Search

CN-121978276-A - Multi-parameter atmospheric environment quality intelligent monitoring system based on Internet of things

CN121978276ACN 121978276 ACN121978276 ACN 121978276ACN-121978276-A

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent multi-parameter atmospheric environment quality monitoring system based on the Internet of things, which comprises a processor and a memory, wherein the processor executes a computer program of the memory to realize the following steps of acquiring target moments of uneven prediction error distribution among various pollutant parameters in the process of predicting concentration data of various pollutant parameters in future time by utilizing an LSTM model, constructing an adaptive weighted mean square error loss function at each target moment according to the difference of the prediction errors between various pollutant parameters and other pollutant parameters at each target moment and the correlation of the concentration data, predicting the concentration data of various pollutant parameters in future time according to the adaptive weighted mean square error loss function at each target moment, and improving the accuracy of monitoring the atmospheric environment quality by utilizing the LSTM model.

Inventors

  • WANG YUHUA
  • ZHANG YOU

Assignees

  • 山东碳排放信息技术有限公司

Dates

Publication Date
20260505
Application Date
20260126

Claims (7)

  1. 1. The multi-parameter atmospheric environment quality intelligent monitoring system based on the Internet of things comprises a memory, a processor and a computer program stored in the memory and running on the processor, and is characterized in that the following steps are realized when the processor executes the computer program: acquiring concentration data of each pollutant parameter at each time in a preset time interval up to the current time in a target area, and acquiring target time when the prediction error among each pollutant parameter is unevenly distributed according to the distribution confusion degree of the prediction error among each pollutant parameter at each time in the preset time interval in the process of predicting the concentration data of each pollutant parameter at the future time by utilizing an LSTM model; Aiming at any target moment, according to the difference of prediction errors between various pollutant parameters and other pollutant parameters at any target moment and the association degree of concentration data between various pollutant parameters and other pollutant parameters in a local time range of the any target moment, acquiring the self-adaptive parameter weight of various pollutant parameters at any target moment; According to the self-adaptive parameter weight of each pollutant parameter at any target time, constructing a self-adaptive weighted mean square error loss function at any target time, according to the self-adaptive weighted mean square error loss function at each target time and the mean square error loss function at each other time, acquiring the concentration predicted value of each pollutant parameter at the future time in the fully-connected layer of the LSTM model, and according to the concentration predicted value of each pollutant parameter at the future time, intelligently monitoring the atmospheric environment quality of the target area at the future time.
  2. 2. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 1, wherein the obtaining the target time of uneven prediction error distribution among the various pollutant parameters according to the degree of confusion of the distribution of the prediction error among the various pollutant parameters at each time in the preset time period comprises: Calculating the average value and the median of the prediction errors of all the pollutant parameters at any moment in a preset period, and taking the absolute value of the difference between the average value and the median as an independent variable of a hyperbolic tangent function to obtain a first distribution discrete degree of the prediction errors among all the pollutant parameters at any moment; Calculating the average difference of the prediction errors of the pollutant parameters at any moment, and taking the accumulated sum of all the average differences as the independent variable of the hyperbolic tangent function to obtain the second distribution discrete degree of the prediction errors among the pollutant parameters at any moment; Calculating the average value between the first distribution discrete degree and the second distribution discrete degree to obtain the distribution confusion degree of the prediction error between each pollutant parameter at any moment, and if the distribution confusion degree is larger than a preset distribution confusion degree threshold value, marking any moment as a target moment.
  3. 3. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 1, wherein the obtaining the adaptive parameter weights of the various pollutant parameters at the arbitrary target time according to the difference of prediction errors between the various pollutant parameters and other pollutant parameters at the arbitrary target time and the correlation degree of concentration data between the various pollutant parameters and other pollutant parameters in the local time range at the arbitrary target time comprises: for any pollutant parameter, acquiring a first weight characteristic value of the any pollutant parameter at any target time according to the difference of prediction errors between the any pollutant parameter and other pollutant parameters at any target time; acquiring a second weight characteristic value of any pollutant parameter at any target time according to the association degree of concentration data between any pollutant parameter and other pollutant parameters in the local time range of any target time; Calculating the sum of the first weight characteristic value and the second weight characteristic value to obtain the comprehensive weight of the pollutant parameter of any item at any target time, obtaining the comprehensive weight of each pollutant parameter of any item at any target time, and calculating the duty ratio of the comprehensive weight of the pollutant parameter of any item at any target time in the comprehensive weight of all pollutant parameters to obtain the self-adaptive parameter weight of the pollutant parameter of any item at any target time.
  4. 4. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 3, wherein the obtaining the first weight characteristic value of the any one contaminant parameter at the any one target time according to the difference of the prediction error between the any one contaminant parameter and other contaminant parameters at the any one target time comprises: Calculating the average value of the prediction errors of all other pollutant parameters at any target time, calculating the absolute value of the difference value between the prediction error of any pollutant parameter at any target time and the average value to obtain the deviation degree of the prediction error between any pollutant parameter at any target time and other pollutant parameters, and taking the opposite number of the deviation degree of the prediction error as an independent variable of an exponential function taking a natural constant as a base to obtain the degree of consistency of the error level between any pollutant parameter at any target time and other pollutant parameters; Acquiring the stability degree of the prediction error of any one pollutant parameter in the local time range of any one target moment according to the fluctuation level of the prediction error of any one pollutant parameter in the local time range of any one target moment; And calculating the sum of the error level consistency degree and the stability degree to obtain a first weight characteristic value of the pollutant parameter of any one of the target time points.
  5. 5. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 4, wherein the obtaining the smoothness of the prediction error of the any one of the contaminant parameters in the local time range of the any one of the target time instants according to the fluctuation level of the prediction error of the any one of the contaminant parameters in the local time range of the any one of the target time instants comprises: Obtaining a prediction error of any pollutant parameter at each moment in a first preset local time range before any target moment, forming a prediction error sequence by the prediction error of any pollutant parameter at each moment in the first preset local time range before any target moment and the prediction error at any target moment, obtaining a first-order differential sequence of the prediction error sequence, calculating a product between any two adjacent data in the first-order differential sequence, if the product is smaller than 0, marking a sign exclusive-OR value between the any two adjacent data as 1, and if the product is larger than or equal to 0, marking the sign exclusive-OR value between the any two adjacent data as 0; Obtaining symbol exclusive-or values between every two adjacent data in the first-order differential sequence, calculating the accumulated sum of all symbol exclusive-or values to obtain the fluctuation degree of the prediction error of any pollutant parameter in the local time range of any target moment, taking the opposite number of the fluctuation degree as the independent variable of an exponential function taking a natural constant as a base number to obtain the stability degree of the prediction error of any pollutant parameter in the local time range of any target moment.
  6. 6. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 3, wherein the obtaining the second weight characteristic value of the any one of the contaminant parameters at the any one of the target moments according to the association degree of the concentration data between the any one of the contaminant parameters and other contaminant parameters in the local time range of the any one of the target moments comprises: For any other pollutant parameter, respectively acquiring concentration data of the any pollutant parameter and the any other pollutant parameter at each moment in a second preset local time range up to the any target moment, respectively acquiring a concentration data sequence corresponding to the any pollutant parameter and the any other pollutant parameter, calculating a pearson correlation coefficient between the concentration data sequence corresponding to the any pollutant parameter and the concentration data sequence corresponding to the any other pollutant parameter, and marking the average value between the pearson correlation coefficient and a constant 1 as the correlation degree between the any pollutant parameter and the any other pollutant parameter; Acquiring the association degree between any one pollutant parameter and each other pollutant parameter, and calculating the average value of all the association degrees to obtain the overall association degree between any one pollutant parameter and the other pollutant parameters in the local time range of any target moment; calculating the range of all the association degrees, and subtracting the range from a constant 1 to obtain the association stability degree between any one pollutant parameter and other pollutant parameters in the local time range of any target moment; and calculating the sum of the overall association degree and the association stability degree to obtain a second weight characteristic value of the pollutant parameter of any item at any target time.
  7. 7. The internet of things-based multi-parameter atmospheric environmental quality intelligent monitoring system according to claim 1, wherein the intelligent monitoring of the atmospheric environmental quality of the target area at the future time according to the concentration predicted value of each pollutant parameter at the future time comprises: According to the concentration predicted value of each pollutant parameter at the future time, acquiring an air quality index of each pollutant parameter at the future time, acquiring an air quality grade of the target area at the future time according to the air quality index, and according to the air quality grade, realizing intelligent monitoring of the atmospheric environment quality of the target area at the future time.

Description

Multi-parameter atmospheric environment quality intelligent monitoring system based on Internet of things Technical Field The invention relates to the technical field of data processing, in particular to an intelligent multi-parameter atmospheric environment quality monitoring system based on the Internet of things. Background With the acceleration of the urban process and the increase of traffic and industrial emission, the atmospheric pollution has an important influence on public health, ecological environment and socioeconomic development, various pollutants in the atmosphere and relevant environmental parameters are currently obtained continuously in real time, the atmospheric environment quality is scientifically evaluated and trend predicted through an LSTM prediction model, the prediction result is mapped into the atmospheric quality index (IAQI) of each pollutant, the primary pollutant is determined according to the maximum value of each index and the integral Atmospheric Quality Index (AQI) is formed, and the corresponding atmospheric quality grade is finally obtained, so that the government and the management department can be helped to master the environmental change rule, the health protection reference is provided for the public, and daily travel and life decisions are guided. The LSTM prediction model takes the prediction error of each pollutant parameter as an optimization target of model back propagation training, and constructs a loss function by using the mean square error. However, since the atmospheric environmental monitoring points are often directly exposed to the complex urban activity environment, the ambient environmental conditions thereof may change significantly in a short time. For example, in a scene of urban road line monitoring, vehicles intensively pass and frequently start and stop in the early and late traffic peak period, the emission intensity of local area tail gas can be obviously increased in a short time, actions such as temporary construction, material stacking and dust raising operation and the like can also influence local atmospheric components in a certain time window around a residential area or a construction area, namely, the response modes and response intensities of different types of pollutants to environmental changes can be obviously different, part of pollutant parameters (such as NO 2、PM10、PM2.5 and the like) directly influenced by local emission can be rapidly increased or obviously fluctuated in a plurality of sampling periods, and pollutant parameters (such as O 3、SO2 and the like) greatly influenced by photochemical reaction, regional transportation or meteorological conditions can be relatively small in change amplitude and relatively stable change trend can be presented in the same time period. In the LSTM prediction model, the prediction error corresponding to the abrupt or severely fluctuating contaminant parameter is significantly larger in value than other contaminant parameters, and the difference of the prediction error between different contaminant parameters can cause the influence degree of the corresponding gradient on the update of the model parameter to be inconsistent in the process of model back propagation, so that the contaminant parameter with larger prediction error has larger influence in the process of model weight and bias update. If the loss function is still constructed by using the mean square error, the pollutant parameter with larger prediction error easily occupies larger proportion in the loss function, so that the model pays more attention to the short-time and local change characteristics, the learning ability of the overall evolution rule of the atmospheric environment quality is further weakened, and finally, the stability of the prediction result is reduced or the overall prediction accuracy is reduced. Therefore, how to adaptively construct a loss function according to the data change characteristics of each pollutant parameter, so as to improve the accuracy of monitoring the atmospheric environment quality by using the LSTM model becomes a problem to be solved. Disclosure of Invention In view of the above, the embodiment of the invention provides an intelligent multi-parameter atmospheric environment quality monitoring system based on the Internet of things, which aims to solve the problem of how to adaptively construct a loss function according to the data change characteristics of various pollutant parameters, and further improve the accuracy of monitoring the atmospheric environment quality by using an LSTM model. The embodiment of the invention provides an intelligent multi-parameter atmospheric environment quality monitoring system based on the Internet of things, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and is characterized in that the processor realizes the following steps when executing the computer program: acquiring concentration