Temperature on-line metering system and dynamic uncertainty evaluation method thereof

Abstract

This invention discloses an online temperature measurement system and its dynamic uncertainty assessment method. The system dynamically fuses the posterior distribution of the (n-1)th iteration and the likelihood function constructed from the temperature measurement data collected in the nth iteration to form the posterior distribution of the nth iteration. Then, after sampling using the Metropolis-Hastings algorithm, combustion and dilution processes are performed to obtain the resulting uncertainty assessment model. This model enables real-time dynamic assessment of the uncertainty of the temperature measurement data collected in the nth iteration. This invention can dynamically assess the uncertainty of the online temperature measurement system, ensuring the accuracy and reliability of the measured values.

Description

Technical Field

[0001] This invention relates to the testing or calibration of online temperature measurement systems, specifically to an online temperature measurement system and a method for dynamically evaluating its uncertainty. Background Technology

[0002] Platinum resistance temperature sensors, with their high accuracy, excellent stability, and wide range of applications, have become the preferred temperature monitoring sensor for harsh environments. Today, as the manufacturing industry moves towards higher precision and intelligence, precision measurement technology has become crucial for improving product quality and production efficiency. Traditional platinum resistance uncertainty assessments can only provide an uncertain static value, which is insufficient to meet the reliability requirements of real-time monitoring values ​​in large-scale industrial applications. To ensure the reliability of the temperature parameters monitored by platinum resistance sensors and avoid the risk of temperature parameter errors, accurate measurement during the temperature monitoring process is a key factor in ensuring the accuracy, reliability, and traceability of platinum resistance sensor measurements.

[0003] In complex operating conditions, there may be dangerous factors that threaten the safety of measurement personnel, making it difficult to monitor the measurement parameters of platinum resistance thermometers in real time. Existing remote measurement systems, based on bus technologies such as GPIB and SCPI and intelligent measurement and control technologies, are limited by the physical constraints of their total cabling. The physical distance of transmission is limited, and the total cabling is susceptible to damage from environmental disasters, which greatly increases the maintenance cost. In remote measurement systems, ensuring the reliability of measured values ​​has always been a challenge. Existing remote measurement systems cannot make a reasonable assessment of the reliability of measured values, and the progress of measurement is not guaranteed. Summary of the Invention

[0004] One of the technical problems to be solved by this invention is to provide a dynamic uncertainty assessment method for an online temperature measurement system, so as to solve the problem of low reliability of existing dynamic uncertainty assessment methods.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0006] like Figure 1 As shown, a method for dynamic evaluation of uncertainty in an online temperature measurement system is characterized by comprising:

[0007] Step S1, Preparation Stage: Based on historical temperature measurement data from the online temperature measurement system, construct a priori distribution for evaluating its uncertainty. Among them, the prior distribution Based on temperature measurement data It is represented as the inverse gamma conjugate distribution of the true temperature value μ and the temperature variance φ; In order to collect temperature measurement data At that time, the temperature measurement values ​​output by the online temperature measurement system are from the 1st to the mth.

[0008] Step S2, Dynamic Evaluation Phase: Whenever a new set of temperature measurement data is acquired from the online temperature measurement system in real time... At this point, an iteration is performed to obtain the posterior distribution of the (n-1)th iteration based on the Bayesian information fusion principle. The likelihood function of the nth iteration The posterior distribution of the nth iteration is formed by fusion. It still takes the form of an inverse gamma conjugate distribution:

[0009] ;

[0010] In the formula, the likelihood function of the nth iteration Temperature measurement data collected during the nth iteration Constructed, and in the n=1th iteration For the prior distribution ;

[0011] Step S3: In each iteration, the posterior distribution obtained from the nth iteration is processed using the Metropolis-Hastings algorithm. Sample sequences were obtained by sampling from the middle. In the formula, the t-th sample The temperature sample value obtained from the t-th sampling. and temperature variance sample values The set, where N is the total number of samplings, typically on the order of hundreds of thousands;

[0012] Furthermore, for the sample sequence The samples were burned and then diluted sequentially to discard the original sequences. The first B samples Then from the remaining samples One sample is drawn every k samples, and the resulting new sample sequence is used as the uncertainty assessment model for the nth iteration. B and k are both preset values.

[0013] Step S4: Using the uncertainty assessment model of the nth iteration, evaluate the temperature measurement data collected during the nth iteration. Uncertainty assessment is performed.

[0014] Therefore, this invention will use the posterior distribution of the (n-1)th iteration. and the temperature measurement data collected during the nth iteration The constructed likelihood function The dynamic fusion forms the posterior distribution of the nth iteration. After sampling using the Metropolis-Hastings algorithm, combustion and dilution processes are performed. The resulting uncertainty assessment model is then used to evaluate the temperature measurement data collected during the nth iteration. The uncertainty is dynamically assessed in real time;

[0015] Among them, due to the posterior distribution and uncertainty assessment model changing with each temperature measurement data The data is collected and iteratively corrected, thus continuously improving the accuracy of uncertainty assessment;

[0016] The Metropolis-Hastings algorithm is used for sampling, and the new sample sequence obtained is used as the uncertainty assessment model. The burning process removes the first B samples that may not converge to a stable distribution to eliminate the influence of the initial value. The dilution process further reduces autocorrelation, so that the uncertainty assessment model meets the independence principle.

[0017] In summary, this invention can dynamically assess the uncertainty of an online temperature measurement system, ensuring the accuracy and reliability of the measured values.

[0018] Preferably, step S1 specifically includes:

[0019] The historical temperature measurement data is obtained from activities in the metrology laboratory using an online temperature measurement system.

[0020] The true temperature value μ and the temperature variance φ are set as the parameters to be estimated, and the conditions are determined by the prior distribution. and marginal prior distribution The prior distribution is constructed in the form of a joint probability distribution. :

[0021] ;

[0022] ;

[0023] ;

[0024] In the formula, the prior distribution The hyperparameters are denoted as:

[0025] The best estimate of the historical temperature, μ0, is equal to the average value of the historical temperature measurements.

[0026] The scaling parameters α0 and β0 are set based on historical experience. Their values ​​jointly determine the confidence strength and central tendency of the prior distribution. They are set as follows: , In the formula, v is the degree of freedom of the prior distribution, and its value is generally set by historical experience. For example, v=2 means that only two data points in the historical temperature measurement data are reliable data. The variance of the historical temperature measurement data;

[0027] Weight k0 is a preset value for the weight of the prior distribution, used to adjust the influence of the prior distribution on the subsequent iterative updates of the posterior distribution. It is generally set to k0=1, indicating that the prior distribution and the temperature measurement data updated in each measurement are related. Assign the same weight;

[0028] Conditional prior distribution This represents the distribution of the true temperature value μ given the temperature variance φ and weight k0, describing the uncertainty of the true temperature value μ relative to the temperature variance φ.

[0029] Marginal prior distribution This represents the independent distribution of the temperature variance φ, which reflects the prior uncertainty of the temperature variance φ. The formula for calculating the Gamma function is defined in integral form: It is used to ensure that the probability density function of the prior distribution is normalized.

[0030] Therefore, this invention uses temperature measurement data obtained by an online temperature measurement system in a metrology laboratory as the historical temperature measurement data, and is based on a conditional prior distribution. and marginal prior distribution The prior distribution is constructed in the form of a joint probability distribution. This reduces the subjectivity of dynamic uncertainty assessment.

[0031] In step S2, the posterior distribution of the nth iteration The hyperparameters are updated as follows:

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] In the formula, The weights for the nth iteration are... This is the best estimate of the historical temperature in the nth iteration. The temperature measurement data collected during the nth iteration The average value, i.e. , Let α0 be the scaling parameter for the nth iteration. Let β0 be the scaling parameter for the nth iteration. The variance of the historical temperature measurement data;

[0037] Therefore, the posterior distribution of the nth iteration The updated hyperparameters can be represented as follows:

[0038] ;

[0039] In the formula, This indicates that the degree of freedom is 2a. n The t-distribution, As a location parameter of the t-distribution As a scaling parameter of the t-distribution.

[0040] Specifically, step S4 includes:

[0041] In the nth iteration, the following uncertainty assessment results are output: ;

[0042] ;

[0043] ;

[0044] In the formula, This represents the optimal temperature measurement estimate at the nth iteration, which indicates the temperature measurement data collected at the nth iteration. The final measurement results Let be the uncertainty at the nth iteration.

[0045] The second technical problem to be solved by this invention is to provide an online temperature measurement system.

[0046] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0047] like Figure 2 and Figure 3 As shown, an online temperature measurement system includes: a temperature acquisition subsystem, used to convert temperature signals into electrical signals, and then output digital signals through modulation amplification and analog-to-digital conversion, serving as the temperature measurement values ​​output by the online temperature measurement system, wherein m temperature measurement values ​​constitute the temperature measurement data. ;

[0048] Its features include: an edge computing subsystem, a cloud subsystem, and a web client;

[0049] The edge computing subsystem connects to the temperature acquisition subsystem via bus communication, and the edge computing subsystem controls the temperature acquisition subsystem through the VISA software interface API. The VISA virtual instrument framework is used to abstract the instrument control behavior into a set of computer-recognizable program commands. The logical control of the temperature acquisition subsystem is based on the SCPI command set. After the edge computing subsystem encapsulates the SCPI program control commands, they are sent to the SCPI command interpreter of the temperature acquisition subsystem through the physical connection of bus technology to complete the program control of the temperature acquisition subsystem.

[0050] The edge computing subsystem communicates with the cloud subsystem based on the MQTT protocol. It is equipped with an MQTT client to enable the uploading of measurement results and the receiving of cloud measurement and control commands.

[0051] The web client communicates with the cloud subsystem via the network. The web client is developed based on the HTML language and uses the HTML language to construct web page content and structure to realize remote display of measurement results and deployment of remote measurement tasks.

[0052] This enables the edge computing subsystem to receive temperature measurement data output by the temperature acquisition subsystem. Based on the aforementioned dynamic uncertainty assessment method, the received temperature measurement data is processed. Uncertainty assessment is performed, and the results are uploaded to the cloud subsystem for web client access; wherein, the edge computing subsystem will use the posterior distribution obtained in step S2. Upload to the cloud subsystem to utilize the computing power of the cloud subsystem to perform the calculation in step S3.

[0053] The cloud server in the cloud subsystem includes an MQTT server (Broker) and a web backend server.

[0054] The MQTT server responds to and distributes system commands in the cloud, is responsible for the efficient relay of MQTT messages, provides message routing for network clients, and distributes these messages to all MQTT clients that have subscribed to the corresponding topics, ensuring that data and programmable commands between edge machines and front-end web MQTT clients are accurately received and properly distributed.

[0055] The web backend server establishes HTTP request-response links, based on Microsoft's IIS server software, and is responsible for managing network client interactions on the measurement and control cloud platform, thereby enabling remote access and control.

[0056] Therefore, to address the issue of remote deployment of testing tasks, cloud technology and cloud computing are combined to ultimately enable remote metering task deployment via a web browser.

[0057] Therefore, the online temperature measurement system of the present invention, combined with cloud technology and Internet of Things+ technology, realizes real-time remote transmission and monitoring of temperature measurement data, greatly expands the physical range of temperature testing task deployment, solves the problem that temperature measurement is limited and cannot be monitored in real time due to personnel safety threats and transmission physical distance under complex working conditions, and can meet the needs of large-scale industrial applications for real-time monitoring of temperature parameters and dynamic evaluation of uncertainty.

[0058] Among them, the computing power of the cloud can be used to control multiple test tasks in parallel, solving the problem of the large amount of traditional test data that is difficult to process;

[0059] Among them, the edge computing subsystem, in conjunction with the cloud subsystem, can automate the testing task process;

[0060] Furthermore, the test results are processed and stored in the cloud, which improves the traceability of the testing process.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] First, this invention will use the posterior distribution of the (n-1)th iteration. and the temperature measurement data collected during the nth iteration The constructed likelihood function The dynamic fusion forms the posterior distribution of the nth iteration. After sampling using the Metropolis-Hastings algorithm, combustion and dilution processes are performed. The resulting uncertainty assessment model is then used to evaluate the temperature measurement data collected during the nth iteration. The uncertainty is dynamically assessed in real time;

[0063] Among them, due to the posterior distribution and uncertainty assessment model changing with each temperature measurement data The data is collected and iteratively corrected, thus continuously improving the accuracy of uncertainty assessment;

[0064] The Metropolis-Hastings algorithm is used for sampling, and the new sample sequence obtained is used as the uncertainty assessment model. The burning process removes the first B samples that may not converge to a stable distribution to eliminate the influence of the initial value. The dilution process further reduces autocorrelation, so that the uncertainty assessment model meets the independence principle.

[0065] In summary, this invention can dynamically assess the uncertainty of an online temperature measurement system, ensuring the accuracy and reliability of the measured values.

[0066] Second, this invention uses temperature measurement data obtained by an online temperature measurement system in a metrology laboratory as the historical temperature measurement data, and is based on a conditional prior distribution. and marginal prior distribution The prior distribution is constructed in the form of a joint probability distribution. This reduces the subjectivity of dynamic uncertainty assessment.

[0067] Third, the online temperature measurement system of the present invention, combined with cloud technology and Internet of Things+ technology, realizes real-time remote transmission and monitoring of temperature measurement data, greatly expands the physical range of temperature testing task deployment, solves the problem that temperature measurement is limited by personnel safety threats and transmission physical distance under complex working conditions and cannot be monitored in real time, and can meet the needs of large-scale industrial applications for real-time monitoring of temperature parameters and dynamic evaluation of uncertainty.

[0068] Among them, the computing power of the cloud can be used to control multiple test tasks in parallel, solving the problem of the large amount of traditional test data that is difficult to process;

[0069] Among them, the edge computing subsystem, in conjunction with the cloud subsystem, can automate the testing task process;

[0070] Furthermore, the test results are processed and stored in the cloud, which improves the traceability of the testing process. Attached Figure Description

[0071] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments:

[0072] Figure 1 This is a flowchart of the uncertainty dynamic evaluation method of the present invention;

[0073] Figure 2 This is a schematic diagram of the online temperature measurement system of the present invention;

[0074] Figure 3 This is a flowchart illustrating the operation of the online temperature measurement system of the present invention. Detailed Implementation

[0075] The present invention will now be described in detail with reference to the embodiments and accompanying drawings to help those skilled in the art better understand the inventive concept of the present invention. However, the scope of protection of the claims of the present invention is not limited to the following embodiments. For those skilled in the art, all other embodiments obtained without creative effort without departing from the inventive concept of the present invention are within the scope of protection of the present invention.

[0076] Example 1

[0077] like Figure 1As shown, this invention discloses a method for dynamic evaluation of uncertainty in an online temperature measurement system, comprising:

[0078] Step S1, Preparation Stage: Based on historical temperature measurement data from the online temperature measurement system, construct a priori distribution for evaluating its uncertainty. Among them, the prior distribution Based on temperature measurement data It is represented as the inverse gamma conjugate distribution of the true temperature value μ and the temperature variance φ; In order to collect temperature measurement data At that time, the temperature measurement values ​​output by the online temperature measurement system are from the 1st to the mth.

[0079] Step S1 specifically includes:

[0080] The historical temperature measurement data is obtained from activities in the metrology laboratory using an online temperature measurement system.

[0081] The true temperature value μ and the temperature variance φ are set as the parameters to be estimated, and the conditions are determined by the prior distribution. and marginal prior distribution The prior distribution is constructed in the form of a joint probability distribution. :

[0082] ;

[0083] ;

[0084] ;

[0085] In the formula, the prior distribution The hyperparameters are denoted as:

[0086] The best estimate of the historical temperature, μ0, is equal to the average value of the historical temperature measurements.

[0087] The scaling parameters α0 and β0 are set based on historical experience. Their values ​​jointly determine the confidence strength and central tendency of the prior distribution. They are set as follows: , In the formula, v is the degree of freedom of the prior distribution, and its value is generally set by historical experience. For example, v=2 means that only two data points in the historical temperature measurement data are reliable data. The variance of the historical temperature measurement data;

[0088] Weight k0 is a preset value for the weight of the prior distribution, used to adjust the influence of the prior distribution on the subsequent iterative updates of the posterior distribution. It is generally set to k0=1, indicating that the prior distribution and the temperature measurement data updated in each measurement are related. Assign the same weight;

[0089] Conditional prior distribution This represents the distribution of the true temperature value μ given the temperature variance φ and weight k0, describing the uncertainty of the true temperature value μ relative to the temperature variance φ.

[0090] Marginal prior distribution This represents the independent distribution of the temperature variance φ, which reflects the prior uncertainty of the temperature variance φ. The formula for calculating the Gamma function is defined in integral form: It is used to ensure that the probability density function of the prior distribution is normalized.

[0091] Therefore, this invention uses temperature measurement data obtained by an online temperature measurement system in a metrology laboratory as the historical temperature measurement data, and is based on a conditional prior distribution. and marginal prior distribution The prior distribution is constructed in the form of a joint probability distribution. This reduces the subjectivity of dynamic uncertainty assessment.

[0092] Step S2, Dynamic Evaluation Phase: Whenever a new set of temperature measurement data is acquired from the online temperature measurement system in real time... At this point, an iteration is performed to obtain the posterior distribution of the (n-1)th iteration based on the Bayesian information fusion principle. The likelihood function of the nth iteration The posterior distribution of the nth iteration is formed by fusion. It still takes the form of an inverse gamma conjugate distribution:

[0093] ;

[0094] In the formula, the likelihood function of the nth iteration Temperature measurement data collected during the nth iteration Constructed, and in the n=1th iteration For the prior distribution ;

[0095] In step S2, the posterior distribution of the nth iteration The hyperparameters are updated as follows:

[0096] ;

[0097] ;

[0098] ;

[0099] ;

[0100] In the formula, The weights for the nth iteration are... This is the best estimate of the historical temperature in the nth iteration. The temperature measurement data collected during the nth iteration The average value, i.e. , Let α0 be the scaling parameter for the nth iteration. Let β0 be the scaling parameter for the nth iteration. The variance of the historical temperature measurement data;

[0101] Therefore, the posterior distribution of the nth iteration The updated hyperparameters can be represented as follows:

[0102] ;

[0103] In the formula, This indicates that the degree of freedom is 2a. n The t-distribution, As a location parameter of the t-distribution As a scaling parameter of the t-distribution.

[0104] Step S3: In each iteration, the posterior distribution obtained from the nth iteration is processed using the Metropolis-Hastings algorithm. Sample sequences were obtained by sampling from the middle. In the formula, the t-th sample The temperature sample value obtained from the t-th sampling. and temperature variance sample values The set, where N is the total number of samplings, typically on the order of hundreds of thousands;

[0105] Furthermore, for the sample sequence The samples were burned and then diluted sequentially to discard the original sequences. The first B samples Then from the remaining samples One sample is drawn every k samples, and the resulting new sample sequence is used as the uncertainty assessment model for the nth iteration. B and k are both preset values.

[0106] Step S4: Using the uncertainty assessment model of the nth iteration, evaluate the temperature measurement data collected during the nth iteration. Uncertainty assessment is performed.

[0107] Specifically, step S4 includes:

[0108] In the nth iteration, the following uncertainty assessment results are output: ;

[0109] ;

[0110] ;

[0111] In the formula, This represents the optimal temperature measurement estimate at the nth iteration, which indicates the temperature measurement data collected at the nth iteration. The final measurement results Let be the uncertainty at the nth iteration.

[0112] Therefore, this invention will use the posterior distribution of the (n-1)th iteration. and the temperature measurement data collected during the nth iteration The constructed likelihood function The dynamic fusion forms the posterior distribution of the nth iteration. After sampling using the Metropolis-Hastings algorithm, combustion and dilution processes are performed. The resulting uncertainty assessment model is then used to evaluate the temperature measurement data collected during the nth iteration. The uncertainty is dynamically assessed in real time;

[0113] Among them, due to the posterior distribution and uncertainty assessment model changing with each temperature measurement data The data is collected and iteratively corrected, thus continuously improving the accuracy of uncertainty assessment;

[0114] The Metropolis-Hastings algorithm is used for sampling, and the new sample sequence obtained is used as the uncertainty assessment model. The burning process removes the first B samples that may not converge to a stable distribution to eliminate the influence of the initial value. The dilution process further reduces autocorrelation, so that the uncertainty assessment model meets the independence principle.

[0115] In summary, this invention can dynamically assess the uncertainty of an online temperature measurement system, ensuring the accuracy and reliability of the measured values.

[0116] Example 2

[0117] like Figure 2 and Figure 3 As shown, the present invention also discloses an online temperature measurement system, comprising: a temperature acquisition subsystem, used to convert temperature signals into electrical signals, and then output digital signals through modulation amplification and analog-to-digital conversion, as temperature measurement values ​​output by the online temperature measurement system, wherein m temperature measurement values ​​constitute temperature measurement data. ;

[0118] It also includes: edge computing subsystem, cloud subsystem, and web client;

[0119] The edge computing subsystem connects to the temperature acquisition subsystem via bus communication, and the edge computing subsystem controls the temperature acquisition subsystem through the VISA software interface API. The VISA virtual instrument framework is used to abstract the instrument control behavior into a set of computer-recognizable program commands. The logical control of the temperature acquisition subsystem is based on the SCPI command set. After the edge computing subsystem encapsulates the SCPI program control commands, they are sent to the SCPI command interpreter of the temperature acquisition subsystem through the physical connection of bus technology to complete the program control of the temperature acquisition subsystem.

[0120] The edge computing subsystem communicates with the cloud subsystem based on the MQTT protocol. It is equipped with an MQTT client to enable the uploading of measurement results and the receiving of cloud measurement and control commands.

[0121] The web client communicates with the cloud subsystem via the network. The web client is developed based on the HTML language and uses the HTML language to construct web page content and structure to realize remote display of measurement results and deployment of remote measurement tasks.

[0122] This enables the edge computing subsystem to receive temperature measurement data output by the temperature acquisition subsystem. Based on the uncertainty dynamic evaluation method described in Example 1, the received temperature measurement data is processed. Uncertainty assessment is performed, and the results are uploaded to the cloud subsystem for web client access; wherein, the edge computing subsystem will use the posterior distribution obtained in step S2. Upload to the cloud subsystem to utilize the computing power of the cloud subsystem to perform the calculation in step S3.

[0123] The cloud server in the cloud subsystem includes an MQTT server (Broker) and a web backend server.

[0124] The MQTT server responds to and distributes system commands in the cloud, is responsible for the efficient relay of MQTT messages, provides message routing for network clients, and distributes these messages to all MQTT clients that have subscribed to the corresponding topics, ensuring that data and programmable commands between edge machines and front-end web MQTT clients are accurately received and properly distributed.

[0125] The web backend server establishes HTTP request-response links, based on Microsoft's IIS server software, and is responsible for managing network client interactions on the measurement and control cloud platform, thereby enabling remote access and control.

[0126] Therefore, to address the issue of remote deployment of testing tasks, cloud technology and cloud computing are combined to ultimately enable remote metering task deployment via a web browser.

[0127] Therefore, the online temperature measurement system of the present invention, combined with cloud technology and Internet of Things+ technology, realizes real-time remote transmission and monitoring of temperature measurement data, greatly expands the physical range of temperature testing task deployment, solves the problem that temperature measurement is limited and cannot be monitored in real time due to personnel safety threats and transmission physical distance under complex working conditions, and can meet the needs of large-scale industrial applications for real-time monitoring of temperature parameters and dynamic evaluation of uncertainty.

[0128] Among them, the computing power of the cloud can be used to control multiple test tasks in parallel, solving the problem of the large amount of traditional test data that is difficult to process;

[0129] Among them, the edge computing subsystem, in conjunction with the cloud subsystem, can automate the testing task process;

[0130] Furthermore, the test results are processed and stored in the cloud, which improves the traceability of the testing process.

[0131] This invention is not limited to the specific embodiments described above. Based on the above content and in accordance with common technical knowledge and conventional methods in the field, without departing from the basic technical concept of this invention, this invention can also make other equivalent modifications, substitutions or alterations, all of which fall within the protection scope of this invention.

Claims

1. A method for dynamic evaluation of uncertainty of an on-line temperature metering system, characterized in that, include: Step S1, Preparation Stage: Based on historical temperature measurement data from the online temperature measurement system, construct a priori distribution for evaluating its uncertainty. Among them, the prior distribution Based on temperature measurement data It is represented as the inverse gamma conjugate distribution of the true temperature value μ and the temperature variance φ; In order to collect temperature measurement data At that time, the temperature measurement values ​​output by the online temperature measurement system are from the 1st to the mth. Step S1 specifically includes: The historical temperature measurement data is obtained from activities in the metrology laboratory using an online temperature measurement system. The temperature true value μ and the temperature variance φ are set as the parameters to be estimated, and the conditional prior distribution and the marginal prior distribution The prior distribution is constituted in the form of a joint probability distribution : ； ； ； In the formula, the prior distribution The hyperparameters of the prior distribution are denoted as: The best estimate of the historical temperature, μ0, is equal to the average value of the historical temperature measurements. The scale parameters a0and b0are set to , , where v is the degrees of freedom of the prior distribution, is the variance of the historical temperature measurement data; Weight k0 is the preset weight value of the prior distribution; Conditional prior distribution This represents the distribution of the true temperature value μ given the temperature variance φ and weight k0. marginal prior distribution an independent distribution representing the variance φ of the temperature; Step S2, dynamic assessment phase: whenever a new set of temperature measurement data is collected in real time by the temperature online metering system an iteration is performed to fuse the posterior distribution of the (n-1)th iteration and the likelihood function of the nth iteration to form the posterior distribution of the nth iteration : ； In the formula, the likelihood function of the nth iteration Temperature measurement data collected during the nth iteration Constructed, and at the n=1th iteration For the prior distribution ; In step S2, the posterior distribution of the nth iteration The hyperparameters are updated as follows: ； ； ； ； wherein is the weight of the nth iteration, is the historical temperature optimum estimate of the nth iteration, is the temperature measurement data collected at the nth iteration is the average value of the temperature measurement data, is the scale parameter a0of the nth iteration, is the scale parameter b0of the nth iteration, is the variance of the historical temperature measurement data; Thus, the posterior distribution of the nth iteration is is expressed with the updated hyperparameters as ； wherein denotes a t-distribution with 2a n degrees of freedom, as a location parameter of the t-distribution, as a scale parameter of the t-distribution; Step S3: In each iteration, the posterior distribution obtained from the nth iteration is processed using the Metropolis-Hastings algorithm. Sample sequences were obtained by sampling from the middle. In the formula, the t-th sample The temperature sample value obtained from the t-th sampling. and temperature variance sample values The set, where N is the total number of samples; Furthermore, for the sample sequence The samples were subjected to combustion and dilution processes sequentially, with the first sample sequence being discarded. The first B samples are used as the uncertainty assessment model for the nth iteration. Then, one sample is drawn from the remaining samples every k samples, forming a new sample sequence. Uncertainty assessment is performed.

2. The method of claim 1, wherein: Step S4 specifically includes: At the nth iteration, output the following uncertainty assessment results: ; ； ； wherein is the temperature measurement best estimate at the n-th iteration, is the uncertainty at the n-th iteration.

3. An online temperature metrology system comprising: The temperature acquisition subsystem converts the temperature signal into an electrical signal, then modulates, amplifies, and converts it into a digital signal, which serves as the temperature measurement value output by the online temperature metering system. The temperature measurement data consists of m temperature measurement values. ; Its features include: an edge computing subsystem, a cloud subsystem, and a web client; The edge computing subsystem is connected to the temperature acquisition subsystem via bus communication, and the edge computing subsystem controls the temperature acquisition subsystem via the VISA software interface API. The edge computing subsystem communicates with the cloud subsystem based on the MQTT protocol. The web client communicates with the cloud subsystem via a network; This enables the edge computing subsystem to receive temperature measurement data output by the temperature acquisition subsystem. Based on the uncertainty dynamic evaluation method described in claim 1 or 2, the received temperature measurement data is processed. Uncertainty assessment is performed, and the results are uploaded to the cloud subsystem for web client access; wherein, the edge computing subsystem will use the posterior distribution obtained in step S2. Upload to the cloud subsystem to utilize the computing power of the cloud subsystem to perform the calculation in step S3.

4. The temperature on-line metering system of claim 3, wherein: The cloud servers of the cloud subsystem include: an MQTT server and a web backend server.

Images