CN112567241A

CN112567241A - Environmental sensor collaborative calibration method

Info

Publication number: CN112567241A
Application number: CN201980006118.6A
Authority: CN
Inventors: 贾帅帅; 刘一平; 万秀福; 司书春; 许军
Original assignee: Nova Fitness Co Ltd
Current assignee: Nova Fitness Co Ltd
Priority date: 2018-07-25
Filing date: 2019-02-15
Publication date: 2021-03-26
Also published as: WO2020021344A1

Abstract

A method for cooperatively calibrating an environmental sensor relates to the field of environmental monitoring. The method relates to an alpha data set from an alpha monitoring station, a beta data set from a beta monitoring station and a gamma data set from a gamma monitoring station, and environmental sensors of the monitoring stations adopt a mode of mutual calibration and comparison of multiple data, so that data deviation complementation and mutual verification are realized, and the reliability, consistency and precision of the sensors are improved.

Description

The invention relates to a method for cooperatively calibrating an environmental sensor, belonging to the technical field of environmental monitoring. Background art atmospheric pollutants monitored in environmental monitoring are sulfur dioxide, nitrogen oxides, ozone, carbon monoxide, PM1 (particles with an aerodynamic particle size of less than 1 micron), PM2.5 (particles with an aerodynamic particle size of less than 2.5 microns), PM10 (particles with an aerodynamic particle size of less than 10 microns), PM100 (particles with an aerodynamic particle size of less than 100 microns), and VOCs (volatile organic compounds) or TVOCs (total volatile organic compounds) in the atmosphere. The atmospheric environment monitoring system can collect and process the monitored data and timely and accurately reflect the air quality condition and the change rule of the regional environment. The existing atmospheric environment monitoring equipment mainly comprises a fixed monitoring station and a movable monitoring device. The existing fixed monitoring stations are mainly divided into large fixed monitoring stations and small stations. The movable monitoring equipment mainly comprises a special atmospheric environment monitoring vehicle, an unmanned aerial vehicle, handheld equipment and the like. The small monitoring station and the handheld device both use the air quality sensor to measure pollutants in the atmosphere. The small sensor has the characteristics of low cost, miniaturization and online monitoring, and can be used in a large scale. The air quality sensor itself may have errors due to inconsistency between the measured value and the actual value for various reasons. Compared with a large-scale precision instrument or a manual monitoring mode, the air quality sensor has the characteristics of lower accuracy, poor stability, large error and frequent calibration. The air pollution particle sensor based on the laser scattering method has wide market prospect due to low cost and portability. However, the portable analysis device adopting the scattering method has the defects of poor measurement consistency, high noise, low measurement precision and the like, and the core device is easily influenced by various environmental factors, so that the fluctuation easily causes misjudgment. When the sensor data change suddenly and greatly, whether the change reason is sensor fault or sudden pollution can be intelligently judged, the data reliability can be greatly improved, and the method has important value for ensuring the quality of environment-friendly monitoring data. When equipment breaks down, if can pass through automatic restoration, also can improve the online rate of data by a wide margin, have important value to the required continuous monitoring of haze work. Meanwhile, manpower and material resources in the aspect of equipment maintenance can be saved, and social waste is reduced.

Disclosure of Invention

Interpretation of terms a dataset: monitoring data of a reference station (a state control station, a city control station and an independently arranged calibration station); al represents the data or data group of the reference station at time T = 1; a1 represents data of a national control station at time T = 1.

P data set: monitoring data of a fixed monitoring station, (31 represents data or a monitoring data group of the fixed atmosphere gridding micro-station at the time T =1, and B1 represents data of the fixed atmosphere gridding micro-station at the time T = 1.

Y data set: monitoring data of the mobile monitoring station, wherein Yl represents data or a monitoring data group of the fixed atmospheric gridding micro-station at the time T = 1; y1 represents data for a fixed atmospheric gridding micro-station at time T = 1. Fixing a monitoring station: the stations with the atmospheric environment monitoring capability can be a state control station, a calibration station and a gridding micro station. The mobile monitoring station: and the monitoring station carries atmospheric environment monitoring equipment and has the mobility. The vehicle can be a social vehicle carrying micro monitoring equipment or a professional atmospheric environment monitoring vehicle. Contrast ratio: the quantity, which represents the degree of linear correlation between the variables, is generally indicated by the letter tl. Calibration coefficients: calibration coefficients are referred to herein as correction coefficients for calibrating, correcting for, or correcting for, deviations in the data set of the sensor. The light scattering method is easy to influence the measurement precision of the particles due to environmental factors, such as humidity and the like. Various ways of improving the accuracy of the sensor have also emerged. The current monitoring station calibration mode mainly adopts regular manual maintenance, and workers clean and maintain equipment on site and carry calibration equipment and standard gas to manually calibrate a sensor on site. Or simply perform coefficient modification on the monitoring device. Aiming at the defects, the invention provides a method for cooperatively calibrating environmental sensors, and the environmental sensors adopt a mode of mutually calibrating and comparing multiple data, thereby realizing data deviation complementation and mutual verification, improving the reliability, precision and service life of the sensors. The calibration method provided by the invention relates to the calibration method from(a dataset from X monitoring station, p dataset from p monitoring station, and p dataset from p monitoring station_YOf monitoring stations_YA data set; firstly, screening out a reference data set serving as a calibration basis; this may be a subset of the a data set; or may be a subset of the (3 data set or a subset of the Y data set.

The method comprises the following specific steps:

1) firstly, a data set, data set,_YAt least two of the data sets;

2) screening out a reference data set serving as a calibration basis and a calibrated data set;

3) obtaining reference data (y) from a reference data set, and obtaining calibrated data (x) from a calibrated data set;

4) obtaining a contrast factor (11) from the reference data (y) and the calibrated data (x);

5) and calculating a calibration coefficient (c), and calibrating by the calibrated monitoring station by using the calibration coefficient (c).

As the reference data set, a certain condition must be satisfied. First, the reference data set, when used to calibrate other monitoring stations at some distance, should have data that is relatively stable over a period of time without significant data fluctuations; this stability of the data over time is indicative of a range of air quality stability due to natural diffusion of the atmosphere, and the reference data set selected at this time should be capable of representing a range of data. We characterize the stability of the data in a subset of a data set by the stability factor of this subset. In the case of a calibration that is not close-range, only a subset of the data set whose stability factor reaches the desired value can be selected as the reference data set. Calibration of P and Y datasets based on cx datasets a first calibration approach proposed by the present invention is to calibrate and Y datasets based on the a dataset. In the case where the uniformity requirement is met for the dataset and the Y dataset, a baseline a dataset is determined by analyzing the data for the a dataset. The method for analyzing the data set a comprises a direct average value method and an average method after the highest value and the lowest value are removedKalman filtering, Bayesian estimation, D-S evidence reasoning, artificial neural network and other methods. After the reference a dataset is determined, the calibration coefficients for the dataset are derived by comparing the dataset with the reference a dataset for calibrating the P dataset. Similarly, the comparison between the Y data set and the reference a data set is used to obtain_YCalibration coefficients of the data set are used to calibrate the Y data set. The comparison method can adopt a linear calibration method, and can also adopt a nonlinear calibration method and other calibration methods. In the calibration process, a plurality of calibration coefficients are generally calculated, the calibration coefficients with the coefficient difference smaller than a certain value are taken as effective calibration coefficients, and the average value of the effective calibration coefficients is taken as a final calibration coefficient to calibrate the calibration object. The calibration coefficients may also need to take into account the spatial distribution. And (3) the calibration coefficients of the data set can be sorted according to the weight of the distance between the 3 station and the a station, the closer the distance is, the larger the weight is, in the case that the station is within a certain distance from the a station, the weighted average value is taken as the accurate value of the calibration target, and in the case that the Y station takes the data passing through the a station within a certain distance, the effective data is taken to participate in the calibration calculation.

The calibration coefficients of the Y data set may also be determined according to different data intervals of the data, i.e. a plurality of calibration coefficients are set in different data intervals. The calibration coefficients in different intervals can still be selected by using a direct average method, an average method after the highest value and the lowest value are removed, and the like. Based on the Y data set, the calibration P data set is based on the Y data set, the calibration (3 data set, when the mobile monitoring station is less than the calibration trigger distance t from the fixed monitoring station, the method comprises the steps of taking a Y data set of a mobile monitoring station as a reference, comparing a (3) data set of the fixed monitoring station with the Y data set of the mobile monitoring station to obtain a calibration coefficient, and calibrating the fixed station by utilizing the calibration coefficient and other calibration factors.And when the station distance is less than the calibration triggering distance t, comparing the Y data set of the selected mobile monitoring station with the Y data sets of other mobile monitoring stations by taking the Y data set of the selected mobile monitoring station as a reference to obtain a calibration coefficient, and calibrating the fixed station by using the calibration coefficient and other calibration factors. The calibration method can select linear calibration or non-linear calibration, and the calibration trigger distance t can be 500m, lkm_s 2km _s5km equidistant. Ranking calibration the third calibration method proposed by the present invention is to rank the P data set and the Y data set with confidence, and then calibrate from a device with high confidence to a device with low confidence. The confidence level may be a comparison of the reference data with the calibrated number, and may also be other parameters of the monitoring device such as a calibration time factor (indicating the time since the last calibration), a stability factor, etc.

Arrangement name indication table for P data set and Y data set

And comparing the P data set with the Y data set with the a data set to obtain a reliability index, wherein the comparison mode can be a correlation coefficient mode, a proportional mean value mode and the like. And after the credibility index is obtained, ranking the credibility from high to low, and calibrating the ranked data set, wherein the calibration mode adopts a first method. And after calibration, recalculating the credibility for ranking. And for the P fixed station, selecting a country a control station within a certain range to carry out reliability calculation. Under the condition that a plurality of national control stations exist in a certain range, the credibility can be the average value of the plurality of national control stations, and can also be calculated by taking the distance as the weight to carry out weighted average; and under the condition that no national control station exists in a certain range, carrying out reliability calculation by using the mean value of the data set a of the whole city. And for the Y mobile station, carrying out reliability calculation on the data after the Y mobile station moves to a country control station a within a certain range. After ranking, the data sets with higher ranking are compared with the data sets with lower ranking, the calibration coefficient is calculated, and the data sets with lower ranking are calibrated by the data sets with higher ranking. The cooperative monitoring station method can communicate with the mobile monitoring station when the monitoring data of the fixed monitoring station is abnormal, control the working state of the sensor of the mobile station and improve the monitoring frequency and the data return frequency. The data abnormity can be judged by judging that the contrast coefficient exceeds a set range, namely, the data abnormity of the station is judged. The related calculation method comprises the following comparison formula:

X

T1 =—

y_c

x is the calibrated data, y_cFor the corrected reference data, ri is a contrast ratio.

Calibration formula:

x' is calibrated data, C is a calibration coefficient, and C can be obtained by an average value of T1 or T1 through other mathematical operations. Other mathematical operations can be an averaging method after the highest value and the lowest value are removed, Kalman filtering, Bayesian estimation, D-S evidence reasoning, artificial neural network and other methods. Method for obtaining the calibration coefficient C: a) the calibration coefficient is the average of the contrast coefficients, c = fi

b) The calibration coefficient is the median of the contrast coefficient c) the calibration coefficient is the mode of the contrast coefficient d) the calibration coefficient is the value of the contrast coefficient calculated by other probability methods. Distance factors can also be taken into account in the calibration process, and the distance factors are introduced. Wherein the distance factor-_dThe method is used for considering the influence on the reliability of the monitoring data caused by the distance factor between the monitored point and the monitoring station. The distance factor can be obtained by the inverse normalization of the distances from the geometric center point of a certain area to each monitoring point according to the weight of the area occupied by the data acquired by the monitoring stations in the area; the distance factor is also embodied in a way that the pollution data of a specific position consists of the monitoring data of a plurality of monitoring stations which are close to each other, the monitoring data can have different weights for the pollution data of the specific position, the weights are obtained by inverse normalization of the distances from the specific position to the monitoring points, and the weights are the distance factorsAnd (4) adding the active ingredients. In the calculation, d represents the distance from the geometric center point of the area to each station in the area or the distance from the specific position to each monitoring point, and the set value of the distance is represented by A. And after the set distance A is exceeded, the farther the distance is, the smaller the weight occupied by the data of the monitored station is, and the closer the distance is, the larger the weight occupied by the data of the monitored station is. The distance factor is calculated by the formula: d is d> A

0 < d <A d indicates the distance from the geometric center point of the area to each station within the area or the distance from the particular location to each monitoring point. The parameter K is a distance weight parameter, and the smaller the distance weight is, the larger the influence of the representative distance is, and in general, after K = a applies a distance factor, the reference value data may also be obtained by normalization operation. And correcting the data of the reference station by applying a normalization method reference station correction calculation formula, wherein the normalization calculation formula comprises the following steps:

y_cin order to obtain the corrected reference data,

y' is the uncorrected reference station data,

n is the number of reference stations meeting the standard. In the case of only one reference station that meets the standard, the correction of the reference data is calculated as follows:

y_c = y - f_dthe x (x-y) stability coefficient A is obtained by the following method: a) The stability factor 2 is a ratio value of the number of reference station data in the set section to the total number of reference station data. If 2 is greater than the set percentage (the set percentage may be 80%, 90%, etc., other percentages), the reference station data set is considered stable, and a higher a represents a more stable data set. The set interval is a range given reference data within a set T time range, and the mathematical expression of the set interval is

(Y-UX Y, Y + UX Y), Y can be obtained by statistical methods such as the average value, median and mode of the data of the reference station in the T time range, and is a section coefficient.

The number of reference station data falling in a set interval in the T time range

The number of reference station data in the T time range b) the stability factor may also be related to the variance of reference data in the set time range if the base in the T time range is set>Variance setting value S3, unstable data variance within the set T time range<Variance setting value B, stability C) the stability factor may be related to a standard deviation of the reference data within a setting interval of a setting 1T time range, and if the variance setting value C is determined by a standard deviation ñ of the reference data within the setting T time range, the stability factor may be unstable if the standard deviation of the reference data within the setting T time range is determined<The variance setting value C needs to have sufficient reliability for the mobile monitoring station to be used as a calibration reference. When the mobile monitoring station adopts the redundancy multi-sensor design, the credibility of the mobile monitoring station can be greatly improved. The high-low-frequency sensor discloses an atmosphere pollution detection device in the prior application PCT/IB2018/05531, wherein the atmosphere pollution detection device, namely a mobile monitoring station comprises a main control module and a detection module; the detection module adopts at least four sub-sensor units to form a sensor module; when the main control module finds that one of the sub-sensor units is suspected to be abnormal and judges that the suspected abnormal sub-sensor is an abnormal sub-sensor, the abnormal sub-sensor is isolated, the abnormal sub-sensor belongs to the isolation region, and the multi-core sensor module continues to normally work after being degraded. The application further discloses another atmosphere pollution detection device, which comprises a main control module and a detection module; the detection module comprises at least two similar sub-sensor units which form a sensor module; the sub-sensor units operate at a normal operating frequency. The detection module alsoThe low-frequency calibration module comprises at least one sub-sensor unit which is similar to the sensor module; the sub-sensor units in the low-frequency calibration module work at a frequency far lower than the working frequency of the sensor module. Therefore, the low frequency calibration module is also called a low frequency set. By contrast, the sensor module is also referred to as a high frequency module. Typically, the operating frequency of the sensor module is 10 times or more the operating frequency of the low frequency calibration module. The ratio of the operating frequencies of the high and low frequency groups, referred to as the high and low frequency ratio, may be selected as: 2:1, 3:1, 4:1, 5:1, 6:1, 7:1, 8:1, 9:1, 10:1, 15:1, 20: l_oThe working frequency of the low frequency group can be consistent with the rhythm of the abnormity judgment. That is, when it is necessary to determine whether there is an abnormal sub-sensor in the sensor module, the low frequency group performs the detection operation. Since laser power decay is slow most of the time during the working life of the laser sensor, the accuracy of its data can be recovered by calibration; i.e. to calibrate sub-sensors with a high degree of attenuation using sub-sensors with no or very low degree of attenuation. In the operation process of the sensor module, the low-frequency group detection data is used as a reference to calibrate the high-frequency group detection data at regular intervals, such as 1 day, 1 week or 1 month, and the calibration coefficient can be obtained by using the ratio of the average value of the detection data of the high-frequency group sensor to the average value of the detection data of the low-frequency group sensor. In addition to the optical attenuation effect of the laser sensor, other types of sensors may have a tendency to have unstable performance or increased data errors under long-term high-load operation. By introducing a low-frequency group, the low-frequency group can be used as a relatively reliable reference for judging whether the sensor module has a data offset phenomenon. Meanwhile, because the data of the low-frequency group is generally higher in credibility, when the sub-sensor unit in the sensor module is judged to be suspected abnormal or abnormal, more credible judgment can be made by increasing the data weight of the low-frequency group. A simple scheme is that all low-frequency group data participate in suspected abnormality judgment according to double weight. Isolation and recovery of the previous application PCT/IB2018/05531 also discloses a set of sub-sensors for identifying their operating states and for identifying their operating statesA method for isolation and recovery. The sensor module obtains a group of detection data at a moment, and the main control module screens out suspected abnormal data from the group of data so as to judge whether the corresponding sub-sensors meet the isolation condition. Judging whether the sub-sensor is an abnormal sub-sensor and then classifying the abnormal sub-sensor into an isolation region; and after the sub-sensor suspected to be abnormal is judged not to meet the isolation condition, the sub-sensor continues to work normally. And judging whether the sub-sensor entering the isolation area can self-heal, and if the sub-sensor can self-heal, performing frequency reduction processing on the self-healable sub-sensor, wherein data output by the sub-sensor does not participate in the calculation of data output by the main control module. And stopping working of the sub-sensors which cannot self-heal, and informing the operation maintenance party to carry out maintenance or replacement. For the sub-sensors after frequency reduction, the main control module detects the data output by the sub-sensors, judges whether the sub-sensors reach the recovery condition, adjusts the sub-sensors reaching the recovery condition from an isolation area, recovers the work, and outputs the data to participate in the calculation of the sensor module data or the main control data; and judging whether the abnormal sub-sensor which does not meet the recovery condition can be self-healed again. After the abnormal sub-sensors in the sensor module are isolated, the average value of the output data of the remaining sub-sensors is used as the output result of the sensor module, and the sensor module can be continuously and normally used.

Description of the drawings fig. 1 is a schematic diagram of the components of a calibration system; FIG. 2 is a schematic diagram of a calibration dataset and a Y dataset based on the a dataset; FIG. 3 is a schematic diagram of a calibration data set based on the a data set; FIG. 4 is a schematic view of a calibration area range; FIG. 5 is a flow chart for calibrating the data set and the Y data set based on the a data set; FIG. 6 is a flow chart of calibrating a data set based on a Y data set; FIG. 7 is a flow chart for calibrating a Y data set based on the Y data set; FIG. 8 is a flow chart of rank calibration; in the figure: reference numeral 10 denotes a reference station, 20 denotes a fixed monitoring station, 30 denotes a mobile monitoring station, 40 denotes a data center, 50 denotes a user, 101 denotes a reference station No. 1, 102 denotes a reference station No. 2, 103 denotes a reference station No. 3, 201 denotes a fixed calibrated station No. 1, 202 denotes a fixed calibrated station No. 2, 203 denotes a fixed calibrated station No. 3, 301 denotes a mobile calibrated station No. 1, 302 denotes a mobile calibrated station No. 2, and 501 denotes a reference station calibration range.

Embodiment mode as shown in fig. 2, there are No. 1, No. 2, and No. 3 fixed domestic control reference stations, and one fixed calibrated station in the area. Taking monitoring values of four national control reference stations at four moments of Tl, T2, T3 and T4 and monitoring data average values of the fixed national control reference stations, and calculating the monitoring data average values of the fixed national control reference stations as shown in the following table, wherein the ratio n of the average values of the calibrated micro station and the national control reference station is respectively 1.2, 1.1 and 1.04 according to the data in the table, so that the average value of n = 1.11 is obtained through calculation. In the comparison of the embodiment, the arithmetic mean is used for calculating the mean value of the monitoring data of the three national control reference stations, and the calculation can also be carried out by using a normalization method. Similarly, the Y data set may be calibrated to obtain the ri average value based on the a data set

Second embodiment as shown in fig. 3, the monitoring data of the fixed station to be calibrated No. 1, station to be calibrated No. 2, and station to be calibrated No. 3 in the known diagram are =120, (3)₂=115、 p₃And =110, the data of the reference station is a =110, the distances from fixed micro stations of

numbers

1, 2 and 3 to the national control reference station are 5km, 6km and 7km respectively, the calibrated stations are sorted according to the data and the credibility, the most accurate micro station data are used for downwards calibrating one by one to obtain calibration coefficients, and the data are counted in the following table:

calibrated after applying a calibration formula

Third embodiment as shown in fig. 3, the monitoring data of the fixed station to be calibrated No. 1, station to be calibrated No. 2, and station to be calibrated No. 3 in the known diagram are =120, (3)₂=115、 p₃And =110, the data of the reference station is a =110, and the distances from fixed

micro stations

1, 2 and 3 to the state control reference station are 5km, 6km and 7km respectively. The spatial distribution may also need to be considered in calibration. The calibration coefficients of the data set can be sorted according to the distance between the P station and the a station, the closer the distance is, the larger the distance factor is, the distance factor calculation formula is applied, A =5 is taken, the calibrated stations are sorted according to the distance factor, and the first ranked calibrated station is used for calibrating the ranked calibrated station.

After applying the calibration formula, the calibrated is

Example four monitoring data of fixed station No. 1 calibrated station, station No. 2 calibrated station, station No. 3 calibrated station, and station No. 4 calibrated station in the known diagram are =120, (3) respectively₂=115、 p₃=110、 p₄=150, the data of the reference station is a =110, the micro stations are sorted according to the above data according to the confidence level, and are calibrated one by one downward by using the most accurate micro station data to obtain the calibration coefficient, and the above data are counted in the schematic table 2. The reliability of the contrast coefficient within the range of 0.95-1.05 is specified to be higher, and calibration is not carried out; the contrast coefficient is between 1.05 and 1.2, and calibration is carried out; the contrast coefficient is more than 1.2 so as not to carry out calibration, serious faults can occur to the equipment, and an alarm is given to the control systemPrompting that the monitoring device requires manual maintenance. The calibration range is determined by a correlation coefficient, the device with the correlation coefficient larger than 0.9 is not calibrated, for the device with the correlation coefficient smaller than 0.9, calibration is performed with the reference data set as a target, in principle, the device with the first credibility is used as a calibration reference, if the device with the first credibility is fixed station, calibration is performed from the station around the (3 fixed station until all completion, if the device with the first credibility is Y mobile station, calibration is performed with the station passing around the device as a priority calibration object until all completion, the determination of the calibration range is determined by a proportional mean coefficient, the device with the proportional coefficient of 0.9-1.1 is not calibrated, and the device with the proportional coefficient of other ranges is calibrated with the reference data set as a target, in principle, the device with the first credibility rank is used as a calibration reference, if the device with the first credibility rank is fixed stations, calibration is performed from the station around the (3 fixed stations until all calibration is completed, and if the device with the first credibility rank is a Y mobile station, calibration is performed by using the station passing around the Y mobile station as a priority calibration object until all calibration is completed.

Example five as shown in fig. 2, in the area, there are reference stations No. 1, 2, and 3, two calibrated stations (31, yl. take monitoring values of four national control reference stations at Tl, T2, T3, and T4, respectively), and the average value of the monitoring data of each fixed national control reference station is counted as the following table, and the ratio n of the average value of the calibrated micro station to the national control reference station is calculated as 1.2, 1.1, and 1.04 according to the data in the table, and the calibration coefficient can be determined according to different data intervals of the data, that is, a plurality of calibration coefficients are set in different data intervals And taking an average value after removing the maximum value of the contrast coefficient at the value above 1.2.

Sixth embodiment knowing that at a certain time, the monitoring data of fixed national

control reference stations

1, 2, 3, 4 and 5 are Cd =120 and a respectively₂=115、 a₃=110 _{s 0(4}=150, a₅=120, when the area average is calculated, the area average =123, and the average of data within ± 10% of the average for each station data is used as a reference value for calibration.

The seventh embodiment provides that when the mobile monitoring device enters the area 5km around the reference station, the calibration procedure is started, and as shown in fig. 4, the vehicle 1 is located in the area 5km around the

monitoring stations

1, 2 and 3, and the vehicle 2 is not located in the area 5km around the

monitoring stations

1, 2 and 3, then the mobile monitoring device 1 starts the calibration procedure, and the mobile detection device 2 does not start the calibration procedure.

Claims

Claims

A method of environmental sensor collaborative calibration involving a data set from an a monitoring station, data set from an monitoring station, and data set from an a monitoring station_YOf monitoring stations_YA data set; the method comprises the following steps:

1) firstly, acquiring at least two of a data set a, a data set and a data set Y;

2) screening out a reference data set serving as a calibration basis and a calibrated data set;

3) obtaining reference data (y) from a reference data set, and obtaining calibrated data (x) from a calibrated data set;

4) obtaining a contrast coefficient (n) according to the reference data (y) and the calibrated data (x);

5) calculating a calibration coefficient (c), and calibrating by the calibrated monitoring station by using the calibration coefficient (c); characterized in that the data in the reference data set meet the requirements of a stability factor U).

2. The method of claim 1, wherein the reference dataset is a subset of an a dataset,

(a subset of the B data set or a subset of the Y data set; the calibrated data set is a subset of the data set or a subset of the Y data set.

3. The method of claim 1, wherein the reference dataset is a subset of an a dataset; the calibrated dataset is a subset of an dataset or a subset of a Y dataset; in calculating the calibration factor (c), the distance factor f is used_dAnd (6) making a correction.

4. The method of claim 1, wherein in step 2), the calibrated data set is from monitoring stations; the reference data set used as the calibration basis is from Y data sets of a plurality of mobile monitoring stations near the calibrated monitoring station; and selecting a reference data subset in the same time period as the data set of the calibrated monitoring station, and combining the reference data subset into a reference data set.

5. The method of claim 1, wherein in step 2), the calibrated data set is from a Y monitoring station; the reference data set used as the calibration basis is from Y data sets of a plurality of mobile monitoring stations near the calibrated monitoring station; and selecting a reference data subset in the same time period as the data set of the calibrated monitoring station, and combining the reference data subset into a reference data set.

6. The method as claimed in claim 1, wherein in step 2), a group of g monitoring stations and mobile monitoring stations with similar distances is selected; then, sequencing each) g monitoring station and each mobile monitoring station according to the credibility; selecting a monitoring station with the lowest reliability as a calibrated monitoring station; selecting a data subset meeting the requirement of a stability coefficient for a period of time as a calibrated data set; and selecting and combining the reference data subsets of a plurality of monitoring stations with high credibility in the same time period with the data set of the calibrated equipment into a reference data set.

7. Method according to one of claims 4 to 6, characterized in that the calculation of the reference data (y) in step 3) is based on a confidence factor and a distance factor (f) for the respective data in the reference data set_d）。

8. Method according to one of claims 1 to 6, characterized in that the stability factor (1) is taken to a value in the range of at least 80% >, the stability factor (A) being calculated by the formula: l T number of reference station data falling within set section within time range A _ T number of reference station data within time range

9. The method according to one of claims 1 to 6, characterized in that the stability factor (1) is related to the variance of the reference data within a set interval of a set T time range; whether the stability coefficient (1) meets the requirement is judged according to the following method:

if the variance of the reference data in the time range T is set to be less than the variance set value B, the requirement is met; otherwise, the requirements are not satisfied.

10. The method according to one of claims 1 to 6, characterized in that the stability factor (1) is related to the standard deviation of the reference data within a set interval of a set T time range; whether the stability coefficient (1) meets the requirement is judged according to the following method:

if the standard deviation of the reference data in the time range T is set to be less than the variance set value C, the requirement is met; otherwise, the requirements are not satisfied.

11. The method of claim 7, wherein the distance factor (f)_d) The calculation method comprises the following steps:

（ K

（ — _' d > A

/d⑷ = ] （ 1 ^d , 0 < d <a wherein the parameter d represents the geometric center point of the area to each station in the areaThe distance between points or the distance from the specific position to each monitoring point; the distance factor is 1 within the set distance A, after the set distance A is exceeded, the farther the distance is, the smaller the weight occupied by the monitoring station data is, and the closer the distance is, the larger the weight occupied by the monitoring station data is; the parameter K is a distance weight parameter-typically K = a.

12. Method according to one of claims 1 to 6, characterized in that the contrast factor (n) is calculated by:
where X is the calibrated data, y is the reference data, and r | is the contrast ratio.

13. The method of claim 7, wherein the baseline data is modified using a normalization algorithm, the normalization calculation formula being:

wherein the reference data is corrected; y _ is uncorrected reference station data; and n is the number of the standard-reaching reference stations.

14. The method of claim 13, wherein when n = l, the normalized calculation formula is:

y_c = y - f_d x (_.x - y)

15. the method according to one of claims 4 to 6, wherein the mobile monitoring station comprises a main control module and a detection module, wherein the detection module comprises at least two similar sub-sensor units which form a sensor module; the sub-sensor units work at a normal working frequency; the detection module also comprises at least one low-frequency calibration module consisting of sub-sensor units similar to the sensor module; the working frequency of the sub-sensor units in the low-frequency calibration module is far lower than that of the sub-sensor units in the sensor module.

16. The method of claim 15, wherein a ratio of the operating frequency of the sensor module to the operating frequency of the low frequency calibration module is: 2:1, 3:1, 4:1, 5:1, 6:1, 7:1, 8:1, 9:1, 10:1, 15:1, or 20: 1.

17. The method according to claim 15, wherein when the main control module finds that a sub-sensor unit in the sensor module is suspected to be abnormal and determines that the sub-sensor is an abnormal sub-sensor, the abnormal sub-sensor is isolated, the abnormal sub-sensor belongs to an isolation region, and the multi-core sensor module continues to work normally after being degraded; if the sub-sensor entering the isolation area can not self-heal, the sub-sensor stops working; if the self-healing is possible, the frequency reduction work processing is carried out, but the data output by the sub-sensors do not participate in the calculation of the data output by the main control module; the main control module monitors data output by the sub-sensors entering the isolation area and judges whether the data reach a recovery condition or not; and (4) the sub-sensors which reach the recovery condition are transferred away from the isolation region, and the work is recovered.