Method and device for predicting concentration of TSP under influence of dust suppression measures in port and storage medium

Abstract

The application discloses a port TSP concentration prediction method and device considering the influence of dust suppression measures and a storage medium, and the port TSP concentration prediction method considering the influence of dust suppression measures comprises the following steps: acquiring historical monitoring data of TSP concentration and port meteorological factor data after dust suppression measures are implemented, and cleaning the data; a meteorological factor feature screening model is constructed to screen out characteristic meteorological factors influencing the TSP concentration; according to the TSP concentration data and the characteristic meteorological factors, a three-layer multi-input single-output prediction model is constructed to predict the future TSP concentration of the port. According to the change of the meteorological factors after the dust suppression measures are implemented, the application can accurately predict the port TSP concentration, provides theoretical support for reasonably selecting the dust suppression measures, and has important significance for the prevention and control of atmospheric pollution of the port.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of atmospheric pollution prevention and control technology in the field of environmental engineering, and particularly relates to a method and device for predicting TSP concentration of a port under the influence of dust suppression measures and a storage medium. BACKGROUND

[0002] The port is an important node of the logistics supply chain and the center of regional economic trade. However, the dust particles generated in the process of loading and unloading, stacking and transporting of goods in the port will have a significant negative impact on the surrounding environment and human health. The inventors found that due to historical reasons, the concentration of total suspended particulate matter (TSP) in the ambient air of the surrounding area of some old ports is generally over standard.

[0003] In order to suppress TSP pollution in the port, the port often uses measures such as spraying atomization, windproof net, and closed treatment. These measures will affect the TSP concentration characteristics of the port, resulting in strong particularity in the time sequence change of the TSP concentration of the port. Therefore, if only the time sequence characteristics of the TSP concentration data are considered without considering the influence of the dust suppression measures, it is difficult to accurately predict the concentration.

[0004] Accurate and efficient prediction of the TSP concentration of the port can help the port managers to adjust the operation and management mode and the dust suppression measures, and is one of the primary tasks of building a smart and green world-class port. The inventors' previous research found that the implementation of the dust suppression measures in the port will directly cause changes in the monitoring data of meteorological factors. Therefore, when predicting the TSP concentration, the influence of the implementation of the dust suppression measures on the change of the TSP concentration can be reflected through the change trend of the meteorological factors. Therefore, when constructing the prediction model of the TSP concentration, considering the correlation between the port meteorological factors and the time sequence non-stationary fluctuation of the TSP concentration data can significantly improve the accuracy and reliability of the short-term prediction of the TSP concentration of the port. SUMMARY

[0005] The technical problem to be solved by the present application is that the implementation of the dust suppression measures will affect the meteorological environment of the port, making the time-varying characteristics of the TSP concentration data have obvious non-stationarity, and causing the problem of decreased accuracy of TSP concentration prediction.

[0006] To solve the above technical problems, the technical scheme adopted by the present application is:

[0007] The present application first provides a method for predicting the TSP concentration of a port considering the influence of dust suppression measures, comprising:

[0008] obtaining a TSP concentration data set X monitored by the port within T hours M and a meteorological factor data set F under the influence of dust suppression measures M,K ;

[0009] predicting the TSP concentration XM and meteorological factor monitoring data F M,K After cleaning and removing outliers, adjust the data time interval to obtain the TSP concentration data set X for analysis T and meteorological factor data set F T,K ;

[0010] A feature meteorological factor feature screening model is constructed to analyze the correlation between the TSP concentration data set X T and meteorological factor data set F T,K , and the feature meteorological factor data set F that can affect the TSP concentration is screened out T,D ;

[0011] A three-layer multi-input single-output TSP concentration prediction model is constructed according to the monitored TSP concentration data set X T and the feature meteorological factor data set F T,D , and the optimal parameters of the model are determined

[0012] According to the TSP concentration data set X T , the feature meteorological factor data set F T,D and the TSP concentration prediction model, the TSP concentration prediction result X T+1 of the port is obtained.

[0013] The application also provides a port TSP concentration prediction device, comprising a processor and a memory; the memory stores programs or instructions, which are loaded and executed by the processor to realize the steps of the port TSP concentration prediction method.

[0014] The application also provides a computer readable storage medium, which stores programs or instructions, and the programs or instructions are executed by the processor to realize the steps of the port TSP concentration prediction.

[0015] The technical scheme adopted by the application has the following beneficial effects:

[0016] (1) Compared with the traditional scheme, when analyzing the port TSP concentration, the change trend of the meteorological factor is used to reflect the influence of the implementation of the dust suppression measure on the TSP concentration change. And at the same time, the size of the linear and nonlinear correlation coefficient is considered to determine the key meteorological factor that has significant correlation with the TSP concentration change.

[0017] (2) According to the correlation between the key meteorological factor of the port and the time series non-stationary fluctuation of the TSP concentration data, a three-layer multi-input single-output TSP concentration prediction model is constructed, which can significantly improve the accuracy and reliability of the port TSP concentration prediction. DETAILED DESCRIPTION

[0018] Figure 1 Basic technical flowchart of the present application;

[0019] Figure 2 The historical monitoring data of TSP concentration and meteorological factors of a certain port;

[0020] Figure 3 The result of cleaning the historical monitoring data; Figure 2 The result of cleaning the historical monitoring data;

[0021] Figure 4 The result of linear correlation coefficient analysis;

[0022] Figure 5 The result of nonlinear correlation coefficient analysis;

[0023] Figure 6 The result of screening of characteristic meteorological factors;

[0024] Figure 7 The comparison chart of measured data and TSP concentration prediction results of different models. DETAILED DESCRIPTION

[0025] The present scheme is further described in conjunction with the accompanying drawings:

[0026] The present embodiment provides a method for predicting the TSP concentration of a port under the influence of dust suppression measures, comprising:

[0027] A. Data acquisition:

[0028] A.1 Data collection method. The data required to be obtained includes the historical monitoring data of TSP concentration and meteorological factors. The data can be monitored by electrochemical and optical sensors at the same location in the port, uploaded to the monitoring cloud platform for storage in a wireless transmission manner according to the RS485 communication protocol, and the sampling frequency of the data is 1 minute per sample. In order to make the model training have enough samples, it is suggested that the length of the historical monitoring data used for analysis should not be less than 15 days.

[0029] A.2 Acquisition of port TSP concentration data. According to the TSP online monitoring system of the port, M TSP concentration data monitored by the port within a time range T constitute a data set X with time series characteristics M = (x1, x2, … x M );

[0030] A.3 Acquisition of port meteorological factor data. Dust suppression measures such as spraying atomization and windproof net will change the meteorological conditions of the port. Therefore, the influence of dust suppression measures can be reflected through the monitoring data of port meteorological factors. Within a time range T, K meteorological factors monitored constitute a data set Y with time series characteristics

[0031]

[0032] B. Data Cleaning:

[0033] The method for cleaning historical monitoring data of TSP concentration and meteorological factors is as follows:

[0034] B.1 Abnormal Data Removal: If there are M TSP concentration datasets X M In the data, x is the monitoring data at the m-th time point (m∈M). m Satisfy x m <A min or x m >A max Then x m This is used to remove outlier data. Similarly, for M×K meteorological monitoring data, the M monitoring data F for the k-th (k∈K) meteorological factor are removed. M,k In the data, f is the monitoring data at the m-th time point (m∈M). m,k Satisfy f m,k <B min or f m,k >B max Then f m,k Removed as abnormal data. Among them, A... min A max B min B max for:

[0035]

[0036] B.2 Time Interval Adjustment: The acquired TSP concentration data and meteorological factor data are adjusted to the concentration data at time t; the TSP concentration x at time t is obtained by arithmetically averaging the q TSP concentration data between time t-1 and t. t Similarly, by taking the arithmetic mean of the p monitoring data of the kth meteorological factor between time t-1 and t, we obtain the monitoring data of the kth meteorological factor at time t as f. t,k .

[0037]

[0038] B.3 Data Cleaning Results: X data from the M TSP concentration data points at the port M =(x1,x2,…x M After cleaning, a TSP concentration dataset X with time-series characteristics is obtained over T hours. T =(x1,x2,…x t ,…x T Similarly, M×K meteorological monitoring data points F M,K After cleaning, T×K meteorological factor monitoring datasets were obtained.

[0039]

[0040] B. Screening of characteristic meteorological factors:

[0041] The method for constructing a feature meteorological factor screening model is as follows:

[0042] B.1 Calculation of linear correlation coefficient: Based on the TSP concentration data X after cleaning T and T×K meteorological monitoring data F T,K Calculate TSP concentration data X T The monitoring data F of the kth meteorological factor T,k The linear correlation coefficient CP between them k for,

[0043]

[0044] Calculate the linear correlation coefficients between TSP concentration data and all K meteorological factors to obtain the vector CP. K =(CP1,CP2,…CP) k ,…CP K ).

[0045] Set the threshold μ for determining linear correlation CP To more fully analyze the impact of different factors, a threshold value of 0.25 is recommended. If the absolute value of the linear correlation coefficient is greater than the threshold, i.e., |CP| > 0.25, then... k |>μ CP If the k-th meteorological factor is considered to have a significant linear correlation with the change in TSP concentration, then G characteristic meteorological factors (G≤K) satisfying this condition are obtained.

[0046] B.2 Calculation of Nonlinear Correlation Coefficient: Based on the cleaned TSP concentration dataset X T and T×K meteorological monitoring datasets F T,K Calculate the TSP concentration dataset X T The monitoring dataset F of the kth meteorological factor T,k The nonlinear correlation coefficient CM between them k for,

[0047]

[0048] In the formula, Pr(x) t ,f t,k Let x be the TSP concentration monitored at a certain time t. t Monitoring data f of meteorological factor k t,k The joint probability of P(x). t (x represents TSP concentration data) t The marginal probability, P(ft,k ) is the monitoring data of meteorological factor k t,k t t,k The distribution range of the Cartesian coordinate axis is divided into LxL grids. At this time, all monitoring data is divided into L intervals in the range of the horizontal axis and the vertical axis, respectively. Pr(x t ,f t,k ) is the value of the number of grids occupied by x t and f t,k divided by the total number of grids, P(x t ) is the value of the number of grids occupied by x t divided by the total number of grids, and P(f t,k ) is the value of the number of grids occupied by f t,k divided by the total number of grids.

[0049] The determination method of interval number L is to set L as (2, 3, 4…, L) respectively, and calculate the corresponding Pr(x t ,f t,k ) for different values of L. The L corresponding to the maximum value of Pr(x t ,f t,k ) is taken as the final interval number.

[0050] The nonlinear correlation coefficients between TSP concentration data and all K meteorological factors are calculated respectively to obtain the vector CM K =(CM1, CM2, … CM k , … CM K ).

[0051] The nonlinear correlation threshold μ CM is set. In order to more fully analyze the influence of different factors, it is recommended that the threshold value is 0.25. If the absolute value of the nonlinear correlation coefficient is greater than the threshold value, that is, |CM k |> μ CM , it is considered that the kth meteorological factor has significant nonlinear correlation with the change of TSP concentration. Thus, E characteristic meteorological factors (E≤K) that meet this condition are obtained.

[0052] B.3 Feature meteorological factor screening: G feature meteorological factors with linear correlation and E feature meteorological factors with nonlinear correlation are combined into C (C=G∪E≤K) feature meteorological factors. That is, there are C feature meteorological factors that may be related to the change of TSP concentration. On this basis, the calculation formula of feature meteorological factor screening index γ k is Δ(γ k ).

[0053] ​​

[0054] where a is a control coefficient, and is recommended to be 0.1; γ k In order to make the calculation formula minΔ(γ k ) value minimum screening index. γ k The determination method is: the setting of γ k The initial value is-1, and is increased to 1.0 with a step of 0.01, and different Δ(γ k ) is calculated, and the γ k corresponding to the minimum value of Δ(γ k ) is selected as the final screening index.

[0055] If γ k =0, the meteorological characteristic factor has no significant influence on the change of TSP concentration, and can be removed. Otherwise, it is considered that the factor is a characteristic meteorological factor. Through Δ(γ k ), the final D characteristic meteorological factors affecting TSP concentration can be determined as F T,D , where D≤C≤K.C. The method for constructing a three-layer multi-input single-output TSP concentration prediction model is:

[0056] C.1 Data conversion: convert the cleaned T TSP concentration data set X T and the characteristic meteorological factor data set F T,D into a series of two-dimensional matrices with a lag time of p as training samples and output labels. Taking the i-th training sample and the output label as an example, the conversion method is:

[0057]

[0058] where i=(1, 2, …, T-p), that is, a total of T-p training samples will be generated, and these training samples constitute the input data of the prediction model as a three-dimensional matrix, and the matrix size is (T-p, p, D+1). Similarly, T-p output labels will be generated, and these output labels constitute the output data of the prediction model as a one-dimensional vector, and the vector size is (T-p).

[0059] C.2 First layer data convolution: for the i-th training sample Set the input data for the first layer convolution Set the total number of first layer neurons to C1. Set the weight matrix of the j-th neuron to The bias matrix is The output value of the j-th neuron is,

[0060] ​

[0061] where is the sum of Hadamard product of elements of two matrices.

[0062] Thus, the input data of the first layer The output data result Y after the convolution i 1 (C1) is,

[0063]

[0064] C.3 Second layer data recursion: for the i-th training sample The output data Y after the first layer convolution i 1 (C1) as the input data X of the second layer data recursion i 2 , i.e., X i 2 = Y i 1 (C1). Let the number of neurons of the second layer be C2, and the input data X i 2 The output data after the data recursion after the state update of the j-th neuron is The final output result of the second layer data recursion is

[0065]

[0066] where tanh is the hyperbolic tangent function, and σ is the sigmoid growth curve function, and are the set state weight values, and are the set bias matrix values, is the output data obtained after the data recursion of the i-1-th training sample When i = 1, is the unit matrix.

[0067] C.4 Third layer weight optimization: taking the output data Y i 2 (C2) as the input data X of the third layer weight optimization i 3 , i.e., X i 3 = Y i 2 (C2). Let the input data X i 3 and the output label Y itrain the relevance score R[X i 3 ,Y i train ] is,

[0068] R[X i 3 ,Y i train ] = FW i 3 X i 3 + Γ i 3

[0069] wherein FW i 3 is a weight matrix, and Γ i 3 is a bias matrix.

[0070] The relevance score R[X i 3 ,Y i train ] for each input data is calculated by a normalized exponential function softmax to obtain the weight index ω i 3 for the input data X i as,

[0071] ω i = softmax(R[X i 3 ,Y i train ])

[0072] The final TSP concentration prediction result X i 3 is obtained by weighted average summation according to the input data X i and the weight index ω t+1 as,

[0073]

[0074] C.5 Model parameter optimization

[0075] Both the weight matrix and the bias term matrix are initial random values, and the optimal parameters need to be determined by an iterative updating method. Using the training sample data, and the constantly iterated weight matrix and bias matrix, the error between the prediction value and the output label is calculated until the prediction result X t+1 (s) of the adjacent s-th step and the prediction result X t+1 (s+1) of the s+1-th step satisfy the constraint condition The parameters at that time are the optimal parameter values. It is recommended to iterate the weight matrix and bias matrix in the negative gradient direction with a step size of 0.01.

[0076] This embodiment also provides a forecasting device for port TSP concentration under the influence of dust suppression measures, including a processor and a memory; the memory stores a program or instructions, which are loaded and executed by the processor to implement the steps of the port TSP concentration forecasting method under the influence of dust suppression measures.

[0077] This embodiment provides a computer-readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the port TSP concentration forecasting method under the influence of dust suppression measures in this embodiment.

[0078] Example 1

[0079] Based on the invention's content, application examples are used to illustrate its application.

[0080] Step 1: Data Acquisition. Taking TSP concentration data monitored at a certain port as an example, historical monitoring data of TSP concentration and meteorological factors at the port from January 1st to March 6th are obtained, such as... Figure 2 As shown.

[0081] from Figure 2 It can be seen that the historical monitoring data of TSP concentration and meteorological factors exhibit time-series characteristics, with an average data acquisition interval of 1 minute. Taking TSP concentration data as an example, a total of 95,040 concentration data points were acquired, i.e., a dataset X of M TSP concentrations. M =(x1,x2,x3,···x M In this context, M is 95040.

[0082] Step Two: Data Cleaning. The acquired TSP concentration and meteorological factor monitoring data are processed by removing outliers and adjusting time intervals to create a time-series dataset with 1-hour intervals. Taking TSP concentration data as an example, a dataset X containing T TSP concentrations... T =(x1,x2,x3,···x T In the sequence, T is 1584, such as... Figure 3 As shown.

[0083] In the subsequent example analysis, 1392 sets of concentration and meteorological factor monitoring data from January 1 to March 6 were used for modeling analysis, and 192 sets of data from March 7 to March 14 were used to compare and analyze the forecast results of the model of this invention, the forecast results of the convolutional neural network model, and the forecast results of the ARIMA model to verify the effectiveness of the method of this invention.

[0084] Step three: feature meteorological factor screening. Linear correlation coefficient of the cleaned TSP concentration and meteorological factor monitoring data, as shown in Figure 4 Nonlinear correlation coefficient of the TSP concentration and meteorological factor monitoring data, as shown in Figure 5 According to the linear correlation coefficient and the nonlinear correlation coefficient of the TSP concentration and meteorological factor monitoring data, the wind speed, temperature and humidity in the meteorological factor are screened to have significant correlation with the change of the TSP concentration, as shown in Figure 6

[0085] Step four: model construction.

[0086] In the optimal parameter iterative optimization process of the model, the error function adopts the mean absolute deviation, and the activation function selects the hyperbolic tangent function. The model parameters constructed are shown in Table 1:

[0087] Table 1: Forecast model parameters

[0088]

[0089]

[0090] The comparison of the measured data of the TSP concentration of 3.7 days and the prediction results of the three prediction models is shown in Figure 7 From the data comparison chart, it can be directly found that the prediction result of the prediction model proposed by the application is closest to the measured data, and can predict the obvious concentration value fluctuation. The prediction results of the convolutional neural network model and the ARIMA model only reflect the general concentration value trend, and many fluctuations are not effectively predicted. Therefore, it can be directly seen that the convolutional neural network model and the ARIMA method are more suitable for data with good stationarity, and it is also verified that the prediction model of the application can better handle the non-stationarity problem of the concentration time series itself.

[0091] In order to further quantitatively judge the prediction results of the three prediction models, the commonly used evaluation indexes: root mean square error (RMSE) and mean absolute error (MAE) are selected for evaluation, as shown in Table 2. The lower the index value, the higher the prediction accuracy, and the formulas of the two evaluation indexes are:

[0092]

[0093]

[0094] In the formula, Y i is the true value, Y i is the predicted value, and N is the sample size.

[0095] Table 2: Comparison table of TSP concentration prediction error

[0096] ​

[0097] Comparing the prediction errors of 192 TSP concentrations during 3.7-3.14, the prediction error of the model of the application is smaller than that of the convolutional neural network model and the ARIMA model. From the evaluation index, compared with the convolutional neural network model, the mean absolute error MAE and the root mean square error RMSE of the model of the application are reduced by 63.7% and 81.8% respectively; compared with the ARIMA method, the mean absolute error MAE and the root mean square error RMSE of the model of the application are reduced by 35.6% and 74.2% respectively. Therefore, the model of the application is effective for predicting the concentration of TSP.

[0098] Under the influence of various dust suppression measures, the TSP concentration data of the port has significant non-stationarity and randomness, and the commonly used prediction methods at the present stage are difficult to meet the demand of TSP concentration prediction of the port, which affects the reliability of dust control of the port. The application mainly designs a port TSP concentration prediction method, device and storage medium considering the influence of dust suppression measures, which can significantly improve the reliability of the prediction of the port TSP concentration, and has important significance for the prevention and control of dust pollution of the port.

Claims

1. A method for predicting port TSP concentration considering the impact of dust suppression measures, characterized in that, include: Obtaining port access Data set of TSP concentrations monitored within hours Data set of meteorological factors under the influence of dust suppression measures ; TSP concentration dataset and meteorological factor dataset After cleaning and removing outliers, the data time intervals were adjusted to obtain the TSP concentration dataset for analysis. and meteorological factor dataset ; Construct a feature meteorological factor screening model and analyze the TSP concentration dataset. and meteorological factor dataset The correlation was analyzed to identify a set of characteristic meteorological factors that could influence TSP concentration. ; Construct a multi-input, single-output TSP concentration prediction model based on the TSP concentration dataset. Feature meteorological factor dataset The concentration forecast results of TSP at the port were obtained. .

2. The port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 1, characterized in that, Construct a feature screening model for characteristic meteorological factors and analyze the TSP concentration dataset. and meteorological factor dataset The correlation was analyzed to identify characteristic meteorological factors that could influence TSP concentration. In the steps, characteristic meteorological factors are constructed. The method for selecting models is as follows: Linear correlation coefficient calculation: based on the cleaned TSP concentration dataset and A meteorological monitoring dataset Calculate TSP concentration dataset With the Monitoring datasets of meteorological factors linear correlation coefficient between : in, x t For a moment Monitored TSP concentration; f t,k For a moment The monitored Data for individual meteorological factors; Calculate TSP concentration data and all The linear correlation coefficients among the meteorological factors are used to obtain vectors. ; Set a threshold for determining linear correlation ,if Then it is believed that the first Several meteorological factors showed a significant linear correlation with changes in TSP concentration, thus yielding results that met this condition. One characteristic meteorological factor, G≤K ; Nonlinear correlation coefficient calculation: based on the cleaned TSP concentration dataset and meteorological monitoring dataset Calculate TSP concentration data With the Monitoring data of meteorological factors Nonlinear correlation coefficient between : In the formula, For a moment Monitored TSP concentration Meteorological factors Monitoring data The smaller value in; For a certain moment Monitored TSP concentration and meteorological factors Monitoring data The joint probability; TSP concentration data The marginal probability, Meteorological factors Monitoring data The marginal probability; Calculate TSP concentration data and all The nonlinear correlation coefficients among various meteorological factors yield vectors. ; Set a threshold for nonlinear correlation determination ,if Then it is believed that the first Several meteorological factors exhibit a significant nonlinear correlation with changes in TSP concentration, thus yielding results that satisfy this condition. One characteristic meteorological factor, E≤K ; Screening of characteristic meteorological factors: linearly correlated Characteristic meteorological factors and nonlinear correlation The characteristic meteorological factors are combined into One characteristic meteorological factor: Calculation formula based on the screening index of characteristic meteorological factors Determine the screening indicators for meteorological factors ; The calculation formula is: In the formula, α For control coefficients; To make the calculation formula The screening criterion with the smallest value; if If the meteorological factor has no significant impact on the change in TSP concentration, it is removed; otherwise, it is considered a characteristic meteorological factor. Determine the factors that ultimately affect TSP concentration The characteristic meteorological factors are ,in .

3. The port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 2, characterized in that, At a certain moment Monitored TSP concentration and meteorological factors Monitoring data joint probability TSP concentration data marginal probability and meteorological factors Monitoring data marginal probability The method for determining this is based on the TSP concentration. and meteorological factors Detection data Within the distribution range of the Cartesian coordinate axes, it is divided into: One grid, for and The value of the number of grid cells occupied divided by the total number of grid cells. for The value of the number of grid cells occupied divided by the total number of grid cells. for The value of the number of grid cells occupied divided by the total number of grid cells.

4. The port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 1, characterized in that, The method for constructing a multi-input single-output TSP concentration prediction model is as follows: Data transformation: Transforming the cleaned TSP concentration dataset and characteristic meteorological factor dataset This is transformed into a series of lag times. A two-dimensional matrix is ​​used as the training sample and output label: First layer of data convolution: targeting the... training samples Set the input data for the first convolutional layer. for Let the total number of neurons in the first layer be denoted as . Let the first The weight matrix of each neuron is as follows The deviation matrix is Then the first Output value of each neuron for: in, It is the sum of the Hadamard products of the elements of the two matrices; Input data for the first layer The output data after convolution for: Second-level data recursion: For the first... training samples The output data after the first convolution layer Input data for the second layer of data recursion ,Right now Let the number of neurons in the second layer be... Input data After the first After each neuron updates its state, the recursive output data is: The final result of the second layer of data recursion is: In the formula, It is the hyperbolic tangent function; This is the sigmoid growth curve function; , , and These are the set state weight values; , , and These are the set deviation matrix values; For the first training samples The output data obtained after data recursion, when hour It is the identity matrix; Third-layer weight optimization: optimize the output data of the second layer. Input data for third-layer weight optimization ,Right now ; Assume input data With output labels correlation score for: In the formula, This is the weight matrix; The deviation matrix; Relevance score for each input data point By normalizing the exponential function Calculate the input data Weighting indicators for: Based on the input data and weighting indicators The final TSP concentration forecast result is obtained by performing a weighted average summation. : 。 5. A port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 4, characterized in that, After cleaning TSP concentration data and characteristic meteorological factors This is transformed into a series of lag times. The transformation method for using a two-dimensional matrix as training samples and output labels is as follows: In the formula, For the first One training sample; For the first One output label; =1, 2, ..., Tp .

6. A port TSP concentration forecasting method considering the impact of dust suppression measures according to any one of claims 1-5, characterized in that, The historical monitoring data on port TSP concentrations and meteorological factors under the influence of dust suppression measures include: Port TSP Concentration Data Acquisition: Based on the port's online TSP monitoring system, within a specified time range... Port monitoring obtained internally The TSP concentration data constitute a dataset with time-series characteristics. ; Port meteorological data acquisition: within the time range Internal monitoring is possible A dataset with time-series characteristics composed of meteorological factors. for: in, f m,k For the first The first meteorological factor One monitoring data point, , .

7. A port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 6, characterized in that, The method for cleaning the historical monitoring data of TSP concentration and meteorological factors is as follows: Abnormal data removal: If in Dataset of TSP concentrations In the middle, the first Monitoring data at each time point satisfy or Then As abnormal data, it was cleared. ;right In the meteorological monitoring data, the first various meteorological factors One monitoring data In the middle, the first Monitoring data at each time point satisfy or Then As abnormal data, it was cleared. ;in , , , for: Time interval adjustment: Adjust the acquired TSP concentration dataset and meteorological factor dataset to any time interval. Concentration data; for time and Between The time was obtained by arithmetically averaging the TSP concentration data. TSP concentration For a moment and Between The first The arithmetic mean of the monitoring data of various meteorological factors was obtained. Time of the first Monitoring data of various meteorological factors are ; Data cleaning results: [The data will be used to clean the port's...] TSP concentration data After cleaning, time-series data with 1-hour intervals were obtained. Dataset of TSP concentrations ; Monitoring datasets of meteorological factors After cleaning Monitoring datasets of meteorological factors .

8. The port TSP concentration forecasting method considering the impact of dust suppression measures according to claim 1, characterized in that, The weight matrix and bias matrix of the constructed three-layer multi-input single-output forecast model are both based on measured TSP concentration data. and characteristic meteorological factor data The optimal parameters are determined through iterative updates.

9. A port TSP concentration forecasting device that takes into account the impact of dust suppression measures, characterized in that, It includes a processor and a memory; the memory stores a program or instructions which are loaded and executed by the processor to implement the steps of the port TSP concentration forecasting method considering the impact of dust suppression measures as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the port TSP concentration forecasting method considering the impact of dust suppression measures as described in any one of claims 1 to 8.

Images