Multi-dimensional drainage port overflow risk assessment system and assessment method

Abstract

The invention discloses a multi-dimensional drainage port overflow risk assessment system and assessment method, and belongs to the technical field of drainage port overflow risk assessment systems, and the system integrates ground monitoring and satellite remote sensing multi-source data, and collects related information of a drainage port in an omnibearing manner. The ground monitoring data provides accurate local hydrology, water quality and drainage facility operation state information, the satellite remote sensing data realizes large-range monitoring, the advantages of the ground monitoring data and the satellite remote sensing data are complementary, the data are fused through a weighted average method and a Kalman filtering method, the data accuracy and reliability are improved, and the weighted average method distributes weights according to data characteristics for preliminary fusion, so that the accuracy and reliability of the data are improved. The Kalman filtering method adapts to data dynamic change, such as water level data fusion according to a state equation and an observation equation optimization result, combines the advantages of ground high-precision monitoring and satellite wide-range coverage, accurately reflects water level change, quantifies the overflow risk of a drainage port, and provides effective support for urban water fine supervision.

Description

Technical Field

[0001] The present invention relates to a drain outlet overflow risk assessment system, and in particular to a multi-dimensional drain outlet overflow risk assessment system and assessment method, belonging to the technical field of drain outlet overflow risk assessment systems. Background Art

[0002] Traditional drainage outlet overflow risk assessments rely solely on data from ground-based monitoring stations, such as simple pipe network water level monitoring, which limits data collection. Due to the complexity of drainage systems and the diverse influencing factors, relying solely on ground-based monitoring fails to capture broad-scale information about water bodies, weather, topography, and land use surrounding the outlets. Consequently, assessing overflow risks across multiple urban outlets hinders accurate macroscopic assessments of drainage system performance and a comprehensive understanding of the relationship between the drainage system and the surrounding environment, leading to incomplete risk assessments.

[0003] Simple data processing methods: In the past, simple data processing methods were often used, and the handling of missing values and outliers in the data was relatively crude. When monitoring equipment malfunctions or is affected by external interference, resulting in data anomalies, simply removing or arbitrarily filling in outliers will cause the data to lose authenticity and integrity, affecting subsequent analysis and assessment results. The lack of effective fusion methods for data from different sources and formats prevents the full utilization of the comprehensive value of multi-source data, reducing the accuracy of risk assessment. Therefore, a multi-dimensional drain overflow risk assessment system and assessment method are designed to address these issues. Summary of the Invention

[0004] The main purpose of the present invention is to provide a multi-dimensional drain outlet overflow risk assessment system and assessment method.

[0005] The purpose of the present invention can be achieved by adopting the following technical solutions:

[0006] A multi-dimensional drain overflow risk assessment system includes a data acquisition module for collecting multiple types of data from ground monitoring sites and satellite remote sensing;

[0007] Data preprocessing module, used for processing ground monitoring data and satellite remote sensing data;

[0008] The data fusion module fuses the data collected by satellite remote sensing and the data collected by ground monitoring stations through a fusion model of weighted average method and Kalman filter model.

[0009] Preferably, when performing data fusion on satellite remote sensing data and ground monitoring site data by using a fusion model of a weighted average method and a Kalman filter model, the following steps are specifically included:

[0010] S11: Use weighted average method to perform preliminary fusion of satellite remote sensing data and ground monitoring station data;

[0011] S12: The preliminary fusion results are then input into the Kalman filter model for optimized fusion;

[0012] S13: Optimize the fusion result through the prediction step and the update step.

[0013] Preferably, the weighted average method in S11 specifically includes the following:

[0014] Assume that the ground monitoring data is x, the weight is w, the satellite remote sensing data is y, the weight is v, and w + v = 1, then the calculation formula of the preliminary fusion result Z0 is:

[0015] Z0=w×x+v×y.

[0016] Preferably, the optimization fusion of the Kalman filter model in S12 specifically includes the following:

[0017] Assume that the state equation of the system is:

[0018] X k =A k X k-1 +W k-1 ;

[0019] Among them, X k is the system state at time k;

[0020] A k is the state transition matrix;

[0021] W k-1 is the process noise, and W k-1 ~N(0,Q k-1 ),Q k-1 is the process noise covariance;

[0022] The observation equation is:

[0023] Z k =H k X k +V k ;

[0024] Among them, Z k is the observed value at time k;

[0025] H k is the observation matrix;

[0026] V k is the observation noise, and V k ~N(0,R k),R k is the observation noise covariance.

[0027] Preferably, in S13, the optimization fusion results of the prediction step and the update step specifically include the following:

[0028] The prediction step involves the a priori state estimation equation:

[0029]

[0030] And the prior estimate error covariance:

[0031]

[0032] in, is the prior state estimate at time k;

[0033] is the prior estimate error covariance;

[0034] is the posterior state estimate at time k-1;

[0035] P k-1 is the posterior estimation error covariance at time k-1;

[0036] The update steps are as follows:

[0037] Kalman gain:

[0038]

[0039] The posterior state estimation equation is:

[0040]

[0041] Posterior estimation error covariance:

[0042] P k =(IK k H k )P k - ;

[0043] Among them, K k is the Kalman gain;

[0044] is the posterior state estimate at time k;

[0045] P k is the posterior estimation error covariance at time k;

[0046] I is the identity matrix;

[0047] In the application of the drain overflow risk assessment system, the state transfer matrix A is reasonably determined according to the specific data characteristics and problem scenarios. k , observation matrix H k , process noise covariance Q k-1 and the observation noise covariance R k At the same time, the weights w and v in the weighted average method also need to be adjusted according to the actual situation of the data.

[0048] Preferably, it also includes a risk assessment module and a fusion model based on a neural network;

[0049] The risk assessment module is used to determine the weight of each assessment indicator using the hierarchical analysis method on the data fused by the data fusion module, and calculate the comprehensive risk index using the weighted comprehensive evaluation method. After standardizing the actual values of each indicator, the module determines the risk level of the drain overflow by comparing it with the risk level classification standard. The risk levels are divided into low risk, medium risk and high risk, and corresponding risk descriptions and response suggestions are given for different risk levels.

[0050] The neural network-based fusion model is used to construct a multi-layer perceptron neural network. The ground-monitored pipe network water level flow, surrounding river water quality, regional meteorological and drainage facility data, and satellite remote sensing water body, meteorological, topographic, land use and drainage facility data are used as input layer neurons. The hidden layer neurons are connected to the input layer through weights, and feature extraction and nonlinear transformation are performed on the input data. The output layer neurons obtain the fused data based on the hidden layer output. The neural network is trained with a large amount of sample data to adjust the weights so that the fused data output by the network is close to the actual situation, thereby realizing data fusion.

[0051] Preferably, the analytic hierarchy process is used to determine the weight of each evaluation indicator, specifically including first establishing a hierarchical structure model, setting the target layer as the risk assessment of the drainage outlet overflow, the criterion layer as the category indicators such as hydrology, water quality, meteorology, and drainage facility conditions, and the indicator layer as the specific evaluation indicators. A judgment matrix is constructed through expert scoring, and then the relative weight of each indicator is calculated;

[0052] The weighted comprehensive evaluation method is used to calculate the comprehensive risk index, specifically including the following formula:

[0053]

[0054] R is the comprehensive risk index;

[0055] w i is the weight of the i-th evaluation indicator;

[0056] x iis the standardized value of the i-th evaluation indicator. The actual values of each indicator are standardized to make them within a unified and comparable range. Based on the calculated comprehensive risk index and the risk level classification standard, the overflow risk level of the drain outlet is determined.

[0057] Determine the drain overflow risk level based on the calculated comprehensive risk index and the risk level classification standards;

[0058] The comprehensive risk index is 0-30 for low risk, 31-60 for medium risk, and 60-100 for high risk. Different risk levels correspond to different risk descriptions and response suggestions.

[0059] Preferably, a fusion effect evaluation module is also included to verify the effectiveness of the fusion of satellite remote sensing data and ground monitoring site data.

[0060] Preferably, after completing the Kalman filter update step, the obtained posterior state estimate is used as the final fusion result. These fusion results are generated in sequence according to the time series, corresponding to different monitoring moments, and are sorted and saved in chronological order to form a complete fusion data sequence for subsequent analysis;

[0061] The actual observation values are obtained by collecting data in the early stage, ensuring that the measurement time of the actual observation values accurately corresponds to the time point of the fusion data. The actual observation values within the same time period are collected and the actual observation value sequence is constructed to provide comparative data for evaluating the fusion effect.

[0062] Calculate the mean square error (MSE) and mean absolute error (MAE);

[0063] The fusion effect is evaluated based on the calculated MSE and MAE values. The smaller the MSE and MAE values are, the closer the fusion result is to the actual observation value, and the better the fusion effect is. If the MSE and MAE values are within an acceptable range, it indicates that the fusion method can effectively improve the data quality. If the values are large, it is necessary to check whether there are problems in the parameter settings, data processing and other links in the fusion process. It is necessary to adjust the weights of the weighted averaging method and the model parameters of the Kalman filter, or reprocess the data.

[0064] A multi-dimensional drain overflow risk assessment method includes the following steps:

[0065] Step 1: Collect data from both ground monitoring stations and satellite remote sensing. Ground monitoring covers pipe network water level and flow, surrounding river water quality, regional meteorology, and drainage facilities. Satellite remote sensing collects data on water bodies, meteorology, topography, land use, and drainage facilities, providing comprehensive information for subsequent analysis.

[0066] Step 2: Process missing values and outliers in ground monitoring data and unify time intervals; perform radiometric correction, geometric correction, and spatial resolution unification on satellite remote sensing data to improve data quality and facilitate data fusion.

[0067] Step 3: First, the satellite remote sensing and ground monitoring data are preliminarily integrated through the weighted average method, and then the fusion results are optimized using the Kalman filter model. In practical applications, relevant parameters need to be adjusted according to the characteristics of the data to adapt to dynamic changes in the data;

[0068] Step 4: Based on the fused data, determine the evaluation indicators from multiple aspects, use the hierarchical analysis method or principal component analysis method to determine the weights, calculate the comprehensive risk index, compare with the risk level classification standards, determine the drain overflow risk level, and give corresponding response suggestions.

[0069] Beneficial technical effects of the present invention:

[0070] This invention provides a multi-dimensional drain overflow risk assessment system and method. This system integrates multi-source data from ground monitoring and satellite remote sensing to comprehensively collect information related to drain outlets. Ground monitoring data provides accurate information on local hydrology, water quality, and the operational status of drainage facilities, while satellite remote sensing data enables large-scale monitoring. The two complement each other's strengths. Data is fused through weighted averaging and Kalman filtering to improve data accuracy and reliability. The weighted averaging method assigns weights based on data characteristics for preliminary fusion, while the Kalman filtering method optimizes the results based on the state equation and observation equation, adapting to dynamic data changes. For example, water level data fusion combines the advantages of high-precision ground monitoring with the wide-area coverage of satellites to accurately reflect water level changes, laying a solid data foundation for risk assessment.

[0071] The system determines evaluation indicators from multiple dimensions such as hydrology, water quality, meteorology, and drainage facility conditions, uses the analytic hierarchy process (AHP) or principal component analysis (PCA) to determine weights, and adopts a weighted comprehensive evaluation method to calculate the risk index and divide the risk level. Compared with single indicator evaluation, the multi-dimensional evaluation comprehensively considers the factors affecting overflow, and the weight determination method makes the evaluation more scientific. For example, it comprehensively considers the difference between the water level of the pipeline network and the warning water level, the ratio of flow rate to design drainage capacity, the water quality category of surrounding rivers and streams, regional rainfall and intensity, and other indicators to accurately quantify overflow risks and provide a reliable basis for drainage management.

[0072] Accurate risk assessment provides key support for drainage management decisions. Based on the risk level, routine monitoring and maintenance can be arranged when the risk is low. When the risk is medium, monitoring can be strengthened, facilities can be inspected, and emergency plans can be formulated. When the risk is high, emergency response can be immediately initiated, drainage systems can be repaired, temporary drainage measures can be taken, and surrounding personnel can be notified. Through timely and effective decision-making, the probability and hazards of overflows can be reduced, the water quality of surrounding rivers and streams can be guaranteed, and effective support can be provided for the refined supervision of urban water bodies. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is an exploded perspective view of the overall structure of a device according to a preferred embodiment of a multi-dimensional drain overflow risk assessment system and assessment method of the present invention;

[0074] Figure 2 This is an enlarged view of the structure at point A of a preferred embodiment of a multi-dimensional drain overflow risk assessment system and assessment method according to the present invention;

[0075] Figure 3 It is a side sectional view of a first telescopic rod according to a preferred embodiment of a multi-dimensional drain overflow risk assessment system and assessment method of the present invention. DETAILED DESCRIPTION

[0076] In order to make the technical solution of the present invention more clear and specific to those skilled in the art, the present invention is further described in detail below with reference to embodiments and drawings, but the embodiments of the present invention are not limited thereto.

[0077] Collect pipeline network hydrological data, surrounding river water quality data, regional meteorological data, and drainage facility-related data through ground monitoring stations;

[0078] Pipeline network hydrological data collection includes the following:

[0079] Water level data: Water level sensors monitor the water level in the pipe network in real time to understand the rise and fall of water levels, determine whether drainage is smooth, and whether there is an overflow risk.

[0080] Flow data: Flow monitoring equipment, such as electromagnetic flow meters and ultrasonic flow meters, is used to measure the amount of water passing through the pipe network per unit time. This is crucial for evaluating the drainage capacity and actual drainage volume of the drainage system, and can help analyze the water flow status and carrying capacity of the drain outlet.

[0081] The water quality data collection of surrounding rivers and streams includes the following:

[0082] pH value: reflects the acidity and alkalinity of water. Changes in pH value can affect the corrosion of drainage system materials and the impact on the surrounding environment and ecology.

[0083] Chemical oxygen demand (COD): It can indirectly indicate the content of organic matter in water. A high COD value means that the water body is seriously polluted by organic matter and there may be overflow at the drain outlet.

[0084] Ammonia nitrogen: It is one of the important indicators to measure the degree of water pollution. Excessive ammonia nitrogen content will have a negative impact on the ecological environment of the water body and also reflect the operation of the drainage system;

[0085] Dissolved oxygen (DO): The dissolved oxygen content affects the survival status of organisms in the water body and is also related to the self-purification ability of the water body. It is of great significance for judging the quality of water and the ecological environment around the drain outlet.

[0086] Regional meteorological data collection includes the following:

[0087] Rainfall: Combined with the surrounding rainwater and sewage diversion effects, rain gauges are used to measure rainfall intensity and accumulated rainfall on the ground. This, combined with the large-scale rainfall information provided by satellite remote sensing, can provide a more accurate understanding of rainfall conditions in the area where the drainage outlet is located, providing key data for analyzing the overflow risk of the drainage outlet;

[0088] Air temperature: Monitor ambient temperature. Air temperature changes will affect the physical and chemical properties of water bodies. Changes in water temperature will affect the dissolved oxygen content and also affect the form of rainfall (rain, snow, hail, etc.), thereby affecting the water flow at the drain outlet;

[0089] Wind direction and speed: Wind direction and speed data help determine the direction and speed of rainfall, as well as its impact on water flow around the drain outlet. Strong winds can cause large fluctuations on the water surface, affecting the accuracy of water level measurements and the diffusion of pollutants near the drain outlet.

[0090] The data collection related to drainage outlet facilities includes the following:

[0091] Equipment operating status: Monitor the operating status of drainage pumps and valves, including whether the equipment is operating normally and whether there are any fault alarm messages, to ensure the normal operation of drainage facilities and promptly identify and resolve drainage problems caused by equipment failures;

[0092] Pipeline pressure: The pressure sensor monitors the pressure changes in the drainage pipe. Abnormal pressure may indicate blockage, leakage or poor water flow in the pipe, which helps to promptly detect hidden dangers in the drainage system.

[0093] By collecting the above multi-dimensional data, ground monitoring stations can provide comprehensive and accurate basic data for drain overflow risk assessment.

[0094] Furthermore, the present invention uses satellite remote sensing to collect water body related data, meteorological data, topographic data, land use data and drainage facility related data;

[0095] Water body related data collection includes the following:

[0096] Water level information: Using optical satellite imagery to identify shoreline changes or using direct measurement with altimeter / lidar satellite data, the water level near the outlet is obtained. This allows for analysis of water level fluctuation trends and determines whether the overflow threshold is approaching.

[0097] Water body area and scope: Using multispectral or hyperspectral data from optical satellites, we can clearly distinguish between water bodies and surrounding land features, determine the area of receiving water bodies around the outfall and the scope of flooding, and provide a basis for assessing the areas that may be affected by overflow;

[0098] Water quality parameter inversion: Hyperspectral satellite data contains rich spectral information. By establishing an inversion model, some water quality parameters of the water body are estimated, such as chlorophyll content, suspended matter concentration, and chemical oxygen demand. This indirectly reflects the water pollution status and eutrophication level, and helps understand the impact of drainage on the water quality of surrounding water bodies.

[0099] Meteorological data collection includes the following:

[0100] Precipitation: Microwave radiation sensors carried by meteorological satellites can monitor rainfall intensity and distribution over a wide area. By analyzing the development and movement of rainfall clouds and the spatial and temporal distribution of rainfall, combined with calibration data from ground-based rain gauges, they provide rainfall input conditions for outfall risk assessment and predict outfall water inflow.

[0101] Cloud amount and cloud phase: Optical and microwave satellite data can obtain cloud amount information, understand the cloud coverage in the sky, and determine whether it will affect optical satellite observations of the ground. At the same time, microwave satellite data can also analyze the cloud phase, such as distinguishing between ice crystal clouds and water droplet clouds, to assist in weather forecasts and rainfall type judgments.

[0102] Topographic and geomorphic data collection includes the following:

[0103] Surface elevation: Using radar satellite interferometry technology, we obtain terrain elevation data for the drainage outlet and surrounding areas and generate a digital elevation model (DEM). The DEM can be used to analyze terrain undulations, slope, and aspect, understand the natural flow direction and confluence of water, and assess drainage patency and potential waterlogging areas under different rainfall conditions.

[0104] Landform types and characteristics: High-resolution images from optical satellites can be used to identify different landform types, such as mountains, plains, and river terraces, as well as landform features around drainage outlets, such as the presence of depressions, gullies, and other terrain prone to water accumulation, providing topographic and geomorphological background information for overflow risk analysis.

[0105] Land use data collection includes the following:

[0106] Land use type: Using multispectral imagery from optical satellites, combined with remote sensing interpretation technology, we categorize land use types into cultivated land, forest land, grassland, construction land, and water areas. We analyze the distribution and changes of different land use types and understand the urbanization process and vegetation cover around drainage outlets. Increased construction land reduces the surface infiltration capacity and increases the runoff coefficient, while vegetation cover helps intercept rainwater and delay runoff, significantly impacting the risk of overflow at drainage outlets.

[0107] Distribution of impermeable layers: Remote sensing images can be used to extract information about impermeable layers, such as the distribution range and proportion of roads and buildings. The presence of impermeable layers will cause surface runoff generated by rainfall to quickly converge into the drainage system, increasing the pressure at the drain outlet. By monitoring changes in the impermeable layers, the load and overflow risk of the drainage system can be better assessed.

[0108] The collection of drainage facility-related data includes the following:

[0109] Drain outlet location and distribution: High-resolution optical satellite imagery can identify the location and shape of drain outlets. Combined with geographic information system (GIS) technology, it can accurately locate and analyze the spatial distribution of drain outlets, providing basic data for subsequent monitoring and management.

[0110] Condition of drainage pipes and facilities: Radar satellite data can penetrate the surface to a certain extent and detect the direction and burial depth of underground drainage pipes. In addition, through long-term remote sensing monitoring of the area around the drainage outlets, it can also be found whether the drainage facilities are damaged or silted up, and potential overflow risks can be discovered in a timely manner.

[0111] Perform data preprocessing on satellite remote sensing data and ground monitoring station data;

[0112] Preprocessing of ground monitoring data: There are missing values and outliers in ground monitoring data. For missing values, mean interpolation, linear interpolation and other methods can be used to supplement them. For outliers, they are judged based on historical data and the normal measurement range of monitoring equipment and removed by statistical filtering. For example, for water level data, if the measured value at a certain moment is much higher than the highest water level in the same period in history and is greatly different from the surrounding data, it can be identified as an outlier and processed. At the same time, the time interval of ground monitoring data is unified, and data collected at different frequencies are resampled to the same frequency to ensure time matching with satellite remote sensing data.

[0113] Satellite remote sensing data preprocessing: Perform radiation correction on satellite remote sensing images. Based on the satellite sensor calibration parameters and atmospheric transmission model, the digital quantization values recorded by the sensor are converted into the true radiation brightness values of the surface. Geometric correction is performed. Ground control points and geometric transformation models are used to eliminate the geometric deformation of the image so that the position of the objects in the image is consistent with the actual geographical location. In addition, the spatial resolution of data from different satellite remote sensing data sources is unified, and the resampling method is used to bring the low-resolution data up to the same level as the high-resolution data for easy fusion.

[0114] Use the weighted average method to perform a preliminary fusion of satellite remote sensing data and ground monitoring station data to obtain a preliminary fusion result, which will be used as the initial state estimate of the Kalman filter method;

[0115] Assume that the ground monitoring data is x, the weight is w, the satellite remote sensing data is y, the weight is v, and w + v = 1, then the calculation formula of the preliminary fusion result Z0 is: Z0 = w × x + v × y;

[0116] It should be noted that the acquisition of ground monitoring data X covers multiple types such as hydrology, water quality, meteorology, and drainage facility-related data. Different types of data are acquired in different ways, involving a variety of professional equipment and technical means;

[0117] Satellite remote sensing data can be obtained through sensors carried by different types of satellites using corresponding technical principles, covering data on water bodies, meteorology, topography, land use, drainage facilities, etc.

[0118] The preliminary fusion results are input into the Kalman filter model, and the fusion results are continuously optimized through the prediction and update steps of the Kalman filter so that it can better track the dynamic changes of the data;

[0119] Assume that the state equation of the system is: k =A k X k-1 +W k-1 ;

[0120] Among them, X k is the system state at time k (i.e. the physical quantity to be estimated after fusion), A k is the state transition matrix, W k-1 is the process noise, and W k-1 ~N(0,Q k-1 ),Q k-1 is the process noise covariance;

[0121] The observation equation is: Z k =H k X k +V k ;

[0122] Among them, Z k is the observation value at time k (including the preliminary fusion results of ground monitoring data and satellite remote sensing data), H k is the observation matrix, V k is the observation noise, and V k ~N(0,R k ),R k is the observation noise covariance;

[0123] Prediction Step

[0124] Prior state estimate:

[0125] A priori estimate of the error covariance:

[0126] in, is the prior state estimate at time k, is the prior estimate error covariance, is the posterior state estimate at time k-1, P k-1 is the posterior estimation error covariance at time k-1.

[0127] The update steps are as follows:

[0128] Kalman gain:

[0129] Posterior state estimate:

[0130] Posterior estimation error covariance:

[0131] Among them, K k is the Kalman gain, is the posterior state estimate at time k (i.e. the final fusion result), P k is the posterior estimation error covariance at time k, and I is the identity matrix;

[0132] In practical applications, it is necessary to reasonably determine the state transfer matrix A according to the specific data characteristics and problem scenarios. k , observation matrix H k , process noise covariance Q k-1 and the observation noise covariance R k At the same time, the weights w and v in the weighted average method also need to be adjusted according to the actual situation of the data.

[0133] The simplicity and intuitiveness of the weighted average method and the precise dynamic processing of the Kalman filter method complement each other, enhancing the reliability of the entire data fusion system. When some data are missing or abnormal, the weighted average method can use the weight distribution of known data to provide a relatively reasonable preliminary result. The Kalman filter method can use its own filtering mechanism to correct and process abnormal data to a certain extent, reducing the impact of abnormal data on the final fusion result. When a water level sensor at a ground monitoring station has a temporary failure and causes data abnormality, the fusion method can still rely on other data and filtering mechanisms to provide a more reliable fusion result, ensuring the stable operation of the drain overflow risk assessment system.

[0134] Output the final fusion result and evaluate the fusion effect by comparing the errors between the data before and after fusion and the actual observation value, such as mean square error (MSE) and mean absolute error (MAE). The specific evaluation includes the following:

[0135] After completing the Kalman filter update step, the posterior state estimate is obtained This is the final fusion result. These fusion results will be generated in sequence according to the time series, corresponding to different monitoring moments. These fusion results will be sorted and saved in chronological order to form a complete fusion data sequence.

[0136] The actual observation values are provided by the above-mentioned collected data. For example, for the water level data of the pipe network, they are obtained by real-time measurement on site through a high-precision water level meter. This ensures that the measurement time of the actual observation value corresponds exactly to the time point of the fused data to ensure the accuracy of the comparison. The actual observation values within the same time period as the fused data are collected to form a sequence of actual observation values.

[0137] The mean square error is used to measure the average value of the square of the error between the fusion result and the actual observation value, which can reflect the degree of discreteness of the data. The calculation formula is: Where n is the number of data samples, is the i-th fused data point, X i is the i-th actual observation data point. During the calculation, the corresponding data in the fused data sequence and the actual observation value sequence are substituted into the formula one by one to calculate the square of the difference, and then all square values are summed and averaged.

[0138] The mean absolute error directly calculates the average value of the absolute value of the error between the fusion result and the actual observation value, which more intuitively reflects the average size of the error. The calculation formula is: The calculation process is similar to MSE, except that the absolute value of the difference is taken and then the sum is averaged.

[0139] The fusion effect is evaluated based on the calculated MSE and MAE values. The smaller the MSE and MAE values, the closer the fusion result is to the actual observation value, and the better the fusion effect. If the MSE and MAE values are within an acceptable range, it indicates that the fusion method can effectively improve data quality. If the values are large, it is necessary to check whether there are problems in the parameter settings and data processing during the fusion process. It is necessary to adjust the weights of the weighted averaging method and the model parameters of the Kalman filter, or reprocess the data. Comparing the MSE and MAE values when using a single weighted averaging method or the Kalman filter method intuitively demonstrates the advantages of the fusion method in reducing errors.

[0140] Establish a fusion model;

[0141] Fusion model based on neural network: A multi-layer perceptron (MLP) neural network is constructed, and the ground-monitored pipe network water level flow, surrounding river water quality, regional meteorological and drainage facility data and satellite remote sensing data on water bodies, meteorology, topography, land use and drainage facilities are used as input layer neurons.

[0142] The neurons in the hidden layer are connected to the input layer through weights to perform feature extraction and nonlinear transformation on the input data.

[0143] The output layer neurons obtain the fused data based on the hidden layer output.

[0144] By training the neural network with a large amount of sample data and adjusting the weights, the fused data output by the network is made closest to the actual situation, thus achieving data fusion.

[0145] For example, when predicting the overflow risk of a drain outlet, the fused data is used as input, processed by a trained neural network, and the overflow risk level is output.

[0146] The establishment of specific overflow risk levels includes the following options:

[0147] Combining ground monitoring and satellite remote sensing data, evaluation indicators are determined from multiple aspects such as hydrology, water quality, meteorology, and drainage facility conditions. In terms of water level data, the difference between the water level in the pipeline network and the warning water level is considered. The smaller the difference, the higher the overflow risk; in terms of water quality data, the difference between the water quality of surrounding rivers and streams and the water quality target is considered. The larger the difference, the higher the overflow risk; flow data focuses on the ratio of the actual flow of the drainage system to the designed drainage capacity. The larger the ratio, the greater the risk. Among meteorological factors, rainfall, its intensity and continuous rainfall time are critical. Short-term heavy rainfall or long-lasting rainfall will increase the overflow risk. The condition of drainage facilities involves the operating efficiency of the drainage pump. Low operating efficiency will lead to poor drainage. The degree of pipeline blockage will also affect the drainage speed. The more serious the blockage, the higher the risk.

[0148] The analytic hierarchy process (AHP) is used to determine the weight of each evaluation indicator. First, a hierarchical structure model is established, with the target layer set as the drainage outlet overflow risk assessment, the criterion layer as category indicators such as hydrology, water quality, meteorology, and drainage facility conditions, and the indicator layer as specific evaluation indicators. A judgment matrix is constructed through expert scoring to calculate the relative weight of each indicator. The principal component analysis (PCA) method can also be used to convert multiple related evaluation indicators into a few unrelated principal components. The weight of each indicator is determined based on the variance contribution rate of the principal component. The larger the variance contribution rate, the more important the indicator represented by the principal component in the risk assessment.

[0149] Based on the value range of the evaluation indicators and actual experience, the risk levels are divided as shown in Table 1;

[0150]

[0151] Table 1 covers the key evaluation indicators in the risk assessment of drain overflow, and divides the risk into three levels: low, medium and high based on practical experience and relevant research.

[0152] According to the determined evaluation index weights, the weighted comprehensive evaluation method is used to calculate the comprehensive risk index. The formula is: R is the comprehensive risk index, w i is the weight of the i-th evaluation indicator, x i is the standardized value of the i-th evaluation indicator. The actual values of each indicator are standardized to make them within a unified and comparable range. According to the calculated comprehensive risk index and the risk level classification standard, the overflow risk level of the drain outlet is determined.

[0153] As shown in Table 2;

[0154]

[0155]

[0156] The above is only a further embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes based on the technical solutions and concepts of the present invention within the scope disclosed by the present invention, which fall within the scope of protection of the present invention.

Claims

1. A multi-dimensional drain overflow risk assessment system, including a data acquisition module for collecting multiple types of data from ground monitoring stations and satellite remote sensing; Data preprocessing module, used for processing ground monitoring data and satellite remote sensing data; Its characteristics are: The data fusion module fuses the data collected by satellite remote sensing and the data collected by ground monitoring stations through a fusion model of weighted average method and Kalman filter model.

2. A multi-dimensional drain overflow risk assessment system according to claim 1, characterized in that: When the satellite remote sensing data and the ground monitoring station data are fused by the fusion model of the weighted average method and the Kalman filter model, the specific steps are as follows: S11: Use weighted average method to perform preliminary fusion of satellite remote sensing data and ground monitoring station data; S12: The preliminary fusion results are then input into the Kalman filter model for optimized fusion; S13: Optimize the fusion result through the prediction step and the update step.

3. The multi-dimensional drain overflow risk assessment system according to claim 2, characterized in that: The weighted average method in S11 specifically includes the following: Assume that the ground monitoring data is x, the weight is w, the satellite remote sensing data is y, the weight is v, and w + v = 1, then the calculation formula of the preliminary fusion result Z0 is: Z0=w×x+v×y.

4. The multi-dimensional drain overflow risk assessment system according to claim 3, characterized in that: The optimization and fusion of the Kalman filter model in S12 specifically includes the following: Assume that the state equation of the system is: X k =A k X k-1 +W k-1 ; Among them, X k is the system state at time k; A k is the state transition matrix; W k-1 is the process noise, and W k-1 ~N(0,Q k-1 ),Q k-1 is the process noise covariance; The observation equation is: Z k =H k X k +V k ; Among them, Z k is the observed value at time k; H k is the observation matrix; V k is the observation noise, and V k ~N(0,R k ),R k is the observation noise covariance.

5. The multi-dimensional drain overflow risk assessment system according to claim 4, characterized in that: In S13, the optimization fusion results of the prediction step and the update step are specifically as follows: The prediction step involves the a priori state estimation equation: And the prior estimate error covariance: in, is the prior state estimate at time k; is the prior estimate error covariance; is the posterior state estimate at time k-1; P k-1 is the posterior estimation error covariance at time k-1; The update steps are as follows: Kalman gain: The posterior state estimation equation is: Posterior estimation error covariance: Among them, K k is the Kalman gain; is the posterior state estimate at time k; P k is the posterior estimation error covariance at time k; I is the identity matrix; In the application of the drain overflow risk assessment system, the state transfer matrix A is reasonably determined according to the specific data characteristics and problem scenarios. k , observation matrix H k , process noise covariance Q k-1 and the observation noise covariance R k At the same time, the weights w and v in the weighted average method also need to be adjusted according to the actual situation of the data.

6. The multi-dimensional drain overflow risk assessment system according to claim 1, characterized in that: It also includes a risk assessment module and a neural network-based fusion model; The risk assessment module is used to determine the weight of each assessment indicator using the hierarchical analysis method on the data fused by the data fusion module, and calculate the comprehensive risk index using the weighted comprehensive evaluation method. After standardizing the actual values of each indicator, the module determines the risk level of the drain overflow by comparing it with the risk level classification standard. The risk levels are divided into low risk, medium risk and high risk, and corresponding risk descriptions and response suggestions are given for different risk levels. The neural network-based fusion model is used to construct a multi-layer perceptron neural network. The ground-monitored pipe network water level flow, surrounding river water quality, regional meteorological and drainage facility data, and satellite remote sensing water body, meteorological, topographic, land use and drainage facility data are used as input layer neurons. The hidden layer neurons are connected to the input layer through weights, and feature extraction and nonlinear transformation are performed on the input data. The output layer neurons obtain the fused data based on the hidden layer output. The neural network is trained with a large amount of sample data to adjust the weights so that the fused data output by the network is close to the actual situation, thereby realizing data fusion.

7. The multi-dimensional drain overflow risk assessment system according to claim 6, characterized in that: The analytic hierarchy process (AHP) method for determining the weights of various evaluation indicators involves first establishing a hierarchical structure model, with the target layer set as the risk assessment of drainage outlet overflow, the criterion layer divided into categories such as hydrology, meteorology, and drainage facility conditions, and the indicator layer as specific evaluation indicators. A judgment matrix is constructed through expert scoring, and the relative weights of each indicator are then calculated. The weighted comprehensive evaluation method is used to calculate the comprehensive risk index, specifically including the following formula: R is the comprehensive risk index; w i is the weight of the i-th evaluation indicator; x i is the standardized value of the i-th evaluation indicator. The actual values of each indicator are standardized to make them within a unified and comparable range. Based on the calculated comprehensive risk index and the risk level classification standard, the overflow risk level of the drain outlet is determined. Determine the drain overflow risk level based on the calculated comprehensive risk index and the risk level classification standards; The comprehensive risk index is 0-30 for low risk, 31-60 for medium risk, and 60-100 for high risk. Different risk levels correspond to different risk descriptions and response suggestions.

8. The multi-dimensional drain overflow risk assessment system according to claim 2, characterized in that: It also includes a fusion effect evaluation module to verify the effectiveness of the fusion of satellite remote sensing data and ground monitoring station data.

9. The multi-dimensional drain overflow risk assessment system according to claim 8, characterized in that: After completing the Kalman filter update step, the obtained posterior state estimate is used as the final fusion result. These fusion results are generated in sequence according to the time series, corresponding to different monitoring moments, and are sorted and saved in chronological order to form a complete fusion data sequence for subsequent analysis. The actual observation values are obtained by collecting data in the early stage, ensuring that the measurement time of the actual observation values accurately corresponds to the time point of the fusion data. The actual observation values within the same time period are collected and the actual observation value sequence is constructed to provide comparative data for evaluating the fusion effect. Calculate the mean square error (MSE) and mean absolute error (MAE); The fusion effect is evaluated based on the calculated MSE and MAE values. The smaller the MSE and MAE values are, the closer the fusion result is to the actual observation value, and the better the fusion effect is. If the MSE and MAE values are within an acceptable range, it indicates that the fusion method can effectively improve the data quality. If the values are large, it is necessary to check whether there are problems in the parameter settings, data processing and other links in the fusion process. It is necessary to adjust the weights of the weighted averaging method and the model parameters of the Kalman filter, or reprocess the data.

10. A multi-dimensional drain overflow risk assessment method, based on the multi-dimensional drain overflow risk assessment system according to any one of claims 1 to 9, characterized in that: The steps include: Step 1: Collect data from both ground monitoring stations and satellite remote sensing. Ground monitoring covers pipe network water level and flow, surrounding river water quality, regional meteorology, and drainage facilities. Satellite remote sensing collects data on water bodies, meteorology, topography, land use, and drainage facilities, providing comprehensive information for subsequent analysis. Step 2: Process missing values and outliers in ground monitoring data and unify time intervals; perform radiometric correction, geometric correction, and spatial resolution unification on satellite remote sensing data to improve data quality and facilitate data fusion. Step 3: First, the satellite remote sensing and ground monitoring data are preliminarily integrated through the weighted average method, and then the fusion results are optimized using the Kalman filter model. In practical applications, relevant parameters need to be adjusted according to the characteristics of the data to adapt to dynamic changes in the data; Step 4: Based on the fused data, determine the evaluation indicators from multiple aspects, use the hierarchical analysis method or principal component analysis method to determine the weights, calculate the comprehensive risk index, compare with the risk level classification standards, determine the drain overflow risk level, and give corresponding response suggestions.

Images