Online instrument automatic inspection and calibration method and system

Abstract

The invention relates to the technical field of instrument testing, in particular to an on-line instrument automatic inspection and calibration method and system, and the method comprises the following steps: obtaining measurement errors, humiture and electromagnetic interference data in real time, carrying out the normalized calculation of an error impact factor distribution value, monitoring an error change rate, and smoothly adjusting the weight according to an index. And analyzing the multi-channel Pearson correlation structure coupling correction matrix, screening the key error source optimization adjustment amount, and judging the error correction stability. According to the method, through real-time measurement error and environmental factor calculation, high-influence error sources are accurately screened, error analysis pertinence is improved, error weights are dynamically adjusted, correction parameters are made to change in a self-adaptive mode along with measurement conditions, compensation accuracy is improved, key error interaction relations are analyzed and screened in combination with multiple measurement channels, non-key factor interference is reduced, and the method is suitable for large-scale popularization and application. The error adjustment calculation is optimized, the correction convergence speed is accelerated, the environmental fluctuation influence is reduced, and the data consistency and stability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of instrument testing, and particularly to an online instrument automatic inspection and calibration method and system. Background Art

[0002] The technical field of instrument testing includes related technologies for detecting, calibrating, and verifying the performance, accuracy, stability, etc. of various measuring instruments. This technical field involves core contents such as measurement signal acquisition, error analysis, data correction, and measurement environment control, and is widely applied in fields such as industrial automation, scientific research, and energy monitoring. In this field, instrument testing methods mainly include manual calibration, automated detection, online monitoring, etc., with the aim of ensuring the accuracy and reliability of instrument measurement data. The traditional manual calibration method relies on manual operation, which has problems such as low efficiency and being easily affected by human factors, while automated detection realizes the rapid acquisition of measurement data and error correction through a combination of hardware and software. The online monitoring technology can collect measurement data in real time during the operation of the instrument, and perform error analysis and correction, thereby improving the long-term stability and measurement consistency of the instrument.

[0003] Among them, the online instrument automatic inspection and calibration method refers to a method of automatically inspecting and calibrating the measurement data of an instrument during the operation of the instrument through specific technical means. This method mainly covers technical matters such as measurement data acquisition, error calculation, reference benchmark comparison, and correction parameter calculation. First, a sensor or a measuring device is used to collect the measured parameters, and the obtained data is preliminarily screened to eliminate outliers. Then, the measurement error is calculated through known standard reference values, and the error correction parameters are determined in combination with a comparison model. Subsequently, the measurement data is calibrated using the correction parameters to adjust the measurement deviation and improve data consistency. During this process, an error tracking method based on time series analysis can be adopted to ensure that the calibration parameters are dynamically adjusted according to the change of the instrument working state. In addition, this method can also be combined with environmental compensation technology to make the measurement and calibration process adapt to different working conditions and ensure the measurement accuracy of the instrument in a complex environment.

[0004] The prior art relies on fixed error compensation parameters and lacks the ability to adapt to changes in the measurement environment, resulting in limited error correction effects in complex environments. The error source screening method is relatively rough, making it difficult to identify key error factors and affecting the correction accuracy. The error interaction relationship is not fully considered during the error calibration process, which may lead to error superposition or insufficient compensation, making the correction effect unstable. The error adjustment strategy lacks the ability of dynamic optimization, resulting in a decline in the adaptability of the correction parameters after long-term operation and affecting the measurement accuracy and consistency. Summary of the Invention

[0005] The purpose of the present invention is to solve the disadvantages existing in the prior art, and to propose an online instrument automatic inspection and calibration method and system.

[0006] To achieve the above object, the present invention adopts the following technical solution: An online instrument automatic inspection and calibration method, comprising the following steps:

[0007] S1: Obtain the error value and environmental impact factors in the real-time measurement data. The environmental impact factors include temperature, humidity, and electromagnetic interference. Calculate the error change trend in the environment, screen the error sources with an error rate exceeding the threshold, perform normalized calculation on the impact factors, and obtain the error impact factor allocation value;

[0008] S2: Invoke the error impact factor allocation value, monitor the change rate of the measurement error of the online instrument, compare the error rate with the threshold, calculate the ratio of the change rate of the difference to the current error weight, and adjust the weight allocation according to the exponential smoothing method to obtain the dynamic error weight adjustment value;

[0009] S3: Invoke the error impact factor allocation value, calculate the Pearson correlation of the error data of multiple measurement channels, screen the associated error pairs, calculate the partial correlation coefficient, eliminate the non-critical interaction terms, use multiple regression operations to calculate the interaction impact factor, invoke the dynamic error weight adjustment value to correct the error impact factor, and calculate the corrected error propagation impact to obtain the error coupling correction matrix;

[0010] S4: Invoke the error coupling correction matrix, sort and screen the error source with the highest contribution rate, calculate the error adjustment amount, and invoke the dynamic error weight adjustment value to optimize the weight to obtain the error optimization calculation value.

[0011] As a further solution of the present invention, the error impact factor allocation value includes error sources with an error rate exceeding the threshold, normalized environmental impact factors, and error impact factor weights. The dynamic error weight adjustment value includes the error change rate, the error weight ratio, and the exponentially smoothed adjusted weight. The error coupling correction matrix includes key interaction error terms, partial correlation error calculation values, and multiple regression interaction impact factors. The error optimization calculation value includes the error source with the highest contribution rate, the error adjustment amount, and the optimized dynamic weight.

[0012] As a further solution of the present invention, the specific steps for obtaining the error impact factor allocation value are as follows:

[0013] S101: Obtain the error value in the real-time measurement data, and collect environmental impact factors, including temperature, humidity, and electromagnetic interference. Match the error value with the environmental impact factors, establish a preliminary correspondence between the error and environmental factors, calculate the correlation coefficient of multiple environmental factors on the error, screen the environmental factors with higher correlation, and obtain the environmental factors with high error correlation;

[0014] S102: Based on the environmental factors with high error correlation, calculate the trend of the error under the change of differential environmental factors. Using the error change rate as a measurement index, calculate the change rate of the error on multiple environmental factors, and screen out the error sources with an error rate exceeding the threshold to obtain the error sources exceeding the threshold;

[0015] S103: Based on the error sources exceeding the threshold, perform normalization processing on multiple environmental factors, calculate the contribution rate of the normalized environmental factors to the error, and use the formula:

[0016]

[0017] Perform operations to obtain the error influence factor allocation value;

[0018] Among them, F env represents the error influence factor allocation value of the environmental factor env on the error, E src represents the error value of the error source src, NF env,src represents the normalized influence factor of the environmental factor env at the error source src, N sc represents the total number of error sources.

[0019] As a further solution of the present invention, the specific steps for obtaining the dynamic error weight adjustment value are as follows:

[0020] S201: Invoke the error influence factor allocation value, monitor the change rate of the measurement error of the on-line instrument, calculate the ratio change amount between the current measurement error value and the previous error value, and obtain the error rate change value. At the same time, compare the error rate change value with the set error change threshold to determine whether the current error state exceeds the threshold range and obtain the error state determination result;

[0021] S202: Based on the error state determination result, calculate the ratio change between the error change rate and the current error weight ratio, adjust the error weight allocation value, and perform operations to obtain the adjusted error weight ratio;

[0022] S203: Invoke the adjusted error weight ratio, and based on the current measurement error data, use the formula:

[0023]

[0024] Perform operations to obtain the dynamic error weight adjustment value;

[0025] Among them, E DWA represents the dynamic error weight adjustment value, W new represents the adjusted error weight ratio, W old represents the error weight ratio before adjustment, ΔE mes represents the error change amount at the mes-th time, N mesRepresents the number of error measurements, T E Represents the error change trend value, T W Represents the error weight smoothing coefficient.

[0026] As a further solution of the present invention, the steps for obtaining the error coupling correction matrix are specifically as follows:

[0027] S301: Invoke the error influence factor assignment value, calculate the Pearson correlation of the error data of multiple measurement channels, screen the error pairs with Pearson correlation coefficients higher than the error correlation threshold, eliminate the low-correlation error pairs, and obtain the error correlation screening result;

[0028] S302: Based on the error correlation screening result, calculate the partial correlation coefficient of the screened error pairs, screen out the error pairs with partial correlation coefficients lower than the interaction influence reference value, and obtain the partial correlation coefficient screening result;

[0029] S303: Based on the partial correlation coefficient screening result, use multiple regression to calculate the interaction influence factor, using the formula:

[0030]

[0031] Obtain the calculation result of the interaction influence factor;

[0032] Among them, I c Represents the calculated value of the interaction influence factor, P prn Represents the Pearson correlation coefficient of the prnth group of error pairs, EIF prn Represents the error influence factor of the prnth group of error pairs, R par Represents the partial correlation coefficient of the parth group of error pairs, A par Represents the normalized error influence factor of the parth group of error pairs, B tot Represents the total amount of error influence factors of the totth group of error pairs, and PCT represents the total number of error pairs.

[0033] As a further solution of the present invention, the steps for obtaining the error optimization calculated value are specifically as follows:

[0034] S401: Invoke the error coupling correction matrix, separate the influences of multiple error sources, obtain the influence amount of multiple error sources on the system error, calculate the contribution rate of multiple error sources to the system error, and at the same time sort the error sources according to the contribution rate, screen the error source with the highest contribution rate, and obtain the error source contribution ranking result;

[0035] S402: Based on the error source contribution ranking result, calculate the error adjustment amount of multiple error sources, and through error adjustment calculation, correct the system error after adjusting the error source, using the formula:

[0036]

[0037] Calculate the correction amplitude of the error source through operations, and adjust the system error value to obtain the system corrected error value;

[0038] Among them, SCE represents the system corrected error value, CC src represents the contribution coefficient of the src-th error source, E src represents the error value of the src-th error source, represents the mean error of all error sources, V src represents the error variance of the src-th error source, N sc represents the total number of error sources;

[0039] S403: Call the system corrected error value, optimize the system error weight based on the dynamic error weight adjustment value, and perform optimization adjustment on the corrected error value to obtain the error optimized calculation value.

[0040] As a further solution of the present invention, the method further includes:

[0041] S5: Call the error optimized calculation value, calculate the error change amount before and after correction, compare the change amount with the threshold value, judge the error convergence situation, and obtain the error correction stability value;

[0042] The error correction stability value includes the error change amount before and after correction, the error convergence determination result, and the error correction value.

[0043] As a further solution of the present invention, the steps for obtaining the error correction stability value are specifically:

[0044] S501: Call the error optimized calculation value, calculate the change amount of the error before and after correction, and at the same time calculate based on the error value before correction, the error value after correction, and the optimized calculation value, using the formula:

[0045]

[0046] Calculate to obtain the error change correction amount;

[0047] Among them, ECC represents the error change correction amount, E prev represents the error value before correction, E corr represents the error value after correction, E EOC represents the error optimized calculation value, and the log operation enhances the calculation adjustability;

[0048] S502: Call the error change correction amount, compare it with the error convergence threshold value, and at the same time calculate the difference between the error change correction amount and the error convergence threshold value, judge whether the error change correction amount is less than the error convergence threshold value, and combine the change rate to perform error convergence determination to obtain the error convergence status flag;

[0049] S503: Call the error convergence status identifier, judge the error change trend based on the convergence status parameter. If the error convergence status identifier meets the stability condition, determine the error correction result and obtain the error correction stability value.

[0050] An on-line instrument automatic inspection and calibration system, which is used to execute the above on-line instrument automatic inspection and calibration method. The system includes:

[0051] The error influence factor analysis module obtains the error value, environmental temperature, humidity, and magnetic interference intensity in the real-time data, converts the environmental parameters into standard values, calculates the multi-dimensional influence ratio based on the error fluctuation range and the standard values, compares the environmental fluctuation correction influence ratio according to the measurement stability, screens the parameters corresponding to the error sources exceeding the threshold, calculates the parameter combination using weighted comparison, and generates the error influence factor allocation value;

[0052] The error dynamic weight adjustment module calls the error influence factor allocation value, monitors the measurement error change rate, compares the error change with the set threshold, adjusts the weight parameter according to the ratio, and generates the dynamic error weight adjustment value;

[0053] The error coupling correction calculation module calls the error influence factor allocation value, calculates the multi-channel Pearson coefficient, screens the correlated error pairs exceeding the threshold, correlates according to the partial correlation comparison data, calls the dynamic error weight adjustment value to correct the error influence factor, and generates the error coupling correction matrix;

[0054] The error optimization adjustment module calls the error coupling correction matrix, compares the contribution rates of multiple error sources, screens the parameters with contributions exceeding the threshold, calculates the weight difference, and generates the error optimization calculation value;

[0055] The error correction stability determination module calls the error optimization calculation value, detects the error difference before and after correction, compares the difference with the convergence threshold, judges stability according to the comparison result, and generates the error correction stability value.

[0056] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0057] In the present invention, through the calculation of real-time measurement errors and environmental factors, the high-influence error sources are accurately screened, making the analysis of error sources more targeted. By using the method of dynamically adjusting the error weight, the error correction parameters can be adaptively adjusted according to the measurement conditions, improving the accuracy of error compensation. Combining the data analysis of multiple measurement channels, the key error interaction relationships are screened, reducing the interference of non-critical error factors, and making the calibration more targeted. Optimizing the calculation of the error adjustment amount, accelerating the convergence speed of error correction, reducing the influence of measurement environment fluctuations, and improving the consistency and long-term stability of measurement data. Description of the Drawings

[0058] Figure 1 It is a schematic diagram of the working process of the present invention;

[0059] Figure 2 It is a flowchart of the steps for obtaining the allocation value of the error influence factor of the present invention;

[0060] Figure 3 It is a flowchart of the steps for obtaining the adjustment value of the dynamic error weight of the present invention;

[0061] Figure 4 It is a flowchart of the steps for obtaining the error coupling correction matrix of the present invention;

[0062] Figure 5 It is a flowchart of the steps for obtaining the optimized calculation value of the error of the present invention;

[0063] Figure 6 It is a flowchart of the steps for obtaining the stable value of error correction of the present invention. Detailed implementation manners

[0064] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0065] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality" is two or more, unless otherwise specifically defined.

[0066] Embodiment 1

[0067] Please refer to Figure 1 , the present invention provides a technical solution: an online instrument automatic inspection and calibration method, including the following steps:

[0068] S1: Obtain the error value and environmental influence factors in the real-time measurement data. The environmental influence factors include temperature, humidity and electromagnetic interference. Calculate the error change trend in the environment, screen the error sources with an error rate exceeding the threshold, and perform normalization calculation on the influence factors to obtain the allocation value of the error influence factor;

[0069] S2: Call the error impact factor allocation value, monitor the change rate of the measurement error of the on-line instrument, compare the error rate with the threshold value, calculate the ratio of the change rate of the difference to the current error weight, and adjust the weight allocation according to the exponential smoothing method to obtain the dynamic error weight adjustment value;

[0070] S3: Call the error impact factor allocation value, calculate the Pearson correlation of the error data of multiple measurement channels, screen the associated error pairs, calculate the partial correlation coefficient, eliminate the non-critical interaction terms, use multiple regression operations to calculate the interaction impact factor, construct a coupling correction matrix, call the dynamic error weight adjustment value to correct the error impact factor, and calculate the corrected error propagation impact to obtain the error coupling correction matrix;

[0071] S4: Call the error coupling correction matrix, sort and screen the error source with the highest contribution rate, calculate the error adjustment amount, call the dynamic error weight adjustment value to optimize the weight, and obtain the error optimization calculation value;

[0072] S5: Call the error optimization calculation value, calculate the error change amount before and after correction, compare the change amount with the threshold value, judge the error convergence situation, and obtain the error correction stability value.

[0073] The error impact factor allocation value includes the error source with the error rate exceeding the threshold, the normalized environmental impact factor, and the error impact factor weight. The dynamic error weight adjustment value includes the error change rate, the error weight ratio, and the exponentially smoothed adjusted weight. The error coupling correction matrix includes the key interaction error term, the partial correlation error calculation value, and the multiple regression interaction impact factor. The error optimization calculation value includes the error source with the highest contribution rate, the error adjustment amount, and the optimized dynamic weight. The error correction stability value includes the error change amount before and after correction, the error convergence determination result, and the error correction value.

[0074] Please refer to Figure 2 , and the steps for obtaining the error impact factor allocation value are specifically as follows:

[0075] S101: Obtain the error value in the real-time measurement data, and collect the environmental impact factors, including temperature, humidity, and electromagnetic interference. Match the error value with the environmental impact factors, establish a preliminary correspondence between the error and the environmental factors, calculate the correlation coefficient of multiple environmental factors on the error, screen the environmental factors with higher correlation, and obtain the environmental factors with high error correlation;

[0076] For example, to measure the reading deviation of a certain device in different environments. Taking a power transformer as an example, its voltage output may vary in different temperature and humidity environments. Therefore, while collecting the error values, it is necessary to record the external environmental parameters. The environmental impact factors include parameters such as temperature, humidity, and electromagnetic interference. These parameters can be obtained through sensors. For example, a temperature sensor can detect the ambient temperature, a humidity sensor is used to measure the air humidity, and electromagnetic interference can be measured by an electromagnetic wave intensity detector. Each data point should be accompanied by a timestamp so that time series correlation can be performed during subsequent analysis. When establishing a preliminary correspondence between the error value and the environmental factors, it is necessary to construct an environmental factor matrix to store the data of the error value and each environmental factor in the form of data points. For example, record the measurement errors at different temperature and humidity levels, and set the error value E i The corresponding environmental factor group F env,i , and then use the correlation analysis method to calculate the influence degree of each environmental factor on the error. Specifically, the linear correlation degree between the error and a single environmental factor can be calculated through the Pearson correlation coefficient, that is:

[0077]

[0078] where, R env,i represents the correlation coefficient of the environmental factor env on the error, E i is the error value, F env,i is the value of this environmental factor, and are the means of the error value and the environmental factor respectively. By calculating the correlation coefficients of multiple environmental factors, select the factors with |R env,i | greater than a certain threshold R thresh as the highly correlated factors. Suppose R thresh = 0.7. If the temperature factor R temp = 0.85, the humidity factor R hum = 0.76, and the electromagnetic interference factor R emi = 0.45, then only temperature and humidity enter the set of environmental factors highly correlated with the error, as shown in Table 1:

[0079] Table 1 Environmental factors highly correlated with the error

[0080]

[0081] As shown in Table 1, finally, temperature and humidity are selected as the environmental factors highly correlated with the error. This result indicates that temperature and humidity are the main environmental factors affecting the measurement error. When calculating the trend change of the error subsequently, the changes of these two factors need to be focused on.

[0082] S102: Based on the environmental factors with high error correlation, calculate the trend of the error under the change of differential environmental factors. Use the error change rate as a measurement index, calculate the change rate of the error on multiple environmental factors, and screen out the error sources whose error rates exceed the threshold to obtain the error sources exceeding the threshold.

[0083] Assume that the error value changes with time as E t , and the environmental factor changes with time as F env,t , and the error change rate is calculated as follows:

[0084]

[0085] For the temperature factor, assume that the measurement error of a certain device is 0.05V at 25°C and 0.08V at 30°C. Then the error change rate:

[0086]

[0087] If the set threshold is 0.005V / °C, then the temperature factor exceeds the threshold and is recorded as an error source exceeding the threshold. Similarly, for humidity calculation, if the humidity increases from 50% to 60% and the error changes from 0.05V to 0.065V, then:

[0088]

[0089] Assume that the error change rate threshold is 0.002V / %, then the humidity does not exceed the threshold. The final result is that the temperature is the error source exceeding the threshold. This result indicates that the influence of temperature on the error is more significant than that of humidity. When analyzing environmental factors, the change of temperature should be focused on to reduce the interference of error sources.

[0090] S103: Based on the error sources exceeding the threshold, perform normalization processing on multiple environmental factors, calculate the contribution rate of the normalized environmental factors to the error, and use the formula:

[0091]

[0092] Perform operations to obtain the error influence factor allocation value;

[0093] Among them, F env represents the error influence factor allocation value of the environmental factor env on the error, E src represents the error value of the error source src, NF env,src represents the normalized influence factor of the environmental factor env at the error source src, N sc represents the total number of error sources.

[0094] The formula is as follows:

[0095]

[0096] Assume the temperature data range is from 20°C to 40°C, and the current temperature is 30°C, then:

[0097]

[0098] After normalization, calculate the contribution rate F of environmental factors to the error env :

[0099]

[0100] Assume the device measurement error values are E1 = 0.05V and E2 = 0.08V respectively, and their corresponding normalized temperatures are NF temp,1 =

[0101] 0.5, NF temp,2 = 0.7, then calculate:

[0102]

[0103] This result shows that the assigned value of the influence factor of temperature on the error is 0.063, which means that in the error source analysis, the influence of the temperature factor on the error accounts for about 6.3%. This value can be used for error correction calculation and compensated or optimized in subsequent steps to reduce the influence of measurement error.

[0104] Please refer to Figure 3 , the specific steps for obtaining the dynamic error weight adjustment value are as follows:

[0105] S201: Call the assigned value of the error influence factor, monitor the change rate of the measurement error of the on-line instrument, calculate the ratio change amount between the current measurement error value and the previous error value, and obtain the error rate change value. At the same time, compare the error rate change value with the set error change threshold to determine whether the current error state exceeds the threshold range and obtain the error state determination result;

[0106] First, the initial value of the error influence factor needs to be obtained. This value can be obtained by statistically analyzing the historical error data of multiple measurement devices under the same environmental conditions. Normalize the error change values of multiple measurement points within a certain time range, and calculate the error influence factor of each device using weighted average. The specific calculation method can be expressed as:

[0107]

[0108] Among them, W init is the initial error influence factor, E i is the historical error value of the i-th measurement device, P i is the proportion weight of the i-th measurement device in the historical data, and N is the total number of measurement devices.

[0109] After obtaining the initial error influence factor, monitor the real-time measurement error change rate of the on-line instrument. The real-time error can be calculated by the deviation of the measurement signal. For example, the measurement error of a temperature sensor can be calculated by:

[0110] E cur =|T meas -T true |;

[0111] Among them, E cur is the current error value, T meas is the measured temperature of the instrument, T true is the standard reference temperature. The calculation of the error change rate is based on:

[0112]

[0113] Among them, R E represents the error change rate, and E prev is the error value at the previous moment. If the error change rate R E exceeds the set threshold T E , it indicates that the current error has exceeded the acceptable range. For example, if the set threshold is 0.05 (i.e., 5%), and the error at a certain measurement point changes from 2.0 °C to 2.2 °C, then calculate:

[0114]

[0115] This value is greater than the threshold 0.05. Therefore, it is determined that the current error state exceeds the threshold, and the error state determination result "exceeding the limit" is obtained. This result indicates that the error of the current measuring instrument fluctuates greatly, and subsequent steps are required to adjust the error weight to ensure more accurate error assessment.

[0116] S202: Based on the error state determination result, calculate the ratio change between the error change rate and the current error weight ratio, adjust the error weight allocation value, and obtain the adjusted error weight ratio through calculation;

[0117] First, the current error weight W old needs to be obtained, and its value can be calculated from the historical measurement error distribution, such as:

[0118]

[0119] Among them, E hist represents the historical error data set. After calculating the error change rate R E , further calculate the error weight adjustment ratio:

[0120]

[0121] The adjusted error weight ratio W new is calculated as:

[0122] W new = W old + α·R W ;

[0123] Where α is an adjustment coefficient, and its value range is generally set between 0.1 and 0.5. For example, if W old = 0.2, the error change rate R E = 0.1, and α = 0.3, then:

[0124]

[0125] W new = 0.2 + 0.3×0.5 = 0.35;

[0126] The adjusted error weight ratio is 0.35. This result indicates that due to the relatively high current error change rate, the error weight increases, making the error weight allocation more inclined to the latest measurement data, providing a new reference basis for subsequent error weight adjustment.

[0127] S203: Call the adjusted error weight ratio, and based on the current measurement error data, use the formula:

[0128]

[0129] Calculate to obtain the dynamic error weight adjustment value;

[0130] Where E DWA represents the dynamic error weight adjustment value, W new represents the adjusted error weight ratio, W old represents the error weight ratio before adjustment, ΔE mes represents the error change amount at the mes-th time, N mes represents the number of error measurements, T E represents the error change trend value, T W represents the error weight smoothing coefficient.

[0131] Formula:

[0132]

[0133] Suppose an instrument obtains error change amounts in 5 measurements:

[0134] ΔE mes = {0.05, 0.07, 0.09, 0.06, 0.08};

[0135] The number of measurements N mes = 5, calculate:

[0136]

[0137] Set W new = 0.35, W old = 0.2, T E = 0.05, T W = 0.1, then:

[0138]

[0139] This result indicates that the current error weight adjustment value is -0.375, which means that the error weight needs to be reduced so that the system can respond more sensitively to error changes. This value directly affects the subsequent error weight allocation strategy. When E DWA is negative, it means that the system needs to reduce the weight assignment to the current measurement data to make the error estimation more stable, thereby optimizing the subsequent measurement accuracy. Further organizing this value can deduce the next adjustment strategy, such as adjusting the smoothing coefficient T W or resetting the error weight ratio to ensure the optimal adjustment plan for measurement error.

[0140] Please refer to Figure 4 , and the specific steps for obtaining the error coupling correction matrix are as follows:

[0141] S301: Call the error influence factor assignment value, calculate the Pearson correlation of the error data of multiple measurement channels, screen the error pairs with Pearson correlation coefficients higher than the error correlation threshold, eliminate the low-correlation error pairs, and obtain the error correlation screening result;

[0142] First, obtain the error data sets of all channels. The error data of each channel consists of multiple sets of time series data. Classify these data according to the channel number, extract the error values within each channel, and calculate the Pearson correlation coefficients between all pairs of channels. The calculation method of the Pearson correlation coefficient is:

[0143]

[0144] where X i and Y i represent the error data points of two different channels respectively, and are the mean values of the error data of the corresponding channels, and n is the number of sampled data. After calculating the Pearson correlation coefficients of all error pairs, set the error correlation threshold T p , and screen out the error pairs with correlation coefficients P prn higher than T p , and record all the error pairs that meet the conditions into the error correlation screening result. For example, set T p= 0.6. If the Pearson correlation coefficient of two error channels A and B is 0.75, then this error pair is retained. If the Pearson correlation coefficient of channels C and D is 0.5, then this error pair is excluded. After screening, an error correlation screening result set is formed, which contains all error pairs that meet the correlation coefficient threshold.

[0145] S302: Based on the error correlation screening result, calculate the partial correlation coefficient of the screened error pairs, exclude the error pairs with a partial correlation coefficient lower than the interaction influence benchmark value, and obtain the partial correlation coefficient screening result;

[0146] Calculate the partial correlation coefficient for the screened error pairs. The calculation of the partial correlation coefficient is to measure whether a pair of errors still maintains a high correlation after removing the influence of other errors. The calculation method is as follows:

[0147]

[0148] Among them, P prn is the Pearson correlation coefficient screened above, P xy,z represents the conditional correlation coefficient under the condition of controlling variable z, P xz and P yz represent the correlation coefficients between error channels x and z, y and z respectively. According to the calculation result of the partial correlation coefficient, set the interaction influence benchmark value T r , if the partial correlation coefficient R par of the error pair is lower than T r , then this error pair is excluded. For example, set T r = 0.4. If the R par calculated for the error pair (A, B) is 0.5, then this error pair is retained. If the R par calculated for (C, D) is 0.35, then the (C, D) error pair is excluded. After screening, retain the error pairs with a partial correlation coefficient greater than or equal to the benchmark value to form the partial correlation coefficient screening result.

[0149] S303: Based on the partial correlation coefficient screening result, use multiple regression to calculate the interaction influence factor, using the formula:

[0150]

[0151] Obtain the calculation result of the interaction influence factor;

[0152] Among them, I c represents the calculated value of the interaction influence factor, P prn represents the Pearson correlation coefficient of the prnth group of error pairs, EIF prn represents the error influence factor of the prnth group of error pairs, R par represents the partial correlation coefficient of the parth group of error pairs, A parRepresents the normalized error influence factor of the par-th group of error pairs, B tot Represents the total error influence factor of the tot-th group of error pairs, and PCT represents the total number of error pairs.

[0153] The formula is as follows:

[0154]

[0155] Where, I c Represents the calculated value of the interaction influence factor, P prn Is the Pearson correlation coefficient of the prn-th group of error pairs, EIF prn Represents the error influence factor of the corresponding error pair, R par Is the partial correlation coefficient, A par Is the normalized error influence factor, B tot Is the total error influence factor. Set the total number of error pairs PCT = 10, and the specific calculation is as follows:

[0156]

[0157] Calculate the first summation term:

[0158]

[0159] Calculate the second summation term:

[0160]

[0161] Set B tot The total amount is 4.5:

[0162]

[0163] Finally, calculate the interaction influence factor:

[0164] I c = 2.701 + 0.321 = 3.022;

[0165] This result shows that the calculated interaction influence factor I c = 3.022. Combining with the set interval of the error influence factor, it can be further used for the construction of the error correction matrix.

[0166] Please refer to Figure 5 , and the specific steps for obtaining the error optimization calculation value are as follows:

[0167] S401: Call the error coupling correction matrix to separate the influences of multiple error sources, obtain the influence amount of multiple error sources on the system error, calculate the contribution rate of multiple error sources to the system error, and at the same time sort the error sources according to the contribution rate, screen the error source with the highest contribution rate, and obtain the error source contribution ranking result;

[0168] First, construct an influence matrix that includes multiple error sources. This matrix is composed of the influence data of multiple error sources on the overall system error. Specifically, first obtain the data of multiple error sources during system operation, including measurement errors, environmental errors, equipment errors, etc. Collect and store the real-time data of these error sources to form an error data set. Subsequently, according to the error performance of each error source under different working conditions, calculate its influence factor on the total system error. For example, if there are five main error sources in the system, and the corresponding error values are E1 = 2.1, E2 = 3.5, E3 = 1.8, E4 = 4.2, E5 = 2.9 respectively, then calculate the mean error of each error source as follows:

[0169]

[0170] Next, calculate the variance of the error sources to measure the dispersion degree of each error source. The calculation formula is:

[0171]

[0172] Substitute the data:

[0173]

[0174] Subsequently, calculate the contribution rate CC of each error source to the system error src , with the formula:

[0175]

[0176] Substitute the data:

[0177]

[0178] According to the calculation results, the error source with the highest error contribution rate is E4. Screen out this error source as the main correction target, and obtain the error source contribution ranking result as: E4, E3, E1, E2, E5.

[0179] This result indicates that the error source E4 has the highest error contribution degree. Therefore, in the subsequent error correction process, it should be adjusted preferentially to reduce the overall error.

[0180] S402: Based on the error source contribution ranking result, calculate the error adjustment amount of multiple error sources. Through error adjustment calculation, correct the system error after adjusting the error sources. Use the formula:

[0181]

[0182] Perform operations to obtain the correction amplitude of the error source and adjust the system error value to obtain the system corrected error value;

[0183] Among them, SCE represents the system correction error value, and CC src represents the contribution coefficient of the src-th error source, and E src represents the error value of the src-th error source, represents the mean error of all error sources, and V src represents the error variance of the src-th error source, and N sc represents the total number of error sources;

[0184] The formula is as follows:

[0185]

[0186] Substitute the data:

[0187]

[0188] The calculated system correction error value SCE = 0.564. Adjust the system error value to reduce the system error and obtain the system correction error value.

[0189] This result indicates that the calculated SCE represents the current error correction amplitude. The larger this value, the greater the error adjustment space. In the next correction process, it is necessary to adjust the error sources based on this value to minimize the error.

[0190] S403: Invoke the system correction error value, optimize the system error weight based on the dynamic error weight adjustment value, and perform an optimized adjustment on the corrected error value to obtain the error optimized calculation value.

[0191] Optimize the system error weight based on the dynamic error weight adjustment value, and perform a secondary correction using the error correction factor λ. The calculation formula of λ:

[0192]

[0193] Substitute the data:

[0194]

[0195] According to the error correction factor λ, adjust the correction amplitude of each error source to the system error. The adjustment formula is:

[0196]

[0197] For the error source E4:

[0198] E adj,4 = 4.2 - 0.565×(4.2 - 2.9) = 4.2 - 0.7345 = 3.465;

[0199] For the error source E3:

[0200] E adj,3 = 1.8 - 0.565×(1.8 - 2.9) = 1.8 + 0.565×1.1 = 1.8 + 0.6215 = 2.4215;

[0201] And so on, correct all error sources to obtain the final optimized error calculation value, as shown in Table 2:

[0202] Table 2 Optimized Error Calculation Value Table

[0203]

[0204]

[0205] As shown in Table 2, after dynamic error weight optimization, the error values of each error source tend to the mean to reach the optimized error calculation value.

[0206] This result shows that after error adjustment, the error values approach the mean, and the error distribution is more balanced, indicating that the correction strategy is effective and can be further optimized and adjusted in subsequent steps to further reduce the system error.

[0207] Please refer to Figure 6 , the steps to obtain the error correction stability value are specifically as follows:

[0208] S501: Call the optimized error calculation value, calculate the change amount of the error before and after correction, and at the same time calculate based on the error value before correction, the error value after correction and the optimized calculation value, using the formula:

[0209]

[0210] Calculate the error change correction amount;

[0211] Among them, ECC represents the error change correction amount, E prev represents the error value before correction, E corr represents the error value after correction, E EOC represents the optimized error calculation value, and the log operation enhances the calculation adjustability;

[0212] First, extract the current measurement error data, including the error value E prev before correction, the error value E corr after correction, and the optimized error calculation value E EOC , calculate the error change correction amount based on these three data. During the calculation process, first obtain the absolute value of the error change |E prev - E corr |, then take the logarithm of the optimized calculation value E EOC and add 1 to enhance the calculation adjustability. Finally, use the formula:

[0213]

[0214] When performing specific calculations, assume that when the measurement system detects the error of the temperature sensor, the error value before correction is 1.5 °C, the error value after correction is 0.8 °C, and the error optimization calculation value is 2.3 °C based on the device adaptive error evaluation model. Then the calculation is as follows:

[0215]

[0216] The specific calculation steps are as follows:

[0217] Calculate the absolute error change value: |1.5 - 0.8| = 0.7;

[0218] Calculate the logarithmic term of the error optimization calculation value: log(2.3) ≈ 0.36;

[0219] Calculate the error change correction amount:

[0220] This result indicates that the error has decreased by 0.7 °C after error correction. However, after adjustment in combination with the error optimization calculation value, the calculated error change correction amount is 0.515. This value is used for subsequent error convergence determination, which reflects the effectiveness of error correction and also indicates that there is still a certain error space after error adjustment, and it is necessary to further determine whether the stability requirement is met.

[0221] S502: Call the error change correction amount, compare it with the error convergence threshold, and at the same time calculate the difference between the error change correction amount and the error convergence threshold, determine whether the error change correction amount is less than the error convergence threshold, and combine the change rate to perform error convergence determination to obtain the error convergence status flag;

[0222] For example, in an industrial temperature monitoring system, the error convergence threshold is set to 0.3 °C, indicating that when the change amplitude after error correction is lower than this value, the error is considered to have stabilized.

[0223] During the comparison process, calculate the difference between the error change correction amount ECC and the threshold T conv between them:

[0224] ΔECC = ECC - T conv ;

[0225] For the above example, set T conv = 0.3, and the calculation is as follows:

[0226] ΔECC = 0.515 - 0.3 = 0.215;

[0227] Judge whether ECC is less than the threshold:

[0228] If ECC < T conv , the error converges;

[0229] If ECC ≥ T conv , the error does not converge.

[0230] In the current calculation, ECC = 0.515 is higher than 0.3, so the error has not converged and further adjustment is needed. At the same time, to consider the error change rate, the error change rate is defined as:

[0231]

[0232] Assume the measurement interval Δt = 2 seconds, then calculate:

[0233]

[0234] When the error change rate R ECC is lower than the set rate threshold (assume 0.2 °C / s), it can be determined that the error is stable, otherwise continue to adjust the error. In the current calculation, R ECC = 0.35 is higher than the set threshold 0.2, so the error is still not stable, and obtain the error convergence status flag.

[0235] This result shows that the current error change correction amount is still large, and the error adjustment still needs to be further optimized. The error has not reached a stable state, so it is necessary to additionally adjust the measurement model or compensation parameters to ensure further error convergence.

[0236] S503: Call the error convergence status flag, judge the error change trend based on the convergence state parameter. If the error convergence status flag meets the stable condition, determine the error correction result and obtain the error correction stable value.

[0237] During the error adjustment process, the system continuously detects the error change situation and uses a sliding window to calculate the error convergence trend. For example, set the window size to 5 measurement periods. If the error change correction amount ECC of 5 consecutive measurements is lower than the set threshold T conv , it is considered that the error correction has stabilized.

[0238] Set the error change correction amount within 5 measurement periods as follows:

[0239] Table 3 Error change correction amount trend table

[0240]

[0241] As shown in Table 3, the error change correction amount shows a downward trend, and the ECC of the 5th measurement is 0.358, which is still higher than the threshold 0.3, so the error is still not stable.

[0242] When the ECC in subsequent measurement cycles further decreases and is less than the threshold for 5 consecutive times, for example:

[0243] ECC = {0.298, 0.290, 0.275, 0.260, 0.250};

[0244] Then it can be judged that the error correction is stable, and finally the error correction stable value E stable , and the calculation method is as follows:

[0245]

[0246] Where N is 5 consecutive measurement cycles, and substitute the data for calculation:

[0247]

[0248] This result indicates that the measurement error has become stable after multiple adjustments. The final error correction stable value is 0.884 °C, indicating that the error adjustment has entered a stable state, the error correction algorithm has effectively converged, and no further adjustment is required.

[0249] An on-line instrument automatic inspection and calibration system, which is used to execute the above on-line instrument automatic inspection and calibration method. The system includes:

[0250] The error impact factor analysis module obtains the error value, environmental temperature, humidity, and magnetic interference intensity in the real-time data, converts the environmental parameters into standard values, calculates the multi-dimensional influence ratio according to the error fluctuation range and the standard values, compares the environmental fluctuation correction influence ratio according to the measurement stability, screens the parameters corresponding to the error sources exceeding the threshold, calculates the parameter combination by weighted comparison, and generates the error impact factor allocation value;

[0251] The error dynamic weight adjustment module calls the error impact factor allocation value, monitors the change rate of the measurement error, compares the error change with the set threshold, and adjusts the weight parameters according to the ratio to generate the dynamic error weight adjustment value;

[0252] The error coupling correction calculation module calls the error impact factor allocation value, calculates the multi-channel Pearson coefficient, screens the error pairs with correlation exceeding the threshold, correlates according to the partial correlation comparison data, and calls the dynamic error weight adjustment value to correct the error impact factor to generate the error coupling correction matrix;

[0253] The error optimization adjustment module calls the error coupling correction matrix, compares the contribution rates of multiple error sources, screens the parameters with contributions exceeding the threshold, calculates the weight difference, and generates the error optimization calculation value;

[0254] The error correction stability determination module calls the error optimization calculation value, detects the error difference before and after correction, compares the difference with the convergence threshold, and judges stability according to the comparison result to generate the error correction stable value.

[0255] The above are only the preferred embodiments of the present invention, and do not impose other forms of limitation on the present invention. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An automatic inspection and calibration method for on-line instruments, characterized in that, It includes the following steps: S1: Obtain the error values and environmental impact factors in the real-time measurement data. The environmental impact factors include temperature, humidity, and electromagnetic interference. Calculate the error change trend in the environment, screen the error sources with error rates exceeding the threshold, normalize the calculation of the impact factors, and obtain the error impact factor allocation values; S2: Call the error impact factor allocation values, monitor the change rate of the measurement error of the online instrument, compare the error rate with the threshold, calculate the ratio of the change rate of the difference to the current error weight, and adjust the weight allocation according to the exponential smoothing method to obtain the dynamic error weight adjustment value; S3: Call the error impact factor allocation values, calculate the Pearson correlation of the error data of multiple measurement channels, screen the associated error pairs, calculate the partial correlation coefficient, eliminate the non-critical interaction terms, use multiple regression operations to calculate the interaction impact factor, call the dynamic error weight adjustment value to correct the error impact factor, and calculate the corrected error propagation impact to obtain the error coupling correction matrix; S4: Call the error coupling correction matrix, sort and screen the error sources with the highest contribution rate, calculate the error adjustment amount, call the dynamic error weight adjustment value to optimize the weight, and obtain the error optimization calculation value.

2. The online instrument automatic inspection and calibration method according to claim 1, characterized in that The error impact factor allocation values include error sources with error rates exceeding the threshold, normalized environmental impact factors, and error impact factor weights. The dynamic error weight adjustment values include error change rates, error weight ratios, and exponentially smoothed adjusted weights. The error coupling correction matrix includes key interaction error terms, partial correlation error calculation values, and multiple regression interaction impact factors. The error optimization calculation values include error sources with the highest contribution rate, error adjustment amounts, and optimized dynamic weights.

3. The online instrument automatic inspection and calibration method according to claim 2, wherein The specific steps for obtaining the error impact factor allocation values are as follows: S101: Obtain the error values in the real-time measurement data, and collect environmental impact factors, including temperature, humidity, and electromagnetic interference. Match the error values with the environmental impact factors, establish a preliminary correspondence between the errors and the environmental factors, calculate the correlation coefficients of multiple environmental factors on the errors, screen the environmental factors with higher correlations, and obtain the environmental factors with high error correlations; S102: Based on the environmental factors with high error correlations, calculate the trend of the errors under the change of different environmental factors. Use the error change rate as a measurement index, calculate the change rates of the errors on multiple environmental factors, and screen the error sources with error rates exceeding the threshold to obtain the error sources exceeding the threshold; S103: Based on the error sources exceeding the threshold, perform normalization processing on multiple environmental factors, calculate the contribution rates of the normalized environmental factors to the errors, and use the formula: Perform operations to obtain the error impact factor allocation values; Among them, F env represents the allocated value of the influence factor of the environmental factor env on the error, E src represents the error value of the error source src, NF env,src represents the normalized influence factor of the environmental factor env at the error source src, N sc represents the total number of error sources.

4. The online instrument automatic inspection and calibration method according to claim 3, characterized in that The specific steps for obtaining the dynamic error weight adjustment values are as follows: S201: Call the error impact factor allocation values, monitor the change rate of the measurement error of the online instrument, calculate the ratio change amount between the current measurement error value and the previous error value, and obtain the error rate change value. At the same time, compare the error rate change value with the set error change threshold to determine whether the current error state exceeds the threshold range, and obtain the error state determination result; S202: Based on the error state determination result, calculate the ratio change between the error change rate and the current error weight ratio, adjust the error weight allocation value, and obtain the adjusted error weight ratio through operation; S203: Invoke the adjusted error weight ratio, and based on the current measurement error data, use the formula: Obtain the dynamic error weight adjustment value through operation; Among them, E DWA represents the dynamic error weight adjustment value, W new represents the adjusted error weight ratio, W old represents the error weight ratio before adjustment, ΔE mes represents the error change amount at the mes-th time, N mes represents the number of error measurements, T E represents the error change trend value, T W represents the error weight smoothing coefficient.

5. The online instrument automatic inspection and calibration method according to claim 4, characterized in that The specific steps for obtaining the error coupling correction matrix are as follows: S301: Invoke the error influence factor allocation value, calculate the Pearson correlation of the error data of multiple measurement channels, screen out the error pairs with Pearson correlation coefficients higher than the error correlation threshold, eliminate the low-correlation error pairs, and obtain the error correlation screening result; S302: Based on the error correlation screening result, calculate the partial correlation coefficient of the screened error pairs, screen out the error pairs with partial correlation coefficients lower than the interaction influence reference value, and obtain the partial correlation coefficient screening result; S303: Based on the partial correlation coefficient screening result, calculate the interaction influence factor using multiple regression, and use the formula: Obtain the calculation result of the interaction influence factor; Among them, I c represents the calculated value of the interaction influence factor, P prn represents the Pearson correlation coefficient of the prn-th group of error pairs, EIF prn represents the error influence factor of the prn-th group of error pairs, R par represents the partial correlation coefficient of the par-th group of error pairs, A par represents the normalized error influence factor of the par-th group of error pairs, B tot represents the total amount of the error influence factor of the tot-th group of error pairs, and PCT represents the total number of error pairs.

6. The online instrument automatic inspection and calibration method according to claim 5, characterized in that, The specific steps for obtaining the error optimization calculation value are as follows: S401: Invoke the error coupling correction matrix, separate the influences of multiple error sources, obtain the influence amount of multiple error sources on the system error, calculate the contribution rate of multiple error sources to the system error, and at the same time sort the error sources according to the contribution rate, screen out the error source with the highest contribution rate, and obtain the error source contribution ranking result; S402: Based on the error source contribution ranking result, calculate the error adjustment amount of multiple error sources, correct the system error after adjusting the error source through error adjustment calculation, and use the formula: Obtain the correction amplitude of the error source through operation, and adjust the system error value to obtain the system corrected error value; Among them, SCE represents the system correction error value, CC src represents the contribution coefficient of the src-th error source, E src represents the error value of the src-th error source, represents the mean error of all error sources, V src represents the error variance of the src-th error source, N sc represents the total number of error sources; S403: Invoke the system corrected error value, optimize the system error weight based on the dynamic error weight adjustment value, and perform optimization adjustment on the corrected error value to obtain the error optimization calculation value.

7. The online instrument automatic inspection and calibration method according to claim 6, characterized in that, The method further includes: S5: Invoke the error optimization calculation value, calculate the error change amount before and after correction, compare the change amount with the threshold, judge the error convergence situation, and obtain the error correction stability value; The error correction stability value includes the error change amount before and after correction, the error convergence determination result, and the error correction value.

8. The online instrument automatic inspection and calibration method according to claim 7, characterized in that, The specific steps for obtaining the error correction stability value are as follows: S501: Invoke the error optimization calculation value, calculate the change amount of the error before and after correction, and at the same time calculate based on the error value before correction, the error value after correction, and the optimization calculation value, using the formula: Calculate the error change correction amount; Among them, ECC represents the error change correction amount, E prev represents the error value before correction, E corr represents the error value after correction, E EOC represents the error optimization calculation value, and the log operation enhances the calculation adjustability; S502: Invoke the error change correction amount, compare it with the error convergence threshold, and at the same time calculate the difference between the error change correction amount and the error convergence threshold, judge whether the error change correction amount is less than the error convergence threshold, and combine the change rate to perform error convergence determination to obtain the error convergence status flag; S503: Invoke the error convergence status flag, judge the error change trend based on the convergence state parameter. If the error convergence status flag meets the stable condition, determine the error correction result and obtain the error correction stability value.

9. An online instrument automatic inspection and calibration system, characterized in that The online instrument automatic inspection and calibration method according to any one of claims 1-8, the system comprising: The error influence factor analysis module obtains the error value, ambient temperature, humidity, and magnetic interference intensity in the real-time data, converts the environmental parameters into standard values, calculates the multi-dimensional influence ratio based on the error fluctuation range and the standard values, compares the environmental fluctuation correction influence ratio according to the measurement stability, screens the parameters corresponding to the error sources with errors exceeding the threshold, calculates the parameter combination using weighted comparison, and generates the error influence factor distribution value; The error dynamic weight adjustment module calls the error influence factor distribution value, monitors the change rate of the measurement error, compares the error change with the set threshold, adjusts the weight parameter according to the ratio, and generates the dynamic error weight adjustment value; The error coupling correction calculation module calls the error influence factor distribution value, calculates the multi-channel Pearson coefficient, screens the error pairs with correlations exceeding the threshold, correlates according to the partial correlation comparison data, calls the dynamic error weight adjustment value to correct the error influence factor, and generates the error coupling correction matrix; The error optimization adjustment module calls the error coupling correction matrix, compares the contribution rates of multiple error sources, screens the parameters with contributions exceeding the threshold, calculates the weight difference, and generates the error optimization calculation value; The error correction stability determination module calls the error optimization calculation value, detects the error difference before and after correction, compares the difference with the convergence threshold, judges stability according to the comparison result, and generates the error correction stability value.

Images