Intelligent leakage detection and alarm system and method for water supply pipeline based on current analysis

Abstract

The invention discloses an intelligent leakage detection and alarm system and method for a water supply pipeline based on current analysis. The method comprises the following steps: S1, collecting current signal data and preprocessing the current signal data; s2, inputting the current signal data into an adaptive neural synaptic network model, and identifying a water leakage signal in the pipeline; s3, optimizing a feature extraction process of the current signal by using a parrot optimization algorithm, and dynamically optimizing a learning strategy of the model; s4, the pipeline water leakage state is classified in real time; s5, the pipeline is divided into a plurality of pipeline sections, and the water leakage state of each pipeline section is monitored in real time; s6, a signal lamp module is used for providing alarm display with different colors for each pipeline section; s7, hydroelectric generation or solar energy technology is adopted, and a rechargeable battery and an energy storage module are arranged; and S8, uploading monitoring data through wireless communication, storing a water leakage report and visually displaying the water leakage report through a display. According to the method, the self-adaptive neural synaptic network model and the parrot optimization algorithm are combined, so that the accuracy and the real-time performance of water leakage detection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water supply pipeline leakage detection, and in particular to a water supply pipeline intelligent leakage detection alarm system and method based on current analysis. Background Art

[0002] With the continuous acceleration of urbanization, the water supply pipeline network is becoming increasingly complex, and the operation status of the pipeline directly affects the quality of life and public safety of urban residents. In the water supply pipeline system, water leakage is a common and serious problem. Leakage not only leads to waste of water resources, but also may cause a series of safety problems such as water pollution and environmental damage. Therefore, water supply pipeline leakage detection and timely alarm are very important. Traditional water supply pipeline leakage detection technology mostly relies on manual inspections or monitoring systems based on traditional methods such as pressure and flow, but these methods often have problems such as poor accuracy, poor real-time performance and limited coverage, which are difficult to meet the high requirements of modern cities for water supply pipeline monitoring.

[0003] Traditional water leakage detection methods usually judge water leakage based on water pressure changes, flow anomalies or sound wave signals. When using pressure sensors for monitoring, if a local water leakage occurs in the pipeline system, the pressure change may not be obvious, and the water leakage problem cannot be discovered in time, especially in underground pipelines or closed areas, where water leakage detection is more difficult. In addition, these detection methods based on physical changes usually require a long time of data accumulation and processing, and can only provide local monitoring information, and cannot achieve comprehensive monitoring of the pipeline system. In some cases, due to the complexity and variability of the pipeline system, traditional methods may produce false alarms or missed alarms, resulting in the failure to detect and deal with water leakage problems in a timely manner, leading to more serious consequences.

[0004] In order to solve the shortcomings of traditional water leakage detection technology, water leakage detection methods based on current signals have been widely studied in recent years. The current signal can reflect the change of water flow conductivity, which is closely related to the water leakage in the pipeline. By installing a current sensor in the pipeline, the change of water flow conductivity can be monitored in real time, thereby providing an early warning signal of water leakage. However, the existing current signal analysis methods usually face problems such as large signal noise and difficulty in signal feature extraction. Especially in complex environments, the current signal is affected by various interference factors, which makes the extraction and analysis process of the water leakage signal more difficult and the accuracy is low.

[0005] In addition, although the current signal analysis method can effectively capture water leakage signals, the existing algorithms often rely on simple statistical methods or rule-based analysis models when processing these signals. These methods cannot fully explore the complex patterns and deep-level features in the current signals. Traditional algorithm models usually have a simple analysis of current signals, lack adaptability, and cannot be dynamically adjusted according to environmental changes or different pipeline conditions. Therefore, the accuracy and real-time performance of water leakage detection are poor, and cannot meet the requirements for efficient, intelligent, and real-time water supply pipeline monitoring systems.

[0006] At present, the application of neural networks and deep learning technologies in signal processing and pattern recognition has made certain progress, especially in the fields of image and speech. However, there are still some challenges when applying these technologies to water supply pipe leakage detection. Existing adaptive neural network models often require a large amount of training data when processing complex current signals, and in the process of model training, they are often easily affected by the quality and quantity of training data, resulting in a certain instability in the final detection results. In addition, when processing signal features, adaptive neural networks often rely on a fixed weight adjustment mechanism and cannot be dynamically optimized according to changes in real-time data. Therefore, it may lead to incorrect judgment or missed detection of water leakage signals.

[0007] In terms of signal optimization, existing technologies usually use traditional optimization algorithms to adjust signal feature parameters. Although these algorithms are effective in some scenarios, they often lack sufficient intelligence and flexibility. For example, although classic optimization algorithms such as genetic algorithms and particle swarm optimization can optimize model parameters to a certain extent, they usually require a lot of computing resources and time, and are sensitive to the setting of initial parameters. They are prone to fall into local optimal solutions, affecting the quality of optimization results.

[0008] Therefore, how to provide a water supply pipeline intelligent leak detection alarm system and method based on current analysis is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0009] One purpose of the present invention is to propose an intelligent leak detection alarm system and method for water supply pipelines based on current analysis. The present invention combines an adaptive neural synaptic network model with the Parrot optimization algorithm to optimize the current signal feature extraction and leak detection process, and can analyze leak signals in water supply pipelines in real time and efficiently. By dynamically adjusting signal analysis parameters and optimizing learning strategies, the present invention improves the accuracy and response speed of leak detection, and avoids the problems of misjudgment and missed judgment in traditional methods. The system has intelligent monitoring, precise classification and multi-segment monitoring capabilities, has significant advantages in leak risk assessment and alarm accuracy, and can provide efficient, accurate and real-time solutions for pipeline management.

[0010] According to an embodiment of the present invention, a water supply pipeline intelligent leakage detection and alarm method based on current analysis includes the following steps:

[0011] S1. Use a current sensor to monitor the conductivity change and water flow state of the water flow outside the water supply pipe in real time, collect current signal data and perform preprocessing;

[0012] S2, inputting the preprocessed current signal data into the adaptive neural synaptic network model, simulating the changes of neuron and synaptic connections, and analyzing the current signal characteristics through the adaptive learning mechanism to identify the water leakage signal in the pipeline;

[0013] S3, using the Parrot optimization algorithm to optimize the feature extraction process of the current signal, adjust the signal analysis parameters, and dynamically optimize the learning strategy of the adaptive neural synaptic network model;

[0014] S4. Based on the output of the optimized adaptive neural synaptic network model, the pipeline leakage status is classified in real time to distinguish leakage signals from normal signals;

[0015] S5. Divide the pipeline into several pipeline sections through the segmented monitoring module, monitor the leakage status of each pipeline section in real time, generate a pipeline leakage report, and issue an alarm signal for the corresponding pipeline section according to the monitoring results;

[0016] S6. Use the signal light module to provide different color alarm displays for each pipeline section. Different colors represent different levels of water leakage risk, and use alarm signals to remind staff to handle the problem;

[0017] S7. Use hydroelectric or solar technology to provide electricity, and equip with rechargeable batteries and energy storage modules as a backup mechanism when energy is insufficient;

[0018] S8. The monitoring data is uploaded to the storage device through wireless communication technology, and the pipeline leakage report is stored in the storage device, which is visualized through the display to update the health status of the pipeline in real time.

[0019] Optionally, the S2 specifically includes:

[0020] S21, the preprocessed current signal data is expressed as a time series in represents the time step t i Current signal, n represents the total length of the time series;

[0021] S22, initializing an adaptive neural synaptic network model, wherein the adaptive neural synaptic network model includes an input layer, a synaptic adjustment layer, a dynamic hidden layer and an output layer, and setting the number of neurons in the output layer to N in , the number of neurons in the dynamic hidden layer is Nhidden , the number of neurons in the output layer is N out , and initialize the weight matrix W syn and a bias vector B, where:

[0022] W syn ={w ij (t)|i∈[1,N in ],j∈[1,N hidden ],t∈[1,n]};

[0023] B = {b j |j∈[1,N hidden ]};

[0024] Among them, w ij (t) represents the synaptic weight at time step t, b j represents the bias value of the jth neuron in the dynamic hidden layer, and t represents the time step;

[0025] S23. Combine the current signal characteristics and dynamically adjust the weight through synaptic efficacy:

[0026]

[0027] in, represents the synaptic efficacy between input neuron i and hidden layer neuron j in time step t, λ represents the adjustment factor of the change rate on the weight, μ represents the adjustment factor of the signal strength, σ represents the standard deviation of the Gaussian distribution, ν represents the adjustment factor of the periodic feature, and τ represents the periodic adjustment parameter;

[0028]

[0029] Among them, w ij (t+1) represents the synaptic weight at time step t+1, and η1 represents the adjustment step size;

[0030] S24. Calculate the weighted input value of the dynamic hidden layer neuron:

[0031]

[0032] Among them, z j represents the weighted input value of the jth neuron in the dynamic hidden layer, k1 represents the regulation factor of synaptic efficacy, and k2 represents the regulation factor of high-order characteristics;

[0033] Calculate the output value of the dynamic hidden layer neuron:

[0034]

[0035] Among them, h jrepresents the output value of the jth neuron in the dynamic hidden layer, tanh represents the hyperbolic tangent function, and β represents the smoothing parameter;

[0036] S25, the output layer combines the dynamic hidden layer output and synaptic efficacy to calculate the leakage signal prediction value:

[0037]

[0038] Among them, y out represents the predicted value of water leakage signal, w jo represents the weight between the jth neuron in the dynamic hidden layer and the output layer, b o represents the output layer bias;

[0039] S26, normalizing the water leakage signal prediction value:

[0040]

[0041] in, represents the normalized water leakage signal prediction value;

[0042] S27. Further optimize the synaptic weight value based on error back propagation:

[0043]

[0044] Among them, L represents the loss function, represents the predicted output value at time step t, y true (t) represents the real water leakage signal at time step t, δ represents the regularization strength, ||W syn (t)||1 represents the weight matrix W syn The L1 regularization term of .

[0045] Optionally, the S3 specifically includes:

[0046] S31. Initialize the characteristics of the current signal data and construct a characteristic vector in represents the time step t i Signal characteristic value:

[0047]

[0048] in, represents the time step t i The characteristic value of the current signal, represents the time step t i The current signal, δ1, δ2, δ3 and δ4 represent the characteristic weight parameters, τ represents the period adjustment parameter, σ represents the standard deviation of the Gaussian distribution, Indicates the rate of change of the current signal;

[0049] S32, initialize the parrot optimization algorithm population P = [F1, F2, ..., F m ], each individual Represents the kth candidate feature vector and initializes the signal analysis parameter Θ k =[θ1,θ2,…,θ p ], where m represents the population size and p represents the dimension of signal analysis parameters;

[0050] S33. Define a fitness objective function to measure the matching degree between the candidate feature vector and the signal characteristics:

[0051]

[0052] Among them, J(F k ,Θ k ) represents the fitness value, β1, β2 and β3 represent the objective function adjustment coefficients, represents the kth candidate feature vector F k At time step t i The eigenvalues ​​on ;

[0053] S34. According to the imitation learning mechanism of the Parrot optimization algorithm, the candidate feature vectors and signal analysis parameters in the population are iteratively optimized:

[0054]

[0055] in, represents the updated value of the kth candidate feature vector, F g represents the optimal solution of the current population, η2 represents the optimal solution learning factor, and ζ represents the random perturbation coefficient;

[0056] S35. Adjust the optimized candidate feature vectors and signal analysis parameters to dynamically optimize the learning strategy of the adaptive neural synaptic network model:

[0057]

[0058] Among them, w ij (t+1) represents the synaptic weight at time step t+1, w ij (t) represents the synaptic weight at time step t, ξ1 represents the fitness learning factor, and ξ2 represents the parameter perturbation factor;

[0059] S36. After multiple iterations, select the feature vector F with the highest fitness value * and signal analysis parameter Θ * , F * Input to the adaptive neural synaptic network model and use Θ *Dynamically adjust the learning rate.

[0060] Optionally, the S4 specifically includes:

[0061] S41, obtaining an output signal from the optimized adaptive neural synaptic network model;

[0062] S42, converting the output signal into a feature vector representation in a high-dimensional space through feature mapping, wherein the dimension of the vector representation is related to the number of features of the current signal;

[0063] S43, use the classification algorithm based on multi-layer perceptron to process the feature vector representation and output the classification vector c t =[c1,c2], where c1 represents the classification score of the water leakage signal and c2 represents the classification score of the normal signal;

[0064] S44. By comparing the classification scores, the category with the highest score is selected as the current state of the pipeline. If c1>c2, the pipeline is determined to be in a leaking state; if c2>c1, the pipeline is determined to be in a normal state.

[0065] Optionally, the S5 specifically includes:

[0066] S51, the segmented monitoring module divides the pipeline into several pipeline segments, each pipeline segment is equipped with a current sensor monitoring node for real-time collection of water flow conductivity data;

[0067] S52, each monitoring node generates real-time current signal data of the pipeline section according to the data collected by the current sensor, and transmits it to the central processing unit for analysis;

[0068] S53, the central processing unit processes the collected current signal data, uses the optimized neural network model to analyze the leakage state of each pipeline section, and identifies the leakage signal and the normal signal;

[0069] S54, generating a pipeline leakage report according to the analysis results, wherein the pipeline leakage report includes the leakage status, leakage location and leakage risk level of each pipeline section;

[0070] S55. When water leakage occurs, an alarm signal is issued and the number of the leaking pipe section and the degree of leakage are indicated.

[0071] Optionally, the S6 specifically includes:

[0072] S61. Use the signal light module to provide different color alarm displays for each pipeline section, and set different colors to correspond to different water leakage risk levels;

[0073] S62. Under normal circumstances, the signal light module displays green, indicating that there is no water leakage in the pipeline section or the risk of water leakage is low;

[0074] S63: When the risk of water leakage reaches a medium level, the signal light module displays yellow to remind staff to conduct further inspection;

[0075] S64: When the risk of water leakage is high, the signal light module displays red, issuing a strong warning and prompting staff to deal with it immediately;

[0076] S65, the signal light module continuously updates the health status of the pipeline section and provides real-time feedback on changes in leakage risks.

[0077] According to an embodiment of the present invention, an intelligent leakage detection and alarm system for water supply pipelines based on current analysis includes the following modules:

[0078] The current signal acquisition module is used to monitor the change of water conductivity and water flow status in the water supply pipeline in real time, collect current signal data and transmit it to the preprocessing module;

[0079] A preprocessing module is used to remove noise, smooth and normalize the current signal data;

[0080] The adaptive neural synaptic network module is used to receive the pre-processed current signal data, and analyze the current signal characteristics through the adaptive learning mechanism by simulating the changes of neuron and synaptic connections to identify the water leakage signal in the pipeline;

[0081] Parrot optimization algorithm module, which is used to adjust signal analysis parameters and dynamically optimize the learning strategy of the adaptive neurosynaptic network model;

[0082] The leakage status classification module is used to classify the pipeline leakage status in real time based on the output of the optimized adaptive neural synaptic network, and distinguish leakage signals from normal signals;

[0083] The segmented monitoring module is used to divide the pipeline into several monitoring sections, monitor the leakage status of each pipeline section in real time, and generate a pipeline leakage report based on the monitoring results;

[0084] The signal light module is used to display alarm signals of different colors according to the water leakage risk level of the pipeline section, and to remind the staff to handle the problem through the alarm signal;

[0085] A power supply module is used to provide electricity, using hydroelectric power or solar energy technology to provide power, and is equipped with rechargeable batteries and energy storage modules to ensure continuous operation without external power supply;

[0086] The wireless communication module is used to upload the monitoring data to the storage device and store the pipeline leakage report in the storage device, which is visualized through the display and updates the health status of the pipeline in real time.

[0087] The beneficial effects of the present invention are:

[0088] First, the present invention adopts an adaptive neural synaptic network model, which can not only accurately analyze the complex features in the current signal by simulating the changes in neurons and synaptic connections, but also dynamically optimize according to the changes in real-time data. Traditional detection methods often rely on fixed models and rules, while the present invention uses an adaptive learning mechanism to enable the model to automatically adjust according to the pipeline environment and operating status, thereby improving the recognition rate of water leakage signals. In addition, the dynamic adjustment mechanism in the model can effectively deal with signal noise caused by external interference and complex environments, further improving the accuracy and reliability of water leakage detection.

[0089] Secondly, the present invention combines the Parrot Optimization Algorithm (PA) to optimize the feature extraction process of the current signal, and can dynamically adjust the signal analysis parameters according to the changes in real-time data. Compared with the traditional optimization algorithm, the Parrot Optimization Algorithm has a stronger global search capability and can avoid the problem of local optimal solutions, thereby obtaining more accurate signal features in the feature extraction process. This optimization capability enables our system to better adapt to the complex and changeable pipeline environment, ensuring the efficient extraction and accurate judgment of water leakage signals.

[0090] In terms of leakage status classification, the present invention uses a multi-layer perceptron (MLP) classification algorithm to process the optimized feature vector, which can quickly and accurately distinguish leakage signals from normal signals. Through this efficient classification method, the present invention can classify the leakage status of the pipeline in real time and accurately determine the leakage risk level of each pipeline segment, thereby effectively reducing false alarms and missed alarms in the leakage detection process. In addition, the system can also provide alarm signals of different colors for each pipeline segment through the signal light module, intuitively display the leakage risk level to the staff, enhance the visualization and real-time nature of the monitoring process, and enable the staff to take corresponding treatment measures in a timely manner.

[0091] In terms of pipeline monitoring coverage, the present invention adopts a segmented monitoring module to divide the pipeline into multiple monitoring sections to ensure comprehensive monitoring of the pipeline. Each monitoring section is equipped with an independent current sensor monitoring node, which can collect water flow conductivity data in real time and generate a leakage status report for the pipeline section. By dividing the pipeline into multiple monitoring sections, the present invention greatly improves the accuracy and real-time performance of leakage detection, avoiding the leakage expansion problem caused by the failure to detect local leakage in time in traditional methods.

[0092] The present invention also combines hydroelectric or solar technology to provide electricity, and is equipped with rechargeable batteries and energy storage modules to ensure that the system can continue to operate without external power supply. This design not only improves the reliability and autonomy of the system, but also enables it to work normally in remote areas or environments with unstable power supply, and has strong universal application. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0094] Figure 1 This is an overall flow chart of a water supply pipeline intelligent leak detection and alarm method based on current analysis proposed by the present invention;

[0095] Figure 2 This is a structural schematic diagram of an intelligent water supply pipeline leakage detection and alarm system based on current analysis proposed by the present invention;

[0096] Figure 3 This is a schematic diagram of an implementation of an intelligent water supply pipeline leak detection and alarm method based on current analysis proposed by the present invention. DETAILED DESCRIPTION

[0097] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0098] refer to Figure 1 , a water supply pipeline intelligent leak detection and alarm method based on current analysis, comprising the following steps:

[0099] S1. Use a current sensor to monitor the conductivity change and water flow state of the water flow outside the water supply pipe in real time, collect current signal data and perform preprocessing;

[0100] S2, inputting the preprocessed current signal data into the adaptive neural synaptic network model, simulating the changes of neuron and synaptic connections, and analyzing the current signal characteristics through the adaptive learning mechanism to identify the water leakage signal in the pipeline;

[0101] S3, using the Parrot optimization algorithm to optimize the feature extraction process of the current signal, adjust the signal analysis parameters, and dynamically optimize the learning strategy of the adaptive neural synaptic network model;

[0102] S4. Based on the output of the optimized adaptive neural synaptic network model, the pipeline leakage status is classified in real time to distinguish leakage signals from normal signals;

[0103] S5. Divide the pipeline into several pipeline sections through the segmented monitoring module, monitor the leakage status of each pipeline section in real time, generate a pipeline leakage report, and issue an alarm signal for the corresponding pipeline section according to the monitoring results;

[0104] S6. Use the signal light module to provide different color alarm displays for each pipeline section. Different colors represent different levels of water leakage risk, and use alarm signals to remind staff to handle the problem;

[0105] S7. Use hydroelectric or solar technology to provide electricity, and equip with rechargeable batteries and energy storage modules as a backup mechanism when energy is insufficient;

[0106] S8. The monitoring data is uploaded to the storage device through wireless communication technology, and the pipeline leakage report is stored in the storage device, which is visualized through the display to update the health status of the pipeline in real time.

[0107] In this implementation, S2 specifically includes:

[0108] S21, the pre-processed current signal data is expressed as a time series X = [x t1 ,x t2 ,…,x tn ], where x ti represents the time step t i Current signal, n represents the total length of the time series;

[0109] S22, initializing an adaptive neural synaptic network model, wherein the adaptive neural synaptic network model includes an input layer, a synaptic adjustment layer, a dynamic hidden layer and an output layer, and setting the number of neurons in the output layer to N in , the number of neurons in the dynamic hidden layer is N hidden , the number of neurons in the output layer is N out , and initialize the weight matrix W syn and a bias vector B, where:

[0110] W syn ={w ij (t)|i∈[1,N in ],j∈[1,N hidden ],t∈[1,n]};

[0111] B = {b j |j∈[1,N hidden ]};

[0112] Among them, w ij (t) represents the synaptic weight at time step t, b j represents the bias value of the jth neuron in the dynamic hidden layer, and t represents the time step;

[0113] S23. Combine the current signal characteristics and dynamically adjust the weight through synaptic efficacy:

[0114]

[0115] in, represents the synaptic efficacy between input neuron i and hidden layer neuron j in time step t, λ represents the adjustment factor of the change rate on the weight, μ represents the adjustment factor of the signal strength, σ represents the standard deviation of the Gaussian distribution, ν represents the adjustment factor of the periodic feature, and τ represents the periodic adjustment parameter;

[0116]

[0117] Among them, w ij (t+1) represents the synaptic weight at time step t+1, and η1 represents the adjustment step size;

[0118] S24. Calculate the weighted input value of the dynamic hidden layer neuron:

[0119]

[0120] Among them, z j represents the weighted input value of the jth neuron in the dynamic hidden layer, k1 represents the regulation factor of synaptic efficacy, and k2 represents the regulation factor of high-order characteristics;

[0121] Calculate the output value of the dynamic hidden layer neuron:

[0122]

[0123] Among them, h j represents the output value of the jth neuron in the dynamic hidden layer, tanh represents the hyperbolic tangent function, and β represents the smoothing parameter;

[0124] S25, the output layer combines the dynamic hidden layer output and synaptic efficacy to calculate the leakage signal prediction value:

[0125]

[0126] Among them, y out represents the predicted value of water leakage signal, w jo represents the weight between the jth neuron in the dynamic hidden layer and the output layer, b o represents the output layer bias;

[0127] S26, normalizing the water leakage signal prediction value:

[0128]

[0129] in, represents the normalized water leakage signal prediction value;

[0130] S27. Further optimize the synaptic weight value based on error back propagation:

[0131]

[0132] Among them, L represents the loss function, represents the predicted output value at time step t, y true (t) represents the real water leakage signal at time step t, δ represents the regularization strength, ||W syn (t)||1 represents the weight matrix W syn The L1 regularization term of .

[0133] In this implementation, S3 specifically includes:

[0134] S31. Initialize the characteristics of the current signal data and construct a characteristic vector in represents the time step t i Signal characteristic value:

[0135]

[0136] in, represents the time step t i The characteristic value of the current signal, represents the time step t i The current signal, δ1, δ2, δ3 and δ4 represent the characteristic weight parameters, τ represents the period adjustment parameter, σ represents the standard deviation of the Gaussian distribution, Indicates the rate of change of the current signal;

[0137] S32, initialize the parrot optimization algorithm population P = [F1, F2, ..., F m ], each individual Represents the kth candidate feature vector and initializes the signal analysis parameter Θ k =[θ1,θ2,…,θ p ], where m represents the population size and p represents the dimension of signal analysis parameters;

[0138] S33. Define a fitness objective function to measure the matching degree between the candidate feature vector and the signal characteristics:

[0139]

[0140] Among them, J(F k ,Θ k ) represents the fitness value, β1, β2 and β3 represent the objective function adjustment coefficients, represents the kth candidate feature vector F k At time step t i The eigenvalues ​​on ;

[0141] S34. According to the imitation learning mechanism of the Parrot optimization algorithm, the candidate feature vectors and signal analysis parameters in the population are iteratively optimized:

[0142]

[0143] in, represents the updated value of the kth candidate feature vector, F g represents the optimal solution of the current population, η2 represents the optimal solution learning factor, and ζ represents the random perturbation coefficient;

[0144] S35. Adjust the optimized candidate feature vectors and signal analysis parameters to dynamically optimize the learning strategy of the adaptive neural synaptic network model:

[0145]

[0146] Among them, w ij (t+1) represents the synaptic weight at time step t+1, w ij (t) represents the synaptic weight at time step t, ξ1 represents the fitness learning factor, and ξ2 represents the parameter perturbation factor;

[0147] S36. After multiple iterations, select the feature vector F with the highest fitness value * and signal analysis parameter Θ * , F * Input to the adaptive neural synaptic network model and use Θ * Dynamically adjust the learning rate.

[0148] In this implementation manner, the S4 specifically includes:

[0149] S41, obtaining an output signal from the optimized adaptive neural synaptic network model;

[0150] S42, converting the output signal into a feature vector representation in a high-dimensional space through feature mapping, wherein the dimension of the vector representation is related to the number of features of the current signal;

[0151] S43, use the classification algorithm based on multi-layer perceptron to process the feature vector representation and output the classification vector c t =[c1,c2], where c1 represents the classification score of the water leakage signal and c2 represents the classification score of the normal signal;

[0152] S44. By comparing the classification scores, the category with the highest score is selected as the current state of the pipeline. If c1>c2, the pipeline is determined to be in a leaking state; if c2>c1, the pipeline is determined to be in a normal state.

[0153] In this implementation manner, S5 specifically includes:

[0154] S51, the segmented monitoring module divides the pipeline into several pipeline segments, each pipeline segment is equipped with a current sensor monitoring node for real-time collection of water flow conductivity data;

[0155] S52, each monitoring node generates real-time current signal data of the pipeline section according to the data collected by the current sensor, and transmits it to the central processing unit for analysis;

[0156] S53, the central processing unit processes the collected current signal data, uses the optimized neural network model to analyze the leakage state of each pipeline section, and identifies the leakage signal and the normal signal;

[0157] S54, generating a pipeline leakage report according to the analysis results, wherein the pipeline leakage report includes the leakage status, leakage location and leakage risk level of each pipeline section;

[0158] S55. When water leakage occurs, an alarm signal is issued and the number of the leaking pipe section and the degree of leakage are indicated.

[0159] In this implementation manner, S6 specifically includes:

[0160] S61. Use the signal light module to provide different color alarm displays for each pipeline section, and set different colors to correspond to different water leakage risk levels;

[0161] S62. Under normal circumstances, the signal light module displays green, indicating that there is no water leakage in the pipeline section or the risk of water leakage is low;

[0162] S63: When the risk of water leakage reaches a medium level, the signal light module displays yellow to remind staff to conduct further inspection;

[0163] S64: When the risk of water leakage is high, the signal light module displays red, issuing a strong warning and prompting staff to deal with it immediately;

[0164] S65, the signal light module continuously updates the health status of the pipeline section and provides real-time feedback on changes in leakage risks.

[0165] refer to Figure 2 , an intelligent leakage detection and alarm system for water supply pipelines based on current analysis, including the following modules:

[0166] The current signal acquisition module is used to monitor the change of water conductivity and water flow status in the water supply pipeline in real time, collect current signal data and transmit it to the preprocessing module;

[0167] A preprocessing module is used to remove noise, smooth and normalize the current signal data;

[0168] The adaptive neural synaptic network module is used to receive the pre-processed current signal data, and analyze the current signal characteristics through the adaptive learning mechanism by simulating the changes of neuron and synaptic connections to identify the water leakage signal in the pipeline;

[0169] Parrot optimization algorithm module, which is used to adjust signal analysis parameters and dynamically optimize the learning strategy of the adaptive neurosynaptic network model;

[0170] The leakage status classification module is used to classify the pipeline leakage status in real time based on the output of the optimized adaptive neural synaptic network, and distinguish leakage signals from normal signals;

[0171] The segmented monitoring module is used to divide the pipeline into several monitoring sections, monitor the leakage status of each pipeline section in real time, and generate a pipeline leakage report based on the monitoring results;

[0172] The signal light module is used to display alarm signals of different colors according to the water leakage risk level of the pipeline section, and to remind the staff to handle the problem through the alarm signal;

[0173] A power supply module is used to provide electricity, using hydroelectric power or solar energy technology to provide power, and is equipped with rechargeable batteries and energy storage modules to ensure continuous operation without external power supply;

[0174] The wireless communication module is used to upload the monitoring data to the storage device and store the pipeline leakage report in the storage device, which is visualized through the display and updates the health status of the pipeline in real time.

[0175] Figure 3 This is a schematic diagram of the implementation of the present invention, showing the specific details of using the present invention for intelligent leak detection and alarm in water supply pipelines. The outside of the pipeline and the module is wrapped with a protective cover, which has a waterproof function, and adopts a segmented monitoring method to divide the pipeline into several monitoring sections to monitor the leakage status of each pipeline section in real time. The sensing module is used to deploy sensors. The power supply module is the power supply module, the display module is visualized through the display, the transmission module is the wireless communication module, and the collection module is the current signal acquisition module. The functional module is used to realize the control function of the entire system.

[0176] Embodiment 1:

[0177] In order to verify the feasibility of the present invention in implementation, the present invention is applied to the intelligent leak detection and alarm system of a city's water supply pipeline. The system is deployed in multiple water supply pipeline sections and uses current signal analysis and intelligent algorithms to perform real-time water leakage detection and alarm, aiming to improve the accuracy and response speed of water leakage monitoring, thereby reducing water resource waste and economic losses caused by pipeline leakage.

[0178] The water supply system of a certain city includes water supply pipelines in multiple areas. These pipelines have been running underground for a long time and have great potential for leakage. In the past, the city used traditional leakage detection methods, such as manual inspections and simple pressure sensors. Although some leakage points can be found, due to the complex pipeline layout and uneven distribution of leakage problems, the traditional methods have low leakage detection efficiency, long response time, and high false alarm rate. In order to improve the efficiency and accuracy of leakage detection, the city decided to introduce an intelligent leakage detection alarm system based on current analysis.

[0179] The present invention uses a current signal acquisition module to monitor the conductivity change of water flow in the pipeline in real time, performs signal analysis through an adaptive neural synaptic network model, and combines the Parrot optimization algorithm to optimize the learning strategy, ultimately achieving real-time monitoring and classified alarm of pipeline leakage status. In a typical water supply pipeline, the system is deployed at multiple monitoring nodes, and uses wireless communication technology to upload monitoring data to a centralized control system, and the health status of the water supply pipeline is updated in real time through the display.

[0180] In practical applications, the current signal acquisition module is placed at key locations in the pipeline, which can monitor the conductivity and flow status of the water in the pipeline in real time. The pre-processed signal is analyzed by an adaptive neural synaptic network model, which automatically learns and identifies the difference between the leakage signal and the normal signal based on the training data. The Parrot optimization algorithm is used to dynamically adjust the parameters in the signal processing and learning process to ensure that the system can maintain high accuracy in different pipeline sections and different water flow conditions.

[0181] In a water leakage accident, the system successfully detected a water leakage in a certain section of the pipeline. Within 30 seconds after the water leakage occurred, the system sent a red alarm signal through the signal light module, indicating that there was a serious water leakage problem in this section of the pipeline, and transmitted the alarm information to the control center in real time through wireless communication. The staff of the control center immediately dispatched maintenance personnel to the scene to repair it, ensuring that the water leakage problem was handled in a timely manner.

[0182] Table 1 Comparison of water leakage detection effects of traditional method and the method of the present invention

[0183] project Traditional methods Method of the present invention Improvement Detection accuracy 65％ 92％ 27％ False Positive Rate 18％ 5％ 13％ Water leakage alarm response time 10-15 minutes Within 30 seconds Quick response Overall water leakage detection rate 70％ 98％ 28％ Leakage repair time 2 hours 45 minutes 75％ Water loss due to leaks 15 tons / hour 2 tons / hour 87％

[0184] According to the comparative data of water leakage detection in Table 1 above, we can analyze the beneficial effects of the present invention from several key indicators.

[0185] First, in terms of the accuracy of water leakage detection, the traditional method is only 65%, while the accuracy of the intelligent leak detection system of the present invention is increased to 92%. This difference shows that the present invention can more accurately identify water leakage signals in pipelines by introducing adaptive neural synaptic networks and parrot optimization algorithms, avoiding the common problem of missed detection in traditional methods. Traditional methods usually rely on relatively basic sensors or manual inspections, which are easily affected by environmental factors, resulting in low accuracy of water leakage detection. The system of the present invention significantly improves the accuracy of detection through real-time current signal analysis.

[0186] In terms of false alarm rate, the false alarm rate of the traditional method is 18%, while the false alarm rate of the present invention is 5%. Too high a false alarm rate usually leads to unnecessary alarms, increases the cost of manual intervention, and reduces the practicality of the system. The present invention reduces unnecessary false alarms by optimizing the signal analysis process, further improving the stability and reliability of the system. This means that the staff can focus more on the actual treatment of water leaks and reduce the waste of resources caused by false alarms.

[0187] Regarding the response time of water leakage alarm, the traditional method takes 10-15 minutes to respond to the water leakage, while the present invention can complete the alarm within 30 seconds. This difference is due to the real-time data processing and rapid response mechanism of the present invention, which enables the water leakage problem to be discovered and handled in the shortest time, avoiding further losses caused by slow response to the water leakage incident.

[0188] In terms of water leakage detection rate, the traditional method has a water leakage detection rate of 70%, while the present invention can increase it to 98%. Comprehensive detection of water leakage is crucial to the stability of the water supply system. Traditional methods are prone to miss certain water leakage points due to their limited coverage, while the present invention can comprehensively improve the water leakage detection rate through segmented monitoring and efficient algorithm optimization.

[0189] Finally, in terms of the time efficiency of water leakage repair, the traditional method takes about 2 hours, while the present invention can shorten the repair time to 45 minutes. This improvement is attributed to the fact that the present invention can help maintenance personnel respond quickly and repair water leakage problems in time through efficient water leakage reporting and accurate water leakage location positioning, thereby reducing water loss and improving maintenance efficiency.

[0190] In general, the present invention shows significant advantages in terms of water leakage detection accuracy, false alarm rate, response time, water leakage discovery rate and repair timeliness, which fully verifies its effectiveness and superiority in practical applications.

[0191] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A water supply pipeline intelligent leak detection and alarm method based on current analysis, characterized in that: The steps include: S1. Use a current sensor to monitor the conductivity change and water flow state of the water flow outside the water supply pipe in real time, collect current signal data and perform preprocessing; S2, inputting the preprocessed current signal data into the adaptive neural synaptic network model, simulating the changes of neuron and synaptic connections, and analyzing the current signal characteristics through an adaptive learning mechanism to identify the water leakage signal in the pipeline; S3, using the Parrot optimization algorithm to optimize the feature extraction process of the current signal, adjust the signal analysis parameters, and dynamically optimize the learning strategy of the adaptive neural synaptic network model; S4. Based on the output of the optimized adaptive neural synaptic network model, the pipeline leakage status is classified in real time to distinguish leakage signals from normal signals; S5. Divide the pipeline into several pipeline segments through the segmented monitoring module, monitor the leakage status of each pipeline segment in real time, generate a pipeline leakage report, and issue an alarm signal for the corresponding pipeline segment according to the monitoring results; S6. Use the signal light module to provide different color alarm displays for each pipeline section. Different colors represent different levels of water leakage risk, and use alarm signals to remind staff to handle the problem; S7. Use hydroelectric or solar technology to provide electricity, and equip with rechargeable batteries and energy storage modules as a backup mechanism when energy is insufficient; S8. The monitoring data is uploaded to the storage device through wireless communication technology, and the pipeline leakage report is stored in the storage device, which is visualized through the display to update the health status of the pipeline in real time.

2. According to claim 1, a water supply pipeline intelligent leak detection and alarm method based on current analysis is characterized in that: The S2 specifically includes: S21, the preprocessed current signal data is expressed as a time series in represents the time step t i Current signal, n represents the total length of the time series; S22, initializing an adaptive neural synaptic network model, wherein the adaptive neural synaptic network model includes an input layer, a synaptic adjustment layer, a dynamic hidden layer and an output layer, and setting the number of neurons in the output layer to N in , the number of neurons in the dynamic hidden layer is N hidden , the number of neurons in the output layer is N out , and initialize the weight matrix W syn and a bias vector B, where: W syn ={w ij (t)|i∈[1,N in ],j∈[1,N hidden ],t∈[1,n]}; B={b j |j∈[1,N hidden ]}; Among them, w ij (t) represents the synaptic weight at time step t, b j represents the bias value of the jth neuron in the dynamic hidden layer, and t represents the time step; S23. Combine the current signal characteristics and dynamically adjust the weights through synaptic efficacy: in, represents the synaptic efficacy between input neuron i and hidden layer neuron j in time step t, λ represents the adjustment factor of the change rate on the weight, μ represents the adjustment factor of the signal strength, σ represents the standard deviation of the Gaussian distribution, ν represents the adjustment factor of the periodic feature, and τ represents the periodic adjustment parameter; Among them, w ij ( t+1 ) represents the synaptic weight at time step t+1, and η1 represents the adjustment step length; S24. Calculate the weighted input value of the dynamic hidden layer neuron: Among them, z j represents the weighted input value of the jth neuron in the dynamic hidden layer, k1 represents the regulation factor of synaptic efficacy, and k2 represents the regulation factor of high-order characteristics; Calculate the output value of the dynamic hidden layer neuron: Among them, h j represents the output value of the jth neuron in the dynamic hidden layer, tanh represents the hyperbolic tangent function, and β represents the smoothing parameter; S25, the output layer combines the dynamic hidden layer output and synaptic efficacy to calculate the leakage signal prediction value: Among them, y out represents the predicted value of water leakage signal, w jo represents the weight between the jth neuron in the dynamic hidden layer and the output layer, b o represents the output layer bias; S26, normalizing the water leakage signal prediction value: in, represents the normalized water leakage signal prediction value; S27. Further optimize the synaptic weight value based on error back propagation: Where L represents the loss function, represents the predicted output value at time step t, y true( t ) represents the real water leakage signal at time step t, δ represents the regularization strength, ∥W syn ( t ) ∥1 represents the weight matrix W syn The L1 regularization term of .

3. The method for intelligent leakage detection and alarm of water supply pipeline based on current analysis according to claim 1 is characterized in that: The S3 specifically includes: S31. Initialize the characteristics of the current signal data and construct a characteristic vector in represents the time step t i Signal characteristic value: in, represents the time step t i The characteristic value of the current signal, represents the time step t i The current signal, δ1, δ2, δ3 and δ4 represent the characteristic weight parameters, τ represents the period adjustment parameter, σ represents the standard deviation of the Gaussian distribution, Indicates the rate of change of the current signal; S32, initialize the parrot optimization algorithm population P = [F1, F2, ..., F m ], each individual Represents the kth candidate feature vector and initializes the signal analysis parameter Θ k =[θ1,θ2,…,θ p ], where m represents the population size and p represents the dimension of signal analysis parameters; S33. Define a fitness objective function to measure the matching degree between the candidate feature vector and the signal characteristics: Among them, J ( F k ,Θ k) represents the fitness value, β1, β2 and β3 represent the objective function adjustment coefficients, represents the kth candidate feature vector F k At time step t i The eigenvalues ​​on ; S34. According to the imitation learning mechanism of the Parrot optimization algorithm, the candidate feature vectors and signal analysis parameters in the population are iteratively optimized: in, represents the updated value of the kth candidate feature vector, F g represents the optimal solution of the current population, η2 represents the optimal solution learning factor, and ζ represents the random perturbation coefficient; S35. Adjust the optimized candidate feature vectors and signal analysis parameters to dynamically optimize the learning strategy of the adaptive neural synaptic network model: Among them, w ij ( t+1 ) represents the synaptic weight at time step t+1, w ij (t) represents the synaptic weight at time step t, ξ1 represents the fitness learning factor, and ξ2 represents the parameter perturbation factor; S36. After multiple iterations, select the feature vector F with the highest fitness value * and signal analysis parameter Θ * , F * Input to the adaptive neural synaptic network model and use Θ * Dynamically adjust the learning rate.

4. The method for intelligent leakage detection and alarm of water supply pipeline based on current analysis according to claim 1 is characterized in that: The S4 specifically includes: S41, obtaining an output signal from the optimized adaptive neural synaptic network model; S42, converting the output signal into a feature vector representation in a high-dimensional space through feature mapping, wherein the dimension of the vector representation is related to the number of features of the current signal; S43, use the classification algorithm based on multi-layer perceptron to process the feature vector representation and output the classification vector c t = [ c1,c 2] , where c1 represents the classification score of the water leakage signal, and c2 represents the classification score of the normal signal; S44. By comparing the classification scores, the category with the highest score is selected as the current state of the pipeline. If c1>c2, the pipeline is determined to be in a leaking state; if c2>c1, the pipeline is determined to be in a normal state.

5. The method for intelligent leakage detection and alarm of water supply pipeline based on current analysis according to claim 1 is characterized in that: The S5 specifically includes: S51, the segmented monitoring module divides the pipeline into several pipeline segments, each pipeline segment is equipped with a current sensor monitoring node for real-time collection of water flow conductivity data; S52, each monitoring node generates real-time current signal data of the pipeline section according to the data collected by the current sensor, and transmits it to the central processing unit for analysis; S53, the central processing unit processes the collected current signal data, uses the optimized neural network model to analyze the leakage state of each pipeline section, and identifies the leakage signal and the normal signal; S54, generating a pipeline leakage report according to the analysis results, wherein the pipeline leakage report includes the leakage status, leakage location and leakage risk level of each pipeline section; S55. When water leakage occurs, an alarm signal is issued and the number of the leaking pipe section and the degree of leakage are indicated.

6. The method for intelligent leakage detection and alarm of water supply pipeline based on current analysis according to claim 1 is characterized in that: The S6 specifically includes: S61. Use the signal light module to provide different color alarm displays for each pipeline section, and set different colors to correspond to different water leakage risk levels; S62. Under normal circumstances, the signal light module displays green, indicating that there is no water leakage in the pipeline section or the risk of water leakage is low; S63: When the risk of water leakage reaches a medium level, the signal light module displays yellow to remind staff to conduct further inspection; S64: When the risk of water leakage is high, the signal light module displays red, issuing a strong warning and prompting staff to deal with it immediately; S65, the signal light module continuously updates the health status of the pipeline section and provides real-time feedback on changes in leakage risks.

7. An intelligent leakage detection and alarm system for water supply pipelines based on current analysis, which implements the intelligent leakage detection and alarm method for water supply pipelines based on current analysis according to any one of claims 1 to 6, characterized in that: Includes the following modules: The current signal acquisition module is used to monitor the change of water conductivity and water flow status in the water supply pipeline in real time, collect current signal data and transmit it to the preprocessing module; A preprocessing module is used to remove noise, smooth and normalize the current signal data; The adaptive neural synaptic network module is used to receive the pre-processed current signal data, and analyze the current signal characteristics through the adaptive learning mechanism by simulating the changes of neuron and synaptic connections to identify the water leakage signal in the pipeline; Parrot optimization algorithm module, which is used to adjust signal analysis parameters and dynamically optimize the learning strategy of the adaptive neurosynaptic network model; The leakage status classification module is used to classify the pipeline leakage status in real time based on the output of the optimized adaptive neural synaptic network, and distinguish leakage signals from normal signals; The segmented monitoring module is used to divide the pipeline into several monitoring sections, monitor the leakage status of each pipeline section in real time, and generate a pipeline leakage report based on the monitoring results; The signal light module is used to display alarm signals of different colors according to the water leakage risk level of the pipeline section, and to remind the staff to handle the problem through the alarm signal; A power supply module is used to provide electricity, using hydroelectric power or solar energy technology to provide power, and is equipped with rechargeable batteries and energy storage modules to ensure continuous operation without external power supply; The wireless communication module is used to upload the monitoring data to the storage device and store the pipeline leakage report in the storage device, which is visualized through the display and updates the health status of the pipeline in real time.

Images