A multi-sensor fusion mechanical metrology device management method and system

Abstract

This invention relates to the field of data fusion technology, specifically to a multi-sensor fusion-based method and system for managing mechanical metrology equipment. The method includes the following steps: acquiring temperature monitoring and indication sequences and generating temperature change rate and indication error sequences; filtering extreme values ​​based on a sliding time window to generate environmental stress and metrological drift probability variable sequences; constructing joint distribution data pairs; inputting an asymmetric Archimedes connection function model to calculate the joint probability density; estimating shape parameters using maximum likelihood to solve for the optimal tail dependency parameters; calculating the upper tail correlation coefficient; and generating a structural instability warning command when the coefficient exceeds a stability safety threshold. In this invention, by introducing a model to analyze the joint distribution characteristics, the method solves the parameter quantification of the correlation between environmental stress and metrological drift extreme effects, accurately identifying instability characteristics caused by thermal stress shock, solving the problem of difficulty in quantifying nonlinear coupling relationships, and achieving advanced prediction of failure risks.

Description

Technical Field

[0001] This invention relates to the field of data fusion technology, and in particular to a method and system for managing mechanical measurement equipment using multi-sensor fusion. Background Technology

[0002] Data fusion technology refers to a data processing technology that uses computer technology to automatically analyze and synthesize several observational information obtained in a time sequence under certain criteria to complete the required decision-making and evaluation tasks. Its aim is to improve the accuracy and reliability of the system by combining data from multiple sensors. Traditionally, the management method for mechanical measurement equipment involves operators using a standard force measuring machine to load the equipment, manually observing the force value data read from a digital display panel connected to a single sensor, and then manually copying the equipment's serial number, calibration date, and the force value error data onto a paper equipment record card, or manually entering them into an Excel spreadsheet stored on a local standalone computer for archiving.

[0003] Traditional mechanical measurement equipment management relies on manual operation and single sensor readings. Simply observing the instrument panel to obtain instantaneous force values ​​cannot capture the continuous impact of environmental factors on measurement performance. Manual copying or single-machine data entry is inefficient and susceptible to human misreading and misrecording. This not only makes it difficult to ensure the real-time and integrity of the data, but also causes the locally stored data to form information silos. It is impossible to effectively assess the long-term stability of the equipment under complex operating conditions, making it difficult for managers to detect the measurement drift trend induced by drastic fluctuations in ambient temperature. Ultimately, the potential structural instability risk of the equipment is ignored, leading to measurement failure. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a multi-sensor fusion-based method and system for managing mechanical measurement equipment.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a multi-sensor fusion method for managing mechanical measurement equipment, comprising the following steps:

[0006] S1: Obtain the temperature monitoring sequence and the indication sequence, generate a temperature change rate sequence by differential analysis of the temperature monitoring sequence, calculate the difference between the indication sequence and the standard value to generate an indication error sequence, and generate an environmental stress extreme value sequence and a measurement drift extreme value sequence by selecting the maximum absolute value from the temperature change rate sequence and the indication error sequence based on a sliding time window.

[0007] S2: Arrange the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine their positions. Calculate the positions based on the total sequence length normalization to generate the environmental stress probability variable sequence and the measurement drift probability variable sequence.

[0008] S3: Pair the environmental stress probability variable sequence and the measurement drift probability variable sequence in the same time window to construct a joint distribution data pair, input it into the asymmetric Archimedes connection function model to calculate the joint probability density, and estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation to maximize the likelihood function in order to solve the optimal tail dependency parameters;

[0009] S4: Calculate the upper tail correlation coefficient based on the optimal tail dependency parameter, and generate a structural instability warning command when the upper tail correlation coefficient is greater than the stability safety threshold.

[0010] As a further aspect of the present invention, step S1 specifically comprises:

[0011] Temperature monitoring data and mechanical indication data are collected by sensor interfaces deployed at key nodes of the mechanical measurement equipment. The temperature monitoring data is processed by a first-order difference algorithm to eliminate baseline trends and retain high-frequency fluctuation characteristics, generating a temperature change rate sequence. At the same time, the numerical deviation between the mechanical indication data and the preset standard measurement reference value is calculated, and the absolute value is taken to generate an indication error sequence.

[0012] The window length and sliding step size of the sliding time window are set according to the preset sampling frequency. The sliding time window is moved synchronously on the temperature change rate sequence and the indication error sequence according to the sliding step size. The data subset in each time window is traversed and the data point with the largest absolute value is extracted. The environmental stress extreme value sequence and the measurement drift extreme value sequence are constructed in sequence.

[0013] As a further aspect of the present invention, step S2 specifically comprises:

[0014] The environmental stress extreme value sequence and the metrological drift extreme value sequence are sorted in ascending order. While maintaining the correspondence between data values, the position index of each data point in its respective sequence is calculated to determine the position value, thus obtaining the environmental stress position sequence and the metrological drift position sequence.

[0015] Obtain the total length of the environmental stress extreme value sequence, and use the empirical distribution function formula to divide each position value in the environmental stress position sequence and the measurement drift position sequence by the sum of the total sequence length and the correction factor, so as to map the discrete position features into probability values ​​with values ​​between zero and one, thereby generating the environmental stress probability variable sequence and the measurement drift probability variable sequence.

[0016] As a further aspect of the present invention, step S3 specifically comprises:

[0017] Based on the timestamp alignment method, the elements in the environmental stress probability variable sequence corresponding to the same time window are paired with the elements in the measurement drift probability variable sequence to construct joint distribution data pairs;

[0018] The joint distribution data pairs are substituted into the probability density function of the pre-constructed asymmetric Archimedes connection function model to construct a log-likelihood function containing the shape parameters to be estimated. The value of the shape parameters is iteratively adjusted using a numerical optimization algorithm until the function value of the log-likelihood function reaches its maximum. The parameter value at this point is taken as the optimal tail dependency parameter.

[0019] As a further aspect of the present invention, step S4 specifically comprises:

[0020] Obtain the optimal tail dependency parameters obtained from the solution, and calculate the limiting conditional probability of the asymmetric Archimedes connection function model in the upper right tail according to the preset extreme value dependency theory formula, and generate the upper tail correlation coefficient.

[0021] The correlation coefficient of the upper tail is compared with the preset stability safety threshold. If the correlation coefficient of the upper tail exceeds the stability safety threshold, the mechanical measuring equipment is determined to be in a high-risk state of thermal structural instability, and the alarm logic is immediately triggered and a structural instability warning command is generated.

[0022] As a further aspect of the present invention, the screening process for the temperature change rate sequence and the indication error sequence specifically includes:

[0023] The window length of the sliding time window is set to cover the entire thermal response cycle of the mechanical measuring equipment, and the sliding step size is set to be less than half of the window length to ensure the continuity and overlap of data sampling.

[0024] Within the coverage area of ​​each sliding window, the absolute values ​​of all data points in the temperature change rate sequence are taken and compared. The point with the largest value is retained as the representative value of environmental thermal stress for that window period. At the same time, the same absolute value maximization filtering operation is performed on the indication error sequence. The two extreme values ​​selected are stored in the corresponding sequence containers in chronological order, thus completing the construction of the environmental stress extreme value sequence and the measurement drift extreme value sequence.

[0025] As a further aspect of the present invention, the asymmetric Archimedes connection function model specifically adopts the Gumbel-Hougaard Copula function structure, and the mathematical expression of its joint distribution function is as follows:

[0026] ;

[0027] in, Represents the joint distribution function, This represents the numerical value in the sequence of environmental stress probability variables. This represents the numerical value in the sequence of the measurement drift probability variables. The shape parameters represent the shape parameters of the asymmetric Archimedean connection function model. Represents an exponential function. This represents the natural logarithm function.

[0028] As a further aspect of the present invention, the process of constructing and solving the log-likelihood function specifically includes:

[0029] The second-order mixed partial derivative of the joint distribution function is obtained to obtain the joint probability density function. All the joint distribution data pairs are substituted into the joint probability density function and the logarithm is taken. The summation is then used to obtain the objective function.

[0030] The Newton-Raphson iterative method is used to optimize the objective function within a preset parameter domain, calculate the parameter estimate that makes the gradient of the objective function zero, and confirm the estimate as the optimal tail dependency parameter.

[0031] As a further aspect of the present invention, the calculation process of the upper tail correlation coefficient specifically includes:

[0032] Based on the optimal tail dependency parameters, and utilizing the tail dependency property of the GumbelCopula function, the upper tail correlation coefficient is calculated according to the following formula:

[0033] ;

[0034] in, This represents the correlation coefficient of the upper tail. This represents the optimal tail dependency parameter.

[0035] A multi-sensor fusion-based mechanical measurement equipment management system, the system being used to implement the aforementioned multi-sensor fusion-based mechanical measurement equipment management method, the system comprising:

[0036] The monitoring data extreme value screening module is used to acquire temperature monitoring sequence and indication sequence, generate temperature change rate sequence by difference of the temperature monitoring sequence, generate indication error sequence by calculating the difference between the indication sequence and the standard value, and generate environmental stress extreme value sequence and measurement drift extreme value sequence by selecting the maximum absolute value from the temperature change rate sequence and the indication error sequence based on the sliding time window, and then transmit them to the sequence position normalization module.

[0037] The sequence position normalization module is used to sort the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine their positions. It calculates the positions based on the total sequence length normalization, generates the environmental stress probability variable sequence and the measurement drift probability variable sequence, and passes them to the connection function parameter estimation module.

[0038] The connection function parameter estimation module is used to pair the environmental stress probability variable sequence and the measurement drift probability variable sequence in the same time window to construct a joint distribution data pair, input it into the asymmetric Archimedes connection function model to calculate the joint probability density, estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation to maximize the likelihood function to solve the optimal tail dependency parameters, and pass them to the structural safety analysis module.

[0039] The structural safety analysis module is used to calculate the upper tail correlation coefficient using the optimal tail dependency parameter, and generate a structural instability warning command when the upper tail correlation coefficient is greater than the stability safety threshold.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] In this invention, temperature and reading data are collected synchronously, and the rate of change and error sequence are calculated. A sliding time window is used to screen the extreme values ​​of environmental stress and metering drift to construct probabilistic variables. An asymmetric Archimedes connection function model is introduced to analyze the joint distribution characteristics of the two. The optimal tail dependency parameter is solved by maximum likelihood estimation to quantify the correlation between the extreme effects of environmental stress and metering drift. Based on this, the upper tail correlation coefficient is calculated and combined with the stability threshold for judgment. This accurately identifies the potential structural instability characteristics of metering equipment caused by thermal stress impact in complex environments, effectively solving the problem that traditional methods are difficult to quantify the nonlinear coupling relationship between environment and performance, and realizing the advanced prediction of the failure risk of metering equipment. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the main steps of the present invention;

[0043] Figure 2 This is a detailed flowchart of step S1 of the present invention;

[0044] Figure 3 This is a detailed flowchart of step S2 of the present invention;

[0045] Figure 4 This is a detailed flowchart of step S3 of the present invention;

[0046] Figure 5 This is a detailed flowchart of step S4 of the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0048] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0049] Please see Figure 1 and Figure 2 This invention provides a technical solution: a method for managing mechanical measurement equipment using multi-sensor fusion, comprising the following steps:

[0050] S1: Obtain the temperature monitoring sequence and the indication sequence, generate the temperature change rate sequence by differential analysis of the temperature monitoring sequence, calculate the difference between the indication sequence and the standard value to generate the indication error sequence, and select the maximum absolute value from the temperature change rate sequence and the indication error sequence based on the sliding time window to generate the environmental stress extreme value sequence and the measurement drift extreme value sequence respectively.

[0051] The process of generating the temperature change rate sequence includes:

[0052] The original temperature monitoring sequence was acquired, and a five-point cubic smoothing algorithm was used to preprocess the sequence for denoising to remove high-frequency electromagnetic interference noise. The backward difference method was used to calculate the temperature difference between adjacent sampling points for the denoised sequence.

[0053] Obtain the sampling time interval parameter of the temperature sensor, divide the temperature difference by the sampling time interval parameter to obtain the temperature gradient value per unit time, and arrange the gradient values ​​in time order to generate a temperature change rate sequence.

[0054] Outlier truncation is performed on the temperature change rate sequence, and distorted data exceeding the physical limit threshold are replaced with the neighborhood mean to ensure that the subsequent extreme value screening process is not affected by sensor fault mutation points, thereby ensuring the authenticity of the environmental stress extreme value sequence.

[0055] The process of selecting the absolute maximum value based on a sliding time window and generating environmental stress extreme value sequences and metrological drift extreme value sequences includes:

[0056] S11: Set the time length parameter and step size parameter of the sliding time window, and divide the temperature change rate sequence and the indication error sequence into multiple local subsequences with time overlap attribute according to the time length parameter;

[0057] S12: Traverse each local subsequence, calculate the absolute values ​​of multiple elements in the subsequence of the temperature change rate sequence and the absolute values ​​of multiple elements in the subsequence of the indication error sequence, and extract the maximum absolute value within each local subsequence as the local representative feature value through comparison operation.

[0058] S13: According to the time sequence of the local subsequences, the extracted local representative feature values ​​are spliced ​​together to construct the environmental stress extreme value sequence and the metric drift extreme value sequence respectively.

[0059] This embodiment selects a mechanical measurement device deployed at a large bridge weighing monitoring station as the execution object. This device is equipped with a high-precision resistance strain gauge load cell and a patch-type PT1000 temperature sensor. The central processing unit executes the temperature monitoring sequence acquisition task, setting the sampling time interval parameter as follows. The continuous sampling length is The original temperature and voltage signals are used to generate temperature monitoring sequences through analog-to-digital conversion. Simultaneously, the central processing unit reads the mechanical readings corresponding to the temperature sampling time, reads the sensor output readings under standard weight loading, and constructs a reading sequence. .

[0060] For temperature monitoring sequences To remove high-frequency electromagnetic interference noise, the central processing unit (CPU) uses a five-point cubic smoothing algorithm for noise reduction preprocessing. The smoothing convolution kernel coefficient vector is set as... For any non-boundary point in the sequence (in Temperature value after noise reduction It is obtained through convolution operations. For example, a segment of the original temperature data is extracted as follows: (Unit: degrees Celsius), where This is the point of interference / noise. Centered on this point... Perform the calculation: Through calculation, the noise amplitude was significantly smoothed, thereby eliminating high-frequency electromagnetic interference noise.

[0061] Central processing unit for the denoised sequence Perform backward difference operation. Extract adjacent sampling points. and Calculate the difference Call the sampling time interval parameter from memory. Calculate the temperature gradient value per unit time. .like , ,but The calculated result The gradient values ​​are arranged by time index to generate an initial temperature change rate sequence.

[0062] The central processing unit then performs outlier truncation on the temperature change rate sequence. The physical limit threshold is set to... Traverse the temperature change rate sequence; if a gradient value is detected at a certain moment... The data was determined to be distorted. The central processing unit (CPU) extracted the gradient values ​​of the three neighboring points before and after the given point and calculated their arithmetic mean. and will Replace with This eliminates false stress characteristics caused by transient circuit impacts, ensuring the authenticity of the environmental stress extreme value sequence.

[0063] Simultaneously, the central processing unit calculates the indication error sequence and obtains the standard value. (Standard load). For any point in the indication sequence Calculation error .like ,but This generates an indication error sequence. .

[0064] Entering the extreme value filtering stage, the central processing unit sets the sliding time window parameters: time length parameter. (Covering 600 sampling points), step size parameter (100 sampling points). The central processing unit (CPU) combines the temperature change rate sequence with the indication error sequence according to... Extract as The overlapping subsequence. (The last part is incomplete and likely refers to a specific sequence or pattern.) Taking a time window as an example, the window covers a time interval. The specific execution of S12 is as follows: The central processing unit traverses the... Temperature change rate subsequence within a window Calculate the absolute value of each element, and extract the maximum value using the bubble sort method as the extreme value of the environmental stress for that window. Similarly, traverse the indication error subsequence within the same window. Calculate the absolute value and extract the maximum value as the extreme value of the meter drift. For example, in a window In the equation, the set of absolute values ​​of the rate of temperature change is: The maximum value is The set of absolute values ​​of indication error is: The maximum value is In S13, the central processing unit follows the window time sequence. The extracted feature values ​​are concatenated sequentially to construct a structure of length . Environmental stress extreme value sequence and measurement drift extreme value sequence Table 1 lists some of the processed extreme value sequence data.

[0065] Table 1. Example of extreme value sequence filtering results:

[0066] ;

[0067] As shown in Table 1, the system successfully extracted representative local extremum features from continuous stream data, completing the transformation from raw physical quantities to statistical extrema.

[0068] The aforementioned five-point cubic smoothing algorithm is a polynomial smoothing filtering method based on the least squares principle. By selecting a data point and five points in its neighborhood, a cubic polynomial is used for fitting, thereby effectively suppressing high-frequency random noise interference while preserving the low-frequency trend characteristics of the signal.

[0069] The aforementioned sliding time window refers to a technique for dynamically capturing data from a time series data stream. By setting a fixed window length and a moving step size, it continuously captures local data segments, resulting in overlapping areas between adjacent windows, thereby capturing the local statistical characteristics of the signal's evolution over time.

[0070] Please see Figure 1 and Figure 3 S2: Arrange the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine their positions. Calculate the positions based on the normalization of the total sequence length, and generate the environmental stress probability variable sequence and the measurement drift probability variable sequence.

[0071] The process of calculating the position based on the total sequence length normalization and generating the environmental stress probability variable sequence and the metrological drift probability variable sequence includes:

[0072] S21: The environmental stress extreme value sequence and the measurement drift extreme value sequence are sorted in ascending order using the quick sorting algorithm. A mapping relationship between numerical value and sequence index is established to determine the specific position of each data point in the original sequence in the sorted sequence.

[0073] S22: Obtain the total data length value of the environmental stress extreme value sequence, divide the position by the sum of the total data length value and the correction factor, calculate the empirical cumulative distribution probability value corresponding to each data point, and eliminate the influence of sample size difference on probability distribution;

[0074] S23: Keeping the time order index of the original sequence unchanged, map the calculated empirical cumulative distribution probability value back to the original time point, and assemble them to generate the environmental stress probability variable sequence and the measurement drift probability variable sequence respectively.

[0075] The central processing unit outputs the environmental stress extreme value sequence in step S1. and measurement drift extreme value sequence Perform ascending sort and position determination operations. Assume the total length of the sequence is... (That is, data from 100 time windows were collected).

[0076] In S21, the central processing unit calls the quicksort algorithm, using the middle element of the sequence as the pivot value, to sort the sequence... Divide the data into two sub-regions: one with values ​​less than the pivot and the other with values ​​greater than the pivot. Repeat this process recursively until the sequence is ordered. The resulting ordered sequence is then generated. ,in Similarly, generate ordered sequences. The central processing unit establishes a mapping table between numerical values ​​and sequence indices. For example, the original sequence... The first data point (From window 1), in the sorted sequence If it is located in the 55th position, then its ranking is... Original sequence The third data point (From window 3), if it is in the 82nd position after sorting, then .

[0077] In S22, the central processing unit performs a normalization calculation based on the total sequence length. The total sequence length value is then obtained. Set the correction factor. The calculation formula is: For the data point in window 1 (environmental stress extreme value 0.32, rank 55): For the data point in window 3 (environmental stress extreme value 0.45, position 82): Similarly, calculate the probability value of the measurement drift extreme value sequence. Assume the measurement drift extreme value of window 1 is 0.15 in the sequence. If the person is ranked 60th in the middle, then: .

[0078] In S23, the central processing unit maintains the time-order index of the original sequence. Without changing the calculated empirical cumulative distribution probability values, the original time points are mapped back. For window 1, the probability variables are assembled. For window 3, assume the probability corresponding to its extreme value of meter drift is... Then the assembly probability variable pairs Ultimately, the central processing unit generates a length of... Environmental stress probability variable sequence and the sequence of probability variables for measurement drift .

[0079] The aforementioned empirical cumulative distribution probability value refers to a nonparametric statistic that approximates the population distribution function value by measuring the proportion of data points in the statistical sample data that are less than or equal to a certain value. It is used to uniformly map physical quantities with different dimensions and ranges of values ​​into a standard probability space of 0 to 1.

[0080] Please see Figure 1 and Figure 4 S3: Pair the environmental stress probability variable series and the econometric drift probability variable series in the same time window to construct a joint distribution data pair, input it into the asymmetric Archimedes connection function model to calculate the joint probability density, and estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation to maximize the likelihood function in order to solve the optimal tail dependency parameters.

[0081] The process of calculating the joint probability density includes:

[0082] Obtain the generator function and its higher-order derivatives of the asymmetric Archimedes connection function model. Combine the environmental stress probability variable sequence and the metrological drift probability variable sequence, and calculate the joint probability density value according to the following formula:

[0083] ;

[0084] in, Represents the joint probability density value. The numerical values ​​represent the input sequence of environmental stress probability variables. This represents the numerical values ​​of the input measurement drift probability variable sequence. This represents the skewness component in the optimal tail dependency parameters. This represents the aggregation degree component in the optimal tail dependency parameter. Represents the value of the connection distribution function. Generative functions representing asymmetric Archimedean connection functions. The first derivative of the generating function. The second derivative of the generating function. The value of the function representing the connection distribution function;

[0085] The process of solving for the optimal tail dependency parameters includes:

[0086] S31: Combine the elements in the environmental stress probability variable sequence and the quantitative drift probability variable sequence at the same time point in pairs to construct a set of joint distribution data pairs including multiple sets of two-dimensional vectors, which serve as input samples for dependency modeling.

[0087] S32: Call the probability density function of the preset asymmetric Archimedes connection function model, substitute the set of joint distribution data pairs into the function, and construct the log-likelihood function about the model shape parameters;

[0088] S33: The Newton-Raphson iterative optimization algorithm is used to find the optimal solution for the log-likelihood function, calculate the parameter solution that maximizes the function value, and determine the parameter solution as the optimal tail-dependent parameter.

[0089] The central processing unit will pair the environmental stress probability variable sequences with the same time window. With the sequence of probability variables of measurement drift Constructing a set of jointly distributed data pairs . Set The input is processed using an asymmetric Archimedes connection function model.

[0090] In calculating the joint probability density, the model employs an asymmetric GumbelCopula as its underlying architecture. The generator functions of the asymmetric Archimedean connection function model are obtained. And its derivative, combined with the environmental stress probability variable series and the measurement drift probability variable series, the joint probability density value is calculated according to the following formula: in, This represents the joint probability density value, which quantifies the probability density of two variables occurring simultaneously. Represents the value of the connection distribution function; and and represent the first and second derivatives of the generator function, respectively. The advantage of this formula is that it directly obtains the joint density through analytical operations on the generators and their derivatives, avoiding complex numerical differentiation and improving computational accuracy.

[0091] During the process of solving the optimal tail dependency parameters (S31-S33): S31: Central Processing Unit Extraction A set of data pairs. For example, data points. S32: Constructing the log-likelihood function S33: Employs the Newton-Raphson iterative optimization algorithm. Initialize parameters. Calculate the gradient vector and Hessian matrix, and perform iterative updates until the parameters converge. Assume convergence occurs after 15 iterations, yielding the optimal solution: , .

[0092] To verify the rationality of the parameter acquisition, a set of high-risk samples was selected. Substitute into the above formula to estimate the density at a single point (assuming asymmetric parameters). The effects have been integrated into the generation meta-parameters. (For simplified demonstration purposes) 1. Calculate the join function value : 2. Calculate the first derivatives of the generators. : . 3. Calculate the second derivative of the generator. : 4. Calculate the joint probability density. : The result of 7.85 is much larger than the density value of 1.0 in the independent case, indicating a very strong positive correlation at high quantiles (i.e., extreme stress and large drift).

[0093] The aforementioned asymmetric Archimedes connection function refers to a class of mathematical models that can construct multidimensional joint distributions through generator functions. Its characteristic is that it allows the dependency structure between variables to be asymmetric (such as the upper tail dependency being stronger than the lower tail dependency), thereby more accurately describing the nonlinear influence of environmental stress on metrological performance.

[0094] The Newton-Raphson iterative optimization algorithm mentioned above is a method for approximating the solution of equations in the real and complex number domains. By utilizing the first and second derivatives (Hessian matrix) of the function, a quadratic approximation model is constructed to determine the search direction, thereby achieving rapid convergence to the extreme points of the objective function.

[0095] Please see Figure 1 and Figure 5 S4: Calculate the upper tail correlation coefficient based on the optimal tail dependency parameter, and generate a structural instability warning instruction when the upper tail correlation coefficient is greater than the stability safety threshold.

[0096] The process of calculating the upper tail correlation coefficient includes:

[0097] Obtain the skewness and aggregation parameters from the optimal tail dependency parameters, construct a tail dependency evaluation model based on extreme value theory, and calculate the upper tail correlation coefficient according to the following formula:

[0098] ;

[0099] in, Represents the upper tail correlation coefficient. Represents the degree of aggregation parameter. Represents the skewness parameter. This represents the structural influence factor function used to correct for asymmetry, where 2 in the formula represents a constant coefficient;

[0100] The process of generating a structural instability early warning command includes:

[0101] S41: Obtain the optimal tail dependency parameters, call the preset limit dependency formula, calculate the limit conditional probability when the probability variable approaches 1, and obtain the upper tail correlation coefficient.

[0102] S42: Obtain the preset stability and safety threshold, compare the upper tail correlation coefficient with the stability and safety threshold, and determine whether there is a high probability of concurrent extreme event risk. If the coefficient value exceeds the threshold, the device is determined to have entered the structurally vulnerable area.

[0103] S43: When it is determined that the equipment has entered the structurally vulnerable area, the alarm logic is immediately triggered, the instability probability level, the current environmental stress intensity and the recommended calibration time are encapsulated, a structural instability early warning command is generated and sent to the remote monitoring terminal;

[0104] The process of generating a structural instability early warning command includes:

[0105] Obtain the upper tail correlation coefficient sequence within the historical time window, calculate the first trend derivative of the coefficient sequence, and if the upper tail correlation coefficient is greater than the stability safety threshold and the first trend derivative is positive, then the device is determined to be in the deterioration acceleration stage.

[0106] Based on the optimal tail dependency parameters, the critical environmental stress value leading to high tail correlation is back-derived. The critical value is marked as the disaster-causing factor of equipment failure, and the disaster-causing factor is associated with and encapsulated with the current metering drift extreme value sequence characteristics.

[0107] The system generates a structural instability warning command through an encrypted communication protocol. The command data packet includes the device's unique identification code, risk level code, and recommended insulation or recalibration measures to ensure that maintenance personnel can specifically block the thermal stress transmission path.

[0108] The central processing unit (CPU) calculates the optimal tail dependency parameters based on step S3. and Perform upper tail correlation coefficient The calculation.

[0109] The formula for calculating the upper tail correlation coefficient is: in, Represents the upper tail correlation coefficient, with a numerical range of ; Represents the degree of aggregation; Represents the skewness parameter; The structural influence factor function used to correct asymmetry is defined in this embodiment as follows: The advantage of this formula is that it compresses the complex asymmetric dependency structure into a scalar index, which can intuitively quantify the conditional probability of metering equipment failure under extreme environmental stress.

[0110] Substitute the parameters for calculation: 1. Calculate the structural influence factor: 2. Calculate the exponential term: 3. Calculate the final coefficients: Calculation results This indicates that, under the current conditions of extremely high environmental stress, the probability of severe metering drift in the equipment is approximately 67.64%.

[0111] The process of generating structural instability warning instructions includes steps S41 to S43. S41: The central processing unit has completed the upper tail correlation coefficient calculation. S42: Obtain the preset stability and security threshold. This threshold is set based on the equipment accuracy class and the statistical failure boundary of the maximum allowable error. The upper tail correlation coefficient is numerically compared with the stability safety threshold: The result is true. This indicates a high probability of concurrent extreme events occurring in the current device, and the device is determined to have entered a structurally vulnerable area. S43: The central processing unit immediately triggers the alarm logic. It retrieves the coefficient sequence from the historical time window (the past 5 calculations). Calculate the first-order trend derivative (slope): .because Furthermore, since the first-order trend derivative is positive, the device is determined to be in the "accelerated deterioration stage".

[0112] The central processing unit (CPU) back-calculates the critical environmental stress value for high tail correlation (e.g., the rate of temperature change needs to be controlled within a certain range) based on the optimal tail dependency parameters. (within), will The factors identified as causative agents of equipment failure are marked. The central processing unit encapsulates an instruction data packet containing: the equipment's unique identifier (UUID-5501), risk level code (Level 3), recommended insulation measures, and recommended calibration time. This instruction is sent to the remote monitoring terminal via an encrypted communication protocol. Table 2 summarizes the key parameters for this warning generation.

[0113] Table 2 Summary of Structural Instability Early Warning Parameters:

[0114] ;

[0115] As shown in Table 2, by comparing the calculation results with the threshold, the system clearly defines the current risk status of the equipment and generates an early warning instruction containing specific maintenance suggestions, effectively blocking the thermal stress transmission path.

[0116] The aforementioned upper tail correlation coefficient refers to a statistical indicator used in extreme value theory to measure the degree of interdependence between two random variables in the tail of their distribution (i.e., the region of maximum values). The higher the value, the greater the probability that the other variable will also have an extreme value when one variable has an extreme value.

[0117] A multi-sensor fusion mechanical measurement equipment management system is provided, which is used to execute the aforementioned multi-sensor fusion mechanical measurement equipment management method. The system includes:

[0118] The monitoring data extreme value screening module is used to acquire temperature monitoring sequences and indication sequences, generate temperature change rate sequences by differential analysis of temperature monitoring sequences, generate indication error sequences by calculating the difference between indication sequences and standard values, and filter the absolute maximum values ​​from temperature change rate sequences and indication error sequences based on sliding time windows to generate environmental stress extreme value sequences and metering drift extreme value sequences, which are then transmitted to the sequence position normalization module.

[0119] The sequence position normalization module is used to sort the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine the position. It calculates the position based on the total sequence length, generates the environmental stress probability variable sequence and the measurement drift probability variable sequence, and passes them to the connection function parameter estimation module.

[0120] The connection function parameter estimation module is used to pair environmental stress probability variable sequences and econometric drift probability variable sequences within the same time window to construct joint distribution data pairs, input them into the asymmetric Archimedes connection function model to calculate the joint probability density, estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation, maximize the likelihood function to solve for the optimal tail dependency parameters, and then pass them to the structural safety analysis module.

[0121] The structural safety analysis module is used to calculate the upper tail correlation coefficient through the optimal tail dependency parameter, and generate a structural instability warning command when the upper tail correlation coefficient is greater than the stability safety threshold.

[0122] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A method for managing mechanical measurement equipment using multi-sensor fusion, characterized in that, Includes the following steps: S1: Obtain the temperature monitoring sequence and the indication sequence, generate a temperature change rate sequence by differential analysis of the temperature monitoring sequence, calculate the difference between the indication sequence and the standard value to generate an indication error sequence, and generate an environmental stress extreme value sequence and a measurement drift extreme value sequence by selecting the maximum absolute value from the temperature change rate sequence and the indication error sequence based on a sliding time window. S2: Arrange the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine their positions. Calculate the positions based on the total sequence length normalization to generate the environmental stress probability variable sequence and the measurement drift probability variable sequence. The specific steps of S2 are as follows: The environmental stress extreme value sequence and the metrological drift extreme value sequence are sorted in ascending order. While maintaining the correspondence between data values, the position index of each data point in its respective sequence is calculated to determine the position value, thus obtaining the environmental stress position sequence and the metrological drift position sequence. Obtain the total length of the environmental stress extreme value sequence, and use the empirical distribution function formula to divide each position value in the environmental stress position sequence and the measurement drift position sequence by the sum of the total sequence length and the correction factor, so as to map the discrete position features into probability values ​​with values ​​between zero and one, thereby generating the environmental stress probability variable sequence and the measurement drift probability variable sequence. S3: Pair the environmental stress probability variable sequence and the measurement drift probability variable sequence in the same time window to construct a joint distribution data pair, input it into the asymmetric Archimedes connection function model to calculate the joint probability density, and estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation to maximize the likelihood function in order to solve the optimal tail dependency parameters. S4: Calculate the upper tail correlation coefficient based on the optimal tail dependency parameter, and generate a structural instability warning command when the upper tail correlation coefficient is greater than the stability safety threshold.

2. The multi-sensor fusion mechanical measurement equipment management method according to claim 1, characterized in that, The specific steps of S1 are as follows: Temperature monitoring data and mechanical indication data are collected by sensor interfaces deployed at key nodes of the mechanical measurement equipment. The temperature monitoring data is processed by a first-order difference algorithm to eliminate baseline trends and retain high-frequency fluctuation characteristics, generating a temperature change rate sequence. At the same time, the numerical deviation between the mechanical indication data and the preset standard measurement reference value is calculated, and the absolute value is taken to generate an indication error sequence. The window length and sliding step size of the sliding time window are set according to the preset sampling frequency. The sliding time window is moved synchronously on the temperature change rate sequence and the indication error sequence according to the sliding step size. The data subset in each time window is traversed and the data point with the largest absolute value is extracted. The environmental stress extreme value sequence and the measurement drift extreme value sequence are constructed in sequence.

3. The multi-sensor fusion mechanical measurement equipment management method according to claim 1, characterized in that, The specific steps of S3 are as follows: Based on the timestamp alignment method, the elements in the environmental stress probability variable sequence corresponding to the same time window are paired with the elements in the measurement drift probability variable sequence to construct joint distribution data pairs; The joint distribution data pairs are substituted into the probability density function of the pre-constructed asymmetric Archimedes connection function model to construct a log-likelihood function containing the shape parameters to be estimated. The value of the shape parameters is iteratively adjusted using a numerical optimization algorithm until the function value of the log-likelihood function reaches its maximum. The parameter value at this point is taken as the optimal tail dependency parameter.

4. The multi-sensor fusion mechanical measurement equipment management method according to claim 3, characterized in that, The specific steps of S4 are as follows: Obtain the optimal tail dependency parameters obtained from the solution, and calculate the limiting conditional probability of the asymmetric Archimedes connection function model in the upper right tail according to the preset extreme value dependency theory formula, and generate the upper tail correlation coefficient. The correlation coefficient of the upper tail is compared with the preset stability safety threshold. If the correlation coefficient of the upper tail exceeds the stability safety threshold, the mechanical measuring equipment is determined to be in a high-risk state of thermal structural instability, and the alarm logic is immediately triggered and a structural instability warning command is generated.

5. The multi-sensor fusion mechanical measurement equipment management method according to claim 2, characterized in that, The screening process for the temperature change rate sequence and the indication error sequence specifically includes: The window length of the sliding time window is set to cover the entire thermal response cycle of the mechanical measuring equipment, and the sliding step size is set to be less than half of the window length to ensure the continuity and overlap of data sampling. Within the coverage area of ​​each sliding window, the absolute values ​​of all data points in the temperature change rate sequence are taken and compared. The point with the largest value is retained as the representative value of environmental thermal stress for that window period. At the same time, the same absolute value maximization filtering operation is performed on the indication error sequence. The two extreme values ​​selected are stored in the corresponding sequence containers in chronological order, thus completing the construction of the environmental stress extreme value sequence and the measurement drift extreme value sequence.

6. The multi-sensor fusion mechanical measurement equipment management method according to claim 3, characterized in that, The asymmetric Archimedes connection function model specifically adopts the Gumbel-Hougaard Copula function structure, and its joint distribution function is mathematically expressed as follows: ； in, Let represent the joint distribution function, u represent the value in the sequence of environmental stress probability variables, and v represent the value in the sequence of measurement drift probability variables. The shape parameters represent the shape parameters of the asymmetric Archimedean connection function model. Represents an exponential function. This represents the natural logarithm function.

7. The multi-sensor fusion mechanical measurement equipment management method according to claim 6, characterized in that, The process of constructing and solving the log-likelihood function specifically includes: The second-order mixed partial derivative of the joint distribution function is obtained to obtain the joint probability density function. All the joint distribution data pairs are substituted into the joint probability density function and the logarithm is taken. The summation is then used to obtain the objective function. The Newton-Raphson iterative method is used to optimize the objective function within a preset parameter domain, calculate the parameter estimate that makes the gradient of the objective function zero, and confirm the estimate as the optimal tail dependency parameter.

8. The multi-sensor fusion mechanical measurement equipment management method according to claim 4, characterized in that, The calculation process of the upper tail correlation coefficient specifically includes: Based on the optimal tail dependency parameters, and utilizing the tail dependency property of the GumbelCopula function, the upper tail correlation coefficient is calculated according to the following formula: ； in, This represents the correlation coefficient of the upper tail. This represents the optimal tail dependency parameter.

9. A multi-sensor fusion mechanical measurement equipment management system, characterized in that, The system is used to implement the multi-sensor fusion mechanical measurement equipment management method according to any one of claims 1-8, the system comprising: The monitoring data extreme value screening module is used to acquire temperature monitoring sequence and indication sequence, generate temperature change rate sequence by difference of the temperature monitoring sequence, generate indication error sequence by calculating the difference between the indication sequence and the standard value, and generate environmental stress extreme value sequence and measurement drift extreme value sequence by selecting the maximum absolute value from the temperature change rate sequence and the indication error sequence based on the sliding time window, and then transmit them to the sequence position normalization module. The sequence position normalization module is used to sort the environmental stress extreme value sequence and the measurement drift extreme value sequence in ascending order and determine their positions. It calculates the positions based on the total sequence length normalization, generates the environmental stress probability variable sequence and the measurement drift probability variable sequence, and passes them to the connection function parameter estimation module. The connection function parameter estimation module is used to pair the environmental stress probability variable sequence and the measurement drift probability variable sequence in the same time window to construct a joint distribution data pair, input it into the asymmetric Archimedes connection function model to calculate the joint probability density, estimate the shape parameters of the asymmetric Archimedes connection function model by maximum likelihood estimation to maximize the likelihood function to solve the optimal tail dependency parameters, and pass them to the structural safety analysis module. The structural safety analysis module is used to calculate the upper tail correlation coefficient using the optimal tail dependency parameter, and generate a structural instability warning command when the upper tail correlation coefficient is greater than the stability safety threshold.

Images