CN114993501A - NTC temperature sensor calibration method based on edge calculation - Google Patents

Abstract

The invention discloses an NTC temperature sensor calibration method based on edge calculation, which comprises the steps that an edge calculation network is deployed on a calibration site, and an equipment end respectively sends data of equipment to be calibrated and a standard device to an edge node through a TCP/IP protocol according to a preset rule; the edge node analyzes the reported data and temporarily stores the data to a local database; selecting the calibrated optimal characteristic point at the edge node through a genetic algorithm; calibrating the calibrator and the standard at the characteristic point according to the characteristic point, and accurately calculating the measured temperature of the characteristic point of the temperature sensor through calibration data; and the edge node sends the calibration data and the calibration result to the cloud end to realize the cloud storage of the calibration record, and simultaneously sends the characteristic function to the equipment end to finish the calibration. Compared with the traditional calibration mode, the period required by calibration can be greatly shortened; the bandwidth consumption of the cloud server is effectively saved, and the computing pressure of the cloud end is reduced; the precision and the accuracy of the NTC temperature sensor can be effectively improved.

Description

A calibration method for NTC temperature sensor based on edge computing

technical field

The invention relates to the technical field of temperature calibration and measurement, in particular to a method for calibrating an NTC temperature sensor using edge computing technology.

Background technique

The traditional method of temperature calibration is laboratory calibration, in which the customer periodically transports his thermometer to be calibrated to the higher-level measurement and calibration institution, and the calibration professional performs the calibration with the high-grade standard platinum resistance in the environment-controlled calibration laboratory. After measurement and comparison, the calibrated thermometer is shipped back to the relevant customer unit, and the calibration certificate is signed and sent to the relevant customer unit. This traditional method has shortcomings such as long calibration period, high uncertainty, difficult management, and low efficiency, which will eventually have a certain impact on the normal production progress of modern industries.

In view of the above problems, the present invention adopts edge computing technology to complete the data fitting with the standard device at the edge of the device and establish an error compensation model, thereby reducing the temperature measurement error of the temperature sensor at the software level.

SUMMARY OF THE INVENTION

The purpose of the present invention is to propose an edge computing-based NTC temperature sensor calibration method for the shortcomings of the traditional temperature sensor calibration method, such as long calibration period, difficult management, and low efficiency.

The object of the present invention is achieved through the following technical solutions: a method for calibrating an NTC temperature sensor based on edge computing, the method comprises the following steps: Step 1: deploying an edge computing network at the calibration site;

Step 2: After the calibration process starts, the device sends the temperature data including the detection index of the NTC temperature sensor to be calibrated and the standard device to the edge node through the TCP/IP protocol;

Step 3: The edge node parses the received data

Step 4: Select the best feature point for calibration by genetic algorithm at the edge node;

Step 5: According to the feature points obtained in step 4 and carry out fitting analysis, the NTC temperature sensor to be calibrated and the standard device are calibrated at the feature points, and the measured temperature of the feature points of the temperature sensor is accurately calculated by the calibration data;

Step 6: The edge node uses the MQTT protocol to send the calibration data and calibration results north to the cloud to realize cloud storage of calibration records, and uses the TCP/IP protocol to send the feature function south to the device to complete the calibration.

Further, the detection indicators included in the step 2 are specifically resistance, temperature, time stamp, device number and ambient temperature.

Further, in the step 4, in order to reduce the time cost of calibration, the genetic algorithm is used to select the optimal feature point for calibration, and the specific steps for obtaining the optimal feature point are as follows:

(4.1) Gene establishment and feature point selection: abstract each temperature data sent by the device as a gene, define x _q to indicate whether the q-th test point is a feature point, and when x _q = 1, it means the q-th test The point is selected as the eigenvalue point, otherwise when x _q = 0, it means that the qth test point is not selected as the eigenvalue point;

(4.2) Definition of fitness function: Based on the given optimization objective, the fitness function is used to measure the quality of chromosomes in the population; since the goal of decision-making is to reduce the total time cost, the fitness function is selected to measure the total time cost of chromosomes C Reciprocal square of _total ,

(4.2.1) Time cost of single measurement: the time cost required to measure a temperature value, denoted as C _q ;

(4.2.2) Time cost of calibration error:

表示传感器读数；The time cost value corresponding to the single point error of the symbol c _j . where i represents the sensor number, j represents the test point, T _i represents the actual temperature value,

Indicates the sensor reading;

计算；The time cost of calibration error for a single sample ontology is

calculate;

(4.2.3) Calibration time cost of individual sample: the calibration time cost C _i of individual sample is the sum of the time cost of individual measurement and the time cost of error calibration;

C _i =C _s +C _q · _ni

n _i represents the number of measurement points in the calibration process of the sample individual;

(4.2.4) Total cost: the total cost C _total is the average value of the calibration time cost of all individuals, and M is the total number of samples;

后代可以表示为

(4.3) Crossover operation: Calculate the offspring c1 and c2 from the parents p1 and p2 using a simulated binary crossover; calculate the parameters of the offspring by selecting a random number u in [0,1)

The descendants can be expressed as

(4.4) Mutation operation: In order to simulate the genetic process, if an individual is to mutate, each characteristic of the individual must be multiplied by a random number within a range [1-m _rgnge , 1+m _range ], and m _range is used as the A multiplicative factor is independently calculated according to each feature. According to the degree of convergence, m _range ∈ [0.01, 0.3, 10% of the outstanding individuals are selected from the current population by the roulette selection method and enter the next iteration to make the outstanding individuals can be preserved as much as possible;

(4.5) Termination condition: beyond the following range, the genetic iteration terminates:

为最小输入温度，

为最大输入温度。σ ² (ri ) is the variance of the _i -th temperature feature point ri, σ ² (W _i ) is the variance of the _i - _th feature point combination Wi, cov(ri -r _j ) is the feature point i, _j coordinator variance,

is the minimum input temperature,

is the maximum input temperature.

Further, in the step 5, the third-order polynomial fitting formula of the Steinhart-Hart algorithm is selected to perform fitting analysis on the temperature data, as follows:

其中T为开式温标，单位为K，R为电阻，单位为KΩ，A，B，C为常数系数；通过分析误差平方和

即有：Select the third-order polynomial fitting formula for curve fitting of the RT table data of the NTC thermistor:

Where T is the open temperature scale, the unit is K, R is the resistance, the unit is KΩ, A, B, C are constant coefficients; by analyzing the sum of squares of errors

That is:

In the formula, R _i is the ith resistance, and T _i is the temperature value matched by R _i ; the constant is replaced by a _i, and brought into the polynomial fitting formula to obtain the constant matrix:

The constant coefficients of the fitting formula can be known:

A=(a23^2*b1-a12*a23*b3+a13*a22*b3-a13*a23*b2+a12*a33*b2-a22*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

B=(a13^2*b2-a12*a13*b3+a11*a23*b3-a13*a23*b1-a11*a33*b2+a12*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

C=(a12^2*b3-a12*a13*b2-a11*a22*b3+a11*a23*b2-a12*a23*b1+a13*a22*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

Thereby, the temperature standard value under a certain resistance value is calculated.

Further, in the step 5, the edge gateway calculates the temperature of the feature point based on the third-order polynomial fitting formula of curve fitting and performs secondary verification, sends the calibration result that has passed the secondary verification to the cloud server, and sends the calibration result after the calibration. The third-order polynomial fitting formula of the curve fitting is sent to the device, and the results that fail the secondary verification are recorded locally and reported to the cloud at the same time.

Further, the secondary verification of the edge gateway needs to satisfy the following two rules at the same time, specifically:

Rule 1: During the temperature calibration process, check whether the historical data reported by the edge gateway receiving device is within the predetermined range of the detection index. If the data value is not within the predetermined range, the verification is deemed to have failed;

Rule 2: Use the third-order polynomial fitting formula of curve fitting to calculate residuals and standard deviations. If the residuals and standard deviations exceed the interval range set by the cloud server, the verification will be regarded as failing.

Beneficial effects of the present invention:

Compared with the traditional calibration method, a method for calibrating an NTC temperature sensor based on edge computing of the present application can greatly shorten the period required for calibration.

A method for calibrating an NTC temperature sensor based on edge computing of the present application, the main process of calibration is all performed on the edge, and the device does not need to send all parameters to the cloud, which effectively saves the bandwidth consumption of the cloud server and reduces cloud computing. pressure.

A method for calibrating an NTC temperature sensor based on edge computing of the present application can effectively improve the precision and accuracy of the NTC temperature sensor.

Description of drawings

FIG. 1 is a flow chart of the method of the present invention.

Detailed ways

The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

As shown in Figure 1, this application proposes an edge computing-based NTC temperature sensor calibration method. First, the edge node obtains the real-time data of the device-side NTC temperature sensor through the TCP/IP protocol, and selects the optimal feature point based on the genetic algorithm. Calibration is performed, and temperature data fitting analysis based on the Steinhart-Hart algorithm is performed at the feature points. Secondly, the calibrated temperature model is updated to the device through the TCP/IP protocol, and the calibration results are sent to the cloud server to realize the edge gateway. The north-south communication completes the calibration of the temperature sensor at the edge. Specific steps are as follows:

Step 1. Deploy an edge computing network at the calibration site where equipment calibration is required.

Specifically, an edge computing network consists of edge computing nodes and device nodes. The deployment of edge nodes uses container technology to build indiscriminate edge computing nodes in different types of devices. The carrier of edge nodes can be devices with certain computing capabilities such as Windows, MacOS, or Linux devices. Each temperature sensor is a device node with wireless communication capability, and forms an edge computing network together with edge nodes through a specific communication protocol.

One or more of the edge computing nodes can be selected according to actual requirements.

Specifically, it is assumed that the number of devices to be calibrated is large and the time is urgent. Therefore, if a single edge computing node is under computational pressure, the number of edge computing nodes can be dynamically increased to perform parallel operations on multiple edge computing nodes to improve calibration efficiency.

Step 2, the device end includes one or more standards and a plurality of devices to be calibrated, the device to be calibrated is the NTC temperature sensor to be calibrated, and the device end can report to the edge node the inclusion detection indicators that need to be calibrated. The temperature data collected is collected, and each device terminal sends the collected temperature data including detection indicators to the edge computing node at a predetermined time interval.

For example, the device sends the collected temperature, resistance value, timestamp, device ID, and ambient temperature to the edge node through the TCP/IP protocol, where the device ID and ambient temperature only need to be sent once before the calibration process starts. , the temperature, resistance value, and timestamp are uploaded to the edge node at a frequency of every 5 seconds.

Step 3. The edge node parses the reported json format data and temporarily stores it in the local redis database. After a week of storage, the local database data is clear, freeing the storage space of the edge node and facilitating short-term traceability. Specifically, the devices in the database use the device number as the primary key to distinguish data such as temperature and resistance reported by different devices. The device takes whether the calibration is completed as the status bit. When the device completes the calibration, the database automatically deletes the data under the device number to ensure the cleanliness of the database; the data on the device side that has not been calibrated is retained in the database for 24 hours. After the time limit is exceeded, the database automatically Clean data.

Step 4. Select the best feature point for calibration by genetic algorithm at the edge node.

Specifically, for mass-produced temperature measurement modules, due to the differences between sensors, the input-output characteristics have obvious nonlinearity, and the individual differences are relatively large. Calibration will greatly increase the time cost. Therefore, when selecting feature points for fitting, there is a certain restrictive relationship between the effect and the selected data. Therefore, the genetic algorithm is used to perform the feature points on the premise of not changing the fitting effect. selection, so as to obtain a solution with high efficiency, low cost and good fitting effect. The specific process is as follows:

(4.1) Gene establishment and feature point selection: abstract each temperature data sent by the device as a gene, define x _q to indicate whether the q-th test point is a feature point, and when x _q = 1, it means the q-th test The point is selected as the eigenvalue point, otherwise when x _q = 0, it means that the qth test point is not selected as the eigenvalue point;

(4.2) Definition of fitness function: Based on the given optimization objective, the fitness function is used to measure the quality of chromosomes in the population; since the goal of decision-making is to reduce the total time cost, the fitness function is selected to measure the total time cost of chromosomes C Reciprocal square of _total ,

(4.2.1) Time cost of single measurement: the time cost required to measure a temperature value, denoted as C _q =50;

(4.2.2) Time cost of calibration error:

表示传感器读数；The time cost value corresponding to the single point error of the symbol c _j . where i is the sensor number, j is the test point, T _i,j is the actual temperature value,

Indicates the sensor reading;

计算；The time cost of calibration error for a single sample ontology is

calculate;

(4.2.3) Calibration time cost of individual sample: the calibration time cost C _i of individual sample is the sum of the time cost of individual measurement and the time cost of error calibration;

C _i =C _s +C _q · _ni

n _i represents the number of measurement points in the calibration process of the sample individual;

(4.2.4) Total cost: the total cost C _total is the average value of the calibration time cost of all individuals, and M is the total number of samples;

后代可以表示为

(4.3) Crossover operation: Calculate the offspring c1 and c2 from the parents p1 and p2 using a simulated binary crossover; calculate the parameters of the offspring by selecting a random number u in [0,1)

The descendants can be expressed as

(4.4) Mutation operation: In order to simulate the genetic process, if an individual is to mutate, each characteristic of the individual must be multiplied by a random number in a range [1-m _range , 1+m _range ], and m _range is used as the A multiplicative factor is independently calculated according to each feature. According to the degree of convergence, m _range ∈ [0.01, 0.3, 10% of the outstanding individuals are selected from the current population by the roulette selection method and enter the next iteration to make the outstanding individuals can be preserved as much as possible;

(4.5) Termination condition: beyond the following range, the genetic iteration terminates:

为最小输入温度，

为最大输入温度。σ ² (ri ) is the variance of the _i -th temperature feature point ri, σ ² (W _i ) is the variance of the _i - _th feature point combination Wi, cov(ri -r _j ) is the feature point i, _j coordinator variance,

is the minimum input temperature,

is the maximum input temperature.

For example, when the selected temperature feature points are 6, the corresponding total cost C _total is 350±10, and the time spent is 350±100s; when there are 5 feature points, the corresponding cost is 260±10 , the time spent is 600±100s, and the increase in the number of times is because the number of iterations increases when the genetic algorithm selects feature points. The total cost is basically close to the number of feature points selected multiplied by the time cost per measurement.

Step 5. Fit the temperature-resistance value change curve of the device to be calibrated and the standard at the characteristic points through the temperature characteristic points obtained by the genetic algorithm, and accurately calculate the characteristic points of the temperature sensor through the temperature model obtained by fitting. measured temperature. Select the third-order polynomial fitting formula of the Steinhart-Hart algorithm to fit and analyze the temperature data, as follows:

其中T为开式温标，单位为K，R为电阻，单位为KΩ，A，B，C为常数系数；通过分析误差平方和

即有：Select the third-order polynomial fitting formula for curve fitting of the RT table data of the NTC thermistor:

Where T is the open temperature scale, the unit is K, R is the resistance, the unit is KΩ, A, B, C are constant coefficients; by analyzing the sum of squares of errors

That is:

In the formula, R _i is the ith resistance, and T _i is the temperature value matched by R _i ; the constant is replaced by a _i, and brought into the polynomial fitting formula to obtain the constant matrix:

The constant coefficients of the fitting formula can be known:

A=(a23^2*b1-a12*a23*b3+a13*a22*b3-a13*a23*b2+a12*a33*b2-a22*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

B=(a13^2*b2-a12*a13*b3+a11*a23*b3-a13*a23*b1-a11*a33*b2+a12*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

C=(a12^2*b3-a12*a13*b2-a11*a22*b3+a11*a23*b2-a12*a23*b1+a13*a22*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33)

Thereby, the temperature standard value under a certain resistance value is calculated.

The edge gateway calculates the temperature of the feature points based on the third-order polynomial fitting formula of curve fitting and performs secondary verification, sends the calibration results that pass the secondary verification to the cloud server, and fits the calibrated curve to the third-order polynomial. The fitting formula is sent to the device, and the results that fail the secondary verification are recorded locally and reported to the cloud at the same time.

The edge gateway secondary verification needs to meet the following two rules at the same time, specifically:

Rule 1: During the temperature calibration process, check whether the historical data reported by the edge gateway receiving device is within the predetermined range of the detection index. If the data value is not within the predetermined range, the verification is deemed to have failed;

Rule 2: Use the third-order polynomial fitting formula of curve fitting to calculate residuals and standard deviations. If the residuals and standard deviations exceed the interval range set by the cloud server, the verification will be regarded as failing. The operating system used in this instance is Linux, the development language used on the edge side is go1.17, and the development language used on the device side is C language. The Raspberry Pi 4B is used as the edge node, and the memory is 4GB. The following 6 points are selected to fit the data reported by the device through the genetic algorithm. the result shows. After calibration, the residual error is controlled within the range of ±1mK, the standard deviation is 0.0049℃, and the calibration effect is good.

Table 1 Data analysis before and after calibration

Step 6: The edge node uses the MQTT protocol to send the calibration data and calibration results to the cloud to realize cloud storage of calibration records, and uses the TCP/IP protocol to send the characteristic function to the device to complete the calibration.

The above-mentioned embodiments are used to explain the present invention, rather than limit the present invention. Within the spirit of the present invention and the protection scope of the claims, any modifications and changes made to the present invention all fall into the protection scope of the present invention.

Claims (6)

1. an NTC temperature sensor calibration method based on edge computing, is characterized in that, this method comprises the steps: Step 1: Deploy the edge computing network at the calibration site; Step 2: After the calibration process starts, the device sends the temperature data including the detection index of the NTC temperature sensor to be calibrated and the standard device to the edge node through the TCP/IP protocol; Step 3: The edge node parses the received data Step 4: Select the best feature point for calibration by genetic algorithm at the edge node; Step 5: According to the feature points obtained in step 4 and carry out fitting analysis, the NTC temperature sensor to be calibrated and the standard device are calibrated at the feature points, and the measured temperature of the feature points of the temperature sensor is accurately calculated by the calibration data; Step 6: The edge node uses the MQTT protocol to send the calibration data and calibration results north to the cloud to realize cloud storage of calibration records, and uses the TCP/IP protocol to send the feature function south to the device to complete the calibration. 2 . A method for calibrating an NTC temperature sensor based on edge computing according to claim 1 , wherein the detection indicators included in the step 2 are specifically resistance, temperature, time stamp, device number and ambient temperature. 3 . 3. a kind of NTC temperature sensor calibration method based on edge computing according to claim 1, is characterized in that, in described step 4, in order to reduce the time cost spent in calibration, adopt genetic algorithm to select optimal feature point to calibrate, The specific steps to obtain the optimal feature points are as follows: (4.1) Gene establishment and feature point selection: abstract each temperature data sent by the device as a gene, define x _q to indicate whether the q-th test point is a feature point, and when x _q = 1, it means the q-th test The point is selected as the eigenvalue point, otherwise when x _q = 0, it means that the qth test point is not selected as the eigenvalue point; (4.2) Definition of fitness function: Based on the given optimization objective, the fitness function is used to measure the quality of chromosomes in the population; since the goal of decision-making is to reduce the total time cost, the fitness function is selected to measure the total time cost of chromosomes C Reciprocal square of _total ,

(4.2.1) Time cost of single measurement: the time cost required to measure a temperature value, denoted as C _q ; (4.2.2) Time cost of calibration error:

The time cost value corresponding to the single point error of the symbol c _j . where i is the sensor number, j is the test point, T _i,j is the actual temperature value,

Indicates the sensor reading; The time cost of calibration error for a single sample ontology is

calculate; (4.2.3) Calibration time cost of individual sample: the calibration time cost C _i of individual sample is the sum of the time cost of individual measurement and the time cost of error calibration; C _i =C _s +C _q · _ni n _i represents the number of measurement points in the calibration process of the sample individual; (4.2.4) Total cost: the total cost C _total is the average value of the calibration time cost of all individuals, and M is the total number of samples;

(4.3) Crossover operation: Calculate the offspring c1 and c2 from the parents p1 and p2 using a simulated binary crossover; calculate the parameters of the offspring by selecting a random number u in [0,1)

The descendants can be expressed as

(4.4) Mutation operation: In order to simulate the genetic process, if an individual is to mutate, each characteristic of the individual must be multiplied by a random number in a range [1-m _range , 1+m _range ], and m _range is used as the A multiplicative factor is independently calculated according to each feature. According to the degree of convergence, m _range ∈ [0.01, 0.3], 10% of the outstanding individuals are selected from the current population by the roulette selection method to enter the next iteration of the next generation, so that the excellent the individual can be retained to the greatest extent possible; (4.5) Termination condition: beyond the following range, the genetic iteration terminates:

σ ² (ri ) is the variance of the _i -th temperature feature point ri, σ ² (W _i ) is the variance of the _i - _th feature point combination Wi, cov(ri -r _j ) is the feature point i, _j coordinator variance,

is the minimum input temperature,

is the maximum input temperature. 4. a kind of NTC temperature sensor calibration method based on edge computing according to claim 1, is characterized in that, in described step 5, selects the third-order polynomial fitting formula of Steinhart-Hart algorithm to carry out fitting analysis to temperature data ,details as follows: Select the third-order polynomial fitting formula for curve fitting of the RT table data of the NTC thermistor:

Where T is the open temperature scale, the unit is K, R is the resistance, the unit is KΩ, A, B, C are constant coefficients; by analyzing the sum of squares of errors

That is:

In the formula, R _i is the ith resistance, and T _i is the temperature value of R _i matching; the constants are replaced by a _i,j and brought into the polynomial fitting formula to obtain the constant matrix:

The constant coefficients of the fitting formula can be known: A=(a23^2*b1-a12*a23*b3+a13*a22*b3-a13*a23*b2+a12*a33*b2-a22*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33) B=(a13^2*b2-a12*a13*b3+a11*a23*b3-a13*a23*b1-a11*a33*b2+a12*a33*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33) C=(a12^2*b3-a12*a13*b2-a11*a22*b3+a11*a23*b2-a12*a23*b1+a13*a22*b1)/(a33*a12^2-2*a12 *a13*a23+a22*a13^2+a11*a23^2-a11*a22*a33) Thereby, the temperature standard value under a certain resistance value is calculated. 5. a kind of NTC temperature sensor calibration method based on edge computing according to claim 4, is characterized in that, in described step 5, edge gateway calculates the temperature of characteristic point based on the third-order polynomial fitting formula of curve fitting and carries out. Secondary verification, send the calibration results that pass the secondary verification to the cloud server and the third-order polynomial fitting formula of the calibrated curve fitting to the device, and record the results that fail the secondary verification locally At the same time report to the cloud. 6. A kind of NTC temperature sensor calibration method based on edge computing according to claim 5, is characterized in that, edge gateway secondary verification needs to satisfy following two rules at the same time, is specifically: Rule 1: During the temperature calibration process, check whether the historical data reported by the edge gateway receiving device is within the predetermined range of the detection index. If the data value is not within the predetermined range, the verification is deemed to have failed; Rule 2: Use the third-order polynomial fitting formula of curve fitting to calculate residuals and standard deviations. If the residuals and standard deviations exceed the interval range set by the cloud server, the verification will be regarded as failing.

Images