Intelligent monitoring method and system for drainage pipe network

Abstract

The invention relates to the field of drainage pipe network monitoring, in particular to an intelligent monitoring method and system for a drainage pipe network, and the method comprises the steps: calculating the weight of each decision tree in a GBDT model, and obtaining the change speed of each parameter of a drainage pipe at the current moment; and inputting the change speed of each parameter into the corresponding GBDT model to judge whether the drainage pipeline is silted or not. According to the method, the silt accumulation condition of the drainage pipe network is comprehensively monitored through multiple parameters, dependence on a single parameter is reduced, meanwhile, the influence of the single parameter on a monitoring result is reduced, the possibility of misjudgment is reduced, and therefore the accuracy and stability of the monitoring result are improved.

Description

Technical Field

[0001] This invention relates to the field of drainage network monitoring, and more particularly to an intelligent monitoring method and system for drainage networks. Background Technology

[0002] As a crucial component of urban infrastructure, drainage pipe networks play a vital role in rainwater drainage, sewage collection, and flood control. However, due to factors such as planning, design, construction, and sludge settling, drainage pipes frequently experience siltation and blockages. This reduces their flow capacity, severely impacting the normal operation of the drainage network and leading to problems like road flooding and urban waterlogging during rainfall. This causes significant inconvenience to traffic and residents' daily lives. The complexity and lack of visibility inherent in urban drainage processes make the management of drainage pipe networks extremely challenging.

[0003] Chinese patent document CN115031776B discloses a method for monitoring and analyzing siltation in drainage pipe networks, including the following steps: S1: Select sample data, process outliers in the sample data, and convert the sample data into dimensionless values; S2: Based on the dimensionless values, construct and train a gated loop unit model using a heterogeneous parallel architecture to obtain the predicted liquid level; S3: Analyze the mean square error and goodness of fit between the predicted liquid level and the actual liquid level to obtain the accuracy of the predicted liquid level. If the accuracy of the predicted liquid level meets the requirements, monitor the actual liquid level in real time; S4: Based on the degree of deviation between the actual liquid level and the predicted liquid level, determine the degree of siltation and classify it.

[0004] However, the above method only monitors and analyzes the siltation of the drainage pipe network by the degree of liquid level deviation. When using a single element for monitoring, the monitoring results mainly come from the processing and analysis of single element data, and are greatly affected by single element data, resulting in poor stability of the monitoring results, easy misjudgment, and low accuracy of drainage pipe network monitoring. Summary of the Invention

[0005] To address the problem of low accuracy in monitoring results when using a single factor to monitor drainage pipe networks, this invention provides an intelligent monitoring method and system for drainage pipe networks.

[0006] In a first aspect, the present invention provides an intelligent monitoring method for drainage pipe networks, employing the following technical solution: Calculate the weight of each decision tree in the GBDT model, obtain the rate of change of various parameters of the drainage pipe at the current moment, and input the rate of change of various parameters into the corresponding GBDT model to determine whether the drainage pipe has accumulated silt. The weights of the decision tree are calculated as follows: Each historical rainfall period is divided into multiple sub-rainfall periods. The rate of change of various parameters of the drainage pipe at each moment in the sub-rainfall period is obtained. The rate of change of the corresponding parameters in the sub-rainfall period is constructed. A label is set for each moment regarding whether the pipe is silted up, and the label is silted up or normal. Calculate the mutual information entropy between the rate of change of each parameter and the label in the sub-rainfall period. The mutual information entropy of the rate of change of the same parameter in the same sub-rainfall period constitutes the mutual information entropy sequence of the corresponding parameter. Calculate the stability of the mutual information entropy sequence. The stability is negatively correlated with the variance of the mutual information entropy sequence. Calculate the importance coefficient of each parameter in the same type of sub-rainfall period. The importance coefficient is positively correlated with the stability of the information entropy sequence of the corresponding parameter and the product of mutual information entropy. The weights of the decision tree are calculated, and the weights are positively correlated with the number of parameters and the importance coefficients of the parameters in the decision tree.

[0007] By comprehensively monitoring the siltation of drainage pipe networks using multiple parameters, the reliance on a single parameter is reduced, as is the impact of a single parameter on the monitoring results. This reduces the possibility of misjudgment and improves the accuracy and stability of the monitoring results.

[0008] Preferably, the method for classifying each historical rainfall period into multiple categories of sub-rainfall periods is as follows: The system obtains the rainfall amount at each moment during historical rainfall periods, calculates the rate of change of rainfall amount at each moment, and clusters the rates of change of rainfall amount to obtain multiple clusters, with each cluster corresponding to a sub-rainfall period.

[0009] By classifying historical rainfall periods into multiple sub-rainfall periods, a theoretical basis is provided for selecting the GBDT model to predict siltation in drainage pipe networks.

[0010] Preferably, the expression for the stability of the information entropy sequence is:

[0011] In the formula, In the j-th sub-rainfall period, the first... The stability of the mutual information entropy sequence is determined by the rate of change of the parameters. The standard deviation of the mutual information entropy sequence represents the rate of change of the x-th parameter during the j-th sub-rainfall period; Let represent the average value of the mutual information entropy sequence of the rate of change of the x-th parameter during the j-th sub-rainfall period, and exp represent an exponential function with base e.

[0012] The stability of the mutual information entropy sequence is calculated by using the standard deviation and mean of the mutual information entropy sequence, which quantifies the stability of the mutual information entropy and improves the accuracy of the calculation results.

[0013] The preferred expression for the importance coefficient is:

[0014] In the formula, Indicates the first The importance coefficient of the x-th parameter in the sub-rainfall period. The mutual information entropy between the rate of change of parameter x in the j-th sub-rainfall period within the k-th rainfall period and the label Y indicating whether the pipeline is silted up is represented. The mutual information entropy sequence represents the stability of the rate of change of the x-th parameter in the j-th sub-rainfall period, and K represents the total number of historical rainfall periods.

[0015] The above formulas demonstrate the importance of various parameters in predicting siltation in drainage pipe networks, thus facilitating the prediction of the current siltation situation and improving the accuracy of the prediction results.

[0016] Preferably, the weight expression for the decision tree is:

[0017] In the formula, This represents the weight of the γ-th decision tree in the GBDT model during the j-th sub-rainfall period; This represents the total number of parameters in the x-th item of the γ-th decision tree; Indicates the first The importance coefficient of the x-th parameter in the sub-rainfall period; Let be the total number of parameter terms on the γth decision tree; Let y be the total number of parameters for all items in the γth decision tree.

[0018] Compared with traditional decision trees, this approach considers the impact of decision trees with different weights on the calculation results. By performing weighted fitting on the decision trees, the final result is obtained, which improves the accuracy of the prediction results.

[0019] Preferably, the method for classifying each historical rainfall period into multiple categories of sub-rainfall periods is as follows: The rate of change of rainfall at each moment in historical rainfall periods is obtained, and the rate of change is clustered to obtain multiple clusters. The time period corresponding to the data points in the cluster is taken as the sub-rainfall period.

[0020] Preferably, the expression for the rate of change of rainfall is:

[0021] In the formula, For the first The rate of change of rainfall at time t in a historical rainfall period; and The first Rainfall amounts at times t-1, t, and t+1 in a given historical rainfall period; This represents the time interval between adjacent moments.

[0022] The preferred method for inputting the rate of change of each parameter into the corresponding GBDT model is as follows: Calculate the probability of the current rainfall falling into the corresponding category of sub-rainfall period, and input the rate of change of each parameter into the GBDT model corresponding to the sub-rainfall period with the highest probability.

[0023] Preferably, the method further includes calculating the probability of the current rainfall belonging to a sub-rainfall period. The calculation method is as follows: obtain the rate of change of the rainfall at the current moment, and the expression for the probability is:

[0024] In the formula, The rainfall at the current moment belongs to the first... The likelihood of a period of similar rainfall. This represents the total number of historical rainfall periods. The rate of change of rainfall at the current moment. Let be the average rate of change of rainfall in the j-th sub-rainfall period within the k-th historical rainfall period, and exp represent an exponential function with base e.

[0025] By assessing the probability, we can determine the category of the sub-rainfall period to which the current rainfall belongs, thus facilitating the selection of the appropriate GBDT model and enabling targeted prediction, thereby refining the prediction process.

[0026] Secondly, the present invention provides an intelligent monitoring system for drainage pipe networks, employing the following technical solution: An intelligent monitoring system for drainage pipe networks includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the intelligent monitoring method for drainage pipe networks described above is implemented.

[0027] The aforementioned intelligent monitoring method for drainage pipe networks is generated into a computer program and stored in a memory for loading and execution by a processor. Thus, a system is created based on the memory and processor for convenient use.

[0028] The present invention has the following technical effects: By comprehensively monitoring the siltation of drainage pipe networks using multiple parameters such as liquid level, flow velocity, and turbidity, the reliance on a single parameter is reduced, as is the impact of a single parameter on the monitoring results. This reduces the possibility of misjudgment and improves the accuracy and stability of the monitoring results. Attached Figure Description

[0029] Figure 1 This is a flowchart of an intelligent monitoring method for drainage pipe networks according to the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] This invention discloses an intelligent monitoring method for drainage pipe networks, referring to... Figure 1 This includes the following steps: S1: Divide each historical rainfall period into multiple sub-rainfall periods and obtain the rate of change of various parameters of the drainage pipe at each moment in the sub-rainfall period.

[0032] The study obtains rainfall data and multiple parameter data of the drainage network at different times during historical rainfall periods. The parameters of the drainage network include liquid level, flow velocity, and turbidity. When there is siltation in the drainage network, the liquid level will rise sharply in a short period of time, the flow velocity will decrease sharply, and the turbidity of the water will change in a wave-like manner. Therefore, the presence of siltation in the drainage network can be determined by liquid level, flow velocity, and turbidity. The rate of change of rainfall at each time moment is calculated, and the rate of change of rainfall is clustered using ordered sample clustering to obtain multiple clusters, each cluster corresponding to a sub-rainfall period.

[0033] The expression for the rate of change of rainfall is:

[0034] In the formula, For the first The rate of change of rainfall at time t in a historical rainfall period; and The first Rainfall amounts at times t-1, t, and t+1 in a given historical rainfall period; This represents the time interval between adjacent moments.

[0035] Similarly, the calculation methods for the rates of change of liquid level, flow rate, and turbidity are the same as those for the rate of change of rainfall; the specific formulas will not be repeated here.

[0036] For example, data from three historical rainfall periods are obtained, with each period including rainfall, liquid level, flow rate, and turbidity at different times.

[0037] For the first historical rainfall period, each moment corresponds to a rainfall change rate data point. The first historical rainfall period corresponds to multiple rainfall change rates. Since rainfall changes during a rainfall period generally follow three cycles: increasing rainfall, stable rainfall, and decreasing rainfall, the K-means algorithm is used to perform ordered sample clustering of the rainfall change rates, resulting in three clusters. The time corresponding to the data points in the first cluster is... - The time corresponding to the data points within the second cluster is - The time corresponding to the data points within the third cluster is - ,So, - This is the first type of sub-rainfall period. - This is the second type of sub-rainfall period. - This is the third type of sub-rainfall period. Then, the rates of change of liquid level, flow velocity, and turbidity are calculated during the first type of sub-rainfall period, the second type of sub-rainfall period, and the third type of sub-rainfall period.

[0038] Similarly, the second and third historical rainfall periods are divided into three sub-rainfall periods. Then, the rates of change of liquid level, flow velocity, and turbidity are calculated in each sub-rainfall period.

[0039] S2: Construct a sequence of the rate of change of the corresponding parameters during the sub-rainfall period, and set a label for each time point regarding whether the pipe is silted up or normal. For example, for the first type of sub-rainfall period in the first historical rainfall period, a sequence of rates of change of liquid level is constructed using the liquid level at each moment; a sequence of rates of change of flow velocity is constructed using the flow velocity at each moment; a sequence of rates of change of turbidity is constructed using the turbidity at each moment; if... If the time marker is siltation, then... The labels for the liquid level, flow rate, and turbidity at any given time all indicate sludge accumulation. If the time label is normal, then The labels for liquid level, flow rate, and turbidity at each time point are all normal. Similarly, the rate of change sequences of liquid level, flow rate, and turbidity in the second and third sub-rainfall periods are the same as those in the first sub-rainfall period.

[0040] The parameter change rate sequence in the sub-rainfall periods of the second and third historical rainfall periods is referenced to the sub-rainfall periods of the first historical rainfall period.

[0041] S3: Calculate the mutual information entropy between the rate of change of each parameter and the label in the sub-rainfall period. The mutual information entropy of the rate of change of the same parameter in the same sub-rainfall period constitutes the mutual information entropy sequence of the corresponding parameter.

[0042] Let mutual information entropy be denoted as Mutual information entropy (MIE) is the mutual information entropy between the rate of change of parameter x in the j-th sub-rainfall period within the k-th historical rainfall period and the siltation status Y in the drainage pipes (Y is 1 when siltation occurs and 0 when normal). A larger value indicates a higher correlation between the change of parameter x in the j-th sub-rainfall period and its ability to distinguish siltation status in drainage pipes, and thus a higher importance for predicting siltation. Conversely, a smaller value indicates a lower correlation between the change of parameter x in the j-th sub-rainfall period and its ability to distinguish siltation status in drainage pipes, and thus a lower importance for predicting siltation. The calculation method for MIE is existing technology, and the specific steps will not be detailed here.

[0043] The mutual information entropy of the rate of change of the same parameter during the same rainfall period constitutes the mutual information entropy sequence of the corresponding parameter.

[0044] For example, the flow velocity in the first type of sub-rainfall period of the first historical rainfall period has a mutual information entropy, the flow velocity in the first type of sub-rainfall period of the second historical rainfall period has a mutual information entropy, and the flow velocity in the first type of sub-rainfall period of the third historical rainfall period has a mutual information entropy. Using the three mutual information entropies obtained, a mutual information entropy sequence about flow velocity is constructed. Similarly, a mutual information entropy sequence about liquid level and a mutual information entropy sequence about turbidity are constructed in sequence.

[0045] Similarly, the second type of sub-rainfall period corresponds to a mutual information entropy sequence with corresponding parameters, and the third type of sub-rainfall period corresponds to a mutual information entropy sequence with corresponding parameters.

[0046] S4: Calculate the stability of the mutual information entropy sequence. The stability is negatively correlated with the variance of the mutual information entropy sequence.

[0047] The expression for the stability of an information entropy sequence is:

[0048] In the formula, In the j-th sub-rainfall period, the first... The stability of the mutual information entropy sequence is determined by the rate of change of the parameters. The standard deviation of the mutual information entropy sequence represents the rate of change of the x-th parameter during the j-th sub-rainfall period; Let represent the average value of the mutual information entropy sequence of the rate of change of the x-th parameter during the j-th sub-rainfall period, and exp represent an exponential function with base e.

[0049] denoted as the coefficient of variation of the mutual information entropy sequence of the rate of change of the x-th parameter under the j-th sub-rainfall period. The larger the value, the greater the volatility and lower the stability of the mutual information entropy of the rate of change of the x-th parameter under the j-th sub-rainfall period in each historical rainfall period; the smaller the value, the smaller the volatility and higher the stability of the mutual information entropy of the rate of change of the x-th parameter under the j-th sub-rainfall period in each historical rainfall period.

[0050] S5: Calculate the importance coefficient of each parameter in the same type of sub-rainfall period. The importance coefficient is positively correlated with the stability of the information entropy sequence of the corresponding parameter and the product of mutual information entropy.

[0051] The expression for the importance coefficient is:

[0052] In the formula, Indicates the first The importance coefficient of the x-th parameter in the sub-rainfall period. The label represents the rate of change of the x-th parameter in the j-th sub-rainfall period within the k-th rainfall period, and whether the pipe is silted up. Mutual information entropy, The stability of the mutual information entropy sequence represents the rate of change of the x-th parameter during the j-th sub-rainfall period. This represents the total number of historical rainfall periods. Therefore, each parameter within each category of sub-rainfall periods corresponds to an importance coefficient. For example, for the first category of sub-rainfall periods out of three historical rainfall periods, the corresponding liquid level, flow velocity, and turbidity each have an importance coefficient; similarly, the second category of sub-rainfall periods, and the third category of sub-rainfall periods, all have their respective importance coefficients.

[0053] S6: Calculate the weight of each decision tree in the GBDT model. The weight is positively correlated with the number of parameters and the importance coefficient of the decision tree.

[0054] For each category of sub-rainfall periods, a GBDT model is constructed. The GBDT model is trained using the rate of change of various parameters within the same category of sub-rainfall periods and their corresponding labels. The weights of each decision tree in the GBDT model are then calculated. The weight expression for each decision tree is as follows:

[0055] In the formula, This represents the weight of the γ-th decision tree in the GBDT model during the j-th sub-rainfall period; This represents the total number of parameters in the x-th item of the γ-th decision tree; Indicates the first The importance coefficient of the x-th parameter in the sub-rainfall period; Let be the total number of parameter terms on the γth decision tree; Let y be the total number of parameters for all items in the γth decision tree.

[0056] S7: Calculate the weight of each decision tree in the GBDT model, obtain the rate of change of various parameters of the drainage pipe at the current moment, and input the rate of change of various parameters into the corresponding GBDT model to determine whether the drainage pipe has accumulated silt.

[0057] To calculate the probability of the current rainfall belonging to a specific sub-rainfall period, the calculation method is as follows: obtain the rate of change of the current rainfall amount. The expression for the probability is:

[0058] In the formula, The rainfall at the current moment belongs to the first... The likelihood of a period of similar rainfall. This represents the total number of historical rainfall periods. The rate of change of rainfall at the current moment. Let be the average rate of change of rainfall in the j-th sub-rainfall period within the k-th historical rainfall period, and exp represent an exponential function with base e. The probability is used to represent the probability that the current rainfall moment belongs to the corresponding sub-rainfall period category.

[0059] For example, if the probability of the current moment being related to the first type of sub-rainfall period is 0.5, the probability of it being related to the second type of sub-rainfall period is 0.8, and the probability of it being related to the third type of sub-rainfall period is 0.7, then the rainfall at the current moment belongs to the second type of sub-rainfall period. Therefore, by inputting multiple parameters of the current moment into the GBDT model corresponding to the second type of sub-rainfall period, the result of whether the drainage pipe has accumulated silt at the current moment can be obtained.

[0060] This invention also discloses an intelligent monitoring system for drainage pipe networks, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an intelligent monitoring method for drainage pipe networks according to the present invention.

[0061] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

Claims

1. A method for intelligent monitoring of a sewer network, characterized in that, The method comprises the steps of: calculating the weight of each decision tree in the GBDT model, obtaining the change rate of each parameter of the drainage pipeline at the current time, and inputting the change rate of each parameter into the corresponding GBDT model to determine whether the drainage pipeline is blocked; The calculation method of the weight of the decision tree is: dividing each historical rainfall period into multiple categories of sub-rainfall periods, obtaining the change rate of each parameter of the drainage pipeline at each time in the sub-rainfall period, constructing the change rate sequence of the corresponding parameter in the sub-rainfall period, setting a label about whether the pipeline is blocked at each time, and the label is blocked or normal; calculating the mutual information entropy between the change rate of each parameter in the sub-rainfall period and the label, the mutual information entropy of the change rate of the same parameter in the same sub-rainfall period constitutes the mutual information entropy sequence of the corresponding parameter; the stability of the mutual information entropy sequence is negatively correlated with the variance of the mutual information entropy sequence; calculating the importance coefficient of each parameter in the same sub-rainfall period, the importance coefficient is positively correlated with the product of the stability of the information entropy sequence and the mutual information entropy of the corresponding parameter; calculating the weight of the decision tree, the weight is positively correlated with the number of parameters on the decision tree and the importance coefficient.

2. A method for intelligent monitoring of sewer networks according to claim 1, characterized in that, The method for classifying each historical rainfall period into multiple categories of sub-rainfall periods is: obtaining the rainfall amount at each time in the historical rainfall period, calculating the change rate of the rainfall amount at each time, clustering the change rate of the rainfall amount to obtain multiple clustering clusters, and each clustering cluster corresponds to a sub-rainfall period.

3. A method for intelligent monitoring of sewer networks according to claim 1, characterized in that, The expression of the stability of the information entropy sequence is: In the formula, In the j-th sub-rainfall period, the first... The stability of the mutual information entropy sequence is determined by the rate of change of the parameters. The standard deviation of the mutual information entropy sequence represents the rate of change of the x-th parameter during the j-th sub-rainfall period; Let represent the average value of the mutual information entropy sequence of the rate of change of the x-th parameter during the j-th sub-rainfall period, and exp represent an exponential function with base e.

4. A method for intelligent monitoring of sewer networks according to claim 1, characterized in that, The expression of the importance coefficient is: In the formula, Indicates the first The importance coefficient of the x-th parameter in the sub-rainfall period. The label represents the rate of change of the x-th parameter in the j-th sub-rainfall period within the k-th rainfall period, and whether the pipe is silted up. Mutual information entropy, The stability of the mutual information entropy sequence represents the rate of change of the x-th parameter during the j-th sub-rainfall period. This indicates the total number of historical rainfall periods.

5. A method for intelligent monitoring of sewer network as claimed in claim 1 wherein, The expression of the weight of the decision tree is: In the formula, represents the weight of the γth decision tree in the GBDT model in the jth sub-rain period; represents the total number of the xth parameter on the γth decision tree; represents the importance coefficient of the xth parameter in the jth sub-rain period; represents the importance coefficient of the xth parameter in the jth sub-rain period; is the total number of parameters on the γth decision tree; is the total number of parameters on the γth decision tree.

6. A method for intelligent monitoring of sewer networks according to claim 1, characterized in that, The method for classifying each historical rainfall period into multiple categories of sub-rainfall periods is: obtaining the change rate of the rainfall amount at each time in the historical rainfall period, clustering the change rate to obtain multiple clustering clusters, and taking the time period corresponding to the data points in the clustering cluster as a sub-rainfall period.

7. A method for intelligent monitoring of sewer networks according to claim 6, characterized in that, The expression of the change rate of the rainfall amount is: In the formula, is the rainfall rate at the tth time in the nth historical rainfall period; is the rainfall rate at the tth time in the nth historical rainfall period; is the rainfall rate at the tth time in the nth historical rainfall period; is the rainfall rate at the tth time in the nth historical rainfall period; is the rainfall rate at the tth time in the nth historical rainfall period;​ 8. A method for intelligent monitoring of sewer networks according to claim 7, characterized in that, The method for inputting the change rate of each parameter into the corresponding GBDT model is: calculating the possible degree of the sub-rainfall period of the corresponding category to which the current rainfall belongs, and inputting the change rate of each parameter into the GBDT model corresponding to the sub-rainfall period with the largest possible degree.

9. A method for intelligent monitoring of sewer networks according to claim 8, characterized in that, The expression of the possible degree is: wherein is the probability that the current time rainfall belongs to the kth rainfall period, is the total number of historical rainfall periods; is the change rate of the current time rainfall, is the average change rate of the jth sub-rainfall period in the kth historical rainfall period, and exp denotes the exponential function with base e.

10. An intelligent monitoring system for sewer networks, characterized in that, comprising: a processor and a memory, the memory stores computer program instructions, when the computer program instructions are executed by the processor, a method for intelligent monitoring of a drainage pipeline network according to any one of claims 1-9 is realized.

Images