Equipment health assessment method and device based on multi-dimensional data fusion and dynamic weight

Abstract

The embodiment of the application discloses a kind of equipment health assessment method and device based on multidimensional data fusion and dynamic weight, it is related to the intelligent monitoring and health assessment technology of equipment complex system in the field of equipment health management, the present application includes: respectively from the regression trend of health parameter with time, the correlation between different health parameters and the angle of health parameter data probability distribution, describe the change with the increase of equipment storage life of health parameter, accurately reflect the difference between health parameter monitoring data and baseline data.Propose storage whole process weight dynamic updating method, scientifically calculate the weight of different health parameters when calculating health index.Thereby effectively solve the problem of health parameter monitoring data scarcity caused by inability to frequently power test during actual storage and use of equipment, improve the evaluation accuracy of equipment health state under the condition of limited monitoring data.

Description

Technical Field

[0001] This invention relates to intelligent monitoring and health assessment technology for complex equipment systems in the field of equipment health management, and particularly to an equipment health assessment method and device based on multi-dimensional data fusion and dynamic weighting. Background Technology

[0002] In modern equipment, the performance degradation and reliability of complex systems during long-term storage and use directly impact equipment efficiency and life-cycle costs. To ensure reliable system operation, extend service life, and achieve precise maintenance, Prognostics and Health Management (PHM) technology has become a core means to improve system reliability, safety, and reduce operation and maintenance costs. It's important to note that health assessment is a core concept in PHM, essentially based on in-depth analysis and decision-making using condition monitoring data. Monitoring is the foundation of "data acquisition and condition awareness" (e.g., acquiring parameters such as vibration, voltage, and resistance), while assessment is key to "condition quantification and trend judgment" (e.g., calculating health indices through multi-dimensional data fusion, determining degradation stages, and predicting remaining lifespan). As a core component of PHM technology, health status assessment identifies system degradation trends through real-time monitoring of multi-source performance parameters, providing decision-making basis for Condition-Based Maintenance (CBM) systems and supporting equipment task prioritization and proactive maintenance optimization. Facing the challenge of multi-level coupled failures in complex systems, highly reliable health assessment methods are of great significance for ensuring equipment availability and efficient resource utilization. Machine learning-based health assessment methods, through data-driven modeling and adaptive feature extraction, have become an important research direction in the field of health management (PHM). While these methods can quantitatively analyze equipment health status based on performance parameters and output the health level of individuals or batches, their practical application still faces many significant bottlenecks.

[0003] Therefore, in further research and development, health assessment methods based on multidimensional data fusion have gradually attracted attention. This method integrates multi-source data throughout the entire life cycle of equipment to construct a cross-level, cross-dimensional performance degradation analysis framework, achieving multi-feature collaborative mining and dynamic fusion. Compared with single-parameter or model-driven assessment methods, multidimensional data fusion technology can more comprehensively capture the coupled failure mechanisms of complex systems, enhance the integrity of health characterization, and improve the model's ability to characterize nonlinear degradation modes through multi-source feature correlation analysis and adaptive weight allocation, providing technical support for accurate health quantification and proactive maintenance decisions for complex systems. Although multidimensional data fusion methods demonstrate theoretical advantages, existing health assessment methods have significant shortcomings in terms of feature parameter completeness, feature fusion mechanisms, and dynamic weight optimization, which restrict the level of precision in health management of complex systems.

[0004] The main constraint lies in the fact that complex equipment systems contain multiple subsystems, including electronic, mechanical, and kinetic systems, which are functionally coupled, classifying them as "multi-domain nonlinear coupled systems." However, more optimized equipment health assessment schemes often require richer monitoring data. Yet, the more complex the equipment system, the less frequent the power-on testing during actual storage and use, leading to a scarcity of health parameter monitoring data. This results in a small data sample size, making it impossible to test each batch of equipment individually. Consequently, it's difficult to accurately reflect the differences between the health parameter monitoring data and baseline data, hindering the improvement of the accuracy of equipment health status assessment. Summary of the Invention

[0005] The embodiments of the present invention provide an equipment health assessment method and apparatus based on multi-dimensional data fusion and dynamic weighting, which can improve the accuracy of equipment health status assessment when monitoring data is limited.

[0006] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0007] Firstly, a method for assessing equipment health based on multidimensional data fusion and dynamic weights is designed, including:

[0008] S1. Establish an equipment health baseline model based on the equipment's health characteristic parameters;

[0009] S2. The health indicators output by the equipment health baseline model are dynamically adjusted based on weights, and then the comprehensive health indicators of the equipment are obtained.

[0010] S3. Determine the health status of the equipment based on the failure criteria for health indicators.

[0011] In this embodiment, S1 includes: collecting key performance monitoring parameters of the equipment and extracting health characteristic parameters that characterize the degradation trend of the equipment. The types of health characteristic parameters include: system characteristic voltage, system characteristic current, resistance, and system power of the equipment. In practical applications, it is necessary to study the equipment's functions and structure and the impact of its failure modes, including electrical systems, mechanical control systems, and kinetic control systems, to locate key performance monitoring parameters. The equipment's mission profile and existing state monitoring parameter system are analyzed, and data acquisition and storage methods are studied. Health characteristic parameters characterizing the degradation trend of the equipment are extracted from three perspectives: electronic products, mechanical control products, and kinetic control products. Specifically, these include equipment system characteristic voltage, characteristic current, resistance, and power. Based on the health characteristic parameters and the nominal data of the target equipment at the time of manufacture and the initial storage state data, an equipment health baseline model is established. The parameter types in the equipment health baseline model include: Mahalanobis distance, mean absolute error, and probability density distribution overlap percentage.

[0012] The Mahalanobis distance index is used to characterize the overall deviation between health characteristic parameters and the standard matrix of the health baseline model. The value of the Mahalanobis distance index is negatively correlated with the health status of the equipment; the Mahalanobis distance index of the i-th health characteristic parameter is expressed as: , This represents the standard matrix of the health baseline model corresponding to the i-th parameter. Let S represent the i-th sample data, and S represent the covariance matrix; the sample set... It includes p test samples, and each test sample contains m health characteristic parameters; The records are stored as a matrix of size p × m;

[0013] The mean absolute error (MAE) index measures the deviation between the actual values ​​of health characteristic parameters and the predicted values ​​of the health baseline model. The MAE index value is negatively correlated with the health status of the equipment. The MAE index for the i-th health characteristic parameter is expressed as: ,in, Represents the true value. This represents the predicted value.

[0014] The probability density distribution overlap percentage index is used to assess the similarity between health characteristic parameters and the distribution shape of standard samples. The value of the probability density distribution overlap percentage index is positively correlated with the health status of the equipment. The probability density distribution overlap percentage index for the i-th health characteristic parameter is expressed as: ,in, It is the probability density value of the health characteristic parameter in the k-th interval. It is the probability density value of the standard sample in the k-th interval. is the interval width of the probability density distribution, and n is the total number of intervals of the probability density distribution.

[0015] In this embodiment, S2 includes: adaptively updating the weights of health indicators using the CRITIC method and the entropy weight method; and generating a comprehensive health indicator based on the updated weights and health indicators. For example, the spatiotemporal characteristics of the health assessment parameter data sample can be expressed using three health indicators: mean absolute error, Mahalanobis distance, and probability density overlap percentage. The weight calculation method is studied to dynamically adjust the fusion weights of health indicators for different categories of parameters. The standard deviation and correlation coefficient of the CRITIC method reflect the variability and conflict of the data, respectively, while the information entropy of the entropy weight method reflects the degree of disorder of the data. Combining the health indicators and weight calculation results of each category of parameters, a comprehensive equipment health indicator is constructed, providing an indicator basis for equipment status assessment and ranking.

[0016] The adaptive update of the weights of health indicators using the CRITIC method and entropy weight method includes: calculating the initial weights. , , and These represent the weights of the i-th parameter calculated using the CRITIC method and the entropy weight method, respectively, based on the baseline data of the first year. The health baseline model is calculated based on the nominal data at the time of equipment delivery and the initial storage condition data; baseline data is a collective term for both nominal data and initial storage condition data. In other words, the health baseline model is calculated based on baseline data. The nominal data at the time of equipment delivery and the initial storage condition data focus on the production and application of the equipment, while baseline data is a term used in the methodology and model. After expanding the parameter data, iterative calculations are performed again, and the parameter weights after the k-th iteration are: .

[0017] For example: Suppose there are p samples to be tested and q evaluation indicators, forming the original indicator data matrix: ,in This represents the value of the j-th evaluation index for the i-th sample.

[0018] The process of calculating objective weights using the CRITIC weighting method includes: dimensionless processing: In order to eliminate the influence of different dimensions on the evaluation results, the CRITIC weighting method generally uses forward or reverse processing to perform dimensionless processing on each indicator.

[0019] If the value of the indicator used is as high as possible (positive indicator): , x max and x min This represents the upper and lower limits; the smaller the value of the indicator used, the better (a reverse indicator, which has an OR relationship with the aforementioned positive indicator; essentially, the positive or negative aspects of this indicator are related by OR, therefore this indicator still uses...).

[0020] express), Indicates the first j One evaluation indicator: .

[0021] The variability of the indicator is expressed in the form of standard deviation: , The standard deviation of the j-th indicator represents the variation and fluctuation of the values ​​within each indicator. A larger standard deviation indicates greater numerical variation in the indicator, reflecting more information and thus a stronger evaluation strength. Therefore, this indicator should be assigned more weight. Indicates the first j The average value of each evaluation indicator This represents the dimensionless processing result of the j-th evaluation index for the i-th sample.

[0022] The conflict of indicators is represented by the correlation coefficient: , This represents the correlation coefficient between evaluation indicators i and j. The correlation coefficient is used to represent the correlation between indicators. The stronger the correlation with other indicators, the less conflict there is between that indicator and the other indicators, the more identical information it reflects, and the more repetitive the evaluation content it conveys. This weakens the evaluation strength of that indicator to some extent, and the weight assigned to that indicator should be reduced.

[0023] Information content , The larger the value, the greater the role of the j-th evaluation indicator in the entire evaluation indicator system, and therefore the more weight should be assigned to it.

[0024] Objective weight: The objective weight of the j-th indicator is: .

[0025] The process of determining the weights of each indicator using the entropy weight method includes:

[0026] Data standardization: Given m samples to be evaluated and n evaluation indicators, form a standardized original data matrix. , ,in Let be the evaluation value of the i-th sample data under the j-th indicator. Calculate the weight (also known as the prior probability) of the indicator value of the i-th sample under the j-th indicator. Therefore, a weighting matrix of the data can be established. .

[0027] Calculate the information entropy of each indicator: Calculate the entropy weight of the j-th indicator. , P ij This represents the contribution (probability value) of the information entropy of the j-th indicator for the i-th sample, a constant. Information utility value .

[0028] The information utility value of a certain indicator depends on the information entropy of that indicator (the j-th indicator). The difference between 1 and 0 directly affects the weight. The greater the information utility value, the greater its importance to the evaluation, and the greater its weight.

[0029] Determine the weight of each indicator: Estimate the weight of each indicator using the entropy weight method. Essentially, this involves calculating the weight using the value coefficient of the indicator's information. The higher the value coefficient, the greater its importance to the evaluation (or the greater the weight, the greater its contribution to the evaluation result). Calculate the entropy weight of the i-th indicator. .

[0030] Determine the comprehensive weight of the indicators Assume the evaluator determines the weight of the indicators' importance based on their own objectives and requirements. Combining the entropy weight of the indicator This will give us the overall weight of index j. When all candidate items have identical values ​​for indicator j, the entropy of that indicator reaches its maximum value of 1, and its entropy weight is zero. This indicates that the indicator fails to provide useful information to the decision-maker; that is, under this indicator, all candidate items are indistinguishable to the decision-maker, and the indicator should be considered for removal. Therefore, the entropy weight itself does not represent the importance coefficient of the indicator, but rather the degree of discrimination of the evaluation objects under that indicator.

[0031] Furthermore, the acquisition of the comprehensive health index of the equipment includes: obtaining the parameter health index of each of the i-th health characteristic parameters. H i Utilizing adaptively updated dynamic weights w i Weighted fusion is performed to obtain preliminary comprehensive health indicators. H param : , parameters health indicators H i Defined as the mean of the m sub-health indicators corresponding to this parameter, i.e., fused using the weighted average method: in, H i,j The j-th sub-health index represents the i-th health characteristic parameter, m is the number of sub-health indexes, and n represents the number of health characteristic parameters; Equipment comprehensive health index H Initial model: This includes Mahalanobis distance, mean absolute error, and the percentage of probability density distribution overlap, therefore m=3.

[0032] Substituting the specific calculation formulas for the weights and sub-health indicators, we obtain the complete expansion of the comprehensive health indicator, that is, the final equipment comprehensive health indicator H equals... .

[0033] S3 includes: selecting the equipment with the best health status based on the obtained comprehensive equipment health indicators; normalizing the calculation results of the comprehensive health indicators of the equipment with the best health status to obtain the health indicator failure judgment threshold and define the failure judgment baseline; evaluating the health status of all equipment to be analyzed using the failure judgment baseline, and then ranking all equipment to be analyzed according to their health status based on their actual years of use. After calculating the comprehensive equipment health indicators, the equipment with the best health status is selected, and its comprehensive health indicator calculation results are normalized to determine the health indicator failure judgment threshold and define the failure judgment baseline. On this basis, the health status of all equipment is evaluated through the failure judgment baseline; the health assessment capability of the proposed method is evaluated through indicators such as accuracy. Finally, the equipment is ranked according to its health status based on its actual years of use, providing a reference for equipment selection.

[0034] Secondly, an equipment health assessment device based on multi-dimensional data fusion and dynamic weights was designed, including:

[0035] The model maintenance module is used to establish a health baseline model for the equipment based on its health characteristic parameters.

[0036] The analysis module is used to dynamically adjust the health indicators output by the equipment health baseline model based on weights, and then obtain the comprehensive health indicators of the equipment.

[0037] The monitoring module is used to determine the health status of the equipment based on the failure criteria of health indicators.

[0038] The model maintenance module is specifically used to collect key performance monitoring parameters of the equipment and extract health characteristic parameters that characterize the degradation trend of the equipment. The types of health characteristic parameters include: system characteristic voltage, system characteristic current, resistance, and system power of the equipment. Based on the extracted health characteristic parameters and the nominal data of the target equipment at the time of manufacture and the initial storage state data, an equipment health baseline model is established. The equipment health baseline model is used to analyze health indicators, which include: Mahalanobis distance index, mean absolute error index, and probability density distribution overlap percentage index.

[0039] In this embodiment, mean absolute error, Mahalanobis distance, and probability density overlap percentage are used to describe the changes in health parameters as the equipment's storage years increase, from the perspectives of the regression trend of health parameters over time, the correlation between different health parameters, and the probability distribution of health parameter data. This accurately reflects the difference between health parameter monitoring data and baseline data. The CRITIC method is used to reflect the variability and conflict between data in the early stages of equipment storage, while the entropy weight method is used to reflect the degree of disorder in data during the later stages of equipment storage. Combining the advantages of the two weighting calculation methods, a dynamic weight update method for the entire storage process is proposed to scientifically calculate the weight of different health parameters in the calculation of health indicators. This effectively solves the problem of scarce health parameter monitoring data due to the inability to frequently power on and test equipment during actual storage and use, as well as the bottleneck of small data samples due to the inability to test each batch of equipment individually. A comprehensive health index is designed to achieve accurate assessment of equipment health status. Thus, the accuracy of equipment health status assessment is improved even with limited monitoring data. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the method flow provided in an embodiment of the present invention;

[0042] Figure 2 A comparison chart showing the variation of mean absolute error between equipment 1 (normal) and equipment 11 (degraded and failed) with storage years in a specific example provided in the embodiments of the present invention;

[0043] Figure 3 A comparison chart of the changes in the Mahalanobis distance of equipment 1 and equipment 11 with storage years in a specific example provided in the embodiments of the present invention;

[0044] Figure 4 A comparison chart showing the change in the probability density overlap percentage of equipment 1 and equipment 11 with storage years in a specific example provided in the embodiments of the present invention.

[0045] Figure 5 A schematic diagram showing the change of the weights of three types of health parameters of equipment 11 with storage years in a specific example provided in the embodiments of the present invention;

[0046] Figure 6 A schematic diagram illustrating the variation of the mean absolute error of equipment with storage years in a specific example provided in this embodiment of the invention;

[0047] Figure 7 A schematic diagram illustrating the variation of the Mahalanobis distance of the equipment in a specific example provided in this embodiment of the invention with storage years.

[0048] Figure 8 A schematic diagram showing how the probability density overlap percentage of equipment changes with storage years in a specific example provided in the embodiments of the present invention.

[0049] Figure 9 A schematic diagram illustrating the change of comprehensive health indicators of failed equipment with storage years in a specific example provided in this embodiment of the invention;

[0050] Figure 10 This is a schematic diagram illustrating the normalized results of the changes in the comprehensive health indicators of failed equipment over storage years in a specific example provided in this embodiment of the invention. Detailed Implementation

[0051] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Embodiments of the present invention will be described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in the specification of the present invention means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or couplings. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless defined as herein.

[0052] The design concept of this embodiment lies in constructing a health assessment system covering the entire life cycle of the system through multi-source parameter collaborative screening, nonlinear feature fusion modeling, physical-data jointly driven dimensional processing, and dynamic weight optimization. This aims to provide generalized technical support for equipment mission planning and maintenance, promote the transformation of health status assessment from qualitative analysis to quantitative evaluation, and ultimately achieve the comprehensive goals of improving equipment utilization efficiency, optimizing operation and maintenance costs, and ensuring reliability throughout the entire life cycle. This solution, based on a multi-dimensional data-driven health assessment method for complex equipment systems, aims to construct a system health assessment model and achieve ranking based on health status. Its technical approach is as follows: Figure 1 As shown, according to the execution logic, it can be divided into three stages: complex system state monitoring parameter system and extraction analysis, construction of health indicators based on regression trends and probability density distributions, and state assessment and ranking based on health indicators. Specifically:

[0053] I. Parameter System and Extraction Analysis for Complex System State Monitoring

[0054] 1.1 Study the functions and structure of the equipment system and their failure mode effects to identify key performance monitoring parameters. Key performance monitoring parameters are parameters that directly reflect the health status of the three major subsystems of the equipment: electronic products, mechanical control products, and kinetic control products, determined through functional-failure analysis. It should be noted that identifying key performance parameters and key performance monitoring parameters refers to a structured method based on systems engineering analysis. The core approach is the integrated "function-structure-failure" analysis, and the specific steps include: "functional and structural analysis," "failure mode effect analysis (FMEA)," and "key parameter screening." The key performance parameters and key monitoring parameters mentioned in this embodiment can be collectively referred to as "key performance monitoring parameters." These refer to the raw physical quantities that are directly measured or monitored by sensors and instruments during the operation of equipment to assess the performance of key equipment and systems. They are the data source for health assessment. The equipment health baseline model refers to a systematic model used in health assessment to establish a "normal health status reference benchmark" based on the nominal data of the target equipment at the time of manufacture and the initial storage status data. This model uses the algorithm described in this paper to quantify and describe the characteristic patterns or health index range of the equipment under normal operating conditions, providing an objective numerical reference for judging whether the equipment has degraded, become abnormal, or failed.

[0055] 1.2 Analysis of Equipment Mission Profile and Existing Condition Monitoring Parameter System. The existing equipment condition monitoring parameter system mainly includes three aspects: electronic products, mechanical control products, and kinetic energy control products. The key health assessment parameters for the equipment's electronic products are determined to be: system power bus voltage and communication unit operating current; the key monitoring health parameters for mechanical control products are inertial measurement unit acceleration and actuator displacement; and the key monitoring health parameters for kinetic energy control products are the ignition circuit resistance of the power unit, the power circuit resistance of the control system, and the power circuit resistance of the auxiliary system.

[0056] 1.3. Collect equipment initial storage status data and construct an equipment health baseline model. Based on the key monitoring health parameters selected in 1.2, collect the nominal data at the time of equipment delivery and the initial storage status data to jointly construct the equipment health baseline model. The manufacturer will provide nominal data covering all subsystems and key components and equipment when designing the equipment. The initial storage status data refers to the test data shortly after the equipment is completed and placed in the warehouse (users generally conduct periodic inspections of the equipment). Both are readily available data.

[0057] II. Construction of Health Indicators Based on Regression Trends and Probability Density Distributions: Capturing the dynamic changes in parameter characteristics and revealing the time-varying characteristics of health status. Mahalanobis distance, mean absolute error, and the percentage of overlap in probability density distributions are used to calculate equipment degradation and collectively measure equipment health indicators.

[0058] 2.1 Calculate three types of health indicators for three categories of parameters: electronic products, kinetic control products, and mechanical control products: mean absolute error, Mahalanobis distance, and probability density overlap percentage. Randomly select several pieces of equipment that failed during the storage period, and calculate the mean absolute error, Mahalanobis distance, and probability density overlap percentage for each of the three health parameters (electronic products, kinetic control products, and mechanical control products). For each selected piece of equipment, a total of nine health indicator calculation results are obtained.

[0059] 2.2 Calculate the weights of electronic products, kinetic control products, and mechanical control products when integrating each type of health indicator, including the initial weights and adaptively dynamically updated values ​​as the storage period changes. The importance of equipment health characteristic parameters varies under different environmental and operating conditions. Therefore, dynamic weight adjustment is crucial for accurately describing changes in component operating status. Combining the CRITIC method and the entropy weight method, an adaptive update algorithm for health indicator weights is studied to address the dynamic changes in the weights of multiple health indicators during the equipment storage period, reflecting the relative importance of different indicators. It should be noted that those skilled in the art may use different terms depending on their work habits. For example, "key (performance) monitoring parameters" refer to the raw physical quantities that are directly measured or monitored by sensors, instruments, etc., during equipment operation to assess the performance of key equipment and systems, and are the data source for health assessment; "health characteristic parameters" refer to derived features that are sensitive to health status and are extracted from key monitoring parameters, and are the information carrier for health assessment; "health indicators" are comprehensive evaluation quantities obtained by fusing health characteristic parameters, and are the basis for health assessment decisions. There is a progressive relationship among the three: key monitoring parameters are essentially data, health characteristic parameters are essentially information, and health indicators are essentially knowledge processed by algorithmic models that can assist in maintenance and decision-making.

[0060] The CRITIC method relies on the data itself, fully utilizing the operational environment and state information it contains. It determines weights based on the variability and conflict of indicators; variability is measured by standard deviation, and conflict is measured by the correlation between indicators. The correlation between indicators is thus reflected. The entropy weight method is based on the concept of information entropy, which measures the degree of uncertainty or disorder within a system. The greater the amount of information, the stronger the interrelationship between parameters. The entropy weight method determines weights based on the data dispersion of each parameter. Combining these two methods can alleviate their respective limitations, thus making weight allocation more scientific. The specific steps of the multi-indicator weight adaptive update process based on the CRITIC and entropy weight methods are as follows:

[0061] Initial weight calculation: After baseline data collection, the weights of each health characteristic parameter are calculated using the CRITIC method and the entropy weight method, and the average value is taken as the initial weight. , , and represents the weights of the i-th parameter calculated using the CRITIC method and the entropy weight method, respectively, based on the baseline data from the first year.

[0062] Parameter data expansion: During the health assessment phase, the actual monitoring data of each health characteristic parameter is recorded for each test and added to the parameter dataset in sequence for weight calculation.

[0063] Dynamic weight update: After parameter data expansion, the weights are recalculated using the CRITIC method and entropy weight method, and the average value is taken as the dynamically adjusted weights. Parameter weight calculation after the first iteration: .

[0064] Iterative loop: Repeatedly expand the parameter data and dynamically update the weights, continuously incorporating new parameter data into the weight calculation dataset to ensure continuous updates of the parameter dataset and adaptive updates of the weights.

[0065] The weights of health characteristic parameters are dynamically updated using the CRITIC method and the entropy weight method. By dynamically integrating new operational data, the weights are iteratively updated, thus achieving adaptive weight updates that accurately reflect changes in equipment health status. This adaptive weight update mechanism ensures that the health assessment model consistently reflects the ability of each parameter to characterize the equipment's health status, thereby providing a more comprehensive and reliable assessment.

[0066] 2.3 Calculate the three types of health indicators for the equipment. Multiply the calculation results of the nine health indicators for each piece of equipment by their corresponding weights to obtain the calculation results of the three types of health indicators for each selected piece of equipment.

[0067] III. Research on Status Assessment and Ranking Methods Based on Health Indicators: The study calculates the results of two health indicators (Mahathano distance and mean absolute error) from the equipment regression trend model and a single health indicator (overlap percentage) from the probability density distribution model. Combining these with the average weight of the three health indicators throughout the entire lifecycle in the equipment health assessment, a comprehensive equipment health index is obtained. H The equipment is assessed and ranked based on comprehensive indicators.

[0068] 3.1 Calculate the comprehensive health index of the equipment. Based on the number of years since each piece of equipment failed, calculate the comprehensive health index of the equipment at the start of the failure period. The health status of the equipment and the comprehensive health index value are negatively correlated; therefore, the smaller the comprehensive health index, the better the health status of the equipment. Thus, the minimum comprehensive health index value of all equipment used for training at the start of the failure period is taken.

[0069] 3.2 Determine the failure threshold for health indicators. The failure threshold for health indicators is determined by comparing the health status of several pieces of equipment participating in the test. Data normalization is performed using the first data point of the equipment with the lowest overall health indicator as the starting point and the value of the equipment with the lowest overall health indicator at the failure initiation year as the ending point. Based on this, and through engineering practice, a failure margin of 5% is defined, i.e., the failure threshold for health indicators is 0.95, thus establishing a unified baseline for equipment failure judgment.

[0070] 3.3 Equipment Health Status Evaluation. The comprehensive health index of the equipment used for testing (including equipment in all health states upon reaching the specified storage period) is calculated. The data samples (including healthy and failed samples) of all equipment are evaluated based on the health index failure judgment threshold. The "health index failure judgment threshold" is a specific quantitative value (0.95 in the text), which is the critical value for judging whether the equipment's health status is "failed." Essentially, it is a failure judgment standard set by normalizing the health index data of the equipment participating in the test and combining it with engineering practice. The "equipment failure judgment baseline" refers to a unified evaluation baseline established based on the failure judgment threshold, serving as a common reference framework for all equipment participating in the evaluation. Essentially, it maps health index data from different equipment and different usage environments to a unified evaluation scale through normalization from "starting point to end point" and the setting of a "5% margin" threshold, forming a baseline of "0 (complete failure) ~ 1 (complete health)," where 0.95 is the "failure critical point" on this line. In this scheme, the threshold is determined by combining the health indicators of a few pieces of equipment with practical engineering practices. The health indicator failure judgment threshold of 0.95 is a specific critical value in the baseline, used for the failure judgment of a single piece of equipment. The equipment failure judgment baseline is a unified evaluation scale used to eliminate differences and achieve horizontal comparison of multiple pieces of equipment. The two have different focuses. Simply put, the baseline is the "ruler," and the threshold is the "scale line" on the ruler—the ruler (baseline) unifies the measurement standard, and the scale line (threshold) clarifies the failure boundary.

[0071] The health assessment capability of the equipment failure determination baseline was then evaluated using four indicators: accuracy, precision, recall, and F1 score. For example, the equipment failure determination baseline refers to the 0-1 interval obtained by "normalizing the data using the first data point of the equipment with the lowest overall health index as the starting point and the value of the equipment with the lowest overall health index at the start of the failure period as the endpoint." To facilitate equipment health assessment, the formulas for accuracy, precision, recall, and F1 score are not included. Referring to the content in the header row of Table 6, these four indicators are calculated based on "the actual number of samples judged as normal," "the actual number of samples judged as failed," "the actual number of samples judged as normal," and "the actual number of samples judged as failed." The lower the values ​​of "the actual number of samples judged as failed" and "the actual number of samples judged as normal," the better; the smaller these values ​​are, the higher the values ​​of the four indicators, indicating a more accurate health assessment and a result that better reflects the actual situation.

[0072] Assume that TP (True Positive) represents a true positive instance, i.e., the number of samples correctly predicted as positive; TN (True Negative) represents a true negative instance, i.e., the number of samples correctly predicted as negative; FP (False Positive) represents a false positive instance, i.e., the number of negative samples incorrectly predicted as positive; and FN (False Negative) represents a false negative instance, i.e., the number of positive samples incorrectly predicted as negative. In the health assessment process of this paper, positive refers to equipment in normal condition, and negative refers to equipment failure. True positive instance TP is equipment that is actually in normal condition and is judged as normal by the health assessment; true negative instance TN is equipment that has actually failed and is judged as failed by the health assessment; false positive instance FP is equipment that has actually failed but is judged as normal by the health assessment; and false negative instance FN is equipment that is actually in normal condition but is judged as failed by the health assessment.

[0073] Accuracy refers to the proportion of correctly classified samples out of the total number of samples. It reflects the classifier's ability to distinguish between different samples. The formula for accuracy is: .

[0074] Precision refers to the proportion of truly positive samples out of a set of samples classified as positive. It measures the accuracy of a classifier's predictions for positive samples, and is therefore also called "precision ratio". The formula for precision is: .

[0075] Recall, also known as "complete detection," refers to the proportion of samples correctly predicted as positive out of all true positive samples. It reflects the classifier's ability to detect all positive samples, hence the name "complete detection." The formula for recall is: .

[0076] The F1 score is the harmonic mean of precision and recall. It comprehensively considers both precision and recall, providing a more holistic evaluation of the classifier's performance. The formula for the F1 score is: .

[0077] The health assessment capability is evaluated using four metrics: accuracy, precision, recall, and F1 score.

[0078] 3.4 Equipment Health Status Ranking. Based on actual equipment usage requirements, four years are defined for ranking health status: the 5th, 10th, 15th, and 20th year of storage. At each of these four years, all equipment is ranked according to comprehensive health index values, and the equipment in the best condition is selected to provide a reference for equipment use.

[0079] In this embodiment, addressing the scarcity of health parameter monitoring data due to the inability to frequently power on equipment for testing during actual storage and use, and the limitation of small data samples due to the inability to test each batch of equipment individually, a comprehensive health index is designed to accurately assess the health status of the equipment. Using mean absolute error, Mahalanobis distance, and probability density overlap percentage, the changes in health parameters with the increase of equipment storage years are described from the perspectives of regression trends of health parameters over time, correlations between different health parameters, and probability distribution of health parameter data, accurately reflecting the differences between health parameter monitoring data and baseline data. The CRITIC method is used to reflect the variability and conflict between data in the early stages of equipment storage, while the entropy weight method reflects the degree of disorder in data during the middle and later stages of equipment storage. Combining the advantages of the two weighting calculation methods, a dynamic weight update method for the entire storage process is proposed to scientifically calculate the weight of different health parameters in the calculation of the health index.

[0080] The following are specific experimental cases for analysis:

[0081] To verify the effectiveness of the proposed health assessment method for complex equipment systems based on multidimensional data fusion and dynamic weights, a study was conducted using simulated monitoring data from a batch of equipment. The health status of equipment, including normal state, degradation failure, sudden failure, and degradation-sudden failure, was evaluated, and a health status ranking was obtained to provide a reliable basis for equipment maintenance decisions. Furthermore, performance comparison experiments were conducted, including ablation experiments and controlled variable experiments, to verify the superiority and robustness of the method. The computer hardware configuration used in the experiments was as follows: AMD Ryzen 7 5800X CPU, 32GB of RAM, and 1TB of hard drive space. The proposed health assessment method was implemented using the Python programming language (version 3.9.19) based on Visual Studio Code.

[0082] Application Scenarios: By integrating multi-dimensional health assessment parameter data, longitudinal analysis of equipment development, production, testing, and usage throughout its entire lifecycle can be conducted. This allows for the capture of early performance degradation characteristics and slowly accumulating degradation patterns that traditional single-parameter threshold health assessment methods cannot identify. It can effectively identify situations where equipment passes inspection but has actually failed. For equipment that passes inspection, health indicators are quantified into precise values, and health status levels (e.g., healthy, sub-healthy, moderate, early warning) are established. Based on the quantification results, health status is evaluated and ranked to determine the equipment with the best current health status, providing a scientific basis for equipment selection.

[0083] Parameter Selection and Data Update: This method is universal and does not rely on specific health assessment parameters. Different monitoring parameters can be selected to construct a health evaluation model based on the user's actual needs. A health baseline model is constructed using nominal health parameter data from the equipment's factory delivery and early detection data. Early health assessments of the equipment are then conducted using comprehensive health indicators. Table 1 uses the nominal data obtained at the equipment's factory delivery as the first detection, and Table 2 uses sample number 1 as the equipment's health baseline sample. The current health status of the equipment is quantified by continuously constructing health assessment samples and calculating comprehensive health indicators. As the equipment's usage time and storage period increase, the health evaluation model can be dynamically updated, thereby constructing more sensitive and generalizable health evaluation indicators and achieving more accurate health status assessments.

[0084] Table 1 Equipment Health Parameter Monitoring Data

[0085]

[0086] Table 2. Equipment Health Assessment Samples and Comprehensive Health Indicators

[0087]

[0088] Data Description: This experimental case study analyzes a batch of 28 pieces of equipment with a normal storage period of 20 years. To simulate actual conditions, this batch of equipment is divided into 10 pieces of normal equipment (numbered 1-10), 9 pieces of equipment experiencing degradation failure (numbered 11-19), 7 pieces of equipment experiencing sudden failure (numbered 20-26), and 2 pieces of equipment experiencing both degradation and sudden failure (numbered 27-28). The health parameters used in this case study are seven key parameters: system power bus voltage, communication unit operating current, power unit ignition circuit resistance, control system power circuit resistance, auxiliary system power circuit resistance, inertial measurement unit acceleration, and actuator displacement. Furthermore, based on the actual equipment testing cycle, monitoring data for these seven key health parameters are obtained every six months. Therefore, all the data simulated in this paper are shown in the appendix, and the health parameter monitoring data for one piece of normal equipment is shown in Table 3.

[0089] Table 3 Health parameter monitoring data for a normal piece of equipment

[0090]

[0091] The nominal data of the equipment products is taken as the standard data, and this standard data is defined as the basis for establishing the health assessment model. When the drift of the health parameter monitoring data of a component exceeds the specified value, the equipment is judged to have degraded and failed. When the health parameter monitoring data of a component suddenly drops to 0 or exceeds the standard data by 1 time, the equipment is judged to have experienced sudden failure. Equipment that experiences both degraded and sudden failure is judged to have experienced both degraded and sudden failure. Therefore, based on this data, the thresholds for determining degraded and sudden failure of this batch of equipment are shown in Table 4.

[0092] Table 4 Thresholds for Degradation Failure and Sudden Failure of Key Parameters of Equipment Systems

[0093]

[0094] Health parameter system and health baseline model construction: The nominal data of 7 health parameters of equipment products at the time of manufacture are defined as the standard data at year 0, and the data format is 7×1. Combined with the data collected by the first test after the equipment enters the storage period, the health baseline model of the equipment product is constructed. The data format of the health baseline model is 7×2. Taking the health parameter monitoring data of this batch of equipment as an example, the sample data of its health baseline model is shown in Table 5.

[0095] Table 5 Sample data of the health baseline model for this batch of equipment

[0096]

[0097] Data samples were constructed sequentially according to the data format of the health baseline model (sample 1 consists of data from year 0 and year 0.5, sample 2 consists of data from year 0.5 and year 1, and so on). Therefore, each piece of equipment had 40 samples, resulting in a total of 1120 data samples for 28 pieces of equipment. Of these, 794 were health data samples and 326 were failure data samples.

[0098] Based on the practical need for equipment health assessment under small sample conditions in engineering, all data samples were divided into training data and test data according to the small sample ratio. 120 data samples from three failed pieces of equipment (equipment 11, 12, and 13 in the appendix) were selected as training samples, and 1000 samples from the remaining 25 pieces of equipment were selected as test samples. Of the 120 training samples, 63 were healthy data samples and 57 were failed data samples.

[0099] Health Indicator Construction: Construct health indicators to evaluate the health status of the equipment system according to the following steps:

[0100] (1) Calculate three types of health indicators for the three types of parameters: electronic product parameters, kinetic control product parameters and mechanical control product parameters: mean absolute error, Mahalanobis distance and probability density overlap percentage.

[0101] The mean absolute error, Mahalanobis distance, and probability density overlap percentage of the electronic, kinetic control, and mechanical control parameters of the three selected failed equipment pieces were calculated respectively. The changes in the mean absolute error, Mahalanobis distance, and probability density overlap percentage of the degraded and failed equipment 11 over storage years were compared with those of normal equipment 1 to demonstrate the effectiveness of constructing three types of health indicators to measure various health parameters of the equipment. The comparison results are as follows: Figure 2 , Figure 3 and Figure 4 As shown.

[0102] (2) Calculate the weights of electronic product parameters, kinetic control product parameters and mechanical control product parameters when each health indicator is integrated, including the initial weights and the adaptive dynamic update values ​​as the storage years change.

[0103] The weighting of the three categories of health parameters—electronic product parameters, kinetic control product parameters, and mechanical control product parameters—of the selected failed equipment 11 when integrating health indicators varies with storage years as follows: Figure 5 As shown. From Figure 5 As can be seen, the dynamic changes in weights differ when using the CRITIC method and the entropy weight method for weight calculation. Specifically: 1) Weight changes and updates in the CRITIC method mainly occur in the early stages of storage, and the weights gradually stabilize as storage time increases. This is because the CRITIC method primarily considers the comparative strength and conflict between indicators when calculating weights. In the early stages, with limited sample data, the comparative strength and conflict between indicators are more pronounced, leading to significant weight changes in the early stages. As storage years increase and data becomes more abundant, the CRITIC method has already determined the general trend of weights based on the characteristics of the initial data, so the changes in the later stages are relatively small. 2) The entropy weight method assigns similar weights to each indicator in the early stages of calculation, and weight changes and updates mainly occur in the middle and later stages. This is because the entropy weight method determines weights based on information entropy. At the beginning, due to limited data, the calculated information entropy results are relatively similar, thus assigning similar weights to each health parameter. As storage years increase and the amount of data gradually increases, the differences in information entropy between different indicators gradually become apparent, leading to more significant weight changes in the middle and later stages. 3) The comprehensive weighting method can combine the advantages of both methods to dynamically adjust the parameter weights at different time periods of equipment storage, thereby more comprehensively and objectively reflecting the objective laws of changes in various health assessment parameters of the equipment throughout the entire storage cycle.

[0104] (3) Calculate the three types of health indicators of the equipment.

[0105] Combining the calculation results of the health indicators of the three types of health parameters in (1) and the weights of the three types of health parameters in the calculation of health indicators in (2), the three types of health indicators of the equipment products are calculated. The calculation results of the three types of health indicators of the three selected failed equipment are as follows: Figure 6 , Figure 7 and Figure 8 As shown.

[0106] Depend on Figure 6 It can be seen that the mean absolute error of equipment 11, 12, and 13 all show an upward trend as the storage period increases. When the storage period is 0 years, all health parameters are the nominal data at the time of manufacture, so the mean absolute error is the smallest. As the storage period increases, the health parameters gradually degrade, so the gap between the data sample and the baseline sample gradually increases.

[0107] Depend on Figure 7 It can be seen that the Mahalanobis distances of equipment 11, 12, and 13 also show an increasing trend with the increase of storage years. The Mahalanobis distance is the smallest when the storage years are 0. As time goes by, the health parameters gradually deteriorate, and the similarity between the data samples and the baseline samples decreases. Furthermore, unlike the calculation results of the mean absolute error index—the error of equipment 13 is greater than that of equipment 12, which is greater than that of equipment 11—the calculation results of the Mahalanobis distances of the three pieces of equipment show that the similarity of the data samples of equipment 11 is less than that of equipment 12, which is less than that of equipment 13.

[0108] Depend on Figure 8 It can be seen that as the storage years increase, the probability density overlap percentage of equipment 11, 12, and 13 shows a downward trend (the indicators in the figure have been normalized, so they show an upward trend). In the early stage of storage, the probability density overlap percentage is relatively high. As time goes by, the degree of overlap of the relevant probability distributions gradually decreases, that is, the difference in distribution between the data sample and the baseline sample gradually increases.

[0109] The equipment health status can then be evaluated and ranked according to the following steps:

[0110] 1. Calculate the comprehensive health index of the equipment: Calculate the comprehensive health index of the three selected failed pieces of equipment, and observe how it changes with the storage years. Figure 9 As shown. By Figure 9It can be seen that for Equipment 11, the power supply loop resistance of the control system and the operating current of the communication unit began to degrade and fail in the 11th and 13.5th years, respectively, with an overall health index of 0.294 at the time of failure and 0.368 after 20 years of storage. For Equipment 12, the power supply loop resistance of the auxiliary system began to degrade and fail in the 11.5th year, with an overall health index of 0.301 at the time of failure. The amount of parameters that degraded in Equipment 12 was different from that in Equipment 11, but the overall parameter drift was similar to that of Equipment 11. Therefore, the overall health index after 20 years of storage was similar to that of Equipment 11, at 0.369. For Equipment 13, the system power bus voltage and the power supply loop resistance of the control system began to degrade and fail in the 10.5th and 11th years, respectively, with an overall health index of 0.298 at the time of failure. Compared with Equipment 11 and 12, the overall parameter drift of Equipment 13 was larger. Therefore, the overall health index after 20 years of storage was the largest, at 0.382. The above analysis demonstrates that comprehensive health indicators can accurately track and objectively reflect changes in equipment health assessment parameters.

[0111] 2. Determining the Failure Threshold for Health Indicators: The failure threshold for health indicators was determined by comparing the health status of three pieces of equipment. It is known that the health status of equipment and the comprehensive health indicator value are negatively correlated; therefore, among the three pieces of equipment, equipment 11 has the best health status. Using the comprehensive health indicator of equipment 11 (0) as the starting point and the comprehensive health indicator of equipment 11 at the start of its failure (year 11) (0.294) as the endpoint, data normalization was performed. The normalized results of the changes in the comprehensive health indicator of the three failed pieces of equipment over the storage years are shown in the figure. Based on this, a 5% margin was taken, resulting in a health indicator failure threshold of 0.95.

[0112] 3. Health Status Evaluation: Calculate the comprehensive health index of the 25 pieces of equipment used for testing (including 10 pieces of equipment 1-10 that did not fail, 6 pieces of equipment 14-19 that degraded and failed, 7 pieces of equipment 20-26 that experienced sudden failure, and 2 pieces of equipment 27 and 28 that experienced both degradation and sudden failure). Evaluate the 1,000 samples contained in the 25 pieces of equipment according to the failure judgment threshold of the health index.

[0113] Table 6 shows the comprehensive health index calculation results and sample judgment results for all 25 tested equipment at the 20-year storage period. The proposed method's ability to assess the health status of equipment products was evaluated using four evaluation indicators: accuracy, precision, recall, and F1 score. Table 6 uses different colors to distinguish equipment states, from top to bottom: normal, degraded failure, sudden failure, and degraded-sudden failure. Analysis of Table 6 shows that out of 1000 samples from the 25 failed equipment, 686 samples were actually normal and were judged normal by the threshold; 45 samples were actually normal but were judged as failed by the threshold; 10 samples were actually failed but were judged normal by the threshold; and 278 samples were actually failed and were judged as failed by the threshold. The calculated accuracy rate for the health index judgment threshold was 94.5%, the precision rate was 98.6%, the recall rate was 93.8%, and the F1 score was 96.1%.

[0114] Table 6. Calculation results of comprehensive health indicators and sample determination for 25 test devices.

[0115]

[0116] Health Status Ranking: Based on the calculation results of the comprehensive health index of the equipment, the health status of all 28 pieces of equipment was ranked to provide a reasonable basis for scientifically carrying out equipment selection. The order of the health status of the 28 pieces of equipment is shown in Table 7.

[0117] Table 7. Ranking of Health Status of 28 Pieces of Equipment

[0118]

[0119] Of the 28 pieces of equipment participating in the health assessment in Table 7, in the 5th year of storage, the top 5 pieces of equipment ranked from lowest to highest comprehensive health index are: Equipment 24, Equipment 3, Equipment 10, Equipment 7, and Equipment 6. Among them, Equipment 24 has the best health status, with a comprehensive health index of 0.3583. In the 10th year of storage, the top 5 pieces of equipment ranked from lowest to highest comprehensive health index are: Equipment 24, Equipment 10, Equipment 3, Equipment 7, and Equipment 6. Among them, Equipment 24 has the best health status, with a comprehensive health index of 0.5304. In the 15th year of storage, the top 5 pieces of equipment ranked from lowest to highest comprehensive health index are: Equipment 3, Equipment 10, Equipment 4, Equipment 1, and Equipment 5. Among them, Equipment 3 has the best health status, with a comprehensive health index of 0.8030. When the storage period reaches the prescribed 20 years, Equipment 8 has the lowest comprehensive health index, therefore it is determined to have the best health status. Furthermore, at this point, the equipment ranked 1-10 corresponds to 10 pieces of equipment that have not failed and are in normal condition, while the equipment ranked 11-28 corresponds to equipment that has failed. This indicates that the proposed health assessment method can correctly distinguish equipment in different states, and the health index calculation results have reference value for the actual use of equipment.

[0120] In summary, a case study was conducted on 28 pieces of equipment in different health states. The proposed health assessment method determined the failure threshold by calculating the comprehensive health index of three pieces of equipment that failed, and then assessed the health status of the remaining 25 pieces of equipment. The accuracy, precision, recall, and F1 score reached 94.5%, 98.6%, 93.8%, and 96.1%, respectively, verifying the accuracy of its health assessment capability under small sample conditions.

[0121] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The above descriptions are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for equipment health assessment based on multidimensional data fusion and dynamic weights, characterized in that, include: S1. Establish an equipment health baseline model based on the equipment's health characteristic parameters; S2. The health indicators output by the equipment health baseline model are dynamically adjusted based on weights, and then the comprehensive health indicators of the equipment are obtained; S2 includes: adaptively updating the weights of the health indicators using the CRITIC method and the entropy weight method; and generating comprehensive health indicators based on the updated weights and health indicators. The adaptive update of the weights of health indicators using the CRITIC method and entropy weight method includes: calculating the initial weights of the health indicators. , , and These represent the weights of the i-th parameter when calculating health indicators using the CRITIC method and the entropy weight method, respectively, based on the baseline data of the first year. After expanding the parameter data, the iterative calculation is performed again. The parameter weights after the k-th iteration are: , and These are the weights calculated for the i-th health feature parameter using the CRITIC method and the entropy weight method, respectively, during the k-th iteration. The process of generating a comprehensive health index includes: combining the health indexes of each of the i-th health characteristic parameters. Utilizing adaptively updated dynamic weights Weighted fusion is performed to obtain preliminary comprehensive health indicators. : , ,in, Let m represent the j-th sub-health index of the i-th health characteristic parameter, m be the number of sub-health indexes, and n represent the number of health characteristic parameters; the final obtained comprehensive equipment health index H is equal to... , H i,MD The Mahalanobis distance index represents the i-th health characteristic parameter. H i,MAE This represents the mean absolute error index of the i-th health characteristic parameter. H i,OP The probability density distribution overlap percentage of the i-th health feature parameter is represented by the following: Iterative loop: Repeated expansion of parameter data and dynamic weight update, incorporating new parameter running data into the weight calculation dataset, so as to continuously update the parameter dataset and adaptively update the weights. S3. Determine the health status of the equipment based on the failure judgment conditions of the comprehensive health index of the equipment; wherein, S3 includes: selecting the equipment with the best health status based on the obtained comprehensive health index of the equipment; after normalizing the comprehensive health index of the equipment with the best health status, determining the failure judgment threshold of the comprehensive health index of the equipment and defining the failure judgment baseline; using the failure judgment baseline to evaluate the health status of the equipment to be analyzed, and then sorting the equipment to be analyzed according to its health status based on the actual years of use of the equipment. The methods for determining the failure threshold of the comprehensive health index of equipment and defining the failure baseline include: determining the failure threshold of the health index by comparing the health status of several pieces of equipment participating in the test, wherein the first data point of the equipment with the smallest comprehensive health index is taken as the starting point, and the value of the equipment with the smallest comprehensive health index at the failure start year is taken as the endpoint for data normalization, and the failure margin of the unified equipment failure baseline is defined as 5%, and the failure threshold of the health index is 0.

95. The health assessment capability of the equipment failure determination baseline was then evaluated using four metrics: accuracy, precision, recall, and F1 score. Accuracy refers to the proportion of correctly classified samples out of the total number of samples, reflecting the classifier's ability to discriminate across the entire sample set. Precision refers to the proportion of truly positive samples among those classified as positive, measuring the accuracy of the classifier's predictions for positive samples. Recall refers to the proportion of correctly predicted positive samples out of all truly positive samples, reflecting the classifier's ability to recall positive samples. The F1 score is the harmonic mean of precision and recall.

2. The method according to claim 1, characterized in that, S1 includes: Key performance monitoring parameters of the equipment are collected, and health characteristic parameters that characterize the degradation trend of the equipment are extracted. The types of health characteristic parameters include: system characteristic voltage, system characteristic current, resistance and system power of the equipment. Based on the extracted health characteristic parameters and the nominal data of the target equipment at the time of manufacture and the initial storage status data, an equipment health baseline model is established. The equipment health baseline model is used to analyze health indicators, which include: Mahalanobis distance index, mean absolute error index, and probability density distribution overlap percentage index.

3. The method according to claim 2, characterized in that, Mahalanobis distance index is used to characterize the overall deviation between health characteristic parameters and the standard matrix of the health baseline model. The value of Mahalanobis distance index is negatively correlated with the health status of the equipment. The mean absolute error index is used to measure the deviation between the actual values ​​of health characteristic parameters and the predicted values ​​of the health baseline model. The value of the mean absolute error index is negatively correlated with the health status of the equipment. The probability density distribution overlap percentage index is used to assess the similarity between health characteristic parameters and the distribution shape of standard samples. The value of the probability density distribution overlap percentage index is positively correlated with the health status of the equipment.

4. The method according to claim 2 or 3, characterized in that, The Mahalanobis distance index for the i-th health characteristic parameter is expressed as: , , represents the standard matrix of the health baseline model corresponding to the i-th parameter. Let S represent the i-th sample data, and S represent the covariance matrix; the mean absolute error index of the i-th health characteristic parameter is expressed as... ,in, Represents the true value. Indicates the predicted value; The percentage overlap of the probability density distribution of the i-th health characteristic parameter is expressed as: ,in, It is the probability density value of the health characteristic parameter in the k-th interval. It is the probability density value of the standard sample in the k-th interval. is the interval width of the probability density distribution, and n is the total number of intervals of the probability density distribution.

5. An equipment health assessment device based on multi-dimensional data fusion and dynamic weighting, characterized in that, include: The model maintenance module is used to establish a health baseline model for the equipment based on its health characteristic parameters. The analysis module is used to dynamically adjust the health indicators output by the equipment health baseline model based on weights, and then obtain the comprehensive health indicators of the equipment; wherein, the weights of the health indicators are adaptively updated using the CRITIC method and the entropy weight method; and the comprehensive health indicators are generated based on the updated weights and health indicators. The adaptive update of the weights of health indicators using the CRITIC method and entropy weight method includes: calculating the initial weights of the health indicators. , , and These represent the weights of the i-th parameter when calculating health indicators using the CRITIC method and the entropy weight method, respectively, based on the baseline data of the first year. After expanding the parameter data, the iterative calculation is performed again. The parameter weights after the k-th iteration are: , and These are the weights calculated for the i-th health feature parameter using the CRITIC method and the entropy weight method, respectively, during the k-th iteration. The process of generating a comprehensive health index includes: combining the health indexes of each of the i-th health characteristic parameters. Utilizing adaptively updated dynamic weights Weighted fusion is performed to obtain preliminary comprehensive health indicators. : , ,in, Let m represent the j-th sub-health index of the i-th health characteristic parameter, m be the number of sub-health indexes, and n represent the number of health characteristic parameters; the final obtained comprehensive equipment health index H is equal to... , H i,MD The Mahalanobis distance index represents the i-th health characteristic parameter. H i,MAE This represents the mean absolute error index of the i-th health characteristic parameter. H i,OP The probability density distribution overlap percentage of the i-th health feature parameter is represented by the following: Iterative loop: Repeated expansion of parameter data and dynamic weight update, incorporating new parameter running data into the weight calculation dataset, so as to continuously update the parameter dataset and adaptively update the weights. The monitoring module is used to determine the health status of equipment based on the failure judgment conditions of the comprehensive health index of equipment. Specifically, based on the obtained comprehensive health index of equipment, the equipment with the best health status is selected. After normalizing the comprehensive health index of the equipment with the best health status, the failure judgment threshold of the comprehensive health index of equipment is determined and the failure judgment baseline is defined. The health status of the equipment to be analyzed is evaluated using the failure judgment baseline, and then the equipment to be analyzed is sorted according to its health status based on the actual years of use of the equipment. The methods for determining the failure threshold of the comprehensive health index of equipment and defining the failure baseline include: determining the failure threshold of the health index by comparing the health status of several pieces of equipment participating in the test, wherein the first data point of the equipment with the smallest comprehensive health index is taken as the starting point, and the value of the equipment with the smallest comprehensive health index at the failure start year is taken as the endpoint for data normalization, and the failure margin of the unified equipment failure baseline is defined as 5%, and the failure threshold of the health index is 0.

95. The health assessment capability of the equipment failure determination baseline was then evaluated using four metrics: accuracy, precision, recall, and F1 score. Accuracy refers to the proportion of correctly classified samples out of the total number of samples, reflecting the classifier's ability to discriminate across the entire sample set. Precision refers to the proportion of truly positive samples among those classified as positive, measuring the accuracy of the classifier's predictions for positive samples. Recall refers to the proportion of correctly predicted positive samples out of all truly positive samples, reflecting the classifier's ability to recall positive samples. The F1 score is the harmonic mean of precision and recall.

6. The apparatus according to claim 5, characterized in that, The model maintenance module is specifically used to collect key performance monitoring parameters of the equipment and extract health characteristic parameters that characterize the degradation trend of the equipment. The types of health characteristic parameters include: system characteristic voltage, system characteristic current, resistance, and system power of the equipment. Based on the extracted health characteristic parameters and the nominal data of the target equipment at the time of manufacture and the initial storage state data, an equipment health baseline model is established. The equipment health baseline model is used to analyze health indicators, which include: Mahalanobis distance index, mean absolute error index, and probability density distribution overlap percentage index.

Images