A dynamic early warning method for atmospheric pollutants

Abstract

The present invention provides a dynamic early warning method for atmospheric pollutants, comprising the following steps: (A1) a monitoring instrument obtains continuous monitoring data Y of atmospheric pollutants; m ; Y m =Y b +Y p +a,Y b is the background value, Y p is the added value affected by pollution, and a is the observation error; (A2) an estimated background value #imgabs0# is obtained based on the continuous monitoring data; a background curve for each pollutant is fitted based on the estimated value #imgabs1#; (A3) a local linear regression algorithm is used to predict the background curve, thereby providing a predicted background concentration value C; (A4) a dynamic threshold is set based on the predicted background concentration value C; if the actual monitoring data is higher than the dynamic threshold, pollution is indicated. The present invention has the advantages of accurate early warning.

Description

Technical Field

[0001] The present invention relates to atmospheric monitoring, and in particular to a dynamic early warning method for atmospheric pollutants. Background Art

[0002] Many cities and industrial areas have established automatic atmospheric monitoring stations and set thresholds to provide early warnings for environmental pollution. However, the concentration of atmospheric pollutants often fluctuates due to seasonal changes, industrial restructuring, and other factors. For continuous online monitoring of atmospheric pollutants, fixed values ​​are usually used as early warning lines. This may result in the inability to identify low-level pollution events or single pollution events when the background concentration is high. Figure 1 As shown; Due to different toxicities, different pollutants have different concentrations of pollution and harm. Currently, there are no relevant technical standards to give concentration limits for many factors; the limits for toxic and harmful gases such as VOCs in relevant standards are mostly set according to the concentrations that affect the body, which is not applicable to assessing the degree of environmental pollution. Summary of the Invention

[0003] In order to solve the deficiencies in the above-mentioned prior art solutions, the present invention provides a dynamic early warning method for atmospheric pollutants.

[0004] The purpose of the present invention is achieved through the following technical solutions:

[0005] A dynamic early warning method for atmospheric pollutants, comprising the following steps:

[0006] (A1) Monitoring instruments obtain continuous monitoring data Y of air pollutants m ;

[0007] Y m =Y b +Y p +a,Y b is the background value, Y p is the added value affected by pollution, and a is the observation error;

[0008] (A2) deriving an estimated value of the background value based on the continuous monitoring data

[0009]

[0010]

[0011] t0, t i is the time point, σ is the standard deviation, k1 is the coefficient, and h is the time length;

[0012] According to the estimated value Fit the background curve of each pollutant;

[0013] (A3) using a local linear regression algorithm to predict the background curve, thereby providing a predicted value C of the background concentration;

[0014] (A4) setting a dynamic threshold value according to the predicted value C of the background concentration;

[0015] If the actual monitoring data is higher than the dynamic threshold, it indicates that pollution exists.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] This technology combines an approximate regression algorithm with the introduction of multiple weight factors W and K in long-term data fitting. Taking into account the weight changes of monitoring data such as outliers, it extracts the background change curve of the gas and predicts the trend of background concentration in a short period of time. At the same time, it provides a reasonable fluctuation range for the background value. Values ​​outside the reasonable change range (dynamic threshold) are considered to be affected by pollution emissions and trigger early warnings, thereby achieving:

[0018] Regardless of long-term high or low concentration periods, the monitoring values ​​of the pollution period can be effectively extracted;

[0019] Due to different sources and properties, the background values ​​of each gas vary greatly. The use of dynamic thresholds can apply a set of exclusive thresholds to each gas, achieving accurate early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The disclosure of the present invention will become more easily understood with reference to the accompanying drawings. Those skilled in the art will readily appreciate that these drawings are merely used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0021] Figure 1 It is a schematic diagram of thresholds and unidentified pollution events according to the prior art;

[0022] Figure 2 is a flow chart of a dynamic early warning method for atmospheric pollutants according to an embodiment of the present invention;

[0023] Figure 3 Schematic diagram of a threshold value, a background curve, and a measured value according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] Figure 2-3The following description describes alternative embodiments of the present invention to teach those skilled in the art how to implement and reproduce the present invention. In order to teach the technical solution of the present invention, some conventional aspects have been simplified or omitted. Those skilled in the art will understand that variations or substitutions derived from these embodiments will be within the scope of the present invention. Those skilled in the art will understand that the following features can be combined in various ways to form multiple variations of the present invention. Thus, the present invention is not limited to the following alternative embodiments, but is limited only by the claims and their equivalents.

[0025] Example 1:

[0026] Figure 2 A flow chart of a dynamic early warning method for atmospheric pollutants according to an embodiment of the present invention is schematically provided. Figure 2 As shown, the atmospheric pollutant dynamic early warning method includes the following steps:

[0027] (A1) Monitoring instruments obtain continuous monitoring data Y of air pollutants m Specific monitoring instruments and monitoring methods are existing technologies in this field;

[0028] Y m =Y b +Y p +a,Y b is the background value, Y p is the added value affected by pollution, and a is the observation error;

[0029] (A2) deriving an estimated value of the background value based on the continuous monitoring data

[0030]

[0031]

[0032] t0, t i is the time point, σ is the standard deviation, k1 is the coefficient, and h is the time length;

[0033] According to the estimated value Use Python or R language to fit the background curve of each pollutant online;

[0034] (A3) using a local linear regression algorithm to predict the background curve, thereby providing a predicted value C of the background concentration;

[0035] (A4) setting a dynamic threshold value according to the predicted value C of the background concentration;

[0036] If the actual monitoring data is higher than the dynamic threshold, it indicates pollution and an early warning is issued.

[0037] In order to set a reasonable dynamic threshold, the dynamic threshold is further set as follows:

[0038] The concentration of k2 times the standard deviation of the predicted value C is used as the dynamic threshold, where k2 is a coefficient and 1.5≤k2≤3.

[0039] In order to accurately predict the background concentration, further, in step (A3), the background curve is continuous and the background curve is estimated using a robust local linear regression algorithm.

[0040] To improve the estimated The accuracy is further improved. In step (A2), the time length h is set as follows:

[0041] The pollutant has been locally contaminated for more than three times the longest time.

[0042] Example 2:

[0043] An example of application of a dynamic early warning method for atmospheric pollutants in an industrial park according to Example 1 of the present invention.

[0044] In this application example, Figure 1 As shown, the dynamic early warning method for atmospheric pollutants includes the following steps:

[0045] (A1) The monitoring instrument obtains continuous monitoring data Y of air pollutants within one week m Specific monitoring instruments and monitoring methods are existing technologies in this field;

[0046] Y m =Y b +Y p +a,Y b is the background value, Y p is the added value affected by pollution, and a is the observation error;

[0047] (A2) Based on the continuous monitoring data, further distinguish the Y in the measured data b and Y p , and obtain an estimate of the background value for,

[0048]

[0049] Y m (t0)≥Y b (t0), t0, t iis a time point, identifying outliers and extreme values ​​in the data set, giving lower weights to values ​​far from the background, and giving weights of 0 to values ​​higher than k1=3 times the standard deviation σ, so that the influence of extreme values ​​and severely polluted points on the overall background fitting process is close to 0, or no influence. For example, when estimating the background value at time t0, if Y m (t0)=5, then the weight is 1, if Y m If (t0)=100, the weight is 0, which has no effect on the calculation of the background value. In this way, the abnormally high value caused by emissions will not increase the background value with a stable trend;

[0050] The position weight factor K is set so that the values ​​far from the center have the smallest weight, and the influence on the fit outside a certain range is zero. For example, when h is set to 24 hours, when calculating the background value at 12:00 on a certain day, according to the following formula, K = 0.96 at 13:00 on the same day, and K = 0 at 12:00 the next day. It can be seen that the closer the time difference, the greater the influence on the current moment. When the time difference is greater than h, the value has no effect on the current background estimate.

[0051] The time length h is set to at least three times the longest local pollution duration of the pollutant. For example, if PM10 pollution in a certain area has lasted for 5 days recently, the time length h is 15 days. VOCs pollution generally lasts for one or two days at most, so the time length h is 7 days. For carbon dioxide, which varies with the seasons, the time length h can be set to one month or more.

[0052] Based on the estimated background value Use Python to fit the background curve of each pollutant online;

[0053] (A3) Since background concentration is the concentration value observed when it is not affected by regional or local pollution sources, it can reflect its long-term trend and seasonal variation characteristics. Therefore, we assume that the background changes continuously and consider the background curve to be a straight line within a sufficiently small range. We use the robust local linear regression algorithm to predict the background curve, thereby providing the predicted background curve, that is, the predicted value C of the background concentration;

[0054] (A4) setting a dynamic threshold value based on the predicted value C of the background concentration: using a concentration k2 times the standard deviation of the predicted value C as the dynamic threshold value, where k2 is a coefficient, 1.5≤k2≤3, and in this embodiment, k2=3;

[0055] If the actual monitoring data is higher than the dynamic threshold, it indicates that there is pollution and an early warning is issued, such as Figure 3 shown.

[0056] Example 3:

[0057] The application example of a dynamic early warning method for atmospheric pollutants in a city according to Example 1 of the present invention is different from Example 2 in that:

[0058] 1. Compared with the larger coefficient k2 value in industrial parks, the coefficient k2 value in urban early warning is smaller. In this embodiment, k2 = 1.5, or 1.8, 2.0, etc.;

[0059] 2. Based on the estimated value The background curves of each pollutant were fitted online using R language.

Claims

1. A dynamic early warning method for atmospheric pollutants, characterized in that: The air pollutant dynamic early warning method comprises the following steps: (A1) Monitoring instruments obtain continuous monitoring data Y of atmospheric pollutants m ; Y m =Y b +Y p +a,Y b is the background value, Y p is the added value affected by pollution, and a is the observation error; (A2) Estimated background value based on the continuous monitoring data ; ， ； ， ； t0, t i is the time point, σ is the standard deviation, the coefficient k1=3, and h is the time length; According to the estimated value Fit the background curve of each pollutant; (A3) Using a local linear regression algorithm to predict the background curve, a predicted value C of the background concentration is obtained; (A4) setting a dynamic threshold value according to the predicted value C of the background concentration; If the actual monitoring data is higher than the dynamic threshold, it indicates pollution and an early warning is issued; The dynamic threshold is set as follows: The concentration of k2 times the standard deviation of the predicted value C is used as the dynamic threshold, where k2 is a coefficient and 1.5≤k2≤3.

2. The dynamic early warning method for atmospheric pollutants according to claim 1, characterized in that: In step (A3), the background curve is continuous and is estimated using a robust local linear regression algorithm.

3. The dynamic early warning method for atmospheric pollutants according to claim 1, characterized in that: In step (A1), obtain continuous monitoring data Y for more than one week m .

4. The dynamic early warning method for atmospheric pollutants according to claim 1, characterized in that: In step (A2), the time length h is set as follows: The pollutant has been locally contaminated for more than three times the longest time.

Images