An adaptive calibration system and method for a combustible gas sensor

Abstract

The application discloses a kind of self-adapting calibration system and method of combustible gas sensor, suitable for catalytic combustion type combustible gas sensor;System includes basic calibration unit, real-time monitoring unit, drift detection and compensation unit, output adjustment unit;Basic calibration unit is used to carry out the basic calibration of sensor, determines the initial sensitivity coefficient of sensor, zero point offset and response time;Real-time monitoring unit is used to monitor the real-time parameter of sensor under different environmental conditions;Drift detection and compensation unit is used to establish sensor performance parameter drift model, calculate calibration compensation coefficient;Output adjustment unit is used to adjust sensor voltage output value in real time according to calibration compensation coefficient;The application can adapt to environmental change and carry out real-time calibration, effectively compensate sensor drift, improve measurement accuracy and reliability, prolong the service life of sensor, reduce maintenance cost, improve system security.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of gas detection, and particularly relates to an adaptive calibration system and method for a combustible gas sensor. BACKGROUND

[0002] Catalytic combustion type combustible gas sensors are widely used due to their simple structure, rapid response and moderate price. However, in actual application, the existing combustible gas sensors have the following problems: first, the performance of the sensor is easily affected by environmental factors such as temperature and humidity, resulting in deviation of the detection result; second, during long-term use, the performance of the sensor will drift, the sensitivity will decrease, and the zero point will shift, affecting the accuracy and reliability of detection; third, the existing calibration method is usually manual calibration at regular intervals, which not only has a large workload, but also cannot timely respond to dynamic changes in environmental factors and attenuation of sensor performance; in addition, the traditional calibration method cannot distinguish target gas and interfering gas, and cannot effectively eliminate the influence of interfering gas on the measurement result, reducing the selectivity of detection.

[0003] Therefore, it is of great significance to develop an automatic calibration system and method that can adapt to environmental changes and compensate for sensor drift in real time, to improve the measurement accuracy and reliability of combustible gas sensors. SUMMARY

[0004] The purpose of the present application is to overcome the shortcomings of the prior art and provide an adaptive calibration system and method for a combustible gas sensor, so as to improve the measurement accuracy of the sensor, prolong the service life, enhance the environmental adaptability and reduce the maintenance cost.

[0005] In a first aspect, the present application provides an adaptive calibration system for a combustible gas sensor, wherein the sensor is a catalytic combustion type combustible gas sensor, and the system comprises, in sequence, a basic calibration unit, a real-time monitoring unit, a drift detection and compensation unit, and an output adjustment unit.

[0006] Further, the basic calibration unit is used for basic calibration of the sensor, recording the response value of the sensor under standard conditions, establishing a calibration curve of the sensor under standard conditions, and determining the initial sensitivity coefficient, zero point shift and response time of the sensor.

[0007] Further, the real-time monitoring unit is used for monitoring real-time parameters of the sensor under different environmental conditions, wherein the real-time parameters include the resistance value change rate, power consumption fluctuation and thermal characteristic change of the sensor, and corresponding environmental temperature, humidity and interfering gas concentration data are recorded.

[0008] Further, the drift detection and compensation unit is configured to establish a sensor performance parameter drift model according to the acquired data, detect and compensate sensor drift, and calculate a calibration compensation coefficient.

[0009] Further, the output adjustment unit is configured to adjust the sensor voltage output value in real time according to the calibration compensation coefficient.

[0010] Further, the system further comprises:

[0011] The performance prediction unit is configured to establish a sensor performance decay prediction model, analyze the parameter variation trend using historical data, predict parameter drift through mathematical statistical methods, and perform parameter adjustment in advance.

[0012] The reliability evaluation unit is configured to evaluate the reliability of the self-calibration result, calculate the calibration uncertainty, and output a sensor state report.

[0013] Further, the sensor performance parameter drift model is expressed as: P(t|T, H, F) = P 0 exp(-γ×t), where P(t|T, H, F) is the performance parameter at time t, temperature T, humidity H, and interference gas concentration F, P 0 is the initial performance parameter of the sensor, γ is the sensor performance drift rate under environmental conditions, and t is time.

[0014] The sensor performance parameter drift rate γ is represented by a multivariate function:

[0015] γ = γ0+ β T (T-T0) 2 + β H (H-H0) 2 + β F × F, where γ0 is the basic drift rate under standard conditions, obtained through long-term stability testing, with a value range of 0.001-0.005 / month; β T , β H , and β F are the drift influence coefficients of temperature T, humidity H, and interference gas concentration F, respectively, T0 and H0 are the standard temperature and standard humidity, respectively; β T has a value range of β H has a value range of β F has a value range of Determined by the control variable method.

[0016]

[0017] Calibration compensation coefficient: Wherein, S0 is the initial sensitivity coefficient, S c is the current sensitivity coefficient, α T , α H and α F are the compensation coefficients of temperature, humidity and interfering gas respectively, which are calculated by multiple regression analysis.

[0018]

[0019] Δγ is the parameter drift rate change; ΔS is the sensitivity coefficient change; ΔT, ΔH and ΔF are the change amounts of temperature, humidity and interfering gas concentration respectively relative to the standard condition.

[0020] Further, the sensor voltage output value is adjusted in real time according to the calibration compensation coefficient, and the mathematical expression of the calibration method is:

[0021] V c =C c ×(V m -O c )+O0;Wherein, V c is the corrected sensor voltage output value, V m is the current sensor measured voltage output value, O c is the current zero point offset, and O0 is the initial zero point offset.

[0022] Further, the detection method of the interfering gas concentration F is: recording the sensor baseline resistance response R0 in clean air, detecting the interfering gas in the environment using the auxiliary sensor array, distinguishing the target gas and the interfering gas through the pattern recognition algorithm, and calculating the interfering gas concentration F according to the sensor cross-sensitivity matrix;

[0023] The quantitative calculation formula of the interfering gas concentration F is: Wherein, R i is the sensor resistance response value when the interfering gas exists; k i is the weight coefficient of the i-th interfering gas, which is obtained by experiment calibration: is the concentration value of the i-th interfering gas under the standard concentration, and R is the corresponding sensor resistance response value under the standard concentration.

[0024] Further, the method for establishing the performance attenuation prediction model comprises: establishing a sensor historical data time sequence, extracting characteristic parameters and performing data normalization, using the exponential smoothing method to analyze the trend of historical data, and predicting the sensor parameter change trend in the future period;

[0025] The mathematical expression of the sensor parameter change trend in the future period is:

[0026] Among them, P t+Δt P represents the predicted sensor parameters at time t+Δt. t T is the current parameter value. t The trend factor is α, and the smoothing coefficients are α and β, with values ​​ranging from 0.2 to 0.3 and 0.1 to 0.2, respectively. They are determined by minimizing the prediction error of historical data.

[0027] in, The actual parameter value at the i-th time point. Here, N represents the parameter values ​​predicted using the exponential smoothing method, and N is the total number of data collection time points.

[0028] Furthermore, the step of calibration uncertainty assessment includes:

[0029] Calculate repeatability uncertainty u r :

[0030] Calculate the system uncertainty u s :

[0031] Calculate the expanded uncertainty U:

[0032] Among them, V i Let i be the value of the i-th measurement. The average value is denoted by m, where m is the number of measurements, and x is the average value. j As an influencing factor, u(x) j ) represents the standard uncertainty of each influencing factor, k p Let k be the inclusion factor. p =2, confidence level greater than 95%.

[0033] Furthermore, the system also includes a lifetime calculation unit for calculating the predicted lifetime of the sensor:

[0034] Where L is the predicted lifetime, L0 is the baseline lifetime under standard conditions (24-36 months), and E a The activation energy is defined as 0.2–0.8 eV, determined based on the sensor material. k is the Boltzmann constant, 8.617 × 10⁻⁶. -5 eV / K, λ is the lifetime influence coefficient of the interfering gas, which is determined by accelerated aging test and has a value range of 0.001-0.01 / (ppm·month), F is the concentration of the interfering gas, and t is the current time, i.e. the time the sensor has been used.

[0035] In a second aspect, the present application provides a combustible gas sensor adaptive calibration method based on the system of the first aspect, the adaptive calibration method comprising:

[0036] Step S1, performing basic calibration of the sensor, recording the sensor response value under standard conditions, establishing the calibration curve of the sensor under standard conditions, and determining the initial sensitivity coefficient, zero point offset and response time of the sensor.

[0037] Step S2, monitoring the real-time parameters of the sensor under different environmental conditions, the real-time parameters including the resistance value change rate, power consumption fluctuation and thermal characteristic change of the sensor, and recording the corresponding environmental temperature, humidity and interference gas concentration data.

[0038] Step S3, according to the obtained data, establishing a sensor performance parameter drift model, detecting and compensating the sensor drift, and calculating the calibration compensation coefficient.

[0039] Step S4, adjusting the sensor voltage output value in real time according to the calibration compensation coefficient.

[0040] Further, the method further comprises:

[0041] Step S5, establishing a sensor performance attenuation prediction model, analyzing the sensor parameter change trend using historical data, predicting the parameter drift through mathematical statistical method, and adjusting the parameters in advance;

[0042] Step S6, performing reliability evaluation of the self-calibration result, calculating the calibration uncertainty, and outputting the sensor state report.

[0043] Compared with the prior art, the present application has the following beneficial effects:

[0044] The present application establishes a sensor performance parameter drift model and a calibration compensation coefficient, realizes real-time compensation of the sensor drift, effectively solves the influence of environmental factor changes and sensor performance attenuation on the measurement accuracy, uses a multivariate function to represent the sensor performance parameter drift rate, comprehensively considers the influence of temperature, humidity and interference gas concentration and other factors, significantly improves the measurement accuracy and stability of the combustible gas sensor, prolongs the service life of the sensor, reduces the maintenance cost, improves the reliability and safety of the system, and has important theoretical significance and application value. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a combustible gas sensor adaptive calibration system composition schematic diagram of the present application;

[0046] Figure 2 It is a combustible gas sensor adaptive calibration method flow chart of the present application. DETAILED DESCRIPTION

[0047] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application are described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the present application.

[0048] Embodiment 1

[0049] As shown in FIG. 1, it is a schematic diagram of a combustible gas sensor adaptive calibration system according to the present application, which comprises, in sequence, a basic calibration unit, a real-time monitoring unit, a drift detection and compensation unit, and an output adjustment unit. Figure 1 These units are connected with each other through a data bus and a control bus, forming a complete calibration system. The system adopts a modular design, facilitating maintenance and upgrading, and can be integrated into a gas detection instrument or used as an independent calibration module. The system hardware adopts a low-power design, and can be powered by a 3.3V or 5V DC power supply, with a power consumption of less than 200mW, suitable for portable and fixed gas detection equipment.

[0050] The basic calibration unit is used for basic calibration of the sensor, recording the sensor response value under standard conditions, establishing a calibration curve of the sensor under standard conditions, and determining the initial sensitivity coefficient, zero point offset and response time of the sensor.

[0051] The unit comprises a high-precision signal acquisition circuit and a standard gas interface. The signal acquisition circuit adopts a 24-bit ADC chip, with a sampling rate of up to 100Hz and a resolution better than 0.1mV, capable of accurately recording the output signal of the sensor under different gas concentrations. The standard gas interface supports connection with an external standard gas generation device or a built-in small gas generation unit, and can generate multiple standard gases for calibration. The calibration process is automatically controlled by a microcontroller, and the least square method is used to calculate the calibration curve parameters, which are stored in a non-volatile memory. For example, the output voltage of a certain sensor under 0, 500, 1000, 2000, 3000 and 5000ppm methane standard gas is 10.2, 260.3, 510.5, 1010.2, 1520.8 and 2530.5mV, respectively. The least square method is used to fit the calibration curve V=0.504xC+10.3mV, wherein the initial sensitivity coefficient S0=0.504mV / ppm and the initial zero point offset O0=10.3mV.

[0052] The unit comprises a high-precision signal acquisition circuit and a standard gas interface. The signal acquisition circuit adopts a 24-bit ADC chip, with a sampling rate of up to 100Hz and a resolution better than 0.1mV, capable of accurately recording the output signal of the sensor under different gas concentrations. The standard gas interface supports connection with an external standard gas generation device or a built-in small gas generation unit, and can generate multiple standard gases for calibration. The calibration process is automatically controlled by a microcontroller, and the least square method is used to calculate the calibration curve parameters, which are stored in a non-volatile memory. For example, the output voltage of a certain sensor under 0, 500, 1000, 2000, 3000 and 5000ppm methane standard gas is 10.2, 260.3, 510.5, 1010.2, 1520.8 and 2530.5mV, respectively. The least square method is used to fit the calibration curve V=0.504xC+10.3mV, wherein the initial sensitivity coefficient S0=0.504mV / ppm and the initial zero point offset O0=10.3mV.

[0053] The real-time monitoring unit is used to monitor the real-time parameters of the sensor under different environmental conditions, including the resistance value change rate, power consumption fluctuation and thermal characteristic change of the sensor, and record the corresponding environmental temperature, humidity and interference gas concentration data; the unit is composed of a sensor parameter monitoring circuit and an environmental parameter monitoring circuit.

[0054] The sensor parameter monitoring circuit adopts a high-precision resistance measurement circuit with a resolution better than 0.01Ω to monitor the resistance change of the sensor in real time; a precision power detection circuit with a resolution better than 0.1mW is used to monitor the power consumption change of the sensor; a thermocouple or infrared temperature measurement element is used to monitor the surface temperature change of the sensor. The environmental parameter monitoring circuit includes an integrated temperature and humidity sensor module, with a temperature measurement range of -40℃ to 85℃ and an accuracy of ±0.3℃; a humidity measurement range of 0-100%RH and an accuracy of ±2%RH. In addition, the unit also includes an auxiliary sensor array composed of 4-8 different types of gas sensors for detecting interference gases in the environment. The data acquisition frequency is adjustable, usually 0.1-1Hz, and all data is stored in the data buffer after filtering and preprocessing for subsequent processing. For example, in a certain industrial environment, the real-time monitoring record shows that the environmental temperature is 32.5℃, the relative humidity is 78.3%RH, the CO concentration is 15ppm, the H2 concentration is 5ppm, the sensor resistance value changes from the initial 8.5Ω to 8.9Ω, and the power consumption decreases from 95mW to 92mW, which will be used for subsequent drift compensation calculation.

[0055] The drift detection and compensation unit is used to establish a sensor performance parameter drift model according to the obtained data, detect and compensate the sensor drift, and calculate the calibration compensation coefficient; it is composed of a high-performance microprocessor and a special algorithm library.

[0056] The microprocessor adopts a 32-bit ARM Cortex-M4 architecture with a working frequency of 120 MHz, built-in DSP and FPU, and has strong data processing capability. The algorithm library includes parameter drift model, multiple regression analysis, pattern recognition and other algorithm modules, which can efficiently perform various calculation tasks. The unit first receives the data collected by the real-time monitoring unit, calculates the drift rate γ under the current environmental conditions according to the sensor performance parameter drift model, and then calculates the calibration compensation coefficient Cc. The calculation process is automatically executed with a period of 1-10 seconds to ensure timely response to environmental changes. For example, when the sensor has been used for 8 months, the current environmental temperature is 35°C (10°C higher than the standard temperature), the relative humidity is 70% RH (20% RH higher than the standard humidity), and the concentration of interfering gas is detected to be 25 ppm, the current drift rate γ = 0.0039 / month is calculated, which is 56% higher than the basic drift rate γ0 = 0.0025 / month under standard conditions, indicating that environmental factors have significantly accelerated the performance degradation of the sensor. Further calculation gives the calibration compensation coefficient Cc = 1.28, which is used for subsequent output adjustment.

[0057] The output adjustment unit is used to adjust the sensor voltage output value in real time according to the calibration compensation coefficient. The output adjustment unit is composed of a signal conditioning circuit and an output interface circuit.

[0058] The signal conditioning circuit uses high-precision operational amplifiers and digital potentiometers to adjust the sensor's original output signal according to the calibration compensation coefficient Cc, eliminating the effects of environmental factors and sensor performance degradation. The output interface circuit provides multiple output methods, including analog voltage output (0-5V or 0-10V), analog current output (4-20mA), and digital communication interface (RS-485, HART or Modbus). The adjustment process is performed in real time with a delay of less than 10ms, ensuring the accuracy and timeliness of the output signal. For example, the sensor measures the methane concentration in a sample, the original output voltage is 500mV, the current zero point offset is 15mV, the initial zero point offset is 10mV, and the calibration compensation coefficient is 1.28, then the corrected output voltage Vc = 1.28 x (500-15) + 10 = 631.8mV, corresponding to the methane concentration of 1254ppm (calculated according to the calibration curve V = 0.504 x C + 10.3mV).

[0059] The system also includes:

[0060] The performance prediction unit is used to establish a sensor performance degradation prediction model, analyze the trend of sensor parameter changes using historical data, predict parameter drift through mathematical statistical methods, and make parameter adjustments in advance; including a data storage module and a prediction analysis module.

[0061] The data storage module uses a large-capacity non-volatile memory (such as EEPROM or Flash), with a capacity of 8MB-32MB, and can store 2-5 years of historical data. The prediction analysis module is based on the exponential smoothing algorithm, and uses the stored historical data to predict the future performance parameters of the sensor. This unit performs prediction calculations regularly (such as every day or every week), generating trend charts and warning information for sensor parameters in the next 1-3 months. For example, by analyzing the sensitivity data of a certain sensor for the past 6 months, the prediction analysis module calculates the smoothing coefficients α = 0.25, β = 0.15, and predicts that the sensor sensitivity will decrease from the current 0.41 mV / ppm to 0.38 mV / ppm in the next 2 months, close to the replacement threshold of 0.35 mV / ppm. The system generates a warning message, suggesting that the user schedule the sensor replacement within 1 month.

[0062] A reliability evaluation unit is used to evaluate the reliability of the self-calibration results, calculate the calibration uncertainty, and output a sensor status report.

[0063] The sensor performance parameter drift rate γ is represented by a multi-element function:

[0064] γ = γ0 + β T (T-T0) 2 + β H (H-H0) 2 + β F × F where γ0 is the basic drift rate under standard conditions, obtained through long-term stability testing, with a value range of 0.001-0.005 / month; β T , β H , and β F are the drift influence coefficients of temperature T, humidity H, and interference gas concentration F, respectively; T0 and H0 are the standard temperature and standard humidity, respectively; β T has a value range of β H has a value range of β F has a value range of It is determined by the method of controlling variables; in the experiment, 5 temperature points (such as 15°C, 25°C, 35°C, 45°C, and 55°C) are usually selected for measurement, and at least 30 days are observed under each temperature condition to record the sensor performance decay, to ensure the accuracy and representativeness of the data.

[0065]

[0066] Calibration compensation coefficient: where S0 is the initial sensitivity coefficient, S c is the current sensitivity coefficient, α T , α H , and α FThe compensation coefficients of temperature, humidity and interfering gas, respectively, are calculated by multiple regression analysis.

[0067]

[0068] Δγ is the parameter drift rate change; ΔS is the sensitivity coefficient change; ΔT, ΔH and ΔF are the changes of temperature, humidity and interfering gas concentration, respectively, relative to the standard conditions; these changes are the deviations relative to the standard conditions (T0=25℃, H0=50% RH, F=0 ppm) and are used to calculate the influence of environmental changes on sensor performance. In practical applications, the temperature change range is usually -20℃ to 70℃, the humidity change range is 10-95% RH, and the interfering gas concentration change range depends on the specific application environment.

[0069] The sensor voltage output value is adjusted in real time according to the calibration compensation coefficient, and the mathematical expression of the calibration method is:

[0070] V c = C c ×(V m -O c )+O0; wherein V c is the corrected sensor voltage output value, V m is the current sensor measured voltage output value, O c is the current zero point offset, and O0 is the initial zero point offset. The current zero point offset Oc can be obtained by regular cleaning air correction or predicted by the zero point drift model. For example, the initial zero point offset O0 of a certain sensor is 10 mV, the initial sensitivity coefficient S0 is 0.8 mV / ppm, the current zero point offset Oc is 15 mV after 6 months of use, the current sensitivity coefficient Sc is 0.65 mV / ppm, the environmental temperature is 10℃ higher than the standard temperature, the humidity is 20% higher than the standard humidity, and the interfering gas concentration is 20 ppm, then the calibration compensation coefficient Cc=1.23, and when the measured voltage output value Vm is 100 mV, the corrected voltage output value Vc=114.2 mV.

[0071] The detection method of the interference gas concentration F is: recording the sensor baseline resistance response R0 in clean air, detecting the interference gas in the environment using an auxiliary sensor array, the auxiliary sensor array is composed of 4-8 different types of gas sensors, including electrochemical sensors, semiconductor sensors and optical sensors, etc., covering common interference gases such as CO, H2, C2H5OH, etc.; distinguishing the target gas and the interference gas through a pattern recognition algorithm, and calculating the interference gas concentration F according to the sensor cross-sensitivity matrix; the pattern recognition algorithm adopts a method combining principal component analysis (PCA) and support vector machine (SVM), and the recognition accuracy can reach more than 90%. The sensor cross-sensitivity matrix is obtained by experiment calibration, and the sensitivity proportion relationship of each sensor to various gases is recorded.

[0072] The quantitative calculation formula of the interference gas concentration F is: Wherein, R i is the sensor resistance response value when the interference gas exists; k i is the weight coefficient of the ith interference gas, which is obtained by experiment calibration: is the concentration value of the ith interference gas under the standard concentration, is the corresponding sensor resistance response value under the standard concentration. For example, the weight coefficient k of a certain sensor to CO is 200 ppm, the baseline resistance R0 in clean air is 10 kΩ, and the resistance becomes 11 kΩ when CO exists in the environment, then the calculated CO concentration is 20 ppm. In actual application, the influence of 3-5 main interference gases is usually considered, such as CO (0-500 ppm), H2 (0-1000 ppm), C2H5OH (0-200 ppm) and the like.

[0073] The method for establishing the performance degradation prediction model comprises: establishing a sensor historical data time series, extracting characteristic parameters and performing data normalization, using exponential smoothing method to analyze the trend of historical data, and predicting the trend of sensor parameter change in the future period; the historical data time series includes the data of the change of sensor sensitivity, zero drift, response time and other parameters with time, and the sampling period is usually 1 day or 1 week. The characteristic parameters include the average value, change rate and fluctuation amplitude of the parameters, which are normalized by the min-max method to eliminate the influence of different parameter dimensions. The exponential smoothing method is a commonly used time series prediction method, which is suitable for processing data with trend and seasonality, and the prediction accuracy is higher than that of the simple moving average method.

[0074] The mathematical expression of the sensor parameter change trend in the future period is:

[0075] Wherein, P t+Δt is the predicted value of the sensor parameter at t+Δt, Pt T is the current parameter value, t is the trend factor, and a and b are smoothing coefficients, with a range of 0.2-0.3 and 0.1-0.2, respectively, determined by minimizing the prediction error of historical data; for example, for the sensitivity parameter, a is usually 0.25 and b is usually 0.15, which can predict the trend of changes in the next 1-3 months, with a prediction error usually less than ±5%.

[0076] wherein, is the actual parameter value at the i-th time point, is the parameter value predicted using the exponential smoothing method, and N is the total number of data time points collected. In the calculation, a grid search method is used to traverse different combinations of a and b in the specified range with a step size of 0.01, and the combination that minimizes the mean square error is selected as the optimal parameters.

[0077] The calibration uncertainty evaluation includes:

[0078] The repeatability uncertainty u r is calculated. In practical applications, usually m = 10-20 measurement values are continuously collected to calculate the standard deviation and evaluate the repeatability of the measurement. For example, for a certain sensor, 10 consecutive measurements were made in a 1000 ppm methane standard gas, resulting in output voltage values of 500.2, 502.1, 499.5, 501.8, 500.7, 503.2, 499.1, 502.5, 501.3, and 500.8 mV, respectively. The calculated ur = 0.42 mV, with a relative standard deviation of 0.08%.

[0079] The system uncertainty u s is calculated. This formula calculates the system error introduced by each influencing factor in the measurement system, using the error propagation law, taking into account the sensitivity coefficients and standard uncertainties of each factor. The influencing factors xj include temperature, humidity, pressure, power voltage, etc. The sensitivity coefficient of each factor is obtained through theoretical analysis or experimental determination, and the standard uncertainty u(xj) is determined according to the instrument accuracy and environmental fluctuations. For example, the sensitivity coefficient of temperature is 1.5 mV / ℃, and the standard uncertainty is 0.3℃; the sensitivity coefficient of humidity is 0.8 mV / %RH, and the standard uncertainty is 2%RH; the calculated system uncertainty us = 1.75 mV.

[0080] The expanded uncertainty U is calculated: wherein, V i is the i-th measurement value, is the average value, m is the number of measurements, x j is the influencing factor, and u(x j) represents the standard uncertainty of each influencing factor, k p Let k be the inclusion factor. p =2, confidence level greater than 95%. Expanded uncertainty U represents the confidence interval of the measurement result and is an important parameter for quality control and reliability assessment. Continuing the previous example, the expanded uncertainty U is calculated to be U = 2 × √(0.42). 2 +1.75 2 = 3.6mV, with a relative expanded uncertainty of 0.72%.

[0081] The system automatically performs an uncertainty assessment monthly and generates an assessment report. When the expanded uncertainty exceeds the allowable range (usually ±2% of the measured value), the system issues a warning, prompting the user to perform maintenance or calibration.

[0082] The method also includes calculating the predicted lifetime of the sensor: Where L is the predicted lifetime, L0 is the baseline lifetime under standard conditions (24-36 months), and E a The activation energy is defined as 0.2–0.8 eV, determined based on the sensor material. k is the Boltzmann constant, 8.617 × 10⁻⁶. -5 eV / K, λ is the lifetime impact coefficient of the interfering gas, determined through accelerated aging tests, with a value ranging from 0.001 to 0.01 / (ppm·month), F is the concentration of the interfering gas, and t is the current time, i.e., the time the sensor has been used. The first exponential term describes the effect of temperature on lifetime; increased temperature accelerates catalyst aging and poisoning, shortening the lifespan. The second exponential term describes the effect of the interfering gas on lifetime; the interfering gas causes catalyst poisoning and reduced activity, accelerating performance degradation. For example, a sensor with a baseline lifetime L0 of 30 months under standard conditions and an activation energy Ea of 0.5 eV, used for t = 10 months in an environment with a temperature T = 45℃ and an interfering gas concentration F = 30 ppm, has a lifetime impact coefficient of λ = 0.005 / (ppm·month). The calculated predicted lifetime L = 16.2 months, with 10 months already used and a remaining lifetime of 6.2 months. By periodically updating the predicted lifetime, the system can plan sensor replacement in advance, avoiding safety risks caused by sensor failure.

[0083] Example 2

[0084] like Figure 2 The diagram shows an adaptive calibration method for a combustible gas sensor according to the present invention, based on the system implementation of Embodiment 1. The sensor is a catalytic combustion type combustible gas sensor, and the adaptive calibration method includes:

[0085] Step S1, perform the basic calibration of the sensor, record the sensor response value under standard conditions, establish the calibration curve of the sensor under standard conditions, and determine the initial sensitivity coefficient, zero point offset and response time of the sensor.

[0086] Place the sensor in a standard environmental condition (temperature 25±1℃, relative humidity 50±5%RH, no interfering gas), use a standard gas generating device to generate 0ppm, 500ppm, 1000ppm, 2000ppm, 3000ppm and 5000ppm standard methane gas, and record the output voltage value of the sensor at each concentration point. The calibration curve of the sensor is obtained by least squares fitting, V=S0xC+O0, where V is the output voltage (mV), C is the gas concentration (ppm), S0is the initial sensitivity coefficient (mV / ppm), the typical value is 0.5-1.0 mV / ppm, O0is the initial zero point offset (mV), the typical value is 5-20 mV. At the same time, record the time required for the sensor to reach 90% of the final value from contacting the gas, and determine the initial response time of the sensor, the typical value is 10-30 seconds.

[0087] Step S2, monitor the real-time parameters of the sensor under different environmental conditions, the real-time parameters include the resistance value change rate, power consumption fluctuation and thermal characteristic change of the sensor, and record the corresponding environmental temperature, humidity and interfering gas concentration data.

[0088] In practical application, the temperature and humidity sensor is used to monitor the environmental temperature and humidity in real time, the temperature monitoring range is -40℃ to 85℃, the accuracy is ±0.5℃; the humidity monitoring range is 0-100%RH, the accuracy is ±3%RH. At the same time, the resistance value change rate ΔR / R0of the sensor is monitored, where R0is the initial resistance value (usually 5-15Ω), ΔR is the resistance value change; the power consumption fluctuation of the sensor is monitored, the normal power consumption range is 80-120 mW, and the fluctuation exceeding ±10% may indicate that the performance of the sensor is abnormal; the thermal characteristic change of the sensor is monitored, including the surface temperature change rate and thermal equilibrium time change of the sensor, these parameters reflect the change of the catalytic activity and thermal conductivity characteristics of the sensor. The auxiliary sensor array is used to detect common interfering gases in the environment, such as carbon monoxide, hydrogen, ethanol, etc., the concentration range is 0-1000ppm.

[0089] Step S3, according to the obtained data, establish the sensor performance parameter drift model, detect and compensate the sensor drift, and calculate the calibration compensation coefficient. The expression of the sensor performance parameter drift model is: P(t|T,H,F)=P 0 exp(-γ×t), where P(t|T,H,F) is the performance parameter at time t, temperature T, humidity H and interfering gas concentration F, P 0For the initial performance parameters of the sensor, γ is the sensor performance drift rate under environmental conditions, and t is time; the model reflects the law of sensor performance attenuation over time. In general, the sensor sensitivity will decrease by 10-20% within 6 months and by 20-35% within 12 months. The exponential function form in the model can accurately describe the nonlinear attenuation characteristics of the sensor performance, which decays faster at the beginning and tends to be stable at the later stage.

[0090] The sensor performance parameter drift rate γ is represented by a multivariate function:

[0091] γ = γ0+ β T (T-T0) 2 + β H (H-H0) 2 + β F × F where γ0 is the basic drift rate under standard conditions, which is obtained by long-term stability testing and has a value range of 0.001-0.005 / month; β T , β H , and β F are the drift influence coefficients of temperature T, humidity H, and interference gas concentration F, respectively; T0 and H0 are the standard temperature and standard humidity, respectively; β T has a value range of β H has a value range of β F has a value range of It is determined by the control variable method; in the experiment, 5 temperature points (such as 15℃, 25℃, 35℃, 45℃, 55℃) are usually selected for measurement, and at least 30 days are observed under each temperature condition, and the sensor performance attenuation is recorded to ensure the accuracy and representativeness of the data.

[0092]

[0093] Calibration compensation coefficient: where S0 is the initial sensitivity coefficient, S c is the current sensitivity coefficient, α T , α H , and α F are the compensation coefficients of temperature, humidity, and interference gas, respectively, which are calculated by multivariate regression analysis;

[0094]

[0095] Δγ represents the change in parameter drift rate; ΔS represents the change in sensitivity coefficient; ΔT, ΔH, and ΔF represent the changes in temperature, humidity, and interfering gas concentration relative to standard conditions, respectively. These changes are deviations from standard conditions (T0 = 25℃, H0 = 50% RH, F = 0 ppm) and are used to calculate the impact of environmental changes on sensor performance. In practical applications, the temperature range is typically -20℃ to 70℃, the humidity range is 10-95% RH, and the range of interfering gas concentration depends on the specific application environment.

[0096] This multivariate function considers the effects of three main environmental factors—temperature, humidity, and interfering gases—on sensor drift. A quadratic term describes the effects of temperature and humidity, while a linear term describes the effects of interfering gases. This is because experiments have shown that the effects of temperature and humidity on drift are non-linear, while the effects of interfering gas concentration on drift are approximately linear. Taking a CH4 sensor as an example, when the temperature increases from 25℃ to 50℃, the drift rate may increase by 0.3-0.6% / month; when the relative humidity increases from 50% to 90%, the drift rate may increase by 0.4-0.8% / month; and when the CO concentration in the environment is 50ppm, the drift rate may increase by 0.2-0.5% / month.

[0097] Step S4: Adjust the sensor voltage output value in real time according to the calibration compensation coefficient. Based on the calibration compensation coefficient Cc calculated in step S3, compensate and correct the measured voltage output value of the sensor to eliminate the influence of environmental factors and sensor performance degradation on the measurement results, and output an accurate gas concentration value. The calibration process is performed in real time at a frequency of 1-10Hz to ensure the timeliness and accuracy of the measurement results.

[0098] The sensor voltage output value is adjusted in real time according to the calibration compensation coefficient. The mathematical expression for the calibration method is as follows:

[0099] V c =C c ×(V m -O c )+O0; where V c V is the corrected sensor voltage output value. m The current measured voltage output value of the sensor, O cOc is the current zero-point offset, and O0 is the initial zero-point offset. The current zero-point offset Oc can be obtained by periodic clean air correction or predicted by a zero-point drift model. For example, the initial zero-point offset O0 of a certain sensor is 10 mV, the initial sensitivity coefficient S0 is 0.8 mV / ppm, the current zero-point offset Oc is 15 mV after 6 months of use, the current sensitivity coefficient Sc is 0.65 mV / ppm, the ambient temperature is 10°C higher than the standard temperature, the humidity is 20% higher than the standard humidity, and the interfering gas concentration is 20 ppm. Then, the calibration compensation coefficient Cc is 1.23, and the corrected voltage output value Vc is 114.2 mV when the measured voltage output value Vm is 100 mV.

[0100] The detection method of the interfering gas concentration F is as follows: the baseline resistance response R0 of the sensor is recorded in clean air, the interfering gas in the environment is detected using an auxiliary sensor array, the auxiliary sensor array is composed of 4-8 different types of gas sensors, including electrochemical sensors, semiconductor sensors, optical sensors, etc., covering common interfering gases such as CO, H2, C2H5OH, etc.; the target gas and the interfering gas are distinguished by a pattern recognition algorithm, and the interfering gas concentration F is calculated according to the sensor cross-sensitivity matrix. The pattern recognition algorithm uses a combination of principal component analysis (PCA) and support vector machine (SVM) methods, with an identification accuracy of over 90%. The sensor cross-sensitivity matrix is obtained by experimental calibration, recording the sensitivity proportion of each sensor to various gases.

[0101] The quantitative calculation formula of the interfering gas concentration F is: wherein, R i is the sensor resistance response value in the presence of interfering gas; k i is the weight coefficient of the oth interfering gas, which is obtained by experimental calibration: is the concentration value of the oth interfering gas at the standard concentration, is the corresponding sensor resistance response value at the standard concentration. For example, the weight coefficient k of CO for a certain sensor is 200 ppm, the baseline resistance R0 in clean air is 10 kΩ, and the resistance becomes 11 kΩ when CO exists in the environment. Then, the calculated CO concentration is 20 ppm. In practical applications, the influence of 3-5 main interfering gases is usually considered, such as CO (0-500 ppm), H2 (0-1000 ppm), C2H5OH (0-200 ppm), etc.

[0102] The method further comprises:

[0103] Step S5, a sensor performance attenuation prediction model is established, historical data is used to analyze the trend of sensor parameter changes, parameter drift is predicted through mathematical statistical methods, and parameter adjustment is performed in advance; through historical data mining, the performance change trend of the sensor in the future period of time is predicted, providing decision basis for active maintenance and replacement. The system collects sensor performance data every day, including sensitivity, zero point offset, response time and other parameters, forming a time series database. By analyzing the change rule of these data, the performance parameters of the sensor in the future 1-3 months are predicted, and when the predicted parameters exceed the allowed range, the system sends a warning information to prompt the user to maintain or replace.

[0104] Step S6, the reliability of the self-calibration result is evaluated, the calibration uncertainty is calculated, and the sensor state report is output. The system regularly (such as every week or every month) evaluates the calibration result, calculates the measurement uncertainty, and generates a sensor state report. The state report includes the current performance parameters of the sensor, the calibration compensation coefficient, the measurement uncertainty, the predicted life, and the corresponding maintenance suggestions. When the measurement uncertainty exceeds the allowed range (usually ±5% of the measured value), the system will issue a warning to prompt the user to perform manual calibration or replace the sensor.

[0105] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. An adaptive calibration system for a combustible gas sensor, wherein the sensor is a catalytic combustion type combustible gas sensor, characterized in that, The system comprises, in sequence, a basic calibration unit, a real-time monitoring unit, a drift detection and compensation unit, and an output adjustment unit; The basic calibration unit is used to perform basic calibration of the sensor, record the sensor response value under standard conditions, establish the calibration curve of the sensor under standard conditions, and determine the initial sensitivity coefficient, zero-point offset and response time of the sensor. The real-time monitoring unit is used to monitor the real-time parameters of the sensor under different environmental conditions. The real-time parameters include the sensor's resistance change rate, power consumption fluctuation and thermal characteristic change, and record the corresponding ambient temperature, humidity and interfering gas concentration data. The drift detection and compensation unit is used to establish a sensor performance parameter drift model based on the acquired data, detect and compensate for sensor drift, and calculate calibration compensation coefficients. The expression for the sensor performance parameter drift model is: ,in, for Time, temperature ,humidity and interfering gas concentration Performance parameters under the conditions, These are the initial performance parameters of the sensor. The sensor performance drift rate under environmental conditions. For time; Sensor performance drift rate Represented by multivariate functions: in, The baseline drift rate under standard conditions is obtained through long-term stability testing, with a value ranging from 0.001 to 0.005 per month. , and Temperature ,humidity and interfering gas concentration The drift influence coefficient, and These are standard temperature and standard humidity, respectively. The range of values ​​is ; The range of values ​​is ; The range of values ​​is The determination was made using the controlled variable method; ； ； ； Calibration compensation coefficient: ;in, This is the initial sensitivity coefficient. This represents the current sensitivity coefficient. , and The compensation coefficients for temperature, humidity, and interfering gases are respectively calculated through multiple regression analysis. ； ； ； This refers to the change in the parameter drift rate; This represents the change in the sensitivity coefficient. , and These represent the changes in temperature, humidity, and concentration of interfering gases relative to standard conditions, respectively. The output adjustment unit is used to adjust the sensor voltage output value in real time according to the calibration compensation coefficient.

2. The adaptive calibration system for a combustible gas sensor according to claim 1, characterized in that, The mathematical expression for the calibration method that adjusts the sensor voltage output value in real time according to the calibration compensation coefficient is as follows: ;in, This is the corrected sensor voltage output value. This is the current measured voltage output value of the sensor. This is the current zero-point offset. This is the initial zero-point offset.

3. The adaptive calibration system for a combustible gas sensor according to claim 1, characterized in that, The concentration of the interfering gas The detection method is as follows: the baseline resistance response R0 of the sensor is recorded in clean air; an auxiliary sensor array is used to detect interfering gases in the environment; a pattern recognition algorithm is used to distinguish between the target gas and the interfering gas; and the concentration of the interfering gas is calculated based on the sensor cross-sensitivity matrix. ; Interfering gas concentration The quantitative calculation formula is as follows: ;in, This is the sensor resistance response value in the presence of interfering gases; For the first The weighting coefficients for the interfering gases were obtained through experimental calibration. ; The first at standard concentration Concentration values ​​of interfering gases This represents the sensor resistance response value at the standard concentration.

4. The adaptive calibration system for a combustible gas sensor according to any one of claims 1-3, characterized in that, The system also includes: The performance prediction unit is used to establish a sensor performance degradation prediction model, analyze the trend of sensor parameter changes using historical data, predict parameter drift through mathematical statistical methods, and adjust parameters in advance. The reliability assessment unit is used to assess the reliability of the self-calibration results, calculate the calibration uncertainty, and output a sensor status report.

5. The adaptive calibration system for a combustible gas sensor according to claim 4, characterized in that, The method for establishing a sensor performance degradation prediction model includes: establishing a time series of historical sensor data, extracting feature parameters and normalizing the data, using exponential smoothing to perform trend analysis on the historical data, and predicting the trend of sensor parameter changes in future periods. The mathematical expression for the trend of sensor parameter changes over the future period is: ;in, For the future Predicted sensor parameters at time [time]. The current parameter value. As a trend factor, and The smoothing coefficient has values ​​ranging from 0.2 to 0.3 and from 0.1 to 0.2, respectively, and is determined by minimizing the prediction error of historical data. ； ;in, For the first The actual parameter values ​​at each time point These are the parameter values ​​predicted using exponential smoothing. This represents the total number of data points collected.

6. The adaptive calibration system for a combustible gas sensor according to claim 5, characterized in that, The calculation of calibration uncertainty includes: Calculate repeatability uncertainty : ; Calculate system uncertainty : ; Calculate the expanded uncertainty U: ; in, For the first This measurement value, This is the average value. To measure the number of times, As influencing factors, The standard uncertainty of each influencing factor, As the inclusion factor, take The confidence level is greater than 95%.

7. The adaptive calibration system for a combustible gas sensor according to claim 6, characterized in that, The system also includes a lifetime calculation unit for calculating the predicted lifetime of the sensor: ;in, To predict lifespan, The baseline lifespan under standard conditions is 24-36 months. The activation energy, ranging from 0.2 to 0.8 eV, is determined based on the sensor material. Boltzmann's constant, , The lifetime impact coefficient of the interfering gas was determined through accelerated aging tests, and its value range was [value range missing]. , To interfere with gas concentration, This represents the current moment, i.e., the time the sensor has been in use.

8. An adaptive calibration method for a combustible gas sensor, based on the system according to any one of claims 1-7, characterized in that, Its features are, The adaptive calibration method includes: Step S1: Perform basic calibration of the sensor, record the sensor response value under standard conditions, establish the calibration curve of the sensor under standard conditions, and determine the initial sensitivity coefficient, zero offset and response time of the sensor. Step S2: Monitor the real-time parameters of the sensor under different environmental conditions. The real-time parameters include the sensor's resistance change rate, power consumption fluctuation and thermal characteristic change, and record the corresponding ambient temperature, humidity and interfering gas concentration data. Step S3: Based on the acquired data, establish a sensor performance parameter drift model, detect and compensate for sensor drift, and calculate the calibration compensation coefficient. Step S4: Adjust the sensor voltage output value in real time according to the calibration compensation coefficient.

9. The adaptive calibration method for a combustible gas sensor according to claim 8, characterized in that, The method further includes: Step S5: Establish a sensor performance degradation prediction model, analyze the trend of sensor parameter changes using historical data, predict parameter drift using mathematical statistical methods, and adjust parameters in advance. Step S6: Perform a reliability assessment of the self-calibration results, calculate the calibration uncertainty, and output a sensor status report.

Images