A data analysis system and method based on an environmental detection cloud platform

Abstract

The present invention discloses a data analysis system and method based on an environmental detection cloud platform, which relates to the technical field of environmental detection. A reference monitoring site is determined according to the correlation between the environmental detection data of all monitoring sites; the first environmental detection data and the second environmental detection data of the reference monitoring site are obtained through transformation, and the data distributions of the first environmental detection data and the second environmental detection data are analyzed; abnormal data is removed and filled; the environmental detection data is transformed to the same reference monitoring site for analysis, and the data distribution at the reference monitoring site is analyzed, which is less affected by the model accuracy, thereby improving the accuracy of the analysis result of the environmental detection data; in the case where abnormalities are determined to exist, the data distribution of the target monitoring site itself is analyzed, making the analysis result more persuasive; the abnormal data is filled, improving the integrity of the data.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental detection, and specifically to a data analysis system and method based on an environmental detection cloud platform. Background Art

[0002] Environmental detection is one of the important means of environmental protection. Through continuous environmental detection, the environmental quality status can be timely understood, problems of environmental pollution and environmental damage can be discovered, and scientific basis can be provided for environmental protection. At present, environmental detection has received extensive attention and implementation globally. Many countries and regions have established perfect environmental detection systems to continuously monitor and analyze environmental elements such as the atmosphere, water, soil, and organisms. At the same time, with the development of technology, the technologies and methods of environmental detection are also constantly updated and improved, improving the accuracy and efficiency of detection; however, due to the complexity of the outdoor environment, when detecting the environment, it is easily affected by natural or biological factors, resulting in abnormal detection data and affecting the detection results. Summary of the Invention

[0003] The purpose of the present invention is to provide a data analysis system and method based on an environmental detection cloud platform to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: A data analysis system based on an environmental detection cloud platform, including an environmental detection module, an environmental detection cloud platform, and a data transmission module; the output end of the environmental detection module is connected to the input end of the environmental detection cloud platform, and is used to obtain environmental detection data of a monitoring site and send it to the environmental detection cloud platform; the environmental detection cloud platform is used to convert the environmental detection data collected by the target monitoring site and other monitoring sites into environmental detection data of a reference monitoring site, judge the abnormal data of the distribution of the environmental detection data obtained by converting the target monitoring site, and analyze the central tendency of the target monitoring site under the condition of the existence of abnormal data, determine the starting point of the abnormal data, and remove and fill the abnormal data; the data transmission module is used to connect the environmental detection module and the environmental detection cloud platform, and support data exchange and sharing between the environmental detection module and the environmental detection cloud platform.

[0005] The environmental detection cloud platform further includes a data storage unit, a correlation analysis unit, a conversion unit, a standardization unit, a first analysis unit, a second analysis unit, and a filling unit; the correlation analysis unit is used for the correlation between the environmental detection data of all monitoring sites, and determines a reference monitoring site based on the correlation; the conversion unit is used to convert the environmental detection data collected by the target monitoring site and other monitoring sites into the first environmental detection data and the second environmental detection data of the environmental detection data of the reference monitoring site; the standardization unit is used to standardize the first environmental detection data and the second environmental detection data; the first analysis unit is used to analyze the data distribution of the first environmental detection data and the second environmental detection data; and determine the abnormal data of the target monitoring site based on the data distribution; the second analysis unit, under the condition that abnormal data of the target monitoring site is determined, analyzes the central tendency according to the mean and variance of the environmental detection data of the target monitoring site, and determines the starting point of the abnormal data; the filling unit is used to remove the abnormal data and fill the removed abnormal data with simulated data.

[0006] The standardization unit performs standardization through the following formula: Z = (z - az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization.

[0007] The conversion unit inputs the environmental detection data collected by the target monitoring site into the first neural network model, and obtains the first environmental detection data of the reference monitoring site; obtains the currently collected environmental detection data from other monitoring stations except the reference monitoring site, inputs it into the second neural network model, and obtains the simulated value of the second environmental detection data of the reference monitoring site; directly obtains the observed value of the currently collected environmental detection data from the reference monitoring site, and performs weighted summation on the simulated value and the observed value of the second environmental detection data of the reference monitoring site to obtain the second environmental detection data of the reference monitoring site.

[0008] To achieve the above object, the present invention provides the following technical solution: A data analysis method based on an environmental detection cloud platform, including the following steps:

[0009] S11, the environmental detection module obtains the environmental detection data of the monitoring site, and the environmental detection module sends the security data to the environmental detection cloud platform, and the monitoring sites are divided into a target monitoring site and other monitoring sites;

[0010] S12, the environmental detection cloud platform analyzes the correlation between the environmental detection data of all monitoring sites, and determines a reference monitoring site based on the correlation;

[0011] In S13, the environmental detection cloud platform converts the environmental detection data of the target monitoring site and the environmental detection data of other monitoring sites into the first environmental detection data and the second environmental detection data of the reference monitoring site, and analyzes the data distributions of the first environmental detection data and the second environmental detection data; determines whether there is abnormal data at the target monitoring site based on the data distributions. If there is abnormal data, determines the starting point of the abnormal data at the target monitoring site, removes and fills the abnormal data, and marks the filled environmental detection data and the environmental detection data before the abnormality as safe data; if there is no abnormal data, marks the environmental detection data as safe data.

[0012] There is a situation where the environmental detection data is first marked as safe data and then detected as abnormal. This is because detecting an abnormality requires the abnormal data to occupy a certain share in the overall data. Therefore, the criterion for marking environmental detection data is last-in-first-out, that is, the new judgment result overwrites the old judgment result.

[0013] In step S12, the steps of analyzing the correlation between the environmental detection data of all monitoring sites and determining the reference monitoring site based on the correlation further include:

[0014] Obtain the historical environmental detection data of all monitoring sites, select consecutive n environmental detection data from the environmental detection data of two monitoring sites, and calculate the correlation T between the environmental detection data of the two monitoring sites ab , where a i and b i are the observed values of the environmental detection data of the two monitoring sites, and are the averages of the observed values of the environmental detection data of the two monitoring sites, n is the sample size, and n is the sample number;

[0015] Add all other monitoring sites to the set U, and obtain the correlation T between the target monitoring site and the jth other monitoring site j where j is the monitoring site number. Remove the jth other monitoring site from the set U, and calculate the average value TU of the correlations between the remaining monitoring sites in the set U and the jth other monitoring site j Add T j and TU j to obtain the estimated value of the jth other monitoring site; select the other monitoring site with the highest estimated value as the reference monitoring site.

[0016] Analyze the correlation between monitoring stations. The higher the correlation with the reference monitoring station, the more the environmental detection data of the target monitoring station can reflect the environmental detection data of the reference monitoring station, and the same applies to other monitoring stations. The purpose of analyzing the correlation is to obtain the data change trend of the reference monitoring station, rather than directly performing error analysis on the transformed data. Since the reference monitoring station is also affected by natural or biological activity factors, in addition to its own environmental detection data, the data of other monitoring stations are also used for transformation. The positive or negative sign of the correlation only affects the direction, and the closer the value is to 0, the weaker the correlation. Therefore, the absolute value of all correlations can be taken.

[0017] In step S13, the conversion of the environmental detection data of the target monitoring station and the environmental detection data of other monitoring stations into the first environmental detection data and the second environmental detection data of the reference monitoring station further includes the following steps:

[0018] S31. Obtain the historical environmental detection data of the target monitoring station and the reference monitoring station with the same collection time, divide the obtained data into a training set and a test set, use the historical environmental detection data of m1 target monitoring stations in the training set as the input, and the historical environmental detection data of m reference monitoring stations as the output to train the first neural network model of the reference monitoring station. Here, m1 is the model order and m is the output dimension. Verify the first neural network model through the test set. Obtain the m1 environmental detection data currently collected by the target monitoring station, input them into the first neural network model to obtain the first environmental detection data of the reference monitoring station. During the operation of the target monitoring station, keep executing step S31.

[0019] S32. Obtain the historical environmental detection data of the reference monitoring station and other monitoring stations except the reference monitoring station with the same collection time, divide the obtained data into a training set and a test set, use the historical environmental detection data of m2 other monitoring stations except the reference monitoring station in the training set as the input, and the historical environmental detection data of m reference monitoring stations as the output to train the second neural network model of the reference monitoring station. Here, m2 is the input feature and is an integer multiple of the model order. Verify the second neural network model through the test set. Obtain the m2 environmental detection data currently collected from other monitoring stations except the reference monitoring station, input them into the second neural network model to obtain the simulated value of the second environmental detection data of the reference monitoring station. Directly obtain the observed value of the m2 environmental detection data currently collected from the reference monitoring station, and perform weighted summation on the simulated value and the observed value of the second environmental detection data of the reference monitoring station to obtain the second environmental detection data of the reference monitoring station. During the operation of other monitoring stations, keep executing step S32.

[0020] There must be errors in the environmental detection data of the reference monitoring sites obtained through transformation. Therefore, instead of analyzing the errors, after eliminating the influence of the mean and variance, the trend after transformation is analyzed. If the trend obtained after the transformation of the target monitoring site is different from the trends obtained after the transformation of other monitoring sites, it indicates that the representativeness of the data of the target monitoring site is poor, that is, the target monitoring site has been interfered with.

[0021] Obtain the mean and variance of the first environmental detection data and the mean and variance of the second environmental detection data, and perform standardization processing on the first environmental detection data and the second environmental detection data. The standardization formula is as follows: Z = (z - az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization.

[0022] Fit the first environmental detection data and the second environmental detection data after standardization to obtain the first regression curve equation and the second regression curve equation, and calculate the root mean square error of the first regression curve equation and the second regression curve equation. If the root mean square error does not exceed the threshold, it is determined that there is no abnormal data at the target monitoring site. If the root mean square error exceeds the threshold, it is determined that there is abnormal data at the target monitoring site.

[0023] In step S13, the step of determining the starting point of the abnormal data of the target monitoring site further includes the following steps:

[0024] S100, obtain the time point t when the abnormal data is determined to exist at the target monitoring site; obtain the historical data when there is no abnormality from the target monitoring site, and calculate the mean mean1 and the standard deviation v1 based on the historical data when there is no abnormality; obtain the historical data when there is an abnormality from the target monitoring site, and calculate the mean mean2 and the standard deviation v2 based on the historical data when there is an abnormality; generate the normal interval [mean1 - k×v1, mean1 + k×v1] and the abnormal interval [mean2 - k×v2, mean2 + k×v2] according to mean1 and v1, mean2 and v2 respectively, where k is a coefficient.

[0025] S200. Starting from time point t, add the environmental detection data collected by the target monitoring site at time point t to set W. Add an environmental detection data collected by the target monitoring site before all time points in set W to set W. Calculate the average value M and standard deviation V of set W. Generate a detection interval [M - k×V, M + k×V] based on M and V. Calculate the coincidence degrees p1 and p2 between the detection interval and the normal interval and the abnormal interval. p1 = c1 / ct1, p1 = c2 / ct2, where c1 is the length of the interval where the detection interval overlaps with the normal interval, c2 is the length of the interval where the detection interval overlaps with the abnormal interval, and ct1 and ct1 are the lengths of the normal interval and the abnormal interval respectively. If p1 is greater than p2, obtain the second smallest value of the collection time corresponding to all environmental detection data in set W to get the starting point of the abnormal data. If p1 is not greater than p2, return to step S200.

[0026] In the case of few abnormal data, it is difficult to affect the overall data distribution. When the abnormal data is detected, it means that a relatively large number of abnormal data has been generated. Therefore, it is necessary to determine when the abnormal data started to be generated. Since the detection data of the target monitoring site comes from the target monitoring site itself, it is difficult to judge the authenticity and accuracy of the detection data. It is difficult to explain whether the target monitoring site has an abnormality based on the data distribution and numerical changes of the target monitoring site, because the data of the target monitoring site itself will also have fluctuations. And if it is determined that there is abnormal data in the target monitoring site, then judging the starting point of the abnormality based on the data distribution of the monitoring site will be more credible.

[0027] In step S13, the removing and filling of abnormal data further includes the following steps:

[0028] Obtain the historical data of the target monitoring site when there is no abnormality, fit the historical data of the target monitoring site when there is no abnormality to obtain a regression curve equation, generate a data baseline according to the regression curve equation, and fill the removed abnormal data with the data of the data baseline.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows: The environmental detection data is transformed to the same reference monitoring site for analysis, and the data distribution at the reference monitoring site is analyzed, which is less affected by the model accuracy, thereby improving the accuracy of the analysis result of the environmental detection data. In the case of determining that there is an abnormality, the data distribution of the target monitoring site itself is analyzed, making the analysis result more persuasive. The abnormal data is filled, improving the integrity of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic structural diagram of a data analysis system based on an environmental detection cloud platform of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0031] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0032] Embodiment: As Figure 1 shown, the present invention provides a technical solution, a data analysis system based on an environmental detection cloud platform, including an environmental detection module, an environmental detection cloud platform, and a data transmission module; the output end of the environmental detection module is connected to the input end of the environmental detection cloud platform, and is used to obtain environmental detection data of a monitoring site and send it to the environmental detection cloud platform; the environmental detection cloud platform is used to convert the environmental detection data collected by the target monitoring site and other monitoring sites into environmental detection data of a reference monitoring site, judge the abnormal data of the distribution of the environmental detection data obtained by converting the target monitoring site, and analyze the central tendency of the target monitoring site under the condition that there is abnormal data, determine the starting point of the abnormal data, and remove and fill the abnormal data; the data transmission module is used to connect the environmental detection module and the environmental detection cloud platform and support data exchange and sharing between the environmental detection module and the environmental detection cloud platform.

[0033] The environmental detection cloud platform further includes a data storage unit, a correlation analysis unit, a conversion unit, a standardization unit, a first analysis unit, a second analysis unit, and a filling unit; the correlation analysis unit is used for the correlation between the environmental detection data of all monitoring sites, and determines a reference monitoring site based on the correlation; the conversion unit is used to convert the environmental detection data collected by the target monitoring site and other monitoring sites into first environmental detection data and second environmental detection data of the environmental detection data of the reference monitoring site; the standardization unit is used to standardize the first environmental detection data and the second environmental detection data; the first analysis unit is used to analyze the data distribution of the first environmental detection data and the second environmental detection data; judge the abnormal data of the target monitoring site based on the data distribution; the second analysis unit, under the condition that it is determined that there is abnormal data in the target monitoring site, analyzes the central tendency according to the mean and variance of the environmental detection data of the target monitoring site, and determines the starting point of the abnormal data; the filling unit is used to remove the abnormal data and fill the removed abnormal data with simulated data.

[0034] The standardization unit performs standardization through the following formula: Z = (z - az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization.

[0035] The conversion unit inputs the environmental detection data collected by the target monitoring site into the first neural network model through the first neural network model to obtain the first environmental detection data of the reference monitoring site; obtains the currently collected environmental detection data from other monitoring stations except the reference monitoring site, inputs it into the second neural network model to obtain the simulated value of the second environmental detection data of the reference monitoring site; directly obtains the observed value of the currently collected environmental detection data from the reference monitoring site, and performs weighted summation on the simulated value and the observed value of the second environmental detection data of the reference monitoring site to obtain the second environmental detection data of the reference monitoring site.

[0036] Embodiment: The present invention provides a technical solution, a data analysis method based on an environmental detection cloud platform, including the following steps:

[0037] S11. The environmental detection module obtains the environmental detection data of the monitoring site, and the environmental detection module sends the security data to the environmental detection cloud platform. The monitoring sites are divided into target monitoring sites and other monitoring sites;

[0038] S12. The environmental detection cloud platform analyzes the correlation between the environmental detection data of all monitoring sites, and determines the reference monitoring site based on the correlation:

[0039] Obtain the historical environmental detection data of all monitoring sites, select consecutive n environmental detection data from the environmental detection data of two monitoring sites, and calculate the correlation T between the environmental detection data of the two monitoring sites ab , , where a i and b i are the observed values of the environmental detection data of the two monitoring sites, and are the averages of the observed values of the environmental detection data of the two monitoring sites, n is the sample size, and n is the sample number; take the absolute value of all correlations;

[0040] Add all other monitoring sites to the set U, and obtain the correlation T between the target monitoring site and the jth other monitoring site j , j is the monitoring site number, remove the jth other monitoring site from the set U, and calculate the average value TU of the correlations between the remaining monitoring sites in the set U and the jth other monitoring site j , add T j and TU j to obtain the estimated value of the jth other monitoring site; take the other monitoring site with the highest estimated value as the reference monitoring site.

[0041] In S13, the environmental detection cloud platform converts the environmental detection data of the target monitoring site and the environmental detection data of other monitoring sites into the first environmental detection data and the second environmental detection data of the reference monitoring site:

[0042] In S31, obtain the historical environmental detection data of the target monitoring site and the reference monitoring site with the same collection time, divide the obtained data into a training set and a test set, use the historical environmental detection data of m1 target monitoring sites in the training set as the input, and the historical environmental detection data of m reference monitoring sites as the output to train the first neural network model of the reference monitoring site; where m1 is the model order and m is the output dimension; verify the first neural network model through the test set; obtain the m1 environmental detection data currently collected by the target monitoring site, input them into the first neural network model to obtain the first environmental detection data of the reference monitoring site; during the operation of the target monitoring station, keep executing step S31;

[0043] In S32, obtain the historical environmental detection data of the reference monitoring site and other monitoring stations except the reference monitoring site with the same collection time, divide the obtained data into a training set and a test set, use the historical environmental detection data of m2 other monitoring stations except the reference monitoring site in the training set as the input, and the historical environmental detection data of m reference monitoring sites as the output to train the second neural network model of the reference monitoring site; where m2 is the input feature and is an integer multiple of the model order; verify the second neural network model through the test set; obtain the m2 environmental detection data currently collected from other monitoring stations except the reference monitoring site, input them into the second neural network model to obtain the simulated value of the second environmental detection data of the reference monitoring site; directly obtain the observed value of the m2 environmental detection data currently collected from the reference monitoring site, and perform weighted summation on the simulated value and the observed value of the second environmental detection data of the reference monitoring site to obtain the second environmental detection data of the reference monitoring site; during the operation of other monitoring stations, keep executing step S32.

[0044] For the first neural network model, optionally, the output dimension is set to 1. By detecting the effects of the first neural network model under different model orders, the optimal model order m1 is determined. For example, the environmental detection data is collected at the monitoring site once per minute. At 8:00 in the historical time, the data of the target monitoring site from 7:41, 7:42, …, 7:59 to 8:00 is used as the input, and the data of the reference monitoring site at 8:00 is used as the output to train the model, and the effect of the first neural network model with a model order of 20 can be obtained. The data of the target monitoring site from 7:40, 7:42, …, 7:59 to 8:00 is used as the input, and the data of the reference monitoring site at 8:00 is used as the output to train the model, and the effect of the first neural network model with a model order of 21 can be obtained. The model order with the best effect is obtained through testing; after obtaining the first neural network model, for example, when the current time is 9:00, the first environmental detection data at 9:00 is obtained through the data before 9:00. At 9:01, the input window is slid backward by one grid to obtain the first environmental detection data at 9:01. Repeating the execution can obtain all the first environmental detection data.

[0045] For the second neural network model, the data of multiple other monitoring sites can be selected as the input. Therefore, m2 is an integer multiple of the model order.

[0046] Analyze the data distributions of the first environmental detection data and the second environmental detection data; based on the data distributions, determine whether there are abnormal data at the target monitoring site:

[0047] Obtain the means and variances of the first environmental detection data and the means and variances of the second environmental detection data, and perform standardization processing on the first environmental detection data and the second environmental detection data. The standardization formula is as follows: Z = (z - az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization;

[0048] Fit the first environmental detection data and the second environmental detection data after standardization to obtain the first regression curve equation and the second regression curve equation, and calculate the root mean square errors of the first regression curve equation and the second regression curve equation. If the root mean square error does not exceed the threshold, it is determined that there is no abnormal data at the target monitoring site. If the root mean square error exceeds the threshold, it is determined that there is abnormal data at the target monitoring site.

[0049] The threshold can be determined according to the historical data of the target monitoring site, that is, in the historical data of the target monitoring site where abnormal data is clearly present, the root mean square errors of the first regression curve equation and the second regression curve equation under the condition of abnormal data can be obtained; the threshold is set according to the obtained root mean square errors.

[0050] If there is abnormal data, determine the starting point of the abnormal data of the target monitoring site:

[0051] S100, obtain the time point t when the target monitoring site is determined to have abnormal data; obtain the historical data when there is no abnormality from the target monitoring site, calculate the mean value mean1 and the standard deviation v1 based on the historical data when there is no abnormality; obtain the historical data when there is abnormality from the target monitoring site, calculate the mean value mean2 and the standard deviation v2 based on the historical data when there is abnormality; generate a normal interval [mean1 - k×v1, mean1 + k×v1] and an abnormal interval [mean2 - k×v2, mean2 + k×v2] according to mean1 and v1, mean2 and v2 respectively, where k is a coefficient;

[0052] S200, starting from the time point t, add the environmental detection data collected by the target monitoring site at the time point t to the set W, add an environmental detection data collected by the target monitoring site before all time points in the set W to the set W, calculate the average value M and the standard deviation V of the set W, generate a detection interval [M - k×V, M + k×V] according to M and V, calculate the coincidence degrees p1 and p2 between the detection interval and the normal interval and the abnormal interval, p1 = c1 / ct1, p1 = c2 / ct2, where c1 is the interval length of the repetition of the detection interval and the normal interval, c2 is the interval length of the repetition of the detection interval and the abnormal interval, and ct1 and ct1 are the interval lengths of the normal interval and the abnormal interval; if p1 is greater than p2, obtain the second smallest value of the collection time corresponding to all environmental detection data in the set W to get the starting point of the abnormal data; if p1 is not greater than p2, return to step S200.

[0053] When the current time is 9:00, calculate the average value and variance of the environmental detection data of the target monitoring site at 8:59 and 9:00 to obtain a detection interval. If the coincidence degree between the detection interval and the abnormal interval is higher, add the data at 8:58 to the set W and calculate again; if the coincidence degree between the detection interval and the normal interval is higher, the starting point of the abnormal data is 9:00, 8:59 is the minimum value, and 9:00 is the second smallest value.

[0054] Remove and fill the abnormal data, and mark the filled environmental detection data and the environmental detection data before the abnormality as safe data:

[0055] Obtain the historical data of the target monitoring site when there is no abnormality, fit the historical data of the target monitoring site when there is no abnormality to obtain a regression curve equation, generate a data baseline according to the regression curve equation, and fill the removed abnormal data with the data of the data baseline.

[0056] If there is no abnormal data, mark the environmental detection data as safe data.

[0057] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, in any regard, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Accordingly, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

Claims

1. A data analysis method based on an environmental detection cloud platform, characterized in that: The following steps are involved: S11, the environment detection module obtains the environment detection data of the monitoring site, and the environment detection module sends the security data to the environment detection cloud platform, and the monitoring site is divided into a target monitoring site and other monitoring sites; S12, the environmental monitoring cloud platform analyzes the correlation between the environmental monitoring data of all monitoring sites, and determines the reference monitoring site based on the correlation; S13, the environmental detection cloud platform converts the environmental detection data of the target monitoring site and the environmental detection data of other monitoring sites into the first environmental detection data and the second environmental detection data of the reference monitoring site, and analyzes the data distribution of the first environmental detection data and the second environmental detection data; based on the data distribution, it is determined whether there is abnormal data at the target monitoring site, and if there is abnormal data, the starting point of the abnormal data at the target monitoring site is determined, the abnormal data is removed and filled, and the filled environmental detection data and the environmental detection data before the abnormality occurs are marked as safe data; If there is no abnormal data, the environmental detection data is marked as safe data; The step of converting the environmental detection data of the target monitoring site and the environmental detection data of other monitoring sites into the first environmental detection data and the second environmental detection data of the reference monitoring site also includes the following steps: S31, obtaining historical environmental detection data of the target monitoring site and the reference monitoring site with the same collection time, dividing the obtained data into a training set and a test set, taking the historical environmental detection data of m1 target monitoring sites in the training set as input, and the historical environmental detection data of m reference monitoring sites as output, and training the first neural network model of the reference monitoring site; wherein m1 is the model order, and m is the output dimension; verifying the first neural network model through the test set; obtaining the m1 environmental detection data currently collected by the target monitoring site, inputting them into the first neural network model, and obtaining the first environmental detection data of the reference monitoring site; during the operation of the target monitoring site, keep executing step S31; S32, obtain historical environmental detection data of the reference monitoring site and other monitoring sites except the reference monitoring site with the same collection time, divide the obtained data into a training set and a test set, use the historical environmental detection data of m2 other monitoring sites except the reference monitoring site in the training set as input, and the historical environmental detection data of m reference monitoring sites as output, and train the second neural network model of the reference monitoring site; wherein m2 is the input feature, which is an integer multiple of the model order; verify the second neural network model through the test set; obtain the m2 environmental detection data currently collected from other monitoring sites except the reference monitoring site, input them into the second neural network model, and obtain the simulated value of the second environmental detection data of the reference monitoring site; directly obtain the observed values ​​of the m2 environmental detection data currently collected from the reference monitoring site, and perform weighted summation on the simulated value and observed value of the second environmental detection data of the reference monitoring site to obtain the second environmental detection data of the reference monitoring site; during the working process of other monitoring stations, keep executing step S32.

2. According to the data analysis method based on the environment detection cloud platform of claim 1, it is characterized in that: In step S12, analyzing the correlation between the environmental detection data of all monitoring sites and determining the reference monitoring site based on the correlation further includes the following steps: Obtain the historical environmental detection data of all monitoring sites, select n consecutive environmental detection data from the environmental detection data of two monitoring sites, and calculate the correlation T between the environmental detection data of the two monitoring sites. ab , , where a i and b i are the observed values ​​of the environmental monitoring data of the two monitoring stations, and is the average value of the environmental monitoring data observations of the two monitoring stations, n is the sample size, and n is the sample number; Add all other monitoring sites to the set U and obtain the correlation T between the target monitoring site and the jth other monitoring site. j , j is the monitoring site number, remove the jth other monitoring site from the set U, and calculate the average value TU of the correlation between the remaining monitoring sites in the set U and the jth other monitoring site j , T j and TU j Add up to get the estimated scores of the jth other monitoring station; take the other monitoring station with the highest estimated scores as the reference monitoring station.

3. A data analysis method based on an environmental detection cloud platform according to claim 2, characterized in that: In step S13, the analyzing the data distribution of the first environment detection data and the second environment detection data further includes the following steps: Obtain the mean and variance of the first environment detection data and the mean and variance of the second environment detection data, and perform standardization on the first environment detection data and the second environment detection data. The standardization formula is as follows: Z=(z-az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization; The standardized first environmental detection data and the second environmental detection data are fitted to obtain the first regression curve equation and the second regression curve equation, and the root mean square error between the first regression curve equation and the second regression curve equation is calculated. If the root mean square error does not exceed the threshold, it is judged that there is no abnormal data at the target monitoring site. If the root mean square error exceeds the threshold, it is judged that there is abnormal data at the target monitoring site.

4. A data analysis method based on an environmental detection cloud platform according to claim 3, characterized in that: In step S13, the step of determining the starting point of abnormal data of the target monitoring site further includes the following steps: S100, obtaining the time point t at which the target monitoring site is determined to have abnormal data; obtaining historical data when no abnormality occurs from the target monitoring site, and calculating the mean mean1 and standard deviation v1 based on the historical data when no abnormality occurs; obtaining historical data when abnormality occurs from the target monitoring site, and calculating the mean mean2 and standard deviation v2 based on the historical data when abnormality occurs; generating the normal interval [mean1-k×v1,mean1+k×v1] and the abnormal interval [mean2-k×v2,mean2+k×v2] based on mean1 and v1, mean2 and v2, respectively, where k is a coefficient; S200, starting from time point t, add the environmental detection data collected by the target monitoring station at time point t to the set W, add one environmental detection data collected by the target monitoring station in the set before all time points in W to the set W, calculate the mean value M and standard deviation V of the set W, generate the detection interval [Mk×V,M+k×V] according to M and V, calculate the overlap p1 and p2 between the detection interval and the normal interval and the abnormal interval, p1=c1 / ct1, p1=c2 / ct2, where c1 is the interval length of the detection interval repeated with the normal interval, c2 is the interval length of the detection interval repeated with the abnormal interval, ct1 and ct1 are the interval lengths of the normal interval and the abnormal interval; if p1 is greater than p2, obtain the next minimum value of the collection time corresponding to all environmental detection data in the set W to obtain the starting point of the abnormal data; if p1 is not greater than p2, return to step S200.

5. The data analysis method based on the environment detection cloud platform according to claim 3 is characterized in that: In step S13, the removal and filling of abnormal data further includes the following steps: The historical data of the target monitoring site when no abnormality occurs are obtained, and the historical data of the target monitoring site when no abnormality occurs are fitted to obtain a regression curve equation, a data baseline is generated according to the regression curve equation, and the removed abnormal data is filled with the data of the data baseline.

6. A data analysis system based on an environmental detection cloud platform, using a data analysis method based on an environmental detection cloud platform according to any one of claims 1 to 5, characterized in that: It includes an environment detection module, an environment detection cloud platform and a data transmission module; the output end of the environment detection module is connected to the input end of the environment detection cloud platform, and is used to obtain the environment detection data of the monitoring site and send it to the environment detection cloud platform; the environment detection cloud platform is used to convert the environment detection data collected by the target monitoring site and other monitoring sites into the environment detection data of the reference monitoring site, judge the abnormal data of the distribution of the environmental detection data converted by the target monitoring site, analyze the concentration trend of the target monitoring site under the condition of the existence of abnormal data, determine the starting point of the abnormal data, and remove and fill the abnormal data; The data transmission module is used to connect the environment detection module and the environment detection cloud platform, and supports data exchange and sharing between the environment detection module and the environment detection cloud platform.

7. A data analysis system based on an environmental detection cloud platform according to claim 6, characterized in that: The environmental detection cloud platform also includes a data storage unit, a correlation analysis unit, a conversion unit, a standardization unit, a first analysis unit, a second analysis unit and a filling unit; the correlation analysis unit is used for the correlation between the environmental detection data of all monitoring sites, and determines the reference monitoring site based on the correlation; the conversion unit is used to convert the environmental detection data collected by the target monitoring site and other monitoring sites into the first environmental detection data and the second environmental detection data of the environmental detection data of the reference monitoring site; the standardization unit is used to standardize the first environmental detection data and the second environmental detection data; The first analysis unit is used to analyze the data distribution of the first environment detection data and the second environment detection data; and judge the abnormal data of the target monitoring site based on the data distribution; The second analysis unit determines the starting point of the abnormal data based on the mean and variance analysis central trend of the environmental detection data of the target monitoring site under the condition that the target monitoring site has abnormal data; The filling unit is used to remove abnormal data and fill the removed abnormal data with simulated data.

8. The data analysis system based on the environment detection cloud platform according to claim 7 is characterized in that: The standardized unit is standardized by the following formula: Z=(z-az) / sz, where Z is the data after standardization, z is the data before standardization, az is the average value of the data before standardization, and sz is the standard deviation of the data before standardization.

9. The data analysis system based on the environment detection cloud platform according to claim 7 is characterized in that: The conversion unit inputs the environmental detection data collected by the target monitoring site into the first neural network model through the first neural network model to obtain the first environmental detection data of the reference monitoring site; obtains the currently collected environmental detection data from other monitoring stations except the reference monitoring site, and inputs it into the second neural network model to obtain the second environmental detection data simulation value of the reference monitoring site; directly obtains the currently collected environmental detection data observation value from the reference monitoring site, and performs weighted summation on the second environmental detection data simulation value and observation value of the reference monitoring site to obtain the second environmental detection data of the reference monitoring site.

Images