Full life cycle management method for electrochemical gas sensors

Abstract

The present invention relates to the field of gas detection technology, specifically to a full life cycle management method for an electrochemical gas sensor, comprising the following steps: S1: real-time monitoring of sensor data noise to obtain sensor noise data; S2: statistics of the accumulated working time of the sensor under different gas concentrations, temperatures, humidity and atmospheric pressure environmental parameters; S3: bringing the sensor production date, data noise and working time under different environments into the sensor life cycle model to calculate the sensor aging degree and remaining service life. With this method, the sensor does not need to be brought back to the laboratory for measurement, and the life cycle of the sensor can be monitored in real time in practical applications.

Description

Technical Field

[0001] The present invention relates to the technical field of electrochemical gas sensors, which belongs to G01N in the IPC classification number, and in particular to a full life cycle management method for electrochemical gas sensors. Background Art

[0002] Electrochemical sensors use electrochemical reactions to detect gas concentrations. Their basic structure includes a working electrode, a counter electrode, and a reference electrode. When the target gas reacts with the surface of the working electrode, the resulting current or potential change is proportional to the target gas concentration. By measuring this current or potential change, the concentration of the target gas can be determined. The life of an electrochemical sensor is usually affected by many factors, including but not limited to the working environment, usage conditions, and the quality of the sensor itself. Therefore, there are many challenges in technically measuring the life of an electrochemical sensor. Currently, the following methods are mainly used for detection:

[0003] 1. Laboratory testing

[0004] By placing the sensor under controlled conditions, simulating actual usage scenarios, and monitoring changes in the sensor output signal, this method can provide a relatively accurate prediction of sensor life, but it requires taking the sensor back to the laboratory environment for measurement, which takes a long time and is costly;

[0005] 2. Analyze and monitor sensor data

[0006] By using the sensor's detection data, combined with the data history log, and using periodic data reference, the sensor's performance can be monitored and evaluated in real time. By establishing appropriate models and algorithms, signs of sensor performance degradation can be discovered in a timely manner and its life can be predicted. However, this method is often discovered after the fact (discovered after the sensor's performance has deteriorated), and it is difficult to achieve pre-warning (discovered before the sensor has problems);

[0007] 3. Model-based lifespan prediction

[0008] Based on the principle and performance parameters of electrochemical sensors, a mathematical model is built, and the life of the sensor under specific conditions is predicted through the model. This method mainly tests the construction of mathematical models. Being able to effectively build a full-parameter data model will directly affect the prediction accuracy. Most of the current model-based predictions are based on laboratory data to build models. With the changes in the use environment and the increase in usage time, the accuracy of the model gradually becomes worse. The present invention is based on the principle of electrochemical sensors, and uses sensor noise, gas concentration, ambient temperature, and ambient humidity as input parameters to establish a sensor full life cycle model. The prediction accuracy will become higher and higher as the sensor application time increases and the use environment changes. Summary of the invention

[0009] In view of the deficiencies in the prior art, the present invention provides a method for managing the entire life cycle of an electrochemical gas sensor, which can perform real-time monitoring in practical applications.

[0010] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for managing the entire life cycle of an electrochemical gas sensor, comprising the following steps:

[0011] S1: monitor the data noise of the sensor in real time to obtain the sensor noise data;

[0012] S2: Count the accumulated working time of the sensor under different gas concentration, temperature, humidity and atmospheric pressure environmental parameters;

[0013] S3: The sensor production date, data noise, and working time in different environments are introduced into the sensor life cycle model to calculate the sensor aging degree and remaining service life.

[0014] Preferably, in S1, a trend prediction algorithm is used to determine the true value of the current sensor data, and the data noise is calculated in real time through data accumulation statistics.

[0015] Preferably, the sensor noise monitoring steps are as follows: first, read the real-time data of the sensor and calculate the 90-second data mean; second, obtain the current data trend according to the trend algorithm, and the data trend is divided into four types: rising, falling, fluctuating and stable; finally, when the data trend is fluctuating or stable, calculate the real-time noise of the data according to Formula 1;

[0016] VN rtd = |VAL rtd -VAL avg | Formula 1

[0017] VN rtd :Data real-time noise

[0018] VAL rtd :Real-time value of data

[0019] VAL avg : Data average value for 90 seconds.

[0020] Preferably, the trend prediction algorithm implementation process includes four data trends, namely: real-time trend, current trend, historical trend and temporary trend; four trend states, namely: rising state, falling state, fluctuating state and stable state; the trend prediction algorithm steps are as follows: first, the real-time trend is obtained according to the real-time data and the data mean, and the real-time data is greater than the data mean in the rising state, otherwise it is in the falling state; secondly, if the historical trend is fluctuating, if the real-time trend is the same as the temporary trend for more than 10 times, the temporary trend will be written into the current data trend, and if it is not more than 10 times, the real-time trend will be written into the temporary trend; if the real-time data trend is the same as the historical trend, the historical trend will be written into the current trend, if they are different, the current trend will be modified to fluctuating; finally, the current trend is output.

[0021] Preferably, in S2, the gas concentration is divided into 5 categories: greater than 80% of the range, greater than 50% of the range and less than 80% of the range, greater than 30% of the range and less than 50% of the range, greater than 15% of the range and less than 30% of the range, and less than or equal to 15% of the range, and the sensor working time is counted once a minute; secondly, the temperature is further divided into 5 categories in each category: greater than 40°C, 30°C to 40°C, 4°C to 30°C, -10°C to 4°C, and less than or equal to -4°C, and the working time is counted respectively; finally, the humidity is divided into 5 categories: 80% to 100%, 50% to 80%, 30% to 50%, 15% to 30%, and less than or equal to 15%, and the time is counted respectively; a total of 125 subcategories are divided to count the sensor working time.

[0022] Preferably, the sensor life cycle model formula is as follows:

[0023] Formula 2

[0024] t 1 : Remaining usage time of the sensor

[0025] t 2 :Sensor design usage time

[0026] n 1 : Expected sensor noise

[0027] n 2 : Actual sensor noise

[0028] k: Correction value of working time in different environments

[0029] t: sensor working time in different environments

[0030] The beneficial effect of the present invention is that by adopting this method, the sensor does not need to be taken back to the laboratory for measurement, and the life cycle of the sensor can be monitored in real time in actual application. The present invention calculates the sensor life cycle curve based on the changing trend of sensor data noise and the summary statistics of sensor working time, gas concentration, temperature and humidity and other parameters, substitutes it into the sensor life cycle model for calculation, and obtains the remaining use time and aging degree of the sensor. At the same time, the measurement data is corrected according to the aging degree of the sensor to optimize the performance of the sensor. The life cycle management of electrochemical sensors is of great value. It not only ensures the performance and reliability of the sensor, but also provides a cost-effective solution for various applications by reducing maintenance costs and improving system design efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Sensor life cycle management flow chart;

[0032] Figure 2 Sensor data noise calculation process;

[0033] Figure 3 Sensor trend algorithm flow chart;

[0034] Figure 4 Statistical sensor working environment data flow chart. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0036] A method for managing the entire life cycle of an electrochemical gas sensor, the implementation steps are shown in Figure 1 First, the sensor noise is monitored in real time to obtain the sensor noise data; then, the sensor’s cumulative working time under different environmental parameters such as gas concentration, temperature, humidity, and atmospheric pressure is counted; finally, the sensor production date, noise, and working time under different environments are brought into the sensor life cycle model to calculate the sensor aging degree and remaining service life.

[0037] The specific implementation is as follows:

[0038] Sensor Data Noise

[0039] There is a certain relationship between the noise of electrochemical sensors and aging. As the use time of electrochemical sensors increases, their electrode materials, electrolytes and other components may have some problems due to aging, resulting in a decrease in sensor performance, which in turn causes an increase in noise. Aging may lead to problems such as a decrease in the electrode surface area of ​​the electrochemical sensor, a decrease in the electrochemical reaction rate, and an increase in electrolyte loss, all of which will affect the noise level of the sensor.

[0040] Sensor data noise is relatively easy to measure in the laboratory, but it is very challenging to measure sensor data noise in applications. This is why few people calculate the degree of sensor aging by monitoring the noise in applications. When monitoring sensor noise, the noise cannot be measured when the data rises or falls, because the calculated noise deviation is large and has nothing to do with sensor performance. The trend prediction algorithm is particularly critical in this step.

[0041] The present invention adopts a trend prediction algorithm to determine the true value of current sensor data, and calculates data noise in real time through data accumulation statistics.

[0042] The implementation steps of sensor noise monitoring are as follows: Figure 2 As shown, first, read the real-time data of the sensor and calculate the 90-second data average. The main reason for using the 90-second average is that the average needs to exceed the minute value and the longest response time of the electrochemical sensor will not exceed 90 seconds; secondly, get the current data trend according to the trend algorithm. The data trend is divided into four types: rising, falling, fluctuating, and stable; finally, when the data trend is fluctuating or stable, calculate the real-time noise of the data according to Formula 1.

[0043] VN rtd = |VAL rtd -VAL avg | Formula 1

[0044] VN rtd :Data real-time noise

[0045] VAL rtd :Real-time value of data

[0046] VAL avg : Data average value for 90 seconds.

[0047] The trend algorithm implementation process includes four data trends: real-time trend, current trend, historical trend, and temporary trend; four trend states: rising state, falling state, fluctuating state, and stable state. The meanings are explained as follows:

[0048] Real-time trend: the trend calculated based on single real-time data and data mean;

[0049] Current trend: Current data trend obtained by integrating historical trend, temporary trend and real-time trend;

[0050] Historical trend: the current trend calculated last time;

[0051] Temporary trend: When the real-time trend differs from the historical trend, the last real-time trend is stored.

[0052] Rising state: VAL rtd >VAL avg And VN rtd >Sensor resolution is in the rising state, and noise is not calculated at this time;

[0053] Descending state: VAL rtd <VAL avg And VN rtd >Sensor resolution is in a decreasing state, and noise is not calculated at this time;

[0054] Fluctuation state: The sensor data is in a fluctuating state due to fluctuations in gas concentration or sensor aging;

[0055] Steady state: VN rtd <Sensor resolution, in a stable state, data fluctuations are small.

[0056] The steps to implement the trend algorithm are as follows: Figure 3 As shown, first, the real-time trend is obtained based on the real-time data and the data mean. If the real-time data is greater than the data mean, it is in an upward state, otherwise it is in a downward state; secondly, if the historical trend is fluctuating, if the real-time trend is the same as the temporary trend for more than 10 times, the temporary trend will be written into the current data trend, and if it is not more than 10 times, the real-time trend will be written into the temporary trend; if the real-time data trend is the same as the historical trend, the historical trend will be written into the current trend, and if they are different, the current trend will be modified to fluctuating. Finally, the current trend is output. Sensor working time statistics

[0057] The aging of electrochemical sensors is closely related to the working time of the sensors and the working environment (temperature, humidity, gas concentration, etc.). Because the attenuation degree of sensors in different environments is different, the method of directly calculating the aging degree by counting the working time of sensors without considering environmental factors will have a large deviation. The present invention uses a hierarchical classification with a total of 125 subdivision levels to count the working time of sensors and achieve a comprehensive and thoughtful calculation. Figure 4As shown, first, the gas concentration is divided into 5 categories: greater than 80% of the range, greater than 50% of the range and less than 80% of the range, greater than 30% of the range and less than 50% of the range, greater than 15% of the range and less than 30% of the range, and less than or equal to 15% of the range, and the sensor working time is counted once a minute; secondly, the temperature is divided into 5 categories in each category: greater than 40℃, 30℃~40℃, 4℃~30℃, -10℃~4℃, and less than or equal to -4℃, and the working time is counted respectively; finally, the humidity is divided into 5 categories: 80%~100%, 50%~80%, 30%~50%, 15%~30%, and less than or equal to 15%, and the time is counted respectively; a total of 125 subcategories are divided into statistical sensor working time.

[0058] Sensor life cycle model

[0059] The life cycle model of an electrochemical sensor is a complex process that involves the entire process from the beginning of use to the final failure of the sensor. The following are some key environmental factors that affect the life of electrochemical sensors:

[0060] Temperature: Temperature is the biggest factor affecting the life of the sensor. For example, at 40°C, the life of the sensor may be 6 months, but at 5 months, the life will be shortened to 3 months, that is, when the temperature is greater than 40°C, the sensor life will be almost halved for every 10°C increase in temperature. Extreme temperatures can cause electrolyte corrosion or electrode corrosion.

[0061] Humidity: Humidity that is too high or too low will have an adverse effect on the sensor. Too high humidity will cause the sensor electrolyte to overflow, while too low humidity will cause the electrolyte to dry up and affect the sensor performance. The ideal operating humidity for electrochemical sensors is 30% to 60%.

[0062] Gas concentration: Frequent or prolonged operation under high-concentration gas will accelerate catalyst deactivation and shorten sensor life.

[0063] Manufacturing quality: The manufacturing process, material selection and packaging technology of the sensor directly affect the initial performance and durability of the sensor.

[0064] The sensor life cycle model of the present invention is designed based on the above factors, and the model formula is as follows:

[0065] Formula 2

[0066] t 1 : Remaining usage time of the sensor

[0067] t 2 :Sensor design usage time

[0068] n 1 : Expected sensor noise

[0069] n 2 : Actual sensor noise

[0070] k: Correction value of working time in different environments

[0071] t: sensor working time under different environments.

[0072] Take a certain brand of SO2 gas sensor as an example. The sensor is designed to be used for a period of time t 2 = 1051200 minutes (2 years), the expected noise level of the sensor is 42 according to factory measurements.

[0073] The actual measured noise of the sensor is 60. Assuming that the sensor only works in the range of >15% and ≤30%, the cumulative working time is 13500.

[0074] According to Formula 1, it is calculated that t1 = 0.9129×(1051200-16302.5)=977743.4 minutes, that is, the remaining usage time of the sensor is 678.99 days.

[0075] The gas concentration measured by the sensor is shown in Formula 3.

[0076] Val = (V cur -V zero ) / Sen Formula 3

[0077] Val: gas concentration

[0078] V cur : Current value of the sensor

[0079] V zero : Sensor zero value (value measured in zero-grade air)

[0080] Sen: sensor sensitivity

[0081] As the use time increases, the measurement accuracy of the sensor will gradually decrease. The main reason for the decrease in measurement accuracy is the increase in sensor noise and the change in sensitivity. The measurement data correction optimization of the sensor is mainly carried out from the following two aspects.

[0082] Based on noise optimization: Increased noise will lead to increased fluctuations in sensor measurement data. The noise optimization method is to increase the sensor mean buffer. The default mean buffer number 90×(actual noise / expected noise) is used to get the adjusted buffer size. By increasing the buffer size, de-jittering is achieved to reduce the impact of noise (increasing the mean buffer is at the expense of sensor response time, so it cannot be adjusted to the maximum by default. When the noise increases, increasing the buffer size is the lesser of two evils, and increasing the response time is used to increase measurement accuracy).

[0083] Sensitivity correction: According to the distribution of sensor working time, the working time of the sensor in humidity greater than 50% and less than 15% is counted. If the former is greater than the latter, the sensitivity value increases (the moisture content of the electrolyte in the sensor increases due to long-term high humidity, resulting in an increase in the measured current signal). If the former is less than the latter, the sensitivity value decreases (the moisture content of the electrolyte in the sensor decreases, resulting in a decrease in the measured current signal). The sensor correction formula is calculated according to Formula 4.

[0084] Formula 4

[0085] Sen 修正 : Corrected sensitivity value

[0086] Sen 原始 : Sensitivity value before correction

[0087] t 高湿 : Accumulated working time when humidity is greater than 50%

[0088] t 低湿 : Less than 15% humidity cumulative working time

[0089] t 总时间 : Total working time of the sensor

[0090] From the data in Table 2 of the previous example, t 高湿 =6000,t 低湿 =1500,t 总时间 =13500, then we can get Sen 修正 =1.33Sen 原始 , use the Sen correction value to enter formula 2 to obtain the corrected gas concentration value.

[0091] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for managing the entire life cycle of an electrochemical gas sensor, characterized in that: The following steps are involved: S1: monitor the data noise of the sensor in real time to obtain the sensor noise data; S2: Count the accumulated working time of the sensor under different gas concentration, temperature, humidity and atmospheric pressure environmental parameters; S3: The sensor design usage time, sensor noise data, and working time under different environments are brought into the sensor life cycle model to calculate the remaining service life of the sensor; The sensor life cycle model formula is as follows: Formula 2 t1: remaining usage time of the sensor t2: sensor design usage time n1: expected sensor noise n2: actual sensor noise k: Correction value of working time in different environments t: sensor working time under different environments; The S1 step is as follows: First, read the real-time data from the sensor and calculate the average value of the 90-second data; Secondly, the current data trend is obtained according to the trend prediction algorithm. The implementation process of the trend prediction algorithm includes four data trends, namely: real-time trend, current trend, historical trend and temporary trend; four trend states, namely: rising state, falling state, fluctuating state and stable state; the steps of the trend prediction algorithm are as follows: first, the real-time trend is obtained according to the real-time data and the data mean value. The real-time data is greater than the data mean value for the rising state, and vice versa for the falling state; secondly, if the historical trend state is a fluctuating state, if the real-time trend is the same as the temporary trend for more than 10 times, the temporary trend is written into the current trend, and if it is not more than 10 times, the real-time trend is written into the temporary trend; if the real-time trend is the same as the historical trend, the historical trend is written into the current trend, and if the number of different times is greater than 10 times, the current trend state is modified to the fluctuating state; finally, the current trend is output; wherein the real-time trend is: the trend calculated according to the single real-time data and the data mean; the current trend is: the current data trend obtained by integrating the historical trend, the temporary trend and the real-time trend; the historical trend is: the current trend calculated last time; the temporary trend is: when the real-time trend is different from the historical trend, the last real-time trend stored; Finally, when the data trend is in a fluctuating state or a stable state, the real-time noise of the data is calculated according to Formula 1; VN rtd = |VAL rtd -VAL avg | Formula 1 VN rtd :Data real-time noise VAL rtd :Real-time value of data VAL avg : Data average value for 90 seconds.

2. The electrochemical gas sensor full life cycle management method according to claim 1, characterized in that: In S2, the gas concentration is divided into five categories: greater than 80% of the range, greater than 50% of the range and less than 80% of the range, greater than 30% of the range and less than 50% of the range, greater than 15% of the range and less than 30% of the range, and less than or equal to 15% of the range. The sensor working time is counted once a minute. Secondly, in each category, the temperature is divided into five categories: greater than 40℃, 30℃~40℃, 4℃~30℃, -10℃~4℃, and less than or equal to -4℃, and the working time is counted respectively. Finally, the humidity is divided into five categories: 80%~100%, 50%~80%, 30%~50%, 15%~30% and less than or equal to 15%, and the working time is counted respectively. A total of 125 subcategories are divided into statistical sensor working time.

Images