A method for tracing, positioning, classifying and evaluating noise complaints

Abstract

The present invention discloses a method for tracing, positioning, classifying and evaluating noise complaints, which relates to the technical field of noise monitoring. The aim is to solve the technical problems of the current noise evaluation method, such as low evaluation efficiency and difficulty in effectively tracing and detecting noise sources widely distributed in society. The method includes the following steps: S1, collecting noise complaint data, analyzing the large-time-scale characteristics of noise according to the noise complaint data to obtain the noise complaint frequency of each grid #imgabs0#; S2, performing time series modeling on the noise complaint frequency of each grid #imgabs1# to obtain the noise complaint time series model of each grid #imgabs2#, and calculating the residual fluctuation degree of the noise complaint time series of each grid #imgabs3#. The present invention solves the problems of high cost, low efficiency and low accuracy of the current noise pollution tracing method, improves the fineness of noise complaint management and the accuracy of tracing, and further improves the management level of noise.

Description

Technical Field

[0001] The present invention relates to the technical field of noise monitoring, and more specifically, to a method for tracing the source, positioning, classifying, and evaluating noise complaints. Background Art

[0002] With the rapid development of social industrialization and urbanization, environmental pollution has become a serious social problem that cannot be ignored. Due to its unique characteristics such as uncertainty, difficulty in measurement, and exposure, noise pollution has become a new form of environmental pollution. In recent years, due to the influence of industrial production, transportation, urban expansion, and residents' lives, there are more and more various noise sources, with increasing intensity, complexity in nature, and severity of harm.

[0003] Currently, the main methods for noise evaluation include manual analysis and evaluation, automatic night noise monitoring stations, noise analysis software, etc. Among them, the manual analysis and evaluation method mainly conducts manual noise detection and evaluation through professional technicians or monitoring personnel using detection instruments.

[0004] Currently, the methods for tracing the source, positioning, classifying, and detecting noise complaints mainly start from traffic noise detection points, monitoring stations, and monitoring areas, construct a rapid traffic noise evaluation system based on the geographic information system, use GIS as a platform, and compare the measured noise values at the monitoring points with the theoretical model noise values to conduct traffic noise evaluation. The evaluation efficiency is relatively low, and it is difficult to effectively trace and detect noise sources widely distributed in society. In view of this, we propose a method for tracing the source, positioning, classifying, and evaluating noise complaints. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for tracing the source, positioning, classifying, and evaluating noise complaints to solve the technical problems of the current noise evaluation method with relatively low evaluation efficiency and difficulty in effectively tracing and detecting noise sources widely distributed in society.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A method for tracing the source, positioning, classifying, and evaluating noise complaints, including the following steps:

[0007] S1: Collect noise complaint data, conduct large-time-scale feature analysis of the noise based on the noise complaint data, and obtain the noise complaint frequency of each grid ;

[0008] S2: Perform time series modeling on the noise complaint frequency of each grid to obtain the noise complaint time series model of each grid ;

[0009] S3: Calculate the residual fluctuation degree of the noise complaint time series of each grid ;

[0010] S4: Extract the characteristics of the residual fluctuation degree of the noise complaint time series to obtain the characteristic data set of the residual fluctuation degree of the noise complaint time series ;

[0011] S5: Collect noise event data, and calculate the influence range of each noise event and the grid at the corresponding position according to the noise event data ;

[0012] S6: According to the influence situation of each grid in the area affected by the noise event, classify the grids affected by each noise event to establish a classification model, and obtain the classification model of the residual fluctuation degree characteristic of each noise complaint time series;

[0013] S7: Calculate the residual fluctuation degree characteristics of the noise complaint time series of each grid and its surrounding grids, merge each grid and its surrounding grids to obtain the merged grid , repeat S6 to establish the classification model of the merged grid ;

[0014] S8: According to the classification model of the merged grid , obtain the noise influence degree characteristics of each grid for each noise event;

[0015] S9: According to the average noise complaint time series residual fluctuation degree characteristic vector of each merged grid , obtain the noise influence degree characteristics of each grid affected by each noise event .

[0016] By collecting noise complaint data and noise event data, combining grid processing to calculate the influence range of each noise event and the grid at the corresponding position, and cooperating with the classification model to obtain the noise influence degree characteristics of each grid, the present invention solves the problems of high cost, low efficiency, and low accuracy of the current noise pollution traceability method. The present invention not only improves the fineness of noise complaint management, improves the accuracy of traceability, but also greatly improves work efficiency, saves manpower and material resources, and further improves the management level of noise.

[0017] Preferably, the step S1 further includes the following steps:

[0018] S101: Collect noise complaint data, grid the location of each complaint event to obtain the grid location, and represent the noise complaint frequency of each grid according to the grid complaint frequency vector. Let the noise complaint frequency of each grid is the noise complaint frequency vector of the grid , and is the noise complaint frequency of the grid . is the number of noise complaints received in the area where the grid is located within a continuous time period . is the total number of grids, and the relationship is: , where represents the number of noise complaints received in the area where the grid is located within the th continuous time period . is the grid where the noise event occurs;

[0019] S102: Construct a sequence of noise complaint frequency vectors for each grid .

[0020] S103: Select the maximum value in the noise complaint frequency vector of each grid to obtain a sequence of maximum value vectors for each grid and construct a noise complaint large time-scale grid feature dataset :

[0021] Among them, the maximum value in the noise complaint frequency vector means: Let ;

[0022] Then represents the noise complaint large time-scale grid feature dataset constructed by taking the maximum value of the complaint frequencies in each continuous time period in the vector . .

[0023] Preferably, step S2 further includes the following steps:

[0024] S201: Perform time series modeling on the noise complaint frequency of each grid .

[0025] Establish an autoregressive moving average model for the noise complaint frequency time series: , where is the noise complaint frequency in the noise complaint large time-scale grid feature, is the noise complaint frequency of the grid at the past time , is the time series model coefficient vector and is the time series model coefficient of the grid in is the autoregressive order;

[0026] S202: Obtain the noise complaint time series model for each grid and each grid based on the time series model coefficient vector : , where represents the predicted value of the noise complaint frequency of grid at time , is the coefficient of the moving average part, is the time and is an element in the model error vector

[0027] Preferably, step S3 further includes the following steps:

[0028] S301: Obtain the time series residual vector for each grid based on the noise complaint time series model and the noise complaint frequency vector of each grid . The time series residual vector is expressed as: , where is the time series residual vector, is the noise complaint frequency vector predicted according to the time series model;

[0029] Among them, the mean of the variance vector of each grid is: , where is the th element in the vector, is the vector dimension;

[0030] S302: Obtain the noise complaint time series residual fluctuation degree of each grid based on the variance vector of each grid :

[0031] ;

[0032] where is the noise complaint time series residual fluctuation degree of each grid .

[0033] Preferably, step S4 further includes the following steps:

[0034] S401: Extract features from the noise complaint time series residual fluctuation degree of each grid to obtain each grid Characteristics of the residual fluctuation degree of the noise complaint time series: , where represents the grid characteristics of the residual fluctuation degree of the noise complaint time series, is the feature extraction function;

[0035] S402: Obtain the dataset of the residual fluctuation degree characteristics of the noise complaint time series based on the characteristics of the residual fluctuation degree of the noise complaint time series for each grid . .

[0036] Preferably, step S5 further includes the following steps:

[0037] S501: Collect noise event data, obtain noise event data and noise event attribute data, and calculate the influence range of each noise event according to the noise event attribute data;

[0038] S502: Obtain the grid coordinates of the location where each noise event occurs according to the longitude and latitude coordinates of the location where each noise event occurs;

[0039] S503: Calculate the grid at the corresponding position according to the grid coordinates of the location where each noise event occurs .

[0040] Preferably, step S6 further includes the following steps:

[0041] S601: Classify the grids in the influence of each noise event according to the grid where each noise event occurs, and the grid where each noise event affects, to obtain the classification result of each noise event on the grid in each noise complaint, and construct the classification training set of the characteristics of the residual fluctuation degree of the noise complaint time series; the classification result is divided into affected by the noise event and not affected by the event; ;

[0042] S602: Establish the SVM classification model for each grid , and obtain the classification model of the influence degree characteristics of each noise complaint event; ;

[0043] Among them, the establishment process of the classification model is as follows:

[0044] Use the radial basis kernel function: , where is the radial basis kernel function, and is the grid where the noise event occurs, is the width parameter of the kernel function, is the Euclidean distance or block distance between two grids;

[0045] ;

[0046] In the formula, is the set of grids and is the number of grids and , sign function;

[0047] ;

[0048] In the formula, is the influence label of the noise event on the grid in each noise complaint. If the classification result is affected by the noise event, then , otherwise, , is the residual of each grid , is the coefficient, threshold parameter.

[0049] Preferably, step S8 further includes the following steps:

[0050] S801: According to the classification model of each merged grid , obtain the average classification result of each noise event for each merged grid . Let be the average classification result of the merged grid , be the classification results corresponding to the original grids included in the merged grid , be the number of original grids included in the merged grid , then ;

[0051] S802: According to the average classification result of each merged grid , obtain the average noise complaint time series residual fluctuation degree eigenvector of each merged grid . If the merged grid is affected by the noise event, then the average classification result of this merged grid is 1, and the average noise complaint time series residual fluctuation degree eigenvector of this merged grid selects the corresponding grid Integrate the eigenvalues, otherwise the merged grid The average classification result is -1, indicating that the merged grid is not affected by noise events.

[0052] Preferably, the noise event attribute data includes the noise event type and the maximum value of the noise volume in the noise event, and the noise event data includes the influence range of the noise event, the geographical coordinates and occurrence time of the noise event, the event type, and the maximum sound level of the noise event.

[0053] A noise complaint traceability positioning and classification evaluation system, including a noise complaint frequency analysis unit, a noise complaint model calculation unit, a residual fluctuation degree feature calculation unit, a residual fluctuation degree feature extraction unit, a noise event data acquisition unit, and a noise event data classification unit;

[0054] The noise complaint frequency analysis unit is used to collect noise complaint data, perform large-time-scale feature analysis of noise according to the noise complaint data, and obtain the noise complaint frequency of each grid. The noise complaint model calculation unit is connected to the noise complaint frequency analysis unit and is used to perform time series modeling on the noise complaint frequency of each grid to obtain the noise complaint time series model of each grid;

[0055] The residual fluctuation degree feature calculation unit is connected to the noise complaint model calculation unit and is used to calculate the residual fluctuation degree of the noise complaint time series of each grid;

[0056] The residual fluctuation degree feature extraction unit is connected to the residual fluctuation degree feature calculation unit and is used to extract features from the residual fluctuation degree of the noise complaint time series of each grid to obtain the residual fluctuation degree feature of the noise complaint time series of each grid;

[0057] The noise event data acquisition unit is used to collect noise event data, calculate the influence range of each noise event and the grid at the corresponding position according to the noise event data. The noise event data classification unit is connected to the residual fluctuation degree feature extraction unit and the noise event data acquisition unit, and is used to classify the grids affected by the noise event according to the affected situation of each grid area by the noise event, establish a classification model, and obtain the noise influence degree feature of each grid.

[0058] Compared with the prior art, the beneficial effects of the present invention are:

[0059] 1. The present invention collects noise complaint data and noise event data, combines grid processing to calculate the influence range of each noise event and the grid at the corresponding position, and cooperates with a classification model to obtain the noise influence degree characteristics of each grid, solving the problems of high cost, low efficiency, and low accuracy in current noise pollution tracing methods. The present invention not only improves the fineness of noise complaint management, improves the accuracy of tracing, but also greatly improves work efficiency, saves manpower and material resources, and further improves the management level of noise.

[0060] 2. The present invention can also extract the characteristics of noise complaints on a large time scale for each grid, thereby obtaining the noise complaint frequency, and through comparative analysis, extract the characteristics of the fluctuation degree of the noise complaint time series for each grid to improve the analysis efficiency, effectively combine the noise event data, and extract the prediction classification model for the influence range of each noise event to effectively evaluate the influence degree of the noise event.

[0061] 3. The present invention also calculates and extracts the characteristics of the residual fluctuation degree of the noise complaint time series for each grid by means of a residual fluctuation degree characteristic calculation unit and a residual fluctuation degree characteristic extraction unit, can deeply understand the fluctuation characteristics of the noise complaint situation in each grid, and at the same time combines a noise event data acquisition unit and a noise event data acquisition classification unit to collect noise event data, calculate the influence range, classify the affected grids and obtain the noise influence degree characteristics, so as to comprehensively and deeply obtain the actual influence characteristics of noise on the surrounding environment and residents in different grid areas, which helps to more scientifically and reasonably evaluate the degree of noise harm. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0063] Embodiment 1: As Figure 1 shown, a method for tracing, positioning, classifying and evaluating noise complaints according to the present invention includes the following steps:

[0064] S1: Collect noise complaint data, perform noise large-time-scale feature analysis based on the noise complaint data, and obtain the noise complaint frequency of each grid ;

[0065] S101: Collect noise complaint data, grid the location of each complaint event to obtain the grid location, and represent the noise complaint frequency of each grid according to the grid complaint frequency vector. Let be the noise complaint frequency vector of grid , be the noise complaint frequency of grid , and be the continuous time period ​ Inner grid The number of noise complaints received in the area where it is located, is the total number of grids, and the relationship is: , where, represents in the th consecutive time period Inner grid The number of noise complaints received in the area where it is located, is the grid where the noise event occurred;

[0066] S102: Construct the noise complaint frequency vector sequence for each grid ;

[0067] S103: Select the maximum value in the noise complaint frequency vector of each grid to obtain the maximum value vector sequence of each grid and construct the noise complaint large time-scale grid feature dataset : :

[0068] Among them, the maximum value in the noise complaint frequency vector means: Let , then represents the noise complaint large time-scale grid feature dataset constructed by taking the maximum value of the complaint frequencies in each consecutive time period in the vector ; ;

[0069] S2: Perform time series modeling on the noise complaint frequency of each grid to obtain the noise complaint time series model of each grid ;

[0070] S201: Perform time series modeling on the noise complaint frequency of each grid ;

[0071] Establish an autoregressive moving average model for the noise complaint frequency time series: , where, is the noise complaint frequency in the noise complaint large time-scale grid feature, is the grid at the past moment 's noise complaint frequency, is the time series model coefficient vector in the grid 's time series model coefficient, is the model error vector, is the autoregressive order;

[0072] S202: According to the noise complaint large time-scale grid feature dataset and each grid The coefficient vector of the time series model to obtain the noise complaint time series model of each grid : , where represents the predicted value of the noise complaint frequency of grid at time , is the coefficient of the moving average part, is the time model error vector element;

[0073] S3: Calculate the residual fluctuation degree of the noise complaint time series of each grid ;

[0074] S301: According to the noise complaint time series model and the noise complaint frequency vector of each grid to obtain the time series residual vector of each grid , the time series residual vector is expressed as: , where is the time series residual vector, is the noise complaint frequency vector predicted according to the time series model;

[0075] Among them, the mean of the variance vector of each grid is: , where is the th element in the vector, is the vector dimension;

[0076] S302: According to the variance vector of each grid to obtain the residual fluctuation degree of the noise complaint time series of each grid :

[0077] ;

[0078] where[[ID=6B]] is the residual fluctuation degree of the noise complaint time series of each grid ;

[0079] S4: Extract features from the residual fluctuation degree of the noise complaint time series to obtain the residual fluctuation degree feature dataset of the noise complaint time series ;

[0080] S401: Extract features from the residual fluctuation degree of the noise complaint time series of each grid to obtain the residual fluctuation degree feature of each grid Characteristics of the residual fluctuation degree of the noise complaint time series: , where represents the residual fluctuation degree characteristic of the noise complaint time series of grid , is a feature extraction function;

[0081] S402: Obtain the dataset of the residual fluctuation degree characteristics of the noise complaint time series based on the residual fluctuation degree characteristics of the noise complaint time series of each grid ; ;

[0082] S5: Collect noise event data, and calculate the influence range of each noise event and the grid at the corresponding location according to the noise event data ;

[0083] S501: Collect noise event data, obtain the noise event data and the noise event attribute data, and calculate the influence range of each noise event according to the noise event attribute data;

[0084] In the embodiments of the present invention, the noise event attribute data includes the type of the noise event and the maximum value of the noise volume in the noise event, and the noise event data includes the influence range of the noise event, the geographical coordinates and the occurrence time of the noise event, the event type, and the maximum sound level of the noise event;

[0085] S502: Obtain the grid coordinates of the location where each noise event occurs according to the longitude and latitude coordinates of the location where each noise event occurs;

[0086] S503: Calculate the grid at the corresponding location according to the grid coordinates of the location where each noise event occurs ;

[0087] [[ID=**********]]S6: Classify the grids affected by the noise events in each grid , establish a classification model, and obtain the classification model of the residual fluctuation degree characteristics of each noise complaint time series;

[0088] S601: Classify the grids affected by each noise event according to the grid where each noise event occurs, and the grid where each noise event affects, and obtain the classification result of the grids affected by each noise event in the noise complaint, and construct the classification training set of the residual fluctuation degree characteristics of the noise complaint time series ; The classification result is divided into affected by the noise event and not affected by the event;

[0089] S602: Classification training set based on the characteristics of the residual fluctuation degree of the noise complaint time series , establish each grid 's SVM classification model , and obtain the classification model of the impact degree characteristics of each noise complaint event;

[0090] Among them, the process of establishing the classification model is as follows:

[0091] Use the radial basis kernel function: , where is the radial basis kernel function, and are the grids of the noise event occurrence location, is the width parameter of the kernel function, is the Euclidean distance or block distance between two grids;

[0092] ;

[0093] In the formula, is the set of grids , is the number of grids , and , sign function;

[0094] ;

[0095] In the formula, is the impact label of the noise event on each grid in each noise complaint. If the classification result is affected by the noise event, then , otherwise, , is the residual of each grid , is the coefficient, threshold parameter;

[0096] S7: Calculate the characteristics of the residual fluctuation degree of the noise complaint time series of each grid and its surrounding grids, merge each grid and its surrounding grids to obtain the merged grid , and repeat the sub-step S602 in step S6 to establish the classification model of the merged grid ;

[0097] S8: According to the classification model of the merged grid , obtain the noise impact degree characteristics of each grid for each noise event;

[0098] S801: According to the classification model of each merged grid to obtain the average classification result of each noise event in each merged grid Let be the average classification result of the merged grid be the merged grid be the average classification result, be the classification results corresponding to the original grids included in the merged grid , be the number of original grids included in the merged grid be the merged grid , then ;

[0099] S802: According to the average classification result of each merged grid to obtain the eigenvector of the average residual fluctuation degree of the noise complaint time series for each merged grid . If the merged grid is affected by a noise event, then the average classification result of this merged grid is 1, and the eigenvalues corresponding to the grid are selected for integration for the eigenvector of the average residual fluctuation degree of the noise complaint time series of this merged grid. Otherwise, the average classification result of this merged grid is -1, indicating that this merged grid is not affected by a noise event;

[0100] S9: According to the eigenvector of the average residual fluctuation degree of the noise complaint time series for each merged grid to obtain the noise impact degree characteristics of each grid

[0101] By collecting noise complaint data and noise event data, combining grid processing, calculating the influence range of each noise event and the grids at the corresponding positions, and cooperating with the classification model to obtain the noise impact degree characteristics of each grid, the present invention solves the problems of high cost, low efficiency, and low accuracy of the current noise pollution tracing method. The present invention not only improves the fineness of noise complaint management, improves the accuracy of tracing, but also greatly improves the work efficiency, saves manpower and material resources, and further improves the management level of noise.

[0102] Embodiment 2: As another embodiment of the present invention, a noise complaint tracing, positioning and classification evaluation system includes:

[0103] ​​A noise complaint frequency analysis unit, which is used to collect noise complaint data, perform large-time-scale feature analysis on the noise complaint data to obtain the noise complaint frequency of each grid;

[0104] A noise complaint model calculation unit, which is connected to the noise complaint frequency analysis unit and is used to perform time series modeling on the noise complaint frequency of each grid to obtain the noise complaint time series model of each grid;

[0105] A residual fluctuation degree feature calculation unit, which is connected to the noise complaint model calculation unit and is used to calculate the residual fluctuation degree of the noise complaint time series of each grid;

[0106] A residual fluctuation degree feature extraction unit, which is connected to the residual fluctuation degree feature calculation unit and is used to extract features from the residual fluctuation degree of the noise complaint time series of each grid to obtain the residual fluctuation degree feature of the noise complaint time series of each grid;

[0107] A noise event data collection unit, which is used to collect noise event data and calculate the influence range of each noise event and the grid corresponding to the location;

[0108] A noise event data classification unit, which is connected to the residual fluctuation degree feature extraction unit and the noise event data collection unit, and is used to classify the grids affected by each noise event according to the influence of the noise event in the area where each grid is located, establish a classification model, and obtain the noise influence degree feature of each grid.

[0109] The present invention can also extract features of noise complaints on a large time scale for each grid, thereby obtaining the noise complaint frequency, and through comparative analysis, extract the feature of the fluctuation degree of the noise complaint time series of each grid to improve the analysis efficiency, effectively combine the noise event data, and extract the prediction classification model of the influence range of each noise event to effectively evaluate the influence degree of the noise event;

[0110] It also calculates and extracts the residual fluctuation degree feature of the noise complaint time series of each grid by means of a residual fluctuation degree feature calculation unit and a residual fluctuation degree feature extraction unit, can deeply understand the fluctuation characteristics of the noise complaint situation of each grid, and at the same time combines a noise event data collection unit and a noise event data collection and classification unit to collect noise event data, calculate the influence range, classify the affected grids and obtain the noise influence degree feature, so as to comprehensively and deeply obtain the actual influence characteristics of noise on the surrounding environment and residents in different grid areas, which is helpful to more scientifically and reasonably evaluate the noise hazard degree.

[0111] The embodiments disclosed in the present invention are preferred embodiments, but not limited thereto. Those of ordinary skill in the art can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. However, as long as they do not depart from the spirit of the present invention, they are within the protection scope of the present invention.

Claims

1. A method for tracing, positioning, classifying and evaluating noise complaints, characterized in that, It includes the following steps: S1: Collect noise complaint data, perform large time-scale feature analysis on the noise based on the noise complaint data, and obtain the noise complaint frequency of each grid ; S2: For each grid , perform time series modeling on the noise complaint frequency to obtain the noise complaint time series model for each grid . S3: Calculate each grid for the residual fluctuation degree of the noise complaint time series; S4: Extract features from the residual fluctuation degree of the noise complaint time series to obtain the feature dataset of the residual fluctuation degree of the noise complaint time series ; S5: Collect noise event data and calculate the influence range of each noise event and the grid at the corresponding location according to the noise event data ; S6: According to the impact of noise events on each grid in the area where it is located, for each grid affected by a noise event is classified, a classification model is established, and a classification model for the residual fluctuation degree characteristics of each noise complaint time series is obtained; S7: Calculate each grid and the characteristic of the residual fluctuation degree of the noise complaint time series of the surrounding grids, and merge each grid and the surrounding grids to obtain the merged grid , repeat S6 to establish a classification model for the merged grid ; S8: According to the classification model of the merged grid obtain the noise impact degree feature of each grid for each noise event; S9: According to the eigenvector of the average noise complaint time series residual fluctuation degree of each merged grid obtain the noise impact degree characteristics of each noise event received by each grid .

2. The noise complaint traceability positioning and classification evaluation method according to claim 1, wherein The step S1 further includes the following steps: S101: Collect noise complaint data, grid the location of each complaint event to obtain grid locations, and represent each grid according to the grid complaint frequency vector. Let be the noise complaint frequency of grid , and be the noise complaint frequency vector of grid . Let be the number of noise complaints received in the area where grid is located within the continuous time period . Let be the total number of grids, and their relationship is: where represents the number of noise complaints received in the area where grid is located within the th continuous time period , and is the grid where the noise event occurred. S102: Construct each grid of the noise complaint frequency vector sequence; S103: Select each grid and obtain the maximum value in the noise complaint frequency vector of each grid to get the maximum value vector sequence of each grid and construct a large time-scale grid feature dataset for noise complaints Among them, the maximum value in the noise complaint frequency vector indicates: Let ; Then, denotes the noise complaint large time-scale grid feature dataset constructed by taking the maximum value of the complaint frequencies in each continuous time period in the vector . .

3. The method for tracing, positioning, classifying and evaluating noise complaints according to claim 2, characterized in that The step S2 further includes the following steps: S201: Perform time series modeling on the noise complaint frequency for each grid ; An autoregressive moving average model is established for the time series of noise complaint frequencies: , where is the noise complaint frequency in the large time-scale grid feature of noise complaints, is the grid at the past time of the noise complaint frequency, is the time series model coefficient vector in the grid of the time series model coefficient, is the model error vector, is the autoregressive order; S202: Obtain the noise complaint time series model for each grid based on the large time-scale grid feature dataset of noise complaints and each grid 's time series model coefficient vector : , where represents the predicted value of the noise complaint frequency of grid at time , are the coefficients of the moving average part is the model error vector at time and is an element in it 4. The noise complaint traceability positioning and classification evaluation method according to claim 3, wherein The step S3 further includes the following steps: S301: According to each grid 's noise complaint time series model and noise complaint frequency vector, obtain each grid 's time series residual vector, and the time series residual vector is expressed as: , where in the formula, is the time series residual vector, is the noise complaint frequency vector predicted according to the time series model; Among them, each grid has a mean of the variance vector as follows: , where is the -th element in the vector, and is the vector dimension; S302: According to each grid to obtain the residual fluctuation degree of the noise complaint time series for each grid from the variance vector: ； In the formula, is the degree of residual fluctuation of the noise complaint time series for each grid .

5. A method for tracing, positioning, classifying and evaluating noise complaints according to claim 4, characterized in that, The step S4 further includes the following steps: S401: For each grid extract the feature of the residual fluctuation degree of the noise complaint time series to obtain the feature of the residual fluctuation degree of the noise complaint time series for each grid : , where represents the feature of the residual fluctuation degree of the noise complaint time series of grid , and is the feature extraction function. S402: According to each grid obtain the dataset of the residual fluctuation degree characteristics of the noise complaint time series based on the residual fluctuation degree characteristics of the noise complaint time series .

6. The method for tracing, positioning, classifying and evaluating noise complaints according to claim 5, wherein The step S5 further includes the following steps: S501: Collect noise event data, obtain noise event data and noise event attribute data, and calculate the influence range of each noise event according to the noise event attribute data; S502: Obtain the grid coordinates of the occurrence location of each noise event according to the longitude and latitude coordinates of the occurrence location of each noise event; S503: Calculate the grid at the corresponding position according to the grid coordinates of the occurrence location of each noise event .

7. A method for tracing, positioning, classifying and evaluating noise complaints according to claim 6, characterized in that The step S6 further includes the following steps: S601: According to the grids of the occurrence locations of each noise event , and the grids of the affected locations of each noise event , classify the grids affected by each noise event to obtain the classification results of the grids in each noise complaint by the noise event, and construct a classification training set for the characteristics of the residual fluctuation degree of the noise complaint time series ; the classification results are divided into affected by the noise event and not affected by the event;​ S602: Classification training set based on the characteristics of the residual fluctuation degree of the noise complaint time series , establish each grid 's SVM classification model , and obtain the classification model for the characteristics of the impact degree of each noise complaint event; Among them, the classification model The establishment process is as follows: Using a radial basis kernel function: , where is the radial basis kernel function,[[]] and are the grids at the locations where noise events occur,[[]] is the width parameter of the kernel function,[[]] is the Euclidean distance or the block distance between two grids; ； In the formula, is the set of meshes , is the number of meshes , and , sign function; ； In the formula, is the impact label of the noise event on the grid in each noise complaint. If the classification result is affected by the noise event, then , otherwise, . , is the residual of each grid , is the coefficient, is the threshold parameter.

8. A method for tracing, positioning, classifying and evaluating noise complaints according to claim 7, characterized in that, The step S8 further includes the following steps: S801: According to the classification model of each merged grid obtain the average classification result of each noise event for each merged grid . Let be the average classification result of the merged grid , be the classification results corresponding to each original grid contained in the merged grid , , be the number of original grids contained in the merged grid , then ; S802: According to the average classification result of each merged grid obtain the eigenvector of the average noise complaint time series residual fluctuation degree of each merged grid . If the merged grid is affected by a noise event, then the average classification result of this merged grid is 1, and the eigenvalue corresponding to the grid is selected for integration of the eigenvector of the average noise complaint time series residual fluctuation degree of this merged grid. Otherwise, the average classification result of this merged grid is -1, indicating that this merged grid is not affected by a noise event. ​ 9. A method for tracing, positioning, classifying and evaluating noise complaints according to claim 6, characterized in that The noise event attribute data includes the noise event type and the maximum value of the noise volume in the noise event. The noise event data includes the influence range of the noise event, the geographical coordinates of the occurrence of the noise event, the occurrence time, the event type, and the maximum sound level of the noise event.

10. A noise complaint traceability, positioning, classification and evaluation system, which uses the noise complaint traceability, positioning, classification and evaluation method described in any one of claims 1-9, characterized in that, It includes: A noise complaint frequency analysis unit, which is used to collect noise complaint data, perform noise large-time-scale feature analysis according to the noise complaint data, and obtain the noise complaint frequency of each grid; A noise complaint model calculation unit, which is connected to the noise complaint frequency analysis unit and is used to perform time series modeling on the noise complaint frequency of each grid to obtain the noise complaint time series model of each grid; A residual fluctuation degree feature calculation unit, which is connected to the noise complaint model calculation unit and is used to calculate the residual fluctuation degree of the noise complaint time series of each grid; A residual fluctuation degree feature extraction unit, which is connected to the residual fluctuation degree feature calculation unit and is used to extract features from the residual fluctuation degree of the noise complaint time series of each grid to obtain the residual fluctuation degree feature of the noise complaint time series of each grid; A noise event data collection unit, which is used to collect noise event data and calculate the influence range of each noise event and the corresponding grid at the position; A noise event data classification unit, which is connected to the residual fluctuation degree feature extraction unit and the noise event data collection unit, and is used to classify the grids affected by the noise event according to the affected situation of the area where each grid is located, establish a classification model, and obtain the noise influence degree feature of each grid.

Images