Environmental quality assessment method and system based on big data

Abstract

The present invention discloses an environmental quality assessment method and system based on big data, and the method includes data collection, data preprocessing, establishing an environmental quality assessment model and environmental quality assessment. The present invention belongs to the field of environmental quality assessment, and specifically refers to an environmental quality assessment method and system based on big data. The scheme introduces a meteorological factor vector, dynamically integrates the influence of environmental variables on pollutant concentrations, and thus performs data correction; based on the dynamic adjustment of distortion and dynamic weight, a dynamic deviation is obtained, thereby improving the accuracy of subsequent environmental quality assessments; by introducing similar deviation values ​​and dynamic factors, the dynamic changes of environmental quality assessment data can be adaptively captured, and the response to environmental data is more accurate; based on the monotonic loss and differential loss design of the introduction of similar deviation values, the monotonic law of pollutants is explicitly utilized, and the assessment ability of complex environmental quality assessment data is improved, thereby improving the accuracy of the final environmental quality assessment results.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental quality assessment, and in particular to an environmental quality assessment method and system based on big data. Background Art

[0002] Environmental quality assessment methods refer to a system of methods that use scientific technical means and models to quantitatively or qualitatively evaluate environmental quality. This method analyzes the state of various physical, chemical and biological factors in the environment, combines relevant standards and laws, determines whether the environment meets the needs of human activities, ecosystems and biological health, and warns of possible environmental problems or pollution risks. However, general environmental quality assessment methods ignore the impact of meteorological conditions and dynamic changes, resulting in poor quality assessment results in actual environments, and ignore the impact of time offsets and amplitude changes, which leads to low accuracy of environmental quality assessments; general environmental quality assessment methods are unable to dynamically combine environmental data, and have a weak ability to capture the interaction between pollutant concentrations and meteorological factors, which leads to unstable environmental quality assessment results and low accuracy. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an environmental quality assessment method and system based on big data. In view of the fact that the general environmental quality assessment method ignores the influence of meteorological conditions and dynamic changes, resulting in poor quality assessment effect in the actual environment, and ignores the influence of time offset and amplitude change, thereby resulting in low accuracy of environmental quality assessment, this scheme introduces meteorological factor vectors, dynamically integrates the influence of environmental variables on pollutant concentration, thereby correcting data and improving the subsequent environmental quality assessment effect; based on the dynamic adjustment of distortion and dynamic weight, normalization processing is performed to obtain the dynamic deviation of the collected environmental quality data, thereby improving the accuracy of subsequent environmental quality assessment; in view of the fact that the general environmental quality assessment method cannot dynamically combine environmental data, and has a weak ability to capture the interaction between pollutant concentration and meteorological factors, thereby resulting in unstable environmental quality assessment results and low accuracy, this scheme can adaptively capture the dynamic changes of environmental quality assessment data by introducing similar deviation values ​​and dynamic factors, and respond more accurately to short-term anomalies and long-term trend changes in environmental data; based on the monotonicity loss and difference loss design of the introduction of similar deviation values, the monotonicity law of pollutants is clearly utilized to improve the assessment ability of complex environmental quality assessment data, thereby improving the accuracy of the final environmental quality assessment results.

[0004] The technical solution adopted by the present invention is as follows: The environmental quality assessment method based on big data provided by the present invention comprises the following steps:

[0005] Step S1: data collection;

[0006] Step S2: data preprocessing;

[0007] Step S3: Establishing an environmental quality assessment model;

[0008] Step S4: Environmental quality assessment.

[0009] Furthermore, in step S1, the data collection is to collect historical environmental quality assessment data; the historical environmental quality assessment data includes environmental pollutant data, meteorological factor data, time, geographic information data and environmental quality assessment level; the environmental quality assessment level is used as a data label; and the collected data is cleaned, converted and standardized; and environmental quality time series data is obtained.

[0010] Furthermore, in step S2, the data preprocessing specifically includes the following steps:

[0011] Step S21: Data correction; calculate the pollutant concentration, add the environmental quality time series data, and the pollutant concentration diffusion is expressed as: ; Where c(t) is the pollutant concentration at time t; c0 is the initial pollutant concentration; is the attenuation coefficient; N is the number of pollution sources; q i1 is the emission intensity of the i1th pollution source; D is the diffusion coefficient; x i1 and x0 are the location of the pollution source and the location of the monitoring point, respectively; It is the vector representation of meteorological factors; is the L2 norm;

[0012] Step S22: Calculate the environmental quality distortion coefficient; the distortion coefficient reflects the change of environmental quality time series data caused by time offset during the alignment process. The environmental quality distortion coefficient of each data point is expressed as: ; ; Calculate the overall environmental quality distortion measure, expressed as: ; In the formula, is the distortion coefficient of the data at the kth time point; k1, k2 and k3 represent the complete alignment between data points, the existence of time offset and the mutation of data respectively; α and β are the adjustment parameters that control the change of distortion coefficient; is the distortion value of the entire data set; m and n are the lengths of the standard environmental quality time series data and the collected environmental quality time series data, respectively; α1 is the distortion parameter; γ is the dynamic adjustment factor; and are the offset and amplitude changes of the kth data point from the standard time series in the time dimension; T is the total time span of the time series; C is the range of data amplitude changes; is the local fluctuation intensity; , and is the weight coefficient;

[0013] Step S23: Amplitude normalization; distance measurement and distortion planning processing, expressed as: ; Based on the difference in the amplitude change and time alignment of the environmental quality data, the similarity deviation value reflecting the overall similarity of the environmental quality is obtained, which is expressed as: ; ;in, and They are the normalized dynamic time warping distance and the distortion measure of the dataset, respectively; is the dynamic time warping distance before normalization; and They are the minimum dynamic time warping distance and the maximum dynamic time warping distance respectively; and are the maximum distortion value and the minimum distortion value respectively; is the similarity deviation value of the data set; is the dynamic bias weight.

[0014] Furthermore, in step S3, the established environmental quality assessment model includes a residual network layer for extracting the characteristics of the pollutant concentration and meteorological factor variable time series; a time series feature extraction layer for capturing the short-term dynamics and long-term trends of environmental quality time series data using a bidirectional GRU; and a fully connected layer for generating environmental quality assessment results; specifically, the following steps are included:

[0015] Step S31: residual network layer design; the residual network layer includes a multi-branch residual network and a layer-by-layer residual network; the multi-branch residual network extracts local features of the environmental quality time series data through a 1-dimensional convolution operation, which is expressed as: ; Each branch uses convolution kernels of different sizes to capture the characteristics of different time scales in the environmental time quality series data; After each branch in the multi-branch residual network, a channel attention mechanism is introduced, which is expressed as: ; ; ; ; The layer-by-layer residual network adds a residual connection after each convolution block, expressed as: ;in, and are the feature outputs of the jth channel of the lth layer and the l-1th layer respectively; is the convolution kernel weight; M j is the input channel set, i2 is the channel index; is the bias term; is a dynamic factor; f(·) is an activation function, using the ReLU activation function; is the feature mean of channel c; n is the length of the time series; is the characteristic value of channel c at time point i; and is the factor weight parameter; mean(·) and std(·) are the feature mean and standard deviation respectively; is the channel attention weight; is the adjusted eigenvalue; y is the output feature after residual connection; It is the feature extracted by 1D convolution;

[0016] Step S32: loss function design; construct monotonic loss, the pollutant concentration has a monotonic characteristic, and the pollutant concentration gradually decreases when there is no new pollution source. It is expressed as: ; Construct difference loss , expressed as: ; Construct basic penalty loss , expressed as: ; The final loss L is obtained, expressed as: ; Where M is the total number of data points; and are the predicted values ​​at the i-th time point and the i-1-th time point respectively; is the rate of change of the predicted value; y i and j2 are the true values ​​at the ith time point and the j2th time point respectively; is the predicted value at the j2th time point; is the weight of the regularization term; is the weight of the noise suppression term; is the variance of the predicted value; and is the loss weight; W is the weight set;

[0017] Step S33: model determination; the data set processed in step S2 is divided into a test set and a training set in advance. When the environmental quality assessment model converges to the training set loss or reaches the maximum number of training times, the environmental quality assessment model training is completed; an accuracy threshold is set in advance. When the prediction accuracy of the trained environmental quality assessment model for the test set is higher than the accuracy threshold, the environmental quality assessment model is established. Otherwise, the model is retrained based on the particle swarm search algorithm to optimize the model parameters.

[0018] Furthermore, in step S4, the environmental quality assessment is based on the established environmental quality assessment model, real-time collection of environmental quality data to be assessed, and the environmental quality assessment grade output by the environmental quality assessment model is used as the environmental quality assessment result of the environmental quality data to be assessed.

[0019] The environmental quality assessment system based on big data provided by the present invention comprises a data acquisition module, a data preprocessing module, an environmental quality assessment model establishment module and an environmental quality assessment module;

[0020] The data acquisition module collects historical environmental quality assessment data and sends the data to the data preprocessing module;

[0021] The data preprocessing module corrects the dynamic changes and deviations of environmental data through dynamic factors, distortion measurement and normalization processing; and sends the data to the environmental quality assessment model establishment module;

[0022] The environmental quality assessment model building module uses layer-by-layer residual networks, dynamic weights and bidirectional GRU to extract time series characteristics of pollutants and meteorological factors to generate environmental quality assessment results; and sends the data to the environmental quality assessment module;

[0023] The environmental quality assessment module implements environmental quality assessment on the historical environmental quality assessment data collected in real time based on the established environmental quality assessment model.

[0024] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0025] (1) In view of the problems that general environmental quality assessment methods ignore the influence of meteorological conditions and dynamic changes, resulting in poor quality assessment results in actual environments, and ignore the influence of time offset and amplitude changes, which leads to low accuracy of environmental quality assessment, this scheme introduces meteorological factor vectors and dynamically integrates the influence of environmental variables on pollutant concentrations to correct data and improve the subsequent environmental quality assessment results; based on the dynamic adjustment of distortion and dynamic weights, normalization processing is performed to obtain the dynamic deviation of the collected environmental quality data, thereby improving the accuracy of subsequent environmental quality assessments.

[0026] (2) In view of the fact that general environmental quality assessment methods are unable to dynamically combine environmental data and have a weak ability to capture the interaction between pollutant concentrations and meteorological factors, which in turn leads to unstable environmental quality assessment results and low accuracy, this scheme introduces similar deviation values ​​and dynamic factors to adaptively capture the dynamic changes of environmental quality assessment data and respond more accurately to short-term anomalies and long-term trend changes in environmental data. Based on the monotonic loss and differential loss design of the similar deviation values, the monotonicity law of pollutants is explicitly utilized to improve the assessment ability of complex environmental quality assessment data, thereby improving the accuracy of the final environmental quality assessment results. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A schematic diagram of the process of the environmental quality assessment method based on big data provided by the present invention;

[0028] Figure 2 A schematic diagram of an environmental quality assessment system based on big data provided by the present invention;

[0029] Figure 3 It is a schematic diagram of the process of step S3.

[0030] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0032] In the description of the present invention, it is necessary to understand that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.

[0033] Example 1, see Figure 1 The present invention provides an environmental quality assessment method based on big data, which comprises the following steps:

[0034] Step S1: Data collection: Collect historical environmental quality assessment data;

[0035] Step S2: Data preprocessing: Correcting the dynamic changes and deviations of environmental data through dynamic factors, distortion measurement and normalization processing;

[0036] Step S3: Establish an environmental quality assessment model; use layer-by-layer residual networks, dynamic weights, and bidirectional GRU to extract time series characteristics of pollutants and meteorological factors to generate environmental quality assessment results;

[0037] Step S4: Environmental quality assessment: Implement environmental quality assessment on the historical environmental quality assessment data collected in real time based on the established environmental quality assessment model.

[0038] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, the historical environmental quality assessment data includes environmental pollutant data, meteorological factor data, time, geographic information data and environmental quality assessment level; the environmental quality assessment level is used as a data label; and the collected data is cleaned, converted and standardized to obtain environmental quality time series data; the meteorological factor data includes temperature, humidity, wind speed, wind direction and atmospheric pressure; the geographic information data includes the location of the pollution source, the location of the monitoring point and the terrain; the environmental quality assessment level includes normal environment, slight abnormality, moderate abnormality and severe abnormality.

[0039] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, data preprocessing specifically includes the following steps:

[0040] Step S21: Data correction; calculate the pollutant concentration, add the environmental quality time series data, and the pollutant concentration diffusion is expressed as: ; Where c(t) is the pollutant concentration at time t; c0 is the initial pollutant concentration; is the attenuation coefficient; N is the number of pollution sources; q i1 is the emission intensity of the i1th pollution source; D is the diffusion coefficient; x i1 and x0 are the location of the pollution source and the location of the monitoring point, respectively; It is the vector representation of meteorological factors; is the L2 norm;

[0041] Step S22: Calculate the environmental quality distortion coefficient; the distortion coefficient reflects the change of environmental quality time series data caused by time offset during the alignment process. The environmental quality distortion coefficient of each data point is expressed as: ; ; Calculate the overall environmental quality distortion measure, expressed as: ; In the formula, is the distortion coefficient of the data at the kth time point; k1, k2 and k3 represent the complete alignment between data points, the existence of time offset and the mutation of data respectively; α and β are the adjustment parameters that control the change of distortion coefficient; is the distortion value of the entire data set; m and n are the lengths of the standard environmental quality time series data and the collected environmental quality time series data, respectively; α1 is the distortion parameter; γ is the dynamic adjustment factor; and are the offset and amplitude changes of the kth data point from the standard time series in the time dimension; T is the total time span of the time series; C is the range of data amplitude changes; is the local fluctuation intensity; , and is the weight coefficient;

[0042] Step S23: Amplitude normalization; distance measurement and distortion planning processing, expressed as: ; Based on the difference in the amplitude change and time alignment of the environmental quality data, the similarity deviation value reflecting the overall similarity of the environmental quality is obtained, which is expressed as: ; ;in, and They are the normalized dynamic time warping distance and the distortion measure of the dataset, respectively; is the dynamic time warping distance before normalization; and They are the minimum dynamic time warping distance and the maximum dynamic time warping distance respectively; and are the maximum distortion value and the minimum distortion value respectively; is the similarity deviation value of the data set; is the dynamic bias weight.

[0043] By performing the above operations, this scheme introduces meteorological factor vectors to dynamically integrate the influence of environmental variables on pollutant concentrations, thereby performing data correction and improving the subsequent environmental quality assessment effect. Normalization processing is performed based on dynamic adjustment of distortion and dynamic weights to obtain the dynamic deviation of the collected environmental quality data, thereby improving the accuracy of subsequent environmental quality assessment.

[0044] Example 4, see Figure 1 and Figure 3Based on the above embodiment, in step S3, the established environmental quality assessment model includes a residual network layer for extracting the characteristics of the pollutant concentration and meteorological factor variable time series; a time series feature extraction layer for capturing the short-term dynamics and long-term trends of environmental quality time series data using a bidirectional GRU; and a fully connected layer for generating environmental quality assessment results. Specifically, the following steps are included:

[0045] Step S31: residual network layer design; the residual network layer includes a multi-branch residual network and a layer-by-layer residual network; the multi-branch residual network extracts local features of the environmental quality time series data through a 1-dimensional convolution operation, which is expressed as: ; Each branch uses convolution kernels of different sizes to capture the characteristics of different time scales in the environmental time quality series data; After each branch in the multi-branch residual network, a channel attention mechanism is introduced, which is expressed as: ; ; ; ; The layer-by-layer residual network adds a residual connection after each convolution block, expressed as: ;in, and are the feature outputs of the jth channel of the lth layer and the l-1th layer respectively; is the convolution kernel weight; M j is the input channel set, i2 is the channel index; is the bias term; is a dynamic factor; f(·) is an activation function, using the ReLU activation function; is the feature mean of channel c; n is the length of the time series; is the characteristic value of channel c at time point i; and is the factor weight parameter; mean(·) and std(·) are the feature mean and standard deviation respectively; is the channel attention weight; is the adjusted eigenvalue; y is the output feature after residual connection; It is the feature extracted by 1D convolution;

[0046] Step S32: loss function design; construct monotonic loss, the pollutant concentration has a monotonic characteristic, and the pollutant concentration gradually decreases when there is no new pollution source. It is expressed as: ; Construct difference loss , expressed as: ; Construct basic penalty loss , expressed as: ; The final loss L is obtained, expressed as: ; Where M is the total number of data points; and are the predicted values ​​at the i-th time point and the i-1-th time point respectively; is the rate of change of the predicted value; y i and j2 are the true values ​​at the ith time point and the j2th time point respectively; is the predicted value at the j2th time point; is the weight of the regularization term; is the weight of the noise suppression term; is the variance of the predicted value; and is the loss weight; W is the weight set;

[0047] Step S33: model determination; the data set processed in step S2 is divided into a test set and a training set in advance. When the environmental quality assessment model converges to the training set loss or reaches the maximum number of training times, the environmental quality assessment model training is completed; an accuracy threshold is set in advance. When the prediction accuracy of the trained environmental quality assessment model for the test set is higher than the accuracy threshold, the environmental quality assessment model is established. Otherwise, the model is retrained based on the particle swarm search algorithm to optimize the model parameters.

[0048] By performing the above operations, the general environmental quality assessment method has the problem of being unable to dynamically combine environmental data and having a weak ability to capture the interaction between pollutant concentrations and meteorological factors, which leads to unstable environmental quality assessment results and low accuracy. This scheme introduces similar deviation values ​​and dynamic factors to adaptively capture the dynamic changes of environmental quality assessment data and respond more accurately to short-term anomalies and long-term trend changes in environmental data. Based on the monotonic loss and differential loss design that introduce similar deviation values, the monotonicity law of pollutants is explicitly utilized to improve the assessment ability of complex environmental quality assessment data, thereby improving the accuracy of the final environmental quality assessment results.

[0049] Example 5, see Figure 1 This embodiment is based on the above embodiment. In step S4, the environmental quality assessment is based on the established environmental quality assessment model, and the environmental quality data to be assessed is collected in real time. The environmental quality assessment level output by the environmental quality assessment model is used as the environmental quality assessment result of the environmental quality data to be assessed; when the environmental quality assessment level is moderately abnormal or severely abnormal, early warning processing is performed.

[0050] Example 6, see Figure 2 , this embodiment is based on the above embodiment, the environmental quality assessment system based on big data provided by the present invention includes a data acquisition module, a data preprocessing module, an environmental quality assessment model establishment module and an environmental quality assessment module;

[0051] The data acquisition module collects historical environmental quality assessment data and sends the data to the data preprocessing module;

[0052] The data preprocessing module corrects the dynamic changes and deviations of environmental data through dynamic factors, distortion measurement and normalization processing; and sends the data to the environmental quality assessment model establishment module;

[0053] The environmental quality assessment model building module uses layer-by-layer residual networks, dynamic weights and bidirectional GRU to extract time series characteristics of pollutants and meteorological factors to generate environmental quality assessment results; and sends the data to the environmental quality assessment module;

[0054] The environmental quality assessment module implements environmental quality assessment on the historical environmental quality assessment data collected in real time based on the established environmental quality assessment model.

[0055] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0056] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0057] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. The environmental quality assessment method based on big data is characterized by: The method comprises the following steps: Step S1: Data collection: Collect historical environmental quality assessment data; Step S2: data preprocessing; Step S3: Establishing an environmental quality assessment model; Step S4: Environmental quality assessment: Based on the established environmental quality assessment model, environmental quality assessment is performed on the historical environmental quality assessment data collected in real time; In step S2, the data preprocessing includes the following steps: Step S21: Data correction; calculate the pollutant concentration, add the environmental quality time series data, and the pollutant concentration diffusion is expressed as: ; Where c(t) is the pollutant concentration at time t; c0 is the initial pollutant concentration; is the attenuation coefficient; N is the number of pollution sources; q i1 is the emission intensity of the i1th pollution source; D is the diffusion coefficient; x i1 and x0 are the location of the pollution source and the location of the monitoring point, respectively; It is the vector representation of meteorological factors; is the L2 norm; Step S22: Calculate the environmental quality distortion coefficient; the distortion coefficient reflects the change of environmental quality time series data caused by time offset during the alignment process. The environmental quality distortion coefficient of each data point is expressed as: ; ; For any two sets of environmental quality time series data, calculate the overall environmental quality distortion measure, expressed as: ; In the formula, is the distortion coefficient of the data at the kth time point; k1, k2 and k3 represent the complete alignment between data points, the existence of time offset and the mutation of data respectively; α and β are the adjustment parameters that control the change of distortion coefficient; is the distortion value of the entire data set; m and n are the lengths of the standard environmental quality time series data and the collected environmental quality time series data, respectively; α1 is the distortion parameter; γ is the dynamic adjustment factor; and are the offset and amplitude changes of the kth data point from the standard time series in the time dimension; T is the total time span of the time series; C is the range of data amplitude changes; is the local fluctuation intensity; , and is the weight coefficient; Step S23: Amplitude normalization; distance measurement and distortion planning processing, expressed as: ; Based on the difference in the amplitude change and time alignment of the environmental quality data, the similarity deviation value reflecting the overall similarity of the environmental quality is obtained, which is expressed as: ; ;in, and They are the normalized dynamic time warping distance and the distortion measure of the dataset, respectively; is the dynamic time warping distance before normalization; and They are the minimum dynamic time warping distance and the maximum dynamic time warping distance respectively; and are the maximum distortion value and the minimum distortion value respectively; is the similarity deviation value of the data set; is the dynamic bias weight; The established environmental quality assessment model includes a residual network layer, which is used to extract the characteristics of the time series of pollutant concentrations and meteorological factor variables; a time series feature extraction layer, which uses a bidirectional GRU to capture the short-term dynamics and long-term trends of environmental quality time series data; and a fully connected layer, which is used to generate environmental quality assessment results. Specifically, it includes the following steps: Step S31: residual network layer design; the residual network layer includes a multi-branch residual network and a layer-by-layer residual network; the multi-branch residual network extracts local features of the environmental quality time series data through a 1-dimensional convolution operation, which is expressed as: ; Each branch uses convolution kernels of different sizes to capture the characteristics of different time scales in the environmental time quality series data; After each branch in the multi-branch residual network, a channel attention mechanism is introduced, which is expressed as: ; ; ; ; The layer-by-layer residual network adds a residual connection after each convolution block, expressed as: ;in, and are the feature outputs of the jth channel of the lth layer and the l-1th layer respectively; is the convolution kernel weight; M j is the input channel set, i2 is the channel index; is the bias term; is a dynamic factor; f(·) is an activation function, using the ReLU activation function; is the feature mean of channel c; n is the length of the time series; is the characteristic value of channel c at time point i; and is the factor weight parameter; mean(·) and std(·) are the feature mean and standard deviation respectively; is the channel attention weight; is the adjusted eigenvalue; y is the output feature after residual connection; It is the feature extracted by 1D convolution; Step S32: loss function design; construct monotonic loss, the pollutant concentration has a monotonic characteristic, and the pollutant concentration gradually decreases when there is no new pollution source. It is expressed as: ; Construct difference loss , expressed as: ; Construct basic penalty loss , expressed as: ; The final loss L is obtained, expressed as: ;in, and are the predicted values ​​at the i-th time point and the i-1-th time point respectively; is the rate of change of the predicted value; y i and j2 are the true values ​​at the ith time point and the j2th time point respectively; is the predicted value at the j2th time point; is the weight of the regularization term; is the weight of the noise suppression term; is the variance of the predicted value; and is the loss weight; W is the weight set; Step S33: model determination; the data set processed in step S2 is divided into a test set and a training set in advance. When the environmental quality assessment model converges to the training set loss or reaches the maximum number of training times, the environmental quality assessment model training is completed; an accuracy threshold is set in advance. When the prediction accuracy of the trained environmental quality assessment model for the test set is higher than the accuracy threshold, the environmental quality assessment model is established. Otherwise, the model is retrained based on the particle swarm search algorithm to optimize the model parameters.

2. The environmental quality assessment method based on big data according to claim 1 is characterized by: In step S1, the historical environmental quality assessment data includes environmental pollutant data, meteorological factor data, time, geographic information data and environmental quality assessment grade; the environmental quality assessment grade is used as a data label; and the collected data is cleaned, converted and standardized; Obtain environmental quality time series data.

3. The environmental quality assessment method based on big data according to claim 2 is characterized by: In step S4, the environmental quality assessment is based on the established environmental quality assessment model, real-time collection of environmental quality data to be assessed, and the environmental quality assessment grade output by the environmental quality assessment model is used as the environmental quality assessment result of the environmental quality data to be assessed.

4. A system for evaluating environmental quality based on big data, for implementing the method for evaluating environmental quality based on big data as claimed in any one of claims 1 to 3, characterized in that: It includes data acquisition module, data preprocessing module, environmental quality assessment model building module and environmental quality assessment module; The data acquisition module collects historical environmental quality assessment data and sends the data to the data preprocessing module; The data preprocessing module corrects the dynamic changes and deviations of environmental data through dynamic factors, distortion measurement and normalization processing; and sends the data to the environmental quality assessment model establishment module; The environmental quality assessment model building module uses a parallel-serial residual network, dynamic weights and bidirectional GRU to extract the time series characteristics of pollutants and meteorological factors, generate environmental quality assessment results, and send the data to the environmental quality assessment module; The environmental quality assessment module implements environmental quality assessment on the historical environmental quality assessment data collected in real time based on the established environmental quality assessment model.

Images