[0017]

[0018] By variance Determine confidence intervals for the predicted output values of a Gaussian regression process.

[0019] In an optional embodiment, the calculation of the predicted value of the margin based on the performance threshold PTL or PTU of the device and the mean value of the device state prediction includes: By formula or Calculate the predicted value M of the margin; in, PTL and PTU They are the lower and upper limits of the device's factory status threshold.

[0020] In an optional embodiment, the calculation of the confidence coefficient CF based on the variance of the equipment state prediction value and the predicted value of the margin, and the use of the confidence coefficient CF to identify the quality state of the equipment include: The confidence coefficient CF is calculated using the following formula:

[0021] Where U represents the uncertainty of the predicted value of the equipment state; If the confidence coefficient CF is greater than 1, it is determined that the equipment quality status meets the reliability requirements; if CF is less than or equal to 1, it is determined that the equipment quality status does not meet the reliability requirements.

[0022] In a second aspect, an embodiment of the present application further provides an equipment quality status assessment device integrating data-driven and QMU, including: The dataset construction module is used to collect equipment test data and task environment data to construct a dataset, and divide the dataset into a training set and a test set; The observation list construction module is used to calculate the importance of multi-dimensional features based on the data set through the extreme gradient boosting tree. According to the importance threshold, it selects features that meet the equipment quality status contribution requirements and generates an observation list for the data set; Model building and optimization module, which is used to build a Gaussian process regression model based on the observation list of the training set and optimize its hyperparameters; The state prediction module is used to predict the equipment state using the optimized Gaussian process regression model, quantify the uncertainty of the prediction results, and output the mean and variance of the equipment state prediction value; The margin prediction module is used to calculate the predicted margin value based on the performance threshold PTL or PTU of the equipment and the average value of the equipment status prediction; The equipment quality status identification module is used to calculate the confidence coefficient CF based on the variance of the equipment status prediction value and the predicted value of the margin, and use the confidence coefficient CF to identify the quality status of the equipment.

[0023] In a third aspect, an embodiment of the present application further provides an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the device quality status assessment method integrating data-driven and QMU as described in any one of the above items are implemented.

[0024] In a fourth aspect, an embodiment of the present application further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the device quality status assessment method integrating data drive and QMU as described in any one of the above items are implemented.

[0025] It can be seen from the above technical solutions that the present invention has the following advantages: The equipment quality status assessment method provided in this application, which integrates data-driven and QMU methods, integrates XGBoost, GPR, and QMU methods to evaluate the equipment quality status, thereby improving the efficiency and accuracy of prediction. The XGBoost method is used to determine the observation list, which can filter out features related to the equipment quality status from a large number of features, simplifying the subsequent equipment status prediction model while improving its prediction speed; the GPR method can not only fit the degradation evolution process of the equipment status, but also quantify the uncertainty of the equipment status prediction results; finally, the QMU method is used to identify the quality status of the equipment in order to provide a clear maintenance plan.

[0026] This application integrates equipment test data and mission environment data to construct a multidimensional dataset, overcoming the limitations of a single data source and comprehensively capturing the factors affecting equipment status. It uses the Extreme Gradient Boosting Tree (XGBoost) algorithm to quantify feature importance, dynamically screen key features, and generate an observation list, effectively eliminating redundant or noisy dimensions. This process not only simplifies the input space of the Gaussian Process Regression (GPR) model and reduces computational complexity, but also avoids the risk of overfitting caused by the curse of dimensionality, significantly improving model training efficiency and prediction accuracy, and achieving a deep synergy between data-driven and physical constraints.

[0027] This application constructs an equipment state prediction model based on Gaussian process regression (GPR). Leveraging its nonparametric nature, it flexibly fits the complex, nonlinear relationships between equipment states and characteristics, overcoming the strong assumptions about data distribution that restrict traditional parametric models. The model outputs dual metrics—mean and variance—to explicitly quantify the uncertainty of the prediction results, providing a probabilistic description for equipment state assessment. By optimizing hyperparameters using a covariance function, the model's robustness to noise and external interference is further enhanced, making predictions more accurate and aligning with actual operating conditions, supporting more precise maintenance decisions.

[0028] This application introduces a Quantified Margin and Uncertainty (QMU) framework, linking equipment performance thresholds (PTL / PTU) with data-driven prediction results to construct a dual-factor assessment system consisting of margin (M) and confidence factor (CF). By calculating the difference between the predicted mean and the threshold, the safety margin is quantified. The confidence factor is dynamically adjusted based on the predicted variance, transforming traditional threshold comparisons into probabilistic judgments (CF > 1 indicates safety). This approach avoids the conservatism or aggressiveness of single-threshold judgments, complementing physical rules with statistical models, and significantly enhancing the interpretability and engineering applicability of the assessment results.

[0029] This application's method supports online data updates and can incorporate newly collected test and environmental data in real time, enabling dynamic monitoring and trend prediction of equipment status. The output of the mean-variance-confidence ternary result provides a quantitative basis for maintenance decisions (e.g., triggering an early warning when the CF approaches 1), promoting the transition from "scheduled maintenance" to "condition-based maintenance." By identifying performance degradation trends early, companies can optimize maintenance plans, reduce unplanned downtime, lower lifecycle costs, and extend equipment lifespan.

[0030] This application constructs a multi-level risk quantification mechanism by jointly assessing margin and uncertainty: margin (M) reflects the distance between the equipment status and the failure boundary, and confidence factor (CF) comprehensively characterizes the prediction credibility. When CF > 1, the equipment status statistically meets the reliability requirements; when CF ≤ 1, a risk warning is triggered. This mechanism not only provides clear maintenance decision thresholds but also reveals potential risk sources through uncertainty propagation analysis, providing data support for equipment upgrades or process optimization, significantly improving the safety redundancy of complex industrial systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for the description. 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.

[0032] Figure 1 This is a flowchart of the equipment quality status assessment method that integrates data-driven and QMU provided in this application.

[0033] Figure 2 A bar chart showing the importance of each feature of the engine provided in this application.

[0034] Figure 3 Schematic diagram of the engine quality state degradation trajectory and uncertainty quantification results provided in this application.

[0035] Figure 4 This is a schematic diagram of the structure of the equipment quality status assessment device that integrates data drive and QMU provided in this application.

[0036] Figure 5 This is a schematic diagram of the structure of the electronic device provided in this application. DETAILED DESCRIPTION

[0037] The specific steps of the device quality status assessment method integrating data-driven and QMU will be described in detail below, and various embodiments of the present disclosure will be described more fully. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but rather that the present disclosure should be understood to cover all adjustments, equivalents, and / or alternatives that fall within the spirit and scope of the various embodiments of the present disclosure.

[0038] Hereinafter, the terms "include" or "may include" as used in various embodiments of the present disclosure indicate the presence of disclosed functions, operations, or elements, and do not limit the addition of one or more functions, operations, or elements. In addition, as used in various embodiments of the present disclosure, the terms "include," "have," and their cognates are intended only to indicate specific features, numbers, steps, operations, elements, components, or combinations of the foregoing, and should not be understood as excluding the presence of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing, or the possibility of adding one or more features, numbers, steps, operations, elements, components, or combinations of the foregoing.

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] See also Figure 1 FIG2 is a flowchart of a method for evaluating equipment quality status by integrating data-driven and QMU in a specific embodiment. The method includes the following steps: S1: Collect the equipment test data and task environment data to build a dataset, and divide the dataset into a training set and a test set.

[0041] In a specific embodiment, the test data and task environment data of the same model equipment in multiple time periods are collected to form a new data table, and a data set is constructed based on the data table. ,in and represent the independent variable and the dependent variable respectively, Indicates the samples, and Represent the number of samples and feature dimensions respectively; 60% to 75% of the data in dataset D are selected as the training set, and the rest are used as the test set.

[0042] S2: Based on the data set, the importance of multidimensional features is calculated through the extreme gradient boosting tree. According to the importance threshold, the features that meet the contribution requirements of the equipment quality status are screened out to generate an observation list of the data set.

[0043] In a specific embodiment, XGBoost-based observation list construction is a technique that uses ensemble learning methods to select important features. Feature importance is measured by calculating the contribution of each feature to the model's objective function when splitting a node in the decision tree. The greater the contribution, the stronger the feature's ability to predict the objective function. The specific process of this step is as follows: S201. Build the initial XGBoost model (all features as observation list): Using the original dataset Form several regression trees (CART) as base models, where and represent the independent variable and the dependent variable respectively, Indicates the samples, and Represent the number of samples and feature dimensions respectively. Usually this model is not optimal, but for each sample , which can provide an initial prediction value .

[0044] S202, evaluate the importance of features: evaluate the importance of each feature by the following method to characterize its contribution to the objective function. The following objective function is used to solve the feature importance (1) (2) (3) in, is the loss function, Represents the predicted value of the sample, in formula (2) is the regularization term, and represents the regularization coefficient, represents the feature importance, For the Granular Regression Tree (CART) model, in formula (3) Indicates the Related functions in the tree, represents the regression tree space, Indicates that the sample A function in the tree structure that maps to the corresponding leaf index, Represents the number of leaves.

[0045] At this point, an iterative algorithm is needed to solve the above model: for each iteration of the model, the objective function of the previous iteration will be optimized. The predicted value for the iteration is: (4) Objective function Can be converted to the following forms: (5) According to the second-order Taylor expansion ,make , ,Right now: (6) In order to minimize the objective function, the second-order Taylor expansion of equation (6) is introduced into the objective function We get: (7) Assume that the current number of iterations is ,So For training The objective function when the tree is is The constant cumulative loss of all true values and errors before the tree is not involved in the optimization process, so this part is discarded. , , the objective function Can be simplified to: (8) Minimize Equation (8): (9) Because in During the iteration process, and It is known that The residual term generated during the iteration process needs to be continued to optimize the objective function. Parameterized.

[0046] like It is During the iteration Samples generated by the tree The prediction result of the tree model is Samples are planned to a leaf node. Define the function Indicates the location of the sample, that is, the sample Falling in the leaf nodes, so The value can be used This completes the Parameterization of . Defined as , indicating a leaf node Which samples are there in the dataset? Reorganize the samples according to the position of the leaf nodes.

[0047] Substitute equation (2) into the simplified objective function The final simplified version is obtained: (10) Minimizing formula (10) can find the importance of each feature: (11) Will Substitute into formula (16) to obtain the corresponding optimal function value : (12) S203. Remove low-importance features: Based on the feature importance scores, filter out low-importance features according to a certain threshold, and generate an observation list of the dataset D based on the filtered features.

[0048] S204. Retrain the model: After removing the features with low importance, we can retrain the XGBoost model.

[0049] S205, repeat the above steps: Repeat the above steps until the requirements are met.

[0050] S3: Based on the observation list of the training set, a Gaussian process regression model is established and its hyperparameters are optimized.

[0051] In a specific implementation manner, the specific process of this step is as follows: S301. Perform factor analysis before prediction. Determine the correlation between variables, determine the input and output for prediction, and select the mean function and covariance function.

[0052] S302: Establish a Gaussian process regression model. Take the feature x of the observation list of the training set as the input of the model and the device status y as the output. The established Gaussian process model is: (13) in, is the selected mean function, and is a constant, is the selected covariance function, is the variance of the noise, is the Dirac function, that is hour ; The hyperparameters that need to be optimized are .

[0053] S303, optimize hyperparameters: Gaussian process regression model optimization, that is, the hyperparameters Solve for the optimal value, where , the optimal value of the hyperparameter can be obtained by maximizing the log-likelihood function of the following formula: The log-likelihood function expression is: (14) in, is the matrix of covariance function, is the unit matrix, is the mean vector. The hyperparameters are initialized to random values. Then, a global optimization algorithm, such as a genetic algorithm or particle swarm optimization, is used to iteratively learn and solve the above equation, searching for the optimal values of the hyperparameters. At this point, the GPR model is the optimal prediction model.

[0054] S4: Use the optimized Gaussian process regression model to predict the device status, quantify the uncertainty of the prediction results, and output the mean and variance of the device status prediction value.

[0055] In a specific embodiment, first, the output value of the test data is Output value of the training data As a basis, it is brought into the optimized Gaussian regression process model, that is, it obeys the joint Gaussian distribution: (15) in, and are the input vectors of the training set and the test set respectively, is the mean vector outputted by the test data, .

[0056] Then, the mean of the equipment state prediction value is obtained through equations (20) and (21): and variance ,Right now (16) and (17) The confidence interval of the predicted output value of the Gaussian regression process can be determined by formula (17).

[0057] S5: Calculate the predicted value of the margin based on the performance threshold PTL or PTU of the device and the average value of the device status prediction.

[0058] In a specific embodiment, based on , calculate the predicted value of the margin ; The calculation formula is: or

[0059] in, PTL and PTU They are the lower and upper limits of the device's factory status threshold.

[0060] S6: Calculate the confidence coefficient CF based on the variance of the equipment status prediction value and the predicted value of the margin, and use the confidence coefficient CF to identify the quality status of the equipment.

[0061] In a specific embodiment, based on the predicted value of the margin and variance , the confidence coefficient CF is calculated, and the reliability of the equipment quality status is determined by the confidence coefficient CF.

[0062] The calculation formula of confidence coefficient CF is: (18) Where U represents the uncertainty of the predicted value of the equipment state.

[0063] If the confidence coefficient CF is greater than 1, it is determined that the equipment quality status meets the reliability requirements; if CF is less than or equal to 1, it is determined that the equipment quality status does not meet the reliability requirements.

[0064] It's important to note that the margin and uncertainty quantification (QMU) used in this invention is based on the physics of failure and margin design. This theory posits that performance parameters all exist within a reliable domain. If a device's initial state is within this reliable domain, the device is considered reliable. However, if it is outside this reliable domain, the product corresponding to this initial state is considered unreliable based on the principle of conservatism. In other words, the device or product can only operate safely within this reliable domain or performance channel; if it exceeds this channel, it is highly likely to fail.

[0065] In this invention, margin is specifically defined as follows: During the design phase, device performance is typically assigned a theoretical design point. Based on the range of a performance channel (established by performance thresholds), the designed performance indicator, or theoretical design point, is typically contained within the channel, but not necessarily located at this design point. Due to design tolerances or the influence of storage and operating environments, this design point typically fluctuates within a range. To ensure that the system operates properly within the specified time and conditions, a safe distance, or performance margin, should be maintained between the design point and the boundaries of the reliability region, or the operating range limits. However, determining the performance channel and margin is subject to random and epistemic uncertainty due to variations in the operating environment and inherent product performance. It is generally assumed that when the performance margin exceeds the uncertainty U, the device may malfunction or fail.

[0066] In summary, the basic idea of the QMU-based equipment quality assessment method of the present invention is to first analyze the equipment or its components to identify performance parameters that can represent the reliability of the equipment or components. Through simulation or testing of the performance of the equipment or components, the performance thresholds of each parameter under normal operation of the equipment or components in a given time and environment are determined. A performance channel is established, and the margin and uncertainty in the assessment process are quantified. The confidence coefficient (CF) is used to determine whether the performance parameters are within the performance channel. If the CF of all performance parameters is greater than 1, the system is considered to meet the reliability requirements; if the CF of any performance parameter is less than or equal to 1, the system is considered to not meet the reliability requirements.

[0067] In this embodiment, by integrating the extreme gradient boosting tree (XGBoost) feature screening, Gaussian process regression (GPR) state prediction and QMU (margin and uncertainty quantification) evaluation framework, the equipment quality state assessment is made more precise and scientific: first, based on data-driven feature importance analysis, redundant dimensions are eliminated to optimize model efficiency and reduce overfitting risks; second, the non-parametric modeling capability of GPR is used to capture the evolution law of equipment states, and the predicted mean and variance are output simultaneously to achieve explicit quantification of uncertainty; third, the physical performance threshold is associated with the statistical prediction results through the QMU framework, and a margin-confidence dual-factor judgment mechanism is constructed to avoid the limitations of single threshold judgment; finally, dynamic data updates and real-time risk warnings are supported, providing a quantitative decision-making basis for predictive maintenance, and significantly improving the accuracy, robustness and engineering applicability of equipment reliability assessment.

[0068] In one embodiment of the present invention, in order to more specifically illustrate the specific execution process of the equipment quality status assessment method integrating data drive and QMU of the present invention, the specific implementation process of the method is given below based on example data.

[0069] For example, a dataset for a specific engine includes basic information such as operating conditions, fault types, and data volume. The dataset consists of 26 columns: the first and second columns are the engine number and detection time, respectively; columns 3 through 5 are engine operating parameters; and the last 21 columns are sensor data (see Table 1). Each engine runs from a healthy state until a fault occurs.

[0070] Table 1: Parameters of an engine sensor

[0071] The experiment selected the first 75% of the monitoring cycle data of the dataset as the training set, and the remaining 25% of the data as the test set. The 21 sensor data and 3 operating condition data were normalized and used as the input of the XGBoost model. The importance of each feature is as follows: Figure 2 As shown in Figure 2, sensors 13, 14, 7, 21, 17, 20, 3, 12, 8, 11, 4, 2, 9 and operating conditions 1 and 2 are finally selected as the observation list. The observation list data of the training set is input into the GPR prediction model, and the hyperparameters are dynamically optimized through the conjugate gradient algorithm to obtain the trained GPR prediction model. The observation list of the test set is input into the trained GPR prediction model, and the mean and variance of the quality state of each numbered engine are calculated. The quality state degradation trajectory and uncertainty quantification results of a certain numbered engine are shown in Figure 2. Figure 3 By calculating the CF value of each monitoring cycle, it can be seen that the CF value after 196 cycles is less than 1, and it should be repaired before this time.

[0072] like Figure 4 As shown, the following is an embodiment of the device quality status evaluation apparatus of the fusion data drive and QMU provided by the embodiment of the present disclosure. The device and the device quality status evaluation method of the fusion data drive and QMU of the above-mentioned embodiments belong to the same inventive concept. For details not fully described in the embodiment of the device quality status evaluation apparatus of the fusion data drive and QMU, please refer to the embodiment of the device quality status evaluation method of the fusion data drive and QMU.

[0073] An equipment quality status assessment device integrating data drive and QMU, comprising: The dataset construction module is used to collect the equipment test data and task environment data to construct a dataset, and divide the dataset into a training set and a test set.

[0074] The observation list construction module is used to calculate the importance of multidimensional features based on the data set through the extreme gradient boosting tree, screen out features that meet the equipment quality status contribution requirements according to the importance threshold, and generate the observation list of the data set.

[0075] The model building and optimization module is used to build a Gaussian process regression model based on the observation list of the training set and optimize its hyperparameters.

[0076] The state prediction module is used to predict the equipment state using the optimized Gaussian process regression model, quantify the uncertainty of the prediction results, and output the mean and variance of the equipment state prediction value.

[0077] The margin prediction module is used to calculate the predicted value of the margin based on the performance threshold PTL or PTU of the equipment and the average value of the equipment status prediction.

[0078] The equipment quality status identification module is used to calculate the confidence coefficient CF based on the variance of the equipment status prediction value and the predicted value of the margin, and use the confidence coefficient CF to identify the quality status of the equipment.

[0079] The equipment quality status assessment device provided in this embodiment, which integrates data-driven and QMU, constructs a closed-loop solution for equipment quality status assessment by integrating gradient boosting tree feature screening, Gaussian process regression prediction and QMU quantitative assessment technology: first, data dimensionality reduction is achieved based on feature importance analysis driven by machine learning, and redundant information is eliminated to improve model training efficiency and avoid overfitting risks; second, a non-parametric Gaussian process is used to model the evolution law of equipment status, and the prediction results and their uncertainty boundaries are simultaneously output to provide probabilistic support for decision-making; then, the physical performance thresholds are deeply coupled with the statistical prediction results through the QMU framework, and a margin-confidence dual-dimensional judgment standard is established, breaking through the limitations of traditional threshold comparison; finally, relying on the dynamic data update mechanism, real-time status monitoring and risk warning are achieved, driving the implementation of predictive maintenance strategies, and comprehensively enhancing the accuracy, anti-interference ability and engineering practice value of equipment reliability assessment.

[0080] Figure 5 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present invention.

[0081] The device quality status assessment method of the fusion data drive and QMU provided in the embodiment of the present application can be applied to electronic devices. Those skilled in the art will understand that the electronic device structure involved in the embodiment of the present invention does not constitute a limitation on the electronic device, and the electronic device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. In the embodiment of the present invention, the electronic device includes but is not limited to a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described and / or required herein.

[0082] The electronic device may include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, a button, a camera, a display, and a SIM card interface, etc.

[0083] A processor may include one or more processing units, such as a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU). Different processing units may be independent devices or integrated into one or more processors.

[0084] The processor can be the nerve center and command center of the electronic device. The controller can generate operation control signals based on the instruction opcode and timing signal to complete the control of instruction fetching and execution.

[0085] The processor may also include a memory for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory can store instructions or data that the processor has just used or is reusing. If the processor needs to use the instruction or data again, it can directly call it from the memory. This avoids repeated accesses, reduces processor latency, and thus improves system efficiency.

[0086] The external memory interface can be used to connect an external memory card, such as a MicroSD card, to expand the storage capacity of an electronic device. The external memory card communicates with the processor through the external memory interface, enabling data storage. For example, files such as music and videos can be stored on the external memory card.

[0087] Internal memory can be used to store computer-executable program code, which includes instructions. The processor executes the instructions stored in the internal memory to perform various functional applications and data processing of the electronic device. The internal memory can include a program storage area and a data storage area. The internal memory can include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, universal flash storage (UFS), etc.

[0088] The wireless communication function of an electronic device can be implemented through an antenna, a wireless communication module, a modem processor, and a baseband processor.

[0089] Wireless communication modules can provide wireless communication solutions for electronic devices, including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc.

[0090] Electronic devices can implement audio functions through audio modules, speakers, receivers, microphones, headphone jacks, and application processors.

[0091] Electronic devices can achieve shooting functions through ISP, camera, video codec, GPU, display and application processor.

[0092] Electronic devices can achieve display functions through GPU, display screen and application processor.

[0093] A GPU is a microprocessor for image processing that connects the display screen to the application processor. It performs mathematical and geometric calculations for graphics rendering. A processor may include one or more GPUs, which execute program instructions to generate or modify display information.

[0094] The display screen is used to display images, videos, etc. The display screen includes a display panel.

[0095] The above-mentioned electronic equipment implements the equipment quality status assessment method of the present application that integrates data-driven and QMU by integrating gradient boosting tree feature screening, Gaussian process regression modeling and quantitative margin and uncertainty assessment framework, realizes dimensionality reduction optimization and prediction uncertainty quantification of the equipment status assessment model, constructs a margin-confidence dual-factor judgment mechanism, and supports dynamic monitoring-driven predictive maintenance strategies, ultimately achieving the comprehensive beneficial effects of improving assessment accuracy, enhancing decision-making robustness and reducing life cycle costs.

[0096] The storage medium provided in the present application stores a program product that can implement an equipment quality status assessment method that integrates data-driven and QMU.

[0097] The equipment quality status assessment method integrating data-driven and QMU includes: Collect equipment test data and mission environment data to build a dataset, and divide the dataset into a training set and a test set; Based on the data set, the importance of multi-dimensional features is calculated through the extreme gradient boosting tree. According to the importance threshold, the features that meet the requirements of the contribution of equipment quality status are screened out to generate the observation list of the data set; Based on the observation list of the training set, a Gaussian process regression model is established and its hyperparameters are optimized; The optimized Gaussian process regression model is used to predict the equipment status, and the uncertainty of the prediction results is quantified to output the mean and variance of the equipment status prediction value; The predicted value of margin is calculated based on the performance threshold PTL or PTU of the equipment and the mean value of the equipment status prediction; The confidence coefficient CF is calculated based on the variance of the equipment status prediction value and the predicted value of the margin, and the confidence coefficient CF is used to identify the quality status of the equipment.

[0098] In some possible implementations, the device quality status assessment method that integrates data-driven and QMU disclosed herein can be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps described in the above "Exemplary Method" section of this specification according to various exemplary implementations of the present disclosure.

[0099] The storage medium of the present disclosure can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, a system, device or component of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0100] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for evaluating equipment quality status by integrating data-driven and QMU, characterized in that: include: Collect equipment test data and mission environment data to build a dataset, and divide the dataset into a training set and a test set; Based on the data set, the importance of multi-dimensional features is calculated through the extreme gradient boosting tree. According to the importance threshold, the features that meet the requirements of the contribution of equipment quality status are screened out to generate the observation list of the data set; Based on the observation list of the training set, a Gaussian process regression model is established and its hyperparameters are optimized; The optimized Gaussian process regression model is used to predict the equipment status, and the uncertainty of the prediction results is quantified to output the mean and variance of the equipment status prediction value; The predicted value of margin is calculated based on the performance threshold PTL or PTU of the equipment and the mean value of the equipment status prediction; The confidence coefficient CF is calculated based on the variance of the equipment status prediction value and the predicted value of the margin, and the confidence coefficient CF is used to identify the quality status of the equipment.

2. The device quality status assessment method integrating data drive and QMU according to claim 1 is characterized in that: The device test data and task environment data are collected to construct a data set, and the data set is divided into a training set and a test set, including: Collect test data and task environment data of the same model equipment in multiple time periods to form a new data table, and build a data set based on the data table ,in and represent the independent variable and the dependent variable respectively, Indicates the samples, and Represent the number of samples and feature dimensions respectively; The dataset D is divided into a training set and a test set according to a preset ratio.

3. The device quality status assessment method integrating data drive and QMU according to claim 2 is characterized in that: Based on the data set, the importance of multi-dimensional features is calculated through the extreme gradient boosting tree. According to the importance threshold, the features that meet the equipment quality status contribution requirements are screened out to generate an observation list of the data set, including: Based on the dataset D, several regression trees are formed to construct the initial XGBoost model, as follows: Where L is the objective function, is the loss function, represents the predicted value of the sample, is the regularization term, and represents the regularization coefficient, represents the feature importance, For the A regression tree model, Indicates the Related functions in the tree, represents the regression tree space, Indicates that the sample A function in the tree structure that maps to the corresponding leaf index, Represents the number of leaves; Solve the initial XGBoost model through an iterative algorithm, optimize the objective function and calculate the feature importance score; Based on the feature importance score, the features with scores below the preset threshold are filtered out according to the preset threshold; Generate an observation list of dataset D based on the filtered features.

4. The device quality status assessment method integrating data drive and QMU according to claim 3 is characterized in that: The Gaussian process regression model is established based on the observation list of the training set and its hyperparameters are optimized, including: The feature x of the observation list of the training set is used as the input of the model, and the device status y is used as the output. The established Gaussian process model is: in, is the selected mean function, and is a constant, is the selected covariance function, is the variance of the noise, is the Dirac function, that is hour ; The hyperparameters that need to be optimized are ; Hyperparameters are obtained by maximizing the log-likelihood function The optimal value of The log-likelihood function is: in, is the matrix of covariance function, is the unit matrix, is the mean vector.

5. The device quality status assessment method integrating data drive and QMU according to claim 4 is characterized in that: The optimized Gaussian process regression model is used to predict the device state, and the uncertainty of the prediction result is quantified to output the mean and variance of the device state prediction value, including: Output value of real-time test data of the equipment Output value of the training set Based on , it is brought into the optimized Gaussian regression process model, which obeys the following joint Gaussian distribution: in, and are the input vectors of the training set and the test set respectively, is the mean vector outputted by the test data, ; The device's input data is obtained through the log-likelihood function and the joint Gaussian distribution. The mean of the state prediction values under and variance ,Right now: By variance Determine confidence intervals for the predicted output values of a Gaussian regression process.

6. The device quality status assessment method integrating data-driven and QMU according to claim 5 is characterized in that: The predicted value of the margin is calculated based on the performance threshold PTL or PTU of the device and the average value of the device status prediction, including: By formula or Calculate the predicted value M of the margin; in, PTL and PTU They are the lower and upper limits of the device's factory status threshold.

7. The device quality status assessment method integrating data-driven and QMU according to claim 6 is characterized in that: The confidence coefficient CF is calculated based on the variance of the equipment state prediction value and the prediction value of the margin, and the confidence coefficient CF is used to identify the quality state of the equipment, including: The confidence coefficient CF is calculated using the following formula: Where U represents the uncertainty of the predicted value of the equipment state; If the confidence coefficient CF is greater than 1, it is determined that the equipment quality status meets the reliability requirements; if CF is less than or equal to 1, it is determined that the equipment quality status does not meet the reliability requirements.

8. An equipment quality status assessment device integrating data drive and QMU, characterized in that: The device adopts the equipment quality status assessment method integrating data drive and QMU as claimed in any one of claims 1 to 7; The device comprises: The dataset construction module is used to collect equipment test data and task environment data to construct a dataset, and divide the dataset into a training set and a test set; The observation list construction module is used to calculate the importance of multi-dimensional features based on the data set through the extreme gradient boosting tree. According to the importance threshold, it selects features that meet the equipment quality status contribution requirements and generates an observation list for the data set; Model building and optimization module, which is used to build a Gaussian process regression model based on the observation list of the training set and optimize its hyperparameters; The state prediction module is used to predict the equipment state using the optimized Gaussian process regression model, quantify the uncertainty of the prediction results, and output the mean and variance of the equipment state prediction value; The margin prediction module is used to calculate the predicted margin value based on the performance threshold PTL or PTU of the equipment and the average value of the equipment status prediction; The equipment quality status identification module is used to calculate the confidence coefficient CF based on the variance of the equipment status prediction value and the predicted value of the margin, and use the confidence coefficient CF to identify the quality status of the equipment.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the device quality status assessment method integrating data-driven and QMU are implemented as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the equipment quality status assessment method integrating data-driven and QMU are implemented as claimed in any one of claims 1 to 7.

Images