Small noninvasive blood glucose detection system and method based on frequency shift and amplitude constraint

Abstract

The invention relates to the technical field of biomedical detection equipment, and provides a small noninvasive blood glucose detection system and method based on frequency shift and amplitude constraint, the small noninvasive blood glucose detection system and method based on frequency shift and amplitude constraint comprises a field programmable gate array (FPGA), a digital-to-analog converter (DAC) and a voltage-controlled oscillator (VCO), the system comprises a double-frequency complementary split-ring resonator (CSRR) sensor, a detector, an analog-to-digital converter (ADC), a liquid crystal display (LCD), a nixie tube and a power supply module for supplying power, after frequency sweeping is completed, the detector output voltage data stored in the RAM is converted into a power value through calculation, the amplitude of the transmission coefficient of the sensor is obtained according to an S21 parameter formula, and the FGPA controls a nixie tube to display the resonant frequency and the resonant depth of the sensor according to a key. The method is mainly applied to design and manufacturing occasions of medical equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of biomedical detection equipment, and in particular to a small non-invasive blood sugar detection system and method based on frequency shift and amplitude constraints. Background Art

[0002] Diabetes is a metabolic disease characterized by high blood sugar and accompanied by multiple complications. Testing blood sugar concentration can effectively help diabetics manage themselves. Currently, invasive blood sugar concentration testing methods are commonly used to test blood sugar concentration, but this method brings pain and infection risks to diabetic patients. In addition, frequent invasive blood sugar testing will lead to a decrease in diabetic patients' compliance with medical advice, a decrease in the level of self-management, and thus a worsening of the disease.

[0003] Non-invasive blood glucose detection methods overcome the shortcomings of invasive blood glucose detection methods. They do not cause pain or discomfort during blood glucose measurement, making the blood glucose detection process safer and more comfortable for diabetic patients. Among the many non-invasive blood glucose detection methods, microwave detection methods show considerable performance and low manufacturing cost, and microwave sensors are small in size and suitable for integration into wearable glucose sensing systems. Microwave sensors detect blood glucose concentration mainly through changes in resonant frequency and resonant depth. Blood glucose concentration detection based on changes in resonant frequency has higher accuracy in predicting low glucose concentrations, while blood glucose concentration detection based on changes in resonant depth has higher accuracy in predicting high glucose concentrations.

[0004] Currently, vector network analyzers (VNA) are mainly used to detect the resonant frequency and resonant depth of sensors. VNA is very accurate in measuring the resonant frequency and resonant depth of sensors, but the high price and large size of VNA have greatly hindered the popularization of non-invasive blood glucose testing based on microwave sensors. Summary of the invention

[0005] In view of the need for popularization of existing non-invasive blood glucose detection technology, the present invention aims to propose a small non-invasive blood glucose detection system and method based on frequency shift and amplitude constraints. The technical scheme adopted by the present invention is as follows: a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints, including: a field programmable logic gate array FPGA, a digital-to-analog converter DAC, a voltage-controlled oscillator VCO, a dual-frequency complementary open resonant ring CSRR sensor, a detector, an analog-to-digital converter ADC, a liquid crystal display LCD, a digital tube, and a power supply module responsible for power supply;

[0006] FPGA controls DAC to output a frequency sweep voltage signal, which is used as a tuning voltage of the VCO, so that the VCO generates a microwave signal corresponding to the tuning voltage;

[0007] The microwave signal generated by the VCO is input into the microwave sensor. The detector measures the power of the microwave signal output by the sensor and converts the power value into an analog voltage signal and outputs it to the ADC. The voltage signal is converted into a digital signal by the ADC module and read by the FPGA. The read data is stored in the FPGA's random access memory RAM. After the storage is completed, the FPGA increases the DAC output voltage, and the ADC module reads the new analog voltage signal. This is repeated until the DAC output voltage reaches the maximum value of the VCO input tuning voltage, that is, the system completes the frequency sweep of the sensor's working frequency band.

[0008] Transmission coefficient S of the sensor 21 The parameters are as follows:

[0009]

[0010] Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor. After the frequency sweep is completed, the detector output voltage data stored in the RAM is converted into a power value through calculation. 21 The parameter formula is used to calculate the amplitude of the sensor's transmission coefficient. FGPA controls the LCD to display the sensor S 21 Amplitude curve of parameter, S 21 The frequency and amplitude at the parameter resonance peak are the resonance frequency and resonance depth of the sensor. FGPA controls the digital tube to display the resonance frequency and resonance depth of the sensor according to the key.

[0011] The dual-frequency complementary open resonant ring CSRR sensor comprises an opener / combiner microstrip line, a substrate, and a complementary open resonant ring unit, wherein: the opener microstrip line / combiner microstrip line is etched on the copper layer at the top of the substrate, and the opener / combiner microstrip line consists of two combiner microstrip lines and two open microstrip lines; two double-ring circular dual-frequency complementary open resonant ring CSRR units and two single-ring circular dual-frequency complementary open resonant ring CSRR units are etched on the copper layer at the bottom of the substrate, the double-ring CSRR unit consists of two circular concentric grooves with symmetrical openings, and the single-ring CSRR unit consists of one circular opening groove; when a microwave signal is applied to the microstrip line, if the symmetry line of the CSRR unit is perpendicular to the axis of the microstrip line, the microstrip line provides electric field excitation for the CSRR unit; if the symmetry line of the CSRR unit is parallel to the axis of the microstrip line, the microstrip line provides electric field excitation and magnetic field excitation for the CSRR unit; the symmetry lines of the four CSRR units are perpendicular to the axis of the opener / combiner microstrip line.

[0012] A small non-invasive blood glucose detection method based on frequency shift and amplitude constraints is implemented using the aforementioned system. The steps are as follows:

[0013] (1) Place the finger on the dual-frequency CSRR sensor. The detection system applies a microwave signal to the sensor to measure the sensor's transmission coefficient S. 21 , the transmission coefficient S 21 Two resonance peaks are displayed, and the minimum points of the two resonance peaks, i.e., the transmission zero points, are named resonance mode 1 and resonance mode 2. The resonance depth of the sensor resonance mode 1 and the resonance frequency of the resonance mode 2 are selected as constraints for data fitting to construct a blood glucose concentration prediction model, and the expression of the prediction model is as follows:

[0014]

[0015] Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1, the resonance frequency of resonance mode 2, and w 0 is the constant to be fitted, w s1 , w f2 is the weight corresponding to each independent variable;

[0016] (2) After the blood glucose concentration prediction model is constructed, the finger is placed on the microwave sensor, and the measured resonance frequency and resonance depth are imported into the blood glucose concentration prediction model to calculate the current blood glucose concentration.

[0017] The detailed steps are as follows:

[0018] (1) After the system is powered on, the ADC and DAC are initialized first. After initialization, the FPGA sets the initial output voltage of the DAC to 0V. When the FPGA completes the setting of the DAC, it sends a data transmission completion signal to the ADC, which is used as the ADC start signal. The FGPA controls the ADC to read the output signal of the detector and stores the read data in the FPGA's RAM. The above operation is repeated until the frequency sweep is completed.

[0019] (2) After the frequency sweep is completed, the excitation voltage value can be converted into frequency according to the mathematical model of the detector output voltage amplitude and the sensor output power. The relationship between the VCO tuning voltage and the output frequency is as follows:

[0020]

[0021] Among them, f VCO is the voltage output by the VCO, in GHz; V T The unit of VCO tuning voltage is V;

[0022] (3) Sensor S 21 The parameters are expressed as:

[0023]

[0024] Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor, P IN Taking the typical output power value of VCO as 11dbm, when the input power of the detector is between 0-10dBm, the relationship between the power of the signal received by the detector and the input voltage of the ADC module is expressed as:

[0025] P OUT =28.6V ADC -33.6

[0026] According to the above formula, the S of the sensor at the corresponding frequency point can be calculated. 21 Parameters; S that traverses all frequency points 21 The amplitude of the parameter is used to find the excitation voltage and resonance depth corresponding to the resonance point. The display content on the digital tube is controlled by the key. After the traversal is completed, the DAC re-outputs the initial sweep voltage signal.

[0027] (4) Place the index finger of the person to be tested in the gap of the pearl cotton foam board, start timing after eating, record the resonant frequency and resonant depth of the sensor every 5 minutes, and use a blood glucose meter to perform an invasive measurement. The blood glucose concentration of the sensor and the resonance characteristics of the sensor are measured by the system within 60 minutes after the volunteer eats;

[0028] (5) The data set recorded in the measurement is divided into a training data set and a test data set for multiple linear regression analysis, where the data sampled at 0, 10, 20, 30, 40, 50, and 60 minutes are used as the training data set, and the data sampled at 5, 15, 25, 35, 45, and 55 minutes are used as the test data set;

[0029] (6) The resonant frequency and resonant depth of resonance mode 1 and the resonant frequency of resonance mode 2 are selected as features, and the blood glucose concentration is used as label data to fit the multivariate linear regression equation. The prediction formula of blood glucose concentration is as follows:

[0030] y=w 0 +w s1 x s1 +w f2 x f2

[0031] Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1, the resonance frequency of resonance mode 2, and w 0 is the constant to be fitted, w s1 , w f2 is the weight corresponding to each independent variable;

[0032] The characteristics and beneficial effects of the present invention are:

[0033] (1) The present invention uses resonance depth and resonance frequency to predict blood glucose concentration, thereby improving the accuracy of blood glucose concentration measurement.

[0034] (2) The present invention measures the S of the microwave sensor 21 The parameters are used to predict blood glucose concentration with little interference.

[0035] (3) The present invention realizes the measurement of the resonant frequency and resonant depth of the sensor at a relatively low cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is the structural block diagram of a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints;

[0037] Figure 2 Schematic diagram of the dual-frequency CSRR sensor structure for non-invasive blood glucose detection provided by the present invention: (a) top structure; (b) bottom structure.

[0038] Figure 3 is the S21 parameter of the dual-frequency CSRR sensor under no-load condition in full-wave simulation

[0039] Figure 4 This is the working flow chart of a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints;

[0040] Figure 5 This is a physical picture of a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints;

[0041] Figure 6 Blood glucose concentration of volunteers within 60 minutes after eating and the resonance characteristics of the sensor measured by the system (a) Comparison between the resonance depth of the microwave sensor and the blood glucose concentration; (a) Comparison between the resonance frequency of the microwave sensor resonance mode 2 and the blood glucose concentration;

[0042] Figure 7 Predict blood glucose concentration and reference blood glucose concentration for a small non-invasive blood glucose monitoring system based on frequency shift and amplitude constraints. DETAILED DESCRIPTION

[0043] In view of the need for popularization of existing non-invasive blood glucose detection technology, the present invention aims to propose a small non-invasive blood glucose detection system and method based on frequency shift and amplitude constraints. The technical scheme adopted by the present invention is as follows: a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints, including: a field programmable logic gate array FPGA, a digital-to-analog converter DAC, a voltage-controlled oscillator VCO, a dual-frequency complementary open resonant ring CSRR sensor, a detector, an analog-to-digital converter ADC, a liquid crystal display LCD, a digital tube, and a power supply module responsible for power supply;

[0044] FPGA controls DAC to output a frequency sweep voltage signal, which is used as a tuning voltage of the VCO, so that the VCO generates a microwave signal corresponding to the tuning voltage;

[0045] The microwave signal generated by the VCO is input into the microwave sensor. The detector measures the power of the microwave signal output by the sensor and converts the power value into an analog voltage signal and outputs it to the ADC. The voltage signal is converted into a digital signal by the ADC module and read by the FPGA. The read data is stored in the FPGA's random access memory RAM. After the storage is completed, the FPGA increases the DAC output voltage, and the ADC module reads the new analog voltage signal. This is repeated until the DAC output voltage reaches the maximum value of the VCO input tuning voltage, that is, the system completes the frequency sweep of the sensor's working frequency band.

[0046] Transmission coefficient S of the sensor 21 The parameters are as follows:

[0047]

[0048] Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor. After the frequency sweep is completed, the detector output voltage data stored in the RAM is converted into a power value by calculation. According to the above formula, the amplitude of the sensor transmission coefficient can be obtained. The FGPA controls the LCD to display the sensor S 21 Amplitude curve of the parameter. S 21 The frequency and amplitude at the parameter resonance peak are the resonance frequency and resonance depth of the sensor. FGPA controls the digital tube to display the resonance frequency and resonance depth of the sensor according to the key.

[0049] The dual-frequency complementary open resonant ring CSRR sensor comprises an opener / combiner microstrip line, a substrate, and a complementary open resonant ring unit, wherein: the opener microstrip line / combiner microstrip line is etched on the copper layer at the top of the substrate, and the opener / combiner microstrip line consists of two combiner microstrip lines and two open microstrip lines; two double-ring circular dual-frequency complementary open resonant ring CSRR units and two single-ring circular dual-frequency complementary open resonant ring CSRR units are etched on the copper layer at the bottom of the substrate, the double-ring CSRR unit consists of two circular concentric grooves with symmetrical openings, and the single-ring CSRR unit consists of one circular opening groove; when a microwave signal is applied to the microstrip line, if the symmetry line of the CSRR unit is perpendicular to the axis of the microstrip line, the microstrip line provides electric field excitation for the CSRR unit; if the symmetry line of the CSRR unit is parallel to the axis of the microstrip line, the microstrip line provides electric field excitation and magnetic field excitation for the CSRR unit; the symmetry lines of the four CSRR units are perpendicular to the axis of the opener / combiner microstrip line.

[0050] A small non-invasive blood glucose detection method based on frequency shift and amplitude constraints is implemented using the aforementioned system. The steps are as follows:

[0051] (1) Place the finger on the dual-frequency CSRR sensor. The detection system applies a microwave signal to the sensor's microstrip line to measure the sensor's transmission coefficient S. 21 , the transmission coefficient S 21 Two resonance peaks are displayed, and the minimum points of the two resonance peaks, i.e., the transmission zero points, are named resonance mode 1 and resonance mode 2; since the blood glucose concentration detection based on the change of the resonance frequency has a higher accuracy in predicting low glucose concentrations, and the blood glucose concentration detection based on the change of the resonance depth has a higher accuracy in predicting high glucose concentrations, and the sensor has a high resonance frequency detection sensitivity in the first resonance mode and a high resonance depth detection sensitivity in the second resonance mode, the resonance depth of the sensor resonance mode 1 and the resonance frequency of the resonance mode 2 are selected as constraints for data fitting to construct a blood glucose concentration prediction model, and the expression of the prediction model is as follows:

[0052]

[0053] Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1, the resonance frequency of resonance mode 2, and w 0 is the constant to be fitted, w s1 , w f2 is the weight corresponding to each independent variable;

[0054] (2) After the blood glucose concentration prediction model is constructed, the finger is placed on the microwave sensor, and the measured resonance frequency and resonance depth are imported into the blood glucose concentration prediction model to calculate the current blood glucose concentration.

[0055] The detailed steps are as follows:

[0056] (1) After the system is powered on, the ADC and DAC are initialized first. After initialization, the FPGA sets the initial output voltage of the DAC to 0V. When the FPGA completes the setting of the DAC, it sends a data transmission completion signal to the ADC, which is used as the ADC start signal. The FGPA controls the ADC to read the output signal of the detector and stores the read data in the FPGA's RAM. The above operation is repeated until the frequency sweep is completed.

[0057] (2) After the frequency sweep is completed, the excitation voltage value can be converted into frequency according to the mathematical model of the detector output voltage amplitude and the sensor output power. The relationship between the VCO tuning voltage and the output frequency is as follows:

[0058]

[0059] Among them, f VCO is the voltage output by the VCO, in GHz; V T The unit of VCO tuning voltage is V;

[0060] (3) Sensor S 21 The parameters are expressed as:

[0061]

[0062] Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor, P IN Taking the typical output power value of VCO as 11dbm, when the input power of the detector is between 0-10dBm, the relationship between the power of the signal received by the detector and the input voltage of the ADC module is expressed as:

[0063] P OUT =28.6V ADC -33.6

[0064] According to the above formula, the S of the sensor at the corresponding frequency point can be calculated. 21 Parameters; S that traverses all frequency points 21 The amplitude of the parameter is used to find the excitation voltage and resonance depth corresponding to the resonance point. The display content on the digital tube is controlled by the key. After the traversal is completed, the DAC re-outputs the initial sweep voltage signal.

[0065] (4) Place the index finger of the person to be tested in the gap of the pearl cotton foam board, start timing after eating, record the resonant frequency and resonant depth of the sensor every 5 minutes, and use a blood glucose meter to perform an invasive measurement. The blood glucose concentration of the sensor and the resonance characteristics of the sensor are measured by the system within 60 minutes after the volunteer eats;

[0066] (5) The data set recorded in the experiment was divided into a training data set and a test data set for multiple linear regression analysis, where the data sampled at 0, 10, 20, 30, 40, 50, and 60 minutes were used as the training data set, and the data sampled at 5, 15, 25, 35, 45, and 55 minutes were used as the test data set;

[0067] (6) The resonant frequency and resonant depth of resonance mode 1 and the resonant frequency of resonance mode 2 are selected as features, and the blood glucose concentration is used as label data to fit the multivariate linear regression equation. The prediction formula of blood glucose concentration is as follows:

[0068]

[0069] Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1, the resonance frequency of resonance mode 2, and w 0 is the constant to be fitted, w s1 , w f2 is the weight corresponding to each independent variable;

[0070] In order to make the purpose, technical solution and advantages of the present invention clearer, the following describes the implementation of the present invention through specific examples. It should be understood that the present invention can also be implemented or applied through other different specific implementations, and is not intended to limit the present invention. A specific technical solution of a small non-invasive blood glucose detection system based on frequency shift and amplitude constraints is as follows:

[0071] (1) The structural block diagram of the non-invasive blood glucose detection system based on microwave sensor is as follows Figure 1 As shown, the system consists of FPGA, digital-to-analog converter, VCO, dual-frequency CSRR sensor, detector, ADC, LCD, digital tube, and power module.

[0072] (2) The structure of the dual-frequency CSRR sensor is as follows Figure 2 As shown in the figure, gray represents metal and white represents the etched part. FR4 is selected as the sensor substrate. The sensor substrate is 40mm long, 20mm wide and 0.8mm thick. The opener / combiner microstrip line is etched on the copper layer on the top of the substrate with a thickness of 35μm. It consists of two widths of W. C The combiner microstrip line and two width W SIn this example, the width of the combiner microstrip line is 1.5 mm, and the width of the splitter microstrip line is 0.3 mm.

[0073] (3) Two double-ring CSRR units and two single-ring CSRR units are etched on the copper layer at the bottom of the substrate with a thickness of 35 μm. The horizontal distance between the two circular double-ring CSRR units is b, and the vertical distance between the two circular single-ring CSRR units is c. The circular double-ring CSRR unit consists of two circular concentric grooves with symmetrical openings with a spacing of t, and the diameter of the circular outer ring is a. 1 , the width of the opening is g, and the width of the groove between the inner and outer metals is s. The circular single-ring CSRR unit consists of a circular opening groove, and the diameter of the circular outer ring is a 2 , the width of the opening is g, and the width of the groove between the inner and outer metals is s.

[0074] (4) S of the sensor in no-load state 21 Parameters such as Figure 3 As shown in the figure, the sensor shows two transmission zeros, which are the same as the analysis results of the sensor equivalent circuit. The two transmission zeros are named resonance mode 1 and resonance mode 2. The resonance frequency and resonance depth of resonance mode 1 are 4.51 GHz and 34.3 dB respectively, and the resonance frequency and resonance depth of resonance mode 2 are 6.35 GHz and 20.1 dB respectively. Compared with resonance mode 2, resonance mode 1 has a larger resonance depth.

[0075] (5) A VCO with an input tuning voltage range of 0-10V, an output frequency of 1.73-2.83GHz, and a power of 11dBm is selected; a DAC module is selected to output a voltage of 0-10V, and its output accuracy reaches 20 bits; the resonance depth of the sensor varies from -2.4dB to -1.8dB when it is working, and a detector with an input frequency range of DC to 6GHz and a nominal range of -30dBm to +15dBm for the root mean square power measurement function is selected to perform power detection on the transmission signal; a 16-bit successive approximation ADC is selected to sample the output signal of the detector;

[0076] (6) The working flow diagram of the non-invasive blood glucose detection system based on microwave sensor is as follows Figure 4 As shown in the figure, after the system is powered on, the ADC and DAC are initialized first. After initialization, the FPGA sets the initial output voltage of the DAC to 0V. When the FPGA completes the setting of the DAC, the FPGA sends a data transmission completion signal to the ADC, which is used as the ADC start signal. The FGPA controls the ADC to read the output signal of the detector and stores the read data in the FPGA's RAM. The above operation is repeated until the frequency sweep is completed.

[0077] (7) After the frequency sweep is completed, the excitation voltage value can be converted into frequency according to the mathematical model of the detector output voltage amplitude and the sensor output power. The relationship between the VCO tuning voltage and the output frequency is shown as follows:

[0078]

[0079] Among them, f VCO is the voltage output by the VCO, in GHz; V T The unit of VCO tuning voltage is V;

[0080] (8) Sensor S 21 The parameters can be expressed as:

[0081]

[0082] Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor, P IN Taking the typical output power value of VCO as 11dbm, when the input power of the detector is between 0-10dBm, the relationship between the power of the signal received by the detector and the input voltage of the ADC module is expressed as:

[0083] P OUT =28.6V ADC -33.6

[0084] According to the above formula, the S of the sensor at the corresponding frequency point can be calculated. 21 Parameters; S that traverses all frequency points 21 The amplitude of the parameter is used to find the excitation voltage and resonance depth corresponding to the resonance point, and the display content on the digital tube is controlled according to the key. After the traversal is completed, the DAC re-outputs the initial sweep voltage signal.

[0085] (9) The actual picture of the non-invasive blood glucose detection system based on microwave sensor is as follows Figure 5 As shown, the power supply of the entire system is provided by a lithium battery. The power management module steps down the voltage of the lithium battery and distributes it to each module. The RF modules are connected by coaxial cables.

[0086] (10) This experiment conducted continuous blood sugar prediction on a healthy volunteer subject. The experiment began after the volunteer subject had rested in his seat for 15 minutes. After the experiment began, the volunteer subject placed his index finger on the empty space of the pearl cotton foam board. The timing began after the meal was completed. The resonant frequency and resonant depth of the sensor were recorded every 5 minutes, and an invasive measurement was performed using a blood glucose meter. The blood sugar concentration of the sensor within 60 minutes after the volunteer had eaten and the resonance characteristics of the sensor measured by the system were as follows: Figure 6 shown.

[0087] (11) The data set recorded in the experiment is divided into a training data set and a test data set for multivariate linear regression analysis, where the data sampled at 0, 10, 20, 30, 40, 50, and 60 minutes are used as the training data set, and the data sampled at 5, 15, 25, 35, 45, and 55 minutes are used as the test data set.

[0088] (12) The Pearson correlation coefficient is used to measure the correlation between two variables. The calculation formula of the Pearson correlation coefficient is as follows:

[0089]

[0090] Among them, x i is the value of variable 1, y i is the value of variable 2, is the mean of variable 1, is the mean of variable 2. The Pearson correlation coefficient between the resonance depth of resonance mode 1 and blood glucose concentration is 0.736, and the Pearson correlation coefficient between the resonance frequency of resonance mode 2 and blood glucose concentration is 0.798. This shows that the resonance depth of resonance mode 1 and the resonance frequency of resonance mode 2 have a good linear relationship with blood glucose concentration. The resonance frequency and resonance depth of resonance mode 1 and the resonance frequency of resonance mode 2 are selected as features, and blood glucose concentration is used as label data to fit the multivariate linear regression equation. The prediction formula for blood glucose concentration is as follows:

[0091]

[0092] Where y is the blood glucose concentration in mg / dL, and They correspond to the resonance depth of sensor resonance mode 1 and the resonance frequency of resonance mode 2, respectively. The unit is dB, The unit is GHz.

[0093] Among them, the resonance depth of the sensor resonance mode 1 and the resonance frequency of the resonance mode 2 are selected as constraints for data fitting in the computer, and a blood glucose concentration prediction model is constructed. The measured resonance frequency and resonance depth are introduced into the blood glucose concentration prediction model to calculate the current blood glucose concentration.

[0094] The comparison between the predicted blood glucose concentration obtained by the regression model and the reference blood glucose concentration is as follows: Figure 7 As shown, from the changing trend in the figure, it can be seen that the change curves of the predicted blood glucose concentration are highly consistent with the reference blood glucose concentration, and the root mean square error and average relative error of the regression model are 2.6 and 2.3% respectively.

[0095] The above embodiments are only used to illustrate the technical solutions of the present invention. Those skilled in the art will easily understand that the above are only preferred embodiments of the present invention and are not limitations of the present invention. The present invention is not limited to the above examples. Any modification or equivalent replacement within the principle of the technical solution of the present invention, as long as it meets the requirements of the method of the present invention, should be included in the protection scope of the present invention.

Claims

1. A small non-invasive blood glucose detection system based on frequency shift and amplitude constraint, characterized in that: include: Field programmable gate array FPGA, digital-to-analog converter DAC, voltage-controlled oscillator VCO, dual-frequency complementary open-ring resonant ring CSRR sensor, detector, analog-to-digital converter ADC, liquid crystal display LCD, digital tube, and power supply module responsible for power supply; FPGA controls DAC to output a frequency sweep voltage signal, which is used as a tuning voltage of the VCO, so that the VCO generates a microwave signal corresponding to the tuning voltage; The microwave signal generated by the VCO is input into the microwave sensor. The detector measures the power of the microwave signal output by the sensor and converts the power value into an analog voltage signal and outputs it to the ADC. The voltage signal is converted into a digital signal by the ADC module and read by the FPGA. The read data is stored in the FPGA's random access memory RAM. After the storage is completed, the FPGA increases the DAC output voltage, and the ADC module reads the new analog voltage signal. This is repeated until the DAC output voltage reaches the maximum value of the VCO input tuning voltage, that is, the system completes the frequency sweep of the sensor's working frequency band. Transmission coefficient S of the sensor 21 The parameters are as follows: Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor. After the frequency sweep is completed, the detector output voltage data stored in the RAM is converted into a power value through calculation. 21 The parameter formula is used to calculate the amplitude of the sensor's transmission coefficient. FGPA controls the LCD to display the sensor S 21 Amplitude curve of parameter, S 21 The frequency and amplitude at the parameter resonance peak are the resonance frequency and resonance depth of the sensor. FGPA controls the digital tube to display the resonance frequency and resonance depth of the sensor according to the key.

2. The small non-invasive blood glucose detection system based on frequency shift and amplitude constraint as claimed in claim 1, characterized in that: The dual-frequency complementary open resonant ring CSRR sensor comprises an opener / combiner microstrip line, a substrate, and a complementary open resonant ring unit, wherein: the opener microstrip line / combiner microstrip line is etched on the copper layer at the top of the substrate, and the opener / combiner microstrip line consists of two combiner microstrip lines and two open microstrip lines; two double-ring circular dual-frequency complementary open resonant ring CSRR units and two single-ring circular dual-frequency complementary open resonant ring CSRR units are etched on the copper layer at the bottom of the substrate, the double-ring CSRR unit consists of two circular concentric grooves with symmetrical openings, and the single-ring CSRR unit consists of one circular opening groove; when a microwave signal is applied to the microstrip line, if the symmetry line of the CSRR unit is perpendicular to the axis of the microstrip line, the microstrip line provides electric field excitation for the CSRR unit; if the symmetry line of the CSRR unit is parallel to the axis of the microstrip line, the microstrip line provides electric field excitation and magnetic field excitation for the CSRR unit; the symmetry lines of the four CSRR units are perpendicular to the axis of the opener / combiner microstrip line.

3. A small non-invasive blood glucose detection method based on frequency shift and amplitude constraint, characterized in that: Using the above system, the steps are as follows: (1) Place the finger on the dual-frequency CSRR sensor. The detection system applies a microwave signal to the sensor to measure the sensor's transmission coefficient S. 21 , the transmission coefficient S 21 Two resonance peaks are displayed, and the minimum points of the two resonance peaks, i.e., the transmission zero points, are named resonance mode 1 and resonance mode 2. The resonance depth of the sensor resonance mode 1 and the resonance frequency of the resonance mode 2 are selected as constraints for data fitting to construct a blood glucose concentration prediction model, and the expression of the prediction model is as follows: Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1 and the resonance frequency of resonance mode 2 respectively. w0 is the constant to be fitted, and w s1 , w f2 is the weight corresponding to each independent variable; (2) After the blood glucose concentration prediction model is constructed, the finger is placed on the microwave sensor, and the measured resonance frequency and resonance depth are imported into the blood glucose concentration prediction model to calculate the current blood glucose concentration.

4. The small non-invasive blood glucose detection method based on frequency shift and amplitude constraint as claimed in claim 3, characterized in that: The detailed steps are as follows: (1) After the system is powered on, the ADC and DAC are initialized first. After initialization, the FPGA sets the initial output voltage of the DAC to 0V. When the FPGA completes the setting of the DAC, it sends a data transmission completion signal to the ADC, which is used as the ADC start signal. The FGPA controls the ADC to read the output signal of the detector and stores the read data in the FPGA's RAM. The above operation is repeated until the frequency sweep is completed. (2) After the frequency sweep is completed, the excitation voltage value can be converted into frequency according to the mathematical model of the detector output voltage amplitude and the sensor output power. The relationship between the VCO tuning voltage and the output frequency is as follows: Among them, f VCO is the voltage output by the VCO, in GHz; V T The unit of VCO tuning voltage is V; (3) Sensor S 21 The parameters are expressed as: Among them, S 21 is the transmission coefficient of the sensor in dB, P IN is the power of the VCO input sensor, P IN Taking the typical output power value of VCO as 11dbm, when the input power of the detector is between 0-10dBm, the relationship between the power of the signal received by the detector and the input voltage of the ADC module is expressed as: P OUT =28.6V ADC -33.6 According to the above formula, the S of the sensor at the corresponding frequency point can be calculated. 21 Parameters; S that traverses all frequency points 21 The amplitude of the parameter is used to find the excitation voltage and resonance depth corresponding to the resonance point. The display content on the digital tube is controlled by the key. After the traversal is completed, the DAC re-outputs the initial sweep voltage signal. (4) Place the index finger of the person to be tested in the gap of the pearl cotton foam board, start timing after eating, record the resonant frequency and resonant depth of the sensor every 5 minutes, and use a blood glucose meter to perform an invasive measurement. The blood glucose concentration of the sensor and the resonance characteristics of the sensor are measured by the system within 60 minutes after the volunteer eats; (5) The data set recorded in the measurement is divided into a training data set and a test data set for multiple linear regression analysis, where the data sampled at 0, 10, 20, 30, 40, 50, and 60 minutes are used as the training data set, and the data sampled at 5, 15, 25, 35, 45, and 55 minutes are used as the test data set; (6) The resonant frequency and resonant depth of resonance mode 1 and the resonant frequency of resonance mode 2 are selected as features, and the blood glucose concentration is used as label data to fit the multivariate linear regression equation. The prediction formula of blood glucose concentration is as follows: Where y is the blood glucose concentration x s1 , x f2 They correspond to the resonance depth of sensor resonance mode 1 and the resonance frequency of resonance mode 2 respectively. w0 is the constant to be fitted, and w s1 , w f2 is the weight corresponding to each independent variable.

Images