Passive RFID temperature measurement method and system based on numerical fitting

Abstract

The invention discloses a passive RFID temperature measurement method and system based on numerical fitting, and the method comprises the steps: enabling a passive RFID tag to be attached to a constant-temperature heating container to sense the temperature, drawing a scatter diagram according to the relation between the temperature under the current tag and the duration time under the current temperature, removing the collected abnormal temperature data, and carrying out the calculation of the temperature of the passive RFID tag. The duration time at the current temperature is the time span of the passive RFID tag from the initial power supply voltage attenuation to the voltage threshold value when volatile data is lost; performing optimal segmentation by using a window variance; segments are obtained according to an optimal segmentation method, and a mapping relation between duration and temperature is established for each segment; and setting early warning of the abnormal state of the label. According to the method, the corresponding abnormal value is obtained according to the duration, the abnormal value is calculated by measuring the duration at the temperature for multiple times and comparing the duration with the corresponding mean value, and the corresponding abnormal value and outlier are eliminated, so that the data value is more effective, and the measurement accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of temperature monitoring, and in particular to a passive RFID temperature measurement method and system based on numerical fitting. Background Art

[0002] In addition to playing an important role in maintaining quality assurance, meeting regulatory compliance in applications such as fire safety and cold chain monitoring, temperature sensors are also widely used in the industrial field. For example, they are used for temperature monitoring and control in industrial processes to ensure that equipment operates within the appropriate temperature range. In some industrial equipment, temperature changes may affect the measurement results. Temperature sensors are used for temperature compensation to eliminate this effect and provide more accurate measurement results. Temperature sensors can also monitor the temperature of the equipment and issue an alarm when the temperature exceeds the set threshold, helping to detect and repair equipment failures in a timely manner and prevent equipment damage.

[0003] Existing temperature measurement equipment is bulky and expensive, making item-level temperature monitoring too expensive. Moreover, most existing sensors cannot directly implement wireless temperature sensing, requiring additional embedded circuits (such as RF transmitters) to perform data transmission, which is complex to implement. Summary of the invention

[0004] Purpose of the invention: In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a passive RFID temperature measurement method based on numerical fitting, and also provides a passive RFID temperature measurement system based on numerical fitting.

[0005] Summary of the invention: The present invention provides a passive RFID temperature measurement method based on numerical fitting, the method comprising: Passive RFID tags are attached to the constant temperature heating container to sense the temperature. The sensing temperature range is from -30℃ to 120℃, with a step length of 5℃. That is, every time the temperature rises by 5℃, the temperature and related data are automatically collected and recorded in a fixed format, and automatically uploaded and stored in the database. The relationship between the temperature under the current tag and the duration at the current temperature is plotted as a scatter plot, and the abnormal temperature data collected is eliminated. The duration at the current temperature is the time span from the initial power supply voltage decay of the passive RFID tag to the voltage threshold when the volatile data is lost; The relationship between the temperature and duration retained after screening is drawn into a relationship curve, and the curve is optimally segmented using the window variance; The segments obtained according to the optimal segmentation method are recorded as low temperature segment, medium temperature segment and high temperature segment, and a mapping relationship between duration and temperature is established for each segment. The fitting functions corresponding to the low temperature segment, medium temperature segment and high temperature segment are recorded as M1, M2 and M3 respectively; Combine the fitting function and the ambient temperature value to set an early warning for abnormal tag status.

[0006] Further, including: After establishing a mapping relationship between duration and temperature for each segment, the method further includes: Since tags with the same model have similar curve patterns and fitting parameters, the duration set {( T i , p i )},in, T i For the i Passive tags at temperature p The duration of the following p i For the i The current temperature of the passive tag; The average values ​​of the fitting parameters under multiple different labels are calculated as the new fitting parameters of the model, and the corresponding fitting functions are recorded as M1', M2', and M3' respectively.

[0007] Further, including: The relationship between the temperature under the current label and the duration under the current temperature is drawn into a scatter plot, and the collected abnormal temperature data is eliminated, specifically including: Calculate the duration of a single measurement at each temperature, where the duration of a measurement of the same tag is calculated as follows: ;in, , For this label at temperature j Next n The duration of the measurement is in seconds. n The number of groups currently measured by the tag. Indicates that the tag n The timestamp of the group data, in milliseconds; Stay at the same temperature for a period of time and collect multiple times to avoid the uncertainty error caused by a single collection. Therefore, for each temperature j , calculate the mean of its duration , which is expressed as: ; Where N is the total number of groups measured under this tag; Delete the same tag with a duration less than 100% at different temperatures. and greater than The data record, that is, when or When , remove the corresponding outliers ;in, in, j The range of is -30 to 120 and is a multiple of 5 and -5; therefore, , ; For each temperature j , delete the outliers, as follows: remember and ‌ is the temperature of the label j The 25th and 75th percentiles of ;when or When , remove the corresponding outlier . Further, including: The relationship between the temperature and the duration retained after screening is drawn into a relationship curve, and the curve is optimally segmented using the window variance, specifically including: Calculate the average duration of the remaining measurements at each temperature, where the remaining measurements are the durations corresponding to the remaining number of measurement groups after the collected data is filtered. Therefore, the corresponding average duration is calculated as follows: The duration calculation formula for the same tag is as follows: ; After removing abnormal values ​​and outliers, For this label at temperature j The average duration of the label is in seconds, M is the total number of groups of measurements remaining after the label is filtered and , a,b,..,m is the number of current measurement groups remaining after the screening process, and ; If there are no available measurement groups after the screening process, all the data collected under this tag will be ignored; The relationship between temperature and duration is drawn as a relationship curve, and the temperature slides along the horizontal axis from the starting point to the end point. The sliding window is 5°C each time. Before each sliding, the average value of the duration corresponding to the five sampling points in the current window segment is calculated, and the sum of the variances of each sampling point and the mean is calculated and saved as the window variance value, which is specifically expressed as: ; ;in, For the temperature i The duration corresponding to Indicates that the window start temperature value is i The average of the 5 values ​​in the window, Indicates that the window start temperature value is i The variance value corresponding to the window; Set the threshold to , calculate the absolute error between the window variance value at the current temperature and the window variance value at the previous moment. If the absolute error is greater than the threshold, the first sampling point of the current temperature window is determined to be the first segmentation point; that is: Otherwise, continue sliding the window until the absolute error is greater than the threshold, and the first sampling point of the current window is the first segmentation point. And so on, get other segmentation points until the window slides to the end; Assume that after the sliding is completed, the curve is divided into k Section, when k ≥3, only 3 segments are retained, that is, the 2 segmentation points with the largest window variance are taken; when k <3, it is expanded to 3 segments, that is, continue to add the windows with the largest window variance value as segmentation points.

[0008] Further, including: The step of establishing a mapping relationship between duration and temperature for each segment includes: For the AB segment, i.e. the low temperature segment, a one-dimensional quadratic regression model is used for fitting. This is because the discharge time at different temperatures is not much different at low temperatures, and the data fluctuates differently. It is impossible to directly use the theoretical mechanism model for fitting. The error caused by diversity will lead to a larger error. A one-dimensional quadratic fitting model is constructed: ;in, , and are the unknown coefficients of the linear fitting model, p 1 is the temperature value, For temperature p 1 duration of the next attack; According to the collected sample data, the fitting function M1 is calculated by the least square method; Then the fitting function M1 is used to find the goodness of fit of the data. , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, and thus the parameters are obtained. , and Otherwise, re-collect data until the model meets the goodness of fit requirements; get the parameter , and After the value of and duration , you can get the temperature value p 1.

[0009] Further, including: The step of establishing a mapping relationship between duration and temperature for each segment further includes: For the BC section, i.e. the medium temperature section, the mechanism model is used. To fit, where a,b,c are 3 constants. In order to obtain a,b,c Three constants; the following steps are used to achieve: Minimize the objective function: , according to the sample data collected under this label at this temperature after screening , the fitting function M2 is calculated by the least squares method; the fitting goodness of the data is calculated using the fitting function , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; Get parameters a,b,c Then according to the formula , according to the duration The temperature value can be directly calculated p 2.

[0010] Further, including: The step of establishing a mapping relationship between duration and temperature for each segment further includes: For the CD segment, a univariate quadratic regression model is used for fitting, and a univariate quadratic fitting model is constructed: Construct a univariate quadratic fitting model: in, , and are the unknown coefficients of the linear fitting model, p 3 is the temperature value, For temperature p 3. Duration of the next attack; According to the sample data collected under the label at the temperature after screening, the fitting function M3 is calculated by the least square method; Then, the fitting function M3 is used to find the goodness of fit of the data. , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; Get parameters , and Then, according to the formula , according to the duration You can get the temperature value p 3. Further, including: The method of combining the fitting function and the ambient temperature value to set an early warning of the abnormal state of the tag includes: Set duration thresholds T1 and T2, and T1< T2. If the duration is less than T1, call model M1' to get the temperature value. p 1; If the duration is greater than or equal to T1 and less than or equal to T2, call model M2' to get the temperature value p 2; When the duration is greater than T2, call model M3' to get the temperature value p 3. The temperature value obtained in this step is recorded as P; Setting the maximum duration threshold th and minimum duration threshold tl , when the duration is less than tl or greater than th When the label is abnormal, it is judged that the label is in abnormal state and a prompt is given to replace the label; Collect ambient temperature values ​​through the reader and set temperature measurement deviation warning values td , when the difference between the obtained temperature value P and the ambient temperature value is greater than td When an alarm is triggered, immediately check the tag status and device status, fill in and report the corresponding fault type and cause.

[0011] On the other hand, the present invention also provides a passive RFID temperature measurement system based on numerical fitting, the system comprising: The data acquisition module is used to attach a passive RFID tag to a constant temperature heating container to sense the temperature. The sensing temperature range is from -30°C to 120°C, with a step length of 5°C. That is, every time the temperature rises by 5°C, the temperature and related data are automatically collected and recorded in a fixed format, and automatically uploaded and stored in the database; A preprocessing module, used to draw a scatter plot of the relationship between the temperature under the current tag and the duration at the current temperature, and to remove the abnormal temperature data collected, wherein the duration at the current temperature is the time span from the decay of the initial power supply voltage of the passive RFID tag to the voltage threshold when the volatile data is lost; The optimal segmentation module is used to draw the relationship between the retained temperature and duration into a relationship curve, and to perform optimal segmentation on the curve using the window variance; A fitting module is used to record the segments obtained according to the optimal segmentation method as low-temperature segments, medium-temperature segments and high-temperature segments, and establish a mapping relationship between duration and temperature for each segment. The fitting functions corresponding to the low-temperature segment, medium-temperature segment and high-temperature segment are recorded as M1, M2 and M3 respectively. The early warning module is used to combine the fitting function and the ambient temperature value to set an early warning for abnormal label status.

[0012] Further, including: In the fitting module, after establishing a mapping relationship between duration and temperature for each segment, the following steps are further included: Since tags with the same model have similar curve patterns and fitting parameters, the duration set {( T i , p i )},in, T i For the i Passive tags at temperature p The duration of the following p i For the i The current temperature of the passive tag; The average values ​​of the fitting parameters under multiple different labels are calculated as the new fitting parameters of the model, and the corresponding fitting functions are recorded as M1', M2', and M3' respectively.

[0013] Beneficial effects: Compared with the prior art, the present invention has the following advantages: (1) The present invention collects temperature data in a preprocessing process, calculates the duration at each temperature, and obtains the abnormal values ​​therein according to the duration. The abnormal values ​​are calculated by comparing the duration at the temperature with the corresponding mean value for multiple times, thereby eliminating the corresponding abnormal values ​​and outliers, making the data value more effective, thereby improving the accuracy of the measurement; (2) The present invention uses a sliding window method to segment the relationship curve between temperature and duration into three segments. Specifically, the mean and variance corresponding to the duration are calculated, and the variance is compared with the corresponding threshold to obtain the corresponding segmentation points. This data processing method provides a basis for subsequent fitting, that is, different temperature segments use different fitting models, so that the final calculated temperature value has higher accuracy. Therefore, temperature sensing of the tag under specific conditions can be achieved based on data fitting. The segmented fitting used in the present invention is more universal and more accurate than pure theoretical calculations and empirical models.

[0014] (3) The present invention obtains corresponding mapping models according to different fitting models to achieve accurate and reliable temperature sensing, and at the same time sets an abnormal status warning for the tag, so that the tag can be replaced or the status of the tag and the device can be checked according to the warning reminder, which is more practical. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a passive RFID temperature measurement method based on numerical fitting according to an embodiment of the present invention; Figure 2 This is an example diagram of the format of the collected temperature data according to the embodiment of the present invention; Figure 3This is an example diagram of temperature and duration fitting curves under three different passive tags in the low temperature section according to an embodiment of the present invention; Figure 4 This is an example diagram of temperature and duration fitting curves for three different passive tags in the medium temperature range according to an embodiment of the present invention; Figure 5 This is an example diagram of temperature and duration fitting curves under three different passive tags in the high temperature section according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of a passive RFID temperature measurement system based on numerical fitting according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example

[0017] like Figure 1 As shown, the present invention provides a passive RFID temperature measurement method based on numerical fitting, the method comprising the following steps: Step S1: Use a passive RFID tag attached to a constant temperature heating container to sense the temperature.

[0018] The sensed temperature range is from -30°C to 120°C with a step size of 5°C, that is, every time the temperature rises by 5°C, the program automatically calls the reader to collect and record data and automatically upload and store it in the database. The model of the reader is selected according to actual conditions and is not limited in this embodiment.

[0019] The specific temperature collection method is: First, a constant temperature heating container is selected, a passive RFID tag is attached to it, and the constant temperature heating container is continuously heated.

[0020] Place the reader in a fixed position. When the temperature rises, the reader is automatically called to obtain the measurement data. Recording starts from -30°C, and the temperature is kept constant for a period of time when the temperature rises by 5°C. After the reader reads the data, it is stored in the database. In this embodiment, 10 data records are collected each time, and the current temperature value is stored at the same time, until 120°C.

[0021] The recorded temperature data format is as follows Figure 2 As shown, it includes: temperature, tag ID, phase value, RSSI, timestamp in milliseconds, frequency, and sensor ID. Figure 2As shown, this is the data format when the temperature is 30°C: the temperature is 30°C, the first column is the tag ID, the second column is the phase value, the third column is the RSSI, the fourth column is the timestamp, the fifth column is the frequency, and the last column is the sensor ID.

[0022] In this embodiment, the present invention uses a common passive RFID tag to sense temperature, which is low-cost, battery-free and stable to environmental conditions. It has been widely used in many applications. The measurement principle is that the resistance of the semiconductor diode in the tag chip is sensitive to temperature, and the temperature is indirectly estimated by measuring the discharge time of the reverse polarized diode. The temperature measurement span of the present invention is large, ranging from -30°C to 120°C. And on the basis of the above, a standard discharge cycle measurement scheme using a volatile memory of the tag is proposed, and a mapping model between the discharge cycle and temperature is established to achieve accurate and reliable temperature sensing.

[0023] Step S2: Draw the relationship between temperature and duration into a scatter plot, and remove abnormal data.

[0024] Step S21: Calculate the duration of a single measurement at each temperature, where the duration is defined as the time span from the initial power supply voltage decaying to the voltage threshold when volatile data is lost. It represents the speed of the discharge process and indirectly reflects the speed of the reverse polarized semiconductor diode leakage current, that is, the current change of the tag circuit. The difficulty and cost of directly monitoring the measured current are not suitable, so consider using the duration to indirectly reflect the current changes around the tag circuit. The duration calculation formula for a certain measurement of the same tag is as follows: ;in, , For this label at temperature j Next n The duration of the measurement is in seconds. n The number of groups currently measured by the tag. Indicates that the tag n The timestamp of the group data is in milliseconds. The total number of experimental groups used in this embodiment is 10. This application collects multiple samples at the same temperature for a period of time to avoid the uncertainty error caused by a single collection. Therefore, for each temperature j , calculate the mean of its duration , which is expressed as: ; Where N is the total number of groups measured under this tag.

[0025] Step S22: Delete the same tag with a duration less than 100% at different temperatures. and greater than The data record, that is, when or When , remove the corresponding outliers .

[0026] in, in, j The range of is -30 to 120 and is a multiple of 5 and -5; therefore, , Step S23: For each temperature j , delete the outliers, as follows: remember and ‌ is the temperature of the label j The 25th and 75th percentiles of ;when or When , remove the corresponding outliers The data obtained after removing abnormal values ​​and outliers are basically valid data, which makes the curve obtained based on temperature and duration more accurate.

[0027] Step S3: Using the values ​​retained after elimination in step 2, the relationship between temperature and duration is plotted as a relationship curve, and the window variance is used to optimally segment the curve.

[0028] Step S31: Calculate the average duration of the remaining measurements at each temperature, where the remaining measurements are the durations corresponding to the remaining number of measurement groups after the measurement data corresponding to the abnormal values ​​and outliers are removed in step S2. Therefore, the corresponding average duration is calculated as follows: The duration calculation formula for the same tag is as follows: ; After removing abnormal values ​​and outliers, For this label at temperature j The average duration of the tag under the condition of S2 is in seconds, M is the total number of groups of measurements remaining after the tag is processed in step S2, and , a,b,..,m is the number of current measurement groups remaining after processing in step S2, and If there are no available measurement groups after step S2, all the data collected under the tag will be ignored.

[0029] Step S32: Draw the relationship between temperature and duration as a relationship curve, and slide the temperature along the horizontal axis from the starting point to the end point. The sliding window is 5°C each time. Before each sliding, calculate the average value of the duration corresponding to the 5 sampling points in the current window segment, and calculate the sum of the variances of each sampling point and the mean, and save it as the window variance value, which is specifically expressed as: in, is obtained according to step S31, indicating that the temperature is i The duration corresponding to Indicates that the window start temperature value is i The average of the 5 values ​​in the window, Indicates that the window start temperature value is i The variance value corresponding to the window.

[0030] Step S33: Set the threshold to , calculate the absolute error between the window variance value at the current temperature and the window variance value at the previous moment. If the absolute error is greater than the threshold, the first sampling point of the current temperature window is determined to be the first segmentation point; that is: Otherwise, continue sliding the window until the absolute error is greater than the threshold, and obtain the first sampling point of the current window as the first segmentation point. And so on, obtain other segmentation points until the window sliding ends. In this embodiment, Set to 0.1. Assume that after sliding, the curve is divided into k Section, when k ≥3, only 3 segments are retained: the 2 segmentation points with the largest window variance are selected; when k <3, it is expanded to 3 segments: continue to add the window with the largest window variance value as the segmentation point.

[0031] Step S4: The segments obtained in step S33 are recorded as AB segment, i.e., low temperature segment, BC segment, i.e., medium temperature segment, and CD segment, i.e., high temperature segment, and a mapping relationship between duration and temperature is established for each segment.

[0032] Step S41: For the AB segment, a one-dimensional quadratic regression model is used for fitting. This is because the discharge time at different temperatures is not much different at low temperatures, and the data fluctuates differently. It is impossible to directly use the theoretical mechanism model for fitting, because the error caused by diversity will lead to a larger error. Construct a one-dimensional quadratic fitting model: in, , and are the unknown coefficients of the linear fitting model, p is the temperature value, For temperature p duration of the following; According to the collected sample data, the fitting function M1 is calculated by the least square method; Then the fitting function M1 is used to find the goodness of fit of the data. , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, and thus the parameters are obtained. , and Otherwise, re-collect data until the model meets the goodness of fit requirement; Get parameters , and After the value of and duration , you can get the temperature value p .like Figure 3 As shown in FIG. 1 , a temperature and duration fitting curve and a corresponding fitting equation of this embodiment are shown, which are specifically expressed as: T = 0.0411 * p ^2 + (-0.025)* p +2.8966.

[0033] Step S42: For the BC segment, use the mechanism model To fit, where a,b,c are 3 constants. In order to obtain a,b,c Three constants; the following steps are used to achieve: Minimize the objective function: , according to the sample data collected under the label at this temperature after screening in step S2 , the fitting function M2 is calculated by the least squares method; the fitting goodness of the data is calculated using the fitting function , if and only if , then the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; get the parameters a,b,c Then according to the formula , according to the duration The temperature value can be directly calculated p .like Figure 4 As shown, a temperature and duration fitting curve and a corresponding fitting equation for a medium temperature section described in an embodiment of the present invention are specifically expressed as: T = 1265.9251 / (power(2, p / 9.4625) + 402.2918).

[0034] Step S43: For the CD segment, a univariate quadratic regression model is used for fitting to construct a univariate quadratic fitting model: Construct a univariate quadratic fitting model: ;in, , and are the unknown coefficients of the linear fitting model, p is the temperature value, For temperature p The duration of the following.

[0035] According to the sample data collected under the label at the temperature after screening in step S2, a fitting function M3 is calculated by the least square method; Then, the fitting function M3 is used to find the goodness of fit of the data. , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; get the parameters , and Then, according to the formula , according to the duration You can get the temperature value p ;like Figure 5 As shown in FIG. 1 , the temperature and duration fitting curve corresponding to the high temperature section and the corresponding fitting equation according to the embodiment of the present invention are specifically expressed as follows: T = 6.5801 * p ^2 + (-17.0672)* p + 11.1994.

[0036] This embodiment similarly considers the corresponding fitting models under all labels according to the above method. Different temperature sections adopt different fitting models according to their own trends. For example, the BC medium temperature section is consistent with the mechanism model, but the relationship between the temperature and duration of the low temperature section and the high temperature section is one more gentle and the other is steeper. If the mechanism model is used, the deviation is too large. Therefore, this application seeks other regression models for fitting, which is more in line with the actual situation and has higher accuracy.

[0037] Step 5: Consider the diversity of temperature measurement labels in actual production. Perform label calibration to obtain the best fit.

[0038] In order to make the model more adaptable to multiple labels, five different labels were used for experiments. Through a large number of experiments, it was found that labels with the same model have similar curve patterns and fitting parameters. Figure 3-5 As shown, a set of durations extracted from three different passive tags with the same model is plotted. Obviously, although label diversity has a certain effect on duration, the curves of different labels also experience a similar downward trend as the temperature increases. In this way, given a label model, it is possible to collect a set of labels from some labels in advance. . After that, the best fit of multiple fitting parameters is obtained. Then the average value of these fitting parameters is used as the fitting parameter of the label model, which is recorded as M1', M2', and M3' respectively. That is, each label has a fitting function, and then multiple labels need to be taken into account when applying it, so at this time, just take the average of their coefficients, specifically: like Figure 3-5 As shown, the corresponding temperature and duration fitting curves of three passive tags with similar models are shown, where: Figure 3 This is the fitting curve graph corresponding to the low temperature section. The label names from top to bottom are A001, A002 and B001. Figure 4 This is the fitting curve corresponding to the medium temperature section. The labels from top to bottom are: A001, A002 and B001. Figure 5 This is the fitting curve corresponding to the high temperature section. The labels from top to bottom are: B002, C001, C002. If the fitting parameters corresponding to three different labels are recorded as , and , therefore, the three parameters corresponding to M1' are expressed as: In this embodiment, the BC segment and the CD segment are obtained by a similar method, which will not be described in detail here.

[0039] Step S6: Based on the mapping models M1', M2', M3' between the discharge time and temperature obtained above, accurate and reliable temperature sensing is achieved, and an abnormal state warning of the tag is set. The steps are as follows: Step S61: setting duration thresholds T1 and T2. When the duration is less than T1, calling model M1' to obtain the temperature value; when the duration is greater than or equal to T1 and less than or equal to T2, calling model M2' to obtain the temperature value; when the duration is greater than T2, calling model M3' to obtain the temperature value; Step S62: Setting the maximum duration threshold th and minimum duration threshold tl , when the duration is less than tl or greater than th When the label is abnormal, it is judged that the label is in abnormal state and a prompt is given to replace the label.

[0040] Step S63: The reader has its own ambient temperature value and sets the temperature measurement value deviation warning value td When the difference between the temperature value obtained in step S61 and the ambient temperature value is greater than td When an alarm is triggered, immediately check the tag status and device status, fill in and report the corresponding fault type and cause.

[0041] In addition, in this embodiment, considering the device measurement deviation during data collection, Kalman filtering is introduced, and the initial value of the measurement noise deviation is , and the model fitting noise deviation initial value is Assuming that the temperature t The fitting duration at +1 is The measurement duration is , t The optimal duration deviation at time is recorded as: .temperature t The optimal temperature value under +1 is: , the optimal temperature value deviation is: . Set the device measurement abnormality threshold DT ,like , the device measurement value at this temperature is considered abnormal data. Check the device status and re-collect data at this temperature to ensure the accuracy of the fitting model. Example

[0042] On the other hand, the present invention also provides a passive RFID temperature measurement system based on numerical fitting, such as Figure 6 As shown, the system includes: The data acquisition module is used to attach a passive RFID tag to a constant temperature heating container to sense the temperature. The sensing temperature range is from -30°C to 120°C, with a step length of 5°C. That is, every time the temperature rises by 5°C, the temperature and related data are automatically collected and recorded in a fixed format, and automatically uploaded and stored in the database; A preprocessing module, used to draw a scatter plot of the relationship between the temperature under the current tag and the duration at the current temperature, and to remove the abnormal temperature data collected, wherein the duration at the current temperature is the time span from the decay of the initial power supply voltage of the passive RFID tag to the voltage threshold when the volatile data is lost; The optimal segmentation module is used to draw the relationship between the retained temperature and duration into a relationship curve, and to perform optimal segmentation on the curve using the window variance; A fitting module is used to record the segments obtained according to the optimal segmentation method as low-temperature segments, medium-temperature segments and high-temperature segments, and establish a mapping relationship between duration and temperature for each segment. The fitting functions corresponding to the low-temperature segment, medium-temperature segment and high-temperature segment are recorded as M1, M2 and M3 respectively. The early warning module is used to combine the fitting function and the ambient temperature value to set an early warning for abnormal label status.

[0043] Further, including: In the fitting module, after establishing a mapping relationship between duration and temperature for each segment, the following steps are further included: Since tags with the same model have similar curve patterns and fitting parameters, the duration set {( T i , p i )},in, T i For the i Passive tags at temperature p The duration of the following p i For the i The current temperature of the passive tag; The average values ​​of the fitting parameters under multiple different labels are calculated as the new fitting parameters of the model, and the corresponding fitting functions are recorded as M1', M2', and M3' respectively.

[0044] Other technical features of the passive RFID temperature measurement system based on numerical fitting described in the present invention are similar to those of the corresponding method and will not be described in detail here.

[0045] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0046] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present invention without departing from the spirit and scope of the embodiments of the present invention. Thus, if these modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A passive RFID temperature measurement method based on numerical fitting, characterized in that: The method includes: Use passive RFID tags attached to the constant temperature heating container to sense the temperature. The sensing temperature range is from -30℃ to 120℃, with a step length of 5℃. That is, every time the temperature rises by 5℃, it will automatically collect and then record the temperature and related data in a fixed format, and automatically upload and store them in the database; The relationship between the temperature under the current tag and the duration at the current temperature is drawn into a scatter plot, and a screening process is performed, that is, the collected abnormal temperature data is eliminated, and the duration at the current temperature is the time span of the passive RFID tag from the initial power supply voltage decay to the voltage threshold when the volatile data is lost; The relationship between the temperature and duration retained after screening is drawn into a relationship curve, and the curve is optimally segmented using the window variance; The segments obtained according to the optimal segmentation method are recorded as low temperature segment, medium temperature segment and high temperature segment, and a mapping relationship between duration and temperature is established for each segment. The fitting functions corresponding to the low temperature segment, medium temperature segment and high temperature segment are recorded as M1, M2 and M3 respectively; Combine the fitting function and the ambient temperature value to set an early warning for abnormal tag status.

2. The passive RFID temperature measurement method based on numerical fitting according to claim 1, characterized in that: After establishing a mapping relationship between duration and temperature for each segment, the method further includes: Since tags with the same model have similar curve patterns and fitting parameters, the duration set {( T i , p i )},in, T i For the i Passive tags at temperature p The duration of the following p i For the i The current temperature of the passive tag; The average values ​​of the fitting parameters under multiple different labels are calculated as the new fitting parameters of the model, and the corresponding fitting functions are recorded as M1', M2', and M3' respectively.

3. The passive RFID temperature measurement method based on numerical fitting according to claim 1 or 2, characterized in that: The relationship between the temperature under the current label and the duration under the current temperature is drawn into a scatter plot, and the collected abnormal temperature data is eliminated, specifically including: Calculate the duration of a single measurement at each temperature, where the duration of a measurement of the same tag is calculated as follows: ;in, , For this label at temperature j Next n The duration of the measurement is in seconds. n The number of groups currently measured by the tag. Indicates that the tag n The timestamp of the group data, in milliseconds; Stay at the same temperature for a period of time and collect multiple times to avoid the uncertainty error caused by a single collection. Therefore, for each temperature j , calculate the mean of its duration , which is expressed as: ; Where N is the total number of groups measured under this tag; Delete the same tag with a duration less than 100% at different temperatures. and greater than The data record, that is, when or When , remove the corresponding outliers ;in, in, j The range of is -30 to 120 and is a multiple of 5 and -5; therefore, , ; For each temperature j , delete the outliers, as follows: and ‌ is the temperature of the label j The 25th and 75th percentiles of ;when or When , remove the corresponding outliers .

4. The passive RFID temperature measurement method based on numerical fitting according to claim 3 is characterized in that: The relationship between the temperature and the duration retained after screening is drawn into a relationship curve, and the curve is optimally segmented using the window variance, specifically including: Calculate the average duration of the remaining measurements at each temperature, where the remaining measurements are the durations corresponding to the remaining number of measurement groups after the filtered data collection. Therefore, the corresponding average duration is calculated as follows: The duration calculation formula for the same tag is as follows: ; After removing abnormal values ​​and outliers, For this label at temperature j The average duration of the label is in seconds, M is the total number of groups of measurements remaining after the label is filtered and , is the number of current measurement groups remaining after the screening process, and ; If there are no available measurement groups after the screening process, all the data collected under this tag will be ignored; The relationship between temperature and duration is drawn as a relationship curve, and the temperature slides along the horizontal axis from the starting point to the end point. The sliding window is 5°C each time. Before each sliding, the average value of the duration corresponding to the five sampling points in the current window segment is calculated, and the sum of the variances of each sampling point and the mean is calculated and saved as the window variance value, which is specifically expressed as: ; ;in, For the temperature i The duration corresponding to Indicates that the window start temperature value is i The average of the 5 values ​​in the window, Indicates that the window start temperature value is i The variance value corresponding to the window; Set the threshold to , calculate the absolute error between the window variance value at the current temperature and the window variance value at the previous moment. If the absolute error is greater than the threshold, the first sampling point of the current temperature window is determined to be the first segmentation point; that is: Otherwise, continue sliding the window until the absolute error is greater than the threshold, and the first sampling point of the current window is obtained as the first segmentation point, and so on, to obtain other segmentation points until the window sliding ends; Assume that after the sliding is completed, the curve is divided into k Section, when k ≥3, only 3 segments are retained, that is, the 2 segmentation points with the largest window variance are taken; when k <3, it is expanded to 3 segments, that is, continue to add the windows with the largest window variance value as segmentation points.

5. The passive RFID temperature measurement method based on numerical fitting according to claim 4 is characterized in that: The step of establishing a mapping relationship between duration and temperature for each segment includes: For the AB segment, i.e. the low temperature segment, a univariate quadratic regression model is used for fitting. Therefore, a univariate quadratic fitting model is constructed: ;in, , and are the unknown coefficients of the linear fitting model, p 1 is the temperature value, For temperature p 1; according to the collected sample data, the fitting function M1 is calculated by the least square method; then the fitting goodness of the data is calculated using the fitting function M1 , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, and thus the parameters are obtained. , and Otherwise, re-collect data until the model meets the goodness of fit requirements; get the parameter , and After the value of and duration , you can get the temperature value p 1.

6. The passive RFID temperature measurement method based on numerical fitting according to claim 5, characterized in that: The step of establishing a mapping relationship between duration and temperature for each segment further includes: For the BC section, i.e. the medium temperature section, the mechanism model is used. To fit, where a,b,c are 3 constants. In order to obtain a,b,c Three constants; the following steps are used to achieve: Minimize the objective function: , according to the sample data collected under this label at this temperature after screening , the fitting function M2 is calculated by the least square method; Use the fitting function to find the goodness of fit of the data , if and only if , then the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; get the parameters a,b,c Then according to the formula , according to the duration The temperature value can be directly calculated p 2.

7. The passive RFID temperature measurement method based on numerical fitting according to claim 6, characterized in that: The step of establishing a mapping relationship between duration and temperature for each segment further includes: For the CD segment, a univariate quadratic regression model is used for fitting, and a univariate quadratic fitting model is constructed: Construct a univariate quadratic fitting model: ;in, , and are the unknown coefficients of the linear fitting model, p 3 is the temperature value, For temperature p 3. Duration of the next attack; According to the sample data collected under the temperature and the label after screening, the fitting function M3 is calculated by the least square method; then, the fitting goodness of the data is calculated using the fitting function M3 , if and only if , then it is considered that the fitting degree of the fitting function meets the requirements, otherwise, re-collect data until the model meets the goodness of fit requirements; Get parameters , and Then, according to the formula , according to the duration You can get the temperature value p 3.

8. The passive RFID temperature measurement method based on numerical fitting according to claim 2, characterized in that: The method of combining the fitting function and the ambient temperature value to set an early warning of the abnormal state of the tag includes: Set duration thresholds T1 and T2, and T1< T2. If the duration is less than T1, call model M1' to get the temperature value. p 1; If the duration is greater than or equal to T1 and less than or equal to T2, call model M2' to get the temperature value p 2; When the duration is greater than T2, call model M3' to get the temperature value p 3； Here, the temperature value obtained in this step is recorded as P; Setting the maximum duration threshold th and minimum duration threshold tl , when the duration is less than tl or greater than th When the label is abnormal, it is judged that the label is in abnormal state and a prompt is given to replace the label; Collect ambient temperature values ​​through the reader and set temperature measurement deviation warning values td , when the difference between the temperature value P and the ambient temperature value is greater than td When an alarm is triggered, immediately check the tag status and device status, fill in and report the corresponding fault type and cause.

9. A passive RFID temperature measurement system based on numerical fitting, characterized in that: The system includes: The data acquisition module is used to attach a passive RFID tag to a constant temperature heating container to sense the temperature. The sensing temperature range is from -30°C to 120°C, with a step length of 5°C. That is, every time the temperature rises by 5°C, the temperature and related data are automatically collected and recorded in a fixed format, and automatically uploaded and stored in the database; A preprocessing module, used to draw a scatter plot of the relationship between the temperature under the current tag and the duration at the current temperature, and to remove the abnormal temperature data collected, wherein the duration at the current temperature is the time span from the decay of the initial power supply voltage of the passive RFID tag to the voltage threshold when the volatile data is lost; The optimal segmentation module is used to draw the relationship between the retained temperature and duration into a relationship curve, and to perform optimal segmentation on the curve using the window variance; A fitting module is used to record the segments obtained according to the optimal segmentation method as low-temperature segments, medium-temperature segments and high-temperature segments, and establish a mapping relationship between duration and temperature for each segment. The fitting functions corresponding to the low-temperature segment, medium-temperature segment and high-temperature segment are recorded as M1, M2 and M3 respectively. The early warning module is used to combine the fitting function and the ambient temperature value to set an early warning for abnormal label status.

10. The passive RFID temperature measurement system based on numerical fitting according to claim 9, characterized in that: In the fitting module, after establishing a mapping relationship between duration and temperature for each segment, the following steps are further included: Since tags with the same model have similar curve patterns and fitting parameters, the duration set {( T i , p i )},in, T i For the i Passive tags at temperature p The duration of the following p i For the i The current temperature of the passive tag; The average values ​​of the fitting parameters under multiple different labels are calculated as the new fitting parameters of the model, and the corresponding fitting functions are recorded as M1', M2', and M3' respectively.

Images