A method and system for assessing flood disaster risk of power facilities considering flood impact

Abstract

The present invention provides a method and system for assessing the risk of flood disasters of electric power facilities taking into account the influence of floods, and relates to the technical field of risk assessment of flood disasters of electric power facilities. The present invention collects pole height, rainfall, topography and soil data in real time, obtains historical mountain torrent influencing factors and height data, pre-processes them, and then constructs a neural network model, uses the processed rainfall, topography and soil data as training sets, and mountain torrent height as a label for training, and then inputs the mountain torrent influencing factors collected in real time into the trained model, predicts the actual mountain torrent height, and compares it with the pole height, and finally judges the safety of the electric power facility according to the comparison result, determines the flood disaster risk level and the overall risk index, thereby providing a quantitative basis for decision-making and improving the flood resistance of the electric power facility.

Description

Technical Field

[0001] The present invention relates to the technical field of flood disasters in electric power facilities, and in particular to a method and system for assessing flood disaster risks in electric power facilities taking flood influence into consideration. Background Art

[0002] Frequent floods pose a severe challenge to the safety of power facilities. Flood disasters not only directly affect the operation of power facilities, but may also cause large-scale power outages, equipment damage and safety accidents, which in turn have a serious impact on social and economic activities and residents' lives. Therefore, it is particularly important to carry out flood disaster risk assessment of power facilities affected by floods. Through scientific and systematic risk assessment methods, power operators can identify and quantify potential risks, formulate effective early warning mechanisms and emergency response measures, and thus ensure the safety and stability of power supply.

[0003] At present, existing technologies have blind spots in analyzing the correlation between flood height and the safety of power facilities, and are unable to fully quantify the specific impact of different risk levels on power facilities, thus affecting the formulation of response strategies. These shortcomings have caused many power facilities to lack sufficient resilience and adaptability in the face of extreme weather. Existing technologies have limitations in dynamic risk assessment and real-time prediction. Many methods rely on historical data and cannot effectively respond to new challenges brought about by climate change. There is also insufficient analysis of the direct correlation between future flood heights and the safety of distribution facilities, which makes it difficult for power facilities to make timely and accurate adjustments when dealing with future uncertainties.

[0004] The present invention effectively solves the shortcomings of traditional technologies in flood risk management by establishing a flood height prediction model and a risk level assessment system, and can provide a scientific and quantitative basis for decision-making. Through intelligent analysis based on real-time data, it can realize timely assessment of potential flood risks of power facilities, ensuring that relevant departments can take preventive measures in advance to reduce the damage caused by mountain torrents to power facilities. Secondly, the introduction of a comprehensive risk index makes risk assessment more quantitative and operational, providing a clear basis for decision makers to reasonably allocate and dispatch resources at critical moments and ensure the safety of power supply. Therefore, the present invention provides a strong technical guarantee for the safe operation of the power system and helps to enhance the overall ability of society to fight natural disasters.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for assessing the risk of flood disasters in power facilities taking into account the impact of floods, so as to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A flood disaster risk assessment method for power facilities considering flood impacts, comprising the following specific steps:

[0009] Step 1: Collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data;

[0010] Step 2: Build a neural network model, take the rainfall data, topography and soil data in the preprocessed historical data as the training set, and the corresponding flash flood height data as the label, and input it into the neural network model to train the neural network model;

[0011] Step 3: Input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data;

[0012] Step 4: Based on the comparison results, the flood disaster risk level is obtained, the ability of the pole height data of the power facilities to cope with the predicted actual flash flood height data is determined, and the overall risk index of the power facilities is calculated, and an assessment is conducted based on the risk index.

[0013] Furthermore, the pole height data includes the heights of all poles in the monitored area and the lowest height of the poles in the area. The step of preprocessing the pole height data is to calculate the average height of all pole height data in the monitored area, based on the formula:

[0014] ;

[0015] in, represents the average height of the poles, represents the height of the i-th pole, Refers to the number of poles within the monitoring area;

[0016] Get the minimum height of all poles in the area to be monitored , the formula based on is:

[0017] ;

[0018] The rainfall data is the total rainfall in the past K days in the monitored area. The steps of preprocessing the rainfall data include:

[0019] The total rainfall for the past K days is calculated based on the following formula:

[0020] ;

[0021] in, is the total rainfall, which is the sum of all rainfall that occurred in the past K days, is the rainfall on the qth day;

[0022] The acquired topographic data in the monitored area include the elevation data of the highest point in the monitored area, the elevation data of the lowest point and the horizontal distance between the highest point and the lowest point, and the step of preprocessing the topographic data includes calculating the slope value by calculating the height difference and the horizontal distance between the highest point and the lowest point, and the formula based on which is:

[0023] ;

[0024] in, Indicates the slope value, Represents the elevation difference between two points, that is , and Respectively represent the elevation data of the highest point and the lowest point in the area to be monitored. Indicates the horizontal distance between two points;

[0025] When obtaining soil data in the area to be monitored, soil samples are collected from the lowest point in the area to be monitored one day before the assessment, and the water volume and soil volume in the soil samples are obtained by high-temperature drying to constant weight. The step of preprocessing the soil data is to calculate the water absorption rate of the soil, and the water absorption rate of the soil is used as the soil data in the area to be monitored. The formula for calculating the water absorption rate is:

[0026] ;

[0027] in, Indicates water absorption (%), is the volume of water in the soil, Represents the volume of soil.

[0028] Furthermore, the rainfall data in the historical data is the total rainfall in the K-day period before the flood, and is calibrated as The soil data in the historical data is the soil sample collected at the lowest point in the monitored area one day before the flood, and the water absorption rate of the soil is calculated as the soil data and calibrated as The topographic data in the historical data are the elevation data of the highest point and the lowest point in the monitored area, as well as the horizontal distance between them. The terrain slope data is calculated as the topographic data and calibrated as .

[0029] Furthermore, the information on factors affecting flash floods in the monitored area in historical data, including rainfall data, topography and soil data, is used as a training set. :

[0030] ;

[0031] in, , and They represent rainfall data, topographic data and soil data in historical data respectively;

[0032] Tagged with :

[0033] ;

[0034] in, Represents the mapping relationship, Represents historical flash flood height data;

[0035] Feature data and label data Pass it as input and output to the neural network model for training, and define the number of input nodes in the input layer as , which represents the rainfall ,slope and soil water absorption ; Number of hidden layer nodes , using the ReLU activation function: , introducing a nonlinear relationship:

[0036] ;

[0037] in, is the output of the jth hidden node, is the weight matrix, connecting the input layer and the hidden layer, is the bias term;

[0038] Output of hidden layer By weight Transformation, and finally get the prediction output of the model:

[0039] ;

[0040] in, is the prediction result of the model, that is, the predicted height of the flash flood, is the weight matrix from the hidden layer to the output layer, mapping the features of the hidden layer to the final output. is the bias term of the output layer;

[0041] Model training, by calculating the square difference between all predicted values ​​and true values, the loss function reflects the prediction accuracy of the model:

[0042] ;

[0043] in, Represents the loss value, measuring the model prediction value With the actual label The gap between Represents the number of training samples, used to calculate the average loss, represents the predicted value of the oth sample, Represents the historical flash flood height of the oth sample.

[0044] Furthermore, by minimizing the loss function To adjust the parameters of the model, namely weights and biases:

[0045] ;

[0046] This formula is the loss function over weights The gradient of To reduce the loss function , It represents the output of the first hidden layer for the oth sample. Represents the sample number in the training set;

[0047] Through the loss function about the output gradient Can be further propagated to the hidden layer to calculate the weight The gradient is:

[0048] ;

[0049] in, is the derivative of the ReLU function, indicating the sensitivity of the hidden layer nodes to the input;

[0050] The logic of updating weight data is to overwrite the original weight with the updated weight. and , the weight update formula is:

[0051] ;

[0052] ;

[0053] in, is the learning rate, which controls the magnitude of each weight update. Indicates the update symbol of the weight;

[0054] use Indicators to evaluate model performance, adjust and optimize the model:

[0055] ;

[0056] in, It is used to measure the difference between the predicted value and the actual value. The smaller the value, the more accurate the prediction.

[0057] Furthermore, the trained model is used for prediction, and the real-time collected data on flash flood influencing factors are input:

[0058] ;

[0059] in, Represents characteristic input data, including data on factors affecting flash floods: rainfall data , terrain slope and soil water absorption ;

[0060] The preprocessed data is input into the neural network for forward propagation to obtain the predicted value of the flash flood height, and the input layer is transferred to the hidden layer through the weight matrix and bias The characteristics Transformed into the output of the hidden layer :

[0061] ;

[0062] in, is the output of the jth hidden node;

[0063] The output of the hidden layer Convert to predicted value :

[0064] ;

[0065] in, Indicates the actual predicted flash flood height;

[0066] Compare the predicted actual flash flood height with the height of the power poles to assess the potential impact of flash floods on power facilities:

[0067] ;

[0068] in, The actual flash flood height data predicted is obtained through model calculation. is the average height of the poles of power facilities, Represents the difference between the predicted actual flash flood height and the height of the pole.

[0069] Furthermore, based on the comparison between the actual mountain torrent height and the pole height, the risk level of flood disasters is defined:

[0070] ;

[0071] Among them, when , indicating high risk, indicating that the height of the mountain torrent is higher than the height of the pole, which will cause damage to the facilities; when , indicating low risk, the pole is safe in flash flood conditions. , indicating medium risk, the pole is in a dangerous state and requires early warning;

[0072] Calculate a composite risk index , comprehensively assess the overall risk of power facilities:

[0073] ;

[0074] in, is the frequency of flood events, defined as the number of flood events in the past 10 years, It is the vulnerability assessment of infrastructure, with a value range of 1 to 5, where 1 indicates low vulnerability and 5 indicates high vulnerability;

[0075] when ,Right now It means that the height and minimum height of the pole can cope with flooding, and the risk is zero; ,Right now , the calculation here reflects the risk when the flood height exceeds the height of the pole, when ,Right now , indicating that the flood is above the lowest pole height but below the average height, and an appropriate risk weighting (multiplied by 0.5) is given.

[0076] The present invention further provides a system for assessing the risk of flood disasters for power facilities taking into account the impact of floods. The system is used to execute the above-mentioned method for assessing the risk of flood disasters for power facilities taking into account the impact of floods, and comprises:

[0077] The data acquisition module is used to collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data;

[0078] The model building module is used to build a neural network model. The rainfall data, topography and soil data in the pre-processed historical data are used as training sets, and the corresponding mountain torrent height data are used as labels, which are input into the neural network model to train the neural network model.

[0079] The flash flood prediction module is used to input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data;

[0080] The risk assessment module is used to derive the flood disaster risk level based on the comparison results, determine the ability of the pole height data of the power facilities to cope with the predicted actual mountain torrent height data, calculate the overall risk index of the power facilities, and conduct assessment based on the risk index.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] The present invention collects the height of power facility poles and the factors affecting flash floods in the detection area in real time, and pre-processes the data in combination with historical flash flood influencing factors and height data, thereby solving the problems of single data source and poor real-time performance of traditional methods. The implementation of this process ensures that the model can be analyzed based on the latest environmental data, and the neural network model is constructed using the pre-processed historical flash flood influencing factor data. The introduction of the neural network enables complex input features to reflect the impact on the height of flash floods with higher sensitivity and accuracy, thereby improving the accuracy of the prediction.

[0083] After completing the model training, the present invention inputs the real-time collected data of flash flood influencing factors into the trained neural network to predict the actual flash flood height and compare it with the height of the electric pole. The comparison result is used to judge the safety of power facilities and calculate the risk level and overall risk index of flood disasters. Through this comprehensive process, the system can not only dynamically monitor and evaluate flood risks, but also provide a quantitative basis for decision-making, enhance the flood resistance of power facilities, and reduce potential losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0085] Figure 2 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION

[0086] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0087] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0088] Example:

[0089] See also Figure 1 , the present invention provides a technical solution:

[0090] A flood disaster risk assessment method for power facilities considering flood impacts, comprising the following specific steps:

[0091] Step 1: Collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data;

[0092] The pole height data includes the heights of all poles in the monitored area and the lowest height of poles in the area. The step of preprocessing the pole height data is to calculate the average height of all pole height data in the monitored area, based on the formula:

[0093] ;

[0094] in, represents the average height of the poles, represents the height of the i-th pole, Refers to the number of poles within the monitoring area;

[0095] average height It reflects the overall height of power facilities. The height of the poles may affect the safety of power lines, the visibility of the area, and the landscape impact. Similarly, higher poles can be free from the impact of flood disasters.

[0096] Get the minimum height of all poles in the area to be monitored , the formula based on is:

[0097] ;

[0098] in, It represents the lowest pole height among all poles in the entire monitoring area. This is an important indicator for evaluating the minimum bearing capacity of power facilities in flood conditions. This formula obtains the lowest pole height in the monitoring area by taking the minimum value of all pole heights. , which is important for assessing potential flood impacts;

[0099] If the height of individual poles Lower, will decrease accordingly; therefore, the independent variable (the height of individual poles) is negatively correlated with the dependent variable (the lowest pole height);

[0100] The rainfall data is the total rainfall in the past K days in the monitored area. The steps of preprocessing the rainfall data include:

[0101] The total rainfall for the past K days is calculated based on the following formula:

[0102] ;

[0103] in, is the total rainfall, which is the sum of all rainfall that occurred in the past K days, is the rainfall on the qth day;

[0104] Total rainfall on day q Indicates the intensity and duration of rainfall on a specific day. Knowing the total rainfall in the past K days is crucial for flood warning, agricultural irrigation, and water resources management. , meaning more frequent or heavier rainfall, which could lead to problems such as soil saturation and increased flood risk;

[0105] The acquired topographic data in the monitored area include the elevation data of the highest point in the monitored area, the elevation data of the lowest point and the horizontal distance between the highest point and the lowest point, and the step of preprocessing the topographic data includes calculating the slope value by calculating the height difference and the horizontal distance between the highest point and the lowest point, and the formula based on which is:

[0106] ;

[0107] in, Indicates the slope value, Represents the elevation difference between two points, that is , and Respectively represent the elevation data of the highest point and the lowest point in the area to be monitored. Represents the horizontal distance between two points, which is calculated using the Euclidean distance formula:

[0108] ;

[0109] in, and are the coordinates of the two collection points;

[0110] slope It reflects the steepness of the terrain. The greater the slope, the more obvious the ground inclination, the faster the water flow speed is usually, and the flow direction may be more direct. The slope affects the gathering and convergence of water flow, and has an important impact on flood formation and soil erosion. A larger slope may cause the water flow to accelerate, increasing the flood risk to downstream areas.

[0111] When obtaining soil data in the area to be monitored, soil samples are collected from the lowest point in the area to be monitored one day before the assessment, and the water volume and soil volume in the soil samples are obtained by high-temperature drying to constant weight. The step of preprocessing the soil data is to calculate the water absorption rate of the soil, and the water absorption rate of the soil is used as the soil data in the area to be monitored. The formula for calculating the water absorption rate is:

[0112] ;

[0113] in, Indicates water absorption (%), is the volume of water in the soil, represents the volume of soil;

[0114] Collecting soil samples the day before means sampling the lowest point within the monitoring area the day before the day of soil analysis. This means that researchers base their assessment of soil properties on the day before or their impact on flood risk based on the soil moisture status of the day before. This process ensures that soil samples reflect the actual moisture status after a rainfall event, thus providing more accurate data for analysis and decision-making;

[0115] Take a soil sample from the area to be monitored, usually using a soil drill or soil sampler, place the sample in an oven, dry it at 105 °C to constant weight, and determine the mass of the dry soil. , then calculate the moisture content using the following formula :

[0116] ;

[0117] in, is the mass of wet soil;

[0118] Volume of water in soil The soil moisture content can be obtained:

[0119] ;

[0120] in, Indicates the density of water (usually 1g / cm³);

[0121] Use a sampler of known volume (such as a soil cylinder) to measure the volume of the soil directly when sampling. If the sample is of regular shape, the volume can be calculated using geometric formulas. For example, for a cylindrical soil sample, the following formula can be used:

[0122] ;

[0123] in, is the radius of the cylinder, is the height;

[0124] Water absorption Indicates how much water the soil can absorb during rainfall, reflecting the soil's water retention capacity and flood risk. A higher water absorption rate means that the soil can absorb rainfall more effectively, thereby slowing down the formation of floods. The higher the water absorption rate, the stronger the soil's water retention capacity.

[0125] The rainfall data in the historical data is the total rainfall in the K-day period before the flood, and is calibrated as The soil data in the historical data is the soil sample collected at the lowest point in the monitored area one day before the flood, and the water absorption rate of the soil is calculated as the soil data and calibrated as The topographic data in the historical data are the elevation data of the highest point and the lowest point in the monitored area, as well as the horizontal distance between them. The terrain slope data is calculated as the topographic data and calibrated as ;

[0126] Precipitation is one of the main factors leading to floods. Knowing the rainfall over the past K days can help researchers analyze the intensity and duration of rainfall, thereby evaluating its impact on soil saturation. By analyzing the relationship between historical rainfall and flood outbreaks, potential early warning signals can be identified, thereby improving the effectiveness of flood warnings. Collecting rainfall data before a flood can reflect the environmental conditions before the event, which helps determine the triggering mechanism of the flood.

[0127] The water absorption rate of soil directly affects its ability to respond to rainfall. A higher water absorption rate means that the soil is close to saturation. Further rainfall may lead to rapid water loss and surface runoff, increasing the risk of flooding. Soil data from the previous day can reflect the impact of recent rainfall on soil moisture and provide instant information for understanding the upcoming flood. Collecting soil samples one day before the flood can ensure the timeliness and relevance of the data, making it possible to more accurately assess the possible response of the soil to the upcoming rainfall.

[0128] Slope affects the speed and direction of water flow. Steep slopes may cause faster water convergence, thereby accelerating the formation of floods. Understanding slope characteristics helps predict the propagation path and intensity of flash floods. Terrains with different slopes have a significant impact on the gathering and dispersion of water after rainfall. By analyzing slope data, possible water convergence areas can be identified. Collecting terrain data before a flood breaks out can provide a stable terrain background, providing a basis for understanding how rainfall triggers floods under different terrain conditions.

[0129] Step 2: Build a neural network model, take the rainfall data, topography and soil data in the preprocessed historical data as the training set, and the corresponding flash flood height data as the label, and input it into the neural network model to train the neural network model;

[0130] Use the information on flash flood influencing factors in the monitored area from historical data, including rainfall data, topography and soil data as the training set :

[0131] ;

[0132] in, , and They represent rainfall data, topographic data and soil data in historical data respectively;

[0133] The greater the rainfall, the higher the risk of flash floods. The greater the slope, the greater the speed and possibility of water convergence. Therefore, a larger slope may lead to a higher risk of flash floods. The higher the soil water absorption rate, the stronger the soil's ability to absorb precipitation, reducing the amount of water lost to the surface. Therefore, a higher water absorption rate may reduce the risk of flash floods.

[0134] Tagged with :

[0135] ;

[0136] in, Represents the mapping relationship, Indicates that the historical flash flood height data is the target variable to be predicted, reflecting the actual height of the flash flood under certain conditions, and the higher the value, the greater the severity of the flash flood;

[0137] Feature data and label data Pass it as input and output to the neural network model for training, and define the number of input nodes in the input layer as , which represents the rainfall ,slope and soil water absorption ; Number of hidden layer nodes , using the ReLU activation function: , introducing a nonlinear relationship:

[0138] ;

[0139] in, is the output of the jth hidden node. This value reflects the nonlinear mapping of the node to the weighted sum of the input features and is the basis for the model to learn the complex relationship between features. is the weight matrix, connecting the input layer and the hidden layer, responsible for adjusting the impact of the input features on the output of the hidden layer. The larger the weight, the more significant the impact of the input feature on the output of the node. is the input feature vector, It is a bias term. The role of the bias term is to allow the model to still have output when there is no input feature;

[0140] Output of hidden layer By weight Transformation, and finally get the prediction output of the model:

[0141] ;

[0142] in, is the prediction result of the model, that is, the predicted flash flood height. It is the final output, indicating the expected flash flood height given the input features. The larger the prediction value, the higher the severity of the flash flood considered by the model. is the weight matrix from the hidden layer to the output layer, which maps the features of the hidden layer to the final output and determines the importance of the hidden layer features in the final output. The bias term of the output layer allows the model to have an output level when there is no input;

[0143] Model training, by calculating the square difference between all predicted values ​​and true values, the loss function reflects the prediction accuracy of the model:

[0144] ;

[0145] in, Represents the loss value, measuring the model prediction value With the actual label The gap between them, the lower the loss value, the closer the model's prediction result is to the true value, which means the better the model's performance is. Represents the number of training samples, which is used to calculate the average loss to ensure that the loss calculation is not affected by the number of samples. represents the predicted value of the oth sample, represents the historical flood height of the oth sample. A larger difference indicates that the model performs poorly on this sample. The lower the loss value, the closer the model's prediction result is to the true value.

[0146] By minimizing the loss function To adjust the parameters of the model, namely the weights and biases, this gradient tells us how to adjust the weights to reduce the loss function and thus improve accuracy:

[0147] ;

[0148] This formula is the loss function over weights The gradient of To reduce the loss function , It represents the output of the first hidden layer for the oth sample. Here, it plays the role of transmitting the information of the previous layer, so that we can adjust the weight of the current layer according to the output of the previous layer, thereby reducing the overall loss. It is used together with the gradient of the loss function to the output of this layer to calculate the gradient of the loss function to the weight of the second layer, thereby realizing the update of the weight, with the purpose of minimizing the loss function. Represents the sample number in the training set;

[0149] Through the loss function about the output gradient Can be further propagated to the hidden layer to calculate the weight The gradient is:

[0150] ;

[0151] in, is the derivative of the ReLU function, indicating the sensitivity of the hidden layer nodes to the input. Represents all samples from 1 to m, that is, the process of traversing all training samples. This formula calculates the The gradient of , and then average the results of all samples; in practice, the accuracy of the formula depends on accumulating and averaging the gradients of all samples to update the weights , the ultimate goal is to adjust the parameters of the model through these gradients to reduce the overall loss;

[0152] The logic of updating weight data is to overwrite the original weight with the updated weight. and , the weight update formula is based on;

[0153] ;

[0154] ;

[0155] in, The learning rate controls the magnitude of each weight update. A too large learning rate may cause the model to not converge, while a too small learning rate may lead to slow convergence. Represents the symbol for updating weights; in each iteration, the gradient of the loss function with respect to the weights is calculated and , we can learn how to adjust the weights to reduce the loss. The updated weights are obtained by subtracting the scaled gradient (via the learning rate) from the current weights. )get;

[0156] Weight and In each iteration, the update is performed according to the gradient of the loss function. This process ensures that the model can gradually learn and reduce the loss. Although in certain cases, the weights may not appear to change (such as when the gradient is very small), the update formula ensures that the weights change as training progresses.

[0157] use Indicators to evaluate model performance, adjust and optimize the model:

[0158] ;

[0159] in, It is used to measure the difference between the predicted value and the actual value. The smaller the value, the more accurate the prediction, and the larger the value, the greater the prediction error. It gives the overall performance of the model and can be used to compare the performance of other models.

[0160] Step 3: Input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data;

[0161] Use the trained model to make predictions and input the real-time collected data on flash flood influencing factors:

[0162] ;

[0163] in, Represents characteristic input data, including data on factors affecting flash floods: rainfall data , terrain slope and soil water absorption ;

[0164] Rainfall data , which can reflect the intensity of precipitation and its impact on the formation of flash floods. Generally speaking, the greater the rainfall, the higher the risk of flash floods that may result;

[0165] Slope data , changes in slope affect the speed of water flow and its flow path. The greater the slope, the faster the flow rate is, which increases the risk of flash floods;

[0166] Soil water absorption data , soils with strong water absorption capacity can better absorb rainwater, reduce surface runoff, and thus reduce the risk of flash floods;

[0167] The preprocessed data is input into the neural network for forward propagation to obtain the predicted value of the flash flood height, and the input layer is transferred to the hidden layer through the weight matrix and bias The characteristics Transformed into the output of the hidden layer :

[0168] ;

[0169] in, is the output of the jth hidden node, indicating that it has been activated by a nonlinear activation function. The value reflects how the input features generate new feature representations according to the designed nonlinear transformation after the linear combination of weights and biases; the output of the hidden layer nodes represents the model's ability to process complex combinations of input features, revealing the potential relationship between input features; when input features (such as rainfall, slope, water absorption rate) increase, assuming that their weights is positive, the node output It will usually increase, thus affecting the final prediction results;

[0170] The output of the hidden layer Convert to predicted value :

[0171] ;

[0172] in, Represents the final prediction value of the model, that is, the predicted flash flood height. This value is obtained after further processing of the output information of the hidden layer, and corresponds to the flash flood height that the input features may cause under specific conditions. The predicted flash flood height directly reflects the expected flood severity of the model under given meteorological and geographical conditions. The weight Determines the contribution of different hidden node outputs to the final prediction. The larger the weight, the more significant the impact of the hidden node on the prediction result.

[0173] Compare the predicted actual flash flood height with the height of the power poles to assess the potential impact of flash floods on power facilities:

[0174] ;

[0175] in, The actual flash flood height data predicted is obtained through model calculation. is the average height of the poles of power facilities, Represents the difference between the predicted flash flood height and the height of the electric pole. This difference is an important indicator for assessing the possible impact of flash floods on power facilities.

[0176] if increases (e.g., increased rainfall), while If unchanged, will also increase, indicating that the risk of facilities being flooded increases as the predicted actual flash flood height increases; if increase (e.g., a pole is raised), and If unchanged, Will be reduced, reducing the risk of facilities being flooded.

[0177] Step 4: Determine the flood disaster risk level based on the comparison results, determine the ability of the power facility's pole height data to cope with the predicted actual flash flood height data, calculate the overall risk index of the power facility, and conduct an assessment based on the risk index;

[0178] Based on the comparison between the height of mountain torrents and the height of electric poles, the risk level of flood disasters is defined as follows:

[0179] ;

[0180] Among them, when , indicating high risk, meaning that the predicted actual flash flood height is significantly higher than the height of the poles, which may cause damage or paralysis of facilities. This state requires immediate action to protect facilities and personal safety; , indicating low risk, indicating that the predicted actual flash flood height does not exceed the height of the poles, and the power facilities are safe under current conditions, which means that the facilities can withstand the current flood conditions and there is no risk of being submerged. , indicating medium risk, which means that the height of the mountain torrent is slightly higher than the height of the power pole, but not enough to cause serious damage. At this time, the power facilities are in a dangerous state and early warning measures need to be taken to remind relevant personnel to be vigilant about possible flood impacts. The larger the value, the higher the risk level, indicating that the poles are less resilient to flood impacts and the greater the risk;

[0181] Calculate a composite risk index , comprehensively assess the overall risk of power facilities:

[0182] ;

[0183] in, is the frequency of flood events, defined as the number of flood events in the past 10 years, It is an assessment of the vulnerability of the infrastructure, with a value range of 1 to 5, where 1 indicates low vulnerability (the facility is robust) and 5 indicates high vulnerability (the facility is easily damaged);

[0184] when ,Right now It means that the height and minimum height of the pole can cope with flooding, and the risk is zero; ,Right now , the calculation here reflects the risk when the flood height exceeds the height of the pole, when ,Right now , indicating that the flood is above the lowest pole height but below the average height, and appropriate risk weighting is given (multiplied by 0.5);

[0185] in, It indicates the difference between the height of the flash flood and the height of the electric pole, which directly reflects the potential water depth risk; is the flood frequency factor. As the number of historical flood events increases, the risk will increase accordingly. Here, "10" is a baseline value used for scaling so that the increase in frequency will not cause a sharp increase in the risk index. is the infrastructure vulnerability factor. As the vulnerability score increases, the risk also increases. The “4” here maps the score range (1 to 5) to an appropriate scale to ensure When the risk factor is 1 (lowest risk), When the risk factor is 2 (highest risk);

[0186] (Risk level) reflects the safety status of power facilities under flood conditions. The value defines the risk level of the facility and helps decision makers take appropriate measures; the independent variable The bigger, if constant, It will also increase, causing the risk level to rise (from low risk to high risk);

[0187] The Composite Risk Index quantifies the overall risk of power facilities to specific flash flood conditions. The greater the risk, the greater the potential threat to power facilities. Increase (flood height increases), if constant, will increase, indicating that the risk is increasing, and the independent variable Increase (increase in the frequency of flood events), the risk index will also increase, and the independent variable An increase (increased infrastructure vulnerability) will also lead to an increase in the risk index.

[0188] See also Figure 2 The present invention further provides a system for assessing the risk of flood disasters for power facilities taking into account the impact of floods. The system is used to execute the above-mentioned method for assessing the risk of flood disasters for power facilities taking into account the impact of floods, and comprises:

[0189] The data acquisition module is used to collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data;

[0190] The model building module is used to build a neural network model. The rainfall data, topography and soil data in the pre-processed historical data are used as training sets, and the corresponding mountain torrent height data are used as labels, which are input into the neural network model to train the neural network model.

[0191] The flash flood prediction module is used to input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data;

[0192] The risk assessment module is used to derive the flood disaster risk level based on the comparison results, determine the ability of the pole height data of the power facilities to cope with the predicted actual mountain torrent height data, calculate the overall risk index of the power facilities, and conduct assessment based on the risk index.

[0193] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0194] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0195] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0196] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for assessing flood disaster risk of power facilities considering flood impact, characterized in that: The specific steps include: Step 1: Collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data; The pole height data includes the heights of all poles in the monitored area and the lowest height of poles in the area. The step of preprocessing the pole height data is to calculate the average height of all pole height data in the monitored area, based on the formula: in, represents the average height of the poles, represents the height of the i-th pole, Refers to the number of poles within the monitoring area; Get the minimum height of all poles in the area to be monitored , the formula based on is: The rainfall data is the total rainfall in the past K days in the monitored area. The steps of preprocessing the rainfall data include: The total rainfall for the past K days is calculated based on the following formula: in, is the total rainfall, which is the sum of all rainfall that occurred in the past K days, is the rainfall on the qth day; The acquired topographic data in the monitored area include the elevation data of the highest point in the monitored area, the elevation data of the lowest point and the horizontal distance between the highest point and the lowest point, and the step of preprocessing the topographic data includes calculating the slope value by calculating the height difference and the horizontal distance between the highest point and the lowest point, and the formula based on which is: in, Indicates the slope value, Represents the elevation difference between two points, that is , and Respectively represent the elevation data of the highest point and the lowest point in the area to be monitored. Indicates the horizontal distance between two points; When obtaining soil data in the area to be monitored, soil samples are collected from the lowest point in the area to be monitored one day before the assessment, and the water volume and soil volume in the soil samples are obtained by high-temperature drying to constant weight. The step of preprocessing the soil data is to calculate the water absorption rate of the soil, and the water absorption rate of the soil is used as the soil data in the area to be monitored. The formula for calculating the water absorption rate is: in, Indicates water absorption (%), is the volume of water in the soil, represents the volume of soil; The rainfall data in the historical data is the total rainfall in the K-day period before the flood, and is calibrated as The soil data in the historical data is the soil sample collected at the lowest point in the monitored area one day before the flood, and the water absorption rate of the soil is calculated as the soil data and calibrated as The topographic data in the historical data are the elevation data of the highest point and the lowest point in the monitored area, as well as the horizontal distance between them. The terrain slope data is calculated as the topographic data and calibrated as ; Step 2: Build a neural network model, take the rainfall data, topography and soil data in the preprocessed historical data as the training set, and the corresponding flash flood height data as the label, and input it into the neural network model to train the neural network model; Use the information on flash flood influencing factors in the monitored area from historical data, including rainfall data, topography and soil data as the training set : in, , and They represent rainfall data, topographic data and soil data in historical data respectively; Tagged with : in, Represents the mapping relationship, Represents historical flash flood height data; Feature data and label data Pass it as input and output to the neural network model for training, and define the number of input nodes in the input layer as , which represents the rainfall ,slope and soil water absorption ; Number of hidden layer nodes , using the ReLU activation function: , introducing a nonlinear relationship: in, is the output of the jth hidden node, is the weight matrix, connecting the input layer and the hidden layer, is the bias term; Output of hidden layer By weight Transformation, and finally get the prediction output of the model: in, is the prediction result of the model, that is, the predicted height of the flash flood, is the weight matrix from the hidden layer to the output layer, mapping the features of the hidden layer to the final output. is the bias term of the output layer; Model training, by calculating the square difference between all predicted values ​​and true values, the loss function reflects the prediction accuracy of the model: in, Represents the loss value, measuring the model prediction value With the actual label The gap between Represents the number of training samples, used to calculate the average loss, represents the predicted value of the oth sample, represents the historical flash flood height of the oth sample; Step 3: Input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data; Step 4: Based on the comparison results, the flood disaster risk level is obtained, the ability of the pole height data of the power facilities to cope with the predicted actual flash flood height data is determined, and the overall risk index of the power facilities is calculated, and an assessment is conducted based on the risk index.

2. A method for assessing flood disaster risk of power facilities considering flood impact according to claim 1, characterized in that: By minimizing the loss function To adjust the parameters of the model, namely weights and biases: This formula is the loss function over weights The gradient of To reduce the loss function , It represents the output of the first hidden layer for the oth sample. Represents the sample number in the training set; Through the loss function about the output gradient Can be further propagated to the hidden layer to calculate the weight The gradient is: in, is the derivative of the ReLU function, indicating the sensitivity of the hidden layer nodes to the input; The logic of updating weight data is to overwrite the original weight with the updated weight. and , the weight update formula is: in, is the learning rate, which controls the magnitude of each weight update. Indicates the update symbol of the weight; use Indicators to evaluate model performance, adjust and optimize the model: in, It is used to measure the difference between the predicted value and the actual value. The smaller the value, the more accurate the prediction.

3. A method for assessing flood disaster risk of power facilities considering flood impact according to claim 2, characterized in that: In step 3, the trained model is used for prediction, and the real-time collected data on flash flood influencing factors is input: in, Represents characteristic input data, including data on factors affecting flash floods: rainfall data , terrain slope and soil water absorption ; The preprocessed data is input into the neural network for forward propagation to obtain the predicted value of the flash flood height, and the input layer is sent to the hidden layer through the weight matrix and bias The characteristics Transformed into the output of the hidden layer : in, is the output of the jth hidden node; The output of the hidden layer Convert to predicted value : in, Indicates the actual predicted flash flood height; Compare the predicted actual flash flood height with the height of the power poles to assess the potential impact of flash floods on power facilities: in, The actual flash flood height data predicted is obtained through model calculation. is the average height of the poles of power facilities, Represents the difference between the predicted actual flash flood height and the height of the pole.

4. A method for assessing flood disaster risk of power facilities considering flood impact according to claim 3, characterized in that: In step 4, the risk level of flood disasters is defined based on the comparison between the actual mountain torrent height and the pole height: Among them, when , indicating high risk, indicating that the height of the mountain torrent is higher than the height of the pole, which will cause damage to the facilities; when , indicating low risk, the pole is safe in flash flood conditions. , indicating medium risk, the pole is in a dangerous state and requires early warning; Calculate a composite risk index , comprehensively assess the overall risk of power facilities: in, is the frequency of flood events, defined as the number of flood events in the past 10 years, It is the vulnerability assessment of infrastructure, with a value range of 1 to 5, where 1 indicates low vulnerability and 5 indicates high vulnerability; when ,Right now It means that the height and minimum height of the pole can cope with flooding, and the risk is zero; ,Right now , the calculation here reflects the risk when the flood height exceeds the height of the pole, when ,Right now , indicating that the flood is above the lowest pole height but below the average height, and an appropriate risk weighting (multiplied by 0.5) is given.

5. A flood disaster risk assessment system for power facilities considering flood impact, characterized by: The system is used to execute a method for assessing flood disaster risk of power facilities taking into account flood impacts as described in any one of claims 1 to 4, comprising: The data acquisition module is used to collect the pole height data and flash flood influencing factor information of the power facilities in the monitored area in real time, collect the flash flood influencing factor information in the historical data, and obtain the historical flash flood height data. The flash flood influencing factor information includes rainfall data, topography and soil data, and pre-process the collected data; The model building module is used to build a neural network model. The rainfall data, topography and soil data in the pre-processed historical data are used as training sets, and the corresponding mountain torrent height data are used as labels, which are input into the neural network model to train the neural network model. The flash flood prediction module is used to input the pre-processed real-time collected flash flood influencing factor data into the trained neural network model, calculate the predicted actual flash flood height data, and use the predicted actual flash flood height data to compare with the pre-processed pole height data; The risk assessment module is used to derive the flood disaster risk level based on the comparison results, determine the ability of the pole height data of the power facilities to cope with the predicted actual mountain torrent height data, calculate the overall risk index of the power facilities, and conduct assessment based on the risk index.

Images